INHIBITION OF CLASS A AND C β-lactamases: CHALLENGES AND PROMISE SARAH MICHEL DRAWZ. Submitted in partial fulfillment of the requirements

Transcription

1 IHIBITI F CLASS A AD C β-lactamases: CHALLEGES AD PRMISE by SARAH MICHEL DRAWZ Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Thesis Adviser: Robert A. Bonomo, M.D. Department of Pathology CASE WESTER RESERVE UIVERSITY May 2010

2 CASE WESTER RESERVE UIVERSITY SCHL F GRADUATE STUDIES We hereby approve the dissertation of candidate for the Ph.D. degree *. (signed) (chair of the committee) (date) *We also certify that written approval has been obtained for any proprietary material contained therein.

3 TABLE F CTETS TABLE F CTETS 1 LIST F TABLES 3 LIST F FIGURES 5 ACKWLEDGEMETS 8 LIST F ABBREVIATIS 9 ABSTRACT 10 CHAPTER 1 - Introduction verview 12 Mechanism of Action of β-lactam Antibiotics 13 Resistance to β-lactam Antibiotics 14 -Lactamases 16 Circumventing β-lactamases 28 β-lactamase Inhibitors in Clinical Practice 30 Inhibitor-Resistant Class A β-lactamases 39 The Promise of ovel β-lactamase Inhibitors 50 Inhibition of Metallo-β-Lactamases 76 Tables 78 Figures 82 CHAPTER 2 - The Role of SHV Asn276 in Clavulanate Resistance Introduction 94 Materials and Methods 97 Results 103 Discussion 107 Conclusion 114 Tables 116 Figures 121 CHAPTER 3 - Inhibition of ADC by Boronates and Carbapenems Introduction 126 Materials and Methods 128 Results 135 Discussion 141 Conclusion 148 Tables 150 Figures 153 CHAPTER 4 - Catalytic and Inhibitory Properties of the Pseudomonas aeruginosa AmpC: Implications for an Inhibitor-Resistant Phenotype Introduction 167 1

4 Materials and Methods 169 Results 181 Discussion 190 Conclusion 205 Tables 206 Figures 216 CHAPTER 5 - Summary, Future Directions, and Lessons Learned Chapter 2 Summary 222 Chapter 2 Future Directions 223 Chapter 3 Summary 225 Chapter 3 Future Directions 226 Chapter 4 Summary 227 Chapter 4 Future Directions 228 Lessons Learned 230 A Perspective 234 Table 236 APPEDIX A Molecular Modeling Terms 237 APPEDIX B Penicillin Sulfone Inhibitors of Class D β-lactamases 238 REFERECES 271 2

5 LIST F TABLES 1-1. Comparison of Ambler and Bush-Jacoby-Medeiros -lactamase classification schemes Kinetic properties of representative -lactamases Kinetic properties of representative class A inhibitor-resistant enzymes Kinetic properties of select inhibitors against different -lactamase Ambler classes 2-1. MIC values (μg/ml) of E. coli DH10B expressing SHV-1 and Asn276 variants 2-2. Kinetic properties of SHV-1 and Asn276Asp for ampicillin, piperacillin, nitrocefin, and cephalothin 2-3. Kinetic properties of SHV-1 and Asn276Asp for clavulanate, cephalothin boronic acid derivatives, methylidene penem, and meropenem Ratio of k cat /K m for IR to wild-type TEM and SHV enzymes Ratio of k inact /K i for IR to wild-type TEM and SHV enzymes K i and K i apps s of inhibitors in direct competition assays with ADC ESI-MS analysis (amu) of ADC alone and incubated with inhibitors MIC values (μg/ml) of ceftazidime and ceftazidime in combination with 4 μg/ml of boronic acid cephalothin analogs Bacterial strains used in Chapter 4 studies Primers used for cloning and mutagenesis in Chapter MIC values (μg/ml) of P. aeruginosa strains PA1 and 18SH, expressing PDC-1 and PDC-3 β-lactamases, respectively PDC-3 -lactamase substrate kinetics K i and K i apps s of inhibitors in direct competition assays with PDC-3 - lactamase and CF k inact rates of inhibitors screened in Chapter ESI-MS analysis (amu) of PDC-3 alone and incubated with inhibitors for 15 min at an I:E of 25:1 for boronate and 20:1 for PSR-3-283a and BAL

6 4-8. Cefotaxime MICs ( g/ml) of P. aeruginosa 18SH and PA1 expressing PDC-1 and PDC-3 β-lactamases, respectively, and E. coli DH10B expressing bla PDC-3 or bla PA1 in pbc SK (-), inhibitors at 4 g/ml 4-9. MIC values (μg/ml) of P. aeruginosa 18SH and E. coli DH10B expressing bla PDC-3 or bla PDC-3 variants at possible carboxylate binding sites. Inhibitors combined with 50 μg/ml ampicillin Experiments planned or in progress for Pseudomonal AmpC 236 4

7 LIST F FIGURES 1-1. Chemical structures of: (1) a penicillin; (2) a third-generation cephalosporin; (3) a monobactam; (4-7) carbapenems; and (8-10) β- lactamase inhibitors in clinical practice. The numbering scheme for penicillins, cephalosporins, and monobactams is shown. Chemical structures of β-lactamase inhibitors in development: (11-14) monobactam derivatives; (15) a penicillin derivative; (16-20) penems; (21-23) penicillin sulfones; (24) a cephalosporin sulfone; (25) a boronic acid cephalothin analog; (26-29) non-β-lactams Family portrait of β-lactamase enzymes: (A) Class A, SHV-1; (B) Class B, IMP-1; (C) Class C, E. coli AmpC; and (D) Class D, XA Proposed reaction mechanism for a penicillin β-lactam substrate by a class A serine β-lactamase enzyme in which Glu166 participates in activating a water molecule for both acylation and deacylation Schmeatic representation of the Zn 2+ -binding site of a dinuclear subclass B1 metallo- -lactamase, such as B. cereus BcII 1-5. Proposed mechanism of inhibition for class A β-lactamases by clavulanate showing the different acyl-enzyme fragmentation products (expressed in amu) that have been experimentally observed 1-6. Tautomers of imipenem hypothesized to form after acylation of carbapenems by serine β-lactamases Molecular representation of SHV-1-meropenem acyl-enzyme Molecular representation of TEM-1 active site showing residues that are most frequently implicated in the development of inhibitor-resistant TEM enzymes 1-9. Representation of proposed Henri-Michaelis preacylation complex of TEM-1 and clavulanate Proposed reaction mechanisms for the inactivation of a serine β- lactamase by: (A) BRL showing formation of the seven-membered thiazepine ring; and (B) L showing intermolecular capture by the pyridyl nitrogen leading to a bicyclic aromatic intermediate Molecular representation of cross-linked active site residues Ser64 and Lys315 formed after aminolysis of -aryloxycarbonyl hydroxamate inhibitor in E. cloacae P99 -lactamase Chemical structures of compounds tested in Chapter

8 2-2. Immunoblot of E. coli DH10B expressing SHV-1, SHV Arg244Ser variant, and Asn276 variants probed with SHV-1 polyclonal antibody 2-3. Deconvoluted ESI-MS spectra of: (A) SHV-1; and (B) SHV Asn276Asp before and after 15 min inactivation with clavulanate at inhibitor:enzyme ratio of 1000: Molecular representation of SHV Asn276Asp based on SHV-1 showing the increased interaction between Arg244 and 276Asp Proposed reaction mechanism for inactivation of SHV-1 by clavulanate Schemes illustrating the interactions of a serine β-lactamase with: (A) the β- lactam cephalosporin ceftazidime; (B) the boronic acid ceftazidime analog, compound 2; and (C) the carbapenem imipenem 3-2. Chemical structures of: (A) commercially available inhibitors and cephalosporin substrate cephalothin; (B) boronic acid derivatives; and (C) carbapenems used in Chapter verlay of the molecular coordinates for the E. coli AmpC covalently bound to cephalothin substrate and boronic acid chiral cephalothin analog 3-4. Deconvoluted mass spectra of: (A) ADC β lactamase alone; (B) ADC after 15 min incubation with compounds 2 and 5; and (C) ADC β- lactamase after 15 min incubation with imipenem, ertapenem, doripenem, and meropenem 3-5. Proposed mechanism of the retroaldolic reaction leading to elimination of C 6 hydroxethyl substituent from the -lactamase: carbapenem acylenzyme 3-6. Discovery Studio multiple sequence protein alignment of crystal structure coordinates for E. cloacae P99, E. coli AmpC, and E. coli CMY-2, and molecular models of PDC and ADC 3-7. verlay of molecular coordinates for the E. coli AmpC-5 complex and generated ADC-5 model colored by atom 3-8. Comparison of the binding site interactions between E. coli AmpC-5 (left panel) and ADC-5 (right panel) 3-9. Molecular representation of: (A) ADC-imipenem acyl-enzyme model; and (B) ADC-meropenem acyl-enzyme model 4-1. Chemical structures of: (A) β-lactam substrates; and (B) investigational inhibitors tested in Chapter Specific activity ( M CF hydrolyzed/sec/ g protein) of crude protein extract from P. aeruginosa PA1 and 18SH without and following 3 hr

9 incubation with 50 µg/ml cefoxitin 4-3. Immunoblots of: (A) purified PDC-3 protein; and (B) purified β- lactamase proteins from all Ambler classes and crude lysates of clinical strains 4-4. Proposed reaction intermediates detected on ESI-MS after 15 min incubation of: (A) PSR-3-283a; and (B) BAL29880 with PDC-3 at an I:E of 20:1. (C) Possible mechanism for elimination of the S 3-2 group from BAL Structure of C 6 α-hydroxymethyl penicillanate studied by Mobashery and colleagues 4-6. Solvent-accessible surfaces of E. coli AmpC bound to the chiral cephalothin analog boronate, compound 4c, and P. aeruginosa PDC-3 with the same inhibitor

10 ACKWLEDGEMETS Many people have helped me complete this project. In particular, I thank my thesis committee for their time and attention to my academic development, and the members of the Bonomo lab for support and scientific guidance. My husband, Paul, has been my strongest source of support through these experiences. The encouragement from my parents has also been instrumental. My children, Maggie and Jack, were born during my pursuit of this degree, and are my greatest joys. Your lives are inextricably woven throughout this work; you are on my mind even when I m thinking about β-lactamases. My utmost thanks to Robert. You have taught me so much about science, and shared your excitement and skill for research. I admire you not just for what you have accomplished as a physician scientist, but also for how you have done it - with consideration and respect for others. Thank you for investing and believing in me. I will always be proud of having been part of your group. 8

11 LIST F ABBREVIATIS ADC AmpC amu bla ESI-MS FPLC IPTG IR IRT LB MIC MDR CF PBS PDB PDC pief SDS-PAGE SB SHV TEM WT Acinetobacter-derived cephalosporinase cephalosporinase β-lactamase belonging to Ambler class C atomic mass units β-lactamase gene electrospray ionization mass spectrometry fast protein liquid chromatography isopropyl-β-d-thiogalactopyranoside inhibitor resistant inhibitor-resistant TEM lysogeny broth minimum inhibitory concentration multidrug resistant nitrocefin phosphate-buffered saline Protein Data Bank Pseuodmonas-derived cephalosporinase preparative isoelectric focusing sodium dodecyl sulfate polyacrylamide gel electrophoresis super optimal broth Ambler class A β-lactamase named after the property sulfhydryl reagent variable Ambler class A β-lactamase named after the patient (Temoneira) from whom the first β-lactamase was isolated wild-type 9

12 Inhibition of Class A and C β-lactamases: Challenges and Promise Abstract by SARAH MICHEL DRAWZ Since the introduction of penicillin, β-lactam antibiotics have been the antimicrobial agent of choice for the treatment of many infections. Unfortunately, the efficacy of these life-saving antibiotics is significantly threatened by bacterial β-lactamase enzymes. To overcome β-lactamase-mediated resistance, β-lactamase inhibitors were introduced (clavulanate, sulbactam, and tazobactam). These inhibitors greatly enhance the efficacy of their partner β-lactams in the treatment of Gram-negative infections. However, selective pressure from excess antibiotic use accelerated the emergence of resistance to β-lactam/βlactamase inhibitor combinations. Furthermore, the prevalence of clinically relevant β- lactamases that are intrinsically resistant to inhibition is rapidly increasing. There is an urgent need for effective inhibitors that can restore the activity of β-lactams. Here, we demonstrate that the Asn276Asp substitution confers resistance to clavulanate in the class A SHV β-lactamase. Unlike the Asn276Asp substitution in the related TEM enzyme, and inhibitor-resistant β-lactamases in general, the SHV variant maintains a high level of catalytic efficiency for penicillins. This fine-tuning of the inhibitor-resistant phenotype may represent a significant evolutionary advance, as the enzyme maintains a balance of desired catalytic properties. By probing Asn276 with selectively designed 10

13 inhibitors, we explored how the configuration of the conserved β-lactam carboxylate impacts binding. Despite relative distance from the active site, this second-shell residue exerts important effects on enzyme-ligand interactions. ur work also addresses the class C β-lactamases from Acinetobacter spp. and Pseudomonas aeruginosa, pathogens of increasing clinical concern for which few effective therapeutic options remain. The currently available β-lactamase inhibitors are inactive against these enzymes, and thus development of second-generation agents is a priority. We first studied boronic acid derivatives bearing side chains of the enzymes substrates as potential inhibitors. Insights about which recognition elements lead to low binding constants provide important leads for class C β-lactamase inhibitor design. We next examined how carbapenems, drugs of last resort for many resistant infections, may also behave as effective β-lactamase inhibitors through unique active site chemistry. Finally, mass spectrometry, kinetics, and susceptibility testing shed light on the possible reaction mechanisms of investigational inhibitors with diverse structures, revealing that inactivation of these enzymes is attainable, but significant barriers to in vivo activity remain. 11

14 CHAPTER 1 Introduction Reproduced in part with permission from THREE DECADES F β-lactamse IHIBITRS Sarah M. Drawz and Robert A. Bonomo Clinical Microbiology Reviews January 2010; Volume 23 (1): verview The development of antibiotics remains one of the most significant advances in modern medicine (413). Antibiotics have saved countless lives and continue to be a mainstay of therapy for bacterial infections. The clinical success of the first β-lactam, Penicillin G (benzylpenicillin, Figure 1-1, 1), prompted the search and development of additional derivatives. This quest gave rise to the β-lactam antibiotics in clinical use today (penicillins, narrow- and extended-spectrum cephalosporins, monobactams, and carbapenems, Figure 1-1, 1-7) (14). The common structural feature of these classes of antibiotics is the highly reactive four-membered β-lactam ring. Unfortunately, β-lactamase mediated resistance to β-lactam antibiotics emerged as a significant clinical threat to these life-saving drugs. In response to this challenge, two strategies were advanced to preserve the utility of β-lactam antibiotics: (i) discover or 12

15 design β-lactam antibiotics that are able to evade bacterial enzymatic inactivation conferred by β-lactamases; (ii) inhibit β-lactamases so the partner β-lactam can reach the penicillin binding proteins (PBPs), the target of β-lactam antibiotics. This chapter begins with a review of the catalytic mechanisms of each β-lactamase class. ext is a summary of the approaches for circumventing β-lactamase-mediated resistance, including mechanisms of action and salient clinical and microbiological features of β-lactamase inhibitors. The following section is a survey of the problem of resistance to β-lactamase inhibitors, explaining the important changes in class A β- lactamases that define this phenotype. The final section reviews compounds that have demonstrated favorable inactivation properties against β-lactamases, and introduces newer compounds showing promise as second-generation inhibitors. Mechanism of Action of β-lactam Antibiotics β-lactam antibiotics exhibit their bactericidal effects by inhibiting enzymes involved in cell wall synthesis. The integrity of the bacterial cell wall is essential to maintaining cell shape in a hypertonic and hostile environment (292). smotic stability is preserved by a rigid cell wall comprised of alternating -acetylmuramic acid (AM) and - acetylglucosamine (AG) units. These glycosidic units are linked by transglycosidases. A pentapeptide is attached to each AM unit, and the cross-linking of two D-alanine-Dalanine AM pentapeptides is catalyzed by PBPs, which act as transpeptidases (172, 427). This cross-linking of adjacent glycan strands confers the rigidity of the cell wall. The β-lactam ring is sterically similar to the D-alanine-D-alanine of the AM pentapeptide and PBPs mistakenly use the β-lactam as a building block during cell wall synthesis (514). This results in acylation of the PBP which renders the enzyme 13

16 unable to catalyze further transpeptidation reactions (152). As cell wall synthesis slows to a halt, constitutive peptidoglycan autolysis continues. The breakdown of the murein sacculus leads to cell wall compromise and increased permeability. Thus, the β-lactammediated inhibition of transpeptidation causes cell lysis, although the specific details of penicillin s bactericidal effects are still unraveling (20). Resistance to β-lactam Antibiotics There are four primary mechanisms by which bacteria can overcome β-lactam antibiotics (14): 1. Production of β-lactamase enzymes. This is the most common and important mechanism of resistance in Gram-negative bacteria and will be the focus of this thesis. 2. Alterations in PBPs. Changes in the active site of PBPs can lower the affinity for β-lactam antibiotics and subsequently increase resistance to these agents, such as those seen in PBP2x of Streptococcus pneumoniae (253). Through natural transformation and recombination with DA from other organisms, eisseria spp. and Streptococcus spp. have acquired highly resistant, low-affinity PBPs (47, 357, 514). In a related manner, penicillin resistance in Streptococcus sanguis, Streptococcus oralis, and Streptococcus mitis developed from horizontal transfer of a PBP2b gene from Streptococcus pneumoniae (129, 396). Methicillin resistance in Staphylococcus spp. is also a significant clinical challenge. While there are many reasons for this resistance, the β- lactam resistance phenotype is also conferred by acquisition of the meca gene which produces PBP2a (also denoted PBP2 ) (88). PBP2a can assemble new cell wall in the presence of high concentration of penicillins and cephalosporins. 14

17 3. Decreased expression of outer membrane proteins (MPs). In order to access PBPs on the inner plasma membrane, β-lactams must either diffuse through or directly traverse porin channels in the outer membrane of Gram-negative bacterial cell walls. Some Enterobacteriaceae (e.g., Enterobacter spp., Klebsiella pneumoniae, and Escherichia coli) exhibit resistance to carbapenems based on loss of these MPs; the loss of prd is associated with imipenem resistance and reduced susceptibility to meropenem in the nonfermenter Pseudomonas aeruginosa (204, 224, 276, 332, 349, 444). Resistance to imipenem and meropenem has also been associated with the loss of the Car MP in clinical isolates of multidrug-resistant Acinetobacter baumannii (326, 394). Point mutations or insertion sequences in porin-encoding genes can produce proteins with decreased function and thus lower permeability to β-lactams (128). f note, the disruption of porin proteins alone is not always sufficient for producing the resistance phenotype, and typically this mechanism is found in combination with β- lactamase expression (128, 274). 4. Efflux pumps. Efflux pumps, as part of either an acquired or intrinsic resistance phenotype, are capable of exporting a wide range of substrates from the periplasm to the surrounding environment (395). These pumps are an important determinant of multidrug resistance in many Gram-negative pathogens, particularly P. aeruginosa and Acinetobacter spp. In P. aeruginosa, upregulation of the MexA-MexB-prD system, in combination with the organism s low outer membrane permeability, can contribute to decreased susceptibility to penicillins, cephalosporins, as well as quinolones, tetracycline, and chloramphenicol ( , 421, 437). To illustrate, an increase in the carbenicillin minimum inhibitory concentration (MIC) from 32 μg/ml to 1028 μg/ml is associated with overproduction of this efflux pump (10, 268). Additionally, an 15

18 upregulated efflux pump (e.g., AdeABC, an RD-type efflux pump in A. baumannii) can augment the carbapenem resistance conferred by a catalytically poor β-lactamase (e.g., XA-23) (195, 377). -Lactamases The first β-lactamase enzyme was identified in Bacillus (Escherichia) coli before the clinical use of penicillin. In a sentinel paper published nearly seventy years ago, E. P. Abraham and E. Chain described the B. coli penicillinase (1). The enzyme was not thought to be clinically relevant since penicillin was targeted to treat staphylococcal and streptococcal infections, and Abraham, Chain, and their colleagues were unable to isolate the enzyme from these Gram-positive organisms (2, 65). It is sobering now to consider the ramifications of this observation. Four years later, Kirby successfully extracted these cell-free penicillin-inactivators from S. aureus which foreshadowed the emergence of a significant clinical problem (242). The growing number of β-lactam antibiotics has since increased the selective pressure on bacteria, promoting the survival of organisms with multiple β-lactamases (292, 303). Currently, more than 850 β-lactamases are now identified (Dr. Bush, personal communication). The rapid replication rate, recombination rates, and high mutation frequency likely permit bacteria to adapt to novel β-lactams by evolution of these β- lactamases (376). Classification Two major classification schemes exist for categorizing β-lactamase enzymes: Ambler classes A through D based on amino acid sequence homology, and Bush-Jacoby- 16

19 Medeiros groups 1 through 4 based on substrate and inhibitor profile (Table 1-1) (8, 68). A family portrait reveals the structural similarity of class A, C, and D serine β- lactamases (Figure 1-2). Class B β-lactamases ( a class apart ) are metallo-β-lactamases (MBLs) (63). MBLs possess either a single or pair of Zn 2+ ions coordinated to His/Cys/Asp residues in the active site. In this review, the Ambler classification scheme will be used. Class A. Serine β-lactamases, e.g., TEM, SHV. In general, these enzymes are susceptible to the commercially available β-lactamase inhibitors (clavulanate, tazobactam, and less so to sulbactam); although the K. pneumoniae carbapenemase KPC may be an important exception to this generalization (363). The first plasmid-mediated β- lactamase was identified in E. coli in 1963 (and reported in 1965), and was named TEM after the patient from whom it was isolated, Temoniera (119). SHV, another common β-lactamase found primarily in K. pneumoniae, was named from the term sulfhydryl reagent variable. Early studies of SHV-1 showed that p- chloromercuribenzoate inhibited the hydrolysis of cephaloridine, but not that of benzylpenicillin (297). TEM and SHV are common β-lactamases detected in clinical isolates of E. coli and K. pneumoniae, pathogens responsible for urinary tract, hospitalacquired respiratory tract, and bloodstream infections (72, 177, 420). While SHV-1 and TEM-1 share 68% sequence homology, the active site of SHV-1 is approximately Å wider than in TEM-1, which has important structural implications, especially related to the substrate profiles of SHV variants (474). Although bla TEM and bla SHV may be found on plasmids, other class A enzymes are encoded on the chromosome, such as bla PenA from Burkholderia pseudomallei, or on 17

20 integrons (e.g., bla GES-1 from K. pneumoniae and bla VEB-1 in P. aeruginosa and A. baumannii) (68, 328). Class A Extended-Spectrum β-lactamases (ESBLs). e.g., TEM-3, SHV-2, CTX-M-15, PER-1. The growing number of β-lactamases in E. coli and K. pneumoniae, as well as the emergence of these enzymes in other pathogens (e.g., H. influenzae and. gonorrhoeae), led to the development of extended-spectrum cephalosporins with an oxyimino-side chain, carbapenems, cephamycins, and monobactams (83, 140, 225, 366). Upon the introduction of the penems and cephems in the early 1980 s, these new agents were effective against many β-lactam resistant bacteria. However, selective pressure quickly fostered the emergence of extended-spectrum β-lactamases (ESBLs) which could hydrolyze many of the oxyimino-cephalosporins (223, 225). In a dramatic parallel to the observations of Abraham and Chain, within two years of the introduction of the extended-spectrum cephalosporins, cefotaxime and ceftazidime, novel ESBLs were reported in E. coli and K. pneumoniae (243). Interestingly, these "new β-lactamases" harbored point mutations in the parent bla TEM-1 and bla SHV-1 genes that led to single amino acid changes in the β-lactamases. ther ESBLs, such as CTX-M, arose by plasmid transfer from preexisting chromosomal ESBL genes from Kluyvera spp., typically non-pathogenic commensal organisms (14, 37). CTX-M ESBLs now represent important enzymes found in isolates from the community and are the most commonly isolated ESBLs in many parts of the world, particularly Europe (387). ESBLs hydrolyze penicillins, narrow- and extended-spectrum cephalosporins (including the anti-mrsa cephalosporin ceftobiprole), and the monobactam aztreonam (9, 366, 405). In contrast, ESBLs cannot efficiently degrade cephamycins, carbapenems, and β- lactamase inhibitors. Since their initial description, many more than 200 different ESBLs 18

21 have been identified, posing a significant risk to public health and hospitalized patients in intensive care units where infection with an ESBL may lead to significant morbidity and mortality (up-to-date listings of verified ESBLs sequences are available on the website maintained by Drs. Jacoby and Bush) (366). The majority of ESBLs are from the SHV, TEM, and CTX-M families, less frequently derived from BES, GES, VEB, and PER enzymes, and sometimes these enzymes do not belong to any defined family (328). Class A Serine Carbapenemases. e.g., MC-A, IMI, SME, KPC. Members of this group of β-lactamases can hydrolyze carbapenems as well as cephalosporins, penicillins, and aztreonam (404). These carbapenem-hydrolzying enzymes have been identified primarily in Enterobacter cloacae, Serratia marcescens, and K. pneumoniae, bacteria which often harbor multiple resistance determinants, narrowing the range of treatment options (335, 336, 408, 509, 510). The bla gene for the former two organisms is typically found on the chromosome, while the K. pneumoniae carbapenemase bla KPC gene is carried on plasmids containing Tn4401 (337). MIC testing for carbapenems in carbapenemase-expressing strains can vary from moderately increased ( 4 µg/ml) to resistant ( 32 µg/ml) (143, 404). Class B. Metallo-β-lactamases, e.g., IMP, VIM, CcrA, L1. Class B enzymes are Zn 2+ - dependent β-lactamases that demonstrate a different hydrolytic mechanism than the serine β-lactamases of classes A, C, and D (68). rganisms producing these enzymes usually exhibit resistance to penicillins, cephalosporins, carbapenems, and the clinically available β-lactamase inhibitors (489). Interestingly, the hydrolytic profile of MBLs does not typically include aztreonam. MBLs likely evolved separately from the other Ambler classes which have serine at their active site (292). The bla MBL genes are located on the 19

22 chromosome, plasmid, and integrons (161, 489). P. aeruginosa, K. pneumoniae, and A. baumannii produce class B enzymes encoded by mobile genetic elements (108, 213, 215). In contrast, Bacillus spp., Chryseobacterium spp., and Stenotrophomonas maltophilia possess chromosomally encoded MBLs, but the majority of these host pathogens are not frequently responsible for serious infections (489). The role these -lactamases in Bacillus spp. and Chryseobacterium spp. will play in the clinical arena is still unknown. Class C. Serine cephalosporinases, e.g., CMY-2, P99, ACT-1, DHA-1. This group includes the AmpC β-lactamases which are usually encoded by bla genes located on the bacterial chromosome, although plasmid-borne AmpC enzymes are becoming more prevalent (384). rganisms expressing the AmpC β-lactamase are typically resistant to penicillins, β-lactamase inhibitors, and cephalosporins including cefoxitin, cefotetan, ceftriaxone, and cefotaxime. AmpCs poorly hydrolyze cefepime and are inhibited by cloxacillin, oxacillin, and aztreonam (Figure 1-1, 3) (68). Production of chromosomal AmpCs in Gram-negative bacteria is at a low level ( repressed ), but can be derepressed by induction with certain β-lactams, particularly cefoxitin and imipenem (14, 26, 180). The mechanism of this regulation have been the subject of intense investigation (the reader is referred to reference (221)). f significant concern is the selection of mutant bacterial populations that are genetically derepressed for AmpCs production, which can cause a dramatic increase in MICs during the course of β-lactam therapy (e.g., after 14 days of ceftazidime therapy, a strain of P. aeruginosa was selected that increased MICs from 1 to 32 µg/ml) (222, 229, 273). Class D. Serine oxacillinases, e.g., XA-1, XA-10, XA-23, XA-48, and XA- 24/40. Class D β-lactamases were initially categorized as oxacillinases because of their ability to hydrolyze oxacillin at a rate of at least 50% that of benzylpenicillin, in contrast 20

23 to the relatively slow hydrolysis of oxacillin by classes A and C (116). In bacteria, XA β-lactamases can also confer resistance to penicillins, cephalosporins, extended-spectrum cephalosporins (XA-type ESBLs), and carbapenems (XA-type carbapenemases). Generally speaking, XA enzymes are resistant to inhibition by clavulanate, sulbactam, and tazobactam, (with some exceptions, e.g., XA-2 and XA-32 are inhibited by tazobactam, but not sulbactam and clavulanate; XA-53 is inhibited by clavulanate) (116, 324, 327, 393). Interestingly, sodium chloride at concentrations > mm inhibits some carbapenem-hydrolyzing oxacillinases (e.g., XA-25 and XA-26) (6). Site-directed mutagenesis studies suggest that susceptibility to inhibition by sodium chloride is related to the presence of a Tyr residue at position 144 (194, 327). Presumably, Tyr144 may facilitate sodium chloride binding better than the Phe residue found in resistant oxacillinases, although the molecular mechanism remains unexplained. Examples of XA enzymes include those rapidly emerging in A. baumannii (e.g., XA- 23, XA-24/40) and constitutively expressed in P. aeruginosa (e.g., XA-50) (170, 394, 490). β-lactamase Hydrolytic Mechanisms Serine β-lactamases acylate β-lactam antibiotics, much like PBPs, and then use strategically positioned water molecules to inactivate the acylated β-lactam (313). In this manner, the β-lactamase is regenerated and can inactivate additional β-lactam molecules. This enzymatic reaction may be represented by the following equation: k 1 E + S E:S E-S E+P k -1 k 2 k 3, H 2 21

24 In this scheme, E is a β-lactamase, S is a β-lactam substrate, E:S is the Henri-Michaelis complex, E-S is the acyl-enzyme, and P is the product devoid of antibacterial activity. The rate constants for each step are represented by: k 1, k -1, k 2, and k 3 ; k 1 and k -1 = association and dissociation rate constants for the preacylation complex, respectively; k 2 = acylation rate constant; and k 3 = deacylation rate constant. More complicated branched reaction mechanisms are seen with some -lactamases (156). In order to be uniform, we now define basic terms that are often used to describe the kinetic behavior of - lactamases. The Michaelis constant, K m, is defined as (156): K m = k 3 K s / (k 2 + k 3 ) where the kinetic constant, K s, is (k -1 + k 2 ) / k 1. The turnover number, k cat, is a composite rate constant that represents multiple chemical steps and is defined as (107): k cat = k 2 k 3 / (k 2 + k 3 ) or: V max = k cat [E] K m, expressed in terms of concentration, represents the relative affinity of the ES encounter and the rate at which the ES is converted to P; a large K m value represents poor affinity (large K s ). Class A. In Figure 1-3, we represent the reaction scheme of a typical class A - lactamase. In brief: (i) after formation of the Henri-Michaelis complex, the active site serine performs a nucleophilic attack on the carbonyl of the β-lactam antibiotic that results in a high-energy tetrahedral acylation intermediate ( 1); (ii) this intermediate transitions into a lower energy covalent acyl-enzyme following protonation of the β- 22

25 lactam nitrogen and cleavage of the C- bond; (iii) next, an activated water molecule attacks the covalent complex and leads to a high-energy tetrahedral deacylation intermediate ( 2); (iv) hydrolysis of the bond between the β-lactam carbonyl and the oxygen of the nucleophilic serine regenerates the active enzyme and releases the inactive β-lactam. Acylation and deacylation require the activation of the nucleophilic serine and hydrolytic water, respectively. An entire body of experimentation is devoted to examining the mechanistic roles of the individual residues in the active site of β-lactamases; here we present only several leading insights to provide a foundation for the discussion of inhibitory pathways. For class A enzymes, a predominant theory holds that Glu166 acts as the activating base of the hydrolytic water in deacylation (3, 199, 441). In contrast, the acylation mechanism is less clear, and several hypotheses and lines of evidence exist. The first notion proposes that Glu166, via a water molecule, deprotonates the Ser70 hydroxyl before addition to the β-lactam bond. This Glu166 general base theory is supported by quantum mechanical/molecular mechanical (QM/MM) calculations and the protonated status of Glu166 in the ultrahigh resolution (0.85 Å) structure of TEM-1 in complex with an acylation transition state analog (Research Collaboratory for Structural Bioinformatics Protein Data Bank entry 1M40) (196, 313). Surprisingly, substitutions at residue 166 do not prevent acylation (168, 178, 247, 441). A second view of acylation maintains that Lys73 can serve as the general base deprotonating Ser70 (199, 308, 450). Recent QM/MM studies by Meroueh and colleagues demonstrate that the pathway involving Lys73 as the activating base is energetically favorable, and may exist in competition with the pathway where Glu166 serves as the activating residue (308). Additional ideas on the mechanism of acylation have been 23

26 proposed, including both a role for the C 3 /C 4 substrate carboxylate and Ser130 in activating Ser70 and a general acid pathway where protonation of the β-lactam nitrogen is the primary event (13, 123). Further clarification of the acylation mechanism may prove useful for the design of inhibitors that capitalize on the enzymes native machinery. Highly conserved structural motifs in class A enzymes include four polar regions: (i) the active site pocket Ser70-Xaa-Xaa-Lys73; (ii) the Ω-loop residues ; (iii) Ser130-Asp131-Asn132; and (iv) Lys234-Thr(Ser)235-Gly236. The roles of residues in the SD loop comprised of Ser130-Asp131-Asn132 are related to maintaining the structure of the active site cavity, enzyme stability, and stabilization of the enzyme transition state, respectively (220, 348). The functional explanation for the conservation of the Lys234-Thr(Ser)235-Gly236 triad is less well understood. Removal of the hydroxyl group from the middle residue has little impact on penicillin hydrolysis in TEM (but more on the catalytic efficiency of cephalosporins), and despite what may be expected from crystal structure alignments, the study of modified β-lactams argues against an interaction between the β-lactam carboxylate and the Thr(Ser) hydroxyl group (137, 219). In SHV, the Thr235Ala substitution lowers MICs to both penicillins and cephalosporins (209). Interestingly, the expression of the structurally conserved SHV Thr235Ser variant has no effect on -lactam MICs, but does lead to an increased susceptibility to the piperacillintazobactam combination (209). The most common mechanism for transformation of a broad-spectrum β-lactamase to an ESBL is point mutations that result in amino acid sequence changes near the enzyme active site, facilitating hydrolysis of oxyimino-cephalosporins (223, 385). However, the specific amino acid replacements and resistance mechanisms vary between enzymes (151). In TEM, changes at Ambler positions Arg164 (-His, -Ser), Gly238 (-Ser, -Ala), and 24

27 Glu240 (-Lys) result in variants that confer the ESBL phenotype ( (48). Several crystallographic studies demonstrated that the new TEM active site is expanded or remodeled, as compared to TEM-1, to accommodate the larger side chain of expanded-spectrum cephalosporins (95, 246, 342, 347). The crystal structure of SHV-2 (Gly238Ser) at 0.91 Å resolution (PDB 19B), as compared to SHV-1, showed a displacement in the b3 β-strand containing residues which created an expanded β-lactam binding site, but preserved the positioning of the essential catalytic residues (151, 342). Similarly, the PER-1 ESBL active site is expanded by a novel fold of the Ω loop and the insertion of four residues after Lys240 (471). In contrast, the crystal structure of the acyl-enzyme intermediate of CTX-M Toho-1 in complex with cefotaxime did not resemble the enlarged active sites seen with the ESBL TEM and SHV β-lactamases. Rather, the Toho-1 enzyme facilitates cephalosporin binding through movement of the Ω-loop towards the active site, interactions between conserved residues Asn104 and Asp240 and the bulky side chain of cefotaxime, and contacts with Ser237 that help position the β-lactam carbonyl in the oxyanion hole (214, 432). The high-resolution crystal structures of four additional CTX-M β-lactamases confirmed that the active sites of these enzymes were not enlarged as compared to the narrow-spectrum TEM-1 and SHV-1 (95, 97). Instead, the structural basis for the extended substrate profiles in the CTX-M enzyme family appears due to specific amino acid (i.e., Ser237, Asp104) interactions with the oxyimino-cephalosporins and increased mobility of the b3 β-strand (95). verall, this increased mobility and activity costs the enzyme thermal stability, a theme recapitulated by the evolution of many resistance phenotypes (433). 25

28 Point mutations are not the only mechanism for the ESBL phenotype, and other alterations in enzyme regulation produce resistance, such as promoter sequence changes leading to increased enzyme expression or alterations in outer membrane porins (338, 346). At this point in time, the biochemical basis for class A serine carbapenemase activity is an area of active investigation. Current evidence suggests that this property rests upon a remodeling of the active site. Unlike class A ESBLs (expanded or enhanced mobility of the b3 β-strand), the active site of class A carbapenemases such as KPC-2, nonmetallocarbapenamase of class A (MC-A), and SME-1 reveal that there are multiple alterations in the spatial position of catalytic residues (239, 382, 436, 451). We are just beginning to understand how this remodeling allows enhanced carbapenemase activity when compared to other class A enzymes. Class C. The prevailing evidence regarding the class C hydrolytic mechanism suggests that Tyr150 behaves as a general base by increasing the nucleophilicity of Ser64 for acylation (343). Early studies of the deacylation mechanism indicated that Tyr150 also acted as the catalytic base, accepting a proton from the deacylating water (136, 343). For example, QM/MM calculations suggested that Tyr150 interacts with Lys67 in a conjugate base manner (165). However, the crystal structure of the E. coli AmpC in complex with a deacylation transition state analog revealed that Tyr150 remains protonated throughout the reaction, and therefore is unlikely to be the anionic base that deprotonates the catalytic water for hydrolysis (96). Instead, deacylation may proceed by substrate-activated catalysis whereby the nitrogen on the β-lactam acts as a base and transiently accepts the proton from the hydrolytic water, while the Tyr150 proton helps stabilize the water s developing negative charge (60, 96). The crystallographic evidence and mutagenesis 26

29 studies do not rule out the role of Lys67 in the coordinate base mechanism, and further studies exploring the role of Lys67 are anticipated (96, 318). Class D. ur understanding of the hydrolytic mechanism of class D β-lactamases is based on the careful study of XA-10, -13, and -1 (299, 353, 378, 429, 447, 448). This class of enzymes is unique because of the direct role of carboxylation of the active site Lys70. The carbamic acid on Lys70 can ionize to yield a carbamate that hydrogen bonds with the nucleophilic Ser67 residue. In this manner, the carboxylated Lys70 may serve as the general base by activating both Ser67 for acylation and the hydrolytic water for deacylation (174, 298, 299). Class B. In contrast to the above serine β-lactamases, class B includes Zn 2+ -dependent enzymes that follow a different hydrolytic mechanism. In general, MBLs use the -H group from a water molecule that is coordinated by Zn 2+ to hydrolyze the amide bond of a β-lactam. MBLs are divided into three classes based on their Zn 2+ dependency, whether they: (i) are fully active with either one or two ions (subclass B1, e.g., IMP-1, VIM-2, BcII, CcrA); (ii) require two ions (subclass B3, e.g., L1); or (iii) employ one ion and are inhibited by binding of an additional ion (subclass B2, e.g., CphA) (158, 197, 198, 369, 370). Crowder and colleagues compiled the recent structural and mechanistic data and summarized the common features for the three subclasses of MBLs (111). MBLs utilizing two Zn 2+ ions for hydrolysis, such as subclass B3 S. maltophilia L1, coordinate the β- lactam substrate by the carboxylate and carbonyl groups, bridged by a hydroxide ion. After substrate binding, one of the Zn 2+ ions, in conjunction with enzyme residues, polarizes the β-lactam carbonyl for attack by the H, which is hydrogen-bonded to deprotonated Asp120. ucleophilic attack by the -H creates a tetrahedral species which rapidly collapses into an intermediate where the β-lactam nitrogen is anionic. Protonation 27

30 of the nitrogen leads to product formation. The source of the proton is not certain, and may come from Asp120 or a water molecule. The B1 enzyme Bacillus cereus BcII is active in both its mononuclear and dinuclear forms, and in the resting state, the Zn 2+ -bound H is hydrogen bonded to the deprotonated Asp120 and coordinated to several other active site residues (Figure 1-4) (111, 495, 496). After attack by the H, the breakdown of the tetrahedral intermediate requires protonation of the β-lactam nitrogen; the source of this proton is under investigation. In the case of the CphA B2 MBL from A. hydrophila, a second bound Zn 2+ ion is inhibitory. The proposed mechanism includes a water molecule activated by either His118 or Asp120, rather than a Zn 2+ -bound H, whereas the singular Zn 2+ appears to help coordinate the -lactam nitrogen (160, 506). Circumventing β-lactamases β-lactamase Inhibitors: Mechanistic Considerations A successful strategy for combating β-lactamase-mediated resistance is the use of agents designed to bind at the active site, and hence are frequently β-lactams. This strategy can take two forms: (i) create substrates that reversibly and/or irreversibly bind the enzyme with high affinity, but form unfavorable steric interactions as the acylenzyme; or (ii) develop mechanism-based or irreversible suicide inhibitors (61). Examples of the former are extended-spectrum cephalosporins, monobactams, or carbapenems which form acyl-enzymes and adopt catalytically incompetent conformations that are poorly hydrolyzed (Figure 1-1, 2, 3, 4-7, respectively). For reversible inhibition, the reaction can be described as was shown above in the equation 28

31 for the hydrolytic scheme, where S represents the (very) slowly hydrolyzed substrate. An equilibrium constant, K i, can be calculated from the pre-steady state rate constants, k -1 /k 1, and yields an estimate of affinity. Irreversible suicide inhibitors can permanently inactivate the β-lactamase through secondary chemical reactions in the enzyme active site. The equation below represents the general mechanism of irreversible inhibitors (I) leading to permanent enzyme inactivation (E-I*): k 1 E + I E:I E-I E-I* k -1 k 2 k 3 Examples of these inactivators are the commercially available class A inhibitors: clavulanic acid, sulbactam, and tazobactam (Figure 1-1, 8-10). As will be described below, these types of inactivators often display additional pathways to inhibition. Irreversible inhibitors can be characterized by first order rate constants for inhibition (k inact, the rate of inactivation achieved with an infinite concentration of inactivator) and K I values (the concentration of inactivator which yields an inactivation rate that is half the value of k inact ) (62, 107). While K I approximates the meaning of K m for enzyme substrates, depending on the individual rate constants comprising the reaction, the K I may or may not equal the equilibrium constant K i (= k -1 /k 1 ) determined under pre-steady state conditions. The 50% inhibitory concentration (IC 50 ) measures the amount of inhibitor required to decrease enzyme activity to 50% of its uninhibited velocity. While an IC 50 can reflect an inhibitor s affinity or partition ratio, these parameters are not always congruent, e.g., an 29

32 inhibitor can have a very poor affinity and acylate the enzyme slowly, but yield a low IC 50 because of very slow deacylation rates. β-lactamase Inhibitors in Clinical Practice Clavulanic Acid, Sulbactam, and Tazobactam Clavulanic acid, the first β-lactamase inhibitor introduced into clinical medicine, was isolated from Streptomyces clavuligerus in the 1970s (409). Clavulanate (the salt form of the acid in solution) showed little antimicrobial activity alone, but when combined with amoxicillin, significantly lowered the amoxicillin MIC against S. aureus, K. pneumoniae, Proteus mirabilis, and E. coli (54). Sulbactam and tazobactam are penicillinate sulfones that were later developed by the pharmaceutical industry as synthetic compounds (144, 148). All three β-lactamase inhibitor compounds share structural similarity with penicillin, are effective against many susceptible organisms expressing class A β-lactamases (including CTX-M and the ESBL derivatives of TEM-1, TEM-2 and SHV-1), and are generally less effective against class B, C, and D β-lactamases (61, 72, 79, 385). The activity of an inhibitor can be evaluated by the turnover number (t n, also equivalent to the partition ratio (k cat /k inact )), defined as the number of inhibitor molecules that are hydrolyzed per unit time before one enzyme molecule is irreversibly inactivated (69). For example, S. aureus PC1 requires one clavulanate molecule to inactivate one β-lactamase enzyme, while TEM-1 needs 160 clavulanate molecules, SHV-1 requires 60, and B. cereus I, more than 16,000 (61, 81, 138, 150, 465). For comparison, sulbactam t n s are 10,000 and 13,000 for TEM-1 and SHV-1, respectively (218, 464). 30

33 The low K I s of the inhibitors for class A β-lactamases (nm to M), ability to occupy the active site longer than β-lactams (rapid acylation and slow deacylation rates), and failure to be hydrolyzed efficiently are integral to their efficacy (191). Clavulanate, sulbactam, and tazobactam differ from β-lactam antibiotics as they possess a leaving group at position C 1 of the five-membered ring (sulbactam and tazobactam have a sulfone, while clavulanate has an enol ether oxygen at this position). The leaving group allows for secondary ring opening and β-lactamase enzyme modification. Compared to clavulanate, the unmodified sulfone in sulbactam is a relatively poor leaving group, a property reflected in the high partition ratios for this inhibitor (e.g., for TEM-1 sulbactam t n = 10,000, clavulanate t n = 160) (217, 218). Tazobactam possesses a triazole group at the C 2 β-methyl position. This modification leads to tazobactam s improved IC 50 values, partition ratios, and lowered MICs for representative class A and C β-lactamases (41, 69, 72). The efficacy of the mechanism-based inhibitors can vary within and between the classes of β-lactamases (Table 1-2). For class A, SHV-1 is more resistant to inactivation by sulbactam than TEM-1, but more susceptible to inactivation by clavulanate (371). Comparative studies of TEM- and SHV-derived enzymes, including ESBLs, found that the IC 50 values for clavulanate were 60- and 580-fold lower than for sulbactam against TEM-1 and SHV-1, respectively (371). In our opinion, the explanations for these differences in inactivation chemistry are likely due to subtle, yet highly important, differences in the enzyme active sites. For example, Thomson and colleagues compared the atomic structure models of TEM-1 and SHV-1 and found that the distance between Val216 and Arg244, residues responsible for positioning of the water molecule important in the inactivation mechanism of clavulanate, was more than 2 Å greater in SHV-1 than 31

34 TEM-1 (465). This increased distance may be too great for coordination of a water molecule, suggesting that the strategic water is positioned elsewhere in SHV-1 and may be recruited into the active site with acylation of the substrate or inhibitor. This variation underscores the notion that mechanism-based inhibitors may undergo different inactivation chemistry even in highly similar enzymes (464, 465). Mechanism of Inhibition Evidence from X-ray crystallography, ultraviolet (UV) difference spectra, isoelectric focusing, mass spectrometry (MS), and Raman microscopy suggests that the inactivators of class A β-lactamases undergo complex reaction schemes with multiple branch points after formation of the acyl-enzyme (52, 90, 91, 190, 352, 445). As represented in the following equation, the acyl-enzyme intermediate can: (i) undergo a reversible change that generates a transiently inhibited enzyme, a tautomer (E-T); (ii) lead to permanent inactivation as a covalent acyl-enzyme species (E-I*); or (iii) regenerate the active enzyme via hydrolysis (E+P): k 4 E-T E-I k -4 k 5 k 3 E+ P E-I* The functional inhibition of the enzyme is determined by the relative rates (k 3, k 4, k -4, and k 5 ) of each of these pathways, and in particular, on the formation of the E-I* species (61). 32

35 Figure 1-5 shows a more detailed mechanism describing clavulanate inhibition of PC-1, TEM-1 or SHV-1 (90, 91, 93, 149, 217, 218). Kinetic and mass spectrometry analysis of inactivation mechanisms, combined with crystallographic studies, suggest that clavulanate, sulbactam, and tazobactam follow similar reaction pathways, beginning with formation of an acyl-enzyme species (55, 93, 350, 351). After acylation, opening of the five-membered ring leads to formation of a transient imine intermediate. This imine species is likely the common intermediate preceding the chemical conversions that lead to transient enzyme inhibition (51, 217, 235). Raman spectroscopy of SHV-inhibitor crystals show the imine species then rearranges to form enamine intermediates (235, 350). This enamine intermediate, either in the trans or cis conformation, represents a second important intermediate in the inactivation mechanism (81, 93, 416). Depending on the properties of the enzyme and inhibitor, the reaction will ultimately proceed to deacylation or irreversible (prolonged) inactivation. In the case of deacylation of the enamine intermediate, the acyl-enzyme undergoes decarboxylation and ester bond hydrolysis, regenerating the active β-lactamase, albeit very slowly (307). The duration of transient inhibition is determined, in part, by the stability of the intermediate species. Preceding acylation of the inhibitor, a persistent noncovalent Michaelis complex may account for inhibition of enzyme activity (234). Following formation of the acyl-enzyme, stabilization of the enamine intermediate is a significant factor for prolonged enzyme inhibition. Crystals of tazobactam in complex with the SHV Glu166Ala variant (PDB 1RCJ) showed that tazobactam formed stochiometric amounts of the trans-enamine (351). The trans-enamine intermediate of tazobactam, as opposed to the trans-enamine intermediates of clavulanate and sulbactam, may be stabilized by intraand inter-molecular interactions. These interactions between the sulfone and triazolyl 33

36 moieties of the tazobactam intermediate and the enzyme active site may explain, in part, tazobactam s potent in vitro and in vivo inhibition of many serine β-lactamases (41, 72, 351, 355). Data from Raman crystallography also supports the hypothesis that, as compared to clavulanate and sulbactam, tazobactam more readily forms the trans-enamine intermediate (235). In this case, Raman spectroscopy allows for the identification of reaction intermediates and calculation of their rates of decay and accumulation by examining single crystals in solution. Studies of SHV-1 in complex with each of the mechanism-based inhibitors revealed that tazobactam forms a predominant population of trans-enamine, as opposed to clavulanate and sulbactam which form a mixture of transenamine and the more chemically labile cis-enamine and imine species (235). Analysis of the inactivation reactions of PC-1, TEM and SHV -lactamases using mass spectrometry has identified a series of intermediates beyond the acyl-enzyme, imine, and enamine (55, 93, 354, 445, 465). The terminal end products of inhibition with these mechanism-based inhibitors suggest that a covalent modification of the active site Ser70 persists through the inhibition process. Mass adducts correspond to different inactivated enzyme species, e.g.: (i) Ser70-Ser130 cross-linked enzyme or propynyl enzyme (Δ + 52 ± 3); (ii) aldehyde (Δ + 70 ± 3); (iii) hydrated aldehyde (Δ + 88 ± 3); (iv) decarboxylated trans-enamine (Δ ± 3) are some of the products (Figure 1-5). β-lactam/β-lactamase Inhibitor Combinations: Clinical Use Generally, the inhibitors do not inactivate PBPs, but notable exceptions include: (i) the intrinsic activities of sulbactam against Bacteroides spp., Acinetobacter spp., and. gonorrhoeae; (ii) clavulanate against Haemophilus influenzae and. gonorrhoeae; and 34

37 (iii) tazobactam inhibition of PBPs in Borrellia burgdorferi (61, 201, 264, 312, 331, 476, 477, 502). As these antibacterial effects are relatively weak, the inhibitors are always combined with β-lactam antibiotics for clinical use. Currently, there are five β-lactam/βlactamase inhibitor formulations available: amoxicillin/clavulanate, ticarcillin/clavulanate, ampicillin/sulbactam, and piperacillin/tazobactam. Cefoperazone/sulbactam is used in several countries, including Japan and India, but is not available in the United States. Inhibitory Activity of Carbapenems The consideration of carbapenems, while not explicitly β-lactamase inhibitors, is essential to a review of β-lactamase inhibition. These thienamycin derivatives are among the last line of defense against many multidrug resistant pathogens. An increasing body of evidence suggests that carbapenems are effective not just for their broad-spectrum antibacterial activity, but also for their multiple β-lactamase inhibitory mechanisms. The insight gleaned from the ability of carbapenems to inhibit serine β-lactamases may serve as a useful starting point for intelligent inhibitor design. And while we await the development and release of novel β-lactamase inhibitors, it behooves us to implement wisely the β-lactams currently in our armamentarium. With this in mind, a discussion of the inhibitory activity of carbapenems follows. Meropenem, imipenem, ertapenem, and doripenem (released on the US and European markets in 2007) are often reserved for the most difficult-to-treat Gram negative infections (Figure 1-1, 4-7) (365). Although resistance to carbapenems is rare in the clinic, several carbapenem-hydrolyzing serine β-lactamases are identified, such as KPC-2, mca-1, SME-1, and IMI-1 (407, 510). These enzymes may be found in organisms that 35

38 also produce additional β-lactamases, conferring broad substrate profiles. For example, E. cloacae isolates can produce IMI-1 or mca-1, which hydrolyze carbapenems, and also TEM- and AmpC-type enzymes (408). In addition to their broad antibacterial spectrum, carbapenems are also effective inhibitors, or slow substrates, of most serine β- lactamases. The stability of thienamycins to β-lactamase hydrolysis was noted in early studies of imipenem, including low k cat values for the class A β-lactamase of B. cereus and a class C P. aeruginosa β-lactamase (182, 248, 317, 414). Charnas and Knowles observed biphasic kinetics when RTEM was acylated by carbapenem compounds; an initial burst was followed by a slower steady-state reaction, strongly suggestive of a branched reaction pathway (92, 139). Further work by Mobashery s group demonstrated that acylation of TEM-1 by imipenem occurs readily, and is followed by Δ 2 - and Δ 1 -tautomerization of the pyrroline double bond in the acyl-enzyme species (Figure 1-6) (456). The rate of deacylation differs for the tautomers (Δ 1 is approximately 10-fold slower than Δ 2 ) and it was initially accepted that molecular rearrangements to the more stable (i.e., more slowly deacylated) species accounted for enzyme inhibition. Site-directed mutagenesis studies of TEM-1 Arg244Ser showed that the guanidinium group on Arg244 provides directly, or indirectly via a water molecule, the necessary proton for tautomerization to the Δ 1 - pyrroline acyl-enzyme (456, 513). Incidentally, this water appears to be the same molecule that serves as the proton donor necessary for clavulanate inhibition in class A enzymes (217). Elucidation of the crystal structures of several β-lactamases in complex with carbapenems has offered additional insight into the versatile inhibition mechanism of these compounds. Maveyraud and colleagues found that in the structure of TEM-1 36

39 complexed with imipenem (PDB 1BT5), the ester carbonyl oxygen of the acyl-enzyme was not located in the oxyanion hole (302). This approximately 180º rotation of the carbonyl was also seen in the crystal structure of acyl-enzyme species of imipenem complexed via Ser64 with the AmpC β-lactamase (PDB 1LL5) (21). In contrast, the SHV-1-meropenem crystal demonstrated two conformations of the intermediate acyl species; one with the meropenem carbonyl oxygen of the acyl-enzyme in the oxyanion hole, and one where it is not (Figure 1-7) (340). Proper position in this oxyanion hole, the binding pocket for the β-lactam carbonyl, is important for both initial enzyme acylation and hydrolysis of the ester bond (325, 478). A conformation which places the β-lactam carbonyl outside of this electrophilic center may be delayed in hydrolysis. Computational models of TEM-1 with imipenem demonstrated that the covalent adduct formed first after acylation did have its carbonyl in the oxyanion hole (302). However, this conformation created a steric clash between the hydroxyethyl group and residues , which forces the rearrangement of the acyl-enzyme rotating the carbonyl outside the oxyanion hole. Thus, while producing a conformational change in the enzyme, the C 6 α- hydroxyethyl substituent also facilitates rearrangement of the carbapenem to a more slowly deacylated form. Subsequent examination of the TEM Asn132Ala variant proved the role of Asn132 by demonstrating that replacement of the residue relieved the steric encumbrance of the TEM-1-imipenem acyl-enzyme and allowed the carbonyl to rotate back into the oxyanion hole (493). Furthermore, the TEM-1-imipenem structure indicated that hydrogen bonds were formed between the hydroxyl group of the C 6 hydroxyethyl substituent on imipenem and both the Glu166-associated hydrolytic water and the side chain oxygen of Asn132 (302). Similarly, the SHV-1-meropenem and class D XA-13- meropenem structures (PDB 2ZD8 and 1H8Y, respectively) revealed that the putative 37

40 deacylation water molecule has an additional hydrogen-bonding interaction with the -H group of meropenem s C 6 hydroxyethyl substituent (340, 378). This interaction weakens the nucleophilicity and/or changes the direction of the lone pair of electrons of the water molecule and results in poor turnover of the carbapenems. As individual hydrogen atoms are difficult to resolve by X-ray crystallography, the structural data of β-lactamases and carbapenems has not been able to discriminate between the Δ 1 - and Δ 2 -pyrroline tautomers. In contrast, Raman spectroscopy allows the monitoring of specific bond character and compound reactivity. Recent studies of Raman difference spectra in SHV-1 reacted with meropenem, ertapenem, and imipenem propose the correlation that the Δ 1 -pyrroline tautomer corresponds to the acyl-enzyme species with the carbonyl outside the oxyanion hole, and the Δ 2 -pyrroline tautomer to the acylenzyme with the carbonyl inside the electrophilic center (233). Further, this work suggests that, in crystal form, the Δ 1 - and Δ 2 - tautomers are present in equal amounts, but Δ 1 -pyrroline predominates in solution. This ratio is consistent with the kinetic data implicating the Δ 1 -tautomer as the primary inhibitory species. The recent structures of XA-1, a class D monomeric β-lactamase, inactivated by doripenem provided further support for the role of carbapenems as inactivators of serine β-lactamases (PDB 3ISG) (430). In this structure by Schneider and colleagues, although the carbonyl is positioned in the oxyanion hole, water molecules are absent. The carbamate of Lys70 forms a hydrogen bond to the C 6 hydroxyethyl group, and Lys212 and Thr213 form a salt bridge and hydrogen bond to doripenem. Furthermore, the bond angles of the acyl-enzyme suggest the Δ 1 -tautomer is formed. In addition to the imipenem-induced conformation change of class A and C β- lactamases, the, albeit slow, turnover of carbapenems leads to a stable enzyme-product 38

41 species that occupies the β-lactamase active site for relatively longer than the enzymeproduct of other β-lactams, such as penicillin (whose dissociation is limited by diffusion for class A enzymes) (181, 301). However, the deacylation rate was still slower than the dissociation rate for imipenem, suggesting that acyl-enzyme rearrangement accounts most significantly for inhibition. otably, the presence of hydrolyzed inhibitor in the active site can contribute to enzyme inhibition. Inhibitor-Resistant Class A β-lactamases Within several years of the introduction of clavulanate combinations for clinical use, resistance to amoxicillin/clavulanate and ticarcillin/clavulanate was observed in isolates of E. coli and K. pneumoniae (290, 291, 425, 501). In 1987, Martinez and colleagues reported resistance rates in E. coli in Madrid hospitals and estimated that of amoxicillinresistant strains, 20-30% were co-amoxicillin/clavulanate resistant (291). Investigations in the late 1980 s demonstrated that this phenotype may result from production of β- lactamases not susceptible to the inhibitors (e.g., AmpCs from Enterobacter spp. or P. aeruginosa or metallo-β-lactamases) or enzyme hyperproduction (80). Hyperproduction of a β-lactamase can be mediated by mutations in the promoter region of the gene and/or high copy numbers of plasmids carrying the bla gene; both scenarios were identified for the TEM-1 enzyme (290, 291, 501, 503, 504). Additional resistance mechanisms were concurrently being examined in research laboratories. In 1989, liphant and Struhl studied clavulanate and sulbactam resistance in TEM-1 by introducing a series of random substitutions into the DA segment which coded for the enzyme s active site (344). Through mutagenesis of a 17-amino acid segment of the TEM-1 active site (Arg61 Cys77) and functional selection assays, they 39

42 found that enzymes with an -Ile, -Leu or -Val substitution at Met69 had increased resistance to ampicillin/clavulanate and ampicillin/sulbactam. This foreshadowed an important amino acid substitution in inhibitor-resistant (IR) TEM. Three years later, Zafaralla and colleagues engineered by site-directed mutagenesis the Arg244Ser variant of TEM-1 to explore the role this key amino acid and its interaction with the penicillin C 3 carboxylate (512). This work advanced the earlier hypothesis by Moews and colleagues that Arg244 played an important role in the recognition of the conserved substrate C 3 /C 4 carboxylate (217, 316). Further studies by Mobashery s laboratory also revealed that Arg244Ser was more resistant to inactivation by clavulanate (217). Interestingly, the first IR class A β-lactamase was isolated from a clinical isolate of E. coli in the same year. The enzyme was determined to be related to TEM-1 or TEM-2, and hence designated inhibitor-resistant TEM (IRT-1) (34, 480). Sequencing information revealed an Arg to Cys or Ser substitution at residue 244 (25, 480). Since this time in the early 1990 s, the number of IRTs has expanded to include more than 35 unique enzymes ( In addition to amoxicillin/clavulanate resistant E. coli, IRTs have been identified in Klebsiella, Proteus, Shigella and Citrobacter spp. (434). IR SHVs have also been isolated from K. pneumoniae, with a present total of 5 IR SHVs described (134). perationally, inhibitor resistance refers to resistance to inactivation by amoxicillin/clavulanate and may, but does not necessarily, include resistance to sulbactam and tazobactam (464). Moreover, this is a relative resistance; provided sufficient clavulanate concentrations are used, these IR enzymes can be inactivated. Currently, reports describing non-tem, non-shv family IR enzymes are becoming more common. The CTX-M ESBLs, which typically hydrolyze penicillins and extended- 40

43 spectrum cephalosporins (preferentially cefotaxime), have not yet shown significant resistance to the β-lactamase inhibitors (37, 80). Present studies indicate that the KPC-2 β-lactamase is not inactivated by clavulanate, sulbactam, and tazobactam, as indicated by high MIC values and turnover numbers of 2500, 1000, and 500, respectively (363). The ability of KPC-2 to avoid inactivation by the currently available inhibitors may represent a novel class of highly IR class A enzymes. The rapid emergence of Enterobacteriaceae harboring these carbapenemases is a significant public health concern (335). Epidemiology and Detection of β-lactamase Inhibitor Resistance Resistance to β-lactam/β-lactamase inhibitor combinations challenges the ability to successfully treat serious urinary tract, respiratory tract, and bloodstream infections (187, 261, 425). The prevalence of organisms expressing inhibitor-resistant enzymes varies throughout the world. In 1993, a French study of 2972 E. coli isolates from urinary tract infections (UTIs) found that 25% and 10% of hospital and community isolates, respectively, achieved amoxicillin/clavulanate MICs > 16/2 μg/ml (193). Characterization of these isolates, including MIC profiles (amoxicillin/clavulanate MIC 90 > 1042 μg/ml and cephalosporin MIC < 32 μg/ml), isoelectric focusing, and DA-DA hybridization, suggested that 27.5% and 45% of hospital and community isolates, respectively, were TEM-1 derived and had substitutions at previously described amino acid positions that confer the IRT phenotype. In this survey, 4.9% of E. coli isolates from UTIs were IRT producing. In 1998, a geriatric department of a French hospital reported an outbreak of amoxicillin/clavulanate resistant isolates which all produced the same IR TEM β- lactamase, TEM-30 (Arg244Ser) (169). Published in 2000, Leflon-Guibout and 41

44 colleagues determined the molecular mechanism of amoxicillin/clavulanate resistance (defined by MICs > 16/2 μg/ml) in E. coli isolates from three French hospitals from 1996 to 1998 (260). The overall resistance rate was 5%, and the majority of these resistant organisms were from patients with respiratory tract infections. Production of IRTs was implicated for resistance in 30-41% of the isolates over the time period, only slightly less frequent than the hyperproduction of chromosomal AmpC enzymes. Two independent Spanish studies reported production of IRTs in 5.4% and 9.5% of amoxicillin/clavulanate resistant E. coli isolates (314, 374). The first IRT identified in the United States was not reported until 2004 in a 14-month survey of E. coli isolates from a northeastern tertiary care clinical microbiology laboratory (237). Kaye and colleagues found that 24% of the 283 isolates classified as ampicillin/sulbactam resistant by disk diffusion and MIC testing were also resistant to amoxicillin/clavulanate by disk diffusion zone diameters 13 mm. f the amoxicillin/clavulanate resistant isolates, 83% were from community-acquired infections. Many of these 69 isolates were recovered from UTIs of outpatients, and two expressed the IR TEM-34 (Met69Val) and one the TEM-122 (Arg275Gln) enzyme. Incidentally, also reported in 2004, was a KPC-2 producing K. pneumoniae isolate from ew York City that contained IRT-2 (50). Isolates producing IRTs are more frequently reported in Europe than in the United States. The explanation for this discrepancy is not clear, and probably involves a combination of environmental, methodological, and clinical practice factors. The phenomenon does not appear to be related to large differences in use of β-lactam/βlactamase inhibitor combinations, as these agents are widely used in both Europe and the United States and likely produce similar selective pressures on bacteria (80, 192). 42

45 The detection of IRTs presents a significant number of operational challenges. Disagreements between the results of MIC testing and disk diffusion assays are often seen, especially among isolates with intermediate resistance phenotypes (80, 237, 345). Currently, an international standard for the amount of β-lactamase inhibitor that should be combined with the partner β-lactam for detection of IR enzymes is not practiced, and different combinations can significantly alter the assigned results (345, 463). For example, the use of the 2:1 ratio for amoxicillin/clavulanate appears to underestimate the presence of IR β-lactamases as compared to employing the fixed concentration of 2 µg/ml clavulanate (463). Further, organisms producing low levels of IR enzymes may be undetectable at the 2:1 ratio (346). As a result of the methodological complications of identifying IR β-lactamase producers, the actual IRT prevalence may be underestimated (80). Definitive identification requires specific enzyme kinetic testing, determination of isoelectric points, and molecular characterization (72, 84, 89). Routine susceptibility tests may not accurately detect strains expressing IRTs. A direct correlation between the level of resistance (MIC) and the predicted clinical outcome is not yet known. Substitutions in Class A Conferring Inhibitor Resistance The amino acid substitutions associated with clavulanate resistance are different than those that produce the ESBL phenotype, and thus IRTs have unique hydrolytic profiles compared to ESBLs. For example, the E. coli isolates that contain amoxicillin/clavulanate IRTs may be more susceptible to cephalosporins, carbapenems, and monobactams (40). However, among TEM and SHV β-lactamases, substitutions that confer an IR phenotype are sometimes found in pairs or are combined with amino acid changes that also increase the ability of the enzyme to hydrolyze oxyimino-cephalosporins. These variants are 43

46 designated complex mutants of TEM (CMTs) and may foreshadow the emergence of a new class of versatile β-lactamases with even broader hydrolytic and resistance profiles (417). Table 1-3 summarizes kinetic properties of representative class A inhibitor-resistant enzymes. IRTs, when derived from mutated wild-type bla genes, typically display a reduction in β-lactam catalytic efficiency (k cat /K m ), mediated by lower k cat and increased K m values for penicillins when compared to the wild-type enzymes (40). IR β-lactamases have increased K I s and IC 50 values for clavulanate and sulbactam; smaller increases are observed for tazobactam. In general, the MICs of organisms expressing IR enzymes are lowered for penicillins and increased for clavulanate and sulbactam. MICs may remain in the susceptible range for tazobactam when obtained with the combination of piperacillin/tazobactam. This susceptibility is likely due to piperacillin s greater potency and rapid cell entry (see above) as well as the relative preservation of tazobactam s K I (41). f significant interest is the recent description of the kinetic properties of SHV-72 (Ile8Phe, Ala146Val, Lys234Arg) (305). In contrast to most other IR enzymes, this unique IR SHV demonstrates increased catalytic efficiency for penicillins. Amino acid substitutions in TEM and SHV result in resistance to clavulanate by two primary mechanisms: (i) alterations in the geometry of shape of the oxyanion hole; and (ii) changes in the position of Ser130 or Arg244 (38, 85, 121, 146, 461, 465, 470). The individual residues that are most commonly substituted in IR TEM enzymes are: Arg244, the nearby Asn276 and Arg275, Met69, and the active site Ser130 (Figure 1-8) ( Arginine

47 To date, 12 of the 36 IRT variants reported have substitutions at residue 244. The guanidinium group (CH ) of Arg244 contributes to substrate affinity through hydrogen bonding to the conserved β-lactam carboxylate (492). Structural and computational studies of the mechanism of clavulanate resistance in TEM-1 reveal that a hydrogen on the η2 atom of the guanidinium group of Arg244 and the carbonyl oxygen atom of the backbone of Val216 coordinate a water molecule (Figure 1-9) (217, 492, 512). When clavulanate is bound, this water molecule is the likely donor of a proton that saturates the double bond at the C 2 position, facilitating the opening of the five-membered ring, formation of an intermediate, and eventual β-lactamase inactivation (217). Loss of the guanidinium group of Arg244 displaces this water molecule and may explain the inhibitor resistance and catalytic impairment that is observed in the TEM Arg244 variants. Without protonation of the double bond, elimination of oxygen at C 1 is thermodynamically unfavored, which may lead to increased inhibitor turnover (217). Interestingly, there are not yet any clinical reports of an SHV enzyme that has the 244 substitution. However, mutagenesis of SHV at Arg244 revealed that the 244Leu and Ser substitutions at this residue also confer resistance to amoxicillin/clavulanate (167, 465). The biochemical correlates of these clavulanate-resistant phenotypes are different for SHV and TEM, and related to slight differences in the inactivation mechanisms for these enzymes (217, 464, 465, 492, 512). Methionine 69 Met69 is the most commonly substituted residue in IR enzyme isolates, appearing either as a single substitution or in combination with additional amino acid replacements in 15 of the 36 presently defined IRTs. Despite the proximity to the enzyme s nucleophilic 45

48 Ser70, the Met69 side chain is oriented away from the catalytic serine and not part of the substrate binding site (228). Residue 69 forms the back wall of the oxyanion hole or electrophilic center that polarizes the β-lactam for nucleophilic attack by Ser70 and in the case of inhibitors, possibly Ser130 as well (246). The residue is not highly conserved in class A β-lactamases, and TEM can tolerate many substitutions at the site and still retain activity against most substrates (8, 121). The SHV Met69Phe, -Lys, and Tyr variants have increased MICs and hydrolytic activity for the extended-spectrum ceftazidime (188). While the residue displays some flexibility for its role in substrate binding, certain substitutions in TEM and SHV confer resistance to the inhibitors. The TEM variants, Met69Ile, -Val, and -Leu, each show an approximately 10-fold increase in K I and 10-fold decrease in k inact, leading to 100-fold decrease in inhibitory efficiency (k inact /K I ) (85). Evidence from X-ray crystallography and molecular dynamic studies has revealed that each substitution introduces subtle active site changes that perturb inhibitor recognition and enzyme catalysis (309, 494). Similarly, SHV Met69Ile, - Val, and Leu have increased MICs and IC 50 s to clavulanate (188). Reduced inhibitor acylation efficiency in SHV was elucidated by observed changes in the oxyanion hole in the crystal structures of SHV Met69Val/Glu166Ala in complex with clavulanate, sulbactam, and tazobactam (PDB 2H0T, 2H0Y, and 2H10, respectively) (470). Asparagine 276 Fourteen TEM variants have been discovered with substitutions at Asn276; seven demonstrate an IR phenotype. Ambler position 276 remains unchanged in currently identified SHV variants. Despite its relative distance (8-10 Å) from the enzyme active site, the crystallographic structure of TEM Asn276Asp (PDB 1CK3) by Swaren and 46

49 colleagues confirmed the importance of the electrostatic environment surrounding residue 276 (449). In TEM, amino acid 276 is located on the enzyme s C-terminal α-helix H11, which lies behind the β-sheet including Arg244. The Asn276Asp substitution creates new interactions with Arg244 and significant displacement of α-helices H1, H10, and H11. The Asp at residue 276 is displaced only 0.75 Å compared to the wild-type Asn, but this movement results in a tighter attraction between Arg244 and 276Asp. This enhanced interaction has two major effects (428, 449). First, there is an effective neutralization of the Arg244 contribution. This decreased positive charge may lower the local affinity for the C 3 carboxylate of clavulanate and penicillins (449). Charge changes in the active site may have a more profound consequence for the inhibitors, which are positioned as small penicillin substrates, and rely heavily on the interactions with Arg244 for binding affinity (217, 428, 512). Second, the TEM Asn276Asp variant s increased clavulanate k cat values may be explained by the displacement of the water molecule necessary for secondary ring-opening of clavulanate, as discussed above for TEM Arg244 variants (428, 449). Similar to Arg244, IR SHV 276 variants have not yet reached clinical attention, and it is unclear whether this reflects an intrinsic difference in enzyme structure and function (465). In Chapter 2, we explore the role of Asn276 in SHV and attempt to explain the residue s relationship to clavulanate resistance and its unique role in this enzyme. Serine 130 Ser130 is a conserved amino acid among all class A β-lactamases, but is substituted for a Gly residue in two IRTs (TEM-59 and -76) and one IR SHV enzyme (SHV-10). The multiple roles of this residue include anchoring β-lactams to and stabilizing the active site 47

50 through hydrogen bonding with the C 3 /C 4 carboxylate of the inhibitors and substrates, and facilitating proton transfer to the β-lactam nitrogen during acylation, leading to β-lactam ring opening (217, 218, 255, 479). Ser130 also serves as a second nucleophile which attacks the imine intermediate and becomes covalently cross-linked to Ser70 in the terminal inactivation of the mechanism-based inhibitors (250). Removing the crosslinking residue seems a clear way to overcome inactivation, but inhibitors, and particularly tazobactam, retain efficacy against the Ser130Gly enzyme. Based on these mechanistic roles, how does the Ser130Gly variant maintain hydrolytic activity and why does it confer resistance to inhibitors (461)? In response to the first part of the question, evidence from X-ray crystallography of SHV and TEM Ser130Gly (PDB 1TDL and 1YT4, respectively) confirmed that a relocated water molecule, initially predicted by Helfand and colleagues, compensates for the loss of the Ser130 hydroxyl group by donating a proton to the β-lactam nitrogen (185, 190, 218, 461, 494). The decreased catalytic efficiencies seen in the TEM and SHV variants are likely a result of perturbations caused by the repositioned water molecule. These perturbations vary somewhat between SHV and TEM Ser130Gly (e.g., reorientation of Ser70 and Lys73 in SHV and TEM Ser130Gly, respectively), but both involve changes in hydrogen bonding networks in the active site, which are modified to accommodate the new water molecule (446, 461). As for the second part of the question concerning inhibitor resistance, the data on Ser130Gly variants argues that cross-linking is not essential for enzyme inactivation. Despite increased MICs to the inhibitors, the turnover numbers for clavulanate and tazobactam by SHV Ser130Gly are equivalent to the wild-type enzyme (185). This paradox of inhibitor resistance illustrates that despite replacement of a key catalytic 48

51 residue, SHV Ser130Gly is still capable of inactivation by the mechanism-based inhibitors (185, 190, 354, 446). The resistant phenotype presents primarily in whole cell assays, where bacteria are exposed to a limited inhibitor concentration. Structural perturbations created by the Ser130Gly substitution discourage formation of the preacylation complex (increased inhibitor K I s), which ultimately leads to fewer inactivation events (190). Thus, while the enzymatic machinery for inhibition is intact, the decreased accumulation of acyl-enzyme intermediates and inactivation products can lead to inhibitor resistance in vivo. Arginine 275 Similar to observations regarding Arg244 and Asn276, the Arg275 variant has only been identified in TEM. Before description of the IR variants TEM-103 and TEM-122, which possess only the Arg275Leu and Gln changes, respectively, substitution at 275 was seen only in combination with the Met69Val and -Leu substitutions (7, 237). Hence, the contribution of Arg275 to the IR phenotype was not appreciated and rigorous kinetic and structural studies of this variant have not yet been completed. However, from crystal structures of the wild-type enzymes and a better understanding of properties at nearby Arg244 and Asn276, Chaibi and colleagues postulated that substitutions at Arg275 (like Asn276) disrupt local electrostatic interactions and displace the water molecule that is key to inactivation by clavulanate (86). Lysine 234 The recent definition of two clinical IR SHV β-lactamases with Lys234Arg substitutions draws attention to this important active site residue (135, 305). Previously, 49

52 TEM or SHV variants obtained from the clinic had not been identified with changes at this highly conserved class A residue. Site-directed mutagenesis of Lys234 to Arg in TEM-1 leads to a 10-fold decreased affinity for penicillins, presumably because of the hydrogen-bonding role of Lys234 to the C 3 carboxylate (Figure 1-3) (262). However, for the clinically isolated SHV-72 enzyme (Ile8Phe, Ala146Val, Lys234Arg), K m values for penicillins were not significantly changed and IC 50 values for clavulanate increased 10- fold compared to SHV-1 (305). The k cat values for the penicillins were increased, up to 5.8-fold for ticarcillin. Molecular dynamics simulations show that Lys234Arg induces movement of Ser130 oxygen away from Ser70. Mendonca and colleagues propose that the mechanism of IR in SHV-72 may be due to movement of Ser130 based on its role in both hydrolysis and terminal inactivation by cross linking with Ser70 (305). The Promise of ovel β-lactamase Inhibitors This section reviews compounds that have demonstrated favorable inactivation properties against β-lactamases, and introduces newer compounds showing promise as second-generation inhibitors. Table 1-4 provides K i (or K I ) and IC 50 values for representative inhibitors against the different -lactamase classes to illustrate the challenge of achieving broad-spectrum inhibition. Monobactam Derivatives Monocyclic, -sulfonated-β-lactams are a family of antibiotic compounds produced by bacteria (216, 453, 454). Structure-activity studies of these monobactams led to the development of the synthetic compound, aztreonam, which interacted with the PBPs of a wide range of aerobic Gram-negative bacteria, including P. aeruginosa, and was 50

53 introduced into clinical practice in 1984 (183, 202, 452). In addition, aztreonam resists hydrolysis by many plasmid-mediated β-lactamases such as TEM-1 and -2, XA-2, nd SHV-1, behaving as a very poor substrate with low affinity, e.g., K m of 2.9 mm for TEM- 2 (452). With class C enzymes, such as E. cloacae P99, aztreonam forms a stable acylenzyme that is very slowly hydrolyzed (low k cat ) (155, 452). Aztreonam also exhibits very low K i values for chromosomally encoded cephalosporinases (e.g., 1.9 nm for P99), and serves as a transient inactivator (66, 155). In vivo, where periplasmic β-lactamase concentrations can be in the low mm range, clinically significant catalytic efficiencies may be obtainable (155). Further modification of the monobactam nucleus was embarked upon to capitalize on the inhibitory activity of aztreonam. The 1,5-dihydroxy-4-pyridone monobactam Syn2190 (Figure 1-1, 11) possesses a novel C 3 side chain that may utilize the iron uptake pathway to enter Gram-negative bacteria. Syn2190 displayed IC 50 values in the nm range for cephalosporinases (e.g., 6 nm for E. cloacae P99), and synergy with ceftazidime and cefpirome in MIC testing of clinical strains, including AmpC derepressed variants of P. aeruginosa (15, 334). In addition, Syn2190, in combination with ceftazidime or cefpirome (at a 1:1 ratio), improved efficacy compared to ceftazidime/tazobactam in murine models of systemic infections with cephalosporin-resistant P. aeruginosa. Syn 2190 has also shown utility for the challenging clinical task of identifying production of plasmid-mediated AmpC β-lactamase in Klebsiella spp. In combination with cefotetan, Syn 2190 achieved 100% specificity and 91% sensitivity for AmpC producers (33). However, Syn2190 has lower affinity than tazobactam for class A enzymes, and could not restore susceptibility to cephalosporins in bacteria expressing these enzymes. This 51

54 limitation against class A enzymes may partly explain the absence of further development of this compound. The design of bridged monobactam derivatives was guided by the crystal structure of the monobactam aztreonam in complex with the class C β-lactamase from Citrobacter freundii (PDB 1FR1) (184, 343). Studies of the structure demonstrated that before deacylation, aztreonam had to rotate around the C 3 -C 4 bond to allow a hydrolytic water access to the ester bond. This rearrangement requires breaking and reforming of the active site hydrogen bond network, and leads to rate-limiting deacylation (245). The introduction of a bridge between C 3 and C 4 limits this rotation and stabilizes the acylenzyme against hydrolysis by AmpC enzymes (117, 205). Heinze-Krauss and colleagues synthesized and evaluated of a panel of bridged monobactams that exhibited very favorable inhibition of the class C C. freundii and P. aeruginosa β-lactamases (IC 50 values as low as 10 nm), but lower affinities for the class A TEM-3 (IC 50 values > 100 μm) (184). At sufficiently high concentrations, class A β- lactamases were acylated rapidly, but the rate of deacylation was higher, leading to high hydrolysis rates and low active site occupancy. Comparison of the crystal structures of class A TEM-1, PC1, and the β-lactamase from Bacillus licheniformis 749/C and the class C β-lactamase from C. freundii revealed that, in class A enzymes, hydrolysis occurs on the opposite side of the bridged monobactam ester by a water molecule activated by hydrogen bond networks not present in class C enzymes (184, 199, 228, 316, 343, 441). Rotation about the C 3 -C 4 bond is less critical to the inhibition mechanism in class A enzymes, and restricting this rotation does less to stabilize the acyl-enzyme. For example, the bridged monobactam Ro (Figure 1-1, 12) (at 4 and 8 μg/ml) effectively potentiated the activity of imipenem against both inducible and derepressed class C- 52

55 expressing P. aeruginosa isolates (277). The activity of piperacillin and ceftazidime were also potentiated by the combination with Ro against the strains expressing elevated quantities of AmpC enzyme. However, Ro did not improve the activity of piperacillin, ceftazidime or carbapenems against organisms expressing class A, B or D β-lactamases. Modification of the monobactam structure at the β-lactam nitrogen has met some success against class A β-lactamases (58, 205). Analysis of a series of -sulfonyloxy-βlactam monobactam derivatives demonstrated rapid acylation of the TEM-1 active site and resistance to deacylation, evident from turnover numbers ranging from 2-8 (58). Additional analysis of one of these derivatives (Figure 1-1, 13) revealed that, following enzyme acylation in the class A carbapenemase mc-a, the tosylate group is eliminated and gives rise to two acyl-enzyme species (322). These inhibited enzyme species are trapped in local energy minima which yield low deacylation rates. Based on molecular simulations, Mourey and colleagues postulated that the acyl-enzyme alternates between two different conformations that move in and out of the oxyanion hole (322). This motion requires breaking and reforming of hydrogen bonds between the acyl-enzyme ester and carbonyl oxygen atoms and the oxyanion hole backbone nitrogens and Ser130 oxygen atom. In both conformations, the water implicated in the deacylation reaction is positioned poorly for hydrolysis, offering a structural explanation for the irreversible inhibition. A series of monobactam derivatives was recently developed to target pathogens harboring multiple β-lactamases. Some of these monobactam analogues act as very effective bactericidal agents for difficult-to-treat Gram-negative pathogens while others are bridged monobactams, such as BAL29880, that are inhibitors of AmpC enzymes 53

56 (117, 122, 361). In Chapter 5, we explore BAL29880 s inhibitory activity against the P. aeruginosa AmpC. Additional derivatives include monobactams bearing a siderophore sidechain which can enhance cell entry through bacterial iron uptake systems (57). f this siderophore type, BAL19764 and BAL30072 (Figure 1-1, 14) have potent antibacterial activity for carbapenem-resistant P. aeruginosa, Acinetobacter spp., S. maltophilia, and S. marcescens as well as against Gram-negative bacilli expressing VIM and IMP - lactamases (45, 122). BAL19764 is stable to hydrolysis by MBLs (358). Further, these siderophore compounds are potent enzyme inhibitors; BAL30072 has low µm IC 50 values for TEM-3 (1.7 µm), and the AmpCs from C. freundii and P. aeruginosa (0.09 and 15 µm, respectively) (163). By combining BAL19764 or BAL30072 with clavulanate and a bridged monobactam (all at a fixed concentration of 4 µg/ml), the MICs of the siderophore compounds was improved for a range of Gram-negative bacteria (including P. aeruginosa) producing AmpCs as well as carbapenemases or over-expression of efflux systems (359, 360). However, the activity of carbapenems is still superior to these new monobactams for many ESBL-producing isolates (e.g., K. pneumoniae expressing CTX- M and SHV-5, E. coli and E. cloacae) (46, 122, 208). The dual antibacterial/inhibitory action of these monobactam derivatives makes them important leads, and the further development of the siderophore compound BAL30072 seems likely. 2-amino-4-thiazolyl methoxyimino Derivatives Certain cephems adopt catalytically incompetent conformations in the active site, effectively trapping the enzyme. Evidence from the crystal structure of an AmpC enzyme complexed with ceftazidime (PDB 1IEL) suggested that the presence of the 2- amino-4-thiazolyl methoxyimino (ATM) R 1 side group, common among extended- 54

57 spectrum cephalosporins (e.g., cefotaxime and ceftazidime), destabilized the formation of the deacylation transition state (398). Synthesis of inhibitors containing this ATM side chain on a penicillin (Figure 1-1, 15) and carbacephem backbone lead to AmpC IC 50 values of 900 nm and 80 nm, respectively (472). Subsequent examination of the crystal structure of the ATM-penicillin in complex with the AmpC (PDB 1LLB) resembled the acyl-enzyme formed by ceftazidime. The ATM groups were located in very similar positions, in both cases leading to steric hindrance between the deacylating water and the β-lactam ring nitrogen, accounting for inhibition. This design highlights the potential of incorporating destabilizing groups onto conserved substrate skeletons that will facilitate recognition by the enzyme (472). Penems BRL and Syn 1012 Methylidene penems are potent inhibitors that have enhanced the inhibitory properties of the penem structure by introducing a double bond at C 6 and a sulfide at penem position 1 (307). These compounds offer advantages over the traditional β-lactamase inhibitors, including a large R 1 side chain that may aid in cell permeability and contribute to active site affinity, and a novel inactivation mechanism. The parent compound from this group, BRL C 6 -( 1 -methyl-1,2,3-triazolylmethylene)penem (Figure 1-1, 16), is an effective inhibitor of class A, C, and D β-lactamases (145, 296). IC 50 values were 10- to 100-fold lower for BRL as compared to clavulanate, tazobactam, and sulbactam for class A TEM-1 and SHV-1, class C P99, class D XA-1, and the S. aureus β- lactamase CTC (104). However, BRL was a substrate for class B metalloβ-lactamases from Aeromonas hydrophila and S. maltophila and BcII (296). In 55

58 susceptibility testing, BRL concentrations of 0.25 μg/ml or lower were able to restore amoxicillin MICs to < 16 μg/ml for TEM-1 and XA-1 producing organisms (104). Further, amoxicillin/brl combinations were more effective than amoxicillin/clavulanate for E. coli, K. pneumoniae, H. influenzae, S. aureus, and. gonorrhoeae producing plasmid-mediated β-lactamases (mostly TEM- and SHVexpressors) and for K. oxytoca, Proteus vulgaris, P. mirabilis producing chromosomal β- lactamases. Kinetic studies of BRL and CTC 11561, TEM-1, and P99 showed rapid, irreversible inactivation with inhibition stochiometery of 1:1, or turnover numbers of 0 (145). Initial studies to elucidate the BRL inactivation product using UV-difference spectroscopy for both enzyme- and base-catalyzed hydrolysis of the methylidene penem suggested the presence of a dihydrothiazepine rearrangement product (53, 296). Accordingly, mass spectrometry studies demonstrated that the methylidene penem inhibition pathway does not involve fragmentation in the active site (145, 296, 455). When TEM-1, P99, and the class A β-lactamase from B. cereus I were inactivated by BRL 42715, spectra showed mass increases equivalent to the molecular masses of the inhibitor plus the enzyme (145). The crystal structure of the class C β-lactamase, E. cloacae 908R, in complex with BRL (PDB 1Y54) confirmed the formation of a stable seven-membered thiazepine ring covalently attached to Ser64 (310). Studies of both BRL and related penems support the mechanism involving formation of an imine intermediate from the acyl-enzyme, which then undergoes rearrangement, and eventually endo-trig cyclization to form the seven-membered thiazepine ring end product of inactivation for class A, C, and D enzymes (Figure 1-10A) (296, 310, 339, 481). 56

59 Crystallization of BLI-489, a related heterocyclic C 6 -substituted penem with IC 50 values of 9.0 and 6.2 nm for SHV-1 and class C GC1, respectively (see below), revealed the common seven-membered thiazepine ring when complexed with the SHV-1 and GC1 enzymes (PDB 1G and 1H, respectively) (339). Interestingly, the orientation of the ring in SHV-1 and GC1 differed by 180º about the bond to the acylated serine. This rotameric preference may be due to the relative openness of the class C binding site at the bottom of the b3 β-sheet. Crystallization of a tricyclic C 6 methylidene derivative in complex with SHV-1 and GC1 (PDB 1Q2P and 1Q2Q, respectively) also showed the common seven-membered thiazepine ring covalently attached to Ser70 and Ser64, respectively (483). The SHV-1-penem complex was similar to what was observed with BLI-489, where the ring had R stereochemistry at the new C 7 moiety (339). However, in the GC1-penem complex, both the R and S chiral forms of the intermediate were found. Taken together, these findings suggest that the C 6 methylidene penems acyl-enzyme stability is likely due to the low occupancy or disorder of the hydrolytic water molecule and, in part, to the conjugation of the ester with the ring system. The nucleophilic attack of the double bond at C 6 appears important to the inhibition mechanism of the methylidene penems, and these novel inhibitors seem to share a similar inactivation mechanism for different β-lactamase classes (483). Most β-lactamases appear to follow this linear pathway to inactivation by the penem compounds, i.e., formation of an acyl-enzyme and rearrangement to the stable dihydrothiazepine ring adduct (145, 296). However, some class A enzymes, such as the β- lactamases from Staphylococcus albus (epidermidis), K. pneumoniae K1, and KPC-2 exhibit more complex kinetics (53, 296, 363). A branched pathway leads to an acyl- 57

60 enzyme hydrolysis product, which also rapidly rearranges to the dihydrothiazepine species. While BRL and the structurally related investigational penem Syn 1012 showed promising in vitro kinetics against a large spectrum of Gram-positive and Gram-negative isolates, they were relatively unstable in human and mouse plasma (386). Further, serum levels of BRL and Syn 1012 were undetectable one hour after intravenous administration in a rabbit model. Thus, the successful elements of BRL have been applied to additional drug-development efforts directed at improving in vivo stability (see below). BRL Derivatives Derivatives of BRL 42715, including tricyclic and bicyclic heterocyclic methylidene penems, have been the focus of a large body of recent structure-activity studies and continue to show promising inhibition properties in investigational studies. The bicyclic heterocycles are synthesized as C 6 methylidene penems attached to thiophene, imidazole, and pyrazole rings (481). Extension of the second heterocyclic ring yields the tricyclic derivatives, which show improved stability in solution and increased lipophilicity for in vivo studies (483). As a group, these compounds have proven to be potent inhibitors of class A, C, and D β-lactamases (30, 189, 380, 481, 497). A panel of one tricyclic and six bicyclic penem compounds displayed IC 50 values of 0.4 to 3.1 nm for TEM-1 (class A), 7.8 to 72 nm for IMI-1 (class A), 1.5 to 4.8 nm for AmpC (class C), and 14 to 260 nm for a CcrA (class B) (497). Compared to tazobactam IC 50 s, these penems were 100- to 56,000-fold more active against TEM-1 and the AmpC enzymes. Each of the seven penems, at a constant concentration of 4 μg/ml, was 58

61 combined with piperacillin, and significantly lowered MICs in ESBL-producers and restored susceptibility in piperacillin-resistant Enterobacter spp., Citrobacter spp., and Serratia spp. organisms. In a mouse model of infection caused by E. coli and Enterobacter aerogenes class A (including ESBL) and C β-lactamase-producing organisms, piperacillin/penem combinations decreased the overall ED 50 of piperacillin from 1.6- to 8-fold (497). However, class B metallo- -lactamase producers were not reported for susceptibility tests or the murine model, and warrant further attention for the clinical translation of these compounds. f the panel of tricyclic and bicyclic penem compounds tested, the bicyclic BLI-489 was chosen as a lead compound to be combined with piperacillin (380). With 4 μg/ml of BLI-489, 92% of piperacillin non-susceptible clinical strains (including expressors of representative β-lactamases from all four classes) had piperacillin-bli-489 MICs 16 μg/ml, in contrast to 66% for piperacillin/tazobactam. Furthermore, the BLI-489 combination demonstrated improved activity over piperacillin/tazobactam against ESBLand AmpC-expressing strains, as well as a panel of staphylococcal and streptococcal isolates. otably, BLI-489 did not improve the activity of piperacillin to the enterococci and penicillin-intermediate and resistant strains of S. pneumoniae (which may reflect that the piperacillin resistance mechanism is not -lactamase mediated) (380). While these practical studies help to bring closer the possible clinical introduction of these novel inhibitors, only a small number of the isolates were KPC- or IRT-producers. These KPCand IRT-producing strains also expressed up to three other β-lactamases and complicate the interpretation of activity against these enzymes. Thus, one cannot yet endorse the efficacy of this inhibitor combination for these relevant resistance threats. 59

62 Two of these bicyclic investigational penems (Figure 1-1, 17 and 18), containing either a dihydropyrazolo[1,5-c][1,3]thiazole or a dihydropyrazolo[5,1-c][1,4]thiazine moiety, are extremely effective inactivators of the class D XA-1 β-lactamase, with nm affinity and low turnover values compared to tazobactam (t n = 0 versus 350, respectively) (30). Additionally, these penems demonstrate efficacy with IR class A β-lactamases. Against the SHV Arg244Ser variant, the penem inhibitors showed significantly lower K I values compared to clavulanate and sulbactam (penems 1 and 2, μm; clavulanate and sulbactam, 63 and 240 μm, respectively) and extremely low turnover numbers (t n = 0 versus 50 and 100 for clavulanate and sulbactam, respectively) (464). xapenems xapenems are derivatives of clavulanic acid, possessing an oxygen atom at penem position 1. Their synthesis was first described in the 1970s, and Cherry and colleagues introduced an oxapenem that had improved in vitro activity over clavulanate against the E. cloacae AmpC and staphylococcal β-lactamases (100). However, the compound was unstable and did not have activity in whole cell assays. In 1993, Pfaendler and colleagues improved the stability of the compounds by adding bulky substituents at the C 2 position (383). The lead compound from this panel of oxapenems, the zwitterionic AM-112 (Figure 1-1, 19), has shown potent β-lactamase inhibitory properties, particularly for class C and D enzymes (226, 227). Against class A enzymes, IC 50 values were 10- to 20-fold higher for AM-112 as compared to clavulanate (227). However, the IC 50 s of AM-112 for the class C enzymes from E. cloacae, S. marcescens, and P. aeruginosa, as well as XA- 1 and XA-5 from E. coli, were to fold lower than clavulanate. AM-112, in combination with ceftazidime at 1:1 or 1:2, reduced ceftazidime MICs against class A 60

63 ESBL producers and the class C enzyme hyperproducers. For the class D enzyme producers and P. aeruginosa strains tested, AM-112 did not enhance the activity of ceftazidime. The susceptibility results were similar for AM-112 s protection of cefepime, ceftriaxone, and cefoperazone (226). In a mouse sepsis model, ceftazidime protection was maintained against E. cloacae P99 and K. pneumoniae SHV-5 (227). The preclinical development of AM-112 has been led by Amura Ltd., and warrants further examination as a potential multi-class β-lactamase inhibitor. Tricyclic Carbapenems (Trinems) Rational structure-based design informed the synthesis of novel tricyclic carbapenems built by the fusion of a cyclohexane ring onto a carbapenem scaffold (105, 106). LK-157 (Figure 1-1, 20) was derived by introduction of a methoxy substituent at position C 4, and shows nm affinity for both the class A TEM-1 and class C E. cloacae 908R β-lactamases (487). IC 50 s were similar between clavulanate and LK-157 for TEM-1 (0.030 μm and μm, respectively), but over 2000-fold better for E. cloacae AmpC (136.2 μm and μm, respectively) (388). In combination with ampicillin, LK-157 (at 30 g/ml) restored susceptibility for the AmpC-overexpressing E. cloacae P99 strain in MIC testing (487). Against a large panel of class A ESBLs (excepting KPC and CTX-M) and class C producing bacteria, LK-157, in combination with cephalosporins, showed improved potency as compared to clavulanate, tazobactam, and sulbactam (368). The lowest MICs (ranging from to 1.6 μg/ml) were achieved for cefepime or cefpirome in combination with LK-157 at 4 μg/ml. Despite this promising data in classes A and C, LK- 157 behaves as a substrate with the carbapenem-hydrolyzing enzymes BcII and mc-a 61

64 (487). Further, while the inhibitor displays nm affinity for and rapidly inactivates the class D XA-10 β-lactamase, the acyl-enzyme is unstable and rapidly hydrolyzed (487). The crystal structure of LK-157 in complex with the E. cloacae P99 (PDB 2Q9), together with spectroscopic data of reaction intermediates, suggests that after deacylation at the active site Ser64, the C 4 methoxy group is eliminated (388, 487). Further, the catalytic water molecule presumed to be responsible for deacylation in class C enzymes is not observed in the AmpC-LK-157 crystal structure. Rotation about the opened β-lactam ring after acylation likely leads to displacement of this water molecule and subsequent enzyme inhibition (388). The C 10 ethyldiene group is probably not bulky enough to induce the conformational change seen with inhibition of class A enzymes by carbapenems, but resonance stabilization of the acyl-enzyme through the acyl carbonyl and ethyldiene group may contribute to stability of the intermediate (302, 388). This series of compounds are worthy of attention for the potential to serve as leads for additional structure-based design of trinem inhibitors, and Lek Pharmaceuticals continues to investigate LK-157. Developed for parental use, the t 1/2 of LK-157 in rats is similar to other -lactams after intravenous administration (402). Additional studies have focused on developing an oral agent with improved permeability over LK-157, and while pro-drug esters have good solubility, the intestinal permeability remains low. 1-β-methylcarbapenems A panel of carbapenem derivatives bearing a methyl group at C 1 and various substituents at C 2 were screened for inhibitory activity against the MBL IMP-1 (329). The study revealed a compound, J-110,441, that had potent inhibitory activity against several MBLs, but also low K i values for class A TEM and class C E. cloacae enzymes (2.54 μm 62

65 and μm, respectively). Susceptibility testing of the E. cloacae AmpC demonstrated an MIC decrease from 64 to 4 μg/ml with imipenem and J-110,411 at 4 μg/ml. Further work with these carbapenem derivatives has focused on their potential as MBL inhibitors. Penicillin and Cephalosporin Sulfone Derivatives Following the success of sulbactam and tazobactam, medicinal chemists have focused much effort on the development of penicillin and cephalosporin sulfones with functionalities that may improve inhibitor efficiency. ne of the first series of investigational sulfones, the C 7 alkylidene cephems, had lead compounds that were potent inhibitors of class C β-lactamases (77). Subsequent reports have explored the roles of C 2 and C 6 penam and C 3 cephem substituents. The findings of each will be discussed in the context of the derivative group. C 2 /C 3 -Substituted Penicillin and Cephalosporin Sulfones β-lactamase inhibitors with C 2 substitutions have shown efficacy against TEM- and SHV-type enzymes, including ESBLs (415, 475). The acrylonitrile derivative Ro (Figure 1-1, 21) achieves an IC 50 for TEM-1 (0.08 μm) that is comparable to clavulanate (0.10 μm) but slightly higher for SHV-1 (0.27 μm) than clavulanate (0.05 μm) (475). Inhibition with Ro was as effective as tazobactam for TEM-1 and SHV-1, with IC 50 values within 0.03 μm for both enzymes. Most class C enzymes tested were inhibited at lower concentrations of Ro than tazobactam (IC 50 range approximately 2- to 30-fold lower) (415, 475). The IC 50 s for X. maltophilia and Bacteroides fragilis class B enzymes were 24 and 200 M, in comparison to 4 and > 10 mm for tazobactam (415). Ro (at 4 µg/ml) performed well in susceptibility 63

66 testing, protecting the activity of ceftriaxone and ceftazidime against class C- and ESBLproducing organisms (475). Central to the potency of this sulfone inhibitor was the slow enzyme recovery rate as compared to tazobactam, especially for class C β-lactamases from C. freundii and E. coli (415). UV absorption spectroscopy and X-ray crystallography revealed the identity of a linear enamine formed by elimination of the sulfate group, ringopening, and rearrangement into a linear conjugated system (415). This particularly stable intermediate presumably accounted for enhanced inhibitor potency. The design of the C 2 -substituted penicillin sulfone SA2-13 (Figure 1-1, 22) drew on the insight that the tazobactam imine is the common intermediate, or gate-keeper to both permanent interaction and enzyme regeneration (69, 352, 507). If the more stable enamine, the trans-enamine, can be favored, then the reaction will be trapped in transient inhibition. In fact, IR enzymes do exhibit decreased trans-enamine formation, as in the case of SHV Met69Val/Glu166Ala variant reacted with tazobactam and clavulanate (470). SA2-13 includes a carboxyl group attached to a linker which is designed to stabilize the trans-enamine conformation in close proximity to conserved active site residues. Interestingly, the crystal structure of SA2-13 in complex with SHV-1 (PDB 2H5S) showed an additional salt bridge with Lys234 and hydrogen bonds with Ser130 and Thr235. As compared to SHV-1 inhibited by tazobactam, SA2-13 had 10-fold slower deacylation rate, presumably mediated by the stabilization of the trans-enamine species (352). However, the dissociation rate for SA2-13 was 17-fold higher than tazobactam (0.100 versus 1.70 μm, respectively) and there was 47% 24-hour recovery of activity of the SA2-13/SHV-1 mixture versus complete inactivation with tazobactam. The 6-alkylidene-2 -substituted penicillin sulfone, L (Figure 1-1, 23), attains nm K I s for the class A SHV-1 and ESBL SHV-2, and class D oxacillinase-, ESBL-, and 64

67 even carbapenemase-type XA enzymes (75, 131, 367). Moreover, L restores the activity of piperacillin against E. coli IR SHV enzyme producers and ceftazidime and cefpirome against E. coli and K. pneumoniae expressing CTX-M, XA, and CMY β- lactamases (75, 367). Meropenem MICs were lowered from 32 to 1 μg/ml with 4 μg/ml of L against S. marcescens possessing SME-1. The C 2 hydrophobic catechol moiety of this investigational compound appears to improve both its entry into cells via siderophore iron channels and affinity for β-lactamase active sites (87). X-ray crystallography of L in complex with SHV-1 (PDB 3D4F) revealed that the inhibitor s C 6 (heteroaryl)alkylidene group plays an important role in the formation of a planar bicyclic aromatic intermediate (367). The pyridyl nitrogen of the C 6 substituent acts as a base and promotes intramolecular capture, leading to the formation of a bicyclic species (Figure 1-10B) (367). The acyl-enzyme ester carbonyl of the intermediate is resonance stabilized by the conjugated π system. Additionally, the carbonyl group of the intermediate is positioned outside the oxyanion hole. The decreased deacylation rates of this species are likely due to resonance stabilization and the location of the carbonyl, at an increased distance from the hydrolytic water and improperly positioned for nucleophilic attack by the enzyme s backbone nitrogens (367). By expanding the 5-membered penicillin sulfone design to a 6-membered cephalosporin sulfone, Buynak and colleagues developed synthetic methodologies allowing modification of C 3 substituents (74, 76). The different functional groups introduced at C 3 had significant impact on whether the inhibitor was selective for either class A or class C β-lactamases, with some features such as electronegativity increasing inhibition of both classes. DVR-II-41S (Figure 1-1, 24) has both the C 7 (heteroaryl)alkylidene group of L and a C 3 vinyl substituent similar to nitrocefin, an excellent substrate for nearly all 65

68 classes of β-lactamases (74, 76). This cephalosporin sulfone demonstrated IC 50 values of 90 nm for class A TEM-1 and 10 nm for class C GC1. The crystal structure of GC1 in complex with DVR-II-41S (PDB 1GA0) revealed a bicyclic aromatic system reminiscent of the L intermediate (109). Additional stability results from the placement of the inhibitor s anionic sulfinate group between Tyr105 and the acyl-enzyme bond, likely blocking the approach of the probable deacylating water molecule. C 6 -Substituted Penicillin Sulfones C 6 -substituted penicillinates are nm to low μm inhibitors of serine β-lactamases, with the sulfone oxidation state at the penam thiazolidine sulfur showing improved affinities over the sulfide state (32, 73, 422, 423). The stereochemistry and specific functionality of the C 6 substituents can affect the specificity of inhibition. For example, β isomer alkenyl derivatives, as opposed to α isomer derivatives, show improved IC 50 values for class B β- lactamases, and derivatives with thiazole rings, particularly with fluorine, amino, or hydroxyl groups, show potency for class A β-lactamases (422). Further, compounds with a mercaptomethyl group at C 6, as opposed to hydroxymethyl, are better inhibitors of class B enzymes, while C 6 mercapto-penicillinates are relatively less active inhibitors of classes A, B, and C (73). The C 6 -hydroxyakylpenicillinates have been employed as tools for studying the mechanism of β-lactamase hydrolysis and inhibition (173). The crystal structure of 6-αhydroxymethylpenicillanate complexed to TEM-1 (PDB 1TEM) suggests that while C 6 - substituted penicillin sulfones are acylated in the same way as β-lactam substrates, the hydroxymethyl group at C 6 impedes the approach of the hydrolytic water and prolongs the lifetime of the acyl-enzyme species, effectively inhibiting the enzyme (301). However, 66

69 the same compound acted as a substrate for the class A mc-a. Crystal structure determination of mc-a complexed with 6-α-[(1-hydroxy-1-methyl)ethyl] penicillanate (PDB 1BUL) showed that the repositioning of residue Asn132 in mc-a, as compared to TEM-1, enlarged the substrate binding cavity and allowed Asn132 to make new hydrogen bonds with the C 6 substituent (323). These interactions permit approach by the hydrolytic water and facilitate deacylation of the acyl-enzyme intermediate for the 6-αhydroxymethylpenicillanate and C 6 -substituted carbapenems like imipenem in mc-a. Upon introduction of a bulkier C 6 hydroxypropyl group, deacylation was impeded, in a mechanism very similar to the inhibition of TEM-1 by imipenem (456). Generalization of these, and other, findings lead to the conclusion that when the hydrolytic water molecule approaches the acyl-enzyme from the α direction, as with class A enzymes, α-hydroxyalkylpenicillinates are effective inhibitors. Conversely, β- hydroxyalkyl penicillinates inhibit the β approach in class C enzymes (173). A group at Wyeth Research found that a 6-β-hydroxymethyl penicillin sulfone has the best activity against both class A and C enzymes; IC 50 s of 8nM for TEM-1 and 1.2 µm for AmpC (31). Presumably, the inhibitor worked like tazobactam or sulbactam in class A enzymes, but the specific crowding of the β-face of the ester prevents approach of the hydrolytic water in class C β-lactamases (60). The class D XA-10 enzyme showed inhibition by both C 6 α- and β-hydroxyisopropylpenicillinates (298). While only the β-isomer could be crystallized with XA-10 (PDB 1K54), this structure and computational models of the α- isomer suggest related, but different, mechanisms of inactivation based on the stereochemistry at C 6. In the acyl-enzyme, the α- and β-substituents appear to interfere with the path of approach and position of the hydrolytic water necessary for deacylation, respectively. 67

70 Based on this previous mechanistic data, we hypothesize in Chapter 5 that a C 6 - hydroxymethyl sulbactam derivative will inhibit the P. aeruginosa AmpC. As will be discussed, these sulfone derivatives are promising compounds with potentially novel inactivation mechanisms in class C enzymes, but entry into the Pseudomonal cell continues to be a great challenge. on-β-lactam Inhibitors Boronic Acid Derivatives In the 1970 s, boronic acid compounds were described as forming reversible, dative covalent bonds with serine proteases, and inhibiting these enzymes by assuming a tetrahedral reaction intermediates (35, 272). Kiener and Waley then applied the chemistry to the class A β-lactamase from B. cereus and showed that boronic acid derivatives, as well as the phenyl- and m-aminophenyl derivatives, were mm inhibitors of the enzyme (240). Further study by other groups demonstrated that these compounds, which lack the β-lactam motif, form adducts that resemble the geometry of the tetrahedral transition state of the β-lactamase hydrolytic reaction (23, 94, 110). By modifying these transition state analogs to contain the R 1 side chains of natural substrates, affinities in the nm range for both class A TEM-1 and ESBL-type SHV and class C enzymes have been reported (82, 441, 443, 466). Weston and colleagues used the crystallographic structure of E. coli AmpC β-lactamase in complex with m-aminophenylboronic acid (PDB 3BLS) as a model for designing boronic-acid based inhibitors specifically for the class C active site (478, 499). The compound showing the greatest affinity was benzo[b]thiophene-2-boronic acid which had a K i for the E. coli AmpC of 27 nm. 68

71 In addition to demonstrating their potential as effective inhibitors, these compounds have been used to probe the important interactions between the inhibitor and the enzyme active site, thereby informing the design of future boronates. Using the individual crystal structures for nine boronates and four β-lactams complexed with the E. coli AmpC β- lactamase, computational analyses have generated consensus binding pockets in the AmpC active site (400). The functionalities recognized by certain residues led to the development of new glycylboronic acid boronates, including a compound containing the R 1 side chain of cephalothin. ne of these cephalothin analogs achieved a K i of 1 nm for the E. coli AmpC (Figure 1-1, 25), a 300-fold improvement over the parent compound which lacked the meta-carboxylphenyl substituent (320). The substituent was designed to resemble both the dihydrothiazine ring and the C 4 carboxylate of the cephalosporin nucleus. In the crystal structure of this cephalothin meta-carboxyphenyl analog complexed with an AmpC (PDB 1MY8), the inhibitor carboxylate interacted with Asn289, a residue not previously implicated in structural or modeling studies as interacting with the cephalosporin s C 4 carboxylate. Asn289 is not a highly conserved residue among class C β-lactamases, but the low K i is maintained when testing different AmpCs with other residues at 289, e.g., 29 nm for E. cloacae P99 (281, 419). High affinity inhibitors will only be clinical useful if they can effectively permeate the bacterial cell membrane and restore susceptibility to partner -lactams. Boronic acid derivatives enter both S. aureus producing class A -lactamases and Enterobacteriaceae such as E. coli, C. freundii, E. cloacae, and P. aeruginosa expressing class C - lactamases, and effectively protect ampicillin and ceftazidime (82, 320, 397, 484, 492). Boronates have not yet been developed for clinical use in combination with a β-lactam, in 69

72 part because of concerns about the toxicity of boron. The safety of boron-containing compounds is currently being addressed by clinical studies sponsored by Anacor Pharmaceuticals. Anacor develops boron compounds for medical uses and has current trials for boron-based topical treatments of fungal, bacterial and anti-inflammatory diseases, as well as research stage data on a systemic antibacterial agent. In Chapters 3 and 4, we develop the notion that the versatility of the cephalothin analog may reflect an important flexibility in AmpC active sites which could impact the recognition of both substrates and inhibitors. We tested a panel of boronate derivatives against both the A. baumannii and P. aeruginosa AmpCs, and use the selective design of the R 1 and R 2 side chains to draw general conclusion about the active site architecture of these diverse enzymes. Phosphonates Phosphonate monoester derivatives can acylate the active site serine of class A, C, and D -lactamases leading to effective inhibition (4, 266, 286, 287, 401). These compounds exhibit branched kinetic pathways, in some cases reflecting inhibition by the products that are formed. Acyl phosphonates inhibit class C enzymes while cyclic acyl phosphonates inhibit both class A and C -lactamases. For diacyl phosphonates, hydrophobic substituents decrease the inhibitor s K i value; and the acylation rates of these compounds follow the order: class D > class C > class A (286, 287). Majumdar and colleagues posit that this relative inhibitor efficiency reflects the general hydrophobicity of the enzyme active sites, and that rational design of diacyl phosphate side chains could lead to nm inhibition of all serine β-lactamase classes. 70

73 Crystal structures with these phosphonic acids have also offered important insight into the reaction mechanisms of class C enzymes (341). The ability of the E. cloacae GC1 - lactamase to hydrolyze extended-spectrum cephalosporins was elucidated by trapping of the enzyme in complex with a phosphonate cefotaxime transition state analog (PDB 1RGZ) (341). A three residue insertion into the Ω-loop of the enzyme, as compared to the C. freundii -lactamase-transition state analog complex, allowed for a better fit of cefotaxime s side chain in the enzyme active site, and facilitating the activation of the hydrolytic water molecule. The clinical potential of phosphonates has been limited by their poor stability in aqueous solution and susceptibility to phosphodiesterases. XL104 XL104 (also known as AVE1330A, Figure 1-1, 26) is a bridged diazabicyclo[3.2.1]octanone non-β-lactam inhibitor with very promising activity against class A, C, and D β-lactamases (36, 141, 279, 438). The majority of the published work on XL104 describes MIC testing, although IC 50 values of 8, 38, and 80 nm have been reported for class A TEM-1 and KPC-2, and class C P99 enzymes, respectively (36, 438). A remarkable property of XL104 is the prolonged deacylation rate. The 50% enzyme recovery time point was seven days for both TEM-1 and P99 (as compared to seven minutes for TEM-1 and clavulanate and 290 minutes for P99 and tazobactam), suggesting the presence of a very stable and long-lived intermediate species (36). The formation of a covalent complex has been supported by an ESI-MS study of TEM-1 and XL104 (439). Chapter 5 will contribute to this growing collection of data on XL104 and examine its kinetic profile with the P. aeruginosa AmpC. 71

74 Currently, the only reported crystal structure of the inhibitor is in complex with CTX- M-15, and shows opening of the XL104 ring upon acylation, interaction with several conserved active site residues, and displacement of the putative deacylation water by approximately 1 Å (125). With ceftazidime as a partner β-lactam in a 4:1 ratio (i.e., ceftazidime 4 µg/ml XL104 1 µg/ml), XL104 lowered the MICs for a series of Enterobacteriaceae strains at least 8-fold as compared to ceftazidime alone, to 4 μg/ml for E. coli, K. pneumoniae, Citrobacter diversus, and P. mirabilis strains and to 2 μg/ml for E. cloacae, and Serratia spp. (36). Against class A β-lactamase producers (primarily TEM- and SHV-types), the ceftazidime/xl104 combination was comparable to ceftazidime/clavulanate, but superior to amoxicillin/clavulanate and piperacillin/tazobactam. Ceftazidime/XL104 was more potent compared to piperacillin/tazobactam for class C enzyme producers such as P. mirabilis, and both ceftazidime-resistant and -susceptible strains of C. freundii and E. cloacae. The XL104 concentration of 4 μg/ml restored MICs into the susceptible range for a collection of clinical P. aeruginosa isolates (21% non-susceptible to ceftazidime versus 6% non-susceptible to ceftazidime/xl104) (431). In murine studies, including models of sepsis with CTX-M Enterobacteriaceae producers, pneumonia with AmpC and SHV-11 K. pneumoniae producers, and thigh infection and sepsis with KPC- 2, TEM-1 and SHV-11 K. pneumoniae producers, the ceftazidime/xl104 combination has demonstrated promising therapeutic efficacy (263, 306, 498). XL104 also enhanced the activity of cefotaxime in a study of Enterobacteriaceae expressing CTX-M, KPC, or XA-48, where 4 μg/ml of the inhibitor lowered MICs to 1 μg/ml. For isolates with ertapenem resistance due to combinations of β-lactamases (ESBLs or AmpCs) and impermeability, the MICs were still 2 μg/ml. While 72

75 cefotaxime/xl104 lowered the MIC of one S. marcescens isolate expressing SME-1 from 0.5 to 0.12 μg/ml, MICs for K. pneumoniae and E. coli expressing class B IMP and VIM metallo-β-lactamases were not lowered by either cephalosporin with XL104. Two recent studies examined more closely the ability of XL104 to restore susceptibility to β-lactams for KPC-producing clinical isolates (141, 438). Stachyra and colleagues demonstrated that XL104 at 4 μg/ml restored susceptibility of six KPC-2 or - 3 producers to piperacillin, ceftazidime, ceftriaxone, and imipenem (438). Endimiani and colleagues tested XL104 at varying concentrations (1, 2, and 4 μg/ml) against a panel of 42 K. pneumoniae isolates with 3 bla genes each and MICs μg/ml for all β- lactams tested. All of these K. pneumoniae strains were susceptible to XL104 at 2 μg/ml with cefotaxime, ceftazidime, cefepime, and aztreonam, and at 4 μg/ml with piperacillin. Collectively, this data suggests that XL104, in combination with several cephalosporins, aztreonam, and even imipenem, may be an excellent candidate for restoring susceptibility against difficult-to-treat emerging carbapenemase- and ESBL-producing organisms. XL104, manufactured by ovexel Inc., is currently in phase II clinical trials in combination with ceftazidime for treatment of complicated urinary tract and intraabdominal infections ( Additionally, Forest Laboratories Inc. intends to begin phase I clinical trials with XL104 in combination with the novel broadspectrum cephalosporin ceftaroline (which is currently in phase III trials as monotherapy for complicated skin infections and community-acquired pneumonia). Results of clinical trials are awaited. Hydroxamates 73

76 -aryloxycarbonyl hydroxamate is an irreversible inhibitor of the class C E. cloacae P99 β-lactamase that utilizes a mechanism of action which differs from the currently available inhibitors (Figure 1-1, 27) (373, 505). The proposed inhibition mechanism involves rate-limiting acylation of the hydroxamate. However, the inhibitor is not stabilized by the traditional oxyanion hole comprised of the backbone nitrogens of Ser64 and Ser318, but rather by the side chains of Tyr150 and Lys 315 (373). As with clavulanate and tazobactam, the acyl-enzyme proceeds to either hydrolysis or inactivation. The inactivation mechanism leads to aminolysis of Lys315 and a covalent cross-linking of Ser64 and Lys315, as seen in the crystal structure of the inhibited P99 (PDB 2P9V) (Figure 1-11) (505). TEM-2 and XA-1 β-lactamases were also inhibited by -aryloxycarbonyl hydroxamates, and the MBL GIM-1 by a reverse hydroxamate (hydroxylamino replacing hydroxylamine) conjugated to a cephalosporin nucleus (159, 373). Additional mechanistic studies of hydroxamates and their derivatives, as well as in vitro activity data, will be of much interest. ther non-β-lactam Inhibitors Inhibitors with a sulfonamide core and varied substituents were developed by structurebased optimization (399, 469). This non-β-lactam design approach is intended to avoid both hydrolysis by β-lactamases and the up-regulation of β-lactamases induced by some β-lactam-based inhibitors. Starting from a map of binding hot spots developed from crystal structures of 13 different ligands in complex with the E. coli AmpC β-lactamase, 200,000 commercially available small molecules were docked into the active site of the enzyme by computer simulation (399, 400). Based on these results, lead compounds were screened in vitro and the molecule with the highest affinity (26 μm) was crystallized in 74

77 complex with the AmpC (PDB 1L2S) (Figure 1-1, 28). The structure closely resembled the prediction of the docking simulation, revealing the plasticity of the enzyme active site to recognize functional groups different from β-lactams. This catalyzed the optimization of a novel inhibitor, and replacement of the chloride substituent with a carboxylate yielded a compound with 26-fold improved K i (1 μm) (469). In MIC testing, the inhibitor restored the susceptibility to ceftazidime (at a 1:1 ratio) in the AmpC producers E. cloacae and C. freundii (but not P. aeruginosa), and, in contrast to clavulanate, did not up-regulate β-lactamase expression in an inducible strain of E. cloacae. β-sultams are the sulfonyl analogs of β-lactams, and -acyl β-sultams (Figure 1-1, 29) inactivate the P99 enzyme (362, 473). ESI-MS studies suggest that the β-sultams sulfonylate the serine residue to form a sulfonate ester. Elimination of the sulfonate anion leads to C- bond fission and formation of a dehydroalanine at the previous Ser64 residue (473). These compounds do not show inhibition of class B MBLs, and there are currently no studies examining their activities in class A or D enzymes. BLIP is a 165 amino acid protein isolated from the same organism that produces clavulanic acid, Streptomyces clavuligerus (127). Studies of BLIP s inhibitory activity have revealed affinities for class A enzymes that vary from pm to M (the enzymes with the highest affinities for BLIP are KPC-2 and the P. vulgaris β-lactamases) (179, 442). Many of the individual class A residues which contribute to BLIP binding have been identified through crystallography and mutagenesis (411, 511, 515). While the clinical viability of a peptide inhibitor based on this protein is limited because of protein degradation in vivo, the high affinity interactions offer direction for the design of novel - lactamase inhibitors. 75

78 The mixture of vanadate and catechol compounds in aqueous solution produces vanadate-catechol complexes which can inhibit class A TEM-2, class C P99, and class D XA-1 β-lactamases with apparent K i s of 62, 0.58, and 1.5 µm, respectively (5). A crystal structure with the complex has not been attained, but molecular modeling suggests that the active site serine is bound to a penta- or hexa-coordinated vanadium. Vandatecatechols represent another group of compounds that are useful as tools for novel inhibitory active site interactions, but lack clear clinical potential. Inhibition of Metallo-β-Lactamases We note that designing and/or discovering inhibitors to MBLs presents special challenges. The diversity of class B enzyme subclasses (i.e., B1, B2, and B3) and the mechanistic complexities regarding the role of one or two Zn 2+ ions have impeded effective inhibitor design. In addition, indiscriminately inhibiting or chelating the metal ion can have untoward effects on natural or endogenous metalloenzymes. For example, similarities between the active site architecture of MBLs and essential mammalian enzymes (e.g., human glyoxalase II) complicate the design an inhibitor that would be safe for human use (112). As a class, the MBLs are resistant to all the mechanism-based inhibitors of the serine enzymes (the inhibitory activity of the methylidene penems against class B β-lactamases may be an important exception) and few investigational inhibitors demonstrate sub-μm efficacy (380, 497). The following data chapters do not address the inhibition of metallo-β-lactamases, and thus the development of inhibitors for this special class of enzymes will not be further covered here. The reader is referred to reference (132) for a review of metallo-β-lactamase inhibitors in development. 76

79 Table 1-1. Comparison of Ambler and Bush-Jacoby-Medeiros -Lactamase classification schemes (8, 68) Ambler Bush- Jacoby- Medeiros Preferred substrates Inhibited by clavulanate Representative enzymes A serine penicillinases 2a Penicillins + PC1 from S. aureus 2b Penicillins, narrow-spectrum cephalosporins + TEM-1, TEM-2, SHV-1 2be Penicillins, narrow-spectrum and extendedspectrum cephalosporins + SHV-2 to SHV-6, TEM-3 to TEM-26, CTX-Ms 2br Penicillins - TEM-30, SHV-72 2c 2e 2f Penicillins, carbenicillin Extend-spectrum cephalosporins Penicillins, cephalosporins, carbapenems + PSE-1 + FEC-1, CphA ± KPC-2, SME-1, MC-A B metallo-βlactamases 3 Most β-lactams, including carbapenems - B1: IMP-1, VIM-1, Ccr A, BcII; B2: CphA; B3: L1 C cephalosporinases 1 Cephalosporins - AmpC, CMY-2, ACT-1 D - oxacillinases 2d Penicillins, cloxacillin ± XA-1, XA-10 ot classified 4 77

80 Table 1-2. Kinetic properties of representative -lactamases a -lactamase, Ambler class K I (μm) Clavulanate Sulbactam Tazobactam IC 50 (μm) t n K I (μm) IC 50 (μm) t n K I (μm) IC 50 (μm) TEM-1, A SHV-1, A t n Ref. (85, 218, 330) (69, 185, 464, 465) SHV-5, A (166) PC-1, A CTX-M-2, A (54, 69, 81) (19) CcrA, B >500 > > (69) P99, C >100 > (69) CMY-2, C (142) XA-1, D XA-2, D a The definition of K I may not be uniform from laboratory to laboratory. The reader is referred to individual citations for the methods used in each instance. (234, 371) (252, 371) 78

81 Table 1-3. Kinetic properties of representative class A inhibitor-resistant enzymes a - lactamase K I (μm) Clavulanate Sulbactam Tazobactam k inact (sec -1 ) t n K I (μm) k inact (sec -1 ) t n K I (μm) k inact (sec -1 ) TEM > t n Ref. (85, 218) TEM Met69Leu TEM Ser130Gly (85) (461) TEM Arg244Ser (217, 218) TEM Asn276Asp (428) TEM Met69Leu, Asn276Asp (435) SHV SHV Arg244Ser (185, 464, 465) (464, 465) SHV Ser130Gly (185) HI (270) HI Met69Ile (270) a The definition of K I may not be uniform from laboratory to laboratory. The reader is referred to individual citations for the methods used in each instance. 79

82 Table 1-4. Kinetic properties of select inhibitors against different -lactamase Ambler classes a Inhibitor, Compound number in Figure 1-1 Kinetic parameter (in μm) Ambler class A B C D Aztreonam, 3 K m or K i 200 > Ref. (66, 257, 381) BAL30072 siderophore monobactam, 14 ATM-carbacephem, 15 BRL methylidene penem, 16 BLI-489 methylidene penem ovel tri and bicyclic methylidene penems, including 17 and 18 LK-157 tricyclic carbapenems, 20 IC (163) IC (472) IC (104) IC (339) IC (30, 497) IC (388) Ro penam sulfone, 21 IC (415, 475) L penam sulfone, 23 Boronic acid metacarboxyphenyl cephalosporin analog, 25 K I or IC IC (75, 131, 367) (320, 466) XL104, 26 IC (36) a The definition of K I may not be uniform from laboratory to laboratory. The reader is referred to individual citations for the methods used in each instance. 80

83 Figure 1-1. Chemical structures of: (1) a penicillin; (2) a third-generation cephalosporin; (3) a monobactam; (4-7) carbapenems; and (8-10) β-lactamase inhibitors in clinical practice. The numbering scheme for penicillins, cephalosporins, and monobactams is shown. Chemical structures of β-lactamase inhibitors in development: (11-14) monobactam derivatives; (15) a penicillin derivative; (16-20) penems; (21-23) penicillin sulfones; (24) a cephalosporin sulfone; (25) a boronic acid cephalothin analog; (26-29) non-β lactams. H S CH S H H S S Ac S 3H H 2 CH H 1 Benzylpenicillin 2 Cefotaxime 3 Aztreonam H 2 H H H S CH H H H H H S CH H H H H H S H CH H H H H H H S CH H H S H 2 H 4 Meropenem 5 Imipenem 6 Ertapenem 7 Doripenem H CH H H S CH H S CH 8 Clavulanic Acid 9 Sulbactam 10 Tazobactam 81

84 H H S H H CH 3 S 3 H H H H H S 3 H CH 3 S CH 3 H 2 S H H S 3 H 11 Syn Ro Monobactam derivative 14 BAL30072 S S H 2 S H S CH S CH S CH S CH 15 ATM-penicillin 16 BRL Penem 1 18 Penem 2 H H C - + H 3 H CH 3 H H S CH C H S CH H 19 AM LK Ro SA2-13 S CH H H S CH CH 2 H S B H H H 23 L DVR-II-41S 25 meta-carboxyphenyl chiral cephalothin boronate Cl H 2 S 3 H H S H S 2 CH S. 26 XL aryloxycarbonyl 28 Sulfonamide hydroxamate 29 -benzoyl β-sultam 82

85 Figure 1-2. Family portrait of β-lactamase enzymes: (A) Class A, SHV-1; (B) Class B, IMP-1; (C) Class C, E. coli AmpC; and (D) Class D, XA-1. For classes A, C, and D, the active site serine is shown in yellow; for class B, the two Zn 2+ ions are shown. Based on PDB entries 1SHV, 1DDK, 2BLS, and 1M6K, respectively. Class A SHV-1 β-lactamase Class B IMP-1 β-lactamase Class C E.coli AmpC β-lactamase Class D XA-1 β-lactamase 83

86 Figure 1-3. Proposed reaction mechanism for a penicillin β-lactam substrate by a class A serine β-lactamase enzyme in which Glu166 participates in activating a water molecule for both acylation and deacylation (additional evidence exists to suggest that this acylation scheme may exist in competition with a mechanism where Lys73 acts as the general base to activate Ser70 (199, 308, 450)). Dashed lines represent hydrogen bonds, and the number labels are included to help the reader appreciate the general order of events, not as discrete steps. Following activation of the hydroxyl group, Ser70 performs a nucleophilic attack on the carbonyl group of the β-lactam antibiotic, resulting in a highenergy acylation intermediate. Protonation of the β-lactam nitrogen leads to cleavage of the C- bond and formation of the covalent acyl-enzyme which adopts a lower energy state. Attack by a catalytic water leads to a high-energy deacylation intermediate, with subsequent hydrolysis of the bond between the β-lactam carbonyl and the oxygen of Ser70. Deacylation regenerates the active enzyme and releases the inactive β-lactam. Lys 73 Lys 73 Lys 73 Ser H 3 H Glu H 166 R 1 S H 4 2 C H 3 H Xaa 237 H H 3 Ser 130 H 2 H 4 H 3 H H Glu R 1 S H 1 C H Xaa 237 H H 3 1 Ser 130 H 3 1 H Glu 166 H 2 3 R 1 S H 4 H C H Xaa 237 H H 3 Ser 70 Lys 234 Ser 70 Lys 234 Ser 70 Lys 234 Lys 73 Ser 130 H 3 H H R 1 Glu 166 S 3 H 2 1 H C H H 3 H Xaa 237 Ser 70 Lys Lys 73 Ser 130 H 3 H R 1 Glu 166 H S H C H H H 3 H Xaa 237 Lys 234 Ser 70

87 Figure 1-4. Schmeatic representation of the Zn 2+ -binding site of a dinuclear subclass B1 metallo- -lactamase, such as B. cereus BcII. H His H 2 His H His H His Zn1 H H Zn2 Asp S Cys 85

88 Figure 1-5. Proposed mechanism of inhibition for class A β-lactamases by clavulanate showing the different acyl-enzyme fragmentation products (expressed in atomic mass units, amu) that have been experimentally observed. Ser 70 propynyl enzyme Δ+ 52 Ser 70 Ser 130 cross-linked enzyme Δ+ 52 Ser 70 H aldehyde Δ + 70 Ser 70 H H hydrated aldehyde Δ + 88 H H Ser 70 decarboxylated trans-enamine Δ Ser Irreversible inactivation products C 2 H Ser 70 C acyl-enzyme Δ Deacylation and regeneration of active enzyme H H Ser 70 imine Δ C H H C Ser70 cis-enamine Δ H Ser70 H C Transient inhibition intermediates H Ser70 H C trans-enamine Δ H 86

89 Figure 1-6. Tautomers of imipenem hypothesized to form after acylation of carbapenems by serine β-lactamases. The deacylation rate of the Δ 1 -pyrroline is significantly slower than the Δ 2 -pyrroline, and is thought to play a large role in the inhibitory activity of these compounds (456, 513). H H H H S H Ser C 2 -pyrroline tautomer H H H H Ser S C 1 -pyrroline tautomer 87

90 Figure 1-7. Molecular representation of SHV-1-meropenem acyl-enzyme, based on PDB coordinates 2ZD8 (340). The conformation with the carbonyl of the acyl-serine bond in the oxyanion hole (formed by H amides of Ser70 and Ala237) is colored teal, while the conformation with the carbonyl flipped out is colored purple. Residues 244, 130, and 105, which demonstrate significant orientation changes as compared to the apo SHV-1 β- lactamase, are labeled. Also shown are Glu166 and Asn132, important for making hydrogen bonds with the deacylation water (approximate location represented as red sphere) and C 6 hydroxyethyl substituent, respectively. The large C 2 R carbamoylpyrrolidinyl group of meropenem was disordered, and thus only partially illustrated. 88

91 Figure 1-8. Molecular representation of TEM-1 active site showing residues that are most frequently implicated in the development of inhibitor-resistant TEM enzymes. Based on PDB coordinates 1TEM. Ser130 Arg244 Asn276 Met69 Arg275 89

92 Figure 1-9. Representation of proposed Henri-Michaelis preacylation complex of TEM-1 and clavulanate. 90

93 Figure Proposed reaction mechanisms for the inactivation of a serine β-lactamase by: (A) BRL showing formation of the seven-membered thiazepine ring; and (B) L showing intermolecular capture by the pyridyl nitrogen leading to a bicyclic aromatic intermediate. (A) Ser-H S C Ser H S C Ser S H C Ser S C Ser S C (B) H Ser 2 S C R Ser H 2 S C R Ser H S 2 C R Ser H S 2 C R -H Ser H H S 2 C R 91

94 Figure Molecular representation of cross-linked active site residues Ser64 and Lys315 (shown in yellow) formed after aminolysis of -aryloxycarbonyl hydroxamate inhibitor in E. cloacae P99 -lactamase (505). Based on PDB 2P9V. 92

95 CHAPTER 2 The Role of SHV Asn276 in Clavulanate Resistance Reproduced in part with permission from THE RLE F A SECD-SHELL RESIDUE I MDIFYIG SUBSTRATE AD IHIBITR ITERACTIS I THE SHV LACTAMASE: A STUDY F AMBLER PSITI AS276 Sarah M. Drawz, Christopher R. Bethel, Kristine M. Hujer, Kelly. Hurless, Anne M. Distler, Emilia Caselli, Fabio Prati, and Robert A. Bonomo Biochemistry June 2, 2009; Volume 48 (21): Copyright 2009 American Chemical Society Introduction The development of -lactam antibiotics (penicillins, cephalosporins, and carbapenems, Figure 2-1) remains one of the most significant advances in modern medicine. β-lactam antibiotics manifest their bactericidal effects by inhibiting enzymes involved in cell wall synthesis. The broad spectrum of activity and low toxicity of β-lactams make them the "antimicrobial agents of choice" in the treatment of many infections. Unfortunately, the efficacy of these life-saving antibiotics is seriously threatened by the evolution of β- 93

96 lactamases (366). β-lactamases are periplasmic bacterial enzymes (hydrolases) that are frequently found in Enterobacteriaceae and are responsible for increasing resistance to penicillins, cephalosporins, and carbapenems. There are four classes of β-lactamases; classes A, C, and D possess Ser as the active site nucleophile, while class B uses a metal ion (Zn 2+ ) to inactivate β-lactams (186). The most prevalent β-lactamases in Enterobacteriaceae (e.g., Escherichia coli and Klebsiella pneumoniae) are the class A enzymes (e.g., TEM and SHV) (366). In order to overcome class A -lactamase-mediated resistance and preserve the efficacy of these antibiotics, β-lactamase inhibitors (clavulanate, sulbactam, and tazobactam, Figure 2-1) were coupled with β-lactam antibiotics. These mechanism-based inhibitors greatly enhance the utility of their partner β-lactams (amoxicillin, ampicillin, piperacillin, cefoperazone, and ticarcillin) in the treatment of serious infections caused by Gramnegative bacteria in the hospital, skilled nursing facilities, and community settings (61). Regrettably, selective pressure from excess antibiotic use accelerated the emergence of resistance to β-lactam/β-lactamase inhibitor combinations. Single amino-acid substitutions in the TEM and SHV families of class A β-lactamases resulted in enzymes with reduced affinity for β-lactamase inhibitors (72, 186). At the time of this writing, thirty-five inhibitor-resistant TEMs (IRTs) and five inhibitor-resistant SHVs have been described ( Inhibitor-resistant refers to resistance to amoxicillin/clavulanate as defined by antimicrobial susceptibility testing and may, but does not necessarily, include resistance to sulbactam and tazobactam combinations (464). Because resistance to β-lactam/β-lactamase inhibitor combinations challenges our ability to successfully treat serious infections, attention has focused on designing and discovering novel inhibitors which can help overcome IR enzymes (71, 187, 320, 497). A 94

97 thorough understanding of the molecular basis of IR β-lactamases is essential to inform inhibitor development and anticipate future resistance determinants. Currently, two primary structural changes explain resistance to clavulanate in TEM and SHV enzymes: (i) alterations in the geometry or shape of the "oxyanion hole" or electrophilic center; and (ii) changes in the position of Ser130 or Arg244 (we use the Ambler numbering system for class A β-lactamases) (121, 461, 465, 470, 494). The substitutions most frequently isolated in the clinic and producing the most pronounced inhibitor resistance in TEM include variants at Ambler positions Met69, Ser130, Arg244, Asn275, and Asn276 (8, 508). Met69 and Ser130 substitutions are seen in clinical IR SHVs, but Arg244, Asn275, and Asn276 substitutions have not yet been detected. Despite 68% sequence identity between SHV and TEM enzymes, these closely related β- lactamases demonstrate differences in the kinetic details which confer resistance to clavulanate and sulbactam (251, 465). Previous studies indicated that Asn276 in TEM-1 plays a critical role in inhibitor inactivation by way of its interaction with Arg244 (428, 449). Arg244 is positioned between the binding pocket and Asn276, and contributes directly to substrate and inhibitor affinity in TEM and SHV through hydrogen bonding with the conserved β- lactam carboxylate (217, 464, 465, 512). The Asn276Asp substitution in TEM results in an amoxicillin/clavulanate-resistant phenotype, reduced k inact, and increased K i values (78, 428, 449, 479). The X-ray crystal structure of TEM Asn276Asp highlighted the importance of the electrostatic environment surrounding residues 244 and 276 (449). The TEM Asn276Asp substitution created a new interaction with Arg244 (276Asp 2:Arg244 2); surprisingly, movement or repositioning of Arg244 was not observed. In the case of the wild-type (WT) TEM-1 enzyme crystal structure, there is only 95

98 a hydrogen bond between Asn276 and Arg244; the TEM Asn276Asp variant has both a hydrogen bond and a salt bridge between these residues (228). Based upon these reported findings, we were compelled to address the role of Asn276 in the SHV family of β-lactamases and discern its contribution to both substrate hydrolysis and clavulanate inhibition. Accounting for the different kinetic parameters of inhibition between TEM and SHV enzymes, we specifically inquired whether Asn276, located remotely from the active site (a second-shell residue) would have a major impact on the ability of SHV β-lactamase to discriminate between various β-lactams and clavulanate. To this end, we performed susceptibility testing on a full panel of amino acid variants at residue 276 constructed by site-saturation and site-directed mutagenesis and completed steady state kinetic characterization of the clavulanate-resistant Asn276Asp variant. To compare the nature of the intermediates in the inactivation process, we inhibited SHV-1 and SHV Asn276Asp with clavulanate and resolved the resulting adducts by timed ESI-MS. In addition, novel boronic acid derivatives, a methylidene penem, and meropenem (Figure 2-1) were tested as probes of the functional consequences of the Asn276Asp substitution on β-lactam carboxylate recognition and affinity. ur data leads us to conclude that in both SHV and TEM, perturbations of the electrostatic relationships between Asn276 and Arg244 create unique structural/conformational changes that have significant consequences on the turnover of substrates and inhibitors. However, the Asn276Asp substitution in SHV "enjoys" clavulanate resistance at a lower cost to catalytic efficiencies than most other IR SHVs and IRTs. Materials and Methods Mutagenesis and Sequencing 96

99 Using the template bla SHV-1 gene in the phagemid vector pbc SK(-) (Stratagene, La Jolla, CA), we employed Stratagene s QuikChange Mutagenesis Kit with degenerate oligonucleotides at Ambler position 276 to make the full complement of amino acid substitutions. After PCR mutagenesis, we electroporated the resultant plasmids into E.coli ElectroMAX DH10B cells (Invitrogen, Carlsbad, CA). These plasmids were isolated and DA sequencing of the bla SHV genes was performed. Variants not acquired after sequencing of 100 clones were created with specific mutagenic oligonucleotides as previously described (210). Immunoblotting E. coli DH10B cells were grown to D 600 = 0.8 and lysed by 10 min incubation at 100 ºC in SDS loading dye buffer. Immunoblots were performed with an anti-shv polyclonal antibody to confirm production of the SHV -lactamases from all 19 constructs as previously described (207). Antimicrobial Susceptibility Tests MICs for E. coli DH10B cells expressing the bla SHV-1 or mutant bla SHV Asn276Xxx genes cloned in phagemid pbc SK(-) were obtained by lysogeny broth (LB) agar dilution. The MICs for various antibiotics were determined using a Steers Replicator that delivered 10 μl of a diluted overnight culture containing 10 4 colony forming units. Ampicillin, piperacillin, cephalothin, and cefotaxime were purchased from Sigma (St. Louis, M); meropenem was purchased from AstraZeneca Pharmaceuticals (Wilmington, DE). Lithium clavulanate (GlaxoSmithKline, Surrey, United Kingdom), sulbactam (Pfizer, La 97

100 Jolla, CA), and tazobactam (Wyeth Pharmaceuticals, Pearl River, Y) were kind gifts and were tested in combination with 50 μg/ml ampicillin. -Lactamase Expression and Purification -Lactamase enzymes were expressed and purified as described previously (271). Briefly, E. coli DH10B cells containing the bla SHV-1 or bla SHV Asn276Asp genes in pbc SK (-) were grown overnight in SB medium, harvested by centrifugation at 4 ºC, and frozen at -20 C. After thawing, β-lactamase was liberated using stringent periplasmic fractionation with 40 μg/ml lysozyme (Sigma) and 1 mm EDTA, ph 7.8. Preparative isoelectric focusing was performed with the lysate in a Sephadex granulated gel (GE Healthcare, Piscataway, J) using ampholines in the ph range and running the gel overnight at a constant power of 8 W on a Multiphor II isoelectric focusing apparatus (GE Healthcare). Purity was assessed by 5% stacking, 12% resolving sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE). -lactamase concentration was determined using the Bio-Rad (Hercules, CA) Bradford protein assay with bovine serum albumin standards. If purity was < 90%, a second purification step was performed using size-exclusion chromatography with a Pharmacia ÄKTA Purifier System (fast protein liquid chromatography instrument) (Uppsala, Sweden) (367). Kinetic Measurements Steady state kinetics were performed using continuous assays at room temperature in an Agilent 8453 diode array spectrophotometer (Agilent, Palo Alto, CA). Each assay was performed in 10 mm phosphate-buffered saline (PBS) at ph 7.4. Measurements were 98

101 obtained using nitrocefin (CF) (BD Biosciences, San Jose, CA) (Δε 482 = M -1 cm -1 ), ampicillin (Δε 235 = -900 M -1 cm -1 ), piperacillin (Δε 235 = -820 M -1 cm -1 ), and cephalothin (Δε 262 = M -1 cm -1 ). In order to specifically investigate the role of the C 3 /C 4 carboxylate in substrate recognition, we used: (i) an achiral and chiral boronic acid cephalothin analog (465); (ii) a methylidene penem inhibitor bearing a dihydropyrazolo[1,5-c][1,3]thiazole moiety (Wyeth Pharmaceuticals) (482); and (iii) the carbapenem meropenem. The kinetic parameters, V max and K m, were obtained with non-linear least squares fit of the data (Henri Michaelis-Menten equation) using rigin 7.5 (riginlab, orthampton, MA): v = (V max [S])/(K m + [S]) The K i values of the boronic acid derivatives were determined by direct competition assays. The initial velocity was measured in the presence of a constant concentration of enzyme (10 nm) with increasing concentrations of inhibitor against the indicator substrate, CF. Because of an incubation effect, SHV β-lactamase and the chiral boronic acid compound were preincubated for 5 min in PBS before initiating the reaction with the addition of substrate (CF), as has been described previously (320, 465). In these competition assays, initial velocities (v 0 ) can be represented by the following equation (107): v 0 = (V max [S]) / {K m (1 + I/K i ) + [S]} To determine K i, the initial velocities immediately after mixing (1/v 0 ) were plotted as a function of inhibitor concentration: (1/ v 0 ) = K m /(V max [S]) (1 + I/K i ) + 1/V max 99

102 Rearrangement of the equation shows: (1/ v 0 ) = K m /(V max [S]) (I/K i ) + 1/V max K m /[S]) Where I is K i obs, the equation is solved by setting 1/v 0 to 0. K i obs = {1/V max K m /[S])} (K m /V max [S]K i ) K i obs = K i ([S]/K m (K m /V max [S]) K i = K i obs /(1 + [S]/K m Thus, the K i obs is the concentration of I that reduces the velocity by 50%, and can be calculated from the x-intercept times -1 (99). Direct competition assays between CF and SHV -lactamases were used in the same manner to measure the inhibitory activity of the meropenem. However, in contrast to the boronic acid derivatives which are reversible inhibitors of class A and C -lactamases, enzyme acylation by carbapenems may not be irreversible within the timeframe of the assay β-lactamase hydrolytic reaction (23, 110). As a true K i determinations depend on reversible equilibria, we refer to this steady-state parameter of the meropenem as K i app (493). The first-order rate constant for enzyme and inhibitor complex inactivation, k inact, was obtained by monitoring the reaction time courses in the presence of clavulanate or the penem inhibitor. Fixed concentrations of enzyme (10 nm) and CF (150 μm) and increasing concentrations of inhibitor were used in each assay. The k obs was determined using a non-linear least squares fit of the data using rigin 7.5 : A = {A 0 + v f t + (v 0 -v f )[1-exp (-k obs t)]} / k obs Here, A is absorbance, v 0 (expressed in variation of absorbance per unit time) is initial velocity, v f is final velocity, and t is time. Each k obs was plotted versus I and fit to 100

103 determine k inact. The value of k inact was used to determine K I, the concentration of inhibitor needed to achieve ½ k inact (107): k obs = (k inact [I])/(K I + [I]) The K I data were corrected to account for the affinity of CF for SHV-1 and Asn276Asp according to the following equations (120): K I (corrected) = K I (observed) / [1 + ([S]/K m CF)] The partitioning of the initial enzyme inhibitor complex between hydrolysis and inactivation, i.e., the partition ratio (k cat /k inact or t n ), was calculated by plotting the relative β-lactamase activity against the inhibitor: enzyme ratio as previously described by Bush and colleagues (69). Briefly, increasing amounts of inhibitor were incubated with a fixed concentration of β-lactamase (10 nm). After 24 hours, an aliquot was removed from the mixture and the initial velocity was measured and compared with a control sample with no inhibitor added. The proportion of inhibitor to β-lactamase that resulted in 90% inactivation after 24 hours was defined here as the partition ratio. Mass Spectrometry For intact protein mass spectrometry to determine intermediates and products of inactivation, we incubated 40 μm of SHV-1 or Asn276Asp with and without addition of 40 mm clavulanate for 15 min. Each reaction was terminated by the addition of 0.1% trifluoroacetic acid and immediately desalted and concentrated using a C 18 ZipTip (Millipore, Bedford, MA) according to the manufacturer s protocol. Samples were then placed on ice and analyzed within 1 hr. 101

104 Spectra of the intact SHV-1 and Asn276Asp proteins were generated on an Q-STAR XL quadrupole-time-of-flight mass spectrometer (Applied Biosystems, Framingham, MA) equipped with a nanospray source. Experiments were performed by diluting the protein sample with acetonitrile/1% trifluoroacetic acid to a concentration of 10 μm. This protein solution was then infused at a rate of 0.5 μl/min and the data were collected for 2 min. Spectra were deconvoluted using the Analyst program (Applied Biosystems). All measurements have an error of ± 3 amu. Molecular Modeling The crystal structure coordinates of the SHV-1 β-lactamase from the Research Collaboratory for Structural Bioinformatics Protein Data Bank (PDB entry 1SHV) were used to generate a representation of SHV Asn276Asp β-lactamase inactivated by clavulanate. The reader is referred to Appendix A for a general description of molecular modeling terms. Hydrogen atoms were added (ph= 7.5), and the structure minimized at a constant dielectric of 1 with the Conjugate Gradient method and a Constant Valence Force Field (Accelrys Insight II, Discover_3, San Diego, CA). The energy minimization of the PDB file was followed by an Asn276Asp substitution. The model for clavulanate was constructed using the Builder Module of Insight II. In order to represent the acyl-enzyme intermediate, a bond was created between C 7 of clavulanate and Ser70. The enzyme-inhibitor complex was energy minimized to a convergence of 0.02 Å (10,000 iterations). Results Construction of Variants at Ambler Position 276; Mutagenesis and Immunoblotting. 102

105 Using degenerate oligonucleotides, we performed site-saturation mutagenesis on the bla SHV-1 gene at Ambler position 276. Seventeen of 19 amino acid substitutions were obtained in the initial sequencing screen (100 bla SHV genes selected and sequenced). bla SHV-1(Asn276Phe) and bla SHV-1(Asn276Thr) were constructed by site-directed mutagenesis. Expression of β-lactamase proteins was confirmed by immunoblotting. With the possible exception of the SHV Asn276Pro variant, all enzymes were expressed at steady state (0.8 D) (Figure 2-2). Antimicrobial Susceptibility Tests To test the impact of the single amino acid substitutions at Asn276 on β-lactam susceptibility, MICs against all 19 variants were determined in a uniform E. coli DH10B background. Results are summarized in Table 2-1. All Asn276 variants possessed lower MICs for the penicillins and cephalothin as compared to SHV-1. The ampicillin and piperacillin MICs (8192 and 512 μg/ml, respectively) were relatively preserved for the Asn276Asp variant compared to the remaining Asn276 variants, suggesting that the Asn276Asp variant retains greater ability to hydrolyze penicillins. It is particularly noteworthy that there is a very wide range of MICs (2 to 8192 μg/ml) and that the pattern seen is not based upon size of the R group, charge, polarity, or hydrophobicity. Each of the variants at Asn276 demonstrated reduced susceptibility to cephalothin (range 4 to 16 μg/ml). MICs against the extended-spectrum cephalosporin cefotaxime and meropenem were also performed and all Asn276 variants had values comparable to SHV-1 (data not shown). We with regard to -lactam/ -lactamase inhibitor combinations, only the Asn276Asp enzyme resulted in a higher MIC value for the ampicillin/inhibitor combinations: 103

106 increased resistance to ampicillin/clavulanate and a value comparable to SHV-1 for ampicillin/sulbactam and ampicillin/tazobactam. Kinetic Behavior of Asn276Asp with β-lactam Substrates Because of its phenotype and potential clinical relevance, we purified the Asn276Asp variant enzyme for kinetic analysis. Except for ampicillin, we observed that the Asn276Asp variant has a higher K m against all substrates (Table 2-2). In contrast, the k cat value is lowered save for piperacillin, which has a value similar to SHV-1. verall, the catalytic efficiencies, k cat /K m, were slightly reduced for the Asn276Asp variant. Kinetic Behavior of Asn276Asp with Clavulanate Table 2-3 summarizes the kinetic data for SHV-1 and the Asn276Asp variant inactivated by clavulanate. The clavulanate K I value was increased 1.6-fold for Asn276Asp as compared to SHV-1. Unlike studies of SHV Arg244Ser, we did not find a difference in clavulanate k inact for Asn276Asp and SHV-1 (465). We measured the partition ratio for both SHV-1 and Asn276Asp and observed only a slight increase in the number of clavulanate molecules hydrolyzed by the variant enzyme before inactivation was achieved (40 versus 50). Similarly, the k cat calculations indicate a small increase in the rate of inhibitor hydrolysis for the Asn276Asp variant. Kinetic Behavior of Asn276Asp with -Lactamase Inhibitors: Probes of the Active Site In order to investigate the functional interactions responsible for the SHV Asn276Asp IR phenotype, we assayed two boronic acid cephalothin analogs, an investigational methylidene penem, and meropenem for their activities against SHV-1 and Asn276Asp 104

107 enzymes (Table 2-3). The boronic acid derivatives tested contain the R 1 side chain of cephalothin and differ only by the addition of a meta-carboxyphenyl group on the chiral compound. The meta-carboxyphenyl moiety contains a carboxylate that is linked to the sp 2 -hybridized carbon atom of a phenyl ring, and is intended to imitate the interactions with the C 3 /C 4 carboxylate of penicillins and cephalosporins. These boronates allow us to assess the extent to which the Asn276Asp substitution contributes to affinity for the substrate and/or inhibitor carboxylate by its interaction with Arg244 (465). The chiral boronate with the meta-carboxyphenyl group demonstrated 62-fold greater affinity for SHV-1 than the achiral compound (42 μm versus 0.68 μm, respectively), suggesting that interactions with the C 3 /C 4 carboxylate are very important for increasing affinity for the enzyme s active site. Testing the boronic acid derivatives with Asn276Asp also showed that the presence of the meta-carboxyphenyl on the chiral compound increased affinity, but the improvement in K i of binding the chiral compound over the achiral was reduced to only 9-fold (36 μm versus 3.8 μm, respectively). We next studied a methylidene penem that contains an sp 2 -hybridized C 3 bearing the carboxylate and a bicyclic R 1 side chain. The K I values of the penem for SHV-1 and SHV Asn276Asp were in the nm range, with a 1.5-fold increase for the Asn276Asp variant. The k inact rate was also slightly higher for the Asn276Asp variant, but the overall inhibitor efficiency (k inact /K I ) and partition ratios were very similar between SHV-1 and Asn276Asp enzymes. Meropenem resists hydrolysis by SHV β-lactamases and forms a stable acyl-enzyme (340). Here, the affinity of meropenem for Asn276Asp was assessed by a competition reaction with CF to obtain a K i app value. The affinity loss of the Asn276Asp enzyme for the carbapenem was greatest, a 46-fold increased K i app as compared to SHV

108 Determining the ature of the Intermediates: ESI-MS of SHV-1 and Asn276Asp with Clavulanate Timed ESI-MS was performed with SHV-1 and Asn276Asp β-lactamase inactivated by clavulanate using an inhibitor:enzyme ratio of 1000:1 to ensure detection of the covalent intermediates in the inactivation pathway. As shown in Figure 2-3, when SHV-1 and Asn276Asp were incubated with clavulanate for 15 min, nearly identical covalent intermediates were formed. This included adducts at Δ +51, Δ +70, Δ +88, Δ +140, and Δ +158 which were observed previously with TEM-2, SHV-1, and the IR SHV Arg244Ser (all measurements have an error of ± 3 amu, see Figure 1-5 for proposed reaction intermediates corresponding to adducts) (55, 445, 465). These similarities in the spectra of inactivation species argue strongly for a common pathway and reaction mechanism. Discussion We undertook an in-depth study of Ambler position 276 in SHV to enhance our understanding of the IR phenotype in class A β-lactamase enzymes, and to explore potentially relevant differences between IR SHVs and IRTs. ur goal was to elucidate the role of this second-shell residue on the catalytic profile of SHV enzymes. The Impact of Substitutions at Ambler Position Asn276 on Phenotype Susceptibility testing of the full repertoire of SHV Asn276 variants showed that only the Asn276Asp enzyme was uniquely able to reduce susceptibility to ampicillin/clavulanate. Furthermore, this variant displayed another distinct feature that 106

109 contrasts what is typically observed with IR enzymes: the Asn276Asp variant retained a high level of resistance to ampicillin and piperacillin (Table 2-1). It is noteworthy that all the other variants at position 276 were uniformly less resistant to ampicillin/sulbactam and ampicillin/tazobactam than SHV-1 expressed in E. coli DH10B. This pattern was observed by Thomson and colleagues in studies exploring the role of Arg244 in SHV (464, 465). Based on the Asn276Asp enzyme s ability to balance these desirable catalytic properties (i.e., avoiding inhibition by clavulanate and retaining hydrolytic activity for penicillins) and its potential clinical relevance, we selected Asn276Asp for detailed study. Steady State Kinetics Steady state kinetic analyses were performed to explain the biochemical correlates of the Asn276Asp phenotype for substrate and inhibitor catalysis. verall, we found that the Asn276Asp β-lactamase is moderately impaired when compared to SHV-1 (k cat /K m ratios are 41-75% of WT) (Table 2-2). f the parameters measured, the K m and catalytic efficiency (k cat /K m ) for the penicillins were least affected by the 276 substitution. CF and cephalothin demonstrated slightly greater reductions in affinities (higher K m ) and turnover (lower k cat ). The most striking difference compared to SHV-1 was that Asn276Asp showed a 46-fold change in K i app for meropenem. Based upon previous analysis and work in SHV performed by ukaga, Thomson, and colleagues, we attribute this pattern (loss of penicillin > cephalothin > meropenem affinities) to the interaction of the Asn276Asp β-lactamase with the C 3 /C 4 carboxylate of these substrates (340, 465). Carboxylates linked to C 3 or C 4 sp 2 -hybridized carbons (meropenem and cephalothin, respectively) have reduced rotational freedom due to conjugation as compared to 107

110 carboxylates linked to a C 3 sp 3 -hybridized carbon (i.e., penicillins). Thomson and colleagues suggested that the carboxylate linked to a C 3 /C 4 sp 2 -configured carbon is brought in closer approximation to Arg244 of SHV in the Henri-Michaelis complex, likely leading to increased hydrogen bonds with this residue (465). This contrasts with the notion that Arg244 may make only one relatively weak hydrogen bond with the penicillin carboxylate linked to a C 3 sp 3 -hybridized carbon; the anchoring interactions come from other residues in the binding pocket, such as Thr235 (217). Thus, by examining substrates with different stereochemistry at the C 3 /C 4 position, we infer that the substitutions at residue 276 in SHV affect the affinity and kinetics of how Arg244 binds to substrates. Resistance to clavulanate in SHV Asn276Asp is mediated by decreased affinity. This is a property shared with the IR TEM Asn276Asp enzyme. However, the magnitude of affinity loss is less for the SHV variant than the TEM variant (428, 449). The inactivation efficiencies (k inact /K I ) reveal that SHV Asn276Asp retains more susceptibility to clavulanate (449). Molecular Modeling Based on phenotypic and kinetic study of SHV Asn276Asp, we created a molecular model to complement our observations. ur model reveals that new electrostatic interactions are created by the Asn276Asp substitution in SHV (Figure 2-4). In SHV Asn276Asp, the distances between atoms 2 of Arg244 and both δ1 and δ2 of 276Asp are decreased as compared to SHV-1, by 0.9 and 0.6 Å, respectively. These observations are reminiscent of the distances observed in the atomic structure of TEM Asn276Asp ( 2 of Arg244 and δ2 of 276Asp are closer by 1.0 Å) (228, 449). Yet, why does Asn276Asp maintain catalytic efficiency against penicillins, demonstrate 108

111 resistance to inactivation by clavulanate, and show such wide changes in K i when probes are tested? Why are all the other 276 substitutions so ineffective? We posit that the Asp substitution is singular in its ability to create a more rigid enzyme by impairing the conformational flexibility of Arg244. Close inspection of the atom of Arg244 in the crystal structures of TEM-1 and TEM Asn276Asp (PDB entries 1BTL and 1CK3, respectively) reveals that this residue exists in multiple conformations. The two new interactions between Arg244 and 276Asp in SHV Asn276Asp restrict facile hydrogen bond formation of the Arg244 guanidinium group with the C 3 carboxylate of clavulanate. As Arg244 plays a key role in substrate recognition, we would anticipate that major effects would be seen for substrates as well. ur MIC data show that 18 of the 19 substitutions at Asn276 result in a less robust penicillinase as a result of this rigidity (the other residues at 276 do not permit productive interactions with Arg244). Most importantly, we also see resistance to clavulanate. This latter property is most likely a consequence or benefit of restricted flexibility. This argument is supported by the increased K m, K I, and changes in MICs. Insight from Inhibitors To explore further the functional interactions of the Asn276Asp substitution, we next turned to inhibitors (boronic acid derivatives, a methylidene penem, and meropenem) with carboxylates linked to sp 2 -hybridized atoms. The inhibitors were chosen because they help us answer: (i) the extent to which the position and stereochemistry of the carboxylate impacts affinity for the Asn276Asp variant; and (ii) if other interactions beside Arg244-Asn276Asp are important for inhibitor binding. 109

112 Boronic acid derivatives were selected because they replace the β-lactam motif with boronic acid and form reversible, dative covalent bonds with the active site serine (94, 110). This leads to formation of an adduct which resembles the geometry of the tetrahedral transition state of the β-lactamase hydrolytic reaction. By modifying these compounds to contain the R 1 side chains of natural substrates at the same distance from the electrophilic moiety, affinities in the nm range for both class A TEM-1 and extendedspectrum-type SHV and class C enzymes were reported (82, 466). Here, we tested one achiral boronate bearing the R 1 amide side chain found in cephalothin and a second chiral boronate that has an additional meta-carboxyphenyl ring on the carbon atom alpha to the boron which resembles the dihydrothiazine ring of cephalosporins. Since the geometry and the distances in the chiral boronate are the same as in the natural substrate, the carboxylate moiety of the meta-carboxyphenyl group corresponds to the C 3 /C 4 carboxylate found on all β-lactams. Thus, comparing K i of the achiral and chiral compounds allowed us to probe selectively how the Asn276Asp substitution affects binding of the carboxylate of cephalosporins. The K i gain of the carboxylate (i.e., the difference etween the achiral and chiral compounds) for the SHV-1 enzyme was almost 7 times increased for the Asn276Asp variant. Because of the selective design of these inhibitors, we attribute this difference to a less favorable interaction between Arg244 and the inhibitor carboxylate in the Asn276Asp variant. In total, these results support our hypothesis that residue 276, despite not making direct bonds with the substrate, plays a significant role in coordinating the carboxylate in the substrate-enzyme complex. Methylidene penems and their derivatives are effective inactivators of class A, C, and D β-lactamases, including IR class A enzymes (30, 339, 464, 481). Assays of one of these 110

113 novel penems yielded similar inhibitor efficiencies for SHV-1 and Asn276Asp. As proposed in studies with the SHV Arg244Ser variant, the affinity losses at the C 3 carboxylate of the penem may be compensated by favorable interactions with the inhibitor s bicyclic R 1 side chain (464). Previous molecular models and structure determinations have shown that the R 1 side chain may be involved in π-π interactions with Tyr105 (339). Despite the sp 2 -hybridized carboxylate, the preserved efficiency of this penem against the clavulanate-resistant Asn276Asp variant stresses that more than one significant (compensatory) interaction is possible in the active site (30). In contrast to stabilizing interactions of the R 1 side chain of methylidene penem inhibitors, carbapenems contain a hydroxyethyl group at the R 1 position and a large R 2 side chain, both of which have destabilizing effects in crystal structures of carbapenem/class A β-lactamase complexes (302, 340, 493). Thus, that SHV Asn276Asp experiences the greatest K i app loss for meropenem, as compared to decreases in the inhibition constants for the other substrates and inhibitors tested, is consistent with the notion that the interactions between the variant enzyme and the ligand carboxylate are impaired. Meropenem has limited flexibility and rigid carboxylate on the sp 2 carbon atom. Comparison to IR SHV and TEM Enzymes Can we compare IR SHV enzymes with IRTs and what do we learn from this? Examined together, our analysis of substrate and inhibitor catalysis in SHV-1 and SHV Asn276Asp represents a fine tuning of the IR phenotype. In general, IR enzymes must establish a balance between retaining the ability to hydrolyze β-lactam substrates, but decreasing efficacy of the structurally related β-lactam inhibitors (72). The majority of the 111

114 IR enzymes "accept" a significant decrease in substrate hydrolytic efficiency in exchange for dramatically decreased inhibitory efficiency. This produces an IR phenotype when organisms expressing these enzymes are tested against β-lactam/β-lactamase inhibitor combinations. MICs represent the sum of an enzyme s ability to hydrolyze the combination of both β-lactam (e.g., ampicillin) and the β-lactamase inhibitor (e.g., clavulanate) in a relatively short time frame. ur data shows that in the case of SHV Asn276Asp, the increased ampicillin/clavulanate resistance results from a small decrease in clavulanate inhibition, and is also due to the preserved ability of the enzyme to hydrolyze the ampicillin present in the β-lactam/β-lactamase inhibitor combination. We further illustrate these unique properties of SHV Asn276Asp by comparison with the kinetic data of a collection of IR SHV and IRT variants; firstly, the relatively high catalytic efficiency for ampicillin; and secondly, the slightly decreased inhibitory efficiency of clavulanate (see Tables 2-4 and 2-5 for substrate and inhibitor efficiencies of a panel of IR SHV and IRT enzymes). The Asn276Asp alteration of the IR phenotype in the SHV β-lactamase may represent a significant evolutionary advance, as the enzyme maintains a balance of the desired catalytic properties. For clavulanate-resistant TEM enzymes, substitutions at Asn276 and Arg244 lead to decreased occupancy of a water molecule essential for inactivation by clavulanate (217, 428, 449, 494). To support our model for SHV, we propose that the alteration imposed by the Asn276Asp substitution may not affect the (yet to be identified) water molecule crucial to clavulanate inactivation in SHV (Figure 2-5). Previous examination of SHV and TEM crystal structures suggests that SHV enzymes may not rely on the water molecule coordinated by Arg244 and Val216 for inactivation by clavulanate (228, 251, 465). Consequently, the attenuated IR SHV Asn276Asp phenotype is produced by 112

115 decreased affinity for the carboxylate of the inhibitors, and not by disruption of the inactivation mechanism from a dislocated water molecule. These important differences in TEM and SHV mechanisms support a divergent evolution of these enzymes viz a viz substrate and inhibitor profiles. Thus, we posit that the SHV Asn276Asp variant may exist in clinical isolates of Enterobacteriaceae. As a result of the attenuated IR phenotype, these variants have not risen to clinical attention. This contrasts the TEM Asn276Asp enzyme, which exhibits a more significant decrease in catalytic activity, but more pronounced IR. In general, the prevalence of IR β-lactamases may be underestimated as identification of IR enzymes requires specific kinetic characterization and susceptibility testing (72, 84). Recognition of existing IR SHV variants, particularly those with a less pronounced phenotype, may have escaped detection. Conclusion In summary, we highlight the important role of a second-shell residue in class A β- lactamases and describe the unique properties of Ambler residue Asn276 in SHV. Structure-function studies of proteins often focus on first-shell residues that make direct interactions with active-site ligands. Recent studies of metallo-enzymes, including metallo-β-lactamases, have emphasized that second-shell residues can contribute to metal binding affinities and active site charge distributions (12, 468). Additional investigations show that second- and third-shell residues in nitric oxide synthase enzymes modulate important conformational changes in invariant first-shell residues (162). In these proteins, new binding pockets are revealed upon ligand binding, and selective inhibitors can be rationally designed to take advantage of how the second-shell residues modulate enzyme 113

116 flexibility. ur data demonstrates that perturbations of the β-lactamase second-shell residue 276 are transferred to the active site by interaction with Arg244 and manifested in decreased affinity for piperacillin, CF, and cephalothin and clavulanate. This new interpretation and deeper understanding for the tertiary structure and interactions between residues is essential to the success of selective and effective inhibitors of these β- lactamase enzymes. The lessons learned from the finely tuned enzyme can focus our efforts to meet the challenges raised by the impressive efficiency of these β-lactamases. 114

117 Table 2-1. MIC values (μg/ml) of E. coli DH10B expressing SHV-1 and Asn276 variants a Ampicillin Piperacillin Cephalothin Ampicillin / clavulanate Ampicillin / sulbactam Ampicillin / tazobactam DH10B SHV-1 > Asn276Asp Asn276Gly Asn276Ser Asn276Val Asn276Thr Asn276Gln, -Glu, -Cys Asn276His Asn276Ala, -Leu, -Ile Asn276Lys, -Arg, -Tyr, -Phe Asn276Met Asn276Pro, -Trp a Inhibitors were evaluated in the presence of 50 μg/ml ampicillin 115

118 Table 2-2. Kinetic properties of SHV-1 and Asn276Asp for ampicillin, piperacillin, nitrocefin, and cephalothin SHV-1 Asn276Asp Ampicillin K m (μm) 183 ± ± 13 k cat (s -1 ) 3100 ± ± 99 k cat /K m (μm -1 s -1 ) 17 ± 4 7 ± 1 Piperacillin K m (μm) 77 ± ± 18 k cat (s -1 ) 898 ± ± 98 k cat /K m (μm -1 s -1 ) 12 ± 1 9 ± 1 itrocefin K m (μm) 21 ± 3 28 ± 3 k cat (s -1 ) 237 ± ± 14 k cat /K M (μm -1 s -1 ) 11 ± 2 5 ± 0.7 Cephalothin K m (μm) 29 ± ± 11 k cat (s -1 ) 14 ± 1 10 ± 1 k cat /K m (μm -1 s -1 ) 0.47 ± ±

119 Table 2-3. Kinetic properties of SHV-1 and Asn276Asp for clavulanate, cephalothin boronic acid derivatives, methylidene penem, and meropenem SHV-1 Asn276Asp Clavulanate K I (μm) 0.72 ± ± 0.11 k inact (s -1 ) ± ± k inact /K I (μm -1 s -1 ) ± ± k cat /k inact (t n ) 40 ± ± 10 k cat (s -1 ) 0.60 ± ± 0.18 Achiral boronic acid cephalothin analog K i (μm) 42 ± 4 36 ± 4 Chiral boronic acid cephalothin analog K i (μm) 0.68 ± ± 0.4 Methylidene penem K I (μm) ± ± k inact (s -1 ) 0.17 ± ± 0.02 k inact /K I (μm -1 s -1 ) 6.5 ± ± 3.0 k cat /k inact (t n ) 2 ± 1 2 ± 1 k cat (s -1 ) 0.34 ± ± 0.24 Meropenem K i app (μm) 31 ± ±

120 Table 2-4. Ratio of k cat /K m for IR to wild-type TEM and SHV enzymes SHV variants Asn276Asp Arg244Ser Ser130Gly Met69Leu, -Val, -Ile Substrate Ampicillin Ampicillin Ampicillin Ampicillin Ratio (%) , 66.6, Ref. (this work) (465) (185) (134, 188) TEM variants Asn276Asp Arg244Thr, -Gln Ser130Gly Met69Leu, -Val, -Ile Substrate Amoxicillin Amoxicillin Ampicillin Amoxicillin Ratio (%) , , 68.6, 31.4 Ref. (428) (121) (461) (85) 118

121 Table 2-5. Ratio of k inact /K i for IR to wild-type TEM and SHV enzymes SHV variants Asn276Asp Arg244Ser Ser130Gly Met69Leu, -Val, -Ile Substrate Clavulanate Clavulanate Clavulanate Clavulanate Ratio (%) A Ref. (this work) (465) (185) TEM variants Asn276Asp Arg244Thr, -Gln Ser130Gly Met69Leu, -Val, -Ile Substrate Clavulanate Clavulanate Clavulanate Clavulanate Ratio (%) 1.5 < 0.008, < , 1.6, 11.1 Ref. (449) (121) (461) (85) 119

122 Figure 2-1. Chemical structures of compounds tested in Chapter 2. The structure of cephalothin is labeled with the accepted ring numbering system. Penicillins H 2 H H Ampicillin H S sp 3 H H Piperacillin H H S sp 3 H Cephalosporins S H H 7 S R 1 side chain sp 2 H Cephalothin R 2 side chain S H H S H itrocefin + + Inhibitors 6 7 H H H H S CH 3 H CH 3 H S CH 3 H Clavulanate Sulbactam Tazobactam S H S B H H Achiral boronic acid cephalothin analog H S B H sp 2 H Chiral boronic acid cephalothin analog H S sp 2 H H H H Methylidene penem sp 2 H S H Meropenem 120

123 Figure 2-2. Immunoblot of E. coli DH10B expressing SHV-1, SHV Arg244Ser variant, and Asn276 variants probed with SHV-1 polyclonal antibody. 121

124 Relative Intensity Figure 2-3. Deconvoluted ESI-MS spectra of: (A) SHV-1; and (B) SHV Asn276Asp before and after 15 min inactivation with clavulanate at inhibitor:enzyme ratio of 1000:1. All measurements have an error of ± 3 amu. Eight distinct mass shifts were identified with both enzymes inactivated with clavulanate. This includes adducts postulated to represent: Δ +51, terminally inactivated cross-linked or propynoyl enzyme species; Δ +70, aldehyde; Δ +88, hydrated aldehyde; Δ +158, decarboxylated imine; and Δ +198, acyl-enzyme, imine, cis- or trans-enamine (see Figure 1-5) (55, 465). (A) (B) SHV SHV Asn276Asp SHV-1 and clavulanate Mass adducts SHV-1 Asn276Asp and clavulanate Mass adducts Mass (amu) Mass (amu) 122

125 Figure 2-4. Molecular representation of SHV Asn276Asp based on SHV-1 (PDB entry 1SHV) showing the increased interaction between Arg244 and 276Asp mediated by shortened distances between Arg244 η2 and both 276Asp δ1 and δ2 (decreased by 0.9 and 0.6 Å, respectively). Residue 216 is shown for reference. SHV-1 amino acid positions in white and SHV Asn276Asp model in color. 123

126 Figure 2-5. Proposed reaction mechanism for inactivation of SHV-1 by clavulanate. Scheme shows the contribution of Arg244 in the coordination of the water molecule hypothesized to donate a proton for C 2 double bond saturation and subsequent secondary ring opening. In SHV-1, it is postulated that the water molecule is recruited or relocated into the active site with the binding of clavulanate (465). The crystal structure of TEM Asn276Asp (PDB entry 1CK3) reveals that this crucial water is missing, offering an explanation for the IR phenotype as protonation of clavulanate would be impaired, and thus lead to more ready hydrolysis of the inhibitor than in TEM-1, where the water is clearly refined (449). SHV-clavulanate acyl-enzyme imine inactivation intermediate 124

127 CHAPTER 3 Inhibition of ADC by Boronates and Carbapenems Reproduced in part with permission from IHIBITI F THE CLASS C -LACTAMASE FRM ACIETBACTER SPP: ISIGHTS IT EFFECTIVE IHIBITR DESIG Sarah M. Drawz, Maja Babic, Christopher R. Bethel, Magda Taracila, Anne M. Distler, Claudia ri, Emilia Caselli, Fabio Prati, and Robert A. Bonomo Biochemistry January 19, 2010; Volume 49 (2): Copyright 2010 American Chemical Society Introduction Acinetobacter spp. are Gram-negative pathogens responsible for an increasing number of serious nosocomial infections including hospital-acquired pneumonia, urinary tract infections, and bacteremia (101, 154, 176, 391). This nonfermentive, aerobic pathogen harbors multiple antibiotic resistance determinants, including chromosomal AmpC β- lactamase enzymes, XA carbapenemases, metallo-β-lactamases, and multidrug resistance (MDR) efflux pumps (377, 486). In addition, changes in outer membrane 125

128 proteins decrease permeability to antimicrobials (280). Besides intrinsic resistance, Acinetobacter spp. possess the ability to acquire new resistance determinants through gene mutations, derepression, and transfer from other organisms. These remarkable attributes can lead to infections resistant to all available β-lactam antibiotics (42, 372). Consequently, treatment of patients with Acinetobacter spp. infections is very challenging, and therapeutic options are severely limited for MDR strains (43, 265, 375). ne strategy for restoring the efficacy of β-lactam antibiotics is the development of novel β-lactamase inhibitors. Boronic acid derivatives are compounds that replace the β- lactam ring with boronic acid. The boron atom forms a reversible, dative covalent bond with the active site serine of class A and C β-lactamases, assuming a geometry that resembles the tetrahedral transition state of the β-lactamase hydrolytic reaction (Figure 3-1) (23, 110). By modifying the boronic acid substituents to resemble in structure, distance, and stereochemical arrangement the R 1 side chains of natural substrates, affinities in the nm range against class C enzymes of Escherichia coli are achieved (82, 320, 499). A second approach to counteracting β-lactamase mediated antibiotic resistance is the design of β-lactams that resist hydrolysis. Through the combined efforts of natural product screens and medicinal chemistry, β-lactamase-stable penem and cephem derivatives have been modified and synthesized. The most potent β-lactams are the derivatives of thienamycin (i.e., imipenem, meropenem, ertapenem, and doripenem). Carbapenems act as inhibitors of class A, C, and certain class D β-lactamases by forming a prolonged acyl-enzyme intermediate with the β-lactamase that is very slowly hydrolyzed (21, 155, 294, 302, 317, 340, 378, 456). 126

129 Acinetobacter-derived cephalosporinases (ADC), a class C -lactamase found in Acinetobacter baumannii and Acinetobacter genomospecies 3, is a unique enzyme responsible for resistance to penicillins, cephalosporins, and β-lactam/β-lactamase inhibitor combinations (212). These AmpC β-lactamases demonstrate a remarkable k cat for first-generation cephalosporins and relatively low affinity for the commercial β- lactamase inhibitors (212). ADC enzymes are important targets for the design of new mechanism-based inactivators. To date, the search for effective inhibitors for the ADC β- lactamase remains challenging. To this end, we synthesized and tested a panel of boronic acid derivatives with specific side chains to serve as chemical probes. We also designed a novel boronate that contained the R 1 side chain of cefoperazone. Concurrently, we explored the role of the R 2 side chain of four different carbapenems as inhibitors of ADC. Taken together, our results indicate that the interactions between the ADC β-lactamase and inhibitors scaffolds and side chains yield important insights into the properties of class C enzyme active sites. Materials and Methods Antibiotics and Inhibitors Ceftazidime and cefotaxime were purchased from Sigma (St. Louis, M). Chemical structures of cephalothin and inhibitors used are shown in Figure 3-2. The boronic acid ceftazidime analog and cephalothin analogs were synthesized as previously described (82, 320). The chiral cephalothin analogs 4 and 5 were obtained in the enatiomerically pure form. Imipenem and ertapenem were obtained from Merck & Co. Inc. (Whitehouse Station, J). Meropenem was purchased from AstraZeneca Pharmaceuticals (Wilmington, DE) and doripenem from rtho-mceil Pharmaceutical Inc. (Raritan, J). 127

130 Synthesis of Cefoperazone Analog The cefoperazone analog 1 was synthesized according to the general protocol for the other boronic acid derivatives by acylation of pinacol bis-(trimethylsilyl)- aminomethaneboronate with the commercially available cefoperazone acid, promoted by isobutyl chloroformate (82). Triethylamine (231 μl, 1.66 mmol) and isobutyl chloroformate (216 μl, 1.66 mmol) were added to a solution of (2R)-2[(4-ethyl-2,3- dioxopiperazinyl)carbonylamino]-2-(4-hydroxyphenyl)acetic acid (530 mg, 1.66 mmol) in anhydrous tetrahydrofuran (60 ml) at 0 C and allowed to react under argon atmosphere for 1 h. A solution of bis-(trimethylsilyl)-aminomethaneboronate (500 mg, 1.66 mmol) in anhydrous tetrahydrofuran (5 ml), previously treated for 30 min with anhydrous methanol (1.74 mmol), was added at the same temperature. After 20 min, the cooling bath was removed and the mixture allowed to react overnight at room temperature. Thereafter, the reaction mixture was diluted with diethylether (60 ml) and the precipitate (trithylammoium chloride) was removed by filtration. The solvent was distilled under reduced pressure and the solid residue crystallized from ethyl acetate, affording the title compound as a whitish solid (54% yield). [α] D 59.0 (c 0.8, CH 3 H). The pinacol ester spontaneously hydrolyzes in the phosphate buffer, generating the corresponding cefoperazone boronic acid. 1 H and 13 C MR spectra of compound 1 were recorded on a Bruker DPX-200 spectrometer. The chemical shifts (δ) are reported in parts per million downfield from the internal standard, tetramethyl silane (TMS) (s, singlet; d, doublet; t, triplet; q, quartet; m, multiplet). Coupling constants (J) are recorded in hertz. Mass fragmentations were determined on a Finnigan MAT SSQ A mass spectrometer (electron impact (EI), 70 ev). 128

131 ¹H MR (200 MHz, DMS-d6): δ 1.07 (3H, t, J = 7.2 Hz, CH 2 CH 3 ), 1.15 (12H, s, CH 3 pic), 2.44 (2H, d, J = 3.7 Hz, CH 2 B), 3.38 (2H, q, J = 7.2 Hz, CH 2 CH 3 ), (2H, m, CH 2 pip), (2H, m, CH 2 pip), 5.32 (1H, d, J = 7.3 Hz, CHPh), 6.71 (2H, d, J = 8.5 Hz, meta), 7.19 (2H, d, J = 8.5 Hz, ortho), 8.41 (1H, t, J = 3.7 Hz, H), 9.14 (1H, s, H), 9.64 (1H, d, J = 7.3 Hz, H). 13 C MR (50 MHz, DMS-d6): δ 12.3, 25.1, 25.2, 40.8, 42.1, 43.3, 56.5, 83.3, 115.5, 128.7, 129.2, 152.3, 155.9, 157.5, 159.8, EI MS: m/z (%) 474 (M+, 0.2), 458 (0.4), 400 (2), 374 (2), 332 (14), 317 (10), 275 (16), 274 (96), 273 (36), 216 (19), 173 (9), 148 (18), 142 (58), 121 (35), 120 (46), 99 (100), 83 (15). Genetic Constructs and Host Strains For protein expression and β-lactamase characterization, the bla ADC gene was cloned into pet24a (+) vector (kanamycin resistant, ovagen, Madison, WI) following a previously published method (212). After sequencing verification, the correct construct was maintained in E. coli DH10B cells and transformed into E. coli BL21(DE3) cells for protein expression. For MIC determinations, bla ADC was directionally cloned into the pbc SK(+) phagemid vector (chloramphenicol resistant, Stratagene, La Jolla, CA) as previously described (212). Briefly, the pet24a (+) bla ADC construct was digested with XbaI and BamHI in Multi-Core buffer (Promega, Madison, WI) preserving the 5 upstream flanking region from the pet24a (+) vector in front of the insert when ligated into pbc SK (+). Antimicrobial Susceptibility (MICs) 129

132 E. coli DH10B cells expressing the bla ADC gene were phenotypically characterized by LB agar dilution MICs. The MICs for various antibiotics were determined using a Steers Replicator that delivered 10 μl of a diluted overnight culture containing 10 4 colony forming units. The cephalothin analogs 3 and 5 were tested at a constant concentration of 4 μg/ml in combination with either ceftazidime or cefotaxime. β-lactamase Purification The ADC β-lactamase was prepared from E. coli BL21(DE3) cells after induction with isopropyl-β-d-thiogalactopyranoside (IPTG). Five hundred milliliter cultures were induced at an optical density at 600 nm of 0.5 to 0.8 (final IPTG concentration, 0.2 mm) at 37 ºC for 4 hr in lysogeny broth. These cells were pelleted and resuspended in 50 mm Tris (ph 7.4) and β-lactamase liberated with lysozyme and EDTA per established methods (210). Accordingly, ADC protein was purified by preparative isoelectric focusing and fast protein liquid chromatography (FPLC) with a Sephadex Hi Load 26/60 column (Pharmacia, Uppsala, Sweden) (367). The enzyme was quantified, purity was assessed by SDS-PAGE, and size verified by mass spectrometry (212). Kinetics Steady state kinetics were performed on an Agilent 8453 diode array spectrophotometer (Palo Alto, CA). Each continuous assay was performed in 10 mm PBS at ph 7.4 at room temperature. K i values was calculated by measuring the initial velocity in the presence of a constant concentration of enzyme (3 nm) and increasing concentrations of the inhibitors (ranging from 50 nm to 500 M depending on compound) competed against the indicator substrate nitrocefin (CF) (BD Biosciences, San Jose, CA) (Δε482 = 17,400 M -1 cm -1 ). 130

133 Due to time-dependent inhibition of the chiral boronic acid derivatives, compounds 4 and 5 were preincubated with enzyme for 5 min in PBS before initiating the reaction with the addition of substrate, as described previously (97, 320, 321, 465, 492). In earlier experiments, preincubation of the achiral compound 3 with enzyme did not affect the K i determination (data not shown). Direct competition assays between CF and ADC were used in the same manner to measure the inhibitory activity of the carbapenems. However, in contrast to the boronates, which are reversible inhibitors of class A and C β-lactamses, enzyme acylation by carbapenems may be irreversible within the timeframe of the assay (23, 110). As a true K i determinations depend on reversible equilibria, we refer to this steady state parameter of the carbapenems as K i app (493). To calculate K i or K i app, the inhibition data were plotted as inhibitor concentration versus observed velocity/velocity without inhibitor (v/v 0 ) and fit to the following hyperbolic equation using rigin 7.5 (riginlab, orthampton, MA): y = C / [C + (x R)] Here, C is a constant representing (K m + [S]) for the substrate. A hyperbolic fit of the data yields R, the ratio of K m to K i (or K i app ). The K i (or K i app ) for each inhibitor was then obtained using the K m value of 50 M for CF (methodology described in ref (212)). Electrospray Ionization (ESI) Mass Spectrometry (MS) Mass spectrometry was performed to detect covalent intermediates in the inactivation pathway. We incubated 14 μm of ADC for 15 min with and without each carbapenem and compounds 2 and 5 at an inhibitor: enzyme ratio of 20:1. Each reaction was 131

134 terminated by the addition of 0.1% trifluoroacetic acid and immediately desalted and concentrated using a C 18 ZipTip (Millipore, Bedford, MA) according to the manufacturer s protocol. Samples were then placed on ice and analyzed within 1 hr. Spectra of the ADC-inhibitor proteins were generated on a Q-STAR XL Quadrupole- Time-of-Flight mass spectrometer (Applied Biosystems, Framingham, MA) equipped with a nanospray source. Experiments were performed by diluting the protein sample with 50% acetonitrile/0.1% trifluoroacetic acid to a concentration of 10 μm. This protein solution was then infused at a rate of 0.5 μl/min and the data were collected for 2 min. Spectra were deconvoluted using the Analyst program (Applied Biosystems). All measurements have an error of ± 3 amu. Molecular Representations The ADC model was generated by the SWISS-MDEL automated protein structure homology-modeling server, available at from the deposited GenBank ADC-7 protein sequence AY and the Enterobacter aerogenes CMY-10 β-lactamase template (Protein Data Bank entry 1ZKJ) (11, 241). We optimized the generated model by energy minimization using Discovery Studio 2.1 software (Accelrys, San Diego, CA, see Appendix A for more information on modeling terms). The minimization was performed in several steps, using Steepest Descent and Conjugate Gradient algorithms to reach the minimum convergence (0.02 kcal/mol*å). The protein was immersed in a water box, 7 Å from any face of the box, and the solvation model used was with periodic boundary conditions (PBC). The force-field parameters of CHARMm were used for minimization and the Particle Mesh Ewald method was used to treat longrange electrostatics. The bonds that involved hydrogen atoms were constrained with the 132

135 SHAKE algorithm. Following equilibration, two separate 2 fs molecular dynamics simulations (Heating/Cooling and Production) at constant pressure and temperature (300 ºK) were carried out for the ADC model. The trajectories were analyzed, and the minimum energy conformation was chosen. To verify the quality of the ADC β-lactamase model, we used the Protein Structure and Model Assessment Tools available at The atomic empirical mean force potential (ALEA) evaluation of the model s packing quality showed that 98% of the amino acids were in the favorable energy environment (304). We validated the stereochemical quality of the ADC model using the Procheck program which compares the geometry of protein residues with the stereochemical parameters of wellrefined, high-resolution structures (258). Additionally, 97.5% of the non-proline, nonglycine residues in the ADC model were in the most favorable region of a Ramachandran Plot. The Align Multiple Sequences function of Discovery Studio 2.1 allowed us to compare the generated ADC protein structure with that of the deposited crystal structure coordinates for the Enterobacter cloacae P99 enzyme (PDB entry 1XX2) and E. coli AmpC (PDB entry 2BLS). The program uses a method based on CLUSTAL W which aligns multiple sequences using a progressive pair-wise alignment algorithm (462). A multiple sequence alignment is generated, and secondary structure matches are graded as identical, strong, weak or non-matching are based on the calculated alignment score. The minimized and equilibrated ADC model was used for constructing the acylation complexes of the ADC β-lactamase and the cephalothin chiral analog 5, imipenem, and meropenem ligands. The ligand structures were built using Discovery Studio Fragment Builder tools. The CHARMm force field was applied; the molecule was solvated with 133

136 PBC and minimized using a Standard Dynamics Cascade protocol (one minimization using Steepest Descent algorithm, followed by Adopted Basis ewton-raphson algorithm and three subsequent dynamics stages at VT and 300 ºK). The minimized ligands were docked in the active site of the enzyme using LibDock (124). The generated conformations (30-40) were manually analyzed and the most favorable ones chosen. The complex between the ligand and the enzyme was created, solvated, and energy minimized. The acyl-enzyme complex was created by making a bond with Ser64 and the assembly was further minimized using Conjugate Gradient algorithm with PBC to minimum derivative. To reach the minimum equilibrium, the complexes were equilibrated using Molecular Dynamic Simulations. Results Kinetics Table 3-1 summarizes our kinetic analysis of the inhibition of ADC β-lactamase. To establish a comparison, we list the previously reported K i app s of the commercially available β-lactamase inhibitors against ADC; these K i s are not in the range that would translate into effective inhibition in MIC testing (212). In contrast, we found that the boronic acid derivatives containing the R 1 side chain of cephalosporins bind the class C ADC with K i s in the nm range. The cefoperazone analog and ceftazidime analog, compounds 1 and 2, respectively, show 1 μm K i s. Compound 5 with the cephalothin R 1 side chain and the meta-carboxyphenyl ring, which has a carboxylate that is designed to mimic the geometry and distances of the conserved C 4 carboxylate of cephalosporin β-lactams (Figure 3-3), had the lowest K i for ADC (11 ± 1 nm). We interpret the 70-fold difference in K i between compounds 3 and 5 to mean that 134

137 the meta-carboxyphenyl moiety contributes significantly to the binding of this inhibitor with the ADC β-lactamase. However, compound 4, which lacks only the metacarboxylate as compared to 5, also had a low K i of 36 ± 8 nm. Because inhibition reactions with the boronic acid derivatives are reversible, we quantified the binding energy contribution of these substituents by using K i as an equilibrium constant in the Gibbs free energy equation (320): G = -RT ln [(K i 5) / (K i 3)] Compared to the achiral cephalothin analog 3, we determined that the metacarboxyphenyl group on the chiral cephalothin analog 5 contributes 2.5 kcal/mol in binding energy to ADC. The presence of the meta-carboxylate provides 0.92 kcal/mol of the 2.5 kcal/mol provided by this substituent: G = -RT ln [(K i 5) / (K i 4)] Based on these calculations, we maintain that the presence of the phenyl group, which approximates the cephalosporin s dihydrothiazine ring, is largely responsible for the low K i s of compounds 4 and 5. As carbapenems are highly effective β-lactam antibiotics in the treatment of Gramnegative bacteria possessing class C β-lactamases, and act as inhibitors of these enzymes, we chose the four commercially available carbapenems to explore the determinants that contribute to the inactivation of ADC β-lactamase. Carbapenems, as ADC AmpC inhibitors, demonstrated high affinities (K i app s = 1.23 ± 0.05 to 24 ± 7 μm). Comparing the carbapenems with different R 2 side chains, we see that the least substituted, imipenem, has the lowest K i app. The penem scaffold on which these β-lactams are 135

138 constructed is similar; thus we assign the differences in K i app among these carbapenems to the interactions of the β-lactamase with the R 2 side chain. ESI-MS and the ature of Inactivation Products We performed timed ESI-MS with ADC, the carbapenems, and compounds 2 and 5 to detect covalent intermediates in the inactivation pathway. As shown in Table 3-2 and Figure 3-4, analysis of the ADC-2 and ADC-5 reactions using ESI-MS shows that the β lactamase is unmodified. This result is expected as boronates undergo reversible inhibition. In contrast, when ADC was reacted with the carbapenems, the predominant mass adduct formed corresponded to the sum of the molecular weights of the enzyme and the inhibitor, suggesting the formation of a non-fragmented covalent acyl-enzyme product. This result is consistent with MS data of the carbapenems forming acyl-enzymes with the class A SHV-1 β-lactamase (340). In addition, the ESI-MS analysis of each ADC-carbapenem spectra included a small adduct which was the mass equivalent of the carbapenem and β-lactamase minus 43 ± 3 amu, an observation made previously (206). We advance that there is a retroaldol elimination of the ligand s C 6 hydroxyethyl substituent (see Figure 3-5 and discussion below). Susceptibility Testing anomolar K i inhibitors are clinically useful only if they can penetrate the outer cell wall of Gram-negative organisms and restore susceptibility to partner β-lactams. To this end, we performed MIC testing using compounds 3 and 5. ur results show that when ADC β-lactamase is expressed in the uniform E. coli DH10B background, the cephalothin 136

139 analogs 3 and 5 lower MICs to ceftazidime and cefotaxime (16 to 8 and 4 μg/ml, and 8 to 2 and 1 μg/ml, respectively) (Table 3-3). ur previous data shows that E. coli DH10B harboring bla ADC have MICs of 0.06 μg/ml to meropenem, ertapenem, and imipenem-cilistatin (212). Thus, we did not perform MIC testing with the carbapenems in combination with the cephalothin analogs as a reduction in susceptibility would be difficult to detect. Molecular Representations To understand the interactions between the carbapenems and high affinity cephalothin analog 5 in the absence of a crystal structure, we constructed a molecular model of ADC β-lactamase from a homology-modeling server. Accurate high resolution protein models can be generated from templates with greater than 50% sequence similarity; our model shared 66% sequence similarity with the template (249). The molecular representations of ADC in complex with compound 5, imipenem, and meropenem were useful for informing our kinetic and ESI-MS data. However, as we develop hypotheses from these generated structures, we remain cognizant of modeling limitations, such as the lack of active site flexibility and the removal of water molecules during the ligand docking protocol. We first compared our ADC model to the defined crystal structures of E. cloacae P99 and the E. coli AmpC. An alignment based on the predicted, and known, secondary structures of ADC, P99, and E. coli AmpC shows that ADC shares 63% sequence similarity with both the P99 and E.coli AmpC β-lactamases (37% and 40% amino acid identity, respectively) (Figure 3-6). 137

140 Using the representation of the ADC-5 acyl-enzyme, we gained insight into how the Acinetobacter cephalosporinase interacts with compound 5. As the crystal structure of the E. coli AmpC in complex with the same boronic acid derivative has been solved (K i = 1 nm, PDB entry 1MX), we overlaid this structure on our generated model (Figure 3-7) (320). The overall tertiary structures of ADC and the E. coli AmpC are similar with conservation of the α-helix and -sheet domains. The loops and turns between these secondary structures follow slightly different paths, but we note that these deviations may be part of the model construction and are allowable (e.g., the permission of increased flexibility for these strand regions). In the active site, the distance between the backbone amides of Ser64 and Ser318, which form the -lactamase oxyanion hole or electrophilic center, are ~ 1 Å further apart in the ADC model than the E. coli AmpC structure (325, 478). The position of the backbones and side chains of the catalytically important Tyr150 and Lys 67 also varies by approximately 2 Å between the enzymes (96, 165). Thus, our model suggests that ADC may harbor a unique binding region as compared to the E. coli AmpC. These tertiary features are reflected in the disposition of 5 in the ADC acyl-enzyme model; in Figure 3-8, we show that the inhibitor may adopt different conformations in these class C - lactamases. Comparing equivalent atoms of the boronic acid derivatives (e.g., the metacarboxylate carbons or thiophene sulfur atoms) reveals a greater than 5 Å deviation in the configuration of the compounds in the overlaid structures. Crystal structures of boronic acid derivatives with AmpC enzymes typically show that one boronic acid oxygen atom is placed in the oxyanion hole formed by residues 64 and 318, and the other oxygen atom hydrogen bonds with Tyr150 (82, 96, 320, 400). In the ADC-5 model, one of the boronic 138

141 acid hydroxyl group interacts with the Ser64 backbone carbonyl oxygen, but as the boronic acid and chiral substituents on the inhibitor are rotated approximately 120º compared to the E. coli -lactamase structure, both oxygens are approximately 5 Å from Ser318 or Tyr150. Instead, Tyr150 is within ~ 3 Å of both the carbonyl oxygen and thiophene ring sulfur atom from the cephalothin R 1 group of the boronic acid derivative. ur model also shows a hydrogen bond between this R 1 carbonyl oxygen and Lys67. Thus, the residues contributing to the low K i of this inhibitor for these two β-lactamases may play different roles in each AmpC. In our kinetic studies, we noted 15-fold differences between the K i app values of the highest K i app carbapenem (meropenem) and the lowest K i app carbapenem (imipenem). As these compounds differ primarily by their R 2 side chains, we created models of the ADCimipenem and ADC-meropenem acyl-enzymes to explore the contributions of these substituents (Figure 3-9). Based on the MS results indicating the presence of species corresponding to the acyl-enzyme both with and without the C 6 hydroxyethyl group of the carbapenems, we constructed representations of intact imipenem and meropenem as well as these compounds without their C 6 hydroxyethyl group. ur models predict that when the hydroxyethyl group is present, the carbonyl oxygen from the β-lactam ring of both imipenem and meropenem is located outside of the enzyme s electrophilic, or oxyanion, hole created by the backbone nitrogen atoms of Ser64 and Ser318 (281, 516). Rather, the imipenem β-lactam carbonyl is hydrogen bonded to Lys67 and the meropenem β-lactam carbonyl is only 1.5 Å from Tyr150. The R 2 side chain for both intact carbapenems is oriented out of the active site in the acyl-enzyme. In contrast, when the C 6 hydroxyethyl group is removed, the conformation of the carbapenem is significantly changed. Most notably, the β-lactam carbonyl rotates towards 139

142 the oxyanion hole, approximately 90º in imipenem, and entirely into the electrophilic pocket for meropenem. Further, the R 2 side chain of imipenem flips back towards the active site so that the terminal amide group is ~ 11 Å from its position in the complex with the C 6 group. The imipenem R 2 group now interacts with Tyr150, Asn152, Lys67, and Gln120. Modeling without the hydroxyethyl group for meropenem changes the conformation of the R 2 group, but it remains oriented outside of the binding site. For both imipenem and meropenem, the model without the C 6 hydroxyethyl group is more energetically favorable, as calculated by the final potential energy of the complexes ( and - 80 kcal/mol, respectively). Discussion ur analysis shows that high affinity inhibition of the ADC β-lactamase, a class C cephalosporinase of increasing medical importance, is a realistic goal. We assayed two types of inhibitors against ADC: (i) compounds that resemble the natural substrate for the Acinetobacter cephalosporinase, and (ii) the currently available carbapenems. This approach teaches us important lessons about the inhibition of this challenging β- lactamase and elucidates the important contribution of R 1 and R 2 side chains. We begin with an examination of the data revealing the low M K i app s of the ADC β-lactamase by the carbapenems, and then discuss how the boronic acid derivatives, as chemical probes, yield important insights into the nature of class C enzyme active sites. After incubation of ADC and each carbapenem, our ESI-MS data reveals the formation of two molecular species. The mass of the predominant species corresponds to the intact carbapenem acylating ADC; the mass of the minor species corresponds to the acylenzyme less 43 amu (Table 3-2). Based on previous MS studies in our and other 140

143 laboratories, we assign the major peak to the carbapenem acyl-enzyme species (206, 340). Formation of a stable acyl-enzyme is supported by previously defined crystal structures of carbapenems and class A and C β-lactamases (21, 302, 340). ur ADC-imipenem and ADC-meropenem acyl-enzyme models show conformations where the β-lactam carbonyl oxygen is not found in the oxyanion hole formed by the backbone nitrogens of residues Ser64 and Ser318. This observation is consistent with the X-ray crystal structure of the E. coli AmpC -lactamase with imipenem where the carbonyl was positioned approximately 180º outside of the oxyanion hole (21). Additional crystal structures of class A β- lactamases in complex with carbapenems have also shown this repositioning of the - lactam carbonyl (302, 340). This displacement is likely precipitated by steric interactions induced by the carbapenems C 6 hydroxyethyl groups, producing a conformational change that forces the carbonyl away from the oxyanion hole and into a position unfavorable for hydrolysis (302, 325, 478). Taken together, this reasoning offers an explanation for the inhibition of the Acinetobacter cephalosporinase by the carbapenems. Secondly, we observed a minor peak in each ADC-carbapenem spectrum that reflects the elimination of the carbapenem C 6 hydroxyethyl group. This observation was reported previously by Hugonnet and colleagues in the class A Mycobacterium tuberculosis blac (206). ur molecular representations of imipenem and meropenem in complex with ADC give us insights into how the carbapenems are behaving in the active site of ADC following this elimination. When the C 6 hydroxyethyl group is removed from the carbapenem, both compounds adopt new positions where the β-lactam carbonyl moves to be either entirely in the oxyanion hole (meropenem) or rotated back toward the hole approximately 90º (imipenem). This prediction is similar to the conclusion drawn from the X-ray crystallographic evidence of the class A Asn132Ala TEM enzyme variant 141

144 which demonstrated that the substitution allowed the β-lactam carbonyl to rotate back into the oxyanion hole (493). We speculate that removal of the C 6 group is an alternative mechanism of alleviating the steric clashes induced by this substituent, allowing repositioning of the acyl-enzyme. When the β-lactam carbonyl is aligned in the oxyanion hole, in a conformation more compatible with hydrolysis, increased turnover is likely, consistent with the relatively small amount of this species evident on ESI-MS. Furthermore, we observe this process after a 15 min incubation, which is within the bacterial generation time (i.e., 20 minutes) and suggests that this elimination is occurring in cells. The mechanism of elimination of the hydroxyethyl group is likely to be a retroaldolictype reaction of the -hydroxyethyl moiety of the β-lactamase acyl-enyzme (Figure 3-5). We propose that Glu272, supported by Lys315, may serve as the base to deprotonate directly the alcoholic function of the C 6 substituent. Alternatively, Glu272 may abstract a proton from the amine side chain of Lys315, and then Lys315 would then deprotonate the carbapenem C 6 alcohol group. In either case, the incipient negative charge on the - lactam carbonyl could be stabilized by Tyr150 and Lys67. Interestingly, this proposes another role for Tyr150, which is already implicated in both acylation and substrateactivated catalysis in AmpC enzymes (60, 96, 165, 343). This mechanism is consistent with our molecular representation which shows Glu272 in hydrogen bond distance of the C 6 hydroxyethyl group for both imipenem and meropenem, and the Lys315, Tyr150, and Lys67 residues are also positioned to support this reaction (see Figure 3-9B). Further investigation of these residues and their potential roles in the retroaldolic-type reaction will help elucidate the C 6 elimination mechanism. 142

145 Structural and kinetic studies provide evidence that, following acylation of carbapenems by β-lactamases, the acyl-enzyme formed can tautomerize between a Δ 2 - and Δ 1 -pyrroline species which have differing rates of deacylation (92, 139, 233, 456, 513). We modeled both the tautomers in our molecular representations with ADC, but the interactions between the enzyme and the carbapenems were not significantly different for either the Δ 1 - and Δ 2 -species. We anticipate that the Δ 2 -Δ 1 tautomerization exists as both a separate and integrated pathway to C 6 hydroxyethyl group elimination, and plan further examination of the reaction and its implications for inhibition. We next turn our attention to the contribution of the R 2 side chain in the differing K i apps of the carbapenems for the ADC β-lactamase. The four carbapenems tested share a common -lactam ring scaffold and vary by their R 2 substituents, yet we observed up to 15-fold differences in K i app values. ur models offer insights for how these side groups interact with the ADC enzyme, and suggest that each carbapenem may behave uniquely. In both the imipenem and meropenem models including the C 6 hydroxyethyl group, the R 2 side chain is oriented out of the active site, and does not engage in significant interactions with the enzyme. This outward conformation is also seen in the E. coli AmpC-imipenem, TEM-imipenem, and SHV-meropenem crystal structures (PDB entries 1LL5, 1BT5, 2ZD8, respectively) (21, 302, 340). However, upon removal of the C 6 hydroxyethyl group, the R 2 side chain of imipenem rotates towards the binding pocket and interacts with several active site residues including Tyr150, Asn152, Lys67 and Gln120. In contrast, the R 2 group of meropenem modeled without the C 6 hydroxyethyl hydroxyethyl group is still positioned away from active site residues. In light of the K i app measurements for these two carbapenems, the molecular representations predict that the interactions between the R 2 of imipenem and ADC may stabilize this form of the 143

146 compound in the active site, while for meropenem, the interactions with the enzyme are less favorable (e.g., electrostatically or sterically). otably, the R 2 group of imipenem is the least substituted of the carbapenems, and the additional atoms and ring structures of meropenem, doripenem, and ertapenem may lower the K i app s by various mechanisms, e.g., limiting conformational flexibility necessary for rotating back towards the active site to make favorable interactions. We note that the R 2 side chain of imipenem in the crystal structure with TEM Asn132Ala remains oriented out of the active site, despite the alleviation of the steric strain caused by the C 6 substituent (PDB entry 1JVJ) (493). That our model of imipenem without the C 6 hydroxyethyl group leads to significant R 2 conformational change may reflect inherent differences between the inhibition of class A and C β-lactamases by carbapenems, perhaps partly due to the increased size of the active site in class C enzymes (222). We now highlight the versatile β-lactamase inhibitory activity of the rationally designed boronic acid derivatives (23, 110). All the cephalosporin analogs had K i values 1 M, and likely these values reflect the naturally high affinity of AmpC enzymes for cephalosporin substrates (155). Interestingly, compounds 1-3 which contain only the cephalosporin R 1 side chains, have very similar K i s, suggesting limited differences in the affinity gains of these R 1 structures. The kinetic substrate profile of ADC, and AmpCs in general, includes increasing affinities for the larger side chains of third-generation cephalosporins, which would include ceftazidime and cefoperazone (155, 212, 356). First-generation cephalosporins, such as cephalothin, typically have lower affinities, but higher hydrolytic rates. Thus, despite these differences in affinity and hydrolytic rates, the first- and third-generation cephalosporin analogs have comparable K i s as inhibitors. 144

147 The introduction of chirality and the substituent resembling the dihydrothiazine ring of the cephalosporin nucleus leads to 22-fold increase in K i (compound 3 vs 4). The further addition of the C 4 carboxylate gains 70-fold over the achiral counterpart (compound 3 vs 5). Thus, compound 5, which incorporates these multiple structural features of the cephalosporins, displays the lowest K i for the ADC enzyme. ur results indicating the contribution of the meta-carboxyphenyl ring on the K i of compound 5 are consistent with previous data demonstrating the importance of this functional group for molecular recognition in class C AmpC β-lactamases (320, 321). The crystal structure of compound 5 in complex with the E. coli AmpC β-lactamase shows a hydrogen bond between the carboxylate of the inhibitor and the amide of Asn289 (320). This interaction is wellstudied and thermodynamic cycle experiments revealed that the hydrogen bond contributes 1.7 kcal/mol to the overall binding affinity, a value within the range for iondipole interaction (419). However, Asn289 is not a well conserved residue among class C β-lactamases. The K i of this cephalothin analog increases from 1 nm with the E.coli AmpC to 29 nm with the E. cloacae P99 AmpC which has a Ser289, suggesting that the amino acid at position 289 plays a role for the affinity of compound 5 (281, 419). Based on amino acid sequence, the Acinetobacter ADC cephalosporinase is not closely related to other AmpCs, and has a Glu residue at position 289 (Figure 3-6). Glu is not a hydrogen bond donor, and our model shows the Glu289 side chain is well outside hydrogen bond distance (~ 7 Å). Thus, we used our molecular representation of the ADC- 5 complex to search for other residues that could be hydrogen bonding with the metacarboxylate. Previous structural and functional studies of AmpC enzymes suggest that the more conserved sites Xaa343, Asn346, Arg349 or Thr316 interact with the C 3 /C 4 carboxylate of the substrate, although none of these residues were involved with the 145

148 meta-carboxylate in the E. coli-5 crystal complex (21, 22, 137, 281, 320, 364, 400, 516). Similarly, each of these residues is at least 5 Å from the meta-carboxylate in our ADC-5 representation, an unlikely distance for a high energy hydrogen bond with the group. Asn287 is an ancillary ADC residue which may be capable of engaging the metacarboxylate in a hydrogen bond with ion-dipole character, but is positioned approximately 8 Å away in our model (Figure 3-8). The molecular explanations of the ADC inhibition by the cephalothin analogs 4 and 5 remain to be validated by further study with boronic acid derivatives as molecular probes, site-directed mutagenesis of ADC, and/or crystallography. AmpC enzyme binding site hot spots were previously identified by a comparison of crystal structures in complex with both boronic acid inhibitors and -lactam substrates (400). ur ADC acyl-enzyme representation reveals that the recognition elements may differ for the ADC -lactamase. For example, the E. coli AmpC hydroxyl binding site was defined by Tyr150 and its hydrogen bond with one of the boronic acid hydroxyls, displacing the deacylation water (96, 320). ur ADC-5 model shows significant repositioning of the boronic acid group, making this interaction with Tyr150 unlikely. Rather, the ADC Tyr150 is within ~ 3 Å of the thiophene sulfur and carbonyl oxygen found in the R 1 side chain of compound 5. Further, the R 1 amide recognition site formed by the interaction of the R 1 carbonyl and Asn152 in the E. coli AmpC differs from our ADC-5 representation, as this same side chain carbonyl hydrogen bonds with Lys67 (400). These consensus binding sites were compiled exclusively from structures of the E. coli AmpC, and our data indicate the ADC -lactamase may interact very differently with boronic acid inhibitors than other class C enzymes (320, 400, 419). 146

149 That the chiral cephalothin analog 5 can maintain low nm K i for several phylogenetically divergent AmpC -lactamases reflects not only the potency of this inhibitor, but also what may be an important plasticity of AmpCs (212, 400, 419). The significant repositioning of the boronic acid derivative revealed in our ADC-5 model may be an indication of this enzyme s versatility, causing the β-lactamase-ligand interactions to have different molecular correlates. We posit that compounds 4 and 5 benefit from the presence of an additional side chain which more closely resembles the dihydrothiazine ring of the natural substrate, cephalosporins. The stereochemistry and conformation of the chiral inhibitors may create a better fit for this enzyme. In part, this improved fit may be due to approximation of the deacylation transition state of the cephalosporinase, a theory which has been previously offered to explain the low K i s of chiral boronic acid derivatives (96, 492). Hence, the notion of dedicated AmpC enzyme R 1 and R 2 binding sites may be especially fluid and adaptable in ADC, permitting the β-lactamase to change recognition elements depending on the ligand (222, 400). These novel interactions may reflect that differences in primary sequence can be compensated by common secondary and tertiary structures, allowing the enzyme to use multiple ancillary residues to make contact with substrates and inhibitors. Alternatively, ADC may have subtle differences in its deacylation mechanism which is suggested by the unanticipated position of the boronic acid oxygen atoms in our model. This AmpC structure-function redundancy merits further study, as it could both lie at the core of why these β-lactamases have evolved as versatile traps of cephalosporin substrates, but also aid the careful design of broad-spectrum inhibitors (424, 460). 147

150 Despite the description of boronates as β-lactamase inhibitors since the 1970 s, boronic acid derivatives have not yet been developed for clinical use in combination with a β- lactam (240). Concerns about the safety and efficacy of boron-containing therapeutics are currently being addressed by clinical studies sponsored by the pharmaceutical industry (17). The data presented in this paper encourages the in vivo study of boronates as β- lactamase inhibitors. Conclusion In summary, we provide important insights into the interaction of two types of inhibitors with the Acinetobacter and other clinically relevant cephalosporinases. Firstly, we present kinetic data and molecular representations that explain why carbapenems are effective inhibitors of class C enzymes, including formation of a stable acyl-enzyme and a role for the compounds R 2 side groups. ur results add to increasing evidence supporting the activity of carbapenems as broad-spectrum β-lactam antibiotics, slow substrates, and as inactivators of class A and C β-lactamases (21, 155, 294, 302, 317, 340, 456). Secondly, our ADC model suggests that inhibitors designed to mimic the structure of natural substrates (i.e., boronic acid derivatives) may adopt unique conformations in different class C active sites. Despite significant sequence and structure dissimilarity between ADC and the E. coli AmpC, the chiral cephalothin analogs attain similar K i s for both enzymes. This versatility may reflect an important plasticity of this cephalosporinase β-lactamase. ur data offers promise for the development of compounds that have an extended inhibition profile across, and within, β-lactamase classes, and specifically against this challenging Acinetobacter spp. target. 148

151 Table 3-1. K i and K i apps s of inhibitors in direct competition assays with ADC Inhibitor K i (μm) Commercially available class A inhibitors Clavulanate Sulbactam Tazobactam 4,275 ± 253 a 109 ± 3 a 91 ± 21 a Boronic acid derivatives Compound ± 0.06 Compound ± 0.01 Compound ± 0.02 Compound ± Compound ± Carbapenems b Imipenem 1.23 ± 0.05 Ertapenem 5.71 ± 0.33 Doripenem 14.5 ± 1.0 Meropenem 24 ± 7 a Values obtained previously in ref (212), and represent K i app. b K i app 149

152 Table 3-2. ESI-MS analysis (amu) of ADC alone and incubated with inhibitors a Predicted molecular weight of β-lactamase or inhibitor Species observed in deconvoluted spectra difference from β- lactamase molecular weight ADC alone 40,631 40,638 7 ADC + Inhibitor Compound ,637 1 Compound ,638 0 Imipenem ,936 40,893 Ertapenem ,112 41,069 Doripenem ,058 41,015 Meropenem ,021 40, a All measurements have an error of ± 3 amu. 150

153 Table 3-3. MIC values (μg/ml) of ceftazidime and ceftazidime in combination with 4 μg/ml of boronic acid cephalothin analogs E. coli DH10B E. coli DH10B bla ADC Ceftazidime 1 16 Ceftazidime/ Compound Ceftazidime/ Compound Cefotaxime Cefotaxime/ Compound Cefotaxime/ Compound

154 Figure 3-1. Schemes illustrating the interactions of a serine β-lactamase with: (A) the β- lactam cephalosporin ceftazidime; (B) the boronic acid ceftazidime analog, compound 2; and (C) the carbapenem imipenem. (A) H 3 S Ser 64 - H S R C (B) H 3 S Ser 64 - H B H H (C) H H Ser 64 - H S C H H H 3 S H Ser 64 S R C H 3 S H B H H H H Ser 64 H S C H H Ser 64 Reaction proceeds to hydrolysis Reversible Formation of stable acyl-enzyme 152

155 Figure 3-2. Chemical structures of: (A) commercially available inhibitors and cephalosporin substrate cephalothin; (B) boronic acid derivatives; and (C) carbapenems used in Chapter 3. Cephalothin structure is labeled with accepted ring numbering system. The C 6 hydroxyethyl group of imipenem, which may be eliminated after formation of the acyl-enzyme, is circled in dashed lines. (A) H CH H H S CH H S CH S R 1 H H S CH Clavulanate Tazobactam Sulbactam Cephalothin (B) H H H B H H Cefoperazone analog, compound 1 H 2 H S H B H H Ceftazidime analog, compound 2 S H B H H Achiral cephalothin analog, compound 3 S H B H H Chiral phenyl cephalothin analog, compound 4 S H B H H H Chiral meta-carboxyphenyl cephalothin analog, compound 5 (C) H H H 1 S CH R 2 H H H H H S H CH H H H H H H S CH H H S H 2 H H H H S CH H H Imipenem Ertapenem Doripenem Meropenem 153

156 Figure 3-3. verlay of the molecular coordinates for the E. coli AmpC covalently bound to cephalothin substrate (colored by atom, PDB entry 1KVM) and boronic acid chiral cephalothin analog, compound 5 (colored green, PDB entry 1MX). The position of cephalothin s dihydrothiazine ring and C 4 carboxylate is shown relative to the metacarboxyphenyl group of compound 5, which is designed to mimic in stereochemistry and geometry the conserved -lactam carboxylate. 154

157 Relative intensity Relative intensity Figure 3-4. Deconvoluted mass spectra of: (A) ADC β lactamase alone; (B) ADC after 15 min incubation with compounds 2 and 5; and (C) ADC β-lactamase after 15 min incubation with imipenem, ertapenem, doripenem, and meropenem. The peak in each ADC-boronate spectrum corresponds to the unmodified ADC enzyme. The major peak in each of the ADC-carbapenem spectrum indicates covalent attachment of the β-lactam with a minor additional peak corresponding to the acyl-enzyme without the carbapenem s C 6 hydroxyethyl substituent. All measurements have an error of ± 3 amu. (A) ADC alone 40, 638 Mass (amu) (B) ADC + compound 2 40, 637 ADC + compound 5 40, 638 Mass (amu) 155

158 Relative intensity Relative intensity (C) ADC + imipenem 40, 936 ADC + ertapenem 41, ADC + doripenem 41, ADC + meropenem 41, 021 Mass (amu) 156

159 Figure 3-5. Proposed mechanism of the retroaldolic reaction leading to elimination of C 6 hydroxethyl substituent from the -lactamase-carbapenem acyl-enzyme. Glu272, supported by Lys315, may serve as the base to deprotonate the alcoholic function hydroxy carbonyl moiety of the C 6 substituent. Alternatively, Glu272 may abstract a proton from Lys315 which subsequently deprotonates the C 6 group (mechanism shown in dashed bold arrows). The incipient negative charge on the -lactam carbonyl could be supported by Tyr150 and a Lys67. Glu272 Ser 64 - H H H S C H H Tyr150 H H 3 Lys67 Ser 64 H H S C H H 2 Lys315 H H Ser 64 H S C H H H Ser 64 H S C H H 157

160 Figure 3-6. Discovery Studio multiple sequence protein alignment of crystal structure coordinates for E. cloacae P99 (PDB entry 1XX2), E. coli AmpC (PDB entry 2BLS), and the ADC model generated by SWISS-MDEL comparative protein modeling server. The E. coli CMY-2 (PDC entry 1ZC2) and P. aeruginosa PDC-3 AmpCs are also included and will be discussed further in Chapter 4. The active site serine, class C motif Y-X- Tyr150, conserved K-T-G Thr316, and amino acids with potential roles in binding compound 5: 289, 346, and Arg349, are labeled. The conserved Glu272 may serve as a base in the retroaldolic reaction leading to elimination of the carbapenem s C 6 hydroxyethyl group. As calculated by a sequence alignment score, residues highlighted in blue indicate identical secondary structure matches, turquoise: strong, red: weak, and white: non-matching. 158

161 E. coli AmpC E. cloacae P99 A.baumannii ADC P.aeruginosaPDC-3 E. coli CMY-2 Ser64 Xxx79 E. coli AmpC E. cloacae P99 A.baumannii ADC P.aeruginosaPDC-3 E. coli CMY-2 Tyr150 E. coli AmpC E. cloacae P99 A.baumannii ADC P.aeruginosaPDC-3 E. coli CMY-2 E. coli AmpC E. cloacae P99 A. baumannii ADC P.aeruginosaPDC-3 E. coli CMY-2 Glu272 E. coli AmpC E. cloacae P99 A. baumannii ADC P.aeruginosaPDC-3 E. coli CMY-2 Xaa289 Thr316 E. coli AmpC E. cloacae P99 A. baumannii ADC P.aeruginosaPDC-3 E. coli CMY-2 Asn346 Arg349 E. coli AmpC E. cloacae P99 A. baumannii ADC P.aeruginosaPDC-3 E. coli CMY-2 159

162 Figure 3-7. verlay of molecular coordinates for the E. coli AmpC-5 complex in yellow (PDB entry 1MX) and generated ADC-5 model colored by secondary structure. The position of α helices and β sheets is generally preserved between the two proteins, but deviations are observed in the strand turns between these secondary structures. Active site differences are illustrated by the altered conformation of 5 (colored green) in ADC as compared to 5 (colored yellow) bound to the E. coli AmpC. 160

163 Figure 3-8. Comparison of the binding site interactions between E. coli AmpC-5 (left panel) and ADC-5 (right panel). Figures have a perspective view to show positions of the residues in relation to the inhibitor. The boronic acid derivative is bound to Ser64 in both structures, but the relative rotation of the inhibitor in ADC changes the relationships with other active site residues. Specifically, the boronic acid oxygens interact with Ala318 and Tyr150 in E. coli AmpC, but the hydrogen bond with Ser318 is lost in ADC. Also in ADC, the meta-carboxylate of the dihydrothaizine ring analog has no clear interaction with previously identified carboxylate binding residues (e.g, Asn346, Arg349, or like Asn289 in E. coli AmpC). Instead, the group may form a long hydrogen bond with Asn287. The carbonyl oxygen of the cephalothin R 1 side chain interacts with Asn152 in E. coli AmpC but is reoriented toward Lys67 in ADC (see dashed red lines). Lastly, the R 1 thiophene ring sulfur in ADC-5 is moved towards Tyr150 as compared to the E. coli AmpC-5 structure. verall, these significant active site differences suggest that while ADC may possess novel architecture, the ability to recognize inhibitors and substrates is preserved because of the versatile functions of the binding site residues. 161

164 162

165 Figure 3-9. Molecular representation of: (A) ADC-imipenem acyl-enzyme model; and (B) ADC-meropenem acyl-enzyme model. The intact carbapenem is shown in green and the carbapenem without the C 6 hydroxyethyl group in orange. Hydrogens are not shown except on the carbapenem C 6 hydroxyethyl which is likely deprotonated by Glu272, leading to elimination of the group. Removal of this C 6 substituent may lead to reorientation of the compound in the active site. Specifically, the β-lactam carbonyl moves back towards the oxyanion hole formed by the backbone nitrogens of Ser 64 and Ser318, approximately 90º rotation for imipenem and entirely into the hole for meropenem. Also after C 6 elimination, the R 2 group of imipenem is repositioned from outside of the binding pocket into a network of interactions with Tyr150, Asn152, Lys67 and Gln120. Active site interactions are not observed for the R 2 group of meropenem, which may have implications for the differing K i app s of these carbapenems. 163

166 (A) Lys67 Glu272 C 6 hydroxyethyl Lys315 Tyr150 Ser64 Asn152 β-lactam carbonyls without C 6 group Ser318 Gln120 R 2 side chains with C 6 group 164

167 (B) Glu272 Lys67 Ser64 Lys315 β-lactam carbonyls C 6 hydroxyethyl without C 6 group Tyr150 Asn152 R 2 side chains Ser318 with C 6 group Gln

168 CHAPTER 4 Catalytic and Inhibitory Properties of the Pseudomonas aeruginosa AmpC: Implications for an Inhibitor-Resistant Phenotype Introduction Pseudomonas aeruginosa is a Gram-negative pathogen responsible for serious nosocomial infections including pneumonia, blood stream, intra-abdominal, wound, eye, ear, and urinary tract infections. The organism, an opportunistic pathogen, is often isolated from patients with multiple illnesses, in-dwelling catheters, burns, and surgical devices (200). P. aeruginosa is one of the most commonly isolated Gram-negative bacilli from patients in intensive care units in the United States, and the incidence has been rising during the past several decades (164). Mortality rates up to 60 % have been reported for P. aeruginosa infections, particularly in patients with immune compromise or underlying comorbidities (359, 485). ften treatment guidelines advise administration of a β-lactam in combination with a fluoroquinolone or aminoglycoside, but despite this combination approach, outcomes are poor (406). The prevalence of P. aeruginosa strains resistant to three or more antimicrobial agents can range from 3 50 % (175, 457). Clearly, treatment of P. aeruginosa infections is challenging, and the dearth of new agents being developed and released into the clinical market stresses the need to use currently available agents in a judicious manner (412). 166

169 The major antibiotic resistance determinants in P. aeruginosa strains are AmpC β- lactamase production, drug efflux pumps (e.g., MexAB-prM), and impermeability of the outer membrane (275). Production of chromosomal AmpCs mediates resistance to β- lactams, and both genetic mutations and induction from certain β-lactams can significantly increase the expression of these drug-hydrolyzing enzymes (180, 229). Protecting the activity of β-lactams by inhibiting β-lactamases is an important approach for preserving our current antimicrobial armamentarium. However, the commercially available β-lactamase inhibitors are less effective at inactivating AmpC enzymes than class A β-lactamases (69, 238, 319). Despite the clear clinical significance of this resistance determinant, little is known about the structure-function relationships of the Pseuodmonal AmpC for either substrates or inhibitors. A better understanding of the molecular details of catalysis and inhibition of this AmpC can help lead design of more effective inhibitors. The work on the P. aeruginosa AmpC in this chapter comprises a new direction of investigation in our laboratory. Presented here are the initial findings on the susceptibility patterns and the detection of β-lactamase protein expression from a laboratory strain of P. aeruginosa, 18SH (27, 153). Also described is the substrate profile of the purified AmpC from this strain, Pseuodmonas-derived cephalosporinase-3 (PDC-3) (418). ur study further details the inhibition of PDC-3 by a panel of small molecules with varied chemical compositions. In particular, we chose three types of compounds with the hypothesis that each agent could contribute to our understanding of the PDC-3 s inhibition profile. First, we included substrate analogs; boronic acid derivatives bearing the R 1 and R 2 side chains of either penicillins or cephalosporins to detect the molecular preferences of the PDC-3 active site. Second, we tested β-lactams with structural features 167

170 previously shown to inhibit AmpC enzymes (i.e., the carbapenems, PSR-3-283a, and BAL29880) with the premise that the compounds would also inhibit PDC-3. The third compound screened was a novel non-β-lactam inhibitor (XL104) for which there is strong in vivo data against class A β-lactamases, but less known about its ability to inactivate AmpCs (498). The latter part of the chapter attempts to correlate each inhibitor s kinetic data with whole cell assays by examining whether the compounds can decrease MICs for a partner β-lactam. Here, we realized that in vitro β-lactamase inhibitory activity may not predict a compound s ability to pass through the P. aeruginosa outer membrane. The final discussion focuses on MIC data that was obtained from PDC-3 variants made by site-directed mutagenesis and expressed in E. coli DH10B cells. Molecular representations of a boronic acid analog of cephalothin in complex with PDC-3 informed the identification of important ligand recognition residues. We also found that mutagenesis at these sites reveals an unexpected increase in β-lactamase inhibitor susceptibility. Materials and Methods Bacterial Strains Table 4-1 summarizes the bacterial strains used in these studies. The P. aeruginosa isolates 18SH, MK1184, and R were obtained as a kind gift from Dr. M. G. P. Page (Basilea). The AmpC β-lactamase from 18SH is produced at constitutively high levels as its regulation is derepressed, i.e., the expression does not increase with induction by β-lactams (27, 153). 168

171 In addition, we used two comparator strains that do not overproduce the AmpC protein: PA1 (obtained from Dr. H. Schweizer, Colorado State University), and the American Type Culture Collection (ATCC) P. aeruginosa strain (203, 440, 459). The AmpC from the PA1 strain has only one amino acid that is different from the 18SH AmpC, and has been designated PDC-1 in the classification system proposed by Rodríguez-Martínez and colleagues (418). By this nomenclature, the 18SH AmpC is designated PDC-3. P. aeruginosa clinical isolates UL140 and DB322 were obtained from and characterized by Dr. J. Quale (State University of ew York Downstate Medical Center). Both isolates are resistant to ertapenem (403). Based on mra expression and regulatory gene sequencing, the primary mechanisms of the phenotypes are: (i) UL140, changes in efflux pump sytems (e.g., MexAB-prM, MexCD-prJ, MexXY-prM, MexEF-pr); and (ii) DB322, a combination of efflux pump and porin changes and increased bla ampc expression. Two non-pseduomonal strains were used as controls for the immunoblotting: a clinical E. coli expressing a CTX-M β-lactamase and K. pneumoniae ATCC expressing SHV-18. Antibiotics and Inhibitors β-lactam substrates were purchased from Sigma (St. Louis, M). Tazobactam and clavulanate were obtained from Wyeth Pharmaceuticals (Pearl River, Y) and GlaxoSmithKline (Surrey, United Kingdom), respectively. Boronic acid derivatives were kind gifts from Drs. Fabio Prati and Emilia Caselli (82, 320, 492). XL104 was received from Dr. Christine Miossec (ovexel), BAL29880 from Dr. M. G. P. Page, and PSR-3-283a from Dr. John Buynak (Southern Methodist University). Imipenem and ertapenem 169

172 were obtained from Merck & Co. Inc. (Whitehouse Station, J). Meropenem was purchased from AstraZeneca Pharmaceuticals (Wilmington, DE) and doripenem from rtho-mceil Pharmaceutical Inc. (Raritan, J). Figure 4-1 shows the structures of substrates and inhibitors (except for carbapenems, see Figure 3-2). Antimicrobial Susceptibility (MICs) Susceptibility profiles were determined by Mueller-Hinton agar dilution MICs according to the Clinical and Laboratory Standard Institute (CLSI) standards (102). The inhibitory activity of β-lactam/investigational β-lactamase inhibitor combinations was evaluated by broth microdilution per CLSI standards. The inhibitors were tested at a constant concentration of 4 μg/ml in combination with cefotaxime. Cefoxitin Induction ne hundred microliters from overnight cultures of P. aeruginosa PA1 and 18SH grown in Mueller-Hinton broth were added to 5 ml of broth and grown to D 600 = 0.5. The cultures were then incubated with 50 µg/ml of cefoxitin or a sterile Millipore water control at 37 ºC for 3 hr (231). Crude extracts were prepared by resuspending the pelleted cells in 50 mm Tris (ph 7.4) and liberating periplasmic proteins with lysozyme (271). Protein in the crude extract was quantified by spectrometric assay using bovine serum albumin (BSA, Bio-Rad, Hercules, CA) as a protein standard. The velocity of nitrocefin (CF) hydrolysis per µg of protein was calculated from the measured change in absorbance (Δε 482 = 17,400 M -1 cm -1 ). Cloning of bla PDC-3 into pbc SK(-) Vector 170

173 P. aeruginosa bla ampc (bla PDC ) genes are encoded on the bacterial chromosome. In order to perform comparative analyses in a uniform genetic background and aid β-lactamase purification, we cloned the bla ampc gene into the pbc SK(-) vector. either the genetic sequence of the bla ampc nor the AmpC β-lactamase from the 18SH strain has been previously described. Thus, in order to design primers for gene amplification, we examined the known P. aeruginosa bla ampc genes deposited in the ational Center for Biotechnology Information (CBI). These analyses revealed conserved regions immediately upstream and downstream from the bla ampc genes of several P. aeruginosa strains, including PA1 (283). Thus, we designed PCR primers to the conserved regions of the bla ampc gene, including approximately 50 base pairs on either side of the start and stop codons (PA1 upstream and PA1 downstream, primers used for cloning and mutagenesis are listed in Table 4-2). The PCR template was prepared from genomic DA isolated from overnight cultures of the P. aeruginosa 18SH strain. Specifically, a 1:10 dilution of an overnight culture was boiled for 10 min (212). Amplification was then performed with 10 µl of this dilution as the DA template. The PA1 upstream and PA1 downstream primers successfully amplified a 1400-base pair amplicon from the 18SH DA. We used the TP TA cloning system (Invitrogen, Carlsbad, CA) to ligate the 18SH PCR product into the cloning site of the pcr 2.1-TP vector. The ligation products were transformed into Escherichia coli DH10B cells (Invitrogen), and plated on agar containing 50 g/ml kanamycin. Plasmids were extracted from ten single colonies of E. coli DH10B cells and sequenced with M13 Universal and M13 Reverse primers, as well as the PA1 upstream and PA1 downstream primers. Primers were also designed internal to the gene near the region encoding the PA AmpC β-lactamase active site to make sure the sequence data 171

174 covered the entire bla ampc on both strands (PA1 AS forward and PA1 AS reverse) (283). Following sequence verification of the 18SH bla ampc in the pcr 2.1-TP vector, we performed a restriction enzyme digest of the recombinant plasmid using the BamHI and XbaI restriction enzymes (Promega, Madison, WI). The digestion product was purified from agarose gel using the Qiagen QIAquick PCR Purification Kit (Valencia, CA). The purified insert was then ligated to the BamHI- and XbaI-cut pbc SK(-) vector and transformed into E. coli DH10B cells. We plated the transformation mixture on agar containing 20 g/ml chloramphenicol, and isolated plasmids from ten single colonies of E. coli DH10B cells. The M13 Universal, M13 Reverse primers, PA1 upstream, PA1 downstream, PA1 AS forward, and PA1 AS reserve primers were used to verify the sequence of 18SH bla ampc (equivalent to bla PDC-3 ) in pbc SK(-). This bla PDC-3 gene was also directionally subcloned into the pet 24a(+) vector (ovagen, Madison, WI) for large scale protein expression from both E. coli BL21(DE3) and BL21(DE3) RP codon+ cells (Stratragene, La Jolla, CA). We designed PCR amplification primers to introduce dei and BamH1 restriction sites into the beginning of the gene sequence that coded for the mature protein (i.e., the -lactamase leader sequence was not included) and the end of the coding sequence, respectively. PCR mutagenesis was performed on the bla PDC-3 in the pcr 2.1-TP vector, and the successful introduction of the restriction sites confirmed by digesting the resultant plasmids with dei and BamH1. The insert was purified from agarose gel, ligated to a dei- and BamH1-digested plasmid prep of the pet 24a(+) vector, and transformed into E. coli DH10B cells. Six colonies containing the plasmid were selected from agar plates 172

175 containing 50 g/ml chloramphenicol. Plasmids from these colonies were isolated and sequenced. E. coli BL21(DE3) and BL21(DE3) RP codon+ cells were transformed with the bla PDC-3 pet 24a(+) vector using a 42 ºC heat shock according to the manufacturer s protocol. Mutagenesis and Sequencing To replace the bla PDC-3 wild-type amino acid codon with the alanine codon at amino acid positions 343, 346, and 349 (based on sequence alignment with P99, E. coli AmpC, and ADC, see Figure 3-6), we designed primers as listed in Table 4-2. Using the template bla PDC-3 gene in the pbc SK(-) vector, we employed Stratagene s QuikChange Mutagenesis Kit to introduce the alternate codon. For the Arg349Ala variant, mutagenesis was also performed in the pet 24a(+) vector using the same primers. To obtain the gene encoding the β-lactamase from the PA1 strain (equivalent to bla PDC-1 ), we performed site-directed mutagenesis on the bla PDC-3 gene to change one nucleotide (G to A) to produce the Thr codon at amino acid 79. For all mutagenesis reactions, PCR products were electroporated into E. coli DH10B cells and plated on 20 g/ml chloramphenicol or 50 g/ml kanamycin agar (depending on host plasmid). Six single colonies of E. coli DH10B cells were selected for each mutagenesis reaction, the plasmids were isolated, and DA sequencing of the bla PDC genes performed using the M13 Reservse and M13 Universal primers. β-lactamase Purification The PDC-3 β-lactamase was purified from the P. aeruginosa 18SH strain and from E. coli DH10B, E. coli BL21(DE3), and E. coli BL21(DE3) RP codon+ cells. P. aeruginosa 173

176 18SH and E. coli DH10B cells were grown overnight in 500 ml cultures in SB. Expression of the protein in the E. coli BL21(DE3) and BL21(DE3) RP codon+ cells was induced with IPTG. Briefly, 12 ml of an overnight culture grown in LB were used to inoculate 500 ml of SB and grown at 37 ºC for 2 hr. IPTG at a final concentration of 1 mm was then added, and the cultures allowed to grow for an additional 2.5 hr. For all the P. aeruginosa and E. coli cultures, the cells were pelleted, resuspended in 50 mm Tris (ph 7.4), and β-lactamase liberated with lysozyme and EDTA per established methods (210). Accordingly, PDC-3 enzyme was purified by pief and FPLC with a Sephadex Hi Load 16/60 column and a HiTrap High Performance sulfopropyl strong cation exchanger (Pharmacia, Uppsala, Sweden) (367). The protein was quantified by BSA assay and purity was assessed by 5% stacking, 12% resolving SDS-PAGE (212). Finally, the atomic mass was verified by mass spectrometry. Polyclonal Antibody Preparation and Purification Anti-PDC-3 polyclonal rabbit antibodies were produced by ew England Peptide (Gardner, MA) from 3.0 mg of PDC-3 protein purified from P. aeruginosa 18SH. The antibodies were isolated from rabbit sera using a Hi-Trap Protein G column (GE Healthcare) as previously described (207). Briefly, rabbit serum was diluted in 20 mm ah 2 P 4 (ph 7.0) and bound to the column. Antibodies were eluted with 0.1 M glycine- HCl (ph 2.7) and added to 1 M Tris-HCl (ph 9.0) to yield a neutralized storage buffer. The concentration was measured by spectrophotometric determination at 280 nm, and the samples aliquoted and frozen at -20 ºC for long-term storage. Immunoblots 174

177 To determine the sensitivity and specificity of the polyclonal antibody, we performed immunoblots with a range of quantities of purified -lactamase from both the P. aeruginosa 18SH strain and the E. coli DH10B cells. Purified -lactamases from our laboratory were used to screen for specificity; specifically, 500 ng each of class A SHV-1 and KPC-2, class B CcrA, class C CMY-2, P99, and ADC, and class D XA-1 were tested. These control proteins were all purified from E. coli cells containing bla genes encoded on expression vectors. In addition to the purified -lactamases we prepared crude cell lysates of a panel of P. aeruginosa isolates by growing each strain to D 600 = 0.8 and lysing the cells by 10 min incubation at 100 ºC in SDS loading dye buffer. These P. aeruginosa strains included PA1, 18SH, MK1184, and R, and clinical isolates UL140 and DB322. Crude lysates were also prepared for the ATCC strains Klebsiella pneumoniae and P. aeruginosa 27853, and a clinical strain of E. coli which produces SHV-, TEM-, and CTX-M -lactamases. Samples were run on 5% stacking, 12% resolving SDS-polyacrylamide gel and protein transferred to a polyvinylidene difluoride membranes according to previously published protocol (211). The membranes were probed with 1 ug/ml of purified anti-pdc-3 antibody and then incubated with a 1:2,000 dilution of horseradish peroxidase-conjugated protein G (Bio-Rad). Kinetics Steady state kinetics were performed on an Agilent 8453 diode array spectrophotometer (Palo Alto, CA). Each continuous assay was performed in 10 mm phosphate-buffered 175

178 saline at ph 7.4 at room temperature in a quartz cuvette with a 1 cm pathlength. Measurements were obtained using CF (Δε 482 = 17,400 M -1 cm -1 ), cephalothin (Δε 262 = M -1 cm -1 ), and cephaloridine (Δε 260 = -10,200 M -1 cm -1 ). The kinetic parameters, V max and K m, were obtained with non-linear least squares fit of the data (Henri Michaelis-Menten equation) using rigin 7.5 (riginlab, orthampton, MA): v = (V max [S])/(K m + [S]) Direct hydrolysis could not be measured at this time for the substrates ampicillin, cefoxitin, cefoperazone, ceftazidime, and cefotaxime (in assays with up to 5 g of protein), and thus the apparent K i values were obtained in competition assays with CF (142). For the boronic acid inhibitors R, 2a, 6a, 7a, and 8a, K i values were calculated by measuring the initial velocity (0-10 sec) in the presence of a constant concentration of enzyme (3 nm) and increasing concentrations of the inhibitors competed against the indicator substrate CF. Due to time-dependent inhibition of the chiral boronic acid derivatives, compounds 1c, 3c, 4c, 5c, and 9c were preincubated with enzyme for 5 min in phosphate-buffered saline before initiating the reaction with the addition of substrate, as described previously (97, 320, 321, 465, 492). In earlier experiments, preincubation of the achiral compound 2a with enzyme did not affect the K i determination (data not shown). Direct competition assays between CF and PDC-3 were used in the same manner to measure the inhibitory activity of the commercially available inhibitors (tazobactam and clavulanate), the carbapenems (imipenem, meropenem, ertapenem, and doripenem), and 176

179 the investigational inhibitors (XL104, BAL29880, and PSR-3-283a). However, in contrast to the boronates, which are reversible inhibitors of class C enzymes, these other inhibitors may engage in chemical reactions with the β-lactamase that are irreversible within the timeframe of the assay (23, 493). As true K i determinations depend on reversible equilibria, we refer to this steady-state parameter of the carbapenems and the commercially available and investigational inhibitors as K i app. To calculate K i or K i app, the inhibition data were plotted as inhibitor concentration versus observed velocity/velocity without inhibitor (v/v 0 ) and fit to the following hyperbolic equation using rigin 7.5 (riginlab, orthampton, MA): y = C / [C + (x R)] Here, C is a constant representing (K m + [S]) for the substrate. A hyperbolic fit of the data yields R, the ratio of K m to K i (or K i app ). The K i (or K i app ) for each inhibitor was then obtained using the K m value for CF. The first-order rate constant for enzyme and inhibitor complex inactivation, k inact, was obtained by monitoring the reaction time courses in the presence of inhibitor. Fixed concentrations of enzyme (10 nm) and CF (150 μm) and increasing concentrations of inhibitor were used in each assay. The k obs was determined using a non-linear least squares fit of the data using rigin 7.5 : A = {A 0 + v f t + (v 0 -v f )[1-exp (-k obs t)]} / k obs Here, A is absorbance, v 0 (expressed in variation of absorbance per unit time) is initial velocity, v f is final velocity, and t is time. Each k obs was plotted versus I and fit to determine k inact. The value of k inact was used to determine K I, the concentration of inhibitor needed to achieve ½ k inact (107): 177

180 k obs = (k inact [I])/(K I + [I]) The K I data were corrected to account for the affinity of CF for PDC-3 according to the following equations (99): K I (corrected) = K I (observed) / [1 + ([S]/K m CF)] Electrospray Ionization (ESI) Mass Spectrometry (MS) Mass spectrometry was performed to determine covalent reaction intermediates with the boronate compound 4c, BAL29880, and PSR-3-283a. We incubated 14 μm of PDC-3 for 15 min with the meta-carboxyphenyl cephalothin boronate 4c at an inhibitor: enzyme ratio (I:E) of 25:1 and BAL29880 and PSR-3-283a at 20:1. Different expression systems were used to purify the enzymes used for each of these MS studies. Specifically, the PDC-3 protein was purified from the P. aeruginosa strain for the boronate experiment, E. coli with the pbcsk vector for the BAL29880 experiment, and the E. coli BL21(DE3) RP codon+ pet 24a(+) system for the PSR-3-283a experiment. Each reaction was terminated by the addition of 0.1% trifluoroacetic acid and immediately desalted and concentrated using a C 18 ZipTip (Millipore, Bedford, MA) according to the manufacturer s protocol. Samples were then placed on ice and analyzed within 1 hr. Spectra of the intact PDC-3: inhibitor proteins were generated on a Q-STAR XL Quadrupole-Time-of-Flight mass spectrometer (Applied Biosystems, Framingham, MA) equipped with a nanospray source. Experiments were performed by diluting the protein sample with 50% acetonitrile/0.1% trifluoroacetic acid to a concentration of 10 μm. This protein solution was then infused at a rate of 0.5 μl/min and the data were collected for 2 178

181 min. Spectra were deconvoluted using the Analyst program (Applied Biosystems). All measurements have an error of ± 3 amu. Molecular Representations The PDC-3 model was generated by the SWISS-MDEL automated protein structure homology-modeling server, available at (11). The determined PDC-3 protein sequence was entered, and a model generated by the software using the Pseudomonas fluorescens class C β-lactamase template (Protein Data Bank entry 2QZ6) (311). PDC-3 and the P. fluorescens protein share 76% sequence similarity; accurate high resolution protein models can be generated from templates with greater than 50% sequence similarity (249). Ms. Taracila optimized the generated PDC-3 model by energy minimization using Discovery Studio 2.1 software (Accelrys, San Diego, CA, see Appendix A for more information on modeling terms). The minimization was performed in several steps, using Steepest Descent and Conjugate Gradient algorithms to reach the minimum convergence (0.02 kcal/mol*å). The protein was immersed in a water box, 7 Å from any face of the box, and the solvation model used was with periodic boundary conditions (PBC). The force-field parameters of CHARMm were used for minimization and the Particle Mesh Ewald method was used to treat long-range electrostatics. The bonds that involved hydrogen atoms were constrained with the SHAKE algorithm. Following equilibration, two separate 2 fs molecular dynamics simulations (Heating/Cooling and Production) at constant pressure and temperature (300 ºK) were carried out for the PDC-3 model. The trajectories were analyzed, and the minimum energy conformation was chosen. 179

182 The minimized and equilibrated PDC-3 model was used for constructing the acylation complexes of the PDC-3 β-lactamase and the chiral cephalothin analog 4c. The ligand structure was built using Discovery Studio Fragment Builder tools. The CHARMm force field was applied; the molecule was solvated with PBC and minimized using a Standard Dynamics Cascade protocol (one minimization using Steepest Descent algorithm, followed by Adopted Basis ewton-raphson algorithm and three subsequent dynamics stages at VT and 300 ºK). The minimized ligand was docked in the active site of the enzyme using LibDock (124). The generated conformations (30-40 poses) were visually inspected and the most favorable ones chosen based on minimum energy. The complex between the ligand and the enzyme was created, solvated, and energy minimized. The acyl-enzyme complex was created by making a bond with Ser64 and the assembly was further minimized using Conjugate Gradient algorithm with PBC to minimum derivative. To reach the minimum equilibrium, the complexes were equilibrated using Molecular Dynamic Simulations. The Accelrys software was used to create a solvent-accessible surface of the PDC-3-4c model using a probe radius of 1.4 Å. Results β-lactam MICs of P. aeruginosa 18SH and PA1 As expected, susceptibility testing reveals that both P. aeruginosa 18SH and PA1 have high MICs to penicillins and the narrow-spectrum cephalosporin cephalothin (Table 4-3). The P. aeruginosa 18SH MICs are in the resistant range for piperacillin and aztreonam as defined by CLSI criteria (resistance breakpoints 64 and 32 g/ml, respectively) (103). The PDC-3 expressing 18SH also demonstrates elevated MICs for the 180

183 extended-spectrum cephalosporins cefotaxime, ceftazidime, and cefepime ( 64, 32, and 64 g/ml, respectively). In contrast, the MICs of the PA1 PDC-1-expressing strain are in the susceptible range for the extended-spectrum cephalosporins and aztreonam. AmpC Expression and Immunoblotting The production of chromosomal AmpCs in Gram-negative bacteria is usually at a low level ( repressed ), but can be increased by induction with certain β-lactams, particularly cefoxitin and imipenem (14, 26, 180). Here we examined whether the AmpC expression of P. aeruginosa 18SH and PA1 strains could be induced by cefoxitin. After a 3-hr incubation with 50 g/ml of cefoxitin, the specific activity of the crude extract from P. aeruginosa 18SH did not change significantly (0.45 ± 0.02 to 0.47 ± 0.04 M CF hydrolyzed/sec/ g protein), while crude extract from PA1 increased from ± to ± 0.01 M nitrocefin hydrolyzed/sec/ g protein (Figure 4-2). We note that the nitrocefinase activity of the induced PA1 crude protein is still significantly lower than of the induced and non-induced 18SH crude protein. The variations in nitrocefinase activity likely reflect not only derepression of AmpC expression in the 18SH strain, but additional intrinsic differences in β-lactamase expression and background between 18SH and PA1. Figure 4-3A shows an immunoblot testing the sensitivity of the anti-pdc-3 antibody. The membrane has a titration of PDC-3 protein isolated from both the P. aeruginosa 18SH strain and the E. coli pbc SK(-) expression construct. The anti-pdc-3 antibody detects as little as 50 ng of purified PDC-3 β-lactamase. We note that when comparing PDC-3 protein purified from E. coli and P. aeruginosa, the background signal for the P. 181

184 aeruginosa 18SH strain is higher. This is likely due to the increased amount of crossreacting proteins from the crude lysate of the clinical strain, as opposed to the commercial E. coli DH10B expression system. The immunoblot in Figure 4-3B demonstrates the specificity of the anti-pdc-3 antibody; included on the membrane are purified β-lactamase proteins from all Ambler classes and crude lysates of clinical isolates. The anti-pdc-3 antibody recognized purified P99 and CMY-2 β-lactamase, but not the other class C β-lactamase, ADC. The class B (CcrA), A (SHV-1, KPC-2), or D (XA-1) β-lactamases were not detected by the antibody. For the crude lysates, bands are observed in all lanes containing P. aeruginosa or E. coli samples carrying a bla ampc gene. However, crude lysates of an SHV-18 producing K. pneumoniae and a CTX-M producing E. coli are not recognized. bla PDC-3 Sequence Analysis Translation of the bla ampc gene amplified from the chromosomal DA of P. aeruginosa 18SH predicts a β-lactamase that shares 100% identity with the bla ampc protein translation from several CBI deposited P. aeruginosa strains, including LESB58 (CBI Reference Sequence: YP_ ), 2192 (YP_ ), and C3719 (YP_ ). The 18SH AmpC β-lactamase (PDC-3) is different by one amino acid from the PA1 AmpC (PDC-1), a Thr to Ala change at residue 79 per alignment and numbering from the active site Ser64 in PA1 (244). Multiple sequence alignment of PDC-3, A. baumannii ADC, E. cloacae P99, E. coli CMY-2, and E. coli AmpC proteins reveal that only ADC has a non- Ala at the equivalent position (ADC has Asn, see Figure 3-6). β-lactam Substrate Kinetics 182

185 Table 4-4 summarizes our data characterizing the PDC-3 β-lactamase. PDC-3 protein was purified using pief, size exclusion, and cation exchange chromatography. In general, the catalytic profile of the penicillin and cephalosporins tested is consistent with the kinetic behavior described for other AmpC cephalosporinases. amely, k cat values are in the order of 10 2 s -1 for narrow-spectrum cephalosporins such as CF, cephalothin, and cephaloridine, and k cat /K m is typically > 1 µm -1 s -1 (142, 155, 212, 418). In contrast, for ampicillin and the extended-spectrum cephalosporins, hydrolysis was not measurable, suggesting k cat is < 1 s -1. The K i app s for these β-lactams remain in the low M range. β-lactamase Inhibitor Kinetics In order to explore the inhibitory profile of PDC-3, we tested three types of compounds as inactivators of this -lactamase: (i) boronic acid cephalosporin and penicillin analogs with both achiral and chiral structures; (ii) the β-lactam carbapenems, PSR-3-283a, and BAL29880; and (iii) the non-β-lactam inhibitor XL104. The inhibition data is compared to that for the currently available class A -lactamase inhibitors and the monobactam aztreonam. The results are presented in Table 4-5. First, of all the compounds examined, the commercially available inhibitors demonstrate the highest K i app s. Reduction of nitrocefinase activity could not be achieved with up to 10 mm of the oxapenam, clavulanate, suggesting that PDC-3 is not inhibited by the compound. The sulfone, tazobactam, has a significantly lower K i app of 24 M. We also tested aztreonam, which behaves as a slow substrate, or transient inhibitor, with class C -lactamases (66). In competition assays with CF, aztreonam achieved a K i app in the range of effective -lactamase inhibition (1.3 ± 0.1 M). 183

186 ext, we screened boronic acid compounds that were synthesized to resemble β-lactam substrates. The relative K i s of these boronates provide insight into the molecular preferences of substrate specificities of the PDC-3 active site. The compounds have either an achiral (compounds with an a suffix: 2a, 6a, 7a, and 8a) or chiral (compounds with a c suffix: 1c, 3c, 4c, 5c, and 9c) configuration at the carbon adjacent to the boron atom. Both groups of compounds have side chains that mimic the R 1 substituents on either penicillin or cephalopsporin -lactams. The chiral compounds incorporate additional functional groups that resemble the R 2 or right-side of pencillins and cephalosporins including the dihydrothiazene ring and conserved C 3 or C 4 -lactam carboxylate. These reversible inhibitors form a covalent bond with the active site serine of class C -lactamases (see Chapter 3 for detailed discussion), and the measured K i s are interpreted relative to a reference compound (compound R) which lacks these side chains. The compounds in the achiral series display K i values that are two orders of magnitude lower than the reference compound (K i s = 33 ± 8 nm to 0.74 ± 0.07 μm). The ceftazidime analog, achiral compound 8a, displayed the lowest K i of the group. In the chiral series, the K i of the nafcillin analog (compound 1c) was comparable to the achiral series (K i = 0.42 ± 0.03 μm). However, the remaining cephalosporin analogs have K i s that are at least 10- fold lower than the achiral compounds. f particularly interest is the cephalothin analog, chiral compound 4c, which achieves a low nm K i for PDC-3, 4 ± 1 nm. We chose to extend our studies of the commercially available carbapenems as inhibitors of the ADC AmpC by measuring their K i app values against PDC-3. All of the carbapenems maintained K i app s < 6 μm. As these β-lactams differ primarily by their R 2 184

187 side chains, it is likely that the K i app s reflect differences in how these substituents interact with the enzyme. The final compounds explored were investigational -lactamase inhibitors with varied chemical structures. Here, we hoped to learn whether the PDC-3 active site could recognize, and be inactivated by, compounds with different molecular features. Specifically, XL104 is a non- -lactam, BAL29880 is a monobactam with a bridge between C 3 and C 4 on the lactam ring, and PSR-3-283a is a novel -lactam sulfone. Despite differences in the reaction mechanisms of each of these investigational inhibitors, the K i app values were all within a 7-fold range (K i app s = 0.87 ± 0.09 to 5.97 ± 0.36 μm). We also measured the inactivation rates and K I s of these investigational inhibitors and compared them to tazobactam and aztreonam (Table 4-6). With the exception of tazobactam, the K I values are remarkably similar for the various compounds. The k inact s are also similar, although the two sulfones (tazobactam and PSR-3-283a) have rates approximately 1.5- to 4-fold lower than the other compounds. The estimate of inactivation efficiency (k inact /K I ) for tazobactam is, by at least an order of magnitude, lower than the other inhibitors. This may explain, in part, why tazobactam is generally considered ineffective against AmpCs (222). The largest k inact /K I values are seen with XL104 and BAL Interestingly, the compound with the highest k inact (aztreonam) had a relatively low k inact /K I, emphasizing the important of both factors in this calculation of efficiency. Furthermore, despite a K I of 0.94 ± 0.19 μm, a low k inact contributed to the k inact /K I value of PSR-3-283a, which is the lowest of the investigational inhibitors (0.049 ± μm -1 s -1 ). 185

188 ESI-MS and the ature of Inactivation Products We used timed ESI-MS to detect covalent reaction intermediates for PDC-3 after incubation with: (i) the chiral boronate 4c; (ii) PSR-3-283a, and (iii) BAL As shown in Table 4-7, the mass of the purified PDC-3 protein in each experiment is within error for the expected mass based on translation of the bla PDC-3 gene from P. aeruginosa 18SH. For the experiments with the boronates, we used PDC-3 purified from the clinical 18SH strain. After 15 min incubation, the chiral cephalothin analog 4c did not form a detectable covalent complex with PDC-3. This finding is consistent with both the reversible mechanism of these inhibitors and our previous data on the ADC -lactamase. For ESI-MS studies of PSR-3-283a, the PDC-3 protein was purified from the pbc SK (-) vector. Based on the mass spectrum of the enzyme alone, we have evidence to suggest that multiple C-terminal truncations may be present. While we cannot conclude definitively that these truncations do not affect the function of the enzyme, kinetic assays examining the nitrocefinase activity of this protein sample were not significantly different than determined for the full-length protein purified from the native P. aeruginosa strain (data not shown). Like us, other labs have found that terminal truncations of -lactamases have been found not to alter catalytic activity (295, 491). Following incubation with PSR-3-283a, PDC-3 forms a adduct on ESI-MS, which we assign to an acyl-enzyme enamine species (which may exist in the trans- or cisconfiguration) (Figure 4-4A). This species reflects the loss of 17 amu from the mass of intact PSR-3-283a, likely indicating the loss of an H group or water molecule. In the experiments which studied BAL29880 and PDC-3, the β-lactamase was purified from the pet 24a(+) vector, and part of the sample contains an additional Met residue on the -terminus (accounting for the expected PDC-3 mass amu). The codon for this 186

189 residue is introduced with the dei restriction site needed to clone the gene into the vector. Previous studies in our lab did not find a functional difference between - lactamases with this extra amino acid. After incubation with BAL29880, the deconvoluted ESI-MS spectrum contains peaks which correspond to the molecular weights of the -lactamase (both with and without the Met) plus 414 and 332 amu (Figure 4-4B). We posit that these adducts reflect the covalent addition of the full BAL29880 compound and the compound without the -S - 3 group, respectively. Figure 4-4C shows a possible mechanism for loss of this group involving hydrolysis of a sulfonamide from the acyl-enzyme and generation of an amine and sulfate ion (142). Susceptibility Testing with Investigational Inhibitors Following the kinetic characterization of the investigational inhibitors, we sought to determine their activity in whole cell assays. Penetration of the outer cell wall presents a significant challenge to the treatment of Gram-negative bacteria, and is a requirement for any potentially successful -lactamase inhibitor. Microdilution MICs were performed for the chiral cephalothin boronate 4c, XL104, BAL29880, and PSR-3-283a in combination with the extended-spectrum cephalosporin cefotaxime (Table 4-8). The inhibitors were held constant at 4 g/ml for each condition. In testing against the PDC β-lactamase expressed in E. coli DH10B cells, each of the inhibitors successfully lowers the cefotaxime MIC by at least two dilutions (boronate) and as many as six dilutions (XL104, BAL29880, and PSR-3-283a). However, the inhibitors were unable to recover cefotaxime susceptibility for P. aeruginosa strains expressing either the PDC-1 or PDC-3 β-lactamases. 187

190 Insights into Substrate and Inhibitor Carboxylate Binding from Molecular Representations f all the inhibitors tested against PDC-3, the chiral cephalothin boronate 4c demonstrated the lowest K i (4 ± 1 nm). How can we explain this result? To better understand the interactions between the β-lactamase and this compound, we created a molecular model of 4c covalently bound to Ser64 of PDC-3. The inhibitor s boron atom has two hydroxyl groups; in our model, one hydroxyl is found in the putative oxyanion hole of PDC-3 formed by the backbone amide groups of Ser64 and Ser318. However, the other hydroxyl oxygen is found hydrogen bonding with side chain of Thr316, an interaction not reported in the crystal structures of boronic acid derivatives with AmpC enzymes. These solved structures typically show that one boronate oxygen atom is placed in the oxyanion hole formed by residues 64 and 318, as we observe in PDC-3, but the other oxygen atom hydrogen bonds with Tyr150 (82, 96, 320, 400). A comparison of the overall dispositions of 4c in the active site of our PDC-3 model and the E. coli-4c crystal reveal significant differences in the positions of the R 1 side chains of the inhibitor (Figure 4-6). In the E. coli AmpC, the R 1 thiophene ring is found in the amide recognition site formed by interactions with Asn152 and Ala318. In our PDC-3 model, the R 1 group is oriented towards the back of the active site pocket in PDC- 3, nearby residues not typically implicated in substrate binding (e.g., Arg149). In contrast, the R 2 meta-carboxyphenyl group is found in the expected carboxylate binding region of both β-lactamases, but the specific contacts of the carboxylate vary between the PDC-3 and E. coli AmpCs. In the E. coli AmpC-4c structure, the meta-carboxylate makes an unexpected bond with Asn289. The PDC-3 model suggests that the same carboxylate 188

191 hydrogen bonds with Asn343 and Arg349, both of which are implicated in the binding the C 3 /C 4 carboxylate in other AmpC structures (21, 22, 400). Asn346, another residue responsible for carboxylate recognition, is also found within ~ 5 Å of the carboxylate, and may play a role in the electrostatic environment of the region. Thus despite possible differences in how the boronate docks in the active site of the PDC-3, compound 4c still retains some of the conserved AmpC-ligand interactions. Susceptiblity Testing of PDC Variants at Potential Carboxylate Binding Residues We applied the insight gained from this model and performed site-directed mutagenesis at the residues possibly involved in carboxylate recognition. Table 4-9 summarizes the MICs values of E. coli DH10B cells expressing PDC-3 Ala variants at residues 343, 346, and 349. As compared to the wild-type PDC-3, the MICs for each of the tested substrates drop at least two dilutions for the Arg349Ala variant. The effect of the substitution at Asn343 and Asn346 is less than Arg349 for the penicillins, and insignificant for cephalothin and aztreonam. Somewhat surprisingly, the largest MIC difference between PDC-3 and the Arg349Ala variant is observed the nine dilution change for ampicillin/clavulanate. The 343- and 346-substituted enzymes also demonstrate increased susceptibility to ampicillin/clavulanate, suggesting an important role for this region of the enzyme in susceptibility to the β-lactam/β-lactamase inhibitor combination. Discussion AmpC verexpression in P. aeruginosa 18SH P. aeruginosa PA1 and 18SH are both derived from isolates recovered from human wound sites (203, 285). While the entire genome of PA1 has been described, the genetic 189

192 background of P. aeruginosa 18SH is not known (440). We cannot assume these two strains possess an equivalent background, but the general mechanisms of resistance are likely to be similar because of shared phylogeny. The important exception to this generalization is that P. aeruginosa 18SH was identified for its high level of carbenicillin resistance after serial subculturing in the penicillin (285). Previous characterization of the 18SH strain confirmed a constitutively high level synthesis of β-lactamase (27, 153). The 18SH strain produces approximately three times the β-lactamase enzyme units/mg of (dry weight) bacteria compared to the parent 18S strain. The genetic explanation for this overexpression has not been elucidated, but based on work on AmpC hyperproduction in clinical P. aeruginosa isolates, mutations in the regulatory ampd genes are likely (230). In contrast, P. aeruginosa PA1, which is generally susceptible to β-lactams, expresses a constitutively low level of the AmpC (16, 293). We confirmed these findings in our laboratory by demonstrating large differences in the basal nitrocefinase activity between crude extracts of 18SH and PA1. After incubation with cefoxitin, an inducer of AmpC-expression, this nitrocefinase activity increased over 10-fold for the PA1 sample, but remained unchanged for 18SH. We cannot conclude definitively how much of the nitrocefinase activity is due to AmpC production, as the presence of other β-lactamases is possible in both strains. However, the initial descriptions of the 18SH strain estimate that > 90 % of the protein extracted from the periplasm was the single cephalosporinase protein species (27). Antibody Detection of AmpC β-lactamases and Relationship to Expression Level Based on the data suggesting significant differences in the amount of β-lactamase produced by the 18SH and PA1 strains, we reasoned that expression levels could be 190

193 measured by immunoblotting. We also considered that polyclonal anti-pdc-3 antibody must recognize other class C β-lactamses based on protein homology. The cross-reactivity of the anti-pdc-3 antibody is limited to other class C β-lactamases, specifically E. cloacae P99 and the plasmid mediated E. coli CMY-2. Interestingly, the antibody does not recognize the ADC β-lactamase at a detectable level. Based on comparisons of primary sequence, PDC-3 shares from 62-65% amino acid similarity with these other AmpCs. While we cannot make conclusions about the epitopes recognized by the polyclonal antibody without further study, this evidence suggests that the linear regions recognized by the PDC-3 antibody are more similar between PDC-3, CMY-2, and P99 than for ADC. Probing the crude lysates from clinical and laboratory isolates with the anti-pdc-3 antibody revealed that PDC-3-like β-lactamases can be detected in all but one of the P. aeruginosa samples. A faint band is observed in the P. aeruginosa ATCC lane. This band may indicate, as has been reported in the literature for this wild-type strain, a very low basal level of AmpC expression (459). This is consistent with the band of comparable intensity in the lane containing lysate from the low level expressor P. aeruginosa PA1. Crude lysate of P. aeruginosa DB322 was recognized by the anti-pdc-3 antibody, and this clinical strain has been characterized as overexpressing the AmpC β-lactamase (403). A detectable signal can not be seen in the lane containing lysate from strain UL140, an isolate with changes in the genes encoding drug efflux pumps. Quale and colleagues reported that UL140 makes 470 times less ampc mra as compared to DB322 (403). Thus, the AmpC expression level of UL140 may be too low for detection by this method. 191

194 We are currently expanding our study of the regulation of the P. aeruginosa bla ampc and will repeat these immunoblots with a broader panel of characterized samples. The crude lysates of the MK1184 and R strains show a band in addition to the one corresponding to the full length PDC-3 protein (as determined by the lane with purified PDC-3 protein). Both - and C-terminal truncations of Pseudomonal AmpC β- lactamases have been described in the literature and observed in our laboratory, and we posit that this additional lower molecular weight band may correspond to one of these truncated species (491). In contrast, the additional bands observed in the lanes containing the crude lysates of the E.coli cells containing either the pbc SK(-) bla PDC-3 or bla PA1 are of higher molecular weight. The approximate weight does not suggest these are aggregates of multiple protein molecules, and may be intrinsic E. coli proteins that crossreact with the antibody. The incidence of strains overproducing the AmpC β-lactamase has been reported as high as 18% of P. aeruginosa isolates, and patients infected with these overexpressors are more likely to receive inappropriate empirical antibiotic treatment and have persistently positive blood cultures (458, 459). These initial immunoblotting results suggest a high sensitivity and specificity for our polyclonal antibody. The development of an antibody which can identify AmpC overexpressing P. aeruginosa isolates offers potential for applications in a clinical assay that could impact and clarify decisions regarding patient care. β-lactam Substrate Kinetics and Implications for the Phenotype of P. aeruginosa 18SH The comparison between 18SH and PA1 affords us an opportunity to observe a phenotype resulting, in part, from overexpression of the AmpC β-lactamase. 192

195 Susceptibility testing of the P. aeruginosa strains 18SH and PA1, together with our kinetic characterization of purified PDC-3, illustrates several important features of cephalosporinases. The predominant difference between the MICs of these two strains is seen with the extended-spectrum cephalosporins (ceftazidime, cefotaxime, and cefepime) and the monobactam, aztreonam. This pattern is typical of Pseudomonal AmpC enzymes with high k cat rates for narrow-spectrum cephalosporins, such as cephalothin and cephaloridine (k cat = 410 ± 41 and 130 ± 13 s -1, respectively). Thus, even at lower levels of expression (e.g., PA1), the enzyme can hydrolyze enough β-lactam to produce a resistant phenotype. However, with cefotaxime, cefatzidime, and aztreonam the hydrolytic rates are much lower (not measurable in our assays; we have not yet tried to measure the hydrolysis of cefepime). The β-lactamase maintains a low K i app for these substrates (0.26 ± 0.03 to 52 ± 2 µm), which, in conjunction with the low k cat s, can yield k cat /K m values of possible in vivo significance. The periplasmic concentration of β-lactamase can reach mm, especially in overexpressing strains such as 18SH, and in the context of decreased membrane permeability often seen in P. aeruginosa, β-lactam concentrations may be low µm (70, 157, 488). Under these conditions, a small substrate K i, even in the setting of low hydrolytic rates, can allow cephalosporinases to trap β-lactams before reaching the penicillin binding protein target (424, 460). Thus, the constitutively high expression of the AmpC in P. aeruginosa 18SH likely plays a large role in these increased MICs. We are mindful that the multidrug resistance phenotype of the P. aeruginosa 18SH strain may be multifactorial, and while the genetic similarity between PA1 and 18SH is likely to be high, there may be other differences (e.g., changes in efflux pump 193

196 regulation or permeability) that contribute to the MIC profile. Furthermore, sequence analysis revealed one amino acid difference between the PA1 AmpC, designated PDC- 1, and the 18SH AmpC, designated PDC-3. In their characterization of a panel of clinically-derived P. aeruginosa AmpC -lactamases, Rodriquez-Martinez and colleagues reported data that this Thr79Ala change may be associated with increased catalytic efficiencies for cefotaxime, cefepime, and imipenem (418). As we have cloned both the bla PDC-1 and bla PDC-3 genes into the E. coli pbc SK(-) vector, it will be possible to test this hypothesis in a uniform background. E. coli strains are generally more permeable to β-lactams than P. aeruginosa, and thus would remove this confounding factor as well. Screening ovel Inhibitors for Activity Against PDC-3 ur panel of investigational inhibitors includes compounds with very different chemical compositions. The boronates are designed to mimic key substituents found on the enzymes natural substrates; PSR-3-283a resembles the sulfone compounds which have been successful inhibitors of class A enzymes; BAL29880 capitalizes on the mechanism of slow hydrolysis observed for monobactams in class C enzymes; and XL104 is a novel non-β-lactam which has restored susceptibility to cephalosporins in many difficultto-treat pathogens, but for which there is little kinetic data. The inhibitors structural and functional diversity can be a valuable tool. By examining the kinetics, reaction intermediates, and whole cell activity of these compounds, lessons emerge about the inhibition of PDC-3. Boronates. We start with the boronic acid derivatives, synthesized specifically to contain the R 1, and in the case of the chiral compounds, the R 2 side chains of β-lactamase 194

197 substrates. As discussed in Chapter 3, this approach has measured K i values in the low nm range for both class A and class C enzymes. Insights from the ADC β-lactamase are recapitulated in the PDC-3 protein: in general, the compounds which include an R 2 side chain display K i s at least 10-fold lower than the achiral counterparts. An important exception is compound 1c which contains the R 1 side chain of nafcillin and the R 2 metacarboxyphenyl group. Here, the K i with PDC-3 resembles more closely the achiral boronates (e.g., compounds 2a, 6a, 7a). The nafcillin boronate is the only compound tested with a pencillin-like R 1 group. Thus, the increased is K i consistent with the notion that the active site of cephalosporinases may best accommodate cephalosporin-like side chains. However, the relationship is clearly complex, as the K i app for the ampicillin β- lactam is still high (and likely for nafcillin as well), and the cephalosporin preference is more prominent in the high k cat s for narrow-spectrum cephalosporins. The mechanisms of catalysis and inhibition extend beyond just ligand recognition. Additional properties such as conformational changes and rate-determining steps influence the measurements we call K i and k cat. These processes are not readily accessible using steady state kinetics, and pre-steady state kinetics can offer important insights with implications for inhibitor design. Examination of the pre-steady state profile of AmpC enzymes is currently underway. We also note that the addition of an extra carbon on the linker between the carbon adjacent to the boron atom and the R 2 group may affect K i measurements. This comparison is afforded by the design of compounds 2a, 3c, 4c, and 5c. As was observed with ADC, the presence of the phenyl group and meta-carboxylate lowers the K i (K i = 0.37 ± 0.02 μm for 2a, but ± and ± μm for 3c and 4c, respectively). However, the K i of 5c, which differs from 4c by a single carbon, is 6-fold 195

198 lower than 3c and 4c. Likely the extra carbon moves the carboxylate into a less favorable position in the enzyme active site. The synthesis of chiral compounds containing the R 1 side group of the nafcillin, cefotaxime, and ceftazidime analogs is challenged by the bulk and size of the oxyimino (7a, 8a, 9c) and fused ring (1c) side chains. The addition of the carbon is a synthetic necessity for chiral 1c and 9c, and these steric difficulties continue to thwart the synthesis of a chiral version of 7a (per Drs. Prati and Caselli). We are reminded that our concept of an ideal inhibitor informed by this screening (e.g., a ceftazidime boronate with a meta-carboxyphenyl group) may not parallel what can be synthesized by medicinal chemists. Another lesson from the boronates is the discrepancy between substrate and inhibitor K i or K m values. For example, both the ceftazidime (8a, 9c) and chiral cephalothin (3c, 4c, 5c) analogs have K i s < 100 nm. Remarkably, even the achiral ceftazidime analog is in this range, making 8a the achiral compound with the lowest K i (30 ± 1 nm.) However, ceftazidime and cephalothin, as β-lactams, have two of the higher K i app or K m s (52 ± 2 and 24 ± 3 µm, respectively). This data suggests that the same R 1 side chain be recognized differently by the β-lactamase when it is presented as a β-lactam substrate or as a boronate. While the boronates are designed to take advantage of an enzyme s natural affinity for a ligand, this may not mean the binding mode or configuration is the same as expected for the substrate. The molecular interactions that mediate recognition of enzyme substrates are the subject of intense study. Despite this intense study, hydrogen bonding, electrostatics, and dynamics are not well understood. ur molecular model of PDC-3 in complex with 4c offers a similar interpretation about substrate and inhibitor recognition, and helps us generate hypotheses about how the inhibitor might interact with the active site. We are unable to compare directly our 196

199 inhibitor model to how a substrate binds, but studies of E. coli AmpC binding hot spots suggests a conserved R 1 recognition region (400). In our PDC-3-4c model, the R 1 of 4c is not found in this location, specifically, interacting with Asn152 and Xxx318. Instead, the R 1 group is oriented towards the back of the active site in a region not previously implicated in ligand binding. We are cognizant that comparisons between crystal structures and molecular models must be made carefully, but in the absence of a structure, models are useful tools. Carbapenems. The K i app s of the carbapenems are clustered in a tight range (0.93 ± 0.17 μm to 5.5 ± 0.4 μm). The penem scaffold on which these β-lactams are constructed is similar; thus we assign the subtle differences in K i app s to the interactions of the β- lactamase with the R 2 side chain. The order of K i app from smallest to largest, ertapenem < imipenem < doripenem < meropenem, is similar to that observed for the ADC β- lactamase (only ertapenem and imipenem were switched). This ranking may suggest a general trend for AmpC-carbapenem R 2 recognition. Further examination of the carbapenems and PDC-3 may help elucidate the many questions concerning the complex active site chemistry of carbapenems, including elimination of the C 6 substituent, conformational positioning of the R 2 group, and Δ 2 to Δ 1 tautomerization (see Chapter 3). Penam Sulfone PSR-3-283a. ext, we move to a discussion of the investigational penicillin sulfone PSR-3-283a. Against PDC-3, PSR-3-283a had a low µm K i app (6.0 ± 0.4 µm), comparable to the carbapenems and significantly lower than the class A sulfone inhibitor tazobactam (24 ± 1 µm). In addition to differences between the R 2 groups of tazobactam and PSR-3-283a (tazobactam has a triazolyl moiety), a β-hydroxymethyl group is found at position C 6 of PSR-3-283a. Attempting to explain the potential 197

200 inhibitory mechanism of this investigational inhibitor is facilitated by a review of the data on similar compounds in class A and C β-lactamases. In the mid-1990 s Mobashery and colleagues studied an inhibitor similar to PSR-3-283a that was a penicillinate (sulfide at C 1 instead of a sulfonyl) with a C 6 α- hydroxymethyl group (Figure 4-5) (301, 315). The crystal structure of this sulfide suggested that the TEM-1 enzyme was inhibited by the formation of a prolonged acylenzyme, permitted by the C 6 group displacing or decreasing the nucleophilicity of the hydrolytic water molecule. Increasing the size of the C 6 group, as is seen with the carbapenems C 6 α-hydroxyethyl group, has different inhibition modes in class A and C enzymes. Carbapenems displace the water molecule in class A enzymes, but carbapenems also force a conformational change in both class A and C acyl-enzymes that moves the β- lactam carbonyl into a hydrolytically unfavorable conformation (21, 302, 493). The hydrolytic water is still observed in the E. coli AmpC-imipenem crystal, reflecting the evidence that the hydrolytic water approaches from the β-face of the acyl-enzyme (59, 60). The synthesis and screening of a panel of sulbactam and tazobactam derivatives with C 6 substituents in the α- and β-configurations by Bitha and colleagues adds additional insight into the behavior of PSR-3-283a (31, 32). Their findings can be summarized as: (i) the β- configuration is superior to α- for inhibiting class C enzymes, but both configurations can inhibit class A enzymes with low µm IC 50 s; (ii) the C 6 hydroxymethyl substituent achieves lower IC 50 s than the hydroxyethyl substituent; and (iii) the sulfone oxidation state at C 1 is also important for enhanced inhibitory activity. Based on this previous data from compounds with similar features, we posit that PSR a s C 6 β-hydroxymethyl group may not be large enough to precipitate movement of 198

201 the β-lactam carbonyl out of the oxyanion hole of PDC-3. Rather, the inhibition mechanism may rely on impeding the approach of the hydrolytic water molecule. We propose that PSR-3-283a finds the binding pocket by taking advantage of conserved active site β-lactam recognition elements. Acylation is likely followed by the formation of a stable reaction intermediate, facilitated by departure of the sulfonyl group as is seen with tazobacam and class A enzymes (351). This acyl-enzyme then slows deacylation by the specific placement of the C 6 group. ur MS data with PSR-3-283a supports the hypothesis that this inhibitor follows a reaction pathway involving formation of a stabilized acyl-enzyme species. We observe the formation of an adduct which corresponds to the molecular weights of the PDC-3 enzyme plus an enamine species (Δ ± 3). (f note, the adduct size reflects the possible loss of an H group or water molecule.) To our knowledge, structural evidence of the cross-linking of a sulfone inhibitor has not been described in class C -lactamases (although hydroxyamates have been shown to cross link Ser64 and Lys315 in the P99 enzyme) (505). As we develop our studies on the inhibition of AmpCs, we are interested in exploring whether an inhibitor such as PSR-3-283a could cross-link PDC-3. We propose that an active site Tyr or Lys could be involved in a manner analogous to the Ser130 cross-linking of clavulanate or tazobactam in class A β-lactamases (55, 250, 354, 445, 465). The role of Tyr150 in the catalytic mechanism of class C enzymes continues to be the subject of much investigation (60, 96, 136, 165, 343). An issue at the core of the discussion is the pk a for the residue, and whether it could exist in the deprotonated state that would be necessary for a role as a proton acceptor (catalytic base) in activating Ser64 and/or the hydrolytic water (236, 256). While the evidence is mixed, a deprotonated Tyr150 may be biochemically possible, which would 199

202 not preclude a role in cross linking to an active site ligand. evertheless, Tyr150 may be a particularly versatile residue. Recent work by Pelto and Pratt demonstrate Tyr150 and Lys315 form the oxyanion hole in recognition of hydroxyamate inhibitors (373). Alternatively, a reactive Lys, such as Lys315, may serve as the cross-linking residue, as Bush and colleagues have suggested for class C enzymes and tazobactam (69). Bridged Monobactam BAL BAL29880, and bridged monobactams in general, are designed to prevent the hydrolytic water access to the ester bond in class C enzymes (184, 277). Specifically, the C 3 -C 4 bond limits rotation of the acylated inhibitor. In our assays with PDC-3, BAL29880 achieved sub-µm K i app and K I values, the lowest in the panel of non-boronate investigational inhibitors. We note that the k inact s for aztreonam and BAL29880 are comparable; the K i app and K I values are improved for BAL29880 over aztreonam. MS spectra indicate the formation of an unfragmented acyl-adduct as well as a species corresponding to the inhibitor following loss of the -linked S - 3. The elimination of the -linked substituent following hydrolysis of the acyl-enzyme s sulfonamide bond has been described for BAL29880 with other class C enzymes, as well as for other monobactams in class A enzymes (142, 323). on- -Lactam XL104. Lastly, the kinetic measurements of PDC-3 and XL104 fall in the middle of the range defined by the BAL29880 and PSR-3-283a. This inhibition profile continues to offer significant improvements over the available class A inhibitors. The mechanism of inhibition by XL104 is not well-studied, but the one reported crystal structure of the inhibitor in complex with a class A enzyme, suggests that the inhibitor does acylate the active site serine, undergo ring opening, and displace the deacylating water molecule (125). We have not yet obtained MS data on XL104 and PDC-3, but 200

203 anticipate the formation of an acyl-enzyme. We further anticipate that we may observe elimination of the -linked S 3 - group as seen with BAL k inact /K I Ratios and Recovery of β-lactam Susceptibility We next tested the inhibitors in whole cell assays. The ultimate goal of understanding the kinetic profiles of novel inhibitors is to aid further improvements in design to enhance enzyme inhibition. However, the identification of a kinetic value that can predict the ability of an inhibitor to protect a partner β-lactam in vivo remains elusive. Here, we tested boronate 4c, PSR-3-283a, BAL29880, and XL104 (all at 4 µg/ml) in combination with cefotaxime against both P. aeruginosa and E. coli strains expressing PDC-1 or PDC- 3. Against the E. coli cells with bla PDC genes, each of the inhibitors demonstrates the ability to enhance β-lactam activity. In particular, the MICs of cells treated with cefotaxime and PSR-3-283a, BAL29880, or XL104 are lowered at least 5 dilutions. The k inact /K I values of these inhibitors ranges from ± to ± μm -1 s -1 (a k inact /K I was not obtained for the reversible inhibitor boronate), what may appear a significant difference in inhibitory parameters. However, the inhibitors perform comparably in whole cells assays, providing an important reminder that MIC testing can level the playing field for even the most kinetically promising prospects. The same β-lactam/β-lactamase inhibitor combinations are unable to increase susceptibility in either P. aeruginosa strain. As mentioned above, the outer membrane permeability of P. aeruginosa presents a substantial challenge to β-lactam access to penicillin binding proteins. Even if the inhibitors retain their inhibitory activity, entry into the cell can limit any potential benefit. These results suggest that inhibition of PDC-1 and PDC-3 is attainable, but cell entry remains a significant barrier for these investigational 201

204 inhibitors. Increasing the amount of inhibitor may meet some success, but there will always be pharmacodynamic and pharmacokinetic challenges to achieving sufficient concentrations in vivo. Insights into Carboxylate Binding and Unanticipated Effects on Clavulanate Susceptibility In our final discussion, we introduce an emerging insight of current studies. The molecular representation of PDC-3-4c allowed us to explore the potential PDC-3 β- lactam carboxylate binding residue(s). Arg244 has been identified as the primary C 3 /C 4 carboxylate binding residue in several class A enzymes (217, 464, 465, 512). The class C correlate is less clear, and different enzymes-ligand pairs implicate residues such as Thr316, Xxx343, Asn346, and Arg349 (21, 22, 137, 281, 320, 364, 400, 516). ur acylenzyme model identified hydrogen bonds between the meta-carboxylate of compound 4c and Asn343 and Arg349. We performed site-directed mutagenesis at Asn 343 and Arg349, as well Asn346, to test whether these residues roles in ligand recognition. ur MIC testing of the three variants expressed in E. coli cells shows that the Ala substitution at residue 349 does have a pronounced effect on substrate MICs. The susceptibility of cells expressing the 343 and 346 variants are less altered by this substitution. Presumably, this discrepancy arises because the charge of Arg349 is electrostatically involved with the C 3 or C 4 carboxylate of the pencillins and cephalosporins, whereas 343 and 346 play an auxiliary role in the shape or integrity of the binding region. Aztreonam does not have an equivalent to the C 3 or C 4 carboxylate, and thus is less affected by all the substitutions. 202

205 f particular interest were our results showing the significant decrease in ampicillin/clavulanate MICs for each of the tested carboxylate-binding variants, reflecting an increased susceptibility to inhibition. That clavulanate may have improved activity against these enzymes is unexpected. The intrinsic resistance of class C enzymes to inhibition by clavulanate is not well understood. Possible explanations include decreased affinity for the class C active site, fewer productive meetings between clavulanate and the active site, infrequent or slow acylation (as is seen for clavulanate and P99), or rapid deacylation (282, 319, 410). Ala residues are considerably smaller than Asn or Arg, and the Ala substitutions may create more space, and perhaps flexibility, in this region of the active site. In fact, the homology-based protein modeling server used to generate the PDC-3 representation identified a P. fluorescens class C β-lactamase to serve as the PDB template. Michaux and colleagues defined and described the structure of this enzyme (which shares 76% sequence similarity to PDC-3) and identify biochemical properties that allow cold-adapted enzymes to optimize catalytic activity at low temperatures (311). Specifically, these psychrophilic enzymes enjoy increased structural flexibility at the price of decreased stability (147). P. aeruginosa is not a psychrophilic organism, but the similarity between the two class C β-lactamases helps generate hypotheses about the biochemical properties of PDC-3. In general, cold-adapted enzymes have decreased numbers of ion pairs and hydrogen bonds, low arginine content, and increased solvent accessibility to the active site (147, 311). ne possible explanation for the increased clavulanate susceptibility in PDC-3 343, 346, and 349 variants requires a rather large assumption that the mechanism of clavulanate inhibition is similar to that outlined for class A. After formation of the acyl enzyme, clavulanate requires saturation of the double bond at C 2, which facilitates ring 203

206 opening and formation of inactivating species. The source of the proton for double bond saturation is likely a well-defined water molecule bound by Arg244 in TEM (see Chapter 2). If PDC-3 shares attributes with its cold-loving counterpart, we may expect it to be a more flexible enzyme. Because of the decreased number of intramolecular interactions and increased solvent accessibility, water molecules may be more available and less tightly bound. Hence, an additional increase in local flexibility (from Ala substitutions) near the putative C 2 and C 3 binding sites may further enhance the availability and movement of potential proton donors (i.e., solvent water molecules). A more flexible active site may also, for reasons not elucidated, effectively increase the affinity for clavulanate. These notions await testing with purified PDC 343, 346, and 349 variants. We are also interested in an examination of the thermal stability and conformational flexibility of PDC-3 through circular dichroism and thermal denaturation experiments. Conclusion In summary, we have presented data describing the inhibition of a β-lactamase with significant clinical significance. Effective inactivation of this β-lactamase is possible; investigational inhibitors including cephalosporin analog boronates, a penicillin sulfone, bridged monobactam, and XL104 show promising kinetic data. The molecular model of the PDC-3 AmpC suggests that this enzyme may have a particularly flexible structure with implications for substrate and inhibitor recognition. ur comparative analysis of different inhibitor structures compels us to keep in mind the complex relationship between active site architecture, tertiary structure, functionality, and stability (467). 204

207 But despite the investigational inhibitors impressive inactivation kinetics, susceptibility testing in P. aeruginosa revealed that a major barrier to restoring β-lactam activity continues to be entry into the P. aeruginosa cell. Efforts to optimize the uptake of β-lactamase inhibitors may require innovative side chain design (such as that seen in the siderophore monobactams and sulfones) and/or additional agents targeted specifically at drug efflux pumps (359, 367). 205

208 Table 4-1. Bacterial strains used in Chapter 4 studies Strain Location of bla ampc PDC-type Relative Level of Chromosomal AmpC Expression a P. aeruginosa PA1 chromosome PDC-1 low P. aeruginosa 18SH chromosome PDC-3 high Used For MICs, cefoxitin induction, immunoblot MICs, cefoxitin induction, immunoblot, cloning, enzyme purification Ref. (418) (this work) P. aeruginosa MK1184 chromosome D b low Immunoblot P. aeruginosa R chromosome D low Immunoblot P. aeruginosa ATCC chromosome ot yet named (not PDC-1 or -3) low Immunoblot (459) P. aeruginosa DB322 chromosome D high Immunoblot (403) P. aeruginosa UL140 chromosome D low Immunoblot (403) E. coli DH10B bla PDC-1 E. coli DH10B bla PDC-3 pbc SK(-) plasmid pbc SK(-) plasmid PDC-1 - MICs PDC-3 - MICs, cloning, enzyme purification E. coli BL21 (DE3) pet 24a(+) plasmid PDC-3 - Enzyme purification E. coli BL21 (DE3) RP codon+ pet 24a(+) plasmid PDC-3 - Enzyme purification E. coli clinical strain, CTX-M expressor none - - Immunoblot K. pneumoniae ATCC none - - Immunoblot a As determined by cefoxitin induction experiment (PA1 and 18SH), immunoblotting (PA1, 18SH, MK1184, R, 27853, DB322, UL140), or as described in the literature based on ampc RA expression (PA1, 27853, DB322, and UL140). b ot determined 206

209 Table 4-2. PCR amplification and sequencing primers used in Chapter 4 experiments Primer name Sequence PA1 downstream 5 GCG GAG GGG CGG GGA AGC GCT CAT 3 PA1 upstream 5 CGT CGT TTG CGG CAA ATC CTG CGC 3 M13 Reverse 5 CAG GAA ACA GCT ATG AC 3 M13 Universal 5 GTA AAA CGA CGG CCA G 3 PA1 AS Forward 5 TTC GAG ATC GGC TCG GTG AGC AAG 3 PA1 AS Reverse 5 CTT GCT CAC CGA GCC GAT CTC GAA3 PDC-3 dei 5 CAT ATG GGC GAG GCC CCG GCG GAT 3 PDC-3 BamHI 5 GGA TCC TCA GCG CTT CAG CGG CAC CTT 3 PDC-3 to PDC-1 Ala79Thr a 5 GCC GGC TAT GCC CTG ACC CAG GAC AAG ATG CG 3 PDC-3 Asn343Ala a 5 ATC CTG GCC AAC CGC GCC TAT CCC AAT GCC GAG 3 PDC-3 Asn346Ala a 5 AAC CGC AAC TAT CCC GCC GCC GAG CGG GTG AAG 3 PDC-3 Arg349Ala a 5 TAT CCC AAT GCC GAG GCG GTG AAG ATC GCC TAC 3 a For the site-directed mutagenesis reactions, both the primer listed and its reverse complement were used. 207

210 Table 4-3. MIC values (μg/ml) of P. aeruginosa strains PA1 and 18SH, expressing PDC-1 and PDC-3 β-lactamases, respectively. Abbreviations: AMP, ampicillin; PIP, piperacillin; CEF, cephalothin; CAZ, ceftazidime; CTX, cefotaxime; FEP, cefepime; ATM, aztreonam. Isolate AMP PIP CEF CAZ CTX FEP ATM P. aeruginosa PA1 (PDC-1) P. aeruginosa 18SH (PDC-3) 2048 D > >1024 >128 > >64 208

211 Table 4-4. PDC-3 -lactamase substrate kinetics Substrate K m or K i app (μm) k cat (s -1 ) k cat /K m (μm -1 s -1 ) Ampicillin 0.31 ± 0.02 M a Aztreonam 1.3 ± 0.1 M itrocefin 21 ± 2 (K m ) 450 ± ± 3 Cephalothin 24 ± 3 (K m ) 410 ± ± 3 Cephaloridine 33 ± 1 (K m ) 130 ± ± 0.4 Cefoperazone 15 ± 1 M Ceftazidime 52 ± 2 M Cefotaxime 0.26 ± 0.03 M a k cat rates are designated not measureable, M, if hydrolysis was not detectable with 5 µg of protein. 209

212 Table 4-5. K i and K i apps s of inhibitors in direct competition assays with PDC-3 - lactamase and CF Commercially available class A inhibitors and monobactam K i app (μm) Clavulanate > Tazobactam 24 ± 4 Aztreonam 1.3 ± 0.1 Boronic acid derivatives R 1 H H R 2 Compound R (reference compound) -lactam analog B H H R 1 R 2 K i (μm) - CH 3 H 25.8 ± 3.2 1c afcillin 0.42 ± 0.03 H 2a H 0.37 ± c ± c Cephalothin S H ± c ± H H 6a Cefoperazon e H 0.74 ± 0.07 H 210

213 Boronic acid derivatives, continued Compound -lactam analog R 1 R 2 K i (μm) 7a Cefotaxime H 2 H 0.22 ± a 9c Ceftazidime H 2 S H S H ± ± H Carbapenems R 2 K i app (μm) Imipenem 1.1 ± 0.3 Ertapenem 0.93 ± 0.17 Doripenem H ± 0.8 S Meropenem 5.5 ± 0.4 S H H S S H H H H H H H H H S Investigational inhibitors K i app (μm) PSR-3-283a 6.0 ± 0.4 H 2 H H S BAL29880 H 0.87 ± 0.09 H C 2 a S H XL ± H 2 S 3

214 Table 4-6. k inact rates Inhibitor K I (μm) k inact (s -1 ) k inact /K I (μm -1 s -1 ) Tazobactam 25.8 ± ± ± Aztreonam 2.71 ± ± ± PSR-3-283a 0.94 ± ± ± BAL ± ± ± XL ± ± ±

215 Table 4-7. ESI-MS analysis (amu) of PDC-3 alone and incubated with inhibitors for 15 min at an I:E of 25:1 for boronate and 20:1 for PSR-3-283a and BAL a Predicted molecular weight of β- lactamase or inhibitor Species observed in deconvoluted spectra difference from β-lactamase molecular weight Boronate PDC-3 alone 40,648 40,646 2 PDC-3 + boronate ,645 1 PSR-3-283a PDC-3 alone 40,648 40,647 40,491 (-Arg from C-term) 40,362 (-LysArg from C-term) 40,249 (-LeuLysArg from C-term) PDC-3 + PSR-3-283a BAL ,892 40,493 (from LeuLysArg species) PDC-3 alone 40,648 40,644 40,775 (with Met) PDC-3 + BAL ,976 41,058 41,107 (with Met) 41,188 (with Met) a All measurements have an error of ± 3 amu

216 Table 4-8. Cefotaxime MICs ( g/ml) of P. aeruginosa 18SH and PA1 expressing PDC- 1 and PDC-3 β-lactamases, respectively, and E. coli DH10B expressing bla PDC-3 or bla PA1 in pbc SK (-), inhibitors at 4 g/ml. Isolate CTX CTX/boronate 4c CTX/PSR-3-283a CTX/BAL CTX/XL E. coli DH10B P. aeruginosa 18SH (PDC-3) E. coli DH10B bla PDC-3 P. aeruginosa PA1 (PDC-1) E. coli DH10B bla PDC-1 > 64 > 64 > 64 > 64 >

217 Table 4-9. MIC values (μg/ml) of P. aeruginosa 18SH and E. coli DH10B expressing bla PDC-3 or bla PDC-3 variants at possible carboxylate binding sites. Inhibitors combined with 50 μg/ml ampicillin. Isolate Ampicillin Piperacillin Cephalothin Aztreonam Ampicillin/ clavulanate Ampicillin/ tazobactam P. aeruginosa 18SH >1024 > E. coli DH10B E. coli bla PDC E. coli bla PDC-3 Asn343Ala E. coli bla PDC-3 Asn346Ala E. coli bla PDC-3 Arg349Ala

218 Figure 4-1. Chemical structures of: (A) β-lactam substrates used in this study; and (B) investigational inhibitors tested. Aztreonam and cephalothin structures are labeled with accepted ring numbering system. (A) S H 2 H 2 H H S H H S 3 H S H H S H S H H S H + Ampicillin Aztreonam Cephalothin Cephaloridine H S H H S H + + H H H S H S H2 S H H S - + H 2 S H S H itrocefin Cefoperazone Ceftazidime Cefotaxime (B) H H H Clavulanate B H H H S B H H S H Tazobactam H H H H B H H 2 S H B H H R 2a 6a 7a 8a H 2 H S H B H H H B H H H H S B H H H S B H H H S H B H H H H 2 H S H B H H H 1c 3c 4c 5c 9c H S C 2 a H 2 H H H S H H 2 S 3 PSR-3-283a BAL29880 XL

219 µm CF hydrolyzed/sec/µg protein Figure 4-2. Specific activity ( M CF hydrolyzed/sec/ g protein) of crude protein extract from P. aeruginosa PA1 and 18SH without and following 3 hr incubation with 50 µg/ml cefoxitin PA1 control PA1 3 hr 50 µg/ml cefoxitin 18SH control 18SH 3 hr 50 µg/ml cefoxitin 217

220 CcrA KPC-2 SHV-1 XA-1 CMY-2 ADC P99 P. aeruginosa UL140 P. aeruginosa 18SH (PDC-3) P. aerufginosa PA-1 (PDC-1) E. coli pbcsk PDC-3 E. coli pbcsk PDC-1 P. aeruginosa MK1184 P. aeruginosa R E. coli CTX-M K. pneumoniae P. Aeruginosa P. aeruginosa DB322 Pure PDC-3 from P. aeruginosa, 600ng 50ng 250ng 600ng 1 g 2 g 600ng 1 g Figure 4-3. Immunoblots of: (A) purified PDC-3 protein; and (B) purified β-lactamase protein from all Ambler classes and crude lysates of clinical strains. Membranes were probed with polyclonal anti-pdc-3 antibody at 1 µg/ml. pbc SK(-) refers to PDC-3 protein purified from E. coli DH10B cells expressing bla PDC-3 on the pbc SK(-) vector. Clinical refers to PDC-3 protein purified from P. aeruginosa 18SH. Bands indicating detection of CMY-2 and P. aeruginosa ATCC proteins are circled. (A) pbc SK(-) clinical (B)

221 Figure 4-4. Proposed reaction intermediates detected on ESI-MS after 15 min incubation of (A) PSR-3-283a and (B) BAL29880 with PDC-3 at an I:E of 20:1. The theoretical masses of the adducts are shown. nly the cis-enamine is shown for PSR-3-283a, but the mass adduct likely reflects both the cis- and trans-configurations. (C) Possible mechanism for elimination of the S - 3 group from BAL29880 involving hydrolysis of the sulfonamide bond of the acyl-enzyme and generation of an amine and sulfate ion. (A) - H S C - Ser (B) H 2 H H H H S H Ser 64 H 2 H H H Ser 64 H 3 (C) H 2 H H H Ser 64 H S H H 2 H 2 H H H Ser 64 H 3 + S

222 Figure 4-5. Structure of C 6 α-hydroxymethyl penicillinate studied by Mobashery and colleagues (301, 315) 220

223 Figure 4-6. Solvent-accessible surfaces of E. coli AmpC bound to the chiral cephalothin analog boronate, compound 4c (from PDB 1MX), and P. aeruginosa PDC-3 with the same inhibitor (320). The surfaces are colored by electrostatic potential. ote the differences in how the inhibitors sit in the active sites, particularly the R 1 amide group which is oriented towards the viewer in the E. coli structure, but towards the back of the binding pocket in the PDC-3 model. K i for PDC-3 from this work; K i for E. coli AmpC from (320). E. coli AmpC Crystal structure, PDB 1MX P. aeruginosa AmpC PDC-3 Molecular model Ser64 Tyr150 Asn152 Ser64 Tyr150 Ser318 Ala318 Asn343 Asn289 Asn343 Thr289 Arg349 K i = 1 nm K i = 4 nm 221

224 CHAPTER 5 Summary, Future Directions, and Lessons Learned Chapter 2 Summary The Role of SHV Asn276 in Clavulanate Resistance Inhibitor-resistant class A -lactamases of the TEM and SHV families that arise by single amino acid substitutions are a significant threat to the efficacy of β-lactam/βlactamase inhibitor combinations. To better understand the basis of the inhibitor-resistant phenotype in SHV, we performed mutagenesis to examine the role of a second-shell residue, Asn276. f the 19 variants expressed in Escherichia coli, only the Asn276Asp enzyme demonstrated reduced susceptibility to ampicillin/clavulanate (MIC increased from 50/2 to 50/8 μg/ml) while maintaining high-level resistance to ampicillin (MIC = 8192 μg/ml). Steady state kinetic analyses of Asn276Asp revealed slightly diminished k cat /K m for all substrates tested. In contrast, we observed a 1.5-fold increase in K I for clavulanate (1.12 ± 0.11 μm for Asn276Asp versus 0.72 ± 0.11 μm for SHV-1) and a 40% reduction in k inact /K I (0.013 ± μm -1 s -1 for Asn276Asp versus ± μm -1 s -1 for SHV-1). Timed ESI-MS of clavulanate-inhibited SHV-1 and SHV Asn276Asp showed nearly identical mass adducts, arguing for a similar pathway of inactivation. Molecular modeling shows that novel electrostatic interactions are formed between Arg244 2 and both 276Aspδ1 and δ2; these new forces restrict the spatial position of Arg244, a residue important in the recognition of the C 3 /C 4 carboxylate of β- lactam substrates and inhibitors. A rigid carbapenem (meropenem) was most affected by 222

225 the Asn276Asp substitution (46-fold increase in K i app versus SHV-1). We conclude that residue 276 is an important second-shell residue in class A β-lactamase mediated resistance to substrates and inhibitors, and only Asn is able to precisely modulate the conformational flexibility of Arg244 required for successful evolution in nature. Chapter 2 Future Directions ur study of the role of Asn276 in the SHV β-lactamase highlights the importance of second-shell residues and how they can affect substrate and inhibitor binding, despite relative distance from the active site. Maracino and colleagues recently published a paper exploring a similar concept in the TEM-1 enzyme (289). Replacement of the Arg244 residue with Ala resulted in an expected reduction of catalytic efficiency for substrates, but introducing an Arg at residue 220 or 276 partly recovered the lost catalytic activity. Their data suggests that while a positive charge facilitates the binding of the C 3 or C 4 β- lactam carboxylate, a certain amount of flexibility exists for the location of the Arg residue. Interestingly, the TEM Arg244Ala/Asn276Arg variant was also able to restore partly TEM Arg244Ala s loss of clavulanate susceptibility. We would like to expand on these inhibitor studies in the SHV enzyme. Specifically, we hypothesize that the decreased susceptibility of SHV Arg244 variants to clavulanate would be abrogated by an additional Asn276Arg substitution. However, the effect would likely be less pronounced than observed for TEM, consistent with the notion that the water molecule important for clavulanate inhibition is found elsewhere in the active site or recruited with the ligand upon binding. Such an exploration may shed additional insight onto the differences between the SHV and TEM enzymes. 223

226 We have also begun studies examining the role of SHV 234. As reviewed in Chapter 1, Mendonca and colleagues reported a new IR β-lactamase, SHV-72 (Ile8Phe, Ala146Val, and Lys234Arg), that displays increased catalytic efficiency for pencillins and resistance to clavulanate (305). The enzyme s optimal profile of maintaining hydrolytic ability for substrates and avoiding inhibition by clavulanate is more pronounced than observed in our SHV Asn276Asp variant. However, the molecular mechanisms have not yet been elucidated. We have constructed all of the SHV Lys234 variants and completed initial MIC testing. These susceptibility tests reproduce the Lys234Arg phenotype documented by Mendonca and colleagues, and reveal that the remaining variants are much less active against substrates and do not display inhibitor resistance. We are in the process of purifying the Lys234Arg enzyme and will study its kinetic characteristics against β- lactams and β-lactamase inhibitors. Molecular dynamics simulations of the SHV-72 variant suggest that 234Arg engages in a new hydrogen bond with Ser130, which causes reorientation of the Ser130 side chain (305). Presumably, this prevents the inactivating mechanism of cross-linking Ser70 and Ser130. We plan on testing this theory by screening the SHV Lys234Arg variant against investigational inhibitors that do not rely on the cross-linking mechanism for inactivation. These include the methylidene penems and L which undergo intramolecular capture for formation of a stabilized reaction intermediate. The crystal structure of SA2-13 in complex with SHV-1 showed a salt bridge with Lys234 (352). Screening with SA2-13 is planned, as we expect that the K I of this inhibitor would increase against SHV 234 variants. Furthermore, the boronate compounds are excellent molecular probes which will allow us to study the effect of the substitution on K i s for compounds bearing an R 2 group that should bind in the vicinity of the 234 residue. 224

227 Chapter 3 Summary Inhibition of ADC by Boronates and Carbapenems The need to develop β-lactamase inhibitors against class C cephalosporinases of Gramnegative pathogens represents an urgent clinical priority. To respond to this challenge, five boronic acid derivatives including a new cefoperazone analog were synthesized and tested against the class C cephalosporinase of Acinetobacter baumannii (ADC). The commercially available carbapenems antibiotics were also assayed. In this series, a chiral cephalothin analog with a meta-carboxyphenyl moiety corresponding to the C 3 /C 4 carboxylate of β-lactams showed the lowest K i (11 ± 1 nm). In antimicrobial susceptibility tests, this cephalothin analog lowered the ceftazidime and cefotaxime MICs of Escherichia coli DH10B cells carrying bla ADC from the resistant to susceptible range (16 4 μg/ml, and 8 1 μg/ml, respectively). Each carbapenem exhibited a K i app < 24 μm, and timed ESI-MS demonstrated the formation of adducts corresponding to acylenzyme intermediates with both intact carbapenem and carbapenem lacking the C 6 hydroxyethyl group. To better understand the interactions between the β-lactamase and the inhibitors, we constructed models of ADC as an acyl-enzyme with: (i) the metacarboxyphenyl cephalothin analog; and (ii) the two carbapenems imipenem and meropenem. ur first model reveals that this chiral cephalothin analog adopts a novel conformation in the β-lactamase active site. Further, the addition of the substituent mimicking the cephalosporin dihydrothiazine ring significantly improves K i for the ADC -lactamase. In contrast, the ADC: carbapenem models reveals a novel role for the R 2 side group in these potent inhibitors, and also suggests that elimination of the C 6 hydroxyethyl group by retroaldolic reaction leads to a significant conformational change of the acyl-enzyme. Lessons from the diverse mechanisms and structures of the boronic 225

228 acid derivatives and carbapenems provide insights for the development of new - lactamase inhibitors against these critical drug resistance targets. Chapter 3 Future Directions In collaboration with Dr. van den Akker s group, our laboratory is working on crystallizing the ADC β-lactamase both apo and soaked with the carbapenems. We are interested in elucidating the overall active site architecture of the enzyme and examining what reaction intermediates will be reflected by a crystallographic structure. Based on our molecular modeling with ADC and the carbapenems, we expect that that location of the β-lactam carbonyl will be influenced by the presence or absence of the inhibitor s C 6 hydroxyethyl group. Dr. Harris has provided much assistance on the kinetic study of ADC including presteady state kinetics. Based on the model: k 1 k 2 k 3, H 2 E + S E:S E-S E-P k -1 k -3 k 4 k -4 E + P we have established k 1, k -1, and k 2 microscopic rate constants for the ADC:CF hydrolytic reaction. Additional studies are planned to study whether the hydrolyzed product rebinds the enzyme at a detectable rate. By allowing the reaction to proceed in D 2, we can use mass spectrometry to analyze incorporation of deuterium into CF. Pending analysis of the data for the substrate reaction, we would like next to study an alternative substrate or inhibitor. ADC achieves very high catalytic rates for CF, so our subsequent plan is to examine the kinetics with a poor substrate such as the extendedspectrum cephalosporin cefepime or a carbapenem. ur ultimate goal is to use the 226

229 information about the reaction mechanism to inform the design of future inhibitors. This may take the form of prolonging k 2 or k 3. The boronate compounds have allowed us to pinpoint several important features for nm inhibition of AmpC enzymes (e.g., cephalosporin-like R 1 and presence of dihydrothiazine ring and carboxylate analogs on the R 2 group). However, the nature of their inhibition for both ADC and PDC-3 is not well understood. In the literature, these compounds are described as reversible inhibitors, yet the chiral compounds bearing R 2 groups have improved apparent K i s following preincubation with the enzyme. Incubation effects are typically characteristic of irreversible inhibition modes, and this time-dependent inhibition is not been observed for the achiral compounds. Thus, the incubation effect may reflect differences in how the chiral and achiral compounds interact with the β- lactamases, and some of the boronates may undergo an additional, slow binding step that induces a conformational change in the enzyme-inhibitor complex. More detailed analysis of these boronates may help elucidate the binding mechanisms, and contribute to the goal of inactivating ADC and/or PDC-3. Relevant studies include re-examining the boronate binding modes by studying all compounds for time-dependent inhibition of these β- lactamases. Addtionally, circular dichroism experiments may expose a conformational change that has not been detected by crystallography with the E. coli AmpC and boronates. Chapter 4 Summary Catalytic and Inhibitory Properties of the Pseudomonas aeruginosa AmpC: Implications for an Inhibitor-Resistant Phenotype osocomial infections caused by P. aeruginosa are on the rise, as is the incidence of strains resistant to multiple antibiotics. The limited treatment options and high rates of 227

230 mortality have driven this study of one of the primary P. aeruginosa resistance determinants, the AmpC β-lactamase. We demonstrate that even with low k cat values for penicillins and extended-spectrum cephalosporins, active site affinity and high AmpC expression levels can lead to MICs in the resistant range for a variety of β-lactam substrates. ur survey of inhibitors provides insight into the important features for inhibition of the PDC-3 enzyme. Specifically, boronates with R 1 and R 2 groups that mimic cephalosporins achieve nm K i s with PDC-3. Sulfones and penems with C 6 modifications (PSR-3-283a and the carbapenems), monobactams with a C 3 -C 4 bridge (BAL29880), and the non-β-lactam XL104 also achieve effective inhibition of the AmpC β-lactamase. Cefotaxime susceptibility is recovered by these investigational inhibitors in whole cell E. coli assays, but the continued challenge appears to be entry into the P. aeruginosa cell. In addition to studying ways to enhance cells accessibility, we have begun to explore the AmpC s intrinsic resistance to clavulanate by site-directed mutagenesis informed from molecular modeling. Chapter 4 Future Directions I will continue my work in the laboratory until late June when I return to medical school. During this time, I plan on studying primarily the Pseudomonal AmpC. Experiments already planned are outlined in Table 5-1, and several high priority analyses are further discussed below. An enhanced understanding of the differences between the PDC-1 and PDC-3 β- lactamases is an immediate focus. The role of the one amino acid change (Thr49Ala) is unclear, but previous studies have suggested this substitution can increase the substrate profile of the PDC-1 enzyme to include extended-spectrum cephalosporins and 228

231 carbapenems. The broadening of the cephalosporinase hydrolytic profile may reflect an inherent ability of these enzymes to increase resistance profiles and acquire novel catalytic properties with significant clinical effects. In our initial studies with cephalosporins, we have not been able to detect PDC-3 hydrolysis of ceftazidime or cefotaxime by monitoring change in UV spectra over time. We plan on expanding these studies to include additional substrates, such as the extended-spectrum cephalosporin cefepime and the carbapenems. Turnover experiments following timed incubation may allow us a better measure of hydrolysis for potentially slow substrates. We will also tease apart whether the increased β-lactam MICs noted for P. aeruginosa 18SH expressing PDC-3 (as opposed to P. aeruginosa PA1 expressing PDC-1) are due primarily to the overexpression of the AmpC in 18SH or to other changes in the genetic background of the strains. We have cloned both of these bla ampc genes into pbc SK(-) vectors and will easily test susceptibilities in a uniform background (as was done for the MIC testing of the investigational inhibitors). Furthermore, moving the bla PDC-1 and bla PDC-3 genes into a uniform P. aeruginosa background is even more desirable, as it would provide a system for testing how well different compounds permeate the P. aeruginosa cell. In this vein, we anticipate screening inhibitors with functional groups specifically designed to increase uptake through iron channels (e.g., siderophore monobactams and the sulfone L-1-255). We are also interested in futher study of how PSR-3-283a inactivates PDC-3. In particular, the possibility of a cross-linking will be explored by performing a tryptic digest of the -lactamase following incubation with PSR-3-283a. Analysis of the digest by ESI- MS will allow us to determine if the inhibitor is covalently attached to any enzyme residues, e.g., Ser64 and or Lys

232 Lessons Learned ext is a summary of the lessons learned from the consideration of β-lactamase inhibitors in this thesis and proposes several important features that must be regarded in order to optimize the development of β-lactamase inactivators. 1. High affinity for the active site of the target β-lactamase. It is axiomatic that in order for an inhibitor to work efficiently, it must exhibit high affinity for its target. This principle is emphasized by the development of clavulanate resistance in TEM-1 and SHV-1 β-lactamase mediated by substitution of amino acid residues 69, 130, 244, and 276, leading to increased inhibitor K I s and/or IC 50 s (85, 185, 428, 465). Particularly promising are those inhibitors that maintain high affinities across different groups of β-lactamases, such as the C 2 -substituted sulfone L-1-255, the methylidene penems, phosphonates and boronates which achieve nm inhibition of class A, C, and D (4, 30, 75, 130, 288, 320, 367, 466, 497). 2. Mimicking the natural substrate. The available inhibitors are effective primarily against class A serine β-lactamases, and the structures of class A β-lactamase inhibitors are similar to penicillin, bearing a fused four-five membered ring system. Similarly, our data on boronic acid derivatives suggests that class C β-lactamases have high affinity for inhibitors that resemble the natural cephalosporin substrates (Chapters 3 and 4) (130, 320). A guideline for inhibitor design may be: penicillin analogs for penicillinases, and cephalosporin analogs for cephalosporinases. This principle also ensures that the inhibition mechanism follows the substrate mechanism save for hydrolysis. While β-lactamases continue to develop resistance to all developed β-lactams, designing inhibitors that follow the hydrolytic mechanism may 230

233 help thwart the emergence of resistant isolates, as the enzyme must find novel ways to preserve function but evade inhibition. We hasten to add that this approach may not be practical in all cases. 3. Stabilizing interactions in the active site. Stabilizing or prolonging the acyl-enzyme intermediate is central to successful inactivation by the currently available class A inhibitors (93, 445). Promising novel inhibitors, such as the methylidene penems (e.g., BRL 42715) and L are designed to capitalize on and enhance this stabilization. ne approach is to prolong the acylation step (k 2 ). Alternatively, acylenzyme intermediates can be stabilized by multiple interactions in the active site, as observed for SHV-1 with tazobactam and SA2-13 (351, 352). The inhibition of mc- A by the monobactam derivative 13 is enhanced by the incorporation of chemical components that confer structural flexibility to the inhibitor, allowing the enzymesubstrate species to adopt multiple conformations and trap the enzyme in energy minima unfavorable for hydrolysis (322, 351, 500). Another viable approach is designing inhibitors that induce conformational changes in the enzyme or displace key water molecules important for hydrolysis, such as seen with the large sevenmembered intermediate formed from the methylidene penems (72, 339, 483). 4. Reaction chemistry that slows deacylation and favors inactivation over inhibitor hydrolysis. Features of an inhibitor which slow deacylation and drive towards terminal inactivation, such as acyl-enzyme stabilization (see above) or the presence of a good leaving group at C 1, contribute to effective inhibition. Inhibitors that can delay deacylation for longer than minutes (k off rate of sec -1 ) may outlast the bacterial generation time, allowing the partner β-lactam time to disrupt the cell wall synthesis necessary for cell division (445, 464). For example, the bridged 231

234 monobactams (such as BAL29880) delay deacylation by limiting the C 3 -C 4 rotation necessary for approach of the hydrolytic water in class C enzymes (184, 277). Additionally, relatively minor changes introduced by the acylation of meropenem by SHV-1, including minor displacements to active site residues (overall 0.29 Å rmsd), and a restructured hydrogen bonding network, lead to persistent complex formation (340). 5. Rapid cell penetration. β-lactamases are typically found in the periplasmic space of Gram-negative bacteria, and the rapidity with which inhibitors can access their target is a prerequisite for successful inhibition. Certain β-lactams are zwitterionic and rapidly penetrate the bacterial outer membrane (e.g., cefepime). The same principle extends to the inhibitors, e.g., the enhanced cell entry of L and the novel monobactams BAL19764 and BAL30072 is likely due to uptake through siderophore channels (87, 360). As highlighted in Chapter 4, novel agents with stellar kinetic profiles may not always successfully enter cells with decreased permeability. Attention must also be paid to designing permeable drugs that are not hindered as they pass through porin channels or quickly returned into the medium by multidrug efflux pumps, an important resistance determinant independent of β-lactamases (333). 6. Low propensity to induce β-lactamase production. Successful inhibitors need to avoid inducing expression of the enzymes they are designed to inhibit, particularly among AmpC β-lactamases. Examples of this approach include the non-β-lactam inhibitors with sulfonamide cores designed to evade hydrolysis as well as avoid induction of AmpC enzyme expression (469). 7. Identification of measureable biological correlates of β-lactamase inhibition in the cell. Regardless of how attractive an inhibitor s laboratory characteristics may be, 232

235 e.g., nm K i values, low IC 50, high k inact s, or low MICs, a successful inhibitor must prove its worth in the clinical setting. In vivo efficacy requires an integration of many complex features including points 1-6 above, partnership with an appropriate β- lactam, sufficient serum and periplasmic concentrations, and factors less well understood (e.g., serum protein binding, individual clearance rates). Many microbiological properties (time kill, mutant prevention concentration, and synergy) are important parameters that need to be considered. There is no certain consensus on how best to identify and measure these attributes in the early research of candidate inhibitors. Previous studies with β-lactam antibiotics have demonstrated that MICs may be poorly correlated with the sensitivity of PBPs for inactivation (254). Finding reliable biological correlates of inhibition is a substantial and relevant challenge, and our current methodological outcome measurements may benefit from reexamination in this vein. 8. In the meantime, use what we have. While scientists and physicians continue struggling to stay ahead, or at least abreast, of the abilities of β-lactamases, we must utilize intelligently and productively the agents are currently available. For example, we can commit to applying the research that demonstrates the potential of novel combinations of β-lactam-β-lactamase inhibitors, such as clavulanate with cefepime, cefpirome, or meropenem, to expand and enhance inhibitors abilities to protect partner β-lactams (206, 278). Several significant hurdles stand in the way of these novel combinations, such as corporate relationships impacting the funding of the trials necessary to demonstrate clinical efficacy, as well as optimizing the pharmacokinetics, pharmacodynamics, and safety of the components. The β- lactamase inhibitory activity of carbapenems should not be overlooked in the face of 233

236 increasing β-lactam resistance. Carbapenems, for the time being, remain effective against many difficult-to-treat infections, and teach us that the distinction between an inhibitor and a slow substrate may be blurred. Data presented in both Chapters 3 and 4 argues strongly for the use of that the carbapenems as both broad-spectrum - lactams and -lactamase inhibitors for difficult to treat pathogens like A. baumannii and P. aeruginosa. A Perspective From our vantage point, the main challenge in β-lactamase inhibitor development is discovering novel inhibitors with activity against a broad spectrum of inhibitorsusceptible and -resistant enzymes from multiple classes (18, 464). ew inhibitors must also target carbapenemases (serine and metallo) as carbapenems are still the most potent β-lactams available for clinical use. Future advances in inhibitor design against versatile β-lactamases must incorporate strategies that address each of the key structural features of these diverse proteins, keeping in mind that mechanism-based inhibitors may engage in unique reaction chemistry with β-lactamases. Compounds with the chemical features of the ideal inhibitor, such as enhanced permeability through the cell membrane, affinity for the active site of the β-lactamase, formation of intermediates that "trap" the enzyme, displacement of critical water molecules, and prolonged time to enzyme recovery, many of which are illustrated by the investigational complexes highlighted herein, may make a significant impact on the development of second-generation inhibitors that target resistant β-lactamases. We also acknowledge the magnitude of this challenge. In nearly 40 years of targeted research on β-lactamase inactivation, new inhibitors have been only slowly introduced 234

237 into clinical trials. There have been many starts, but few agents cross the finish line. This is a testament to the versatility and complexity of β-lactamases, and the incredible evolutionary ability of the organisms harboring them. Is the perfect β-lactamase inhibitor an unattainable goal? Perhaps. Yet, we anticipate with excitement the achievements in the next three decades, and suspect better chemistries and combinations will ultimately be found. 235

238 Table 5-1. Experiments planned or in progress for P. aeruginosa AmpC Hypothesis Thr79Ala plays a role in PDC-3 hydrolysis of extended-spectrum cephalosporins and carbapenems The overexpression of AmpC in P. aeruginosa 18SH is due to changes in AmpC regulatory genes, and these can be correlated to protein expression levels There are differences in the genetic backgrounds of PA-1 and 18SH Describing microscopic rate constants of substrates and inhibitors can inform selective design of inhibitors XL104 and PSR-3-283a undergo unique inactivation chemistry with PDC-3 We can improve the permeability of β- lactam/β-lactamase inhibitor combinations Elucidation of the structural features of PDC-3 will reveal a relatively flexible enzyme PDC-3 Arg349 plays a role in clavulanate resistance 236 Specific experiments 1. Complete MIC testing with extended-spectrum cephalosporins and carbapenems in uniform genetic background (both E. coli and P. aeruginosa) 2. Purify PDC-1 and characterize substrate and inhibitor kinetics 1. Use panel of clinical P. aeruginosa strains available in lab for sequencing of regulatory genes, e.g., ampd, ampr 2. RT-PCR of ampc mra in strains, including after cefoxitin induction 3. Correlate above with whole cell protein expression levels as detected by immunoblotting 4. Estimate concentration of AmpC β-lactamase cell periplasm 1. PCR for efflux pump and porin genes, e.g., MexAB-prM, prd 2. PCR for other β-lactamase genes, e.g., bla PER, bla XA, bla VIM 3. aief of PA-1 to look for β-lactamase protein expression 1. Characterize substrate kinetics hydrolysis of extended-spectrum cephalosporins including cefepime 2. Look for evidence of conformational change (e.g., non stochiometric bursts) with certain substrates as noted in (356) 3. Study inhibitor kinetics, especially boronates for binding modes and possible conformational change 1. ESI-MS of PDC-3 and XL Tryptic digest of PDC-3 and PSR-3-283a 1. Repeat all whole cell MIC assays: include inhibitor alone (boronates, PSR-3-283a, XL104, or BAL29880 have no activity alone in disc assays); and try new partner β-lactam 2. Add MIC studies of inhibitors with side chains that may enhance uptake, e.g., L-1-255, BAL btain P. aeruginosa strain which can be used for cloning bla ampc, e.g., P. aeruginosa KG2505 Δ bla ampc (418) 1. rder P. fluorescens strain to purify and study cold-loving β- lactamase, compare to PDC-3 substrate and inhibitor profile 2. Explore possibility of MR studies to probe protein flexibility 3. Provide van den Akker lab with more PDC-3 protein purified from RP codon+ cells for purification (in progress) 1. Purify PDC Arg349Ala (in progress) 2. Kinetic analysis of variant with substrates and clavulanate 3. ESI-MS to see if detectable covalent adducts formed

239 APPEDIX A Molecular Modeling Terms Molecular Dynamics Molecular Mechanics CHARMm SHAKE Steepest Descent A computational method that calculates the time dependent behavior of a molecular system and provides detailed information on the fluctuations and conformational changes of proteins Modeling system based on ewtonian mechanics that applies a force field to minimize potential energies, providing detailed structure and physical properties of a molecule (e.g., dipole moment) Set of force fields for molecular dynamics and also the name of the commercial molecular mechanics and dynamics program available through Accelrys An algorithm which sets bond geometry constraints during molecular dynamics simulations, especially important for constraining the energy and force of high-frequency motions involving bonds containing hydrogens A relatively simple algorithm to find the local energy minimum in a potential energy surface Conjugate Gradient Periodic Boundary Conditions Particle Mesh Ewald VT Solvent LibDock Another algorithm for minimization of a biomolecular structure which improves convergence as compared to Steepest Descent by using history of previous minimization steps Set of conditions allowing the simulation of proteins in a solvent by representing the system as a unit cell Method for calculating interaction energies (e.g., long range electrostatics) of a periodic system Molecular dynamics run at constant moles (), volume (V), and temperature (T) As it applies to the influence of solvent on the structure, dynamics and thermodynamics of biological molecules, may be implicitly or explicitly simulated An algorithm allowing the site-directed docking of ligands into biomolecules (protein) a This appendix is not meant to be a comprehensive review of molecular modeling tools and programs, but rather a general guide for the reader. The information was gathered from the Help menu of the Accelrys Studio 2.1 software (Accelrys, San Diego, CA), where more detailed descriptions are available. 237

240 APPEDIX B Penicillin Sulfone Inhibitors of Class D β-lactamases Reproduced in part with permission from PEICILLI SULFE IHIBITRS F CLASS D β-lactmases Sarah M. Drawz, Christopher R. Bethel, Venkata R. Doppalapudi, Anjaneyulu Sheri, Sundar Ram Reddy Pagadala, Andrea M. Hujer, Marion J. Skalweit, Vernon E. Anderson, Shu G. Chen, John D. Buynak, and Robert A. Bonomo Antimicrobial Agents and Chemotherapy January 19, 2010; doi: /aac The previous chapters focused on the inhibition of class A and C β-lactamases. In Appendix B, we present work that is recently published on the inhibition of class D β- lactamases. This paper was not included as one of my formal thesis projects, but the exercise allowed me to explore the challenge of inhibition in another group of enzymes with unique features. Completing the work was a productive and valuable experience, and thus is included for the committee s reference and interest. Abstract XA -lactamases are largely responsible for -lactam resistance in Acinetobacter spp. and Pseudomonas aeruginosa, two of the most difficult to treat nosocomial pathogens. In general, the -lactamase inhibitors used in clinical practice (clavulanic acid, sulbactam, and tazobactam) demonstrate poor activity against class D -lactamases. To overcome 238

241 this challenge, we explored the efficacy of -lactamase inhibitors of the C 2/3 -substituted penicillin and cephalosporin sulfone families against XA-1, extended-spectrum (XA- 10, -14 and -17), and carbapenemase-type (XA-24/40) class D -lactamases. Three C 2 - substituted penicillin sulfone compounds (JDB/L-1-255, JDB/L-III-26, and JDB/ASR-II-292) showed low K i values for the XA-1 -lactamase (0.70 ± 0.14 to 1.60 ± 0.30 μm), and demonstrated significant improvements compared to the C 3 -substituted cephalosporin sulfone (JDB/DVR-II-214), tazobactam, or clavulanic acid. The C 2 - substituted penicillin sulfones JDB/ASR-II-292 and JDB/L also demonstrated low K i s for XA-10, -14, -17, and -24/40 -lactamases (0.20 ± 0.04 to 17 ± 4 M). Furthermore, JDB/L displayed stochiometric inactivation of XA-1 (t n or k cat /k inact = 0) and t n s ranging from 5-8 for the other XA enzymes. Using mass spectroscopy to study the intermediates in the inactivation pathway, we determined that JDB/L inhibited XA -lactamases by forming covalent adducts that do not fragment. Based upon substrate and inhibitor kinetics of XA-1, we constructed a model showing that the C 3 carboxylate of JDB/L interacts with Ser115 and Thr213, the R 2 group at C 2 fits between the space created by the long B9 and B10 strands, and stabilizing hydrophobic interactions and formed between the pyridyl ring of JDB/L and Val116 and Leu161. By exploiting conserved structural and mechanistic features, JDB/L is a promising lead compound in the quest for effective inhibitors of XA type -lactamases. Introduction 239

242 The complexity and diversity of -lactamases (EC ) is largely determined by the unique manner in which the various amino acids define the active site of these enzymes. XA -lactamases are class D enzymes (Bush Jacoby Group 2bf) that are structurally related to the class C penicillin binding proteins (64, 68, 292, 517). f primary importance, XA -lactamases are found in some of the most difficult to treat nosocomial pathogens (Pseudomonas aeruginosa and Acinetobacter spp.) (56, 186, 490). In these pathogens, XA -lactamases may confer resistance to penicillins, cephalosporins, and carbapenems (6, 133, 151, 170, 327, 394). The bla XA genes encoding resistance to -lactams are located in the chromosome, on plasmids, or in integrons, and may be inducible (118, 133, 171, 327, 389, 393). Currently, there are more than 140 different XA -lactamases reported ( XA-1 is a penicillinase, found in Escherichia coli, Klebsiella pneumoniae, and P. aeruginosa, and exists as a monomer (447). Among the most studied XA enzymes are XA-10, a dimer, and its clinically important derivatives, XA-14, and -17. These latter three enzymes confer resistance to ceftazidime and are regarded as extended-spectrum -lactamases (ESBLs) of the class D family ( ). XA-14 differs from XA-10 by the substitution Gly167Asp, and XA-17 has the same amino acid sequence as XA-10 with an Asn76Ser substitution. The major XA enzymes that confer resistance to carbapenems include XA-23, -24/40, -48, -51, and -58 (44, 126, 194, 284, 390, 394). XA carbapenemases are mostly found in Acinetobacter baumannii (377). The growing number of XA lactamases found in nature generates considerable interest in both understanding the mechanistic basis of resistance to inactivation and developing effective inhibitors (30). When found in clinical isolates, 240

243 XA lactamases are poorly inhibited by the currently available -lactam/ -lactamase inhibitor combinations (ampicillin/sulbactam, amoxicillin/clavulanic acid, ticarcillin/clavulanic acid, and piperacillin/tazobactam) (49, 67, 68, 327, 392). For the XA-1 lactamase, the affinity of tazobactam is reduced, and the turnover of the inhibitor is significantly elevated (30). However, little is known about the inactivation kinetics of other XA -lactamases with experimental inhibitors (4, 286, 300, 379). In the quest for new inhibitors, Buynak and co-workers designed and synthesized C 2 - substituted 6-alkylidene penicillin sulfones and C 3 -substituted 7-alkylidene cephalosporin sulfones as mechanism-based inactivators of class A -lactamases (24, 71, 74, 367). In general, these -lactamase inhibitors derive their success from high affinities for the active site and ability to form stable reaction intermediates (352, 367). The aim of this present work is to determine the relative efficacy of C 2/3 -substituted 6/7-alkylidene penicillin and cephalosporin sulfones as -lactamase inhibitors of XA-1, XA-10, -14, and -17, and 24/40 -lactamases. In contrast to what has been determined in the inactivation of class C CMY -lactamase and class A SHV and TEM -lactamases by mechanism-based or suicide inhibitors (clavulanic acid, tazobactam, and sulbactam, Figure B-1, compounds 1-3), we show that C 2/3 -substituted penicillin and cephalosporin sulfone inhibitors form a covalent adduct that undergoes unique reaction chemistry and does not fragment (55, 445, 465). This behavior may prove to be an important characteristic of successful -lactamase inhibitors against class D enzymes. Materials and Methods Chemical synthesis 241

244 The chemical structures of Pencillin G, ampicillin, oxacillin, cephaloridine, and nitrocefin are shown in Figure B-1 (compounds 4-8). The chemical structures of the C 2/3 - substituted sulfone -lactamase inhibitors tested in this study are also illustrated in Figure B-1, compounds The synthesis and initial evaluation of compounds 9-12 were reported and reviewed by Buynak and co-workers (74-76). Bacterial strains and plasmids The bla XA-1 gene was cloned from plasmid R G238 into the vector pet 12a (+) (ovagen, Gibbstown, J), as described previously (447). Plasmid R G238 containing bla XA-1 was maintained in E. coli DH10B cells (Invitrogen, Carlsbad, CA). For protein purification, E. coli BL21(DE3) cells (ovagen) were transformed with the pet 12a (+) bla XA-1 plasmid by electroporation. Using recombinant plasmid pxa-40 obtained from ordmann and co-workers as a template (195), bla XA-24/40 was amplified with primers containing 5 dei and 3 BamHI restriction sites. The primers were designed to amplify the bla XA-24/40 gene including start and stop codons for the mature protein, without including the upstream leader sequence. The PCR product was subcloned directionally into pet 24a (+) (ovagen) and transformed into E. coli DH10B cells. Following sequence verification of the construct, the plasmid was transformed into E. coli BL21(DE3). This plasmid was used to express and purify XA-24/40 -lactamase. Protein purification Preparation of XA-1 and XA-24/40 -lactamases was performed by inducing E. coli BL21(DE3) cells containing either the pet 12a (+) or pet 24a (+) vector with the cloned 242

245 bla XA-1 or bla XA-24/40, respectively (30). Briefly, 500 ml culture flasks containing cells were induced at D 600 of 0.8 with isopropyl- -D-galactopyranoside (IPTG) to a final concentration of 0.2 mm. The culture was harvested 3-6 hours after IPTG induction, centrifuged, frozen at -20 C, and the next day resuspended in 50 mm Tris HCl buffer, ph 7.4. Lactamase was liberated using lysozyme and EDTA (39, 209). All cell manipulations to isolate XA lactamases were done in 50mM sodium phosphate [monobasic and dibasic] buffer, ph 7.2. XA-1 and XA-24/40 -lactamases were purified by preparative isoelectric focusing (209). We assessed the purity of each preparation by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE). Samples were resolved on a 5% stacking 12% separating SDS-PAGE gel and were stained with Coomassie Brilliant Blue R250 (Fisher, Pittsburgh, PA) to visualize proteins. Further purification of XA-1 and XA-24/40 -lactamases was performed using size exclusion chromatography with a Pharmacia ÄKTA Purifier System (GE Healthcare, Piscataway, J). We employed a Hi Load 16/60 Superdex 75 column (GE Healthcare) and eluted with 20 mm phosphate buffered saline, PBS, (ph 7.4). Protein was concentrated and dialyzed with 50 mm sodium phosphate buffer, ph 7.2. Protein concentrations were determined with Bio-Rad s Protein Assay using bovine serum albumin (Sigma, St. Louis, M) as a standard. XA-10, -14, and 17 -lactamases were purified to homogeneity according to the method described by Danel and co-workers, and were kind gifts from Professor M. G. P. Page (114, 115). Kinetic parameters 243

246 Kinetic constants of the XA lactamases were determined by continuous assays at room temperature using an Agilent 8452 diode array spectrophotometer (Agilent, Palo Alto, CA) (185). Each experiment was performed in triplicate in 50 mm sodium phosphate buffer supplemented with 20 mm bicarbonate (174). itrocefin (CF) was purchased from Becton Dickinson (Cockeysville, MD), and oxacillin, ampicillin, and cephaloridine from Sigma. Clavulanic acid and tazobactam were generous gifts from GlaxoSmithKline (Surrey, United Kingdom) and Wyeth Pharmaceuticals (Pearl River, J), respectively. The kinetic parameters, V max and K m, were determined by measuring the hydrolysis of CF ( 482 = 17,400 M -1 cm -1 ), oxacillin ( 263 = 258 M -1 cm -1 ), ampicillin ( 235 = -900 M -1 cm -1 ), and cephaloridine ( 260 = -10,000 M -1 cm -1 ), and obtaining the non-linear least squares fit of the data to the Henri-Michaelis-Menten equation (Equation 1) using the program Enzfitter (Biosoft Corporation, Ferguson, M): v = (V max [S]) / (K m + [S]) Equation 1 The reaction between the -lactamase (E) and mechanism-based inhibitors (I) studied in this paper can be represented by the follow equation: In this model, E:I represents the formation of the preacylation complex and E-I, the acyl-enzyme species. The acyl-enzyme (E-I) can proceed to hydrolysis (E + P 1 ) or undergo rearrangement to a transiently inhibited species (E-I T ). The E-I T intermediate 244

247 may then return to E-I, proceed to hydrolysis (E + P 2 ), or form an inactivated acyl-enzyme (E-I*). The rate constants, k, describing each of these steps are represented in Equation 2. We determined the apparent K i for the inhibitors using competition assays that employed enzyme, the reporter substrate CF, and each inhibitor. We measured initial velocities (v 0 ) in the presence of a constant concentration of enzyme (11 nm) and increasing concentrations of inhibitor (ranging from 4 μm - 9 mm, depending on the inhibitor and enzyme) that were competed against 50 μm of CF. In these competition assays, the K i value approximates K m, and initial velocities can be represented by the following equation: v 0 = (V max [S]) / (K m (1 + I/K i ) + [S]) Equation 3 To determine the apparent K i, the initial velocities immediately after mixing were plotted, 1/velocity, as a function of inhibitor concentration, fitted to a linear equation, and the value of K i app determined by dividing the y-intercept by the slope of the line. The first-order rate constant for enzyme and JDB/L complex inactivation, k inact, was obtained by monitoring the reaction time courses in the presence of inhibitor and CF until the reaction reached the final steady state. Fixed concentrations of enzyme (11 nm) and CF (100 μm) and increasing concentrations of JDB/L ( μm, depending on the enzyme) were used in each assay. The k obs was determined using a nonlinear least squares fit of the data using rigin 7.5 (riginlab, orthampton, MA): A = A 0 + v f t + (v 0 -v f )[1-exp (-k obs t)] / k obs Equation 4 Here, A is absorbance, v 0 (expressed in variation of absorbance per unit time) is initial velocity, v f is final velocity, and t is time. Each k obs was plotted versus I and fit to determine k inact as the maximum asymptote. 245

248 The turnover number, t n (i.e, partitioning of the initial enzyme inhibitor complex between hydrolysis and enzyme inactivation, k cat /k inact ), was derived by incubating increasing amounts of inhibitor with a fixed concentration of β-lactamase. After 24 hours, an aliquot was removed from the mixture and the initial velocity was measured and compared with a control sample with no inhibitor added. The ratio of inhibitor to β- lactamase (I:E) that resulted in greater than 90% inactivation after 24 hrs was defined here as the t n (69). Minimum Inhibitory Concentration (MICs) The susceptibility of E. coli DH10B cells harboring plasmid R G238 bla XA-1 was determined by MICs performed in lysogeny broth agar using a Steers Replicator that delivered 10 μl of a diluted overnight culture containing 10 4 colony forming units (28, 29). Piperacillin was purchased from Sigma. Tazobactam and JDB/L were tested at a constant concentration of 4 μg/ml in combination with increasing concentrations of piperacillin. Electrospray ionization mass spectroscopy (ESI-MS) For mass spectrometry studies, we incubated 20 μm of each XA enzyme for 15 min with and without JDB/L at an I:E ratio of 30:1. Each reaction was terminated by the addition of 0.1% trifluoroacetic acid and immediately desalted and concentrated using a C 18 ZipTip (Millipore, Bedford, MA) according to the manufacturer s protocol. Spectra of XA alone and XA: JDB/L proteins were generated on a Q-STAR XL Quadrupole Time-of-Flight mass spectrometer (Applied Biosystems, Framingham, MA) equipped with a nanospray source. Experiments were performed by diluting the 246

249 protein sample with 50% acetonitrile/0.1% trifluoroacetic acid to a concentration of 10 μm. This protein solution was then infused at a rate of 0.5 μl/min and the data were collected for 2 min. Spectra were deconvoluted using the Analyst program (Applied Biosystems). All measurements on the Q-STAR have an error of ± 4 m/z. Ultraviolet Difference (UVD) Spectroscopy Per previously published methods, UVD spectra were obtained for JDB/L reacted with XA-1 -lactamase at a 1000:1 inhibitor to enzyme ratio (50 M: 0.05 M) for 30 min (367, 445). Wavelengths from nm were measured. Molecular modeling To create overlays of -lactamase structures, the Research Collaboratory for Structural Bioinformatics Protein Data Bank (PDB) entries for TEM-1 (1ZG4), XA-1 (1M6K), and PC-1 (3BLM) were manipulated with the Sybyl molecular modeling program (Tripos Inc., St. Louis). We also created models of XA-1 with oxacillin, cephaloridine, and the inhibitor JDB/L docked in the active site. These studies utilized the docking protocol FlexX module (BioSolveIT), within the Sybyl program. The β-lactam carboxylate is a key ligand recognition point which is held in place primarily by the side chain hydroxyl groups of Ser115 and Thr213, as well as by its proximity to Lys212. As the FlexX program automatically sets all (Ser and Thr) side chain C-C--H torsion angles at 180, we manually adjusted the torsional angle of the hydroxyl side chains of Ser and Thr residues in close contact with the β-lactam carboxylate. This allowed us to obtain the most ideal fit of the antibiotic as judged by 247

250 the proximity of the -lactam carbonyl oxygen (of the docked antibiotic) into the oxyanion hole pocket formed by the backbone -H s of Ser67 and Ala215. Accordingly, the torsional angle of Thr213 was set to 90 to provide a logical conformation where the positive end of the -H dipole faces the incoming carboxylate of the antibiotic. Ser67 is the active site nucleophile and its side chain oxygen partially negatively charged. Thus, the C-C--H torsional angle of this residue was set at -30 to allow the positive end of the -H dipole moment to face the negatively charged C - 2 of carboxylated Lys70, in a position for the proton to be abstracted. Further, at this angle, the negative end of the Ser67 -H dipole faces the -lactam carbonyl carbon preparing for acylation. The C-C- -H torsional angle of the remaining side chain hydoxylic residue, Ser115, was left at 180, since this angle positions the proton toward the -lactam carboxylate. Results XA β-lactamase substrate kinetics XA-1, -10, -14, -17, and -24/40 -lactamases were purified to greater than 95% homogeneity. XA-1 s high k cat for oxacillin translated to an elevated catalytic efficiency (k cat /K m = 19 ± 1 μm -1 s -1, Table B-1), while for ampicillin, the catalytic efficiency was 3-fold lower. The k cat /K m of XA-1 for CF was 10 ± 3 μm -1 s -1, and 45-fold lower for cephaloridine. As the substrate profiles of XA-10, -14, -17, and -24/40 were previously reported, we focused upon the behavior of CF as this cephalosporin served as our indicator substrate (113, 114, 174, 195). The catalytic efficiencies of CF for the ESBL XAs (XA-10, - 248

251 14, and -17) and XA-24/40 ranged from 4.0 ± 0.1 to 9.6 ± 0.3 μm -1 s -1 (see Table B-2 legend). Previous studies reported biphasic kinetic profiles for some XA enzymes and particular substrates, and while this behavior is not completely understood, it appears related to the equilibrium of enzyme monomer/dimer species in solution as well as concentration of bicarbonate (113, 174, 259, 353). Under the conditions tested, we did not observe biphasic kinetics with the XA enzymes against any of the tested substrates. ur assays used a nm concentration of enzyme that is likely well below the dissociation constant for any dimer species, and a relatively high concentration of supplemental bicarbonate (20 mm). These experimental conditions may offer an explanation for the lack of biphasic kinetic behavior. Inhibition of XA-1 β-lactamase Each of the C 2/3 -substituted sulfone compounds had a lower K i app for the XA-1 β- lactamase than clavulanic acid and tazobactam (Table B-2). The C 3 cephalosporin inhibitor DVR-II-214 had the highest K i app of the panel, 57 ± 5 μm, while the C 2 penicillin inhibitors showed K i app s < 2 μm (JDB/L-III-26 = 1.60 ± 0.30 μm, remaining data shown in Table B-2). JDB/L demonstrated 830- and 270-fold lower K i s than tazobactam and clavulanic acid, respectively, for XA-1. Determination of 24 hr t n values for JDB/L against XA-1 revealed stochiometric inactivation of the β- lactamase (t n = 0) (Table B-3). These values are compared to t n s > 100 for both clavulanic acid and tazobactam. While the first order rate constant, k inact, of JDB/L for XA-1 (0.094 ± s -1 ) was not significantly different than the previously determined k inact for tazobactam (0.12 ± 0.01 s -1 ), the decreased K i app of this novel inhibitor yields an 249

252 89-fold increase in inhibitor efficiency (k inact /K i app = ± versus ± M -1 s -1 ) (30). Inhibition of ESBL XA-10, -14, and -17 β-lactamases Based on our kinetic data with XA-1, we decided to focus our attention on the C 2 - substituted sulfones JDB/ASR-II-292 and JDB/L against the ESBLs XA-10, - 14, and -17. Table B-2 summarizes our data showing that these investigational inhibitors demonstrated decreases in K i app s for the XA enzymes compared to clavulanic acid and tazobactam, from 6 to 225-fold and 10 to 200-fold, respectively. verall, the K i app s of the sulfones were nm to low M. As shown in Table B-3, the k inact /K i app ratios of JDB/L for these ESBL XAs varied from ± M -1 s -1 for XA-10 to 1.51 ± 0.34 M -1 s -1 for XA-17. This 84-fold difference is due primarily to JDB/L s larger K i app for XA-10. However, the JDB/L t n remains low for XA-10, as well as for the other ESBL-type enzymes (t n s = 6-8). Inhibition of the carbapenemase XA-24/40 β-lactamase The C 2 -substituted penicillin sulfones, JDB/ASR-II-292 and JDB/L-I-255, demonstrated low K i app values for XA-24/40, 2.4 ± 0.4 and 0.65 ± 0.05 M, respectively. f all the -lactamases studied, only XA-17 has a higher JDB/L k inact /K i app than XA-24/40. Susceptibility testing 250

253 -lactamase inhibitors with low K i app s and t n s are clinically useful only if they can protect partner -lactam antibiotics from inactivation by -lactamase enzymes. Based on the kinetic profile of the C 2 -substituted penicillin sulfone inhibitors, as well as previously published data demonstrating the efficacy of JDB/L with class A -lactamases, we chose to compare the in vitro activity of JDB/L to tazobactam (367). We reasoned that the unique C 2 - catechol moiety of JDB/L may facilitate the uptake of the inhibitor by bacterial siderophore (iron) channels (71, 367). Against E. coli DH10B cells harboring plasmid R G238 bla XA-1, susceptibility tests revealed that piperacillin with 4 μg/ml of JDB/L was more potent than piperacillin combined with 4 μg/ml of tazobactam (MICs of 128 μg/ml compared to 512 μg/ml, respectively). Mass spectroscopy studies To gain insight into the nature of the inhibition reactions and the intermediates that are generated, we performed ESI-MS of each XA -lactamase after 15 min incubation with JDB/L at an I:E of 30: 1. ESI-MS of each XA -lactamase alone showed a mass peak corresponding to the predicted molecular mass of each apo-enzyme within experimental error (± 4 m/z) (mass data summarized in Table B-4; representative spectra for XA-1 and XA-10 shown in Figure B-2). For the enzyme and inhibitor assays, we observed a species corresponding to the covalent addition of the -lactamase inhibitor to XA-1, -10, -14, -17 and -24/40. nly the XA-10 + JDB/L spectra showed an additional peak corresponding to the molecular mass of the apo-enzyme, suggesting that not all of the enzyme was covalently inhibited at the 15 min timepoint. This result is 251

254 consistent with the kinetic data revealing lower inhibitor efficiency of the JDB/L compound for XA-10 as compared to the other XA -lactamases. UVD Spectroscopy UVD spectroscopy can be used to provide insight into the presence of intermediates formed during the reaction of an inhibitor and a -lactamase (69, 90, 367, 445). In our study of JDB/L and XA-1 reacted at an I:E of 1000:1, we observe the formation of two chromophores at nm and nm. Each of these chromophores develops by 12 s, and continues to increase in intensity up to 30 min (Figure B-3). In accordance with previous data on this inhibitor with the class A -lactamases, we tentatively assign the chromophore at nm to the reaction intermediate containing a bicyclic aromatic ring system (see below). It is also possible that these UVD spectra represent other chemical changes to the enzyme-inhibitor complex. Molecular modeling After consideration of our substrate and inhibitor profiles, we chose to focus our molecular modeling on XA-1, a monomer. To this end, we selected a good substrate (oxacillin), a poor substrate (cephaloridine), and JDB/L-1-255, the overall good inhibitor. Using our kinetic data to rationalize our models, we propose a mechanism by which JDB/L inhibits this enzyme. To start, we studied the structure of XA-1 in comparison to the class A enzymes TEM-1 and the Staphylococcal -lactamase PC1. verlaying XA-1 with TEM-1 and PC1 elucidated significant structural differences which may be related to the substrate 252

255 profiles of classes A and D. oticeably, the turn between the B9 and B10 -strands is significantly longer in the case of the XA-1 enzyme as compared to both TEM-1 and PC1. This results in an effective widening of the active site, since, as the turn begins in the class A -lactamases the (somewhat displaced) side chains of residues Glu240 (TEM- 1) and especially Ile239 (PC1) become angled back toward the active site serine and significantly restrict the substrate binding site (Figure B-4). By contrast, the closest corresponding residue of XA-1 (Phe217) is oriented away from the site, thus providing additional space for recognition of sterically bulky substrates, such as oxacillin. The entry and binding of oxacillin (compound 6) in the active site of XA-1 is depicted in Figure B-5A. The carbonyl oxygen of the -lactam is positioned in the oxyanion hole, and both oxygens of the sp 3 -hybridized C 3 carboxylate interact with Ser115 and Thr213. In this orientation, the bulky hydrophobic phenyl group at C 3 of the isoxazole substituent creates favorable interactions with the side chains of Trp102 and Met99. Such interactions of residues on this loop would not be possible with penicillin G (compound 4), nor is the loop of the TEM-1 and PC1 class A -lactamases properly positioned to interact with this substrate, thus potentially explaining the higher hydrolytic efficiency of oxacillin versus penicillins for XA-1. In contrast, the carboxylate of cephaloridine (compound 7) does not appear to achieve optimal interactions with both Ser115 and Thr213 of XA-1 (Figure B-5B) due to the relatively flat geometry of cephaloridine imposed by the sp 2 -hybridized C 4 of the cephem. The docked cephaloridine molecule suggests that additional hydrophobic residues at C 3 (as is the case with CF, a better substrate of XA-1) might favorably interact with a number of residues in this part of the site. 253

256 The modeling of these XA substrates helped us to better understand the affinity contributions of JDB/L modeled in the active site of XA-1 (Figure B-6). In our model, the C 3 carboxylate of JDB/L interacts with Ser115 and Thr213. This also allows optimal room for the C 2 substituent to be placed in the entry port between Gln113 and Leu259. Potential hydrophobic interactions between the pyridyl ring with hydrophobic residues Val116 and Leu161 may also contribute to the inhibitor s overall affinity. Discussion This study compares the activity of a novel class of -lactamase inhibitors, C 2/3 - substituted penicillin and cephalosporin sulfones, to the class A mechanism-based inactivators, clavulanic acid and tazobactam, against select XA -lactamases. Class D -lactamases possess significant differences in primary sequence and structure as compared to the other serine -lactamases of classes A and C. Since these XA - lactamases are found in highly resistant bacteria encountered in the hospital setting, the need to understand the kinetic behavior of these enzymes is an important first step in designing effective class D -lactamase inhibitors. The ultimate goal of this work is to find inhibitors that can inactivate all three types of XA β-lactamases penicillinases (XA-1), ESBLs (XA-10, -14, -17), and the carbapenemases (XA-24/40). ur kinetic data show that the C 2 -substituted penicillin sulfones (JDB/L-1-255, compound 9; JDB/L-III-26, compound 10; and JDB/ASR-II-292, compound 11) inhibit the XA-1 lactamase with low K i app values. These novel compounds are more effective than the C 3 -substituted cephalosporin sulfone (JDB/DVR-II-214, compound 12) 254

257 and have significantly lower K i app s than clavulanic acid and tazobactam when tested against XA-1. Susceptibility testing shows that JDB/L in combination with piperacillin is superior to the piperacillin/tazobactam combination for lowering the MIC to E. coli DH10B cells expressing the XA-1 -lactamase. Against the ESBL XA-10, -14, and -17 lactamases, JDB/ASR-II-292 and JDB/L demonstrated low K i app s ranging from 0.20 ± 0.04 to 8.0 ± 0.2 M, significantly improved compared to tazobactam and clavulanic acid. JDB/ASR-II-292 and JDB/L showed the lowest K i app s for XA-17, and the highest K i app values for XA-10. The amino acid substitutions that characterize XA-14 and -17, as compared to XA-10 (Gly167Asp and Asn76Ser, respectively), suggest a role for the Ω loop region in the inactivation mechanism of these inhibitors. The Ω loop is known to follow different paths in XA-10 as compared to XA-1, and these deviations may also influence the cephalosporin substrate profile of the β-lactamases (259, 447). Similarly, the XA-14 and -17 substitutions may affect affinities for the penicillin sulfone inhibitors; but despite these variations, the range of K i app values for the three ESBLs and XA-1 offers substantial gains over the available mechanism-based inhibitors. These low K i app s stand out when compared to previously reported data for XA-10 inhibited by 6- hydroxyalkylpenicillanates (K i s = μm) (298). Furthermore, JDB/ASR-II-292 and JDB/L maintained their potency with the carbapenamase XA-24/40. The versatility of these compounds to inhibit XA enzymes with different substrate profiles is of particular relevance considering the increasing clinical impact of these carbapenemases (195, 377). The crystal structure of XA-48, a carbapenemase of increasing clinical importance, indicates that this enzyme possesses 255

258 unique active site shape and charge features as compared to the XA-24/40 β-lactamases (126, 426). This comparison suggests the evolution of multiple mechanisms conferring carbapenem resistance in class D -lactamases, underscoring the need for inhibitors which can maintain activity across diverse active sites (126). Studies directed against XA-48 and other XA carbapenemases that are evolving in the clinic are planned. f the compounds tested against the XA enzymes, JDB/L achieved the lowest K i app values and demonstrated in vivo activity (lowered MICs). ur data shows that JDB/L demonstrates significant improvements in turnover and inhibitor efficiency (k inact /K i app ) for XA-1 compared to tazobactam. JDB/L extends these low t n s to the ESBL- and carbapenemase-type β-lactamases. Further, the JDB/L inhibitor efficiency is improved for XA-17 and -24/40 as compared to XA-1. ESI-MS of each of the XA -lactamases tested with JDB/L revealed a covalent adduct corresponding to the addition of inhibitor to the β-lactamase. This observation is consistent with previous MS and X-ray crystallography data showing the formation of a stable acyl-enzyme product consisting of unfragmented JDB/L covalently bound to SHV class A -lactamases (367). Based upon past studies with JDB/L and class A -lactamases, we posit that this compound forms a stable acyl-enzyme intermediate with XA enzymes that does not fragment along the inhibition pathway or reaction coordinate (367). Crystallography reveals that the SHV-1 -lactamase and JDB/L form a bicyclic aromatic intermediate that stabilizes the acyl-enzyme by a large conjugated system. We propose that reaction of XA -lactamases with JDB/L leads to a similar intermediate where the acyl-enzyme ester carbonyl is stabilized by both the bicyclic 256

259 aromatic system and resonance interactions with the nitrogen atoms on the inhibitor. UVD spectroscopy of XA-1 with JDB/L reveals the formation of reaction intermediates with characteristic absorption spectra. The appearance of the chromophore at nm could be consistent with formation of an aromatic intermediate. A similar UVD profile was observed with JDB/L and SHV-1 (367). In the case of irreversible inhibitors of serine -lactamases, efficacy is determined by a combination of factors, including inhibitor recognition, speed and efficiency of acylation, and hydrolytic stability of the acyl-enzyme (185, 367, 445). Class D -lactamases are not usually inhibited by commercial inhibitors such as clavulanic acid or tazobactam, while class A enzymes are generally susceptible (68). Clavulanic acid and tazobactam were shown to be mechanism-based enzyme inactivators that require either isomerization of an intermediate imine (represented by E-I T in Equation 2) to a resonance-stabilized - aminoacrylate enamine, or interception of this reactive intermediate by a second nucleophilic amino acid side chain, such as Ser130 (leading to formation of the species represented by E-I* in Equation 2) (Figure B-7A). In either case, the inhibitory process requires appropriately placed enzymatic machinery (e.g. a base to promote the isomerization, or an appropriately positioned nucleophilic residue). Glu166 of the class A β-lactamases promotes the imine to enamine isomerization of clavulanic acid (217, 232). In class D enzymes, the carboxylated active site Lys70 may serve as the general base by activating both the nucleophilic serine for acylation, as well as the hydrolytic water for deacylation. However in reference to the mechanism-based inhibitors, the carboxylated lysine of the class D enzymes may not be properly positioned, or otherwise unable, to promote the second isomerization to the stabilized intermediate enamine (174, 298, 300). 257

260 In contrast, the 6-(pyridylmethylidene)penicillin sulfone JDB/L possesses an intramolecular nucleophile in the form of the pyridine nitrogen. Previous work in SHV - lactamases suggests that the formation of the proposed inactivating species, an aromatic indolizine ring system (represented by E-I* in Equation 2), involves intramolecular capture of the imine (represented by E-I T in Equation 2) followed by the final loss of a proton (367). This proton is lost from C 5 (not from C 6 as is the case for clavulanic acid and sulbactam), which is directly adjacent to the newly basic nitrogen (formerly the - lactam nitrogen). This mechanism of intramolecular rearrangements leading to the JDB/L bicyclic intermediate is probably common to SHV class A and XA class D β lactamases (98, 367). However, the details of -lactam acylation and deacylation differ between these two enzyme classes, and we propose a general mechanism of XA inactivation by JDB/L as outlined in Figure B-7B (298, 447). Taken together, our kinetics, spectroscopic and susceptibility data, and molecular representations serve as guides for important active site/inhibitor interactions with class D β-lactamases. From these studies, we highlight four important features of JDB/L-1-255: 1) the ability of the conserved β-lactam carboxylate to make multiple interactions with active site residues, dictated in part by the stereochemistry and rotational freedom of the substituent (e.g., sp 2 - versus sp 3 -hybridization); 2) high affinity interactions between the C 6 group and residues evolved for recognizing the enzymes natural substrate, oxacillin; 3) a C 2 group which may enhance cell entry, as has been supported by this and other MIC studies (367); and 4) a novel inhibition mechanism that is less reliant on appropriately placed enzymatic machinery than the currently available suicide inhibitors. 258

261 Conclusion In summary, the JDB/L inhibitor is likely to be a broad spectrum class D β- lactamase inhibitor with activity against the pencillinase-, ESBL- and carbapenemasetype XA enzymes. This C 2 -substituted 6-alkylidene penicillin sulfone and its novel inactivation chemistry may offer significant advantages to commercially available inhibitors of XA enzymes. Table B-1. Kinetic properties of XA-1 lactamase Substrate K m (μm) k cat (s -1 ) k cat /K m (μm -1 s -1 ) xacillin 31 ± ± ± 1 Ampicillin 64 ± ± 38 6 ± 1 Cephaloridine 80 ± ± ± 0.02 itrocefin 9 ± 2 94 ± ± 3 259

262 Table B-2. K i app s of inhibitors for penicillinase-, ESBL- and carbapenemase-type XA -lactamases K i app μm a Inhibitor XA-1 XA-10 XA-14 XA-17 XA-24/40 Tazobactam 585 ± ± ± 4 40 ± ± 37 Clavulanic acid 195 ± ± 2 19 ± 5 45 ± 7 51 ± 2 JDB/L ± ± ± ± ± 0.05 JDB/ASR-II ± ± ± ± ± 0.4 a The K m values for CF used to determine the K i app for XA-10, -14, -17, and -24/40 were 13 ± 0.6, 12.5 ± 0.8, 9.5 ± 0.8, and 28 ± 3 M, respectively, and k cat values were 125 ± 10, 50 ± 5, 41 ± 4, 269 ± 20 s -1, respectively. 260

263 Table B-3. Kinetic parameters of inhibition for JDB/L Kinetic parameter β-lactamase XA-1 XA-10 XA-14 XA-17 XA-24/40 t n k inact s ± ± ± ± ± k inact /K i app M -1 s ± ± ± ± ±

264 Table B-4. Mass spectrometry analysis (atomic mass units) of XA-1 -lactamase alone and after incubation with inhibitors at 30:1 inhibitor: enzyme for 15 min a Predicted molecular mass of apo-enzyme Species observed in deconvoluted spectra XA-1 alone 28,132 28,131 difference from apo-enzyme XA-1 + JDB/L , b XA XA-10 + JDB/L XA XA-14 + JDB/L XA XA-17 + JDB/L XA-24/ XA-24/40 + JDB/L a All measurements have an error of ± 4 m/z. b Molecular mass of JDB/L is 488 g 262

265 Figure B-1. Chemical structures of commercially available inhibitors: clavulanic acid, 1; tazobactam, 2; and sulbactam, 3. Substrates used in this study: Penicillin G, 4; ampicillin, 5; oxacillin, 6, cephaloridine, 7; and nitrocefin, 8. ovel compounds used in this study: C 2 -substituted 6-alkylidene penicillin sulfones, 9-11; and C 3 -substituted 7-alkylidene cephalosporin sulfone, 12. The accepted ring numbering system is shown for cephaloridine, and C 3 of the isoxazole substituent on oxacillin is labeled. CH H S CH S CH H S CH Clavulanic Acid, 1 Tazobactam, 2 Sulbactam, 3 Penicillin G, 4 H 2 H S Ampicillin, 5 CH 3 H S CH xacillin, 6 Cephaloridine, 7 S H 7 S CH S H H S CH itrocefin, S Ca H H S Ca H S Ca S CH 2 Ca CH 3 JDB/L-I-255, 9 JDB/L-III-26, 10 JDB/ASR-II-292, 11 JDB/DVR-II-214,

266 Relative intensity Relative intensity Relative intensity Relative intensity Figure B-2. Deconvoluted mass spectra of: (A) XA-1 -lactamase; (B) XA-1 - lactamase after 15 min incubation with 30:1 I:E ratio of JDB/L-1-255; (C) XA-10 - lactamase; and (D) XA-10 -lactamase after 15 min incubation with 30:1 I:E ratio of JDB/L Measurements have an error of ± 4 amu. A. C. XA-1 alone 28, 131 XA-10 alone 27, 546 B. Mass (amu) D. Mass (amu) XA-1 + JDB/L , 618 XA-10 + JDB/L , , 035 Mass (amu) Mass (amu) 264

267 Figure B-3. UVD spectroscopy of XA-1 reacted with JDB/L ote chromophore formation at nm and nm. 265

268 Figure B-4. Molecular representations depicting overlays of the structures of XA-1 in yellow with (A) TEM-1 in red and (B) PC1 in blue. The Ω loop and residues at the turn between the B9 and B10 -strands are labeled to illustrate the relative XA-1 active site widening. A Ω-loop C XA-1 Lys70 PC1 Ser70 TEM-1 Glu166 B Ω-loop XA-1 Lys70 XA-1 Lys70 TEM-1 Ser70 PC1 Ile239 TEM-1 Glu240 XA-1 Phe

269 Figure B-5. Stereoviews of molecular representations of substrates: (A) oxacillin; and (B) cephaloridine docked in the active site of XA-1. A phenyl group sp 3 C 3 carboxylate interactions B sp 2 C 4 carboxylate interactions 267

270 Figure B-6. Stereoview of molecular representation of JDB/L docked in the active site of XA-1. pyridyl ring sp 3 C 3 carboxylate C 2 substituent 268