A Typing System for Neisseria gonorrhoeae Based on Biotinylated Oligonucleotide Probes to PIB Gene Variable Regions

Transcription

1 1652 A Typing System for Neisseria gonorrhoeae Based on Biotinylated Oligonucleotide Probes to PIB Gene Variable Regions Dorothea K. Thompson, 1,a Carolyn D. Deal, 1 Catherine A. Ison, 3 Jonathan M. Zenilman, 2 and Margaret C. Bash 1 1 Division of Bacterial, Parasitic, and Allergenic Products, Center for Biologics Evaluation and Research, Bethesda, and 2 Division of Infectious Diseases, Johns Hopkins University School of Medicine, Baltimore, Maryland; 3 Department of Infectious Diseases and Microbiology, Imperial College School of Medicine, St. Mary s Campus, London, United Kingdom The porin proteins PIA and PIB of Neisseria gonorrhoeae are serotyping antigens for the serovar classification system and leading candidates for gonococcal vaccine development. Although serotyping has been a useful tool, this method can be insensitive to critical sequence changes in the por gene, including those in surface-exposed variable regions (VRs). A sensitive and specific typing system for N. gonorrhoeae has been developed that uses biotin-labeled oligonucleotide probes with chemiluminescence detection to type PIB gene VRs. The PIB VR types of geographically and temporally diverse gonococcal strains and sexual contact isolates were determined. por VR typing discriminated between most unrelated isolates and provided information about individual VR type that was not apparent from serovar designations. PIB VR typing avoids limited monoclonal antibody availability, interlaboratory variation, and the requirement for culture-based surveillance associated with gonococcal serotyping, and provides useful information about the molecular epidemiology of individual por gene VRs. Neisseria gonorrhoeae, an obligate human pathogen, causes one of the most common sexually transmitted diseases worldwide. No vaccine currently exists for N. gonorrhoeae infections, although several major cell surface antigens have been evaluated as potential targets [1 6]. The outer membrane porin (Por) trimer is a promising candidate for inclusion in a gonococcal vaccine, because the protein is antigenically stable during the course of infection [7]. In addition, Por constitutes the predominant antigen on the gonococcal cell surface [8] and is encoded by an essential, single-copy gene. The 2 variants of the Por protein, PIA and PIB, are the structural and functional equivalents of the meningococcal PorB proteins [9 12]. A 2-dimensional model of the topology of the meningococcal and gonococcal porins was proposed by van der Ley et al. [11]. The model predicts 8 surface-exposed loops that display extensive variation in length and amino acid Received 5 November 1999; revised 11 January 2000; electronically published 15 May Presented in part: 99th General Meeting of the American Society for Microbiology, Chicago, May 1999 (abstract C-370). Financial support: Food and Drug Administration Office of Women s Health, Rockville, MD. D.T. was supported by a postdoctoral fellowship administered by the Oak Ridge Institute for Science and Education. The work of C.I. was funded by the Wellcome Trust. a Present affiliation: Oak Ridge National Laboratory, Environmental Sciences Division, Oak Ridge, Tennessee. Reprints or correspondence: Dr. Margaret C. Bash, Division of Bacterial, Parasitic and Allergenic Products, HFM-428, Center for Biologics Evaluation and Research, 1401 Rockville Pike, Rockville, MD The Journal of Infectious Diseases 2000;181: by the Infectious Diseases Society of America. All rights reserved /2000/ $02.00 sequence. By contrast, membrane-spanning sequences are highly conserved among neisserial porins [11]. In support of this structural model, epitopes for antibodies that possess bactericidal activity were mapped to regions that reside within the putative external loops [5, 11, 13]. Sequence variations are nonrandom [14, 15] and occur primarily at the predicted loop apex. Por sequence diversity among strains serves as the basis for serological typing. In this system, the PIA and PIB classes are subdivided into serovars defined by their coagglutination patterns with a panel of antigonococcal monoclonal antibodies (MAbs) [16]. When combined with auxotyping, serovar classification provides a useful degree of discrimination among unrelated gonococcal isolates [17]. However, serotyping can be insensitive to sequence changes in the Por-encoding gene, including potentially significant differences among major variable regions (VRs) suspected to be antigenic determinants. In addition, only a fraction of the Por-typing MAbs have been mapped [18], thus limiting the information supplied by serovar determinations. Various DNA-based typing schemes have emerged to provide higher resolution of the genotypic variability found among gonococcal strains [18 29]. Although DNA sequencing gives complete information on por VRs, this technology is generally not readily available or affordable. A molecular typing system based on probes specific for por VRs, however, might provide a reasonable approximation of por sequence. In this study, we describe the development and utility of a genetic typing system for N. gonorrhoeae based on biotinlabeled oligonucleotide probes to PIB gene VRs. The PIBencoding gene was initially selected for testing the feasibility of

2 JID 2000;181 (May) Gonococcal por Typing System 1653 this typing system, because PIB strains occur at a greater frequency than PIA strains in gonococcal infections. The discriminatory power of the method was examined by analyzing strains known to be temporally and geographically diverse, unrelated gonococcal strains within a defined population, and epidemiologically linked isolates. Furthermore, the oligonucleotidebased typing system was compared with PIB gene sequence information and with the classical serovar system. Materials and Methods N. gonorrhoeae strains and growth conditions. Four collections of gonococcal isolates were analyzed in this study: (1) 17 geographically and temporally diverse strains [17]; (2) 21 PIB strains isolated from US military personnel stationed in South Korea who had participated in a pilus vaccine trial [4]; (3) 13 reported sexual contact pairs isolated in Baltimore, Maryland; and (4) 12 sets of strains isolated from reported sexual contacts in England. The following gonococcal strains served as positive and negative controls for oligonucleotide probe hybridization and were kindly provided by M. M. Hobbs (University of North Carolina School of Medicine, Chapel Hill): 1861, 5441, 5589, 6200, 9087, and 9299 [29]. J. Cannon (University of North Carolina School of Medicine) provided gonococcal strains FA1090 and FA19, and strain MS11mkC was from our collection. N. gonorrhoeae strains were cultivated from either lyophilized cultures or frozen stocks that had been stored at 70 C in Greaves solution (5% bovine serum albumin, 5% monosodium glutamate, and 10% glycerol). Cells were grown overnight at 37 C in 5% (vol/ vol) CO 2 on GC medium base agar (Difco, Detroit) supplemented with the following components (per liter): 3.0 g glucose, 8.3 g Fe(NO 3) 3 9H 2O, 0.1 g L-glutamine, 0.1 ml cocarboxylase (2%), and 40 mg L-cysteine hydrochloride. DNA preparation. Genomic DNA was isolated from N. gonorrhoeae cells either by phenol extraction, as described elsewhere [30], or by use of the Wizard Genomic DNA purification kit (Promega, Madison, WI) according to the manufacturer s instructions. The concentration and purity of genomic DNA was determined spectrophotometrically at wavelengths of 260 nm and 280 nm. Polymerase chain reaction (PCR) amplification of the PIB gene was done with forward primer PIB.Fpr (5 -ATTGCCCTGA- CTTTGGCAGCCCTTCCT-3 ) and reverse primer PIB.Rpr (5 - TTGCAACCAGCCGGCAGAAACCAAGGC-3 ) by use of a DNA Thermal Cycler (Perkin-Elmer, Norwalk, CT). Forward and reverse PCR primers were complimentary to nucleotide sequence regions located at 16 to 42 and 961 to 982 relative to the transcription start site of the MS11 por gene. Primers were complementary to conserved regions of the PIB gene on the basis of comparisons of published por sequences and corresponded to regions within the signal peptide and loop 8 coding regions, respectively. Amplification reactions were done by use of 1 U of Taq DNA polymerase (Boehringer Mannheim, Indianapolis), 200 ng of template genomic DNA, 1 mm of each primer, 200 mm deoxynucleotide triphosphates, and 1 PCR buffer containing MgCl 2 (Boehringer Mannheim) in a total reaction volume of 100 ml. PCR reactions were programmed for 30 sequential cycles of denaturation at 94 C (1 min), primer annealing at 50 C (1 min), and primer extension at 72 C (2 min), followed by a final extension reaction at 72 C (5 min). Reactions generated single PCR products of the predicted size of 1 kb, as determined by 1% agarose gel electrophoresis followed by ethidium-bromide staining. Multiple-sequence alignment analysis. All sequence analyses were done by use of GCG Wisconsin Package programs (Genetics Computer Group, Madison, WI). Fifty-two full-length PIB sequences and 21 partial PIB sequences were identified in GenBank, excluding sequences that represent the same strain and sequences from strains subjected to transformation. Gen- Bank accession numbers of sequences analyzed are as follows: AF AF [29], AF AF [15], M21289 [31], J03017 [32], U07887, U09044 U09047, U75631 U75643 [18], U17019 U17029 [23], U17234 U17238 [23], U07827 U07829 [33], X52823, X52825, X52827 [13], AF015117, and AF [34]. Multiple sequence alignment of all PIB sequences done by use of the PILEUP program (Genetics Computer Group) was followed by individual multiple sequence alignments of the predicted encoding regions for loops 1, 5, and 6 of the mature protein. The consensus sequences in figure 1A 1C are from multiple sequence alignments done by using 1 representative strain for each unique VR sequence. PIB oligonucleotide probes. The 5 -end biotinylated oligonucleotides were synthesized by the Facility for Biotechnology Resources at the Center for Biologics Evaluation and Research, Food and Drug Administration. Oligonucleotide probes were designed to different PIB VRs found within the predicted surface-exposed loops 1, 5, and 6 on the basis of manual sequence alignment of 76 PIB genes deposited in the GenBank. The VR-specific probes used in this study are listed in table 1. In addition, probe PIB-univ, which specifically recognizes a conserved sequence that flanks the VR in PIB loop 5, was used to identify gonococcal strains that contain the PIB gene. This oligonucleotide was specific for both genomic DNA prepared from PIB strains and PCR-amplified PIB DNA. Probe PIA-3, which is complementary to a common sequence located in loop 3 of PIA genes, was specific for PCRamplified PIA DNA but cross-hybridized to genomic DNA prepared from PIB strains. A BLASTN search against the N. gonorrhoeae FA1090 genome database (Gonococcal Genome Sequencing Project [http// revealed that the oligonucleotide sequence for PIA-3 had sequence homology (78%) to an unknown region (contig 38) within the genome. DNA sequence analysis of PIB VRs. The loop 5 regions from gonococcal PIB strains 0068, 204, 288, 375, 446, 721, 948, and TR137 were sequenced by ACGT (Northbrook, IL). The PIB gene from these strains was amplified by PCR as described, and then PCR DNAs were purified by use of QIAquick PCR Purification kit (Qiagen, Chatsworth, CA) according to the manufacturer s instructions. PCR DNA served as the template in single-sequencing reactions that contained either the forward primer PIB5seq.F (5 -GGCGAATCTTACCACGTTGG-3 ) or the reverse primer PIB5seq.R (5 -GGCATTATTGTCGTAACCGC-3 ). Dot blotting, hybridization, and signal detection. Before it was blotted, genomic or PCR-generated DNA was denatured in a final alkali concentration of 0.4 M NaOH containing 0.4 mm EDTA and then boiled at 100 C for 5 min. DNA was applied to Zeta- Probe-GT nylon membrane (Bio-Rad Laboratories, Hercules, CA)

3 1654 Thompson et al. JID 2000;181 (May) in amounts of 20 mg per well (for genomic DNA) or in equal volumes of 20 ml (for PCR DNA) by use of a 96-well manifold vacuum apparatus and then fixed to the membrane by UV irradiation (120 mj/cm 2 ) by use of the UV Stratalinker (Stratagene, La Jolla, CA). Membranes were placed in sealable plastic bags containing 150 ml of hybridization solution (0.25 M Na 2 HPO 4 [ph 7.2], 7% sodium dodecyl sulfate [SDS]) per cm 2. The hybridization temperature (T H ) for each oligonucleotide probe was determined empirically by use of positive and negative control DNAs. Prehybridization was done at the appropriate T H for min before biotinylated oligonucleotide probes (30 ng/ml of hybridization solution) were added to the membranes and permitted to anneal to their complementary sequences for 3 4 h. After probe hybridization, membranes were washed twice in 2 SSC (1 containing 150 mm NaCl and 15 mm trisodium citrate) and 0.1% SDS for 5 min each at room temperature and then once in 0.1 SSC and 0.1% SDS for 15 min at the T H. For signal detection by chemiluminescence, membranes were washed for 2 min in 1 maleic acid buffer (Boehringer Mannheim) and 3% Tween-20 (Sigma, St. Louis) before being blocked in 1 maleic acid buffer containing 5% nonfat dry milk (blocking grade; Bio-Rad) for 30 min at room temperature. Maleic acid buffer could be replaced by 1 Trisbuffered saline (Bio-Rad). After it was blocked, the membrane was incubated for 30 min at room temperature in the presence of streptavidin-horseradish peroxidase conjugate (Boehringer Mannheim) that was diluted 1 : 500 in blocking solution. Membranes were washed twice in 1 maleic acid buffer containing 3% Tween-20 for 15 min at room temperature, to remove nonspecifically bound conjugate. Membranes then were immersed in a solution containing equal volumes of ECL detection reagents (Amersham Pharmacia Biotech, Piscataway, NJ) for 1 min at room temperature followed by exposure to Kodak X-Omat-AR film. Membranes could be probed at least 5 times without any detectable reduction in target DNA retention after stripping in 0.1 SSC and 0.5% SDS for 20 min at 80 C 90 C. Results Figure 1. Multiple sequence alignment of regions coding for predicted PIB loops 1 (A), 5 (B), and 6 (C). N is the no. of strains reported in GenBank that are identical in the regions shown. Deletions are represented by a dash ( ), and a dot (.) represents agreement with the consensus sequence. Nos. located below the consensus sequence indicate the nucleotide position on the basis of sequence M21289 (MS11) [31]. Oligonucleotide probes used in this study are indicated by boxes. Analysis of PIB VRs and probe selection. The deduced amino acid sequences of N. gonorrhoeae PIB genes in GenBank were aligned in regions that corresponded to the 8 external loops predicted by neisserial porin models [11, 12]. The sequence of loops 2, 4, and 8 were mostly conserved among the different analyzed PIB sequences, whereas loops 1, 3, 5, 6, and 7 displayed sequence variability among strains (data not shown). Nucleic acid sequence alignments of PIB genes reported in GenBank were done for VRs encoding predicted loops 1, 5, and 6 of the porin protein. Strains isolated in the United States [29], the United Kingdom [13, 18, 34], Kenya [15], and Singapore [23, 33] were represented. Fifty-two Gen- Bank entries that represented 71 strains had sequence data available for the loop 1 region (figure 1A). Three common sequence types represented 46 strains (65%), and all remaining sequences differed from 1 of the common sequences by only 1 or 2 nucleotides. Sequences from 73 GenBank entries that represented 92

4 JID 2000;181 (May) Gonococcal por Typing System 1655 Table 1. Oligonucleotide probes used in gonococcal PIB gene variable region (VR) typing. Probe Sequence (5 r3 ) Translated VR Reference sequence T H, C a strain GenBank accession no. Universal PIB-univ AGATACGGCGAAGGCACTAAAAAA RYGEGTKK 55 PIA-3 TTAAGCAACATTGCCCAACCCGAAGAA LSNIAQPEE 59 Loop ACAGACGGCAAGGTAAGTAAA TDGKVSK 52 MS11 M CGGGAAGGCAAAGTAGTTGGC REGKVVG AF ACAAAAGGCAAGGTAAGTAAA TKGKVSK AF Loop GATGATCAAAATTATAGTATA DDQNYSI AF GATGGTCAAGCTTATAGTATG DGQAYSM 48 S1 U ATGGAAGGATATCTATATAATATCCCC MEGYLYNIP AF ATCGAATACGATAATCAATTT IEYDNQF AF ATGGAAGGATATACATATAAT MEGYTYN 43 S26 U ATGGAAGGATATGCATATAAT MEGYAYN 45 S11 U GATAGTCAATATTATAGTATC DSQYYSI 43 SU50 U GAATACGAACATCAAGTTTAT EYEHQVY 45 MS11 M21289 Loop GGAGCAAGGAGGGCTAATTCG GARRANS AF GGAGCAACGAGGGTTAATTCG GATRVNS AF GGAGCAATGAGCGGTAATTCG GAMSGNS AF GGAACATGGCGTGCTAATTCG GTWRANS AF (ms11) b CAAAATCAAATAGTGCGTGAT QNQIVRD 50 S11 U07828 a Hybridization temperature (T H ) for each oligonucleotide probe was determined empirically. b Probe 6-5 recognizes strain MS11 loop 6 VR (QNQLVRD), even though the oligonucleotide differs from the MS11 PIB gene by a single base. Sequence for oligo 6-5 occurs in clinical isolates whose PIB gene sequence had been deposited in GenBank. strains were aligned at the loop 5 VR and loop 6 VR. Multiplesequence alignment of loop 5 VRs revealed 38 sequence types; however, many of these sequences were closely related (figure 1B). Six sequences were found in which the given sequence represented 5 strains. For each of these, strains from at least 2 geographic locations were represented. Strains from all 4 geographic areas were represented in the largest group, all having identical loop 5 VR sequences to M21289 (strain MS11). The 6 common sequence types together represented 49 of the strains (52%), and an additional 20 strains differed from these common sequence types by 2 nucleotides. In the alignment of loop 6 VRs, all 92 strains were represented by 15 sequence types (figure 1C). The loop 6 VR sequences of AF and AF represented 30 and 17 strains, respectively, and AF044788, AF044794, and U07828 were also common. All other sequences, except AF015117, showed 2 nucleotide differences from the most common motifs. AF is the por sequence of strains RT117 and RT380, which were reported by Cooke et al. [34] as PIA/PIB hybrids. To examine the utility of a genotyping system based on nucleotide differences in por VRs, we used biotinylated oligonucleotide probes designed to predominant variable sequence motifs located in loops 1, 5, and 6 (figure 1A 1C). Probes were initially tested against genomic DNA isolated from 6 standard PIB strains (MS11, 1861, 5441, 5589, 9087, and 9299), the PIA strain FA19, and Neisseria meningitidis strain H355. Figure 2 shows typical hybridization patterns for genomic DNA dot blots of standard PIB strains probed with 3 VR-specific oligonucleotides. Each hybridization analysis of clinical isolates included known PIB strains that served as positive and negative controls for probings. Cross-hybridization was not observed for DNA target sequences that differed by 3 nucleotides from the oligonucleotide sequence. Some probes for example, 1-1 and 1-2 clearly distinguished VR sequences differing by 2 nucleotides. Standard PIB DNAs with VR sequences identical to the probe showed strong hybridization signals (see MS11, 1861, and 9087 in figure 2A; 9299 in figure 2B; and 5589 in figure 2C). In the case of probe 6-2, control PIB DNAs differing by 2 nucleotides from the probe showed positive hybridization signals of weaker intensity indicating sequence similarity to the probe (see 1861 in figure 2C). Temporally and geographically diverse strains. Genomic DNA from 17 geographically and temporally diverse N. gonorrhoeae strains that had been previously serotyped, auxotyped, and opa typed [17] were por VR typed in a blinded analysis to assess the utility of our genotyping system. PIB strains were distinguished from PIA strains by hybridization with the PIBuniv oligonucleotide probe that was complementary to a conserved sequence in the loop 5 region of PIBs. Of the 12 identified PIB strains, 9 distinct hybridization types were found, compared with 10 previously identified serotypes (table 2) and 12 opa types. Three sets of strains (3790/855, 899/3797, and 3224/ B95) displayed identical hybridization patterns that suggest similar por VR types for the 3 loops tested. These similarities were not identified by serovar type. In addition, 2 sets of strains (PPNG949/3797 and 855/899) with identical serovar designations were shown to differ in 2 of 3 VRs examined. Three of the PIB strains (3633, 3224, and B95) did not recognize any of the 8 loop 5 probes, and work on sequencing these strains is in progress. All of the geographically and temporally diverse

5 1656 Thompson et al. JID 2000;181 (May) Figure 2. Three separate genomic DNA dot blots (A C) of standard Neisseria gonorrhoeae PIB strains probed with variable region specific oligonucleotides. Genomic DNA from known PIB strains MS11 (M21289), 1861 (AF044790), 5441 (AF044795), 5589 (AF044794), 6200 (AF044784), 9087 (AF044796), and 9299 (AF044788) was probed with the following biotin-labeled oligonucleotides in separate hybridization assays: (A) 1-1, (B) 5-1, and (C) 6-2 (table 1). Hybridized probes were detected by chemiluminescence. PIB strains could be typed with a loop 1 and loop 6 probe from our panel. Furthermore, none of the genomic DNAs from strains identified as PIAs hybridized a PIB probe, indicating that all of the oligonucleotides were specific for PIB genes. Strains from a gonococcal pilus vaccine trial. The complete panel of oligonucleotide probes was used to analyze genomic DNA prepared from 21 unrelated PIB strains isolated over a 6-week period from US military personnel stationed in South Korea (table 3). These Korean strains represented a temporally and geographically defined population. As the probing results in table 3 show, the Korean strains constitute a relatively homogeneous population with only 5 distinct hybridization patterns exhibited by the 21 strains; 100% of the DNAs hybridized to a loop 1 probe and to a loop 6 probe, whereas 81% recognized a loop 5 probe. DNA sequencing of loop 5 either verified probe hybridizations (strains 288, 721, 0068, and 375) or revealed the presence of a loop 5 VR sequence (5 -GGCTATCAGTATAA- TATCCCC-3 ) that was not included in our panel of oligonucleotides (strains 446 and 204) [table 3]. Epidemiologically linked isolates. Hybridizations with biotinylated probes were done to type PCR-amplified por DNA from 13 reported gonococcal sexual contact pairs isolated at a sexually transmitted diseases clinic in Baltimore, Maryland. By using the PIB-univ probe, we found 9 of the reported contact pairs expressed PIB, whereas 2 putatively related strains, 783 and 950, were shown to be a PIA/PIB discordant pair (table 4). PIA strains were identified by hybridization with the PIA-3 probe, which specifically recognizes a common loop 3 sequence in PCR-amplified PIA DNAs. No hybridization of PIB VR specific probes occurred among PIA strains. With the exception of the 1 discordant pair, all of the PIB partner strains gave identical hybridization patterns for all 3 PIB loops analyzed by the oligonucleotide-based typing method (table 4). Only 1 contact pair, 692 and 559, did not hybridize a PIB loop 5 probe. The por genes of the Baltimore sexual contact pairs have been sequenced by other investigators as part of another study (R. Viscidi, personal communication). Comparison of the hybridization assay results with the PIB gene sequences revealed that the VR sequence was identical to the nucleotide sequence of the hybridizing probe in 50% of the cases for loop 5 and 170% of the cases for loops 1 and 6. In those instances when the sequences of the PIB VR and the gene probe were not identical, not more than 2 nucleotide differences were present. por Sequencing showed identical loop 5 VR sequences for isolates 692 and 559; this sequence was not represented in our panel of loop 5 probes. The 1 discordant contact pair, 783 and 950, identified in this study was also confirmed by por sequencing (R. Viscidi, personal communication). In addition, the oligonucleotide-based typing scheme was used to analyze PCR-amplified por DNA prepared from 11 sexual contact pairs and 1 triplet from 2 different locations in England. These strains had been serologically characterized by use of classical methods (table 5), and each contact group was different by opa-typing (data not shown). Hybridization with the PIA-3 and PIB-univ probes indicated that of the 12 contact groups of isolates, 9 expressed PIB and 3 expressed PIA. This Table 2. por Variable region (VR) typing of geographically and temporally diverse gonococcal strains. DNA isolate Year isolated Location VR-specific probes a Loop 1 Loop 5 Loop 6 Serovar UK IB UK IB5 PPNG UK IB UK IB UK IB UK IB UK IB UK IB UK IB7 B Bahrain IB19 TR Tanzania IB22 D S. Africa IB13 a Complete panel of oligonucleotide probes (table 1) was tested against genomic DNA isolated from each gonococcal strain. Only listed probes hybridized. A dash ( ) denotes no hybridization signal for the probes used.

6 JID 2000;181 (May) Gonococcal por Typing System 1657 Table 3. Korean gonococcal strains por variable region (VR) typed by oligonucleotide-based method. No. of isolates (%) VR-specific probes a Loop 1 Loop 5 Loop 6 7 (33.3) (28.6) (19.0) (9.5) b (9.5) c a Complete panel of biotin-labeled oligonucleotide probes (table 1) was tested against genomic DNAs isolated from each gonococcal strain. Probes that hybridized are listed. A dash ( ) denotes no hybridization signal for the probes tested. b PIB loop 5 gene region from representative strains not recognizing a loop 5 probe were sequenced. Strain 446 contained the loop 5 variable region GY- QYNIP, which was not represented in the panel of oligonucleotide probes. c PIB loop 5 gene region from representative strains not recognizing a loop 5 probe were sequenced. Strain 204 contained the loop 5 variable region GY- QYNIP, which was not represented in the panel of oligonucleotide probes. finding agreed with the serotype designation for these strains. As shown in table 5, all of the PIB strains hybridized to a loop 1, 5, and 6 probe. Furthermore, strains reported to be epidemiologically linked exhibited identical hybridization patterns. This finding was consistent with the opa-type data and with the serovar classification, except in the case of partner strains 4541 (IB2) and 4498 (IB26). PIB VR typing of the 2 groups of sexual contact isolates suggested that the collection of isolates from England were less diverse than those from Baltimore. The 9 groups of PIB sexual contact isolates from England were divided into 6 different por VR genotypes, and the 9 PIB contact pairs from Baltimore exhibited 8 different por VR genotypes (tables 4 5). The strains from England showed only 3 loop 5 VR types and 3 loop 6 VR types (table 5). Among these strains, individual VRs 5-2 (42%), 5-4 (47%), and 6-4 (58%) were common. The Baltimore partner strains exhibited greater heterogeneity in their individual loops, with 7 distinct loop 5 VR types and 4 loop 6 VR types observed within this population (table 4). Serovar data were not available for the Baltimore strains. Serovar typing of the England strains, however, did not distinguish group 7 from group 9 or group 8 from groups 1, 2, and 3 (table 5). Discussion The current standard method for differentiating antigenic variants of Neisseria gonorrhoeae is the Por-based serovar classification system in which serovars are defined by their reactivity with Por-specific MAbs [16]. In combination with auxotyping, serological characterization has proved useful for elucidating the epidemiology of N. gonorrhoeae [17] but has received criticism for not providing sufficient discrimination among unrelated strains. In addition, serotyping depends on the availability of standard MAbs, requires maintenance or transportation of live gonococcal isolates, and is difficult to standardize among laboratories. Genetic typing methods can avoid many of these problems. With the large amount of accumulating knowledge in regard to gonococcal por genes, it was thought that a genotyping system based on por could be developed that would provide the technical advantages of DNA-based methods while also providing epidemiological data relevant to Por-based vaccine development. In this study, we have shown the utility of a DNA-based por VR typing system for N. gonorrhoeae that uses biotin-labeled oligonucleotide probes. A limited number of probes were chosen on the basis of common VR sequence motifs seen among PIB sequences in GenBank. Eighty-eight gonococcal isolates were por VR typed by use of either genomic DNA or PCRamplified por DNA. Cross-hybridization of PIB gene probes with DNA from PIA strains was not observed. The oligonucleotide-based por typing system discriminated among unrelated gonococcal strains and identified epidemiologically related isolates. Oligonucleotide probing distinguished 9 hybridization patterns compared with 10 serovar types and 12 opa-types for the 12 diverse PIB strains (table 2). The PIB isolates from 9 different sexual contact groups from England showed 6 PIB VR hybridization patterns, compared with 4 serovars (table 5) and 9 opa-types (data not shown). The 9 sexual contact pairs of PIB strains from Baltimore showed 8 different por VR genotypes. With the exception of 1 IA/IB discordant pair (isolates 783 and 950 in table 4), all of the reported sexual contact pairs or triplets exhibited identical hybridization patterns. Oligonucleotide-based VR typing also provided information in regard to the prevalence of individual por VR types among Table 4. Oligonucleotide probing of sexual contact PIB isolates from a sexually transmitted diseases clinic in Baltimore, Maryland. VR-specific probes Sexual contact group a Isolate b Loop 1 Loop 5 Loop , d 950, 783 (PIA) 1-1, 5-2, 6-2, 3 407, , , 692 e 1-1 e , , , , , NOTE. Por class for each isolate was determined by oligonucleotide probe hybridization with PIB-univ and PIA-3 and was then subsequently verified by comparison with por sequences. VR, variable region. a Each group consisted of 1 male and 1 female partner. b All strains were isolated from June 1990 through July Most contact pair isolates were obtained on the same day; the maximum interval was 5 days. c Complete panel of oligonucleotide probes (table 1) was tested against polymerase chain reaction amplified por DNA from each gonococcal isolate. Only probes that hybridized are presented. A dash ( ) denotes no hybridization signal for the probes tested. d Discordant pair. e Sexual contact isolates 559 and 692 contained the deduced PIB loop 5 variable region DDQAYYS, as determined by DNA sequencing. VR was not represented in the panel of probes. c

7 1658 Thompson et al. JID 2000;181 (May) Table 5. Sexual contact PIB isolates from England typed by por variable region (VR) oligonucleotide probes. Sexual contact group (sex) Clinic a Date isolated VR-specific probes b Loop 1 Loop 5 Loop 6 Serovar 1 (M, F) L , IB3 2 (M, F) S , IB3 3 (F, M) L IB3 4 (M, F) S , IB2 5 (M, F) L , IB26, IB2 c 6 (F, F, M) L , IB2 7 (M, M) L IB1 8 (M, F) S , IB3 9 (M, M) L IB1 a Clinic location: Sheffield, England (S); St. Mary s in London (L). b Complete panel of oligonucleotide probes (table 1) was tested against polymerase chain reaction amplified por DNA from each gonococcal isolate. Only probes that hybridized are presented. c Two strains from group 5 had identical opa types. different groups of strains. For all N. gonorrhoeae strains analyzed in this study, probe 1-3 consistently hybridized to a lower percentage of gonococcal DNAs than did probes 1-1 and 1-2 (tables 2 5). For loop 5, the Baltimore sexual contact isolates exhibited a high degree of diversity: probes 5-8, 5-1, and 5-5 each hybridized to 10% of the por DNAs; 20% typed as 5-2, 5-4, or 5-7; and 10% displayed a loop 5 VR type not represented in our probe panel (table 4). By contrast, the 19 PIB contact isolates from England exhibited only 3 loop 5 VR types (table 5), and 17 (81%) of 21 Korean isolates hybridized to either probe 5-2 or probe 5-6. This suggests that for a given time or geographical origin, even the highly variable loop 5 region may display a limited number of por VR types. Finally, the loop 6 VR type that corresponded to oligonucleotide 6-4 appeared to be predominant, whereas hybridization with the MS11-like probe 6-5 was uncommon (tables 2, 4, 5). Recent studies have indicated that both point mutations and horizontal genetic exchange form the molecular basis for the diversity of gonococcal Por proteins [15, 29, 34]. Although point mutations contribute to Por diversity, multiple sequence alignments suggest that sequence variation of por may be limited [15, 29, this study]. Many identical nucleotide sequences for individual VRs are found among strains isolated in distinct geographic regions. Hybridization results in this study suggested that there is sufficient stability of individual VRs to allow for por VR typing. For example, all PIB strains were typeable for the VRs corresponding to loops 1 and 6, and 87% (62/71) of the gonococcal DNAs hybridized to 1 of 8 loop 5 probes. Sequence data were available for some strains and showed that oligonucleotide hybridization approximated PIB VR sequence to within 1 or 2 nucleotides. This genetic typing system, however, cannot replace por sequencing in studies interested in identifying single point mutations. Horizontal genetic exchange also contributes to Por diversity. Exchange of the entire por gene results in genetically closely related strains with markedly different Por proteins [29]. Exchange of portions of por result in mosaic patterns in which some regions of por are identical among strains, but other regions are very different. The frequency of mosaic por genes found among strains isolated from a group of commercial sex workers in Nairobi, Kenya, suggests that horizontal genetic exchange contributes significantly to Por diversity [15]. The gonococcal strains typed in this study showed patterns of mosaicism. Hybridization of a particular probe for 1 VR did not predict which probes hybridized to other VRs. Horizontal genetic exchange of partial por genes can be rapidly identified by por VR typing, as described here, because each VR is examined independently. By contrast, por mosaicism may not be identified by serovar designation, because the panel of serotyping MAbs does not completely define all of the loops of Por that display variation. The relationship between por VR typing and PIB serovar is complex and in many cases unclear. A direct relationship between PIB serovar and por VR type is only apparent when the typing MAb identifies a specific VR epitope. This is the case with typing antibody 2D4, whose epitope has been mapped to the loop 6 VR of gonococcal strain MS11 [18]. Two strains in this study have serovars that indicate 2D4 binding: strain 3633 and strain D12 (table 2). As expected, genomic DNA from these strains hybridized to the 6-5 probe, which indicates sequence similarity to MS11 in the loop 6 VR. The serovar of strains hybridizing to probes 5-2 or 6-4 suggested a relationship between these VR sequences and MAbs 3C8 and 2D6, respectively. However, in both cases there were examples of strains that did not follow this pattern. The differences between serovar and por VR type are highlighted by 3 strains in table 2. Strains 855 and 899 are both serovar IB2, which indicates that they bind MAbs 3C8, 2D6, and 2H1. These 2 strains have different loop 5 and loop 6 VR types, which suggest that the 1B2- typing antibodies do not recognize epitopes that involve the VRs of either loop 5 or loop 6. Strain 3797 has identical por VR types to strain 899 for loops 1, 5, and 6; however, strain 3797 binds an additional MAb, 1F5, classifying it as serovar 1B1. Expansion of the por VR typing system to include VR-

8 JID 2000;181 (May) Gonococcal por Typing System 1659 specific oligonucleotide probes for loops 3 and 7 is underway and may help explain some of the differences between serotype and por gene VR type. However, MAb binding may be affected by more than the primary amino acid sequence, and single point mutations not identified by VR genotyping may also result in changes in MAb binding. In summary, a PIB gene VR typing method was developed and applied to a wide variety of clinical strains. This method identified differences among unrelated strains and similarities among sexual contact strains and provided information about the molecular epidemiology of por VRs. VR genotyping has the advantages of DNA-based typing systems because oligonucleotide probes are easily synthesized, transfer of live cultures is not necessary, and typing of PCR-amplified por DNA from urine samples should be possible. Expansion of this system to include PIA typing probes and its application to specific epidemiologic studies, such as typing cohorts of strains isolated over time in a community or typing isolates from repeated infections in core groups, will test the usefulness of the oligonucleotide-based por VR typing method as an epidemiologic tool. Acknowledgments We are grateful to Freyja Lynn, Carl E. Frasch, Michael Brennan, and E. Scott Stibitz for critical comments on the manuscript, and Katherine Schmidt and Jeffrey Yuenger for strain preparation. References 1. Brinton CC, Bryan J, Dillon J. Uses of pili in gonorrhea control: role of bacterial pili in disease, purification, and properties of gonococcal pili, and progress in the development of a gonococcal pilus vaccine for gonorrhea. In: Brooks GF, Gotschlich EC, Holmes KK, eds. Immunobiology of Neisseria gonorrhoeae. Washington, DC: American Society for Microbiology, 1978: Tramont EC, Sadoff JC, Boslego JW, et al. Gonococcal pilus vaccine: studies of antigenicity and inhibition of attachment. J Clin Invest 1981;68: Heckels JE, Virji M, Tinsley CR. Vaccination against gonorrhoeae: the potential protective effect of immunization with a synthetic peptide containing a conserved epitope of gonococcal outer membrane protein IB. Vaccine 1990;8: Boslego JW, Tramont EC, Chung RC, et al. Efficacy trial of a parenteral gonococcal pilus vaccine in men. Vaccine 1991;9: Elkins C, Carbonetti NH, Varela VA, Stirewalt D, Klapper DG, Sparling PF. Antibodies to N-terminal peptides of gonococcal porin are bactericidal when gonococcal lipopolysaccharide is not sialylated. Mol Microbiol 1992;6: Parmar MM, Blake MS, Madden TD. Biophysical and antigenic characterization of gonococcal protein I incorporated into liposomes. Vaccine 1997;15: Zak K, Diaz JL, Jackson D, Heckels JE. Antigenic variation during infection with Neisseria gonorrhoeae: detection of antibodies to surface protein in sera of patients with gonorrhea. J Infect Dis 1984;149: Douglas JT, Lee MD, Nikaido H. Protein I of Neisseria gonorrhoeae outer membrane is a porin. FEMS Microbiol Lett 1988;12: Blake MS, Gotschlich EC. Functional and immunologic properties of pathogenic Neisseria surface proteins. In: Inouye M, ed. Bacterial outer membranes as model systems. New York: John Wiley and Sons, 1987: Judd RC. Protein I: structure, function, and genetics. Clin Microbiol Rev 1989;2:S41 S van der Ley P, Heckels JE, Virji M, Hoogerhout P, Poolman JT. Topology of outer membrane porins in pathogenic Neisseria spp. Infect Immun 1991;59: Derrick JP, Urwin R, Suker J, Feavers IM, Maiden MCJ. Structural and evolutionary inference from molecular variation in Neisseria porins. Infect Immun 1999;67: Butt NJ, Virji M, Vayreda F, Lambden PR, Heckels JE. Gonococcal outermembrane protein PIB: comparative sequence analysis and localization of epitopes which are recognized by type-specific and cross-reacting monoclonal antibodies. J Gen Microbiol 1990;136: Smith NH, Maynard Smith J, Spratt BG. Sequence evolution of the porb gene of Neisseria gonorrhoeae and Neisseria meningitidis: evidence of positive Darwinian selection. Mol Biol Evol 1995;12: Fudyk TC, Maclean IW, Simonsen JN, et al. Genetic diversity and mosaicism at the por locus of Neisseria gonorrhoeae. J Bacteriol 1999;181: Knapp JS, Tam MR, Nowinski RC, Holmes KK, Sandstrom EG. Serological classification of Neisseria gonorrhoeae with use of monoclonal antibodies to gonococcal outer membrane protein I. J Infect Dis 1984;150: Van Looveren M, Ison CA, Ieven M, et al. Evaluation of the discriminatory power of typing methods for Neisseria gonorrhoeae. J Clin Microbiol 1999;37: Cooke SJ, de la Paz H, Poh CL, Ison CA, Heckels JE. Variation within serovars of Neisseria gonorrhoeae detected by structural analysis of outermembrane protein PIB and by pulsed-field gel electrophoresis. Microbiology 1997;143: Falk ES, Danielsson D, Bjornvatn B, Melby K, Sorensen B, Kristiansen BE. Genomic fingerprinting in the epidemiology of gonorrhoea. Acta Derm Venereol 1985;65: Poh CL, Ocampo JC, Sng EH, Bygdeman SM. Rapid in-situ generation of DNA restriction endonuclease patterns for Neisseria gonorrhoeae. J Clin Microbiol 1989;27: Poh CL, Lau QC. Subtyping of Neisseria gonorrhoeae auxotype-serovar groups by pulsed-field gel electrophoresis. J Med Microbiol 1993;38: Poh CL, Loh GK, Tapsall JW. Resolution of clonal subgroups among Neisseria gonorrhoeae IB-2 and IB-6 serovars by pulsed-field gel electrophoresis. Genitourin Med 1995;71: Poh C, Lau QC, Chow VTK. Differentiation of Neisseria gonorrhoeae IB-3 and IB-7 serovars by direct sequencing of protein IB gene and pulsedfield gel electrophoresis. J Med Microbiol 1995;43: Xia M, Whittington WL, Holmes KK, Plummer FA, Roberts MC. Pulsedfield gel electrophoresis for genomic analysis of Neisseria gonorrhoeae. J Infect Dis 1995;171: Ng LK, Dillon JA. Typing by serovar, antibiogram, plasmid content, ribotyping, and isoenzyme typing to determine whether Neisseria gonorrhoeae isolates requiring proline, citrulline, and uracil for growth are clonal. J Clin Microbiol 1993;3: Li H, Dillon JAR. Utility of ribotyping, restriction endonuclease analysis and pulsed-field gel electrophoresis to discriminate between isolates of Neisseria gonorrhoeae of serovar IA-2 which require arginine, hypoxanthine or uracil for growth. J Med Microbiol 1995;43: Dasi MA, Camarena JJ, Ledesma E, et al. Random amplification of polymorphic DNA of penicillinase-producing Neisseria gonorrhoeae strains. Genitourin Med 1993;69: O Rourke M, Ison CA, Renton AM, Spratt BG. opa-typing: a high-resolution tool for studying the epidemiology of gonorrhoeae. Mol Microbiol 1995; 17: Hobbs MM, Alcorn TM, Davis RH, et al. Molecular typing of Neisseria

9 1660 Thompson et al. JID 2000;181 (May) gonorrhoeae causing repeated infections: evolution of porin during passage within a community. J Infect Dis 1999;179: Bash MC, Lesiak KB, Banks SD, Frasch CE. Analysis of Neisseria meningitidis class 3 outer membrane protein gene variable regions and type identification using genetic techniques. Infect Immun 1995;63: Carbonetti NH, Simnad VI, Seifert HS, So M, Sparling PF. Genetics of protein I of Neisseria gonorrhoeae: construction of hybrid porins. Proc Natl Acad Sci USA 1988;85: Gotschlich EC, Seiff ME, Blake MS, Koomey M. Porin protein of Neisseria gonorrhoeae: cloning and gene structure. Proc Natl Acad Sci USA 1987;84: Lau QC, Chow VT, Poh CL. Polymerase chain reaction and direct sequencing of Neisseria gonorrhoeae protein IB gene: partial nucleotide and amino acid sequence analysis of strains S4, S11, S48 (serovar IB4) and S34 (serovar IB5). Med Microbiol Immunol (Berl) 1993;182: Cooke SJ, Jolley K, Ison CA, Young H, Heckels JE. Naturally occurring isolates of Neisseria gonorrhoeae, which display anomalous serovar properties, express PIA/PIB hybrid porins, deletions in PIB or novel PIA molecules. FEMS Microbiol Lett 1998;162:75 82.