Novel Vi conjugate vaccines against typhoid

Transcription

1 Novel Vi conjugate vaccines against typhoid by Melissa Arcuri A thesis submitted to the University of Birmingham for the degree of DOCTOR OF PHILOSOPHY Institute of Immunology and Immunotherapy MRC Centre for Immune Regulation University of Birmingham Edgbaston B15 2TT January 2017 i

2 University of Birmingham Research Archive e-theses repository This unpublished thesis/dissertation is copyright of the author and/or third parties. The intellectual property rights of the author or third parties in respect of this work are as defined by The Copyright Designs and Patents Act 1988 or as modified by any successor legislation. Any use made of information contained in this thesis/dissertation must be in accordance with that legislation and must be properly acknowledged. Further distribution or reproduction in any format is prohibited without the permission of the copyright holder.

3 Abstract Typhoid fever remains a major public health concern in low-income countries affecting millions of people each year. Research on effective vaccines against Salmonella Typhi has been directed toward the development of glycoconjugates. Several conjugation parameters affect the magnitude, quality and persistence of immune response. A systematic investigation of the effect of each variable on immunogenicity was conducted, synthesizing and testing in mice a panel of Vi-based conjugates differing for saccharide size, carrier protein, saccharide to protein ratio and conjugation chemistry. Saccharide size and carrier protein were the parameters mainly impacting immunogenicity. Differently from full-length Vi (165 kda) conjugates, fragmented Vi (< 82 kda) CRM197 showed anamnestic response and induced minimal antibody responses in T-cell knockout mice. Unexpectedly, fragmented Vi conjugates induced lower persistent antibody levels compared to full-length Vi conjugates. Fragmented Vi conjugates offer benefits in terms of manufacture, and random chemistry is preferable because of higher conjugation yields obtained compared to selective chemistry. In view of Vi conjugate vaccine introduction into vaccination schedules, CRM197 may have advantages as an optimal carrier protein to avoid any negative effect of pre-existing anti-carrier immunity. Therefore, this systematic investigation of conjugation parameters represents a model for rational design of glycoconjugates. ii

4 Acknowledgements The research leading to the results of this thesis has received funding from the People Programme (Marie Curie Actions) of the European Unions Seventh Programme FP7/ / under REA grant agreement n I would like to thank for conducting in vivo studies and GSK Vaccines Institute for Global Health (GVGH) Immunoassay unit for performing ELISA assays in Siena (Italy). I would like to thank also all GSK approvals for their comments and suggestions and a special gratitude for my supervisors Cunningham, MacLennan and Micoli that guided me during all PhD project. A special gratitude for all people I worked with: my VADER colleagues with whom I enjoyed several time inside and outside laboratories, people working inside GVGH and University of Birmingham laboratories. Each one of them was essential for my personal and scientific growth. Thank you so much. iii

5 Table of Contents ABSTRACT II ACKNOWLEDGEMENTS III LIST OF ACRONYMS AND ABBREVIATIONS 1 1 INTRODUCTION Global burden of infection Poverty and infection Mechanism of action of vaccines The humoral response The different types of vaccines used in humans Evasion of immune response associated with bacterial capsules From polysaccharide to glycoconjugate vaccines Polysaccharide and T-Independent (TI) antigens Benefits and limits of polysaccharide vaccines Comparison between TI and T-Dependent (TD) mechanisms Key parameters affecting the immunogenicity of glycoconjugate vaccines Conjugation chemistry Saccharide chain length Saccharide to protein ratio Carrier protein Salmonella enterica causes a spectrum of infectious diseases The genus Salmonella Differences in clinical symptoms between non-typhoidal and typhoidal fever The expression and regulation of Vi antigen and its role in immune evasion The global burden of Salmonella Typhi Licensed vaccines against Salmonella Typhi Typhoidal vaccine live oral attenuated Ty21a Unconjugated Vi Vi glycoconjugates Parameters affecting the immunogenicity of Vi glycoconjugate vaccines 44 iv

6 1.9 PhD projects aims 46 2 MATERIAL AND METHODS Synthesis of Vi glycoconjugates Chemicals used for the synthesis Reagents Vi polysaccharide (PS) Fragmented Vi (fvi) Proteins CRM197 formylation Protein derivatization with ADH Protein characterization Synthesis of full-length and fragmented Vi glycoconjugates Vi-CRM197, Vi-CRMf, Vi-DT, Vi-TT, fvi-crm197, fvi-dt and fvi-tt: Vi activation with EDAC/NHS followed by conjugation to the protein derivatized with ADH linker fvi-crm197 conjugates differing for linkers length: fvi -CRMODH, fvi-crmsdh and fvi- CRMPDH fvi-(adh)-crm197: fvi randomly derivatized with ADH linked to CRM197 after activation of protein COOH groups with EDAC/NHS fvi(dmt-mm)-crmadh: fvi randomly activated with DMT-MM linked to CRM197 after its derivatization with ADH ViADHN3CRMalkyne: fvi linked to CRM-alkyne after random derivatization with azido groups fvi s(adh)crm197: fvi activated with ADH at the reducing end and linked to CRM197 after activation of COOH groups on the protein with EDAC/NHS fvi sshcrmsbap: fvi activated with cysteine at the reducing end and conjugated to CRM 197 previously derivatized with SBAP Characterization of glycoconjugates Analytical methods Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) MALDI-TOF High pressure liquid chromatography-size exclusion chromatography (HPLC-SEC) Ion pair high pressure liquid chromatography-reversed phase (HPLC-RP) Determination of amount of free Vi in the conjugates by Capto Adhere/HPAEC-PAD method H NMR spectroscopy Immunogenicity studies in mice ELISA assays conducted by GVGH Immunoassay unit in Italy Immunological assessments performed at the University of Birmingham Conjugates and antigens Biotinylation of fvi Buffer preparations PBS at ph v

7 Buffers used for ELISA Buffers for Immunohistochemistry ELISA Preparation of cell suspension from spleen and bone marrow Enzyme-Linked ImmunoSpot (ELISpot) Immunohistochemistry (IHC) Cryostat sectioning Staining Bacteria growth and detection of Vi expression Bacterial infection Burden of bacteria Statistics 71 3 INFLUENCE OF CONJUGATION VARIABLES ON THE IMMUNOGENICITY OF A VI VACCINE AGAINST SALMONELLA TYPHI Introduction Summary Results Optimization conjugation process and characterization of Vi -CRM197 conjugate obtained Main conjugate strategy adopted Reproducibility of conjugate formation at small scale Scale-up of the process for Vi-CRM197 conjugate production and conjugate characterization Impact of saccharide chain length Production of fragmented Vi populations at reduced molecular weight compared with full - length Vi Investigation of conjugate conditions in order to obtain fvi -CRM197 conjugate vaccines Immunogenicity of fragmented Vi -CRM197 conjugates compared with full-length Vi-CRM197 in mice Influence of crosslinking/size and saccharide to protein ratio Investigation over conjugates obtained with full -length Vi Investigation over conjugates obtained with fvi with NAMW 43 kda Impact of carrier protein Impact of conjugation chemistry Discussion COMPARISON OF DIFFERENT CARRIER PROTEINS FOR VI CONJUGATE VACCINES AND THEIR PRIMING EFFECT Introduction 123 vi

8 4.2 Summary Results Vi-CRM197, Vi-DT and Vi-TT conjugates differing for protein derivatization degree with ADH linker The formylation process on CRM197 determines similarities with DT physical properties Synthesis of Vi conjugate with CRMf Immune response induced by Vi conjugates differing for carrier protein and carrier priming effect induced by CRM197, CRMf, DT and TT Discussion INVESTIGATION OF THE EARLY- AND LONG-TERM IMMUNE RESPONSES INDUCED IN MICE BY SELECTED VI CONJUGATES Introduction Summary Results Investigation of the early-response induced by Vi conjugates Full-length Vi conjugates are well-tolerated and induce predominantly anti -Vi IgG1 antibodies Full-length Vi conjugates, but not fragmented Vi conjugates, induce Vi-specific immune response seven days post-immunization Antibodies induced by Vi-CRM197 and Vi-TT cross-react with other Vi conjugates Assessment of the longevity of the response to Vi conjugate vaccines Anti-Vi IgG antibody titres are similar if the carrier protein is CRM197 or TT Boosting with full-length Vi conjugates induces higher anti -Vi IgG antibody responses than Vi PS Anti-carrier protein antibody responses are detectable after boosting with full -length Vi conjugates Immunizations with Vi conjugates and Vi PS induces ASCs in the spleen and bone marrow Assessment of protection after immunization with different Vi based vaccines Induction of Vi expression by Salmonella Typhimurium strains Discussion FINAL DISCUSSION AND FUTURE WORK 179 LIST OF REFERENCES 183 vii

9 List of acronyms and abbreviations Reagents CRM197 Cross Reacting Material 197 CRMf DT fvi fvis PS TT CRM197 subjected to formylation process Diphtheria Toxoid fragmented Vi fragmented Vi activated by selective approach polysaccharide Tetanus Toxoid Analytical methods for conjugate characterization BCA HPLC-SEC HPLC-RP HPAEC-PAD MALDI-TOF NMR SDS-PAGE TFF bicinchoninic acid High Pressure Liquid Chromatography-Size Exclusion Chromatography Ion pair High Pressure Liquid Chromatography-Reversed Phase High Performance Anion-Exchange Chromatography Matrix Assisted Laser Desorption/Ionization-Time Of Flight Nuclear Magnetic Resonance Sodium Dodecyl Sulfate-Polyacrylamide Gel Electrophoresis Tangential Flow Filtration Acronyms used for immunological methods BSA CFU ELISA EU Bovine Serum Albumin Colony Forming Units Enzyme Linked Immunosorbent Assay ELISA Units 1

10 IHC SBA SFU Immunohystochemistry Serum Bactericidal Antibody Spot Forming Units Immunological acronyms ASCs BCRs DCs EF GCs LPS MAC MLN MHCII MZ PAMPs SPI TCR TD Tfh TI TLRs T3SS Antibody Secreting Cells B cell receptors Dendritic cells extrafollicular Germinal Centers lipopolysaccharide membrane attack complex mesenteric lymph nodes major-hystocompatibility complex class II Marginal Zone pathogen-associated molecular patterns Salmonella pathogenicity island T cell receptor T-dependent T follicular helper T-independent Toll-like receptors type III secretion systems Acronyms of pathogens/diseases IDoP Infectious Diseases of Poverty 2

11 ints MDR NTS invasive Non-Typhoidal Salmonella Multi-Drug Resistant Non-Typhoidal Salmonella Acronyms of vaccines or vaccine subunits OMP PD PRP PspA repa rp40 outer membrane protein non-typeable Haemophilus influenzae-derived protein D polyribosylribitol phosphate pneumococcal surface protein A Pseudomonas aeruginosa exotoxin A recombinant OMP of Klebsiella pneumoniae General Acronyms CIES GVGH h HMW i.p. Kd LMICs LMW NAMW min MW MWCO ON Carrier-Induced Epitopic Suppression GSK Vaccines Institute for Global Health hour/s high molecular weight intraperitoneally distribution coefficient low- and middle-income countries low molecular weight number average molecular weight minute/minutes molecular weight molecular weight cut-off overnight 3

12 RT SARs s.c. v/v WT w/v w/w room temperature structure-activity relationship subcutaneously volume to volume wild type weight to volume weight to weight 4

13 1.1 Global burden of infection 1 Introduction Infectious diseases have been a bane of humankind (1, 2) and affect the poorest communities disproportionately (2). Estimates suggest that 25% of all deaths are attributable to infectious diseases, although this is a gross under estimate since infection contributes markedly to deaths included in noncommunicable diseases like diabetes, cancer and cardiovascular diseases (2). Even though vaccination adds unquestionable value against disease, preventing 700 million cases and more than 150 million of deaths during the last century (3), the need for effective vaccines against many diseases still persists (4). 1.2 Poverty and infection Poverty is closely linked to the burden of infection. The term IDoP or infectious diseases of poverty defines infectious diseases dominant inside poor and marginal populations (5). IDoP and poverty co-exist in a vicious cycle: persistent poverty creates favourable conditions for the spread and the persistence of infectious diseases in poor communities lacking prevention and health care assistance. Moreover there are limited resources to tackle diseases in poor countries (5). Tackling infectious diseases is a key aim for the improvement of health in endemic countries (5). Environmental, social, cultural, economic factors and political instability influence the extent of government resources allocated to prevent the risk of infection and transmission (6). Accessible medicines for use in at risk populations alongside safe-water provision and improved sanitation are all ways to tackle infectious diseases in low-income countries (7, 8). Vaccination is an enormous potential tool to fight infectious diseases (9). 5

14 1.3 Mechanism of action of vaccines The term vaccine identifies a substance inducing immune responses against a specific pathogen when it is administered in a host. Vaccination typically works through for inducing antibodies and herd immunity. Herd immunity works through the reduction of disease incidence among non-vaccinated people living in contact with vaccinated individuals (10). This public benefit can be conferred by mass-immunization programs supported by broad vaccination coverage and by cost-effectiveness analysis (11, 12). Long-term protection is achieved by induction and maintenance of antigen-specific immune effectors (13). Immune effectors are composed essentially of antibodies that provide the humoral immune response. Antibodies derive from B lymphocytes, cells produced in the bone marrow that mature in the secondary lymph nodes. They can become activated after the encounter with antigen. After activation, B lymphocytes can differentiate into plasma cells and memory B cells. The plasma cells secrete antigen-specific antibodies into extracellular spaces meanwhile memory B cells are pointed to provide protection against further encounters with the antigen by proliferation and differentiation into antibody-secreting plasma cells (13). The generation of antigen-specific antibody alongside cell-mediated immunity are key elements for the induction of effective immune response after vaccination (3) The humoral response The efficacy of the humoral response toward a specific pathogen can be assessed through at least four approaches: a) passive administration of pathogen-specific antibodies effective against the course of infection, b) detection of an inverse relationship between the presence of pathogen-specific antibodies and the susceptibility to infection, c) increased susceptibility to 6

15 disease associated with a deficit of humoral and cell-mediated immunity and d) immunization against the pathogen on the absence of infection (14). The intrinsic properties of the antibodies (specificity, isotype and affinity), the immune status and the genetic background of the host are all factors determining the efficacy of the humoral response (15). Antibodies can act through direct and indirect mechanisms (Table 1.1). Direct mechanisms involve the recognition and binding of the antibodies toward a pathogenic component. Direct mechanisms include the complement activation through the classical pathway, agglutination and neutralization of a toxin or a virus. Indirect mechanisms are antimicrobial effects that involve the participation of other effector cells and/or modification of the inflammatory environment (15). Table 1.1. Classification of antimicrobial activities of antibodies (Table adapted from (15)) Type of mechanism Direct Indirect Action Opsonization Complement activation Viral neutralization Toxin neutralization Antibody-dependent cellular cytotoxicity Bactericidal Fungistatic Interference with microbial activities Generation of oxidants Changes in cytokine expression Changes in costimulatory molecule expression Changes in Fc R expression Enhancement of lysosome-phagosome fusion 7

16 1.4 The different types of vaccines used in humans Vaccines can be classified as live, inactivated, subunit, recombinant and glycoconjugate vaccines (10). Live or inactivated vaccines are composed of pathogenic organisms that have been respectively weakened or killed. Subunit vaccines are formed of components of the pathogen such as proteins or polysaccharides (Table 1.2) expressed by the pathogen. Unfortunately polysaccharide (PS) vaccines are frequently ineffective in young children and possibly older adults (16). Glycoconjugate vaccines have been developed to overcome this limitation. 8

17 9

18 1.5 Evasion of immune response associated with bacterial capsules Many Gram-negative and Gram-positive bacteria are surrounded by capsules that are composed of PS chains made up of repeating units of one up to six monosaccharides linked by glycosidic bonds. The particular chemical composition and structure to the PS (such as presence of enantiomeric centers, stereoisomers, specific glycosidic linkage and positions of specific residues on the ring unit) can vary significantly between bacteria/serovars (Table 1.2) (17, 18). The different structures characterizing bacterial capsules can determine protection against host immunity (19). Pathogenic bacteria are able to elude host immune defenses through different strategies that can be provided by bacterial capsules (19). Capsules can act as cloaks masking antigenic molecules such as pathogen-associated molecular patterns (PAMPs) that could be easily recognized by host immune system through Toll-like receptors (TLRs). Bacterial capsules can help evade the complement system and related host defence actions such as phagocytosis, killing through membrane attack complex (MAC) formation, clearance of harmful immune complexes and modulation of inflammatory response (19). Furthermore bacterial capsules can inhibit the migration of phagocytes toward the site of infection or impair the attachment/ingestion of bacteria (19). 10

19 1.6 From polysaccharide to glycoconjugate vaccines Polysaccharide and T-Independent (TI) antigens Vaccine candidates are often structure surfaces including PS capsules (20, 21). Protection is often related to different aspects of immune response induced. For example Streptococcus pneumoniae and Haemophilus influenzae type b protection is correlated to antibody levels measured by ELISA. Bacterial antibody titres for Neisseria meningitidis are measured by serum bactericidal antibody (SBA) assay (20). Capsular polysaccharides are species characterized by varied degree of polymerization and repeating antigenic epitopes spatially close to each other. They can cross-link multiple surface immunoglobulins expressed on B cells (BCRs) and provoke a response without involving the participation of T cells. Capsular polysaccharides are classified as T-cell-independent type 2 (TI-2) antigens (22-29). Such responses are associated with little IgG switching, affinity maturation of BCRs and production of memory B cells (30). Particular subsets of B cells, such as B1-b and marginal zone (MZ) B cells are associated with TI responses (31, 32). As a consequence of cross-linking on BCRs, B cells are activated and differentiate into short-lived plasma cells secreting low-affinity IgM antibodies (20, 27, 33) Benefits and limits of polysaccharide vaccines Whilst the use of plain PS vaccines has been beneficial, there are limitations associated with the limited persistence of antibody responses and the hyporesponsiveness, a phenomenon occurring by repeated administrations when immune response is not equal or greater in magnitude compared to primary response (34-36). 11

20 The drawbacks of the immune response induced by pure PS vaccines can be overcome by chemical conjugation of the saccharide, which contains the required B-cell epitopes covalently linked to an appropriate protein to provide T-cell help. The concept of glycoconjugate vaccines has the origin from Avery and Goebel. (37). They coupled derivatives of galactose and glucose to globulin from horse serum and crystalline egg albumin. The glycoconjugates injected intravenously were able to induce specific anti-carbohydrate antibodies in rabbits. The first protein-polysaccharide conjugate vaccines were developed against Haemophilus influenzae type b and licensed in USA between 1987 and 1990 for children and infants (38). Clinical trials conducted in the 1980s in Finland, North America and UK demonstrated the efficacy of these glycoconjugate vaccines in infants (39). The introduction of Hib glycoconjugate vaccines into national immunization programs has led to a dramatic decline in the incidence of the disease worldwide (40). Glycoconjugate vaccines have been developed to target other organism such as Neisseria meningitidis, Streptococcus pneumoniae and Group B Streptococcus and have been shown to provide protection (41, 42). Clinical trials performed directly in children (43-46) or in pregnant women (47) showed glycoconjugate vaccines can protect against infection in vulnerable age groups Comparison between TI and T-Dependent (TD) mechanisms TI-2 antigens, such as PS vaccines, cross-link BCRs of MZ and B1-b cells. This interaction generates B cell activation and migration to the T zone of secondary tissues where proliferation occurs. The B cell fate can follow two possible pathways that develop with parallel kinetics (Figure 1.1). Some B cells migrate to the follicles where potentially germinal centers (GCs) can develop, but lacking T-cell support GCs involute within 4-5 days after 12

21 immunization (48, 49). In parallel plasmablasts migrate from the T zone to the splenic red pulp or medulla of lymph nodes where extrafollicular (EF) responses develop (50). Even in the absence of T cells this latter mechanism can results in IgM production with little somatic hypermaturation (51). Extrafollicular responses can be considered a source of adaptive humoral immunity against TI-2 antigens such as PS capsules (49). Figure 1.1. Possible B cell developments occurring kinetically in parallel (50). B cells can be activated both by T-dependent and T-independent antigens. Migration to T zone is followed by two possible pathways leading to antibody production. Tfh: T follicular helper cells Differently to TI-2 antigens, the carrier protein moiety of glycoconjugates involves preliminary maturation of a specialized subset of antigen-presenting cells called dendritic cells (DCs) designed to take up, process and present antigen protein to naїve CD4 + T cells via major-hystocompatibility complex class II (MHCII) (52-54). This interaction generates CD4 + T cells that proliferate and differentiate into effector T cells accordingly to the Th subset profile induced by cytokines (55). In parallel, as reported previously for TI-2 antigens, B cells 13

22 encountering the PS portion of the conjugate are activated and migrate to the T zone (50) and possible pathways for B cell development are the migration to EF sites and to GCs (Figure 1.1). A crucial factor in immune response induced by glycoconjugates is the recruitment of T follicular helper (Tfh) cells that sustain GC formation. Germinal centers develop later that EF responses (50). Therefore TD responses induced by glycoconjugate vaccines can promote saccharide specific IgM-to-IgG class switching and GC reaction thus generating both highly specific antibodysecreting plasma cells and memory B cells (Figure 1.2) (13, 17, 56-59). Extrafollicular responses to capsular saccharide antigens are of modest longevity, in contrast plasma cells from GCs can be long-lived and when they home to the bone marrow they can survive for decades (60). The prolonged antibody persistence from plasma cells in the bone marrow provides protection against future exposure to the pathogen, whereas memory cells represent sentinels ready to proliferate and differentiate into plasma cells (13). 14

23 Figure 1.2. The immune responses to T-independent (A) and T-dependent (B) antigens (59) Recently a new model has been proposed for presentation of glycoconjugate vaccines to T cells. Particular polysaccharides, called zwitterionic polysaccharides, have been found able to be presented to CD4 + T cells even if not conjugated to a carrier protein (17, 61). In line with this concept it was suggested that through B cell endolysosomal digestion glycoconjugates are processed into glycopeptides where both carbohydrate and peptide epitopes are presented to T 15

24 cell receptors (TCRs) (61-63) (Figure 1.3 B). According to this model the glycoconjugate design determines the presentation of epitopes that stimulates TD immune response (62, 64). Figure 1.3. Classical (A) and novel (B) models of interaction between B- and T-cells (61) 16

25 1.7 Key parameters affecting the immunogenicity of glycoconjugate vaccines The immune response induced by glycoconjugates is affected by several conjugation parameters, such as the preservation of saccharide immunodeterminants (65), conjugation chemistry, linkers used, saccharide chain length, saccharide to protein ratio and carrier protein (33, 66, 67) Conjugation chemistry Glycoconjugate synthesis is usually performed by one of two main saccharide activation approaches: random activation of multiple points along the saccharide chain, or selective activation at the terminal end. The conjugation strategy used determines the conjugate structure. A cross-linked net is characteristic of a random saccharide activation (Figure 1.4 A), a sun-structure is typical of a selective approach (Figure 1.4 B). Bioorthogonal reactions have been exploited in different scientific areas, including the synthesis of glycoconjugates (68). Several different chemistries have been proposed both for random and selective conjugation approaches (66). The natural presence of carboxylic groups in the saccharide backbone can be potentially used for direct conjugation to the amine groups of the protein. However many conjugation processes involve previous chemical modification of saccharide (69) and/or protein (68). Furthermore the usage of adequate linkers can facilitate the conjugation, reducing steric hindrance between protein and saccharide (66). 17

26 Figure 1.4. Possible conjugate structures depending on the conjugation strategy applied: A) crosslinked net and B) sun-structure. Figure adapted from (33) 18

27 Figure 1.5 reports some of the strategies most commonly used for the synthesis of glycoconjugate vaccines. The first conjugate vaccine against Haemophilus influenzae type b was composed of the polyribosylribitol phosphate (PRP) PS conjugated to Diphteria Toxoid (DT) protein (70). The absence of carboxylic or aldehyde groups on the PS does not allow direct conjugation to amine groups on the protein, thus the PS was activated by cyanogen bromide with formation of reactive cyanate ester groups (71) (Figure 1.5, mechanism 1). Introduction of a dihydrazide spacer on the protein facilitates the linkage to the activated PS (70). Subsequently the conjugation protocol was modified in order to better preserve protein structure and immunogenicity (72). The PS was activated through cyanogen bromide, linked to a dihydrazide linker and then bound to the protein via carbodiimide chemistry (72). The carbodiimide reaction involves the conversion of carboxylic groups into O-acyl isourea intermediates that can react with amine groups resulting in amide formation (Figure 1.5, mechanism 2 A). This amidation reaction has been widely investigated for modifying polysaccharides (69). The reaction is strongly ph-dependent: carboxylic group activation by carbodiimide occurs preferably in acid conditions, meanwhile high ph facilitate amide formation when the amine group is not protonated (Figure 1.6, mechanism 1) (73). Hydrazides are characterized by lower pka compared to amines, thus use of hydrazide groups instead of amines can lead to higher conjugation efficacy (Figure 1.6, mechanism 2) (69). 19

28 Figure 1.5. Conjugation strategies mostly used in glycoconjugate synthesis. Mechanism 1: cyanobromide activation and formation of cyantate ester groups. Mechanism 2: carbodimmide activation and direct conjugation to amine group (A) or after activation with N-hydroxysuccinimide (B). Mechanism 3: saccharide oxidation and reductive amination. Mechanism 4: coupling between succinimidyl ester and amino groups. Mechanism 5: possible thiol-reactions. Mechanism 6: coupling involving tyrosine groups. Mechanism 7: thiol-ene coupling. Mechanism 8: azido-alkyne cycloaddition 20

29 In order to avoid the side formation of stable N-acylurea after carbodiimide activation (Figure 1.6, mechanism 3), the addition of N-hydroxysuccinimide results in the formation of a more stable NHS ester intermediate (Figure 1.5, mechanism 2 B) that can be subjected to nucleophilic attack by amines (Figure 1.5, mechanism 4) (69). Figure 1.6. Possible products obtained activating carboxylic groups via carbodiimmide reaction. Mechanism 1: coupling with amine. Mechanism 2: linkage to dihydrazide linker. Mechanism 3: side formation of stable N-acylurea 21

30 Another possible saccharide modification involves aldehydes formation as the aldehyde groups can then be linked to lysine amino acids through reductive amination. According to the PS structure, when vicinal hydroxyl groups are present the treatment with sodium periodate can result in cleavage of C-C bonds and aldehydes formation (Figure 1.5, mechanism 3). Periodate chemistry approach has been exploited to conjugate Hib oligosaccharides to Cross Reacting Material 197 (CRM197) (74) and DT (75) proteins. Following another conjugation strategy, serogroup C meningococcal oligosaccharides were obtained by acid hydrolysis and selectively modified at the terminal end with an amino group by reductive amination. Through the reaction between the terminal amino group and N-hydroxysuccinimide diester, this further linker was used for conjugation with CRM197 protein (76). The same approach was followed to obtain the quadrivalent A,C,W135,Y meningococcal vaccine (Menveo) (77) (Figure 1.7). 22

31 Figure 1.7. Conjugation process for meningococcal glycoconjugate vaccines (77) 23

32 Since amino and carboxylic groups are abundant in proteins sequence, they have been widely exploited for protein modification (78, 79). Lysine groups are commonly used for linkage to electrophiles molecules such as activated esters (80). In general carboxylic groups are coupled to amino-linkers through carbodiimide reaction (Figure 1.5, mechanism 2) (81). Other protein modifications occur through reactions involving amino acids less abundant in proteins sequence, such as cysteines and tyrosines. Cysteines can be coupled to - halocarbonyls (Figure 1.5, mechanisms 5 A) (78), maleimides (Figure 1.5, mechanisms 5 B) (82, 83) or to glyco-methanethiosulfonates, glycophenilthiosulfonates and glycoselenenylsulfides leading to disulfide formation (Figure 1.5, mechanism 5 C) (84, 85). Tyrosine groups represent an alternative target for protein modification (Figure 1.5, mechanism 6) (86). For instance tyrosine-protein modification has been exploited to enrich GBS67 pilus protein with maleimido or azido groups in order to couple modified type V PS respectively through thiol-maleimide addition (Figure 1.5, mechanism 5 A) or copper-free strain-promoted azido-alkyne cycloaddition (Figure 1.5, mechanism 8) (87). The latter strategy is an example of the click chemistry approach applied for biomedical research and glycoconjugate synthesis (88). Even if copper-catalysed azide-alkyne cycloaddition has been demonstrated to be an efficient click chemistry approach for protein-polymer conjugation (89), copper-free chemistry has been preferred for obtaining higher conjugation efficacies with high molecular weight (MW) polysaccharides (90, 91). Free-radical thiol-ene coupling is another click chemistry approach using alkenyl glycines and glycosyl thiol (92, 93) (Figure 1.5, mechanism 7). 24

33 Glycoconjugate design requires the selection of conjugation strategies optimizing antigen presentation to the immune system with identification of appropriate linkers. Linkers are commonly used as spacers between carbohydrates and protein moieties to reduce steric hindrance and facilitate conjugation (66). The presence of spacers characterized by rigid and constrained structure can induce anti-linker immune response (66) in particular when conjugations involve not immunogenic antigens such as meningococcal PS serogroup B (94). Investigation of immunogenicity of tumour-associated saccharides conjugated through maleimide linkers to the carrier protein revealed the induction of a linker-specific immune response and suppression of antigen-specific antibodies (95). Recently it was observed that glycoconjugates obtained through azido-alkyne cycloaddition induce specific triazole ring antibodies although no deviation of the antigen antibody response was found (87, 96). In the last decade several studies have been conducted to synthesize well-defined constructs obtained by coupling saccharides selectively activated at the terminal end with modified protein in correspondence of specific amino acids, in order to control the exact position of the linkage and/or the number of attachments points (87, 91, 96). Well-defined constructs obtained by site-specific conjugation represent a potential tool for investigating structureactivity relationship (SARs) with immune response. These conjugations are preferable to random cross-linking for manufacturing, enable batch-to-batch consistency and easier characterization (86, 87, 91). Manufacturing cost is an important aspect for vaccine production. Glycoconjugates are expensive products due to the complexity of their multi-step synthesis and analytic analysis (97). Polysaccharides are reagents with varied degree of polymerization (98), thus glycoconjugates obtained through random chemistry consist of heterogenous compounds 25

34 difficult to fully characterize and to produce consistently (99). In contrast both selective chemistry and amino acid-specific conjugation are consistent processes producing welldefined constructs. Bioconjugation technology is emerging in response to the high cost of glycoconjugate production simplifying multi-step synthesis and guaranteeing well-controlled conjugation through enzymatic site-specific conjugation (61) Saccharide chain length The influence of saccharide chain length on immunogenicity is closely correlated to conjugation chemistry. Long polysaccharides are usually conjugated by random chemistries, meanwhile selective approaches are preferred for short oligosaccharides. Random chemistry through multi activation points on the saccharide backbone could markedly alter epitopes of oligosaccharides composed of a few repeating units. Some investigations have compared immune responses induced by polysaccharide or oligosaccharide conjugates obtained by random chemistries ( ). Higher anti-saccharide specific antibody responses were induced by conjugates composed of longer saccharide chains both in mice (101, 102) and in infants (100). Comparisons between polysaccharide and oligosaccharide conjugates are hard to perform especially when obtained by random chemistry: they often differ in saccharide size, saccharide-loading and degree of cross-linking. Selective conjugation using oligosaccharides could be a useful approach to obtain welldefined constructs and to investigate the effect of saccharide chain length on immunogenicity. Some efforts were made to obtain sets of glycoconjugates differing in only oligosaccharide chain length. The effects of saccharide chain length were different and saccharide-specific (75, ). 26

35 1.7.3 Saccharide to protein ratio Some studies have been performed to investigate the effect of saccharide chain length and saccharide loading on immunogenicity. A set of conjugates through random conjugation linking TT protein to pneumococcal type 4 PS (saccharide repeating units ranging from ) and oligosaccharide (composed by 12 repeating units). The hapten-specific antibody response was modulated by saccharide loading depending on the length of the carbohydrate (saccharide/protein w/w ratio ranged 0.2 to 2.2 and 0.9 to 3.2 respectively for polysaccharide and oligosaccharide conjugates). The lower the saccharide loading, the higher IgG immune response induced by the PS conjugate, meanwhile the opposite correlation was found for oligosaccharide conjugates (101). The effect of saccharide size on immunogenicity through selective chemistries is closely related to saccharide loading (109). An excess of saccharide could mask the carrier protein hampering the participation of T cells in the immune response. This concept is consistent with findings from conjugates of Shigella dysenteriae type 1 tetra-, octa-, dodeca- and hexadecasaccharide differing for antigen loading (saccharide/protein molar ratio ranged from 4 to 23) (109). For long oligosaccharides the optimal response was obtained with intermediate carbohydrate loading, meanwhile optimal carbohydrate density shifts to higher values for shorter saccharide chains (109) Carrier protein Potentially any protein with T helper cell epitopes can act as carrier protein. The selection of appropriate carrier proteins is limited by basic requirements involving non-toxicity, nonreactogenicity, adequate level of purity and stability under conjugation conditions (67). Among the proteins used as carrier for glycoconjugate vaccines, five carrier proteins have 27

36 been widely exploited: TT, DT, CRM197, Neisseria meningitidis outer membrane protein (OMP) and non-typeable Haemophilus influenzae-derived protein D (PD) (110). Tetanus Toxoid and DT have been used as carrier proteins for licensed vaccines against Haemophilus influenzae type b and Neisseria meningitidis diseases (111). In contrast, CRM197 protein is non-toxic and thus the manufacturing process does not require chemical detoxification, allowing the protein to be obtained in homogenous preparations with high levels of purity (111). Several manufacturers have widely used CRM197 as a carrier protein for both routine childhood and adulthood vaccines (111). Furthermore several syntheses have been conducted using the recombinant Pseudomonas aeruginosa exotoxin A (repa) as carrier protein for glycoconjugate vaccines against S. Typhi (112), Staphylococcus aureus (113), Shigella (114) and E. coli (115). Some animal studies investigated the influence of carrier protein on immunogenicity elicited by glycoconjugate vaccines (72, ). In some cases investigations did not consider the influence of the other conjugation parameters (72, 117, 118) and the superiority of one protein over another results dependent on the saccharide to which it is conjugated (72, 116, 122). As components of glycoconjugates, carrier proteins can induce specific anti-carrier antibody responses (123). It is generally considered that vaccination must elicit protection against the bacteria whose PS is the component of the vaccine, but with a low/absent response against carrier epitopes in order to avoid immune interference with the anti-ps response (124). Other studies have investigated capsular PS conjugated to protein derived from the same pathogen in order to try to elicit a synergistic immune response against both components ( ). Recently a bivalent conjugate vaccine, obtained by coupling Vi capsular PS from S. Typhi to pneumococcal surface protein A (PspA), was found to be potentially protective against 28

37 typhoid fever and infection from a broad range of Streptococcus pneumoniae strains (128). Interest over new proteins as potential carriers has been growing, in particular for development of multivalent vaccine formulations and more elaborated immunization schedules (129). 1.8 Salmonella enterica causes a spectrum of infectious diseases The genus Salmonella Salmonella is a Gram-negative bacteria able to colonize humans and animals (130). Two distinct species of Salmonella can be distinguished: Salmonella bongori and Salmonella enterica, the latter divided into six subspecies: enterica (I), salamae (II), arizonae (IIIa), diarizonae (IIIb), hotenae (IV), and indica (V) (131). Each subspecies is divided in serogroups and serovars on the basis of antigenic composition accordingly to Kauffman-White scheme (130). Distinct antigens shared among Salmonella serovars have been used for their classification (Table 1.3). The current nomenclature reports subspecies designation followed by O and H antigens that define respectively O-antigen of lipopolysaccharide (LPS) and flagellin (132). Salmonella enterica species comprises more than 2660 serovars which have different antigenic properties, host specificity and clinical symptoms. The majority of S. enterica serovars consists of non-typhoidal and typhoidal serovars. Table 1.3 shows the variations in expression for some antigens among the main serovars. In addition to those reported in Table 1.3, other S. enterica serovars that express Vi PS include serovar Paratyphi C (133) and serovar Dublin (134). Citrobacter freundii is a bacterium able to express Vi antigen and it has been used as suitable source for producing Vi (135). 29

38 Invasive non-typhoidal Salmonella (ints) disease represents a neglected disease most affecting sub-saharan Africa (136). Salmonella Typhimurium and Enteritidis are the two main ints serovars (Table 1.3) mainly affecting children under 2 years old and HIV-infected people (137). Typhoidal Salmonella clinically manifests as enteric fever, that can be typhoid or paratyphoid fever depending on the responsible pathogen: S. Typhi or S. Paratyphi (Table 1.3) (137, 138). Table 1.3. Variation in expression found for some of antigens among the main Salmonella serovars (Table adapted from (137)) Clinical presentation Serovar of S. enterica Antigen Enteric fever ints disease Typhi Paratyphi A Typhimurium Enteritidis 1.Vi O: O:4, O: Protein 5.OmpF/OmpC OmpD Other +/- +/- +/- +/- 1.Capsular Vi antigen, 2,3,4. O-antigen of the lipopolysaccharide, 5,6.Porin proteins 30

39 Although enteric fever can be easily recognized from gastroenteritis associated with NTS, common clinical symptoms and overlapping geographic distribution do not permit discrimination between enteric fever caused by S. Typhi and S. Paratyphi A (139), without blood cultures (140). Enteric fever is endemic in the developing world and in poor-resources settings where both adequate laboratory diagnostics and supplying system for safe water are not accessible (140, 141) Differences in clinical symptoms between non-typhoidal and typhoidal fever Non-typhoidal Salmonella (NTS) diseases are caused by pathogens able to infect a broad range of host species. NTS salmonellosis is common worldwide and transmission can occur through contaminated food and contact with animals such as cats, dogs, rodents, reptiles or amphibians (138). NTS typically manifests as a gastroenteritis localized to the intestine and mesenteric lymph nodes (MLN) and generally the infection is self-limiting (137). Up to 5% of cases bacteria can spread to become invasive NTS (ints) infections characterized by bacteremia and focal systemic infections (138). The burden of ints disease is concentrated in Sub-Saharian Africa and associated with S. Typhimurium and S. Enteritidis (142). Fatality occurs in up to 25% of cases (143), and ints has been strongly associated with African children living in regions with a high incidence of malaria and among HIV-infected adults (142). Salmonella serovars can also cause enteric fever. These serovars are more host restricted, primarily infecting humans, and can be divided into four phylogenetically unrelated clonal lineages (144). Salmonella enterica Typhi is human-specific and represents one of these clonal 31

40 lineages. Salmonella enterica Paratyphi C and B serovars represent two lineages, meanwhile S. enterica Paratyphi A and Sendai constitute the fourth one. Salmonella Paratyphi A, B, C and Sedai serovars cause paratyphoid fever, that is typically milder but clinically identical compared to typhoid fever (144). In contrast to gastroenteritis which can occur from the first day of ingestion of non-typhoidal Salmonella contaminated food (144), the average incubation time of S. Typhi is two weeks (145). Typhoid fever is not considered a typical diarrheal disease because only one third of typhoid fever patients present diarrheal symptoms (146). Whereas there are localized gastroenteritis symptoms after ingestion of NTS serovars, in typhoid fever patients lack the robust initial intestinal inflammation and neutrophil accumulation allowing bacteria to invade deeper tissues of gut (147). In fact S. Typhi spreads systemically from intestine to MLN, liver, spleen, bone marrow and gall bladder (Figure 1.8) (138). Cytokines promote fever that is sustained in the second week after symptoms appear (148). If untreated or not correctly diagnosed, the symptoms can persist beyond the third week leading to gastroenteritis, bleeding and intestinal perforation. Fever typically decreases in the fourth week of disease without antibiotic therapy, although some patients (10-15 %) who do not receive antibiotic treatment can die (148, 149). It has been estimated up to 10% of convalescent untreated patients shed S. Typhi in stool within three months after infection, and 1-4% of infected subjects can be asymptomatic chronic carriers for more than 12 months (150). The gall bladder is considered the reservoir for S. Typhi persistence where bacterial biofilm formation shields the pathogen from antimicrobial compounds and the immune system (151). Long-term persistence of S. Paratyphi A has not been studied as much as for S. Typhi, 32

41 but in a recent study a similar rate of persistence has been observed between these two serovars in endemic countries such as Nepal (152). Figure 1.8. Representation of systemic infection caused by Salmonella enterica Typhi (153) The expression and regulation of Vi antigen and its role in immune evasion The different clinical symptoms observed in non-typhoidal and typhoidal patients are associated with the presence of antigens and the modulation of virulence gene expression specific for the serotype investigated (154, 155). The neutrophil accumulation in the intestinal mucosa, characteristic of S. Typhimurium, has been associated with the bacterial ability to evade the immune system and through a mechanism involving the recognition of patternassociated molecular patterns such as flagellin and LPS (155, 156). In particular the lipid A moiety of LPS results in Toll-like receptor 4 (TLR4) ligation on monocytes (156). The 33

42 production of the neutrophil chemoattractant CXCL-8 after TLR4 activation (157) can be amplified through the complement system, which is itself activated by the O-antigen sugars of LPS that contain multiple free hydroxyl groups (158). Neutrophil recruitment in the intestine does not occur in typhoid fever patients (159), and this is associated with the expression of the virulence factor Vi capsular antigen. Vi is a linear homopolymer of (1 4) -D-galacturonic acid fully N-acetylated at position 2 and variably O-acetylated at position 3 (Figure 1.9). Figure 1.9. A. chemical composition of Vi repeating unit, B. Vi polysaccharide sequence composed by repeating units 34

43 The absence of hydroxyl groups in the Vi repeating unit structure (Figure 1.9) enables encapsulated bacteria to evade C3b fixation impairing the bacterial opsonization by macrophages (160). Reduced neutrophil migration associated with Vi occurs through preventing C5aR-dependent neutrophil chemotaxis (161). Furthermore the reduced oxygen consumption of neutrophils observed after ingestion of encapsulated S. Typhi bacteria (162) has been related to the inhibition of C5aR host cell recruitment (163). The early production of IL-10 and the scarcity of neutrophil recruitment were revealed in mice infected by S. Typhimurium strains expressing Vi. These factors could help explain the progressive spread in to the host by S. Typhi (164). Bacterial Vi capsule is encoded by the viab locus located in the Salmonella pathogenicity island 7 (SPI-7). The viab locus is composed by genes involved in regulation (tvia), biosynthesis (tvibcde) and cell surface localization (vexabcde) of Vi PS (Figure 1.10). Figure Schematic representation of the composition of viab locus (165) The gene and the correspondent enzyme determining the degree of the O-acetylation has not yet been identified (165). The viab locus is not exclusive to S. Typhi genome but is also present in typhoidal S. serotype Paratyphi C and in some non-typhoidal isolates of S. serotype Dublin. 35

44 Intriguingly, other serovars that cause enteric fever, such as Paratyphi A and B, lack the viab locus (159). It was suggested that all four clonal lineages can be related to typhoid fever through acquisition of viab locus or alternatively through different sets of genes (166). Vi capsule is not a virulence factor constantly expressed on the surface, in fact bacteria need to control Vi expression during different phases of pathogenesis to maximize infection and evade immune defenses as necessary (167). Bacteria have to adapt to changes in environmental conditions by switching specific sets of genes on/off. For instance environments characterized by high osmolarity provoke Vi downregulation that, together with expression of components of the type III secretion system (T3SS), determines the bacterial invasiveness of M cells in Peyer s Patches (168). Conversely the absence of detectable levels of Vi under low osmolarity conditions suggests that other stimuli can impinge the expression of Vi antigen (167). The TviA regulatory protein, encoded within the viab locus, not only modulates Vi expression but alters the expression of other virulence factors. TviA induces both flagellin repression and Vi upregulation in order to respectively inhibit TLR5 and TLR4 signaling (169, 170). TivAmediated flagellin repression is associated with CD4 + T cell inhibition and therefore it is considered responsible for the bacterial dissemination from MLN to the spleen (171). The mechanism by which Vi capsule can interfere with TLR4 signaling has not yet been clarified (170). Furthermore TivA mediates the repression of pro-inflammatory responses associated with T3SS-1, which helps the transport of other effector proteins to the cytoplasm of host cells promoting intestinal inflammatory response (172). Both S. Typhimurium and S. Typhi express T3SS-1 (155), but only S. Typhimurium induces local inflammatory response in the intestine (172). Therefore Vi expression contributes to the different clinical presentation associated with non-typhoidal and typhoidal fevers, and helps the evasion of host immunity by 36

45 S. Typhi bacteria to blunt the intestinal inflammatory response and promote the systematic dissemination of bacteria (163) The global burden of Salmonella Typhi Determining the absolute numbers of typhoid cases worldwide is problematic, not least because of the other enteric fever causing bacteria and ints (141). Crump et al estimate there were 21 million cases of typhoid fever during 2000 with 1% mortality (173). The high prevalence of typhoid cases in Asia is consistent with another investigation analysing blood-cultures taken from febrile patients hospitalized in South- and South-East Asia. Salmonella enterica serovar Typhi was the most common bacterial pathogen with 30% of cases in adults and 25% in children (174). In several recent studies, similar endemic incidence of S. Typhi was found for Africa and Asia ( ). Another aspect that must not be underestimated is the emergence of multi-drug resistant (MDR) S. Typhi strains. Before 1993 more than 70% of Kenyan isolates from blood cultures were fully sensitive to all antimicrobials, but after 2000 nearly 75% of isolates were MDR (175, 182). More accurate epidemiological analysis estimates the incidence of typhoid fever of low- and middle-income countries (LMICs) during 2010 at 11.9 million cases with 129,000 deaths (Figure 1.11) (183). 37

46 Figure Typhoid incidence in low- and middle-income countries (risk-adjusted and corrected for blood culture sensitivity)(183) A constant and accurate check of global epidemiology is important for disease prevention, but it must be linked to significant solutions against the spreading of Salmonellae disease. In the early twentieth century typhoid fever was endemic in Europe and North America (184). Improved access to clean water was the key intervention responsible for the decline of typhoid incidence in Europe and North America (184). A similar improvement is needed to see such a change in developing countries ( ). In addition the implementation of appropriate vaccination strategy targeting the populations vulnerable to the disease will help curb its spread. Several investigations in low-income countries report on the age groups most affected by typhoid. For instance data collected from a community-based laboratory network in Pakistan between has revealed that 70% of typhoid cases occurred in under 15 years old (188) (Figure 1.12). 38

47 Figure Age distribution of enteric fever in Karachi (Pakistan) between (188) A passive-population-based surveillance study was conducted in rural communes in Southern Vietnam in 2000, and the highest typhoid incidence was reported among 5-9 years old children (531/ and 358/ persons per year respectively in 5-9 and 2-4 year-olds) (181). Similar conclusions were found in other studies (189). The apparent absence of typhoid among infants could be due to a minor number of very young children admitted to other social reasons rather than an absence of infections (181, 189). This is supported by a recent study conducted to determine the typhoid incidence in pre-school children in Pakistan which showed the incidence of typhoid in children < 2 years old and <1 year old to be 443/ and 506/10000 per year respectively (190). There is a significant economic consequence to typhoid. Each blood-culture confirmed typhoid fever results in a mean cost of approximately 100$ in outpatient treatment, hospitalization increases this 5 fold (191). The increased number of admissions with MDR 39

48 infections will lead to a higher cost to manage (191). Vaccination of pre-school children could help reduce these costs (187, 188, 190, ) Licensed vaccines against Salmonella Typhi There are several vaccines against S. Typhi infections. The whole-cell killed vaccine is highly reactogenic, thus it is no longer available for use (196). Currently three types of vaccines are licensed against S. Typhi: the live attenuated vaccine Ty21a, unconjugated Vi PS and Vi-TT glycoconjugate vaccine ( ) Typhoidal vaccine live oral attenuated Ty21a In the 1970s Ty21a was developed as an attenuated version from the wild-type strain Ty2 by mutation leading to a loss of an isomerase responsible for LPS production, and by a separate mutation that prevents expression of Vi antigen (200). Further mutations are present and these contribute to the attenuation of virulence (201). Protection can be elicited by multiple doses given orally of Ty21a formulated as entericcoated capsules or liquid solutions (196, ). The cumulative efficacy elicited by a threedose schedule of Ty21a is estimated at 79% and 62% over 5 years and 7 years respectively after vaccination (196). Furthermore there is some benefit of herd immunity effect from this vaccine and some cross-protection against S. enterica serovar Paratyphi B (201, 205). Ty21a induces both humoral and cell-mediated immunity ( ). The plasmablasts induced can have an intestinal homing profile suggesting Ty21a vaccination can show similarities with a natural typhoid infection (209). Enteric-coated Ty21a (with the brand name Vivotif TM ) is manufactured by Crucell Switzland LTD and is indicated for adults and children 6 years aged or older (210, 211). Immunization 40

49 consists of ingestion of four capsules on alternate days, where each dosage contains x 10 9 colony-forming units with re-immunization recommended every 7 years if necessary (211). The vaccine needs cold storage (211) Unconjugated Vi Vi PS makes up the capsule surrounding S. Typhi, Paratyphi C and Dublin serovars (137), and is a virulence antigen with the capacity to evade innate immune response through several mechanisms (163). Vi PS can impair bacterial LPS recognition by TLR4 (170), inhibit host opsono-phagocytosis activity (160) and abrogate neutrophil chemotaxis toward infected tissues (161). All these prevent the innate immune response from acting effectively against S. Typhi (212). Original attempts to obtain Vi from typhoid bacilli denatured the antigen resulting in a loss of antigenic sites and consequently immunogenicity (212). Later methods prevented denaturation and provided high levels of saccharide purity (212). Trials both in animals (213) and in humans ( ) followed with protection associated with anti-vi antibodies (214, 215). The use of Vi antigen as a vaccine was first registered by Sanofi Pasteur in 1988 and then in the USA in 1994 (217). Initially the Vi vaccine was used for travellers going to typhoid endemic areas (218). Two factors provided incentives for Vi vaccine production worldwide (219): reports from WHO that highlighted the necessity of immunization for school-aged children in areas where typhoid is prevalent (220, 221) and that standardized manufacturing of Vi is not patented (222). Several vaccination programs were launched providing Vi vaccine produced locally at prices more affordable for poor populations (223). A typhoid vaccination program targeting children 41

50 aged 3-10 years old was carried out by a National Immunization Program (NIP) in Vietnam providing Vi vaccine at approximately US $ 0.5 per dose (217), with other programs following in Asia (195, 217, 224). The immune response to Vi vaccination is fast: a significant antibody response is observed one week after immunization and the maximal response is achieved after only one month (225) in both endemic (226) and non-endemic areas (225). The minimal antibody level associated with protection after Vi vaccination was estimated at 1 g/ml (226). A recent meta-analysis found the protection of the vaccine afforded after 1, 2 and 3 years as 69%, 59% and 50% respectively (196). Thus re-immunization after 2-3 years is generally recommended for travellers going to endemic areas (227). Vi PS does not elicit immune responses in children below 2 years old and is unsuitable for this age group (227). It is unclear how much protection Vi PS confers in the 2-5 year age group (193, 196, 228). Nor is it clear whether re-immunization with Vi PS induces a response or if there is hyporesponsiveness (229). Vi vaccine is administered subcutaneously or intramuscularly as a single dose containing 25 g of Vi. The storage temperature is between 2 and 8 C (221). The limited persistence of protection and inefficiency in young children are factors strongly affecting the usage of both Ty21a and Vi PS as vaccines for mass vaccination in endemic areas. Currently two main strategies are being undertaken to overcome these drawbacks: the improvement of live oral vaccines (197, 199) and the development of Vi glycoconjugates (196). 42

51 Vi glycoconjugates Two Vi-based glycoconjugates are licensed locally in India. Both are composed of TT protein conjugated to Vi PS and are identified by the trade names Typbar-TCV and Pedatyph TM (198, 199). Typbar-TCV glycoconjugate has been tested in India in a double-blind, randomized controlled trial recruiting people aged 2-45 years. It was performed in parallel with an openlabel trial enrolling only infants and toddlers aged between 6 and 23 months. A single dose of Typbar-TCV showed induction and maintenance of seroprotection among all age groups for two years after injection (230). Immunizing children 6 months to 12 years of age with 2-doses of Pedatyph TM, Vi-TT conjugate vaccine conferred 100% protection during the first year post immunization (231). A second vaccine dose injected 30 months post-primary vaccination did not enhance antibody levels significantly compared to children receiving a single Pedatyph TM dose (232). Currently several efforts for the development of better Vi glycoconjugates are taking place (198, 199). A glycoconjugate composed by Vi PS covalently linked to repa has already been tested in humans with some success, but is still not licensed ( ). The International Vaccine Institute has worked on the development of a Vi glycoconjugate vaccine with DT as carrier protein (236, 237). Clinical trials of the vaccine are being conducted by Shanta Biotech (198). Another Vi glycoconjugate was developed by GSK Vaccines Institute for Global Health (GVGH), using Citrobacter freundii as bacterial source for Vi production, and CRM197 as carrier protein ( ). The vaccine has been tested in Phase I and II clinical trials (241, 242) and recently the Vi-CRM197 technology has being transferred to the major Indian vaccine manufacturer, Biological E (137). 43

52 1.8.6 Parameters affecting the immunogenicity of Vi glycoconjugate vaccines Several investigations have been conducted on the impact of conjugation parameters on the immunogenicity of Vi-based glycoconjugate vaccines. The chemical preservation of the structure of the repeating unit is crucial and de-o acetylation drastically decreases the antibody response to Vi (243). For this reason guidelines on quality, safety and efficacy of typhoid conjugate vaccines, require O-acetyl content of purified Vi PS to be at least 2.0 mmol/g PS (corresponding to 52% Vi repeating units O-acetylated) (244). According to the Courtauld-Koltun space-filling atomic model, O-acetyl moieties of Vi are extended outward compared to the saccharide backbone and constitute exposed immunodominant epitopes. In contrast carboxyl groups are not fundamental for the immune response as they are hidden and closer to the saccharide axis (243). Random conjugation strategies use carboxyl groups for conjugation. In 1987 Szu et al. compared two different conjugation strategies: the direct binding of protein amino groups to Vi via carbodiimide activation (Figure 1.5, mechanism 2) and disulfide formation between derivatized protein and thiolated Vi (Figure 1.5 mechanism 5 C) (245). Despite the more complex conjugation procedure the use of linker spacers helped alleviate steric hindrance and facilitated coupling (66). Another conjugation scheme was developed based on carbodimmide chemistry between hydrazide-derivatized protein and Vi PS (Figure 1.6, mechanism 2) (246). A study conducted by Kossaczka compared anti-vi IgG antibody response induced by repa glycoconjugates obtained using adipic acid dihydrazide (ADH) or succimidyl 3-(2-pyridyldithio)propionate (SPDP) as linkers attached to the protein. ADH conjugate induced higher anti-vi IgG antibody responses in Vietnamese children (112). 44

53 The conjugation strategy with ADH as protein linker was applied to repa (112), DT (236), CRM197 and TT proteins (240). Investigations on the influence of the carrier protein on the immune response induced were conducted in mice using the same conjugation chemistry and resulting in similar Vi to protein ratio of the obtained conjugates. No significant differences were revealed between repa and DT (247) or among repa, TT and CRM197 (240). In order to investigate the Vi to protein ratio effect on immune response, a set of Vi-CRM197 conjugates, with ratio values ranging from 0.9 to 10.1, were synthesized and tested in mice. The study examined different Vi doses. Differences in the immune response were only evident at lower dose and the anti-vi antibody response declined with increased Vi to protein ratios (248). In another study, Vi-DT conjugates with different Vi to protein ratios were compared. The lower Vi to protein ratio, the higher was the degree of conjugate cross-linking. Higher the amount of DT, higher the anti-vi IgG response induced (236). Only one study has investigated the effect of Vi chain length on the immunogenicity induced (249). Full-length and shorter Vi, with number average molecular weight (NAMW) 45 kda (250)) were conjugated to Cholera toxin subunit B protein via disulphide formation (Figure 1.5, mechanism 5 C) using the heterobifunctional linker SPDP as spacer. The shorter Vi conjugate induced a lower anti-vi IgG response compared to that induced by full-length conjugate both in mice and monkeys. In contrast, higher anti-toxin levels were detected from animals injected with shorter Vi conjugate compared to levels elicited by full-length Vi conjugate. This difference in anti-toxin response could be explained by more exposure of the toxin molecule if conjugated to shorter Vi (249). 45

54 Understanding the relation between conjugate constructs and immune responses is difficult if multiple parameters are changed simultaneously: for instance some studies were conducted modulating both the Vi to protein ratio and the conjugate size (236), or Vi to protein ratio and carrier protein (251), or even differing three parameters simultaneously like the Vi to protein ratio, the carrier protein and the Vi chain length (249). 1.9 PhD projects aims In this thesis I aimed to perform a systematic investigation to define the major parameters affecting the immunogenicity induced by Vi glycoconjugates, trying to identify the best combination of parameters for an improved glycoconjugate vaccine against S. Typhi. The three main objectives of this study were: 1) To examine the effect of saccharide chain length, carrier protein, saccharide to protein ratio and conjugation chemistry on the immunogenicity induced by Vi glycoconjugates. 2) To investigate the influence of the carrier protein on the immunogenicity induced by Vi glycoconjugates and the carrier priming effect in mice. 3) To evaluate the early- and long-term immune response induced by selected Vi glycoconjugates in mice. 46

55 2 Material and Methods 2.1 Synthesis of Vi glycoconjugates Chemicals used for the synthesis The following chemicals were used for the glycoconjugate synthesis: adipic acid dihydrazide (ADH), oxalildihydrazide (ODH), pimelic acid dihydrazide (PDH), succinic dihydrazide (SDH), N-(3-dimethylaminopropyl)-N -ethylcarbodiimide hydrochloride (EDAC), N- hydroxisuccinimide (NHS), 4-(4,6-dimethoxy-1,3,5-triazin-2-yl)-4-methylmorpholinium chloride (DMT-MM), biotin hydrazide, succinimidil 3-(bromoacetamido) propionate (SBAP), cystamine dihydrochloride, sodium cyanoborohydride (NaBH3CN), formalin, L-Lysine, 4- morpholine ethanesulfonic acid (MES), sodium chloride (NaCl), sodium hydroxide (NaOH), hydrochloride acid 37% (HCl), dimethyl sulfoxide (DMSO), sodium acetate (AcONa), sodium phosphate monobasic monohydrate (NaH2PO4. H2O), 2,4,6-trinitrobenzenesulfonic acid (TNBS) [Sigma], dithiothreitol (DTT) [Invitrogen], phosphate buffered saline tablets (PBS) [Fluka], ethylenediaminetetraacetic acid (EDTA) disodium salt [Merk], acetonitrile (ACN) [Prolabo], NHS-PEG4-N3 [Thermo & Fisher], click easy BCN NHS ester I alkyne linker [Berry & Associate] Reagents Vi polysaccharide (PS) Vi PS was purified at GVGH from Citrobacter NVGH328 strain as previously described (135, 239). In particular the lot used for this study was characterized by contaminations: 0.3% protein (assessed by micro BCA), 0.001% nucleic acid (assessed by picogreen) (w/w respect 47

56 to the sugar) and endotoxin level of 2.6 EU/ g of sugar (assessed by LAL test). O-acetylation level was > 90% as detected by 1 H NMR and the NAMW was of 165 kda, as estimated by HPLC-SEC analysis (TSKgel 3000 PWXL column). Determination of saccharide NAMW was conducted by calibration with molecular weight dextran standards for the similar conformation of these molecules with Vi saccharide Fragmented Vi (fvi) Fragmented Vi (fvi) NAMW ranging from 82 to 8.6 kda were produced and characterized. Full-length Vi was solubilized in water and H2O2 was added in order to obtain a final concentration of Vi and H2O2 respectively of 2.5 mg/ml and 5% (w/v) in water. The mixture solution was heated at 80±0.5 C for 2h, subsequently it was injected in Hiscreen Capto Q [GE Healthcare] column (4.7 ml of resin loading up to 100 mg of fragmented Vi mixture) equilibrated with buffer A (20 mm NaH2PO4 ph 7.2). Four populations characterized by different NAMW were separated by a gradient step method using buffer A and B (20 mm NaH2PO4 1M NaCl ph 7.2). fvi pools of increasing NAMW were eluted at 25, 30, 37 and 45% of buffer B and labelled from pool1 to pool4 increasing the NAMW. Each pool was desalted on Sephadex G-25 column [GE Healthcare] equilibrated with water. The NAMW of fvi pools was determined by HPLC-SEC equipped with a TSKgel 3000 PWXL column and a TSKgel PWXL guard column [Tosoh Bioscience]. Dextrans (NAMW 5, 25, 50, 80, 150 kda) were used as standards due to the similar conformation with Vi saccharides [Sigma Aldrich]. The mobile phase was 0.1 M NaCl, 0.1 M NaH2PO4, 5% ACN, ph 7.2, at the flow rate of 0.5 ml/min (isocratic method for 30 min). Vi structure integrity and O-acetylation level were determined by 1 H NMR. 48

57 Proteins CRM197, DT and TT were obtained from GSK Vaccines, Siena. TT protein was purified by gel filtration through Sephacryl S300 [GE Healthcare] equilibrated in 0.15 M NaCl, 10 mm NaH2PO4, ph 7.2. The fractions corresponding to the monomeric MW of TT were pooled and used for conjugation CRM197 formylation CRM197 protein was formylated as reported in the literature (252). In detail CRM197 was purified three times against 100mM NaH2PO4 ph 7.0 by Vivaspin 20 30k MWCO PES [Sartorius Stedim], and diluted in 0.7% (v/v) formalin, 25 mm L-Lysine, 67 mm NaH2PO4 ph 7.8 in order to obtain protein concentration of 8.3 mg/ml. Mild stir was conducted at RT for one week. After this period dialysis by Spectra/Por dialysis membrane (MWCO ) was performed against seven cycles of water, followed by Vivaspin 20 30k against 100 mm MES ph Protein derivatization with ADH CRM197, DT, CRMf (formylated CRM197) and TT were derivatized by the same treatment with ADH and EDAC (239, 240). ADH (3.5 mg per mg of protein) was added to the protein (10 12 mg/ml in mm MES buffer ph ) and mixed. As soon as solution became clear, EDAC (EDAC/protein = 0.15, w/w) was added. The reaction was allowed to proceed for 1h at RT with slow mixing. In order to introduce more ADH linkers on CRM197, EDAC/protein ratio in the mixture was increased to 0.4 or 1 w/w. The reaction mixture was dialyzed against 0.2 M NaCl, 5 mm MES buffer, ph 7.0, 2-8 ºC and 5 mm MES buffer, ph 7.0, 2-8 ºC (Spectra/Por dialysis membrane MWCO ). Alternatively, for CRM- 49

58 ADH tangential flow filtration (TFF) was performed using a 10k membrane (Hydrosart 200 cm 2 in stabilized cellulose). The TFF was done keeping the retentate volume constant with a protein concentration of 10 mg/ml and 20 cycles diafiltration against 5 mm MES buffer ph 7.0 were done. Derivatized protein solutions were 0.22 µm filtered Protein characterization Protein recovery was calculated by micro BCA (using BSA as standard and following Thermo Scientifics instructions). The CRMf protein was characterized by HPLC-SEC, SDS-PAGE gel and MALDI-TOF in comparison to CRM197 and DT proteins. Protein derivatization degree with ADH was quantified by MALDI-TOF analysis. SDS-PAGE and HPLC-SEC were run to verify proteins integrity after derivatization Synthesis of full-length and fragmented Vi glycoconjugates Vi-CRM197, Vi-CRMf, Vi-DT, Vi-TT, fvi-crm197, fvi-dt and fvi-tt: Vi activation with EDAC/NHS followed by conjugation to the protein derivatized with ADH linker Fragmented Vi with NAMW 43 kda was solubilized in 100 mm MES ph 6 at a concentration of 50 mg/ml. NHS and then EDAC were added to have 0.33 M NHS and EDAC/saccharide repeating units molar ratio of 5. After the reaction was mixed at RT for 1h, the protein previously derivatized with ADH (239, 240), was added to have fvi concentration at 7.8 mg/ml in 20 mm MES ph 6. The mixture was mixed at RT for 2h. The same conjugation procedure was applied for fvi with shorter length (NAMW 9.5 and 22.8 kda), while saccharide concentration during the activation step with EDAC/NHS was reduced to 15 mg/ml for fvi with NAMW 82 kda. For full-length Vi, the PS concentration in the 50

59 EDAC/NHS activation step was reduced to 4.2 mg/ml and to mg/ml in the conjugation step. Different ratios of saccharide to protein were used: 1:1, 2:1 or 1:2 in weight. Full-length Vi-CRM197 was purified by TFF by using a 300k membrane (Sartocon Slice Cassette 200 cm 2 PES). Twenty cycles of diafiltration against 1M NaCl 20 mm NaH2PO4 ph 7.2, and subsequently twenty cycles of diafiltration against 20 mm NaH2PO4 ph 7.2 were performed. For full-length Vi-DT and Vi-CRMf conjugates the purification was performed with a 100k membrane (Hydrosart 200 cm 2 in stabilized cellulose). Full-length Vi-TT conjugate and fvi conjugates were purified by size exclusion chromatography on a 1.6 cm x 60 cm Sephacryl S300 column or 1.6 cm x 60 cm Sephacryl S100 HR column respectively [GE Healthcare] eluted at 0.5 ml/min in PBS. Fractions at higher MW, not overlapping free PS and free protein run on the same column in the same conditions, were collected. Vi-NHS and fvi-nhs intermediates were not isolated before protein addition, but a fraction of the mixture was sampled in process and characterized for quantifying the % of activated saccharide repeating units (molar ratio % of NHS/saccharide repeating units). Samples were purified by 14.5 x 50 mm Disposable PD10 Desalting Column Sephadex G-25 [GE Healthcare] against HCl 55 ppm, not to hydrolyze NHS-ester groups; NHS was quantified by ion pair HPLC-RP and PS was quantified by HPAEC-PAD (240) fvi-crm197 conjugates differing for linkers length: fvi-crmodh, fvi- CRMSDH and fvi-crmpdh CRM197 was derivatized with ODH, SDH or PDH, as previously described for ADH (239, 240). Conjugation step with fvi NAMW 43 kda was performed as described for CRM-ADH, with a fvi to protein ratio 1:1 in weight, but increasing Vi concentration to 15 mg/ml to have 51

60 all the protein conjugated after 2h mixing at RT. The conjugates were purified by size exclusion chromatography on a 1.6 cm x 60 cm Sephacryl S100 HR column eluting at 0.5 ml/min in PBS fvi-(adh)-crm197: fvi randomly derivatized with ADH linked to CRM197 after activation of protein COOH groups with EDAC/NHS Fragmented Vi NAMW 43 kda was solubilized in 100 mm MES ph 6 at a concentration of 15 mg/ml. NHS and then EDAC were added to have 0.1 M NHS and EDAC/fVi repeating units molar ratio of 5. After the reaction was mixed at RT for 1h, ADH was added (molar ratio between ADH and Vi repeating units equal to 1.5). The mixture was mixed at RT for 2h and then desalted by PD10 column. No crosslinking of fvi(adh) was confirmed by HPLC-SEC and 22% repeating units resulted activated by TNBS colorimetric method (253). For the step of conjugation, CRM197 was diluted with 600 mm MES ph 6 at a concentration of 15.5 mg/ml. NHS and then EDAC were added to have 0.1 M NHS and EDAC/COOH groups molar ratio of 5. After the reaction was mixed at RT for 1h, fvi(adh) was added to have fvi concentration of 10 mg/ml and with a fvi to protein ratio 1:2 in weight in 100 mm MES ph 6. The reaction was mixed at RT for 2h. The conjugate was purified by size exclusion chromatography on a 1.6 cm x 60 cm Sephacryl S300 column eluting at 0.5 ml/min in PBS. It was verified by HPLC-SEC that no protein aggregation happened in the reaction conditions used (data not shown) fvi(dmt-mm)-crmadh: fvi randomly activated with DMT-MM linked to CRM197 after its derivatization with ADH Fragmented Vi NAMW 43 kda was solubilized in NaH2PO4 100 mm ph 7 to have a fvi concentration of 10 mg/ml and DMT-MM was added with a molar ratio of 5 compared to Vi 52

61 repeating units. The reaction proceeded at RT for 10 min and CRM-ADH was then directly added to the solution to have a fvi to protein ratio 1:1 in weight with a fvi concentration of 3.8 mg/ml. After mixing at RT for 2h, the conjugate was purified by size exclusion chromatography on a 1.6 cm x 60 cm Sephacryl S100 HR column eluting at 0.5 ml/min in PBS ViADHN3CRMalkyne: fvi linked to CRM-alkyne after random derivatization with azido groups Fragmented Vi NAMW 43 kda was randomly activated with ADH as previously described. Derivatized fvi (25% fvi repeating units activated according to TNBS) was then mixed with the linker NHS-PEG4-N3 (25 mg/ml in DMSO) in 100 mm NaH2PO4 ph 7.2 at a concentration of 3.6 mg/ml fvi and with a molar ratio azido linker to NH2 groups on fvi 2:1. The reaction was mixed at RT for 4h and the product purified by PD10 column eluting with NaH2PO4 10 mm ph 7.2. Ninety % NH2 groups introduced on fvi through ADH resulted derivatized, as verified by TNBS method (253). CMR197 was diluted in PBS and click easy BCN NHS ester I alkyne linker (10 mg/ml in DMSO) was added (molar ratio linker to lysines on CRM197 of 0.76) resulting in a protein concentration of 8.6 mg/ml. After mixing at RT for 5h, the mixture was purified by PD10 column eluting with PBS. An average of 12 linkers resulted introduced per CRM197 molecule by MALDI-TOF (240). Conjugation was performed in PBS with a final concentration of protein at 12 mg/ml and a molar ratio of azido to alkyne groups 6:1. The solution was mixed at RT ON and the resulting conjugate purified by hydrophobic interaction chromatography on a Phenyl HP column [GE Healthcare], loading 300 μg of protein for ml of resin in 50 mm NaH2PO4 3 M NaCl ph

62 The purified conjugate was eluted in water and the collected fractions were dialyzed against PBS fvis(adh)crm197: fvi activated with ADH at the reducing end and linked to CRM197 after activation of COOH groups on the protein with EDAC/NHS Fragmented Vi NAMW 8.6 kda was dissolved in AcONa 20 mm ph 4.5 at a concentration of 30 mg/ml. Reductive amination was performed by adding ADH and NaBH3CN (respectively 6 and 17 fold molar excess respect to fvi chains). The reaction proceeded for 3 days at 30 C. The solution was then diluted in 3 M NaCl and desalted twice by PD10. Ninetyfive % of fvi chains resulted activated by TNBS. Fragmented Vi derivatized with ADH was added to CRM197 activated with EDAC/NHS as previously described, in order to obtain fvi and CRM197 concentrations respectively of 30 and 10 mg/ml and a molar ratio of fvi chains to CRM197 of 20:1. The reaction proceeded for 2h at RT. Conjugate was purified by size exclusion chromatography on a 1.6 cm x 60 cm Sephacryl S100 HR column eluting at 0.5 ml/min in PBS fvisshcrmsbap: fvi activated with cysteine at the reducing end and conjugated to CRM197 previously derivatized with SBAP CRM197 was solubilized in 100 mm NaH2PO4 1 mm EDTA ph 8.0 (5.1 mg/ml); SBAP was added (0.3 mg/ml, molar ratio SBAP/lysine groups on CRM197 = 0.3) after being solubilized in DMSO (final DMSO concentration of 4% v/v). Mixture was stirred for 3h at RT, and then purified by PD10 column eluting with 100 mm NaH2PO4 1 mm EDTA ph 7.0. An average of 9 linkers was introduced per CRM197, as determined by MALDI-TOF (240). Fragmented Vi NAMW 8.6 kda was solubilized in 100 mm NaH2PO4 ph 7.0 (20 mg/ml) and then cystamine (112.5 mg/ml, cystamine/fvi (w/w) = 5.6) and NaBH3CN (50 mg/ml, 54

63 NaBH3CN/fVi (w/w) = 2.5) were added. The mixture was stirred for 5 days at 30 C and then desalted after diluting the sample with 6 M NaCl by PD10 column. Cystamine disulfide bond was reduced by mixing derivatized fvi at a concentration of 20 mg/ml with 100 mm DTT in 100 mm NaH2PO4 5 mm EDTA ph 7.2 for 1h at RT. The derivatized fvi was purified by desalting on PD10 column against 10 mm NaH2PO4 5 mm EDTA ph 7.5. Seventy % fvi chains resulted activated with cystamine, according to TNBS, and reduction with DTT was complete, as verified by Ellman colorimetric method (254). Conjugation was performed in 100 mm NaH2PO4 1 mm EDTA ph 7.2 with protein concentration of 12 mg/ml and using a molar ratio of fvi chains to CRM197 20:1. After mixing 3h at RT the conjugate was purified by hydrophobic interaction chromatography on a Phenyl HP column, as previously described Characterization of glycoconjugates Purified conjugates were characterized by HPAEC-PAD for total Vi content (240), micro BCA for total protein content, HPLC-SEC for determining NAMW distribution of the conjugate and to assess the amounts of free protein and free saccharide (for fvi pool 1 and pool 2). For conjugates prepared with longer fvi chains as for full-length Vi conjugates, free saccharide was estimated by Capto Adhere/HPAEC-PAD method. 1 H NMR analysis was performed to verify Vi PS identity and O-acetylation degree retention after conjugation (240). 55

64 2.1.3 Analytical methods Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS- PAGE) Protein integrity after derivatization or formylation process was verified by SDS-PAGE. Five- 20 g of protein with a concentration of 1-2 mg/ml were mixed with 0.5 M DTT (1/5 v/v) and NuPAGE LDS sample buffer (1/5 v/v). Mixture solution was heated at 100 C for 1 min. Sample and pre-stained protein standard (SeeBlue Plus2, Thermo Fisher) were loaded in 7% Tris-acetate gel (NuPAGE, Invitrogen) and electrophoresed at 45 ma in NuPAGE Tris- Acetate SDS running buffer (20x, Invitrogen). Gel was stained using Simply Blue Safe Stain [Invitrogen] MALDI-TOF Proteins were diafiltrated against 10 mm NaH2PO4 ph 7.2. Two L of the final solution of protein (at the concentration of 5 mg/ml) were mixed with 2 L of saturated solution of sinapinic acid composed by 50% ACN and 0.1% TFA. Two L of the final mixture were spotted on MTP 384 stainless steel target [Bruker Daltonics GmBH] and allowed to dry. Data were recorded on Ultraflex III [Bruker GmbH] MALDI-TOF/TOF MS in linear mode. Calibration was performed using protein calibration standard II [Bruker Daltonics] composed by: trypsinogen (23982 Da), protein A (44613 Da) and bovine serum albumin (66431 Da). Mass spectra were obtained recording up to 400 laser shots, and data were processed by Flex Analysis Software. 56

65 High pressure liquid chromatography-size exclusion chromatography (HPLC-SEC) HPLC-SEC is a chromatographic method widely used for separating analytes based on their size. Herein HPLC-SEC analysis was performed to characterize full-length Vi, fvi, proteins, conjugation mixtures and purified conjugates. Unconjugated proteins, Vi and fvi conjugates were eluted on a TSK gel G3000 PWXL column (30 cm x 7.8 mm; particle size 7 m; cod ) with TSK gel PWXL guard column (4.0 cmx6.0 mm; particle size 12 m; cod ) [Tosoh Bioscience], whereas full-length Vi conjugates were eluted on a TSK gel 6000 PW (30 cm x 7.5 mm) column (particle size 17 m; Sigma ) connected in series with a TSK gel 5000 PW (30 cm x 7.5 m) column (particle size 17 m; Sigma ) with TSK gel PWH guard column (7.5 mm IDx7.5 cm L; particle size 13 m; Sigma ). The mobile phase was 0.1 M NaCl, 0.1 M NaH2PO4, 5% ACN, ph 7.2 at the flow rate of 0.5 ml/min (isocratic method for 60 min on TSK gel and for 30 min on TSK gel G3000). Superose 6 10/300 GL column [GE Healthcare] was also used for conjugates analysis, eluted in PBS and 0.3 ml/min. Void and bed volume calibration was performed with λ-dna (λ-dna Molecular Weight Marker III kb; Roche) and sodium azide [Merck], respectively. Vi and proteins were detected setting UV detector at 214 nm (PS and protein) and at 280 nm (protein). Protein peaks were also detected using tryptophan fluorescence (emission spectrum at 336 nm with excitation wavelength at 280 nm). Vi PS (also detected by refractive index) NAMW was calculated using dextrans (5, 25, 50, 80, 150 kda) as standards [Sigma Aldrich]. The distribution coefficient Kd was calculated following the equation: 57

66 Kd =(Te-T0)/(Tt-T0) where: Te represents the elution time of the analyte, T0 the elution time of the biggest fragment of -DNA and Tt the elution time of sodium azide. HPLC-SEC was also used to estimate the amount of unconjugated CRM197 (fluorescence emission detection) and unconjugated fvi (for pool 1 and 2; refractive index detection). The area of unreacted protein was quantified using a calibration curve built with protein samples in the range 5 50 µg/ml. The percentage of unconjugated protein was calculated dividing the amount of free protein detected by HPLC-SEC by the total amount of protein quantified in the sample by micro BCA. Similarly the amount of unconjugated fvi was quantified using a calibration curve of fvi (having same NAMW than in the conjugate) in the range µg/ml. The percentage of unconjugated saccharide was calculated dividing the amount of free Vi detected by HPLC-SEC by the total amount of saccharide quantified in the sample by HPAEC-PAD Ion pair high pressure liquid chromatography-reversed phase (HPLC-RP) The quantification of NHS ester groups introduced on Vi PS was determined by ion pair HPLC-RP analysis. Same analysis was used to quantify residual NHS groups in the conjugates. Samples were eluted on a C18 column (Phenomenex, Gemini-NX 5 with 80% 10 mm TBABr, 0.17% NH4OH, 20% ACN in isocratic condition with a flow rate of 1 ml/min. Eluent ph allowed ester-nhs groups hydrolysis and formation of N- hydroxysuccinimidate anion that was detected at 260 nm eluted as ion pair with TBA. Calibration curve was built using NHS as standard in the range 3-50 nmol/ml. 58

67 Determination of amount of free Vi in the conjugates by Capto Adhere/HPAEC-PAD method The packed pellet derived from a low speed centrifugation of 500 µl of Capto Adhere resin suspension [GE Healthcare], corresponding to about 50 µg of vacuum dried resin, was washed twice with 20 mm AcONa 30% ACN ph 5. ACN was added to the glycoconjugate sample in 20 mm NaH2PO4 ph 7.2 (total Vi approximately 150 µg) to give a 30% concentration. The sample was added to the resin and incubated at RT for 30 min on a rotating wheel, then centrifuged (5 min at 4 C rpm) and the supernatant (indicated as unbound ) collected. Under these conditions, the presence of quaternary ammonium residues allows binding of the conjugate, but also of the polyanionic unconjugated Vi. The pellet was washed twice by resuspending in 1 ml 20 mm AcONa 30% ACN ph 5 and centrifuging for 5 min at 4 C rpm and the two wash solutions were collected and pooled to give a volume of 2 ml. Unconjugated Vi was stripped by vortexing the resin for few seconds in 500 µl of 1 M AcONa 30% ACN ph 5, then centrifuged for 5 min at 4 C rpm. This was repeated 5 times and the 6 supernatants ( strip solution ) pooled (total volume 3 ml). Incrementing the ionic strength of the stripping buffer, the interaction between carboxylic groups of Vi and quaternary ammonium residues on the resin is lost, while Vi-CRM197 remains bound to the matrix due to the presence of alkyl-hydroxyl groups. The 2.5 ml of the strip solution was desalted by PD10 column recovering 3.5 ml final volume. Samples of unbound (0.5 ml), wash solution (0.5 ml), and desalted strip solution (3.5 ml) were dried under reduced pressure (SpeedVac) and reconstituted in 0.5 ml (unbound and wash solutions) or in 0.65 ml water (strip solution). One hundred microliters of each sample were analysed by HPLC-SEC (fluorescence emission), eluted with 0.1 M NaCl, 0.1 M NaH2PO4, 5% ACN ph 7.2, with a 59

68 fluorescence detector, to verify the absence of protein and thus conjugates. Starting sample and strip solution were assayed for Vi content by HPAEC-PAD (240), using a Vi calibration curve with range of µg/ml. The ratio of Vi content in the strip solution (unconjugated Vi) and in the starting sample (total Vi), corrected for dilution, represents the proportion of free Vi in the sample H NMR spectroscopy Spectra were recorded as previously described (240). Dried Vi unconjugated or conjugated samples were solubilized in D2O and transferred to 5 mm NMR tube (Wilmad, 535-PP-7). Two spectra were collected: the first one in D2O and the second one after addition of NaOD to a final concentration of 200mM. The first spectrum was recorded in order to assure that other impurities did not fall at the same chemical shift of the acetate anion (released after de-o acetylation of the sample), with implications for the quantification of O-acetyl content. Addition of NaOD to the sample caused rapid de-o acetylation and consequently sharper peaks appeared in the proton spectrum. The spectrum of the de-o acetylated PS in alkaline medium was better suited for use as an identification test than that of the untreated sample, based on the presence of five typical signals (from about 5 to about 4 ppm) corresponding to the protons of the carbohydrate ring. Well resolved N-acetyl and acetate peaks were integrated so that the degree of O-acetylation of Vi repeating units was determined by comparison of these two integrals. EDAC derivatives % was calculated on the same spectrum in NaOD, by comparison of the integral of the Vi N-acetyl peak with that of the two methyl groups of the isopropyl moiety of EDAC. 60

69 2.2 Immunogenicity studies in mice The animal studies reported in this thesis were performed in different laboratories in Italy and UK. In all the immunogenicity studies performed in Italy eight mice per group were injected subcutaneously (s.c.) two times, at 4-5 week intervals, with 200 L/dose of Vi antigens diluted in saline solution NaCl 0.9% w/v without adjuvants. Female 10 week old outbred CD-1 mice, both wild type (WT) and T-cell deficient mice, were purchased from Laboratory. Injections, collection of bleeds and euthanasia were carried out by. Mice were bled and sera collected before first immunization (day 0), two weeks after the first immunization, 28 or 35 days (depending on the animal study) after first immunization and two weeks after second immunization. All animal protocols were approved by the local animal ethical committee and by the Italian Minister of Health in accordance with Italian law. For the animal studies performed in UK, WT C57BL6/J male mice aged 6-8 weeks were obtained from the. All animal studies performed were covered by Home Office approval. Mice were sacrificed seven or 145 days after immunization for investigating early- and long-term response respectively. Sacrifice was conducted by a schedule 1 method following the Home Office guidelines (UK). All animal studies performed were covered by Home Office approval. Mice were immunized intraperitoneally (i.p.) with conjugates diluted in sterile PBS without adjuvant. Each mouse received 200 L/dose of solution containing 2 g Vi. Some mice were also immunized with 61

70 unconjugated Vi. Vi PS was diluted in PBS without adjuvant and each mouse received 200 L/dose of solution containing 25 g Vi ELISA assays conducted by GVGH Immunoassay unit in Italy Serum IgG levels against Vi and carrier proteins were measured by ELISA using the method previously described (240). Round bottom 96-well ELISA plates [Maxisorp Nunc] were coated ON at 4 C with Vi or protein (CRM197, DT or TT) at 1.0 and 2 ug/ml in PBS respectively. Plates were blocked with 5% milk powder in PBS for 1h at RT and washed with PBS containing 0.05% Tween-20. Dilutions of mouse sera (1:100 or 1:4000) in PBS containing 0.1% BSA and 0.5% Tween 20 were deposited in triplicate on 3 different plates. Each plate contained a serially diluted anti-antigen standard serum in duplicate. One ELISA unit (EU) is defined as the reciprocal of the standard serum dilution that gives an absorbance value of 1 in this assay. Sera and standard sera were incubated for 2h at RT. After washing step, goat anti-mouse IgG alkaline phosphatase-labelled secondary antibody (Sigma A3438) was diluted 1: in PBS containing 0.1% BSA and 0.5% Tween 20, and incubated 1h at RT. After washing, alkaline phosphatase substrate (Sigma N2770) was dissolved in water, added in each well and incubated for 1h at RT. Absorbance at 405 nm and 490 nm was read using an ELISA reader. The absorbance values were calculated from the subtraction A405-A490. Absorbance of each well was converted in EU relative to anti-vi or anti-protein standard serum curves generated from the absorbance values of standard serum using a 4 parameter fit determined by modified Hill Plot. Data are presented as scatter plots of the individual IgG EU and bars represent the geometric mean EU of the group in log scale. 62

71 2.2.2 Immunological assessments performed at the University of Birmingham Conjugates and antigens All followed conjugates and antigens were used for the immunological assays as reported in Table 2.1. Table 2.1. List of conjugate and antigens used for the immunological assays conducted at the University of Birmingham Procedure/immunological assay Immunizations ELISA/ELISpot Immunohistology conjugate/antigen required Vi-CRM197 Vi-TT fvi-crm197 fvi-tt Vi CRM197 TT fvi-biotin Biotinylation of fvi Fragmented Vi with NAMW 43 kda was solubilized in 400 mm MES ph 6 at a concentration of 50 mg/ml. NHS and then EDAC were added to obtain 0.33 M NHS and EDAC/saccharide repeating units molar ratio of 5. After the reaction was mixed at RT for 1h, the linker (+)- biotin hydrazide [Sigma Aldrich], solubilized in DMSO, was added (molar ratio between biotin and Vi repeating units equal to 1). Biotinylation reaction proceeded for 2h at RT and the solution was then diluted in 3 M NaCl and desalted by PD10. The degree of saccharide repeating units biotinylated was of 20% as estimated by 1 H NMR analysis and saccharide amount was quantified by HPAEC-PAD. 63

72 Buffer preparations PBS at ph M NaCl, M KCl, M Na2HPO4, M KH2PO4 in MilliQ water Buffers used for ELISA ELISA carbonate coating buffer M Na2CO3, M NaHCO3 and 0.1% NaN3 in MilliQ water ELISA wash buffer 0.1% Tween 20 in PBS ELISA blocking buffer 1% BSA dissolved in PBS ELISA dilution buffer 0.05% Tween-20 and 1% BSA in PBS Buffers for Immunohistochemistry Tris buffer ph 7.6 Tris buffer ph 7.6 was obtained mixing: 1L of 200 mm Tris Base, 1.5 L of 152 mm Phisiological NaCl and 1.5 L of 0.1 M HCl Tris buffer ph 9.2 This buffer was prepared as Tris buffer ph 7.6, but adding 1 M HCl dropwise to reach ph 9.2 DAB solution A tablet of 3-3 -diaminobenzidine tetrahydrochloride (DAB) [Sigma Aldrich] was dissolved in 15 ml of Tris buffer ph 7.6 and filtered. Two drops of hydrogen peroxide [Sigma Aldrich] was added using a pasteur pipette and 2 drops of the solution were laid over each splenic section. Blue fast solution 64

73 Solution was prepared dissolving 8 mg of Levamisole [Sigma Aldrich] in 10 ml of Tris buffer ph 9.2. In a fume hood 4 mg of Napthol AS-MX [Sigma Aldrich] was dissolved in a glass bijoux containing 340 L of dimethyl-formamide and mixed with Levamisole solution. Ten mg of Fast Blue salt was added, vortexed and filtrated. One drop of the solution was then added to each splenic section ELISA Mouse sera were obtained from blood collected through tail bleed or cardiac puncture procedures. The sera were separated from clotted blood through centrifugation for 10 min at x and stored at -80 C. 96-well ELISA plates [Maxisorp NUNC] were coated ON at 4 C with Vi or protein (CRM197 or TT) at 5 g/ml in carbonate coating buffer. Plates were incubated for 1h at RT with blocking buffer. Dilutions of mouse sera in dilution buffer were deposited into a single well and diluted stepwise 1:3. Incubation time lasted 1h in humidified chamber at 37 C. Wells were washed with wash buffer. The goat anti-mouse alkaline phosphatase-labelled secondary antibody [Southern Biotech] was diluted in ELISA dilution buffer and added in each well. Dilution was dependent on the isotype involved in the ELISA assay as reported in Table 2.2. Secondary antibody was incubated for 1h at RT in humidified chamber at 37 C. After washing, p-nitrophenyl Phosphate tablets [Sigma FAST TM ] were dissolved in water, added in each well and incubated for 1h at 37 C in humidified chamber. Absorbance at 405 nm was read using the Emax precision microplate reader [Molecular Devices]. ELISA titres were calculated as inversion of the dilution at which sera reached a defined value of OD. 65

74 Table 2.2. Concentration of the secondary antibody correspondent to each isotype Secondary antibody isotype Catalogue number dilution IgM :2000 IgG :1000 IgG :1000 IgG :2000 IgG2b :2000 IgG2c :2000 IgA : Preparation of cell suspension from spleen and bone marrow Spleens were disrupted in a solution composed of RPMI media, 10% heat-inactivated fetal bovine serum and 1% penicillin/streptomycin. Cells from the bone marrow were collected from one hind leg, removing femur and tibia and flushing RPMI media through the bones. The cell suspensions obtained were filtered through 70 m cell-strainers and centrifuged at 394 x g for 4 min. Supernatants were discarded and pellets were re-suspended in 500 L ammoniumchloride-potassium lysis buffer for 5 min at RT. Cell suspensions were centrifuged as before and washed in RPMI media. Ten L of sample were diluted in Trypsan Blue staining and cell density was measured counting the cells by eye on fast reading counting slides. The final volume of the cell suspensions was adjusted depending on the required cell density. 66

75 Enzyme-Linked ImmunoSpot (ELISpot) Cells isolated from spleens and bone marrows were assessed by ELISPOT to reveal the presence of antigen-specific antibody secreting cells (ASCs). Multiscreen filter plates were coated ON at 4 C with Vi or protein (CRM197 or TT) at 5 or g/ml in PBS. Plates were washed with PBS and blocked with 10% heat-inactivated fetal bovine serum and 1% penicillin/streptomycin in RPMI media for 1 h at 37 C. After washing with PBS, 5x10 5 cells isolated from spleen or bone marrow were deposited in triplicate. Plates were incubated in a chamber with 37 C and 0.05% CO2 for 6h. Cells were lysed and washed away with a solution composed of PBS and 0.05% Tween 20. The goat anti-mouse alkaline phosphatase-labelled secondary antibody was diluted in PBS and added in each well, plates were incubated ON at 4 C. The final concentration of the secondary antibody was dependent on the isotype involved in the ELISA assay as reported in Table 2.2. Plates were washed three times with a solution of PBS with 0.05% Tween 20 and subsequently with PBS. Tablets of Sigma FAST 5-Bromo- 4-chloro-3-indolyl phosphate / Nitro blue tetrazolium (BCIP/NBT) [Sigma Aldrich] were dissolved in sterile water according to manufacturer s instructions and developed for 5 min Immunohistochemistry (IHC) Cryostat sectioning Mouse spleens were frozen in liquid nitrogen and stored at -80 C. Frozen spleens were sectioned into 5.5 m thick slices and mounted on four-spot slides. Slides were air dried for 30 min, fixed in acetone for 20 min at 4 C and air dried again. The cryosections were then stored at -20 C. 67

76 Staining Slides were let to thaw for 20 min at RT and subsequently hydrated in Tris buffer ph 7.6. Four different solutions of primary antibodies were prepared in Tris buffer ph 7.6 as follows: fvibiotin and IgD, fvi-biotin and IgM, fvi-biotin and IgG and fvi-biotin and IgA. Each mixture solution was dropped on sections and incubated for 1h in humidified chamber. Sections were washed in Tris buffer ph 7.6 for 5 min, incubated for 45 min with sheep anti-biotin diluted in Tris buffer ph 7.6 and washed again by Tris buffer ph 7.6. Sections were then incubated for 45 min with secondary antibodies that were previously adsorbed with 10% normal mouse serum in Tris buffer ph 7.6. Primary and secondary antibodies are reported in Table 2.3. Sections were washed by Tris buffer ph 7.6 and incubated with VectaStain streptavidinbiotinylated alkaline phosphatase complex for 40 min. Subsequently they were washed with Tris buffer ph 7.6, incubated for 1-2 min with DAB solution (see ) and washed again. The final development step occurred through incubation for 5-8 min with Blue Fast salt solution (see ). Sections were washed in Tris buffer ph 7.6, water and dried at RT. Finally sections were covered using coverslips and VectaMount [Vecta Laboratories]. Images of the stained sections were collected by Leica CTR6000 microscope and QCapture software. 68

77 Table 2.3. Reagent components used for staining procedure Isotype Source Dilution or concentration Primary antibodies IgD Rat antimouse BD Pharmingen, Clone 11-26c.2a, :500 IgM IgG Rat antimouse Rat antimouse AbD Serotec, MCA199 1:500 AbD Serotec, MCA424 1:300 IgA Rat antimouse BD Pharmingen, Clone C10-1, :300 Secondary antibodies Donkey anti Sheep-biotin / Jackson Immuno Research Laboratories, :100 Rabbit anti-rat- Px / DAKO, P0450 1:50 Others fvi-biotin / GVGH 1 g/ml Sheep antibiotin / Bethyl Laboratories, A A 1:800 69

78 Bacteria growth and detection of Vi expression The Salmonella enterica serovar Typhimurium strains reported in Table 2.4 were kindly provided by Andreas Bäumler (255). Table 2.4. List of Salmonella Typhimurium strains growth under different osmolarity conditions Salmonella Typhimurium strains IR715 fepe IR715 fepe phon::viab IR715 fepe phon::viab with aroa mutation IR715 phon::viab with aroa mutation IR715 phon::viab rpos::cm r with aroa mutation IR715 phon::viab IR715 phon::viab rpos:: Cm r phon::viab mutants obtained by insertion of viab into phon gene (255, 256) fepe mutants lacking very long O-antigen chains expression (256) rpos is a master regulator required to survive under extreme conditions (257) All strains were incubated ON at 37 C with constant agitation (180 rpm) in Luria-Bertani (LB) broths with differing NaCl concentration: 8.5, 85 and 500 mm. When OD reached 0.8 value, the culture was centrifuged, washed with PBS and re-suspended in 1 ml PBS with 0.05% sodium azide. The protein concentration was assessed by micro BCA following manufacturer s instruction [Thermo Fisher]. 96-well ELISA plates were coated with each bacterial strain at 10 g/ml in PBS and incubated ON at 4 C. ELISA assay was conducted as 70

79 reported before, but a single anti-vi mouse serum was used as primary antibody diluted stepwise Bacterial infection For infection with S. Typhimurium IR715 phon::viab rpos::cm r strain, bacteria were grown in LB broth with 85 mm NaCl ON at 37 C. When OD reached 0.8 value, 1 ml of culture broth was collected. Bacteria were harvested by centrifugation at 6000 x g for 5 min and washing with PBS three times. Bacteria were diluted in PBS in order to infect each mouse with a fixed bacterial dose in 200 L of saline solution. Bacterial solution was serially diluted and plated on agar plates to verify the infection dose Burden of bacteria Spleen and liver samples were disrupted in PBS. Suspensions were filtered through 70 m cell-strainer and serially diluted in RPMI media. Suspensions were diluted stepwise 1:10, plated on LB agar plates and incubated ON at 37 C. Bacterial burden was measured by counting bacterial colonies developed on the plates. The total colony forming units (CFUs) per organ were calculated considering the whole organ mass and the dilution. 2.3 Statistics Statistical and graphical analysis was performed using GraphPad Prism 7 software. The nonparametric Mann-Whitney test and Kruskal-Wallis analysis with Dunn s test for post hoc analysis were used to compare respectively two or multiple groups. Wilcoxon matched-pairs signed rank two-tailed test was used to compare results from the same group at different time points. 71

80 3 Influence of conjugation variables on the immunogenicity of a Vi vaccine against Salmonella Typhi 3.1 Introduction The persistent high occurrence of Salmonella enterica serovar Typhi in LMICs (183) and the emerging appearance of MDR strains (182, 258) have guided the development of vaccines against S. Typhi. Currently two vaccines are licensed, the live attenuated Ty21a strain (Vivotif TM ) and the purified Vi PS; but both vaccines are not reliable for children under 2 years of age because of the lack of protection induced (187, 259). It is well known that glycoconjugates can overcome immunological drawbacks typical of PS vaccines, converting TI into TD response (66). Different Vi-based conjugates are in development and some of them have been tested in clinical trials with promising results in children and infants (230, 235, 241). In particular two glycoconjugates have been launched in the last decade. Vi-TT conjugate with the trade name Pedatyph TM has been licensed for in-country use in India since 2008, and another Vi-TT conjugate (Typbar-TCV) was officially launched in 2013 in India (198, 199). GVGH has developed a Vi-CRM197 conjugate vaccine against S. Typhi, by linking Vi PS from Citrobacter freundii to CRM197 protein (239, 240). This conjugate has been tested in mice revealing safety and good immunogenicity in absence of adjuvant (238, 240, 248) and inducing good protection against challenge with a Vi + S. Typhimurium strain (248). Furthermore this Vi conjugate (240, 248) has already been tested in Phase 1 clinical trials in 72

81 healthy European adults inducing higher immunogenic response compared with unconjugated Vi PS administered at higher dose (5 g Vi-CRM197 against 25 g Vi) (242). Phase 2 clinical trials in South and Southeast Asia have been conducted to test this Vi conjugate in adults, children and older and young infants (241). In adults, one month after vaccination, anti-vi responses were significantly higher for Vi-CRM197 versus Vi PS. However, after 6 months, a significant decrease of antibody titres was observed and similar levels were revealed in adults immunized with Vi-CRM197 or Vi PS. Children and older infants were vaccinated twice meanwhile young infants received three injections of Vi-CRM197: after re-immunization no booster effect was observed in all age groups and antibody level declined gradually during 6 months after last immunization (241). Similarly absence of antibody enhancement response was revealed in children receiving secondary Vi-TT (Pedatyph TM ) vaccination (232). Several conjugation parameters can affect the immunogenicity induced by glycoconjugates (66), such as carrier protein, saccharide chain length, conjugation chemistry, saccharide to protein ratio and size of the conjugate. These conjugation parameters result interconnected among each other, therefore several times studies over the effect of a single conjugation parameter on immunogenicity have been conducted changing other variables simultaneously. The aim of this part of my PhD project was to test the impact of different conjugation variables on the immunogenicity of Vi conjugates. The investigation was conducted following a systematic approach changing one variable per time in order to identify the major parameters impacting the immune response and to develop a vaccine with improved ability to protect. 73

82 3.2 Summary The objective of this Chapter was the investigation of the impact of different conjugation variables on the immunogenicity of Vi glycoconjugates, in order to develop improved vaccines against S. Typhi. Some studies have investigated the influence of such parameters on the immune response induced by Vi conjugate vaccines, but they have limited the studies focusing on the effect of a single parameter or changing multiple parameters within a single conjugate vaccine. In contrast, this investigation was conducted following a systematic approach, altering just one parameter in each set of vaccine candidates tested, whilst keeping the others constant. Vi glycoconjugates that differ in Vi chain length (full and fragmented), carrier protein, conjugation chemistry, saccharide to protein ratio and size were produced and tested in mice. The length of Vi chains, but not the ultimate size of the conjugate, has an impact on the anti- Vi IgG immune response induced. Full-length Vi conjugates, independent of the carrier protein, induce peak IgG responses rapidly after just one immunization, and secondary immunization does not enhance the magnitude of these responses. Fragmented Vi linked to CRM197 and diphtheria toxoid, but not to tetanus toxoid, gives lower anti-vi antibody responses after the first immunization than full-length Vi conjugates, but antibody titers are similar to those induced by full-length Vi conjugates following a second dose. The chemistry to conjugate Vi to the carrier protein, the linker used, and the saccharide to protein ratio do not significantly alter the response. We conclude that Vi length and carrier protein are the variables that influence the anti-vi IgG response to immunization the most, while other parameters are of minor importance. 74

83 3.3 Results Optimization conjugation process and characterization of Vi-CRM 197 conjugate obtained Main conjugate strategy adopted The original method used at GVGH for the synthesis of Vi-CRM197 conjugate vaccine (Scheme 3.1 A) (240, 248) has been slightly modified (Scheme 3.1 B) in order to increase the conjugation efficiency of Vi PS, with the aim to have less than 20-30% of unconjugated Vi at the end of the reaction making conjugate purification easier. The new method consists of the following steps: reacting COOH groups of Vi with EDAC to obtain an O-acylisourea ester intermediate; reacting the O-acylisourea ester intermediate with NHS to form a semi-stable NHSester (69); reacting the NHS-ester with CRM197 previously derivatized with ADH. 75

84 Scheme 3.1. Optimization of the conjugation process through the formation of NHS-ester. A) Original method: Vi is activated with EDAC, giving unstable reactive O-acylisourea esters on Vi, and conjugated to CRM-ADH. B) Modified method: Vi is activated with EDAC and NHS, with formation of more stable and amine-reactive NHS-ester group on Vi that can react with CRM-ADH Several reactions were tested following Scheme 3.1 B and optimal conditions, such as percentage of activated Vi repeating units, protein derivatization level and reagent concentrations, were identified. In the step of Vi activation, the PS has a concentration of 4.2 mg/ml with a molar ratio of carbodiimide to Vi repeating units of 5:1. The reaction is performed at ph 5-6, optimal for carbodiimide reactions (69). In the same step, the concentration of NHS used is 0.1 M. The resulting intermediate (Vi-NHS) can be isolated and analysed, after desalting at low ph, for total sugar content by HPAEC-PAD and ion pair 76

85 HPLC-RP for NHS quantification, so to determine the percentage of activated Vi repeating units. The reaction conditions selected resulted in an activation of Vi repeating units close to 20-30%, thought to be a good compromise for efficient conjugation without impacting too much on Vi epitopes by the modification of carboxylic groups on the saccharide backbone. Good reproducibility was obtained when the reaction was scaled from 10 (18% Vi repeating units activated) to 40 (18-21% Vi repeating units activated) and 150 mg Vi (26-28% Vi repeating units activated). Derivatization of CRM197 with ADH was performed at ph 6.0 rather than in order to avoid protein precipitation (240), and the use of MES buffer guaranteed maintaining the ph constant during the entire reaction time. The derivatization was conducted both at 100 and 500 mg protein scale. In all preparations, the final protein recovery after purification was good (>80%) with an average number of 6 ADH linkers introduced per protein, as verified by MALDI-MS analysis. The conjugation step of activated Vi-NHS with CRM-ADH was performed at ph 6, with a Vi concentration between mg/ml. Higher Vi concentrations resulted in too viscous solutions and too much crosslinking of the resulting conjugate, with gel formation and conjugate precipitation Reproducibility of conjugate formation at small scale Several lots of Vi-CRM197 conjugates were synthesized at 10 mg Vi scale to verify the reproducibility of the process. Even using slightly different Vi concentrations in the conjugation mixture, data confirmed the consistency of the process at small scale (Table 3.1). All preparations contained undetectable levels of free protein and < 20% free Vi, complying with the relevant WHO Technical Report (244). 77

86 Table 3.1. Reproducibility of Vi-CRM 197 conjugates at small scale using two different Vi to CRM 197 ratios Lot conjugate Vi/CRM 197 w/w Conj. Mixture [Vi] mg/ml Vi/CRM 197 ratio w/w % free Vi Purified conjugate % free CRM 197 A <20 nd B nd 1:1 C <20* nd D <16* nd E nd F <20* nd G 1: <20* nd H nd I nd nd: not detectable, * % free Vi values refer to the ratio between the minimum quantifiable amount of free Vi by HPAEC-PAD to total amount of polysaccharide in the conjugate, w/w: weight to weight Scale-up of the process for Vi-CRM 197 conjugate production and conjugate characterization The conjugation process was scaled up to 150 mg Vi, working with both the Vi to protein ratios tested at smaller scale. Conjugations conducted with Vi to CRM197 w/w ratios 1:1 or 1:2 gave consistent results in terms of recoveries and quality of the final product. Table 3.2 reports the full characterization of some lots produced at mg Vi scale compared with lots at 10 mg Vi scale. 78

87 Table 3.2. Scale up and consistency of the process for Vi-CRM 197 conjugates production Characterization Tests Vi scale 10 mg 150 mg 10 mg 100 mg Vi to CRM 197 w/w ratio in conjugate mixture 1:1 1:2 Total Vi content ( g/ml) % Vi recovery post-purification Total protein content ( g/ml) % protein recovery post-purification w/w Vi per CRM Free Vi 16%* 20% <10%* 9% Free protein Not detectable Not detectable Not detectable Not detectable MW distribution (Kd on Superose 6) % Vi O-acetylation level Saccharide composition Vi identity confirmed Vi identity confirmed Vi identity confirmed Vi identity confirmed EDAC derivatives (mol/mol Vi RU) 0.6% 0.3% 1.2% 0.6% Residual NHS (mol/mol Vi RU) <1% 0.9% 0.3% <1% * % values refer to the ratio between the minimum quantifiable amount of free Vi by HPAEC-PAD to total amount of polysaccharide in the conjugate, w/w: weight to weight Figure 3.1 shows the HPLC-SEC profile in fluorescence emission of a representative Vi- CRM197 conjugate. Before purification by TFF, the conjugation mixture showed the presence 79

88 of unreacted protein, that was completely removed after purification. The peak of the conjugate was shifted at higher MW compared with that of the unconjugated protein. Analysis by 1 H NMR (Figure 3.2 reports the spectrum of a representative Vi-CRM197 conjugate) confirmed that O-acetylation level remained high after conjugation, while percentage of EDAC derivatives was very low (<1% in moles compared to Vi repeating units for all prepared lots). Low residual percentages were found for NHS groups (by ion pair HPLC-RP analysis), confirming that the method used did not alter the Vi PS structure. 80

89 81

90 82

91 3.3.2 Impact of saccharide chain length Production of fragmented Vi populations at reduced molecular weight compared with full-length Vi Full-length Vi with NAMW 165 kda, (Figure 3.3) was fragmented using hydrogen peroxide (260). Fragmented Vi populations of different NAMW were separated by ion exchange chromatography. Four different fvi (pools 1-4) were isolated, ranging from 9.5 to 82 kda, as determined by HPLC-SEC (Figure 3.3; Table 3.3). Dextrans were used as MW standards due to the similar conformation with Vi saccharides. Analysis by 1 H NMR confirmed that Vi PS structure was not altered by fragmentation and that greater than 60% of monosaccharides retained O-acetylation, a critical parameter for immunogenicity of Vi PS based vaccines (243) (Table 3.3). Table 3.3. Physical characteristics of fragmented Vi pools compared to full-length Vi Saccharides PS NAMW (kda) % OAc level fvi pool fvi-crm 197 pool fvi-crm 197 pool fvi-crm 197 pool full-length Vi-CRM NAMW: number average molecular weight, PS: polysaccharide, fvi: fragmented Vi, % Oac level: percentage of Vi repeating units O-acetylated 83

92 84

93 Investigation of conjugate conditions in order to obtain fvi-crm 197 conjugate vaccines The conjugation procedure applied to full-length Vi (NAMW 165 kda) was shifted to fragmented Vi chains (Table 3.3), but conjugate formation did not occur. As reported in section , full-length Vi has been activated by NHS in order to form semi-stable NHSester on saccharide repeating units (Scheme 3.1 B). The activation step for full-length Vi was conducted fixing Vi PS and NHS concentration respectively at 4.2 mg/ml and 0.1 M in order that 20-30% saccharide repeating units were activated. In order to define optimal conditions for conjugating fragmented Vi chains with CRM197 protein, different conditions were tested by working with fvi NAMW 43 kda: NHS concentration was increased to 0.3 M, fvi concentration and saccharide repeating units to EDAC molar ratio were modulated in the range mg/ml and 1-10 respectively. Comparing activation degrees on saccharide chains obtained fixing NHS concentration at 0.33 M, both fvi concentration and EDAC to fvi repeating units molar ratio enhanced the percentage of activated saccharide repeating units (Table 3.4). Table 3.4. Different conditions tested for fvi NAMW 43 kda activation with EDAC/NHS and resulting in different percentage activated fvi repeating units [NHS] M [fvi] mg/ml EDAC/ fvi RU molar ratio % activated RU RU: repeating units, fvi: fragmented Vi, % activated RU: percentage of saccharide repeating units activated by NHS (Scheme 3.1 B) 85

94 CRM197 protein was linked to fvi with 13 and 23 % of activated repeating units. Conjugations were conducted with fvi concentration of 7.8 mg/ml and different fvi to protein w/w ratios (Table 3.5). Table 3.5. Physical characteristics of fragmented Vi-CRM 197 conjugates using fvi with NAMW 43 kda, different percentage of activated repeating units and w/w fvi to CRM 197 ratio fvi/crm 197 w/w ratio in conjugate mixture % activated RU lot Purified conjugate fvi/crm197 w/w ratio CRM197 % free % recovery fvi 1:1 13 no conjugate formation CRM197 fvi 23 A 0.66 nd < :2 13 no conjugate formation 23 B 0.26 nd < :1 13 no conjugate formation 22 C 0.85 nd < nd: not detectable, RU: repeating units, fvi: fragmented Vi, w/w: weight to weight The optimal percentage of activated fvi repeating units for conjugation was 23%, instead fvi with 13% activated repeating units did not give conjugate formation in all the conditions tested (Table 3.5). The fvi to CRM197 ratio used in conjugation impacted on the ratio in the purified conjugates: higher the ratio in the conjugate mixture, higher the ratio in the purified conjugate (Table 3.5). Also the influence of fvi concentration on conjugation efficiency was evaluated by working with fvi to CRM-ADH w/w ratio equivalent to 1:1 (Table 3.6). Conjugate could be formed if the saccharide concentration was equal or major to 7.8 mg/ml (Table 3.6). 86

95 Table 3.6. Physical characteristics of fragmented Vi-CRM 197 conjugates obtained by working with fvi to CRM 197 w/w ratio of 1:1 and modulating the fvi concentration in the reaction mixture [fvi] in conjugate mixture mg/ml lot Purified conjugate fvi/crm197 w/w ratio % free % recovery 2.5 no conjugate formation 5 no conjugate formation 7.8 D 0.59 nd < E 0.78 nd F 0.66 nd nd: not detected, fvi: fragmented Vi, w/w: weight to weight CRM197 fvi CRM197 fvi Based on these results for conjugating CRM197 protein with fvi (NAMW 45kDa), conjugations for all fvi chains at different NAMW (Table 3.3) were conducted as follows: the EDAC-NHS activation step was performed by working with NHS and fvi concentrations of 0.33 M and 50 mg/ml respectively and EDAC to fvi repeating units molar ratio of 5:1. Percentages of activated fvi repeating units ranged from 20 to 30%. The conjugation step was performed with fvi concentration of 7.8 mg/ml and fvi to protein w/w ratio of 1:1. With the longer saccharide chains (pool4 NAMW 82 kda), these conditions led to gel formation during the conjugation step, thus fvi concentration was decreased to 15 mg/ml during the step of saccharide activation. Even though the percentage of activated fvi (pool4) repeating units was low (only 9.3%), it allowed fvi-crm197 conjugate formation. 87

96 All the conjugates were characterized by less than 0.02 moles of carbodiimide derivatives (e.g. the N acyl urea groups) per mole of Vi monosaccharide and less than 0.01 moles per mole of residual ester groups from NHS. All conjugates contained undetectable levels of unconjugated protein and <20% free Vi. Table 3.7 lists characteristics of obtained conjugates. Table 3.7. Physical characteristics of Vi-CRM 197 conjugates differing for Vi chain length Conjugate PS NAMW (kda) PS to CRM 197 ratio (w/w) % free PS fvi-crm 197 pool fvi-crm 197 pool <5 * fvi-crm 197 pool fvi-crm 197 pool full-length Vi-CRM <13 * NAMW: number average molecular weight, PS: polysaccharide, * % values refer to the ratio between the minimum quantifiable amount of free Vi by HPAEC-PAD to total amount of polysaccharide in the conjugate, w/w: weight to weight Immunogenicity of fragmented Vi-CRM 197 conjugates compared with full-length Vi-CRM 197 in mice Conjugates composed by CRM197 linked to Vi of different chain lengths (Table 3.7) and their corresponding unconjugated saccharide chains were tested in WT mice at 8 g Vi/dose (Figure 3.4 A). Full-length Vi-CRM197 was able to induce a greater anti-vi IgG response compared with unconjugated full-length Vi (12.8 times higher). As for unconjugated Vi, the peak response 88

97 occurred after one injection, with no enhancement observed following the second vaccination (Figure 3.4 A). This lack of anamnestic response was not due to the maximal response being achieved after the first dose. When WT mice were immunized with the full-length conjugate at lower doses of Vi (0.04, 0.35 and 2.8 g), the anti-vi IgG response was significantly lower than at higher Vi dose after one injection, but again there was an absence of increase in antibody following the second injection (Figure 3.5). Conversely to full-length Vi-CRM197 conjugate, for all fragmented Vi conjugates, despite a lower anti-vi IgG response 14 days after the first injection compared to the full-length Vi conjugate (significant differences for fvi NAMW 9.5, 22.8 and 42.7 kda compared to fulllength Vi, p = 0.04, <0.0001, 0.03 respectively), there was a significant increase in anti-vi IgG following the second injection with an average 13 fold increase between day 35 and day 49 (p = ). Final anti-vi IgG levels (at day 49) were comparable to those induced by the full-length Vi conjugate (Figure 3.4 A). The conjugate of fvi with NAMW 82 kda gave intermediate results compared with the conjugates of fvi with lower NAMW and full-length Vi conjugate: the response at 14 days after first injection was higher than with shorter fragmented Vi conjugates and not significantly different from that of the full-length Vi conjugate (Figure 3.4 A). Among unconjugated saccharides characterized by different chain lengths, only full-length Vi induced substantial anti-vi IgG response 14 days following the first injection, with no enhancement of antibody levels after the second injection (Figure 3.4 A). In contrast, the fvi with NAMW 9.5, 22.8 or 42.7 kda induced a minimal response 14 days after the first injection, and for each of these, the anti-vi IgG levels were significantly less than the response 89

98 to the full-length (165 kda) Vi (p = 0.002, , 0.03, respectively). Following a second injection, these three fvi populations induced increased levels of IgG antibody to Vi PS, but remained about three times lower than the response to full-length Vi at the same time point (Figure 3.4 A). The response of the fvi with NAMW 82 kda was intermediate between the lower-sized fvi populations and full-length Vi. At day 14, this was not significantly different to the response of the shorter fvi or the full-length Vi (Figure 3.4 A). To further investigate on the nature of the immune response induced by fvi-crm197 conjugates compared to full-length Vi-CRM, both WT and T-cell deficient mice were immunized with fvi-crm197, full-length Vi-CRM197 conjugates and full-length unconjugated Vi (Figure 3.4 B). Only unconjugated and conjugated full-length Vi were able to induce a response significantly higher than time 0 background in T-cell deficient mice, while no response was induced by any of the fragmented Vi conjugate vaccines. The response induced by conjugated full-length Vi was higher in WT compared with T-cell deficient mice (p=0.05 at day 14; p= at day 42); in contrast higher antibody response induced by unconjugated full-length Vi was observed in T-cell deficient compared to WT mice (p = day 14, p = day 42) (Figure 3.4 B). 90

99 Figure 3.4. Identification of a critical size (approximately NAMW 82 kda) below which both unconjugated and conjugated fragmented Vi polysaccharide can no longer act as T-independent antigens. A) Anti-Vi IgG response induced in WT mice following immunization with unconjugated fulllength or fragmented Vi and their corresponding CRM 197 conjugates expressed in ELISA units. Wild type mice are injected subcutaneously at days 0 and 35 with 8 µg of Vi antigen/dose. Individual animals are represented by the dots in log scale and horizontal bars represent Geometric Mean ELISA Units. Statistical analysis is conducted as reported in paragraph

100 Figure 3.4. Identification of a critical size (approximately NAMW 82 kda) below which both unconjugated and conjugated fragmented Vi polysaccharide can no longer act as T-independent antigens. B) Anti-Vi IgG response induced in WT and T-cell deficient mice following immunization with unconjugated full-length Vi, or conjugates obtained linking CRM 197 with both full-length and fragmented Vi. Mice are injected subcutaneously at days 0 and 28 with 8 µg of Vi antigen/dose. Antibody responses are expressed in ELISA units, individual animals are represented by the dots in log scale and horizontal bars represent Geometric Mean ELISA Units. Statistical analysis is conducted as reported in paragraph

101 Figure 3.5. Absence of anamnestic anti-vi antibody response persists when Vi-CRM 197 conjugate is administered at different Vi doses. Anti-Vi IgG response induced in WT mice following subcutaneous immunization with Vi-CRM 197 conjugate at days 0 and 28 with 0.04, 0.35 and 2.8 µg of Vi antigen/dose. Antibody responses are expressed in ELISA units, individual animals are represented by the dots in log scale and horizontal bars represent Geometric Mean ELISA Units Influence of crosslinking/size and saccharide to protein ratio Based on the results obtained with Vi conjugates of different chain length (section ), the impact of conjugate crosslinking/size and saccharide to protein ratio was evaluated on the immunogenicity of full-length (165 kda) and fvi NAMW 43 kda, by using CRM197 as carrier protein. 93

102 Investigation over conjugates obtained with full-length Vi Full-length Vi-CRM197 was run on a Sephacryl S column and the leading (HMW) and tailing fractions (LMW) of the resulting peak were collected separately. Fractionated conjugates were characterized by same Vi to CRM197 w/w ratio of 0.8 (as for non-fractionated conjugate) but different size (Figure 3.6), likely related to the different level of crosslinking. Figure 3.6. Selected fractions, collected from the elution of a representative Vi-CRM 197 conjugate, compose two Vi conjugates characterized by different size. The Vi-CRM 197 conjugate is eluted on Sephacryl S1000 column flushing the mobile phase constituted by 10 mm NaH 2 PO 4 ph 7.2. Fractions collected in the yellow area compose the Vi-CRM 197 conjugate with high molecular weight (HMW), meanwhile the fractions pooled in the red area constitute the Vi-CRM 197 conjugate with low molecular weight (LMW) 94

103 Wild type mice were immunized in a Vi dose escalating experiment (0.125, 1 and 8 Vi g/dose), comparing HMW and LMW fractionated conjugates with the non-fractionated conjugate (Figure 3.7 A). The anti-vi IgG antibody response was related to the dosage of Vi for all the conjugates tested: the higher the dosage, the higher the level of antibodies after each of the two injections (Spearman rank test). HMW and LMW Vi-CRM197 did not induce a different anti-vi IgG response 14 days after neither first nor second injection. HMW and LMW conjugates did not induce a different anti-vi IgG response from non-fractionated Vi-CRM197 at all days and all doses tested (for day 42 p= 0.08 at g/dose; p= 0.05 at 1 g/dose, p= 0.80 at 8 g/dose). Anti-CRM197 IgG response was minimal 14 days after first injection, and increased after the second one. Fourteen days after the second injection the anti-crm197 response was proportional to the amount of protein injected (data not shown). To look at the impact of Vi to protein ratio on immune response, full-length Vi-CRM197 conjugates characterized by w/w Vi to protein ratio equal to 1.4 and 0.8 respectively (Table 3.1) and showing similar size by HPLC-SEC (data not shown), were compared in mice at 0.35 and 1 g Vi/dose. At 0.35 g Vi/dose, the conjugate with a lower Vi to protein ratio was able to induce stronger anti-vi IgG response after the first and second injection (p = at day 42) (Figure 3.7 B), and stronger anti-crm197 IgG response after the second injection (p = at day 42) compared to the conjugate with the higher Vi to protein ratio (Figure 3.7 C). However, when the conjugates were compared at higher dose (1 g Vi), anti-vi and anticarrier protein IgG responses were similar (Figure 3.7 B and C). 95

104 Figure 3.7. A) Comparable anti-vi IgG antibody responses are induced by Vi-CRM 197 conjugate non-fractionated and Vi-CRM 197 conjugates obtained pooling the fractions with different conjugate size. All conjugates are characterized by same Vi to protein ratio. Blue area collects anti- Vi IgG responses elicited by non-fractionated conjugate, yellow and red areas report anti-vi IgG responses elicited by high molecular weight (HMW) and low molecular weight (LMW) conjugate respectively. Wild type mice are immunized subcutaneously at days 0 and 28 with 0.125, 1 and 8 µg of Vi antigen/dose. Antibody responses are expressed in ELISA units, individual animals are represented by the dots in log scale and horizontal bars represent Geometric Mean ELISA Units 96

105 Figure 3.7. Vi-CRM197 conjugates, with similar conjugate size but differing for Vi to protein ratio do not alter anti-vi (B) and anti-carrier (C) IgG antibody responses when injected at 1 µg of Vi antigen/dose. Wild type mice are immunized subcutaneously at days 0 and 28 with 1.4 and 0.8 µg of Vi antigen/dose. Antibody responses are expressed in ELISA units, individual animals are represented by the dots in log scale and horizontal bars represent Geometric Mean. Statistically significant differences are represented by asterisks (** and *** are correspondent to p<0.005 and < respectively) and statistical analysis is conducted as reported in paragraph

106 Investigation over conjugates obtained with fvi with NAMW 43 kda Fragmented Vi-CRM197 conjugates characterized by the same fvi to protein w/w ratio but different size (obtained by using a fvi concentration in the conjugation mixture of 7.8 or 20 mg/ml and a fvi to protein ratio 1:1 w/w in both cases, Table 3.5 and 3.6), and by different fvi to protein ratios with similar size (obtained by using different fvi to protein ratios in reaction, Table 3.5) were tested in mice (Figure 3.8, Table 3.8). Table 3.8. Fragmented Vi-CRM 197 conjugates differing for fvi to CRM 197 w/w ratio or size fvi-crm 197 lot [fvi] in conjugation mixture mg/ml fvi to CRM 197 w/w ratio in conjugation mixture fvi to CRM 197 w/w ratio in purified conjugate A B C F w/w: weight to weight 98

107 99

108 All the conjugates were tested in WT mice both at 1 and 8 g fvi/dose. For all the conjugates variability in anti-vi antibody response among individual mice, with some non-responders was observed. However, at both doses tested, no statistically significant differences were observed among anti-vi antibody responses elicited by all the groups (Figure 3.9 A). Differences in geometric means among the groups are reflection of non-responders mice. At 8 g fvi/dose, anti-crm197 IgG response was minimal after one only injection for all the constructs, and was similar after re-immunization independent of the dose of carrier injected (Figure 3.9 B). Only at 1 g fvi/dose, the anti-crm197 IgG response induced by the conjugate with 0.85 w/w fvi to protein ratio was significantly lower (p = 0.047) than the response induced by the conjugate with 0.26 w/w ratio (Figure 3.9 B). Otherwise responses were similar. Comparable anti-vi IgG responses were induced by fragmented Vi-CRM197 conjugates with same fvi to protein ratio (0.66 w/w) but different size (lot A and F in Table 3.8, Figure 3.9 A). As reported before, differences in geometric means among the groups are reflection of non/low-responders mice. At 1 g fvi/dose, the larger conjugate did not induce an anti- CRM197 IgG response (Figure 3.9 B). There was no statistically significant difference in the anti-crm197 IgG response when the conjugates were tested at 8 g fvi/dose. This study confirmed the ability of fragmented Vi conjugates to induce anti-vi booster response after a second dose (Figure 3.9 A). 100

109 Figure 3.9. A) Comparable anti-vi IgG response induced by fragmented conjugates differing for saccharide to protein ratio or conjugate size. Wild type mice were immunized at day 0 and 28 at 1 and 8 g fvi/dose. Bars represent the Geometric Mean ELISA Units of the group in log scale, individual animals are represented by the scatter plots. Statistically significant differences are represented by asterisks (* and ** are correspondent to p<0.05 and <0.005 respectively) and statistical analysis is conducted as reported in paragraph

110 Figure 3.9. B) Comparable anti-carrier IgG response are induced by fragmented conjugates with similar size but differing for saccharide to protein ratio. Wild type mice are immunized at day 0 and 28 at 1 and 8 g fvi/dose. Bars represent the Geometric Mean ELISA Units of the group in log scale, individual animals are represented by the scatter plots. Statistically significant differences are represented by asterisks (* and ** are correspondent to p<0.05 and <0.005 respectively) and statistical analysis is conducted as reported in paragraph Impact of carrier protein The impact of the carrier protein on immunogenicity was investigated on full-length saccharide (165 kda) and fvi with NAMW 43 kda. CRM197 was compared to DT and TT, two other proteins commonly used in the production of glycoconjugate vaccines (33, 66), including Vi conjugate vaccines (137, 198). 102

111 The same protocol of derivatization with ADH used for CRM197 (240) was applied to DT and TT, resulting in a different average number of ADH molecules per mole of protein. In fact, an average number of 12 and 24 ADH linkers was introduced on DT (236) and TT respectively, compared to 6 linkers introduced on CRM197 (Figure 3.10). Figure Same derivatization protocol applied on CRM 197, DT and TT (240) determines different derivatization degrees. The number average molecular weight of the protein (derivatized and not) is determined by MALDI-TOF spectrometry. The average number of ADH linkers introduced on the protein is calculated from the shift of the molecular weight 103

112 DT-ADH and TT-ADH were used for the synthesis of the corresponding conjugates, with a saccharide to protein ratio of 1:1 (w/w), and using the protocols optimized for CRM197 with full- length Vi and fvi. With full-length Vi, higher percentage of DT and TT remained unconjugated after 2h of reaction compared to CRM197 (close to 50% for TT and 70% for DT compared to values of 25-30% usually found with CRM197 working in the same conditions). The purification of Vi-DT and Vi-TT through TFF 300k, used for Vi-CRM197 conjugate, failed as the conjugates passed through the membrane, probably because of a different conformation of the products obtained. For this reason, Vi-DT was purified by TFF 100k and Vi-TT by size exclusion chromatography due to the higher MW of the protein itself. HPLC- SEC profile of the resulting purified conjugates showed that Vi-DT and Vi-TT were characterized by two main populations at different NAMW, while Vi-CRM197 conjugate had a unique population at higher MW (Figure 3.11 A). 104

113 Figure Profiles of full-length conjugates, except CRM 197 conjugate, are characterized by two molecular weight populations. In contrast, all fragmented Vi conjugate are characterized by a unique molecular weight population. A) HPLC-SEC profiles (fluorescence emission detection) of full-length conjugates: Vi-CRM 197 (red), Vi-DT (brown) and Vi-TT (green). Full-length conjugates are eluted on TSK gel PW columns and mobile phase (0.1 M NaCl, 0.1 M NaH 2 PO 4, 5% ACN, ph 7.2) flushes in isocratic condition at 0.5 ml/min. B). HPLC-SEC profiles (fluorescence emission detection) of fragmented conjugates: fvi-crm 197 (red), fvi-dt (brown), fvi-tt (green). Fragmented conjugates are eluted on TSK gel 3000 PWxl column and mobile phase (0.1 M NaCl, 0.1 M NaH 2 PO 4, 5% ACN, ph 7.2) flushes in isocratic condition at 0.5 ml/min Working with fvi, with all the three proteins no free protein was detected in the reaction mixture at the end of conjugation reaction. As for the full-length conjugates, DT conjugate was characterized by higher saccharide to protein w/w ratio than fvi-crm197 (Table 3.8, Figure 3.11 B). All conjugates reported in Table 3.9 were compared in WT mice receiving 1 g Vi/dose. 105

114 Table 3.9. Characterization of full-length Vi and fragmented Vi conjugates by using different carrier proteins and saccharide to protein w/w ratio 1:1 in the conjugation mixture Conjugate n ADH per protein Total saccharide to protein w/w ratio % free saccharide % free protein Vi-CRM nd Vi-DT nd Vi-TT nd fvi-crm <15 nd fvi-dt <20 nd fvi-tt <6.8 nd nd: not detectable All full-length Vi conjugates induced high anti-vi antibody responses 14 days after the first injection, with no booster after reinjection (Figure 3.12 A). Anti-Vi IgG response was similar independent of the carrier protein. Looking at the impact of the different carrier proteins on immune response elicited by fragmented Vi conjugates, no differences in the anti-vi IgG response induced were observed two weeks after second injection. Fourteen days following first injection, however, the anti-vi IgG response induced by fvi-tt conjugate was significantly higher than that induced by fvi-dt conjugate (p = 0.012). A significantly higher anti-vi IgG response was reveled after re-injection compared with that elicited 14 days after first injection for groups receiving fvi-crm197 (p= ) and fvi-dt (p =0.039) conjugates. Differently fvi-tt did not give booster after the second injection, but similarly to full-length conjugates the response peaked after the first injection. For all the fragmented Vi conjugates the ability to induce similar anti-vi IgG response to the corresponding full-length Vi conjugates after two 106

115 injections was confirmed (Figure 3.12 A). All the full-length conjugates induced an anti-carrier response higher than corresponding fragmented conjugates even though the amount of protein injected was lower (Figure 3.12 B), expect for CRM197 for which the response was also very low with full-length Vi. Figure A) Full-length Vi conjugates induce similar anti-vi IgG responses independent of the carrier protein used in the conjugation. Differently from fvi-crm197 and fvi-dt, fvi-tt does not give a booster response after the second injection. Wild type mice are immunized at day 0 and 35 at 1 g saccharide/dose. Bars represent the Geometric Mean ELISA Units of the group in log scale, individual animals are represented by the scatter plots. Statistically significant differences are represented by asterisks (* and ** are correspondent to p<0.05 and <0.005 respectively) and statistical analysis is conducted as reported in paragraph

116 Figure B) The full-length conjugates induce anti-carrier IgG responses higher than the corresponding fragmented conjugates, except for CRM 197 conjugates for which the response is low with both full-length and fragmented Vi. Wild type mice are immunized at day 0 and 35 at 1 g saccharide/dose. Bars represent the Geometric Mean ELISA Units of the group in log scale, individual animals are represented by the scatter plots. Statistically significant differences are represented by asterisks (* and ** are correspondent to p<0.05 and <0.005 respectively) and statistical analysis is conducted as reported in paragraph

117 3.3.5 Impact of conjugation chemistry Different conjugation chemistries were tested to link fvi NAMW 43 kda to CRM197 as carrier protein (Scheme 3.2 A-D, Table 3.10). Scheme 3.2. A set of fragmented Vi conjugates is obtained by differing conjugation synthesis (Table 3.10). Fragmented Vi activation occurs randomly along the saccharide backbone (box A, B, C and D) or selectively in the terminal end of the saccharide chain (box E and F). Different random chemistry approaches are applied: varying the length of the linker (box A) or the activation molecule (box B) or derivatizing the saccharide instead of the protein (box C). Furthermore amine (box D and E) or carboxylic (box A and F) groups on CRM 197 protein are targeted and conjugated to fvi 109

118 Table 3.10 Characteristic of the fragmented Vi conjugates obtained using different conjugation strategies Scheme fvi NAMW Amino acids activated Selective Random terminal A B C D F E Asp Glu 43 kda 8.6 kda Lys Asp Glu Lys Component derivatized (activation degree) CRM197 (7 linkers) CRM197 (6 linkers) CRM197 (6 linkers) CRM197 (9 linkers) CRM197 (6 linkers) fvi (22% RU) CRM197, fvi (12 linkers, 22.5% RU) fvi (1 linker/chain) CRM197, fvi (9 linkers, 1 linker/chaim) Conjugate (n CH2 in the linker) fvi-crmodh (x=0) fvi-crmsdh (x=2) fvi-crmadh (x=4) fvi-crmpdh (x=5) fvi(dmt- MM)CRMADH fvi(adh)-crm197 fviadhn3-crmalkyne fvis(adh)-crm197 fvissh-crmsbap fvi to CRM 197 w/w ratio fvi to CRM 197 molar ratio % free Vi na na na na na na na < <20 <20 <20 <20 NAMW: average molecular weight, na: not applicable, w/w: weight to weight, no free CRM197 detected in all the conjugates 110

119 In particular, conjugates differing for the length of the linker introduced on CRM197 (NH2NHCO(CH2)xCONHNH2, x = 0, 2, 4, 5) before conjugation to fvi were prepared (Scheme 3.2 A). By using same reaction conditions for CRM197 derivatization with ADH (240), an average of 9 linkers was introduced per molecule of protein with PDH (x=5) and an average of 6-7 linkers with all the other linkers. There was no correlation between the number of linkers per protein or their length and the saccharide to protein ratio of corresponding conjugates (Table 3.10). An additional conjugate was synthesized using DMT-MM (261, 262) to activate COOH groups on fvi instead of EDAC/NHS before linkage to CRM-ADH (Scheme 3.2 B). This reagent has been tested as an alternative coupling reagent to carbodiimide, allowing the activation step to be reduced to few minutes instead of one hour with EDAC/NHS (261, 262). The resulting conjugate has the same structure of fvi-crmadh and was also characterized by a similar fvi to protein ratio (Table 3.10). Another conjugate was synthesized by derivatizing fvi instead of CRM197 with ADH before performing conjugation (Scheme 3.2 C). Derivatization of fvi must be controlled because elevated numbers of ADH introduced on the saccharide chain can give cross-linking among fvi(adh) molecules. By working with a fvi concentration of 50 mg/ml, NHS concentration of 0.3 M and ADH to fvi repeating units molar ratio of 1.5, 50% of fvi repeating units were activated (Table 3.11), but HPLC-SEC analysis showed crosslinking of the saccharide chains (Figure 3.13 A). By increasing the ADH to fvi repeating units molar ratio from 1.5 to 5 or by reducing fvi and NHS concentrations to 15 mg/ml and 0.1 M respectively, crosslinking was reduced (Figure 3.13 B) or completely avoided (Figure 3.13 C). The latter condition was 111

120 preferred resulting in 20% fvi repeating units activated versus 40% (thus less impact on saccharide structure and epitopes) and no fvi(adh) cross-linked species. Table Different conditions tested for random fvi activation with ADH linker EDAC/RU molar ratio [NHS] M [fvi] mg/ml ADH/RU molar ratio % activated fvi RU RU: repeating units Figure Modulation of [fvi] and ADH/RU molar ratio determines the degree of fvi(adh) crosslinking. HPLC-SEC profiles of fvi(adh) batches are obtained with these sets of parameters: 0.3 M NHS, [fvi]=50 mg/ml, ADH/RU molar ratio 1.5 (profile A), 0.3 M NHS, [fvi]=50 mg/ml, ADH/RU molar ratio 5 (profile B), 0.1 M NHS, [fvi]=15 mg/ml, ADH/RU molar ratio 1.5 (profile C). fvi(adh) batches are eluted on TSK gel 3000 PWxl column with the mobile phase (0.1 M NaCl, 0.1 M NaH 2 PO 4, 5% ACN, ph 7.2) flushing in isocratic condition at 0.5 ml/min 112

121 Two fvi to CRM197 w/w ratios were tested for conjugation (1:1 and 1:2), and HPLC-SEC profiles of the conjugation mixtures revealed conjugate formation with no residual free saccharide only when using a 1:2 ratio (Figure 3.14). 113

122 The conjugate obtained (fvi(adh)-crm197) was characterized by a higher fvi to protein ratio than fvi-crmadh (Table 3.10). Even though there was no statistically significant difference among the anti-vi IgG response induced in WT mince by the different conjugates, conjugates obtained using the longer linker between the saccharide and protein moiety (ADH or PDH) gave a more homogeneous response (Figure 3.15 A). fvi(adh)-crm197 conjugate was also characterized by a higher fvi to protein w/w ratio than all the other conjugates (Table 3.10). The fact that it did not elicit a significantly different immune response from the other conjugates also confirms that the saccharide to protein ratio is not a critical parameter for the immunogenicity of either fulllength Vi (Figure 3.7) or fvi-crm197 conjugate vaccines (Figure 3.9 and 3.15 A). Two further conjugates were prepared by selective terminal linkage of fvi chains to CRM197, one targeting lysine groups (fvissh-crmsbap) and the other one carboxylic residues (fvis(adh)-crm197) on CRM197 (Scheme 3.2 E and F; Table 3.10). The selective conjugates were prepared using fvi with NAMW 8.6 kda, as selective chemistries did not work with longer Vi chains. When tested in WT mice, the different conjugation chemistries had no impact on anti-vi IgG responses (Figure 3.15 B). Two weeks after the second injection, the anti-crm197 response induced by fvis(adh)-crm197 and fvissh-crmsbap was significantly higher than the response induced by the random conjugate fvi-crmadh (p= and 0.01 respectively) (data not shown). 114

123 Figure No major impact of conjugation chemistry approach is revealed on the anti-vi antibody response. Anti-Vi IgG antibody responses elicited in WT mice by fragmented Vi-CRM 197 conjugates obtained through selective and random chemistries and targeting different amino acids on CRM 197 protein (Table 3.10). Wild type mice are immunized at days 0 and 28 at 1 g Vi/dose. Bars represent the Geometric Mean ELISA Units of the group in log scale, individual animals are represented by the scatter plots 115

124 3.4 Discussion It is well documented that several parameters can affect the immunogenicity of glycoconjugate vaccines (66). Here a systemic approach was used to look at the impact of the main variables and their combination on the immunogenicity of a glycoconjugate vaccine against S. Typhi. Vi-CRM197 conjugate vaccine has been already tested in Phase 1 and 2 clinical trials in Europe (242) and endemic countries (241). Superior immunogenicity of Vi-CRM197 compared with unconjugated Vi PS was revealed and the vaccine induced an anti-vi specific antibody response in infants, but second injection of the conjugate did not generate a booster response and anti-vi antibody persistence was similar to that induced by Vi PS (241). Similarly absence of antibody enhancement response was revealed in children receiving secondary Vi-TT (Pedatyph TM ) vaccination (232). The hypothesis is that, despite being conjugated to a carrier protein, the large size of the fulllength Vi PS could permit crosslinking of B-cell receptors thereby inducing a TI response. On the other hand, conjugation could have an impact on the apparent size of Vi PS, making crosslinking of BCRs much more efficient. For this reason we started by looking at the impact of Vi chain length and conjugate size/crosslinking on the immunogenicity of Vi-CRM197. As observed in humans (242), in WT mice full-length Vi-CRM197 induced a much larger anti- Vi IgG response that the unconjugated PS, and no booster response following a second immunization. In T-cell deficient mice, the full-length Vi conjugate gave a significant response suggesting that Vi maintains its ability to work as a TI antigen even if conjugated to a carrier protein. 116

125 By producing fvi with differing average chain lengths, a critical size (approximately NAMW 82 kda) was identified below which both unconjugated and conjugated fvi can no longer act as TI antigens. Unlike the full-length Vi conjugate vaccine, the fragmented Vi conjugates could boost specific anti-vi IgG antibody levels following a second injection in WT mice. This seems to support the hypothesis that when the ability of the conjugated Vi to act as a TI antigen is lost, the detrimental effects of a TI response on memory and subsequent boosting are also negated. This is consistent with other studies of boosting mice with TI antigens, resulting in the apoptosis of memory B-cells and the ablation of GCs (263, 264). Concerning Vi-DT conjugate, An et al. showed that bigger and more crosslinked the conjugate, the higher the anti-vi response induced after one injection (236). Also Wessels at al. claimed larger and more crosslinked GBS type III-TT conjugates were more immunogenic (102). Different results were obtained in this study; in fact no differences were found between Vi-CRM197 conjugates differing for size. The conjugate size also did not impact the immunogenicity of fragmented Vi conjugates. Herein data suggest that it is the length of Vi chains that have the major impact on the specific anti-ps immune response induced, and not the apparent size deriving from crosslinking after conjugation to the carrier protein. Based on the results obtained with Vi of different chain length, full-length Vi and fvi of NAMW 43 kda were selected for investigating the impact of the carrier protein on the immunogenicity in WT mice. In all the studies performed, Vi and fragmented Vi conjugates were compared at 1 g PS/dose and with no adjuvant so to highlight the impact of the conjugation parameters on immunogenicity results. 117

126 Theoretically, any protein with T helper cell epitopes can be used as a carrier protein. Few proteins have been used as carriers in licensed glycoconjugate vaccines to date, with DT, TT and CRM197 used for Hib, pneumococcal and meningococcal conjugate vaccines (33, 66). TT and DT have traditionally been used because of safety data collected with tetanus and diphtheria vaccination. CRM197 does not require chemical detoxification, facilitating production and resulting in homogeneous preparations (111). There has been particular interest in the influence of carrier protein on immunogenicity of conjugate vaccines. In the context of Vi conjugate vaccines, several proteins, such as repa, DT, TT, Cholera toxin subunit B protein, the B subunit of the heat-labile toxin of E. coli, recombinant OMP of Klebsiella pneumoniae (rp40) and iron-regulated OMP of S. Typhi have been tested as carriers (198, 240, 245, 247, 249, ). Two studies have investigated the impact of carrier protein on the immunogenicity of Vi conjugates in mice. Both studies found no effect of the carrier protein (CRM197, TT, DT or repa) on the immunogenicity of full-length Vi conjugate vaccines obtained by EDAC random chemistry with ADH as linker (240, 247). In this study DT, TT and CRM197 were used as carrier proteins. Irrespective of the carrier protein used, similar anti-vi IgG response was elicited by full-length Vi conjugates. For all of them there was no booster response after reinjection. Similarly fragmented Vi conjugates elicited anti-vi IgG response after the second injection independent of the carrier protein, and similar to anti-vi IgG response elicited by the corresponding full-length conjugates. However, differently from fvi-crm197 and fvi-dt, fvi-tt did not give a booster response after the second injection, and similarly to full-length conjugates the response peaked already after the first injection. The different behaviour of TT compared to CRM197 and DT could be related to the bigger size of the protein. For pneumococcal 3 PS it was shown the ability of a TT 118

127 conjugate to induce higher IgG response compared to an analogous DT conjugate after primary vaccination in mice. No effect of a second vaccination was evaluated (117). Higher protection rate and SBA titres were detected in toddler immunized with meningococcal group C conjugate composed by TT, instead of CRM197, as carrier protein (269). It would be interesting to perform additional studies by comparing CRM197 and TT as carrier and by testing fvi of shorter chain length (< 43 kda) with TT protein. The impact of conjugate size on the immunogenicity of Vi-TT could be different to what is seen for Vi-CRM197. All the full-length Vi conjugates induced an anti-carrier response higher than corresponding fragmented Vi conjugates, except CRM197 for which the response was very low also with fulllength Vi. We expected the opposite considering the higher amount of protein administered with fragmented conjugates and that longer Vi chains should mask more the protein (249). However, in a study comparing pneumococcal PS and oligosaccharide conjugates with TT, an inverse correlation was found between protein-specific IgG response and the protein dose administered for polysaccharides but not for oligosaccharides (101). Furthermore in a study comparing long and short PRP PS conjugated to TT carrier protein, it was found that anti-tt antibody response was higher for the longer saccharide conjugate, in line with our results. It was suggested that during the conjugation small saccharides could have a major impact on protein epitopes comparing with longer saccharides (270). In a recent analysis of structureantibody recognition relationships in nine licensed polysaccharide-tt conjugate vaccines, it was shown that recognition of the carrier epitopes was not necessarily hampered by the size of the conjugate or the saccharide loading (271). An alternative possible explanation is that longer Vi chains may better stabilize the protein with production of higher anti-carrier protein antibody levels. 119

128 The use of shorter Vi chains could have advantages in terms of immunogenicity and has clear advantages in terms of manufacturability: the conjugation reaction can be performed with a higher degree of control and better consistency due to the higher solubility of the shorter saccharide chains. Yields of conjugate (expressed in terms of the fvi in the conjugation mixture) are higher and the product is easier to sterile filter, purify (particularly from unreacted PS) and characterize. Herein HPLC-SEC analysis was used for determining saccharide NAMW through calibration with molecular weight dextran standards. Other methods for MW analysis, such as analytical ultracentrifugation or size exclusion chromatography coupled to multi-angle laser light scattering, could be useful for characterizing both fragmented Vi and the corresponding Vi conjugates ( ). The impact of conjugation chemistry on shorter Vi chains and with CRM197 carrier protein was another parameter investigated. Random and selective conjugation chemistries were applied to conjugation, by targeting different amino acids on CRM197 (91) and by using linkers of different length, and it was verified that there was no major impact on anti-vi IgG response in mice. The only study investigating the impact of conjugation chemistry on the immunogenicity of full- length Vi conjugate vaccines showed greater immunogenicity of VirEPA conjugate in humans when EDAC/ADH chemistry was used instead of cystamine/spdp (112). Selective chemistries avoid chemical modification in multiple points of the saccharide backbone and thus produce better-defined structures in contrast with random approaches that result in high MW, crosslinked and rather undefined and heterogeneous structures. Anyway considering the difficulty we found to extend terminal selective chemistries to Vi chains longer than 8.6 kda, and the lower yields that characterized these conjugates (20-30% fvi 120

129 recovery compared to 60-70% with random approaches), random chemistries should be preferred for manufacturing. The initial plan of this study was to compare glycoconjugates differing for one only parameter at a time, keeping all the others constant. However, the saccharide to protein ratio was influenced by the saccharide chain length, protein and conjugation chemistry used and so was difficult to control. Both for full-length and shorter Vi chains (NAMW 43 kda) no major impact of saccharide to protein ratio on the immunogenicity in WT mice was observed. Few studies have looked at the impact of Vi to protein ratio on the immunogenicity of Vi conjugate vaccines (236, 248). Rondini et al. showed that for Vi-CRM197, a different Vi to protein ratio did not impact greatly on the immune response (248). In fact, a Vi to CRM197 ratio of 10.1 was suboptimal compared to conjugates with a Vi to CRM197 ratio of 0.9 and 2.1 only at a low dose of g Vi. An et al. (236) showed in contrast that the amount of DT conjugated to Vi have an impact on the magnitude of the response to the PS. The more DT was bound (range of Vi/DT w/w tested was ), the greater the anti-vi response after the first two injections. Anyway the conjugates tested were different not only for the Vi to protein ratio, but also for crosslinking degree. Thus it is difficult to identify the effective parameter affecting the immune response. In conclusion the effects of different parameters on the immunogenicity of Vi conjugate vaccines are interconnected, as shown by the different behaviour of fragmented Vi when linked to CRM197 or TT. Carrier protein and saccharide chain length play a major role on the immunogenicity of Vi conjugate vaccines. 121

130 The kind of approach used here can be extended to other conjugate vaccines in order to identify critical parameters and select their optimal combination to obtain improved glycoconjugate vaccines both in terms of production and efficacy. 122

131 4 Comparison of different carrier proteins for Vi conjugate vaccines and their priming effect 4.1 Introduction Carrier protein is one of the main parameters can affect the immune response elicited by glycoconjugate vaccines (110, 275, 276). Three proteins, DT, TT and CRM197, have been mainly used to date as carriers for licensed glycoconjugate vaccines (110). Diphtheria toxin is a single polypeptide chain synthesized and released by Corynebacterium diphteriae (110). If subjected to a mild treatment with trypsin, diphtheria toxin assumes the nicked form composed by fragment A, containing the catalytic domain, and fragment B, which carries transmembrane and receptor-binding domains (111). Inactivation of catalytic activity occurs through a detoxification process with formaldehyde, with formation of stable methylene bridges between lysines and tyrosines or histidines (111). The resulting protein has been identified as DT (277). Similarly TT, obtained from Clostridium tetani, derives from the respective toxin after chemical detoxification with formaldehyde (278, 279). The detoxification step, that is a crucial step for toxoids manufacturing, generates crosslinking in the protein structure resulting in heterogeneous preparations difficult to produce with consistency (111, 278, 279). In the 70s CRM197, a genetically detoxified mutant of diphtheria toxin, was produced (280). This protein does not require detoxification treatment because the single amino acid substitution from Gly to Glu in position 52 of the mutant sequence determines a fully loss of catalytic activity of fragment A (111). Even though the fragment B is not directly altered, the introduction of a negative amino acid in fragment A, perturbs fragment B structure (111) 123

132 resulting in a more open conformation of CRM197 compared to DT protein (281). Unlike DT or TT, CRM197 is a homogeneous and well-defined protein easier to consistency produce and characterize (111). Glycoconjugate vaccines are often co-administered or given in combination with other vaccine antigens for preventing multiple diseases and when possible without overloading the number of injections (275, 276, ). Design of appropriate immunization schedules is critical to avoid undesired interferences on immune responses (124, , 287).The main mechanisms at the base of these interferences have been identified ( ). Suppression of the immune response to a specific-antigen has been associated to two main mechanisms: Carrier-Induced Epitopic Suppression (CIES), when different saccharides are independently coupled to a common carrier protein, and bystander effect, when un-related polysaccharides are coupled to different proteins (282, 283). Several factors can be associated to CIES such as: the presence of pre-existing anti-carrier antibodies, the monopolization of T-cell effector cells by specific-carrier memory B cells at expenses of specific-saccharide B cells or the inhibitory effect of T regulatory cells (282, 283, 288). On the other hand bystander effect can be explained by local competition of common resources such as activation signals, chemokines, follicular DCs and T helper cells (283). Alternative explanations include the imbalance among Th1/Th2/Th0 responses and/or action of T regulatory cells (283). Also a positive effect on immune response to antigens conjugated to the same protein has been reported through carrier-driven T-helper cell stimulation (283). For example co-administration or vaccine combination of MenC-TT and Hib-TT conjugates enhance anti-prp immune response in infants ( ). The spread of DT, TT and CRM197 as carrier proteins for glycoconjugate vaccines has raised concerns about possible interference. The terminology 124

133 carrier priming effect identifies the impact of previous protein administration on the immune response induced by glycoconjugate vaccines (124). Since saccharide loading on the protein may suppress carrier-specific antibody response, prior administration of the unconjugated protein permits to emphasize possible CIES effect on following glycoconjugate vaccinations (288, 292). Many clinical studies have examined the effect of carrier pre-existing immunity on the immunogenicity of glycoconjugate vaccines, but with contradictory results ( ). Administration of Vi conjugate vaccine alongside EPI schedule or in combination with measles immunization has been proposed respectively for young and older infants in endemic countries (188). Vi-rEPA and Vi-CRM197 have shown to be compatible with routine infant immunizations, when tested in endemic countries (241, 297). A Phase IV Open labelled study is ongoing to evaluate the non-interference of Vi-TT (Typbar-TCV) conjugate vaccine in immune response in children receiving measles vaccine concominantly (298). No specific studies have been reported on the effect of carrier priming on the immunogenicity elicited by Vi conjugate vaccines. In the present chapter, carrier priming effect induced by DT, TT and CRM197 on the immune response induced by homologous Vi conjugates and by Vi-CRM197 conjugate is compared. Furthermore formylated CRM197 has been compared with DT and CRM197 in terms of physical properties and carrier priming effect induced, in order to see if differences observed between DT and CRM197 could be mainly implicated to the detoxification process to which DT is subjected. 125

134 4.2 Summary Recently it was proposed the introduction of Vi conjugate vaccines within infant scheme immunizations. The use of DT, TT and CRM197 as carrier proteins for an increasing number of glycoconjugate vaccines has raised concerns about possible interference. The term carrier priming effect is used to describe the impact of previous carrier protein administration on the immune response induced by glycoconjugate vaccines. Herein carrier priming effect induced by DT, TT and CRM197 on the immune response elicited by homologous Vi conjugates and by Vi-CRM197 was investigated. CRM197 is the only protein able to induce a positive carrier priming effect on the homologous Vi conjugate and CRM197 glycoconjugate results not influenced by DT or TT priming. This means that CRM197 pre-existing immunity could have a positive effect on the immune response induced by a Vi-CRM197 glycoconjugate vaccine and previous immunizations with DT and TT proteins would not have negative impact on the immune response induced by Vi-CRM197. Furthermore in the present chapter it was investigated how the nature of the carrier protein could affect the physical properties of Vi polysaccharide conjugate vaccines. In particular the physical differences observed between DT and CRM197 glycoconjugates were found to be related to the formylation process used to detoxify diphtheria toxin. 126

135 4.3 Results Vi-CRM 197, Vi-DT and Vi-TT conjugates differing for protein derivatization degree with ADH linker Vi-CRM197, Vi-DT and Vi-TT conjugates were synthesized with the aim to look at the impact of different carrier proteins on the immune response (as reported in Chapter 3, section 3.3.4). The conjugates, obtained by using same chemistry and same conjugation conditions, showed different MW distribution (Figure 4.1), with Vi-TT and Vi-DT having an additional population at lower MW compared to Vi-CRM197. In particular, Vi-DT was characterized by a higher Vi to protein w/w ratio than Vi-CRM197, even if the two proteins differ for only one amino acid in position 52 (Gly and Glu respectively for DT and CRM197). This could be the result of higher average number of ADH linkers introduced on DT compared with CRM197, 12 and 6 respectively (Table 4.1). The relation between the degree of protein derivatization with ADH and the Vi to protein ratio of the corresponding conjugate was investigated. CRM197 derivatization process with ADH linker was modified by increasing the ratio of EDAC/CRM197 (w/w) in the reaction mixture from 0.15 to 0.4 and 1. The average number of linkers introduced per molecule of CRM197 increased from 6 to 12 and 35 respectively (Figure 4.2). Conjugation with Vi (EDAC/NHS chemistry working with a Vi to protein w/w ratio of 1) worked for the protein with 12 linkers, while precipitation occurred when too many linkers were on the protein favouring too much crosslinking. Similarly to what observed for DT (average number of 12 linkers introduced) and TT (average number of 24 linkers introduced), when an average number of 12 ADH linkers was introduced on CRM197, the percentage of unreacted protein remained higher (approximately 50%) than when conjugation was performed with a less derivatized protein (25-30%). The resulting conjugate Vi-CRM197(12ADH), purified by TFF 300k, was 127

136 characterized by a higher w/w Vi to protein ratio compared to the conjugate Vi-CRM197(6ADH), but still lower than Vi-DT conjugate (Table 4.1). HPLC-SEC profile was similar to that of Vi- CRM197(6ADH) (Figure 4.1), but shifted at slightly lower MW. Figure 4.1. Only CRM 197 conjugates are characterized by a unique molecular weight population. HPLC-SEC profiles of purified conjugates: Vi-CRM 197 (6ADH) (red line), Vi-CRM 197 (12ADH) (blue line), Vi-CRM f, Vi-DT and Vi-TT. Conjugates are eluted on TSK gel PW columns with mobile phase (0.1M NaCl, 0.1M NaH 2 PO 4, 5% ACN, ph 7.2) flushing at 0.5 ml/min 128

137 Figure 4.2. Different average number of ADH linkers are introduced on CRM 197 by modulating derivatization reaction conditions. Different levels of derivatization are obtained by increasing the amount of EDAC in the reaction mixture: A) EDAC/CRM 197 w/w 0.15, B) EDAC/CRM 197 w/w 0.4 and C) EDAC/CRM 197 w/w 1. The average numbers of ADH linkers introduced on the protein is determined from the shift of the molecular weight Table 4.1. Main characteristics of Vi conjugates differing for carrier protein Conjugate Vi-CRM 197 (6ADH) Average n of ADH linkers introduced per protein Total Vi to protein ratio (w/w) % free Vi % free CRM nd Vi-CRM 197 (12ADH) nd Vi-DT nd Vi-TT nd Vi-CRM f <20 nd nd: not detectable 129

138 Conjugates produced (Table 4.1) were compared in WT mice (Figure 4.3). Fourteen days after the first injection, Vi-TT induced an anti-vi IgG antibody response higher to that induced by Vi-CRM197(12 ADH) (p=0.0003, by comparing all the groups with Kruskal-Wallis test). No differences were revealed among the different groups at day 49. Exclusively the conjugate Vi- CRM197(12 ADH) induced an anamnestic response after injection (p= ) (Figure 4.3 A). Figure 4.3 B reports anti-carrier protein IgG response induced by the different conjugates. Differently from anti-vi IgG response, Vi-CRM197(6 ADH), Vi-DT and Vi-TT showed an increased anti-carrier protein response after the second injection (p= , p= and p= respectively). Vi-CRM197(6 ADH) induced higher anti-crm197 IgG response than Vi- CRM197(12 ADH) at day 49 (p= ). 130

139 Figure 4.3. For full-length Vi conjugates, the carrier protein does not have major impact on the anti-vi IgG response elicited. Wild type mice are immunized at days 0 and 35 with 1 g Vi/dose. Bars represent the Geometric Mean ELISA Units of the group in log scale, individual animals are represented by the scatter plots. Statistically significant differences are represented by asterisks (*, ** and *** are correspondent to p<0.05, <0.005 and < respectively) and statistical analysis is conducted as reported in paragraph

140 4.3.2 The formylation process on CRM 197 determines similarities with DT physical properties Differently to CRM197, DT is subjected to detoxification with formaldehyde (277). In order to investigate if this formylation process was responsible for the different behaviour of DT and CRM197 in the process of derivatization with ADH and following conjugation to Vi, CRM197 was subjected to the same protocol applied for DT detoxification (277). SDS-PAGE, HPLC- SEC and MALDI-MS analysis (Figure 4.4) showed a more similar profile of formylated CRM197 (CRMf) to DT than to CRM

141 133

142 Derivatization of CRMf with ADH, following the same protocol of derivatization applied to DT and CRM197 (240), resulted in an average number of 15 ADH linkers per protein molecule (Figure 4.5), higher than the average number introduced on CRM197 (6 linkers), and more similar to the number of linkers introduced on DT (12 linkers) (Figure 3.10, section 3.3.4). This confirmed a more similar reactivity of CRMf to DT than to CRM197. Figure 4.5. The degree of derivatization of CRM f results higher to the average number of linkers introduced on CRM 197. MALDI-TOF spectra of CRM f (black line) and corresponding ADH derivative (red line) synthesized using the same protocol applied to DT and CRM 197 (240). The average numbers of ADH linkers introduced on the protein is determined from the shift of the molecular weight 134

143 4.3.3 Synthesis of Vi conjugate with CRM f The Vi-CRMf conjugate showed a HPLC-SEC profile very similar to that of Vi-DT, with two main populations at different NAMW compared to mainly one population at higher MW for Vi- CRM197(12 ADH) and Vi-CRM197(6 ADH) (Figure 4.1). As for Vi-DT (see Chapter 3, section 3.3.4), purification by TFF 300k was not possible as the conjugate passed through the membrane and Vi-CRMf was purified by TFF 100k. Efficiency of conjugation was lower for CRMf (with 15 ADH linkers introduced; about 50% of free protein in the conjugation mixture) than for CRM197(6 ADH) (about 25-30% of free protein in the conjugation mixture) and more similar to CRM197(12 ADH) and to DT (about 50% and 70% of free protein in the conjugation mixture respectively). The purified conjugate was characterized by a similar Vi to protein ratio compared to Vi- CRM197(12 ADH) (Table 4.1) Immune response induced by Vi conjugates differing for carrier protein and carrier priming effect induced by CRM 197, CRM f, DT and TT All the conjugates obtained with the same protocols for protein derivatization and for conjugation were tested in WT mice (Table 4.2). Also, the priming effect of the four different proteins was evaluated on the immune response elicited by the homologous conjugates and by Vi-CRM197. Wild type mice were primed at day 0 with saline or 5 g of protein and then immunized with the Vi conjugates at day 28 and 56 (Table 4.2). Sera were collected at day 0, 14, 28, 42, 56 and

144 Table 4.2. Immunization scheme for investigating the impact of carrier protein on the immune response of Vi conjugates and the carrier priming effect Group # Antigen for I immunization Protein dose (I imm) Antigen for II and III immunizations 1 Saline / Vi-CRM197(6 ADH) 2 CRM197 5 g Vi-CRM197(6 ADH) 3 Saline / Vi-DT 4 DT 5 g Vi-DT 5 Saline / Vi-CRMf 6 CRMf 5 g Vi-CRMf 7 Saline / Vi-TT 8 TT 5 g Vi-TT 9 DT 5 g Vi-CRM197(6 ADH) 10 CRMf 5 g Vi-CRM197(6 ADH) 11 TT 5 g Vi-CRM197(6 ADH) Imm= immunization Vi dose (II and III imm) 1 g By comparing all the conjugates that did not receive priming with the protein, at day 42, Vi- TT induced a higher response than both Vi-DT (p = 0.022) and Vi-CRMf (p = ). Anti- Vi IgG response induced was instead similar for all the conjugates at day 70. When the groups were primed with the protein, Vi-CRM197(6 ADH) induced higher anti-vi IgG response than Vi-CRMf at day 42 (p = ), while no differences were observed among all the groups at day 70. Priming with the protein before conjugate injection did not have any impact on the immune response induced by Vi-DT, Vi-CRMf and Vi-TT conjugates. Conversely, priming with CRM197 had a positive impact on the anti-vi IgG response induced by the homologous conjugate, both at day 42 (p = ) and at day 70 (p = ). Vi-CRMf was the only conjugate able to boost the response after re-injection (p= comparing the response at day 42 and at day 70), but only when primed with saline. In this study, for Vi-CRM197(6 136

145 ADH) and Vi-TT a decline of the response after the second conjugate injection was observed, with or without carrier priming (Figure 4.6 A). Priming with the protein had a positive effect on the anti-carrier protein IgG response induced by all the conjugates at day 42, while there was an increase of the response at day 70 only for Vi-TT (data not shown). The priming effect of different carrier proteins was evaluated on the immune response induced by Vi-CRM197 conjugate. The only protein inducing an increase of the anti-vi IgG response was confirmed to be CRM197 (p= and p= after first and second conjugate immunizations respectively), meanwhile no effect was associated to the other carrier proteins (Figure 4.6 B). 137

146 138

147 139

148 4.4 Discussion To date, CRM197, DT and TT are the main carrier proteins used in commercial glycoconjugate vaccines (110). All the three proteins have been tested for Vi conjugates (230, 231, 236, 241). Two studies have investigated the impact of carrier protein on the immune response elicited by Vi conjugates in mice, showing no impact of this variable. In particular DT and repa or CRM197, repa and TT conjugates were compared (240, 247). This is the first study showing a direct comparison between DT and CRM197. CRM197 and DT differ for one only amino acid in their sequence, but they showed a different behaviour already in the derivatization step with ADH linker before conjugation. In fact, by using the same protocol, DT introduced an average number of 12 linkers against 6 ADH for CRM197. This could be the result of different accessibility and exposure of carboxylic reactive groups on the two proteins, confirming their different conformation (111), as also verified by SDS-PAGE, MALDI-MS and HPLC-SEC analysis. Also resulting Vi conjugates were characterized by different MW distribution and Vi to protein w/w ratio. By subjecting CRM197 to the formylation process needed for DT detoxification, we found that SDS-PAGE, MALDI- MS and HPLC-SEC profiles of the protein became much similar to that of DT alongside its reactivity with ADH and Vi conjugate structural characteristics. By applying the conjugation protocol optimized for CRM197 (6 ADH) to DT, CRM197 (12 ADH) and CRMf, presence of a higher average number of ADH linkers per protein resulted in an increased amount of unconjugated protein in the conjugation mixture. It seems that conjugation of first Vi chains favours further linkage on the same protein, compared to free carrier molecules. Resulting conjugates are characterized by higher Vi to protein w/w ratio and less crosslinked structures. 140

149 Differently from Vi-CRM197(6ADH), Vi-CRM197(12ADH) was able to induce an anti-vi IgG anamnestic response after reinjection, possibly as a consequence of different Vi chains conformation. Also the anti-crm197 IgG response induced was lower. This could be due to the lower dose of protein injected, but also the result of an increased masking of the protein from Vi chains or the result of higher structural modification of the protein itself. Independently from the structural differences among all the conjugates produced, we confirmed that, in the case of full-length Vi, the protein does not have major impact on the anti-vi IgG response elicited. The introduction of Vi conjugate vaccine in infants immunization has been suggested (188). It becomes important to design appropriate vaccination schedules, considering possible interference of concomitant or previous vaccinations. For this reason, we investigated the possible priming effect of the different carrier proteins on the immune response induced by Vi conjugates. At the doses tested, CRM197 was the only protein to positively prime the anti-vi response elicited by its conjugate, while DT and TT did not have an impact on the immune response elicited by their corresponding conjugates or Vi-CRM197. CRMf showed a more similar behaviour to DT than to CRM197 also in terms of priming effect. Recent studies in mice have shown that CRM197 has a clear propensity to prime the antisaccharide immune response elicited by the homologous serogroup A meningococcal conjugate vaccine, as opposed to DT and TT (116, 299). Furthermore, DT priming can suppress the response to MenA-DT conjugate, but not the response elicited by MenA-CRM197 (299). 141

150 Several factors can be associated to CIES such as the presence of pre-existing anti-carrier antibodies that could hamper the recognition of glycoconjugate by the saccharide-specific B cells (282, 283, 288). Other perspective involves the monopolization of T-cell effector cells by specific-carrier memory B cells elicited by prior vaccinations at expenses of specificsaccharide B cells or, otherwise, presence of T regulatory cells having a negative effect on the immune response (283). In the case of DT, more stable structure induced by chemical detoxification could promote the retention of carrier-specific B cell epitopes after conjugation, while stronger protein conformation changes have been suggested after CRM197 conjugation. This can affect binding features and avidity of pre-existing anti-carrier antibodies (299). It would be interesting to verify if the priming effect of the proteins remains the same when conjugated to unrelated polysaccharides. Also the conjugation chemistry used and the glycosylation level of the resulting conjugates could have an impact (116). Furthermore it has been shown that DT is poorly presented to T cells and that its priming stimulates anti-carrier more than antisaccharide B cell response, meanwhile CRM197 priming induce both carrier- and saccharidespecific B cell response (300). Results reported here may have implications for human vaccination: CRM197 pre-existing immunity could have a positive effect and previous immunizations with DT and TT proteins would not have negative impact on the immune response induced by Vi-CRM

151 5 Investigation of the early- and long-term immune responses induced in mice by selected Vi conjugates 5.1 Introduction The development of Vi conjugates has focused on their production of humoral responses against Vi antigen expressed by S. Typhi (112, 230, 240, 241). Several studies have been conducted in mice investigating the immune responses induced by Vi conjugates like Vi-rEPA (246, 268), Vi-CRM197 (238), Vi-DT (236) and Vi-PspA (128). Furthermore the efficacy of protection induced by Vi conjugates has been evaluated by challenging mice with S. Typhimurium strains genetically modified to express Vi antigen (248, 266). This chapter reports the results of investigation into the immune responses induced in WT mice by four Vi conjugates. These conjugates differ for Vi saccharide chain length and carrier protein (Table 5.1). In particular Vi of two different saccharide chain lengths were used for conjugation: the full-length Vi (NAMW 165 kda) and the fragmented Vi with NAMW 43 kda. The carrier proteins CRM197 and TT were conjugated to both full-length and fragmented Vi. The conjugates Vi-CRM197 and Vi-TT were synthesized using full-length Vi meanwhile fvi-crm197 and fvi-tt were obtained by conjugation of the carrier protein to the fragmented Vi. The selection of these Vi conjugates was based on previous serological data that have highlighted both saccharide chain length and carrier protein as the key parameters affecting the immune response elicited by Vi conjugates (see Chapter 3). Adjuvants can modulate the efficacy of Vi vaccines (248), but in this chapter investigation was focused only on the role of 143

152 saccharide chain length and carrier protein on immunogenicity, thus the effect of adjuvants like alum was not investigated. All mice studies and experiments were performed in the to investigate both the early- and long-term immune responses elicited by these different Vi conjugates (Table 5.1). Table 5.1. Characterization of full-length and fragmented Vi conjugates tested in wild type mice Conjugate saccharide to protein w/w ratio % free saccharide % free protein Vi-CRM nd Vi-TT nd fvi-crm <20 nd fvi-tt 0.35 <20 nd nd: not detectable 5.2 Summary The immune response induced by few selected Vi conjugate vaccines was studied in the current chapter. Full-length and fragmented Vi conjugates of CRM197 and TT were studied, with the aim to further investigate the role of saccharide chain length and carrier protein. The investigation was focused on the early- and long-term immune responses induced by Vi conjugates in WT mice. The analysis of sera and splenic sections identify a rapid anti-vi immune response developed in just one week after single full-length Vi conjugate immunization. In contrast fragmented Vi conjugates do not induce an anti-vi immune response characterized by the same magnitude. 144

153 Preliminary results on long-term response indicate full-length Vi conjugates promote comparable anti-vi immune response to much higher dose of Vi polysaccharide 145 days after immunization, meanwhile lower response is elicited by fragmented Vi conjugates. The protection efficacy of full-length conjugates was tested by a bacterial challenge with a S. Typhimurium strain expressing Vi. Full-length Vi conjugates and Vi polysaccharide can provide comparable protection even though Vi conjugates are administered at lower doses. In conclusion Vi chain length seems to have an impact on both early and long term responses, but further investigation are needed to confirm these results. 5.3 Results Investigation of the early-response induced by Vi conjugates Full-length Vi conjugates are well-tolerated and induce predominantly anti-vi IgG1 antibodies To assess the safety of Vi conjugates, a pilot study was performed by immunizing WT mice i.p. with Vi-CRM197 and Vi-TT (Table 5.1). This preliminary study was conducted using two mice per group and the animal health status was monitored for 14 days. No complications were identified and the animal weights were maintained in the range g. Sera collected 14 days post-immunization were assayed by ELISA to detect anti-vi IgM, IgG and IgA antibody titres and IgG subclasses (Figure 5.1). No anti-vi IgA antibodies were detected (data not shown), but robust IgG antibody titres were detected, with anti-vi IgG1 most readily detected. 145

154 Figure 5.1. Antibody subclasses induced by immunization of full-length conjugates. Wild type mice are immunized with Vi-CRM 197 or Vi-TT intraperitoneally with 2 g Vi. Anti-Vi antibody titres for IgM, IgG and subclasses are assayed by ELISA analysis from sera collected 14 days post-immunization. Data are representative of two mice per group and one experiment 146

155 Full-length Vi conjugates, but not fragmented Vi conjugates, induce Vi-specific immune response seven days post-immunization After verifying Vi conjugates were well-tolerated when injected i.p., a study was performed to compare the early immune response induced in WT mice seven days after immunization with full-length or fragmented conjugates (Table 5.1). Mice were i.p. immunized with Vi conjugates at 2 g dose. Sera and spleens were collected for ELISA, ELISpot and immunohistology assays. No statistically significant differences were revealed for anti-vi IgM and IgG antibody responses induced by Vi-CRM197 or Vi-TT. Similar results were found in a repeat study. Both full-length conjugates induced higher anti-vi IgM and IgG titres compared with the matched fragmented conjugates (Figure 5.2). Anti-carrier protein antibody responses from mice immunized with both full-length and fragmented conjugates were low or not detectable at this stage (Figure 5.3). To assess whether immunization induced specific B cell responses in the spleen, ELISpot was performed to detect splenic Vi-specific antibody secreting cells (ASC). Comparable numbers of anti-vi IgM ASCs were detected among all the immunized groups, and more Vi-specific IgG cells were detected in the mice that received full-length Vi conjugates (Figure 5.4). 147

156 Figure 5.2. Full-length conjugates induce stronger anti-vi immune response compared with fragmented conjugates seven days post-immunization. Anti-Vi antibody titres for IgM and IgG assayed induced in WT mice immunized intraperitoneally with Vi-CRM 197, Vi-TT, fvi-crm 197 or fvi- TT glycoconjugates with 2 g Vi. Symbols in dark red, dark green, light red and light green are correlated to mice respectively immunized with Vi-CRM 197, Vi-TT, fvi-crm 197 and fvi-tt. * represents statistically significant difference between two groups (p<0.05) and statistical analysis is conducted as reported in paragraph 2.3. Data are representative of four mice per group and one experiment 148

157 Figure 5.3. Anti-carrier protein IgG antibody levels induced in WT mice seven days postimmunization. Wild type mice were immunized intraperitoneally with Vi-CRM 197, Vi-TT, fvi-crm 197 or fvi-tt glycoconjugates with 2 g Vi. Data are representative of four mice per group and one experiment 149

158 Figure 5.4. Full-length conjugates induce stronger anti-vi immune response compared to fragmented conjugates. Spot forming units (SFU) of anti-vi IgM and IgG secreting cells isolated from spleens of WT mice immunized intraperitoneally with Vi-CRM 197, Vi-TT, fvi-crm 197 or fvi-tt glycoconjugates with 2 g Vi. Immune responses are assessed seven days post-immunization by ELISA and ELISpot assays. Symbols in dark red, dark green, light red and light green are correlated to mice respectively immunized with Vi-CRM 197, Vi-TT, fvi-crm 197 and fvi-tt. * represents statistically significant difference between two groups (p<0.05) and statistical analysis is conducted as reported in paragraph 2.3. Data are representative of four mice per group and one experiment 150

159 To investigate the response in the spleen in greater depth, sections were cut from frozen spleens and stained for Vi-specific cells using fvi-biotin. These sections were also stained using IgD, IgM, IgG and IgA immunoglobulins. Blue and black spots corresponded respectively to fvi-biotin + cells and to fvi-biotin + cells expressing the immunoglobulin (Figure 5.5). Spots were counted in different areas of each section. Staining for fvi was not observed on sections from mice immunized with fragmented conjugates (data not shown), meanwhile the frequency of fvi-biotin + IgD +, fvi-biotin + IgM +, fvi-biotin + IgG + and fvibiotin + IgA + cells stained on sections of mice demonstrated that both Vi-CRM197 and Vi-TT immunizations generated Vi-specific antibody secreting cells (Figure 5.6). The majority of Vi + cells were either IgM + or IgG + but surprisingly some Vi + cells also expressed IgA immunoglobulins. 151

160 Figure 5.5. Anti-Vi antibody producing cells can be detected on spleen sections seven days after immunization. Representative spleen sections from WT mice sacrificed seven days post-immunization with Vi-TT. Splenic sections were stained for fvi-biotin (blue spots) and IgD, IgM, IgG and IgA (brown spots, from the top to bottom) 152

161 Figure 5.6. Anti-Vi antibody producing cells can be detected on spleen sections seven days after immunization. Relative population of anti-vi cells expressing IgD, IgM, IgG and IgA immunoglobulins on splenic sections from individual mice immunized with Vi-CRM 197 or Vi-TT conjugates 153

162 Antibodies induced by Vi-CRM 197 and Vi-TT cross-react with other Vi conjugates To investigate whether the Vi epitopes are maintained unaltered on the different Vi conjugates, ELISA assays were performed using the different sera generated by coating ELISA plates with the different conjugates. The highest IgG titres were found in sera of Vi-CRM197 and Vi-TT immunized mice and these sera reacted also with fragmented Vi conjugates (Figure 5.7). This means fragmented Vi conjugates contain epitopes that are shared with full-length Vi conjugates. Furthermore it was demonstrated the low anti-vi IgG titres elicited by immunization with fragmented Vi conjugates are not due to a technical issue with the ELISA assay (Figure 5.2) nor to changes in the conformation of Vi in the different conjugates. 154

163 Figure 5.7. Different conjugates maintain shared Vi-specific epitopes. ELISA assays are performed by coating ELISA plates with Vi-CRM 197, Vi-TT, fvi-crm 197 or fvi-tt conjugates and IgG antibodies are detected. Circles in dark red, dark green, light red and light green are related to sera from WT mice immunized with Vi-CRM 197, Vi-TT, fvi-crm 197 and fvi-tt respectively 155

164 5.3.2 Assessment of the longevity of the response to Vi conjugate vaccines Inducing long-lived antibody is a key aim of vaccination programmes, thus to examine the longevity of antibody responses induced by Vi conjugates a study of 145 days was planned involving different immunization schemes. Wild type mice received a single dose of one of the following conjugates: Vi-CRM197, Vi-TT, fvi-crm197 or fvi-tt at day 0, with or without a secondary immunization at day 35. The booster immunization time was fixed as reported in studies investigating the impact of saccharide chain length (Chapter 3, section 3.3.2) and carrier protein (Chapter 3, section 3.3.4) on immunogenicity. Wild type mice immunized at day 0 with Vi PS constituted the control group. All immunizations were i.p. with 2 g Vi at except for purified Vi PS, which was administered at 25 g, as reported in the control group in the clinical trial conducted in South and Souteast Asia (241). All mice were sacrificed at day 145 post-primary immunization (Figure 5.8). Figure 5.8. Immunization scheme for WT mice immunized once at day 0 or receiving a booster dose at day 35. The conjugates used for immunizations were: Vi-CRM 197, Vi-TT, fvi-crm 197 and fvi- TT, all mice received 2 g Vi. Control group was constituted by WT mice receiving one dose of 25 g Vi PS at day 0 156

165 Anti-Vi IgG antibody titres are similar if the carrier protein is CRM 197 or TT Sera at day 145 post-primary immunization were analysed by ELISA to determine antibody titres in WT mice receiving one or two immunizations of full-length (Figure 5.9 A) or fragmented Vi conjugates (Figure 5.9 B). Comparison between mice immunized twice revealed anti-vi IgM antibody titres from Vi-TT immunized mice were significantly higher compared to Vi-CRM197 immunized mice. No statistically significant differences were revealed among anti-vi IgM antibody titres for all other groups (Figure 5.9 A and B). Anti-Vi IgG antibody titres were detected in most mice in all groups, but antibody response resulted variable for groups receiving booster dose of CRM197 conjugates. It cannot be defined if booster may enhance anti-vi IgG antibody in the Vi-CRM197 or fvi-crm197 groups (Figure 5.9 A and B). After a secondary immunization with Vi-CRM197 the anti-vi IgG antibody titres were comparable to those induced by Vi-TT immunization (Figure 5.9 A). 157

166 Figure 5.9. Secondary immunization of Vi-CRM 197 or fvi-crm 197 is related to an enhanced trend of anti-vi IgG antibody persistence compared to single immunization. Sera from WT mice immunized intraperitoneally with Vi-CRM 197, Vi-TT, fvi-crm 197 or fvi-tt with 2 g Vi are collected at day 145 to determine antibody titres between groups receiving one or two immunizations. Graphs report anti-vi IgM and IgG antibody titres comparing antibody persistence induced by A) full-length conjugates and B) fragmented conjugates. Symbols in dark red and dark green are correlated to mice respectively immunized with Vi-CRM 197 and Vi-TT, meanwhile light red and light green are correlated to mice immunized respectively with fvi-crm 197 and fvi-tt. * represents statistically significant difference between two groups (p<0.05) and statistical analysis is conducted as reported in paragraph 2.3. Data are representative of four mice per group and one experiment 158

167 Boosting with full-length Vi conjugates induces higher anti-vi IgG antibody responses than Vi PS Anti-Vi IgG responses were examined in sera, collected by the final bleed at day 145, from mice boosted at day 35 with Vi conjugates and compared to responses in mice receiving a single dose of 25 g Vi PS at day 0. Reduced anti-vi antibody was detected in WT mice immunized with fragmented conjugates compared to other groups (Figure 5.10 A and B). Compared to Vi PS, booster enhanced anti-vi IgG titres when full-length conjugates were administered (Figure 5.10 A and B). 159

168 Figure Secondary immunization of full-length Vi conjugates with 2 g Vi/dose can induce higher anti-vi IgG antibody titres compared with single 25 g Vi PS immunization. Sera from WT mice immunized intraperitoneally with Vi PS, Vi-CRM 197, Vi-TT, fvi-crm 197 or fvi-tt are collected at day 145. ELISA assay is conducted to compare the anti-vi antibody persistence between mice receiving a single dose of 25 g Vi PS and mice receiving conjugate immunization at day 0 and 35 with 2 g Vi/dose. Anti-Vi IgM and IgG antibody persistence induced by Vi PS is compared with those induced by: A) CRM 197 conjugates or B) TT conjugates immunizations. Symbols in blue, dark red and dark green are correlated to mice respectively immunized with Vi PS, Vi-CRM 197 and Vi-TT, light red and light green are correlated to mice immunized respectively by fvi-crm 197 and fvi-tt. * represents statistically significant difference between two groups (p<0.05) and statistical analysis is conducted as reported in paragraph 2.3. Data are representative of four mice per group and one experiment 160

169 Anti-carrier protein antibody responses are detectable after boosting with full-length Vi conjugates Anti-carrier protein antibody responses were assessed from sera, collected by the final bleed at day 145, in WT mice receiving one or two immunizations with full-length or fragmented conjugates. No statistically significant differences were observed for anti-carrier protein IgM antibodies among all groups (Figure 5.11 A and B). Anti-CRM197 IgG antibody titres were variable (Figure 5.11 A). Anti-TT IgG titres induced after immunization with fvi-tt were below level of detection (Figure 5.11 B). As reported in Chapter 3 section 3.3.4, the higher anti-protein response induced in mice immunized with Vi-TT compared with those immunized with fvi-tt was confirmed in Figure 5.11 B. A trend of enhanced anti-protein IgG antibody titres was revealed for Vi-TT immunized mice receiving secondary immunization compared to those receiving single dose (Figure 5.11 B). 161

170 Figure Anti-protein IgG antibody persistence induced by secondary immunization results dependent of the carrier protein and saccharide length. Sera from WT mice immunized intraperitoneally by Vi-CRM 197, Vi-TT, fvi-crm 197 or fvi-tt are collected at day 145 to compare the anti-protein antibody persistence between groups receiving a single or a secondary immunization. Graphs report data comparing A) anti-crm 197 antibody titres induced by CRM 197 -conjugates and B) anti-tt antibody titres induced by TT-conjugates. Symbols in dark red and dark green are correlated to mice respectively immunized with Vi-CRM 197 and Vi-TT, meanwhile light red and light green are correlated to mice immunized respectively by fvi-crm 197 and fvi-tt. * represents statistically significant difference between two groups (p<0.05) and statistical analysis is conducted as reported in paragraph 2.3. Data are representative of four mice per group and one experiment 162

171 Immunizations with Vi conjugates and Vi PS induces ASCs in the spleen and bone marrow Vi-specific antibody secreting cells were assessed at day 145 post-primary immunization in bone marrows and spleens of mice immunized with Vi PS, Vi-CRM197, Vi-TT, fvi-crm197 or fvi-tt. Each group was composed by four WT mice, thus experiments were performed on two different days, due to the high number of mice involved in the study. Vi PS and full-length conjugate immunizations induced the Vi-specific ASCs in the spleens and bone marrow. Vi-specific cells expressing IgM, IgG and IgA were detected. In contrast mice receiving fragmented Vi conjugates lacked anti-vi IgG ASCs. Little difference was seen between the conjugated vaccines (administered at 2 g Vi/dose) and Vi PS (administered at 25 g dose). Comparable anti-vi IgM ASCs were revealed in mice immunized by the different Vi conjugates (Figure 5.12), meanwhile a trend of enhanced anti-vi IgG and IgA ASCs was revealed for Vi-TT immunized mice compared to other groups (Figure 5.13 and 5.14). 163

172 Figure Vi-specific IgM antibody secreting cells levels in mice immunized with Vi PS and Viconjugates. Spot forming units (SFU) of Vi-specific secreting cell levels in spleen and bone marrow expressing IgM immunoglobulins. Anti-Vi antibody secreting cells are collected 145 days postimmunization from spleens and bone marrows of WT mice receiving Vi PS or one of these conjugates: Vi-CRM 197, Vi-TT, fvi-crm 197 or fvi-tt. Symbols in blue, dark red and dark green are correlated to mice respectively immunized with Vi PS, Vi-CRM 197 and Vi-TT, meanwhile light red and light green are correlated to mice immunized respectively with fvi-crm 197 and fvi-tt. Data are representative of four mice per group and two experiments: vertical lines divide the data obtained from different experiments 164

173 Figure A trend of enhanced Vi-specific IgG antibody secreting cells levels revealed in mice immunized with Vi-TT conjugate (highlighted by dotted line). Spot forming units (SFU) of Vispecific secreting cell levels in spleen and bone marrow expressing IgG immunoglobulins. Anti-Vi antibody secreting cells are collected 145 days post-immunization from spleens and bone marrows of WT mice receiving Vi PS or one of these conjugates: Vi-CRM 197, Vi-TT, fvi-crm 197 or fvi-tt. Symbols in blue, dark red and dark green are correlated to mice respectively immunized with Vi PS, Vi- CRM 197 and Vi-TT, meanwhile light red and light green are correlated to mice immunized respectively with fvi-crm 197 and fvi-tt. Data are representative of four mice per group and two experiments: vertical lines divide the data obtained from different experiments 165

174 Figure A trend of enhanced Vi-specific IgA antibody secreting cells levels revealed in mice immunized with Vi-TT conjugate (highlighted by dotted line). Spot forming units (SFU) of Vispecific secreting cell levels in spleen and bone marrow expressing IgA immunoglobulins. Anti-Vi antibody secreting cells are collected 145 days post-immunization from spleens and bone marrows of WT mice receiving Vi PS or one of these conjugates: Vi-CRM 197, Vi-TT, fvi-crm 197 or fvi-tt. Symbols in blue, dark red and dark green are correlated to mice respectively immunized with Vi PS, Vi- CRM 197 and Vi-TT, meanwhile light red and light green are correlated to mice immunized respectively with fvi-crm 197 and fvi-tt. Data are representative of four mice per group and two experiments: vertical lines divide the data obtained from different experiments 166

175 5.3.3 Assessment of protection after immunization with different Vi based vaccines Induction of Vi expression by Salmonella Typhimurium strains We had obtained a number of S. Typhimurium strains that had been reported to express Vi after their genetic modification (255). To identify which one could be used as Vi source for bacterial challenge, all strains (Table 2.4) were grown under different osmolarities (Figure 5.15). Among the strains, only IR715 phon::viab rpos::cm r reproducibly induced significant Vi expression as shown by ELISA (Figure 5.15). We then infected WT mice by IR715 phon::viab rpos::cm r strain grown in culture medium with 85 mm NaCl. Pilot studies were conducted to test the optimal bacterial dose. Initially two doses were tested and animal welfare was monitored for seven days post-challenge. The challenge with 10 5 CFU/mouse of S. Typhimurium IR715 phon::viab rpos:: Cm r induced clinical signs including weight loss, reduced mobility and swollen abdomen. In contrast mouse challenged with 10 4 CFU/dose of the same S. Typhimurium strain showed no signs and no reduced weight, and bacteria were readily detectable in the spleen (Table 5.2). In order to enhance the bacterial dose without enhancing clinical signs, another pilot study was conducted in WT mice challenged by 10 5 CFU/dose but the experiment was only four days post-infection. No reduction in weight or other clinical signs were observed (Table 5.2). 167

176 Figure Salinity of the culture medium modulates Salmonella Typhimurium Vi expression. ELISA plates were coated with each bacterial strain and incubated with serum from a WT mouse immunized with Vi-TT with 2 g Vi/dose and sacrificed seven days post-immunization. The absorbance signal at 405 nm detected indirectly the expression of Vi from each bacteria. As positive control 5 g of Vi PS were coated in the same ELISA plate 168

177 Table 5.2. List of the pilot studies performed to define bacterium dosage and duration of the study post-challenge with Salmonella Typhimurium strain IR715 phon::viab rpos::cm r Days postchallenge Bacterium dosage Mouse n Spleen mass (mg) CFUs per spleen x x x x10 3 Based on these preliminary results the bacterial challenge study was designed as follows. WT mice were immunized at day 0 with 25 g of Vi PS or with full-length conjugate (Vi-CRM197 or Vi-TT) with 2 g Vi given. Non-immunized mice were used as a control group. All mice were challenged at day 14, because it was demonstrated both full-length conjugates can induce high anti-vi antibody response two weeks after immunization (Chapter 3, section 3.3.4). All mice were challenged i.p. with 10 5 CFU/dose S. Typhimurium IR715 phon::viab rpos::cm r for three days (Figure 5.16 A). Challenge studies were performed twice, one experiment is shown. ELISA assays for anti-vi IgM antibody were performed using sera collected at day 17 (3 days post-infection). Anti-Vi IgM antibody titres were detected in all mice (Figure 5.16 B). Sera collected through bleeds at day 0, 14 and 17 (Figure 5.16 A) were assayed by ELISA to detect anti-vi IgG antibody titres pre- and post-infection. No Vi-specific IgG antibody titres were detected from sera bled before challenge at day 0 and at day 14 for non-immunized mice. Compared to non-immunized both Vi-CRM197 and Vi-TT immunized groups had induced significantly higher anti-vi IgG antibody titres at 14 days (Figure 5.16 C). Reflecting findings in Chapter 3 section 3.3.4, comparable anti-vi IgG antibody titres were induced by Vi-CRM

178 and Vi-TT immunization at 14 days. Lower IgG titres were detected from Vi PS immunized mice (Figure 5.16 C). In all groups anti-vi IgG antibodies were not enhanced three days post-challenge (Figure 5.16 C). Spleen mass and bacterial CFUs per spleen and liver were collected for all mice (Figure 5.17). The spleen mass was greater in non-immunized than in immunized mice demonstrating bacterial infection occurred in non-immunized mice (Figure 5.17 A). Furthermore bacterial CFU levels in spleens and livers from immunized mice were significantly lower compared to non-immunized mice. There were no significant differences between the immunized groups in bacterial number (Figure 5.17 B and C). This suggests immunization of conjugates with 2 g Vi/dose produced comparable protection to 25 g Vi PS immunization in immunized mice for two weeks. 170

179 Figure Anti-Vi IgG antibody secreting cell levels isolated from spleens are comparable between mice immunized with Vi-CRM 197 or Vi-TT, and higher compared with non-immunized and Vi PS immunized mice. A) Wild type mice are non-immunized or immunized with Vi PS, Vi- CRM 197 or Vi-TT at day 0. For all immunizations Vi dose is 2 g at exception of Vi PS that is administered at 25 g. Mice, immunized or not, are challenged with 10 5 CFU/dose of S. Typhimurium IR715 phon::viab rpos:: Cm r 14 days post-immunization. B) Anti-Vi IgM antibody titres from sera collected three days post-infection. Symbols in grey, blue, red and green are correlated to mice respectively non-immunized, immunized with Vi PS, Vi-CRM 197 and Vi-TT. C) Anti-Vi IgG antibody titres from sera collected through bleeds at day 0, 14 and 17. The response of individual animals is represented by circle symbol with scaled colours accordingly to the time for bleeding.. * represents statistically significant difference between two groups (p<0.05) at day 14, statistical analysis is conducted as reported in paragraph 2.3. Data are representative of four mice per group and one experiment 171

180 Figure Vi-CRM 197 or Vi-TT immunization with 2 g Vi dose can impair infection with Salmonella Typhimurium strain expressing Vi equally to 25 g Vi PS immunization. Wild type mice are immunized with 25 g of Vi PS or with Vi conjugates (Vi-CRM 197 or Vi-TT) with 2 g Vi/dose. Non-immunized and immunized WT mice are challenged 14 days post-immunization with 10 5 CFU/dose of S. Typhimurium IR715 phon::viab rpos:: Cm r. Spleens and livers are collected three days postinfection. A) Spleen masses, B) bacterial burdens detected in spleens and C) in livers are represented by grey, blue, red and green circles correspondent respectively to non-immunized, Vi PS, Vi-CRM 197 and Vi-TT immunized mice. * represents statistically significant difference between two groups (p<0.05) and statistical analysis is conducted as reported in paragraph 2.3. Data are representative of four mice per group and one experiment 172

181 Vi-specific ASCs number were examined in the spleen and bone marrow of immunized mice subjected to bacterial challenge. In order to demonstrate the effect of immunization and infection, non-immunized and non-challenged WT mice were used as additional control groups (Figure 5.18 and 5.19). Infection induced anti-vi IgM ASCs in the spleen (Figure 5.18). Whereas ELISpots showed that anti-vi IgG ASCs were reproducibly detected in the spleen, meanwhile no relevant anti-vi IgG ASCs were detected in the bone marrow after immunization with the conjugate vaccines (Figure 5.19). 173

182 Figure Infection induces anti-vi IgM ASCs in the spleen. Spot forming units (SFU) of Vispecific secreting cells expressing IgM immunoglobulins in spleen and bone marrow. Wild type mice were non-immunized or immunized with Vi PS, Vi-CRM 197 or Vi-TT at day 0. For all immunizations Vi dose was 2 g at exception of Vi PS that was administered at 25 g. Mice, immunized or not, were challenged with 10 5 CFU/dose of S. Typhimurium IR715 phon::viab rpos:: Cm r 14 days postimmunization. Anti-Vi antibody expressing cells were isolated and enumerated by ELISpot three days post-challenge from spleens and bone marrows. Circles in grey, blue, red and green are correlated to infected mice respectively non-immunized, immunized with Vi PS, Vi-CRM 197 and Vi-TT. Empty circles correspond to non-immunized and non-infected mice. * represents statistically significant difference between two groups (p<0.05) and statistical analysis is conducted as reported in paragraph 2.3. Data are representative of four mice per group at exception of two mice non-immunized and nonchallenged used as control group 174

183 Figure Vi-CRM 197 and Vi-TT induce comparable anti-vi IgG antibody secreting cells isolated from spleens, and higher anti-vi IgG antibody secreting cells compared with nonimmunized and Vi PS immunized mice. Spot forming units (SFU) of Vi-specific secreting cells expressing IgG immunoglobulins in spleen and bone marrow. Wild type mice were non-immunized or immunized with Vi PS, Vi-CRM 197 or Vi-TT at day 0. For all immunizations Vi dose was 2 g at exception of Vi PS that was administered at 25 g. Mice, immunized or not, were challenged with 10 5 CFU/dose of S. Typhimurium IR715 phon::viab rpos:: Cm r 14 days post-immunization. Anti-Vi antibody expressing cells were isolated and enumerated by ELISpot three days post-challenge from spleens and bone marrows. Circles in grey, blue, red and green are correlated to infected mice respectively non-immunized, immunized with Vi PS, Vi-CRM 197 and Vi-TT. Empty circles correspond to non-immunized and non-infected mice. * represents statistically significant difference between two groups (p<0.05) and statistical analysis is conducted as reported in paragraph 2.3. Data are representative of four mice per group at exception of two mice non-immunized and non-challenged used as control group 175