Discovery of Human-Specific Immunodominant Chlamydia trachomatis B Cell Epitopes

Current serological assays for species-specific detection of anti-Chlamydia species antibodies suffer from well-known shortcomings in specificity and ease of use. Due to the high prevalences of both anti-C. trachomatis and anti-C. pneumoniae antibodies in human populations, species-specific serology is unreliable. Therefore, novel specific and simple assays for chlamydial serology are urgently needed. Conventional antigens are problematic due to extensive cross-reactivity within Chlamydia spp. Using accurate B cell epitope prediction and a robust peptide ELISA methodology developed in our laboratory, we identified immunodominant C. trachomatis B cell epitopes by screening performed with sera from C. trachomatis-infected women. We discovered 38 novel human host-dependent antigens from 20 immunodominant C. trachomatis proteins, in addition to confirming 10 host-independent mouse serum peptide antigens that had been identified previously. This extended set of highly specific C. trachomatis peptide antigens can be used in simple ELISA or multiplexed microarray formats and will provide high specificity and sensitivity to human C. trachomatis serodiagnosis.

cross-reactivity concerns after human exposure to these animal chlamydiae, while it also may be absent in certain C. trachomatis strains following the loss of the plasmid (37,38). Thus, assays that determine antibody responses against a wide spectrum of C. trachomatis-specific antigens are still needed to improve the specificity and sensitivity of C. trachomatis serology.
We previously identified Chlamydia species-specific immunodominant B cell epitopes using mouse hyperimmune sera generated by three high-dose intranasal inoculations of mice with live chlamydial organisms (39). These B cell epitopes were used as peptide antigens in ELISAs for detection of species-specific anti-Chlamydia antibodies. Reactivities of C. pecorum-specific peptide antigens were confirmed with sera from cattle, the natural host of C. pecorum (39). However, reactivity of C. trachomatis and C. pneumoniae peptide antigens has not been confirmed with sera of naturally infected humans.
In the present study, we first expanded our bank of Chlamydia species-specific mouse sera to all 11 chlamydial species and used it to identify additional immunodominant B cell epitopes. We then confirmed that the mouse-identified C. trachomatis B cell epitopes (referred to as "host-independent" epitopes) are similarly immunodominant in humans, using sera from young women actively infected with C. trachomatis, many of whom also had documented a prior chlamydial genital infection(s). Since antibody responses can be influenced by host-dependent expression of protein antigens in natural versus natural hosts (40), we also sought to identify "host-dependent" C. trachomatis B cell epitopes that might be immunodominant in human infection but fail to elicit antibodies in the heterologous murine host. Since C. pneumoniae is a common infectious pathogen in children and young adults, we also utilized the human sera to probe for host-independent and -dependent B cell epitopes specific for this respiratory pathogen. Using this approach, we have identified highly species-specific C. trachomatis and C. pneumoniae peptide antigens.

RESULTS
Mouse-reactive Chlamydia species-specific peptide antigens. Using monospeciesspecific mouse antiserum pools, species-specific reactivities of Chlamydia peptide antigens were first confirmed or identified for C. avium and C. gallinacea ( Fig. 1 and Table S1). The majority of peptide antigens (49 of 60; Fig. 1 and Table S1) reacted with high specificity only with homologous sera. Only 11 peptides (18%) showed crossreactivity with 1 or more of the 10 heterologous serum pools, in addition to reacting with the corresponding homologous serum pools ( Fig. 1 and Table S1). However, these cross-reactivities were observed among closely related chlamydial species, and the majority of the observed cross-reactivities were weak. These results clearly show that these peptide antigens can be used for Chlamydia species-specific detection of anti-Chlamydia antibodies.
Antibodies against Chlamydia spp. in women with C. trachomatis infections. To ascertain the suitability of the sera for identification of immunodominant C. trachomatis B cell epitopes, the set of 60 Chlamydia species peptide antigens ( Fig. 1) was tested with polyclonal anti-IgG and monoclonal anti-IgG1 and anti-IgG3 conjugates for reactivity with the C. trachomatis-positive serum pools (Fig. 2). All 10 C. trachomatis peptides showed consistently strong IgG, IgG1, and IgG3 reactivity with the C. trachomatispositive pool (Fig. 2). This result confirmed that both long-lived IgG1 and short-lived IgG3 were present in the C. trachomatis-positive serum pool, and this finding was in agreement with the recent C. trachomatis infection history of the study subjects. None of the C. trachomatis peptide antigens showed reactivity with the C. trachomatisnegative serum pool (Fig. 2), confirming the specificity of these antigens. Further evidence for the overall specificity of peptide antigens is the fact that all 10 peptides of C. muridarum, a chlamydial species closely related to C. trachomatis, showed no cross-reactivity with the anti-C. trachomatis sera.
C. pneumoniae peptides from non-OmpA source proteins (PmpD, IncA, CT529, and CT618) showed substantial reactivities with the C. trachomatis-positive serum pools, particularly for long-lived total IgG and IgG1 antibodies (Fig. 2). These data suggest that the study human sera, particularly the sera from 125 women with active C. trachomatis infection, also retained moderate amounts of anti-C. pneumoniae antibodies. Combined with the reactivity of only a single CT529 peptide with short-lived IgG3 antibodies, these results mainly represent past C. pneumoniae infections. Importantly, weak reac- tivity of only a few peptide antigens of C. abortus, C. felis, C. caviae, C. pecorum, C. pecorum, and C. suis suggests sporadic exposure of human hosts to Chlamydia spp. from nonhuman hosts (Fig. 2), given that these peptide antigens showed highly species-specific reactivity with mouse sera (Fig. 1). Reactivities of host-independent peptide antigens with anti-C. trachomatis MIF titer-ranked subpools. To determine anti-C. trachomatis antibody-dependent signal intensity, the set of 60 mouse-reactive Chlamydia species peptides was also tested for reactivity with the four human serum subpools ranked by the C. trachomatis MIF titers of the constituent individual sera (Fig. 3). The reactivities of 10 C. trachomatis peptide antigens correlated strongly with the C. trachomatis MIF titers of 4 C. trachomatispositive serum subpools (R ϭ 0.80; P Ͻ 10 Ϫ6 ) (Fig. 3). For instance, the serum subpool with the highest anti-C. trachomatis MIF titers (1:256 to 1:2,048) showed the highest reactivity with all 10 individual C. trachomatis-specific peptides (Fig. 3). Signals above background were observed for 3 C. pneumoniae peptide antigens and for 6 from 5 other chlamydial species (C. abortus, C. felis, C. suis, C. pecorum, and C. psittaci). As expected, these reactivities did not correlate with the anti-C. trachomatis MIF titers of the 4 subpools (P ϭ 0.78). Thus, the host-independent peptide antigens of C. tracho-   Fig. 2. All 10 C. trachomatis peptides showed a signal above background, and the signals correlated highly significantly with MIF titers (R ϭ 0.80; P Ͻ 10 Ϫ6 ). Three C. pneumoniae peptides and six peptides from other Chlamydia species showed a signal above background that did not correlate with MIF titers (P ϭ 0.78).
matis, but not those of other chlamydial species, showed anti-C. trachomatis MIF titer-dependent reactivity.
Overall, these results confirmed the specific reactivity of the C. trachomatis peptide antigens ( Fig. 1 to 3). Additionally, the presence of high-titer anti-C. trachomatis antibodies in the serum pools from infected women ( Fig. 2 and 3) indicated that these sera were suitable for discovering additional immunodominant C. trachomatis B cell epitopes that had not previously been recognized by mouse hyperimmune sera (39). Antibodies against C. pneumoniae were the second most prevalent antichlamydial antibodies in the human sera ( Fig. 2 and 3), in agreement with the high prevalence of C. pneumoniae infections in humans (1,2,(5)(6)(7). This opened the possibility to also identify, by the use of these sera, C. pneumoniae B cell epitopes that were not recognized by anti-C. pneumoniae sera from the heterologous murine host.
Identification of host-dependent B cell epitopes of C. trachomatis and C. pneumoniae. In the preceding study that identified the set of Chlamydia species-specific peptide antigens (39), many more predicted peptides did not react with hyperimmune mouse sera. Therefore, in the present study, we rescreened the C. trachomatis and C. pneumoniae peptides of the library of mouse nonreactive antigens with the human C. trachomatis-positive and -negative serum pools. Of the 271 C. trachomatis and 153 C. pneumoniae peptides tested, the set of 38 reactive C. trachomatis and 8 reactive C. pneumoniae peptides is shown in Fig. 4. These C. trachomatis peptides reacted, almost uniformly, with the C. trachomatis-positive serum pool but not with the C. trachomatis-negative pool, indicating high specificity (Fig. 4). As expected, mouse nonreactive C. pneumoniae peptide antigens also reacted with these human serum pools (Fig. 4).

Reactivities of host-dependent peptide antigens with anti-C. trachomatis MIF titer-ranked subpools.
To determine the C. trachomatis MIF titer-dependent reactivity of the 46 host-dependent peptide antigens (Fig. 4), they were tested with the four C. trachomatis MIF titer-ranked human serum subpools (Fig. 5). Similarly to the hostindependent peptides (Fig. 3), the C. trachomatis peptides, except for the Hsp60 and Pgp3 peptides, reacted most intensely with the subpool with the highest MIF titer (Fig. 5). The reactivities of C. trachomatis peptides also correlated strongly with the MIF titers of the serum subpools (R ϭ 0.79; P Ͻ 10 Ϫ6 ), except for Hsp60 and Pgp3 peptides. However, the reactivities of C. pneumoniae peptides did not correlate with the C. trachomatis MIF titers (P ϭ 0.61). Thus, host-dependent peptide antigens of C. trachomatis, but not of C. pneumoniae, showed anti-C. trachomatis MIF titer-dependent reactivity.
Ranking of candidate C. trachomatis and C. pneumoniae peptide antigens for serological assays. Table 1 presents the summary of the serological screening results for C. trachomatis and C. pneumoniae peptide antigens that may be suitable for development of species-specific serology. Ten human host-independent and 38 hostdependent peptide antigens of C. trachomatis were ranked by an overall reactivity score (Table 1; see also Table S2 in the supplemental material). Similarly, 10 human hostindependent and 8 host-dependent peptide antigens of C. pneumoniae were ranked (Table 1). Only 10 C. trachomatis peptides and 10 C. pneumoniae peptides had been previously identified as B cell epitopes by their reactivity with murine hyperimmune sera (39).
Sequence conservation and probability of peptide antigen cross-reactivity. The intended use of these antigens for species-specific detection of antichlamydial antibodies requires maximum divergence from all other chlamydial species but highest sequence conservation within the target species. By these criteria (Ͻ50% sequence identity [SeqID], probability of peptide cross-reactivity [P cross ] Յ 0.04), the vast majority of these peptide antigens were highly specific for C. trachomatis and were conserved within all C. trachomatis serovars but not in the remaining Chlamydia spp. (Table 1; see  also Table S2). The 30 C. trachomatis peptides from the following 16 proteins fit these criteria: CT442, IncE, PmpD, TarP, CT618, CT529, CT813, CT875, CT795, CT223, PmpC, IncG, LcrE, IncA, CT694, and CT226. The exceptions with a high likelihood of crossreactivity with other chlamydial species were two C. trachomatis Hsp60 peptides (74% to 85% SeqID) and two pGP3 peptides (40% to 80% SeqID). In addition, 9 peptides of 7 C. trachomatis proteins (OmpA, CT618, CT143, CT529, CT579, LcrE, CT875) had P cross values of 0.16 to 0.89 only with C. suis or C. muridarum (40% to 93% SeqID), which are closely related to C. trachomatis.

FIG 4
Host-dependent C. trachomatis and C. pneumoniae peptide antigens specifically reactive with human C. trachomatis-positive sera but not with immune mouse sera. A library of 271 C. trachomatis and 153 C. pneumoniae peptide antigens was screened with the C. trachomatis-positive and -negative human serum pools. These peptides had previously been nonreactive in screens performed with hyperimmune monospecies-specific anti-C. trachomatis and anti-C. pneumoniae mouse sera (39). A total of 100 peptides showed reactivity above background, and the top-ranked 38 C. trachomatis and 8 C. pneumoniae peptide antigens derived from immunodominant proteins are shown. Signal intensities are shown in Log 2 RLU bars as described for Fig. 2. indicated in Table S2, the OmpA peptides (variable domain 1 [VD1] to VD4) showed high levels of sequence polymorphism within 22 strains that represent the major clades of C. trachomatis. The highest levels of sequence polymorphisms (Table 1; see also  Table S2) were observed within the VD1 region, with only 64% average sequence identity across these 22 C. trachomatis strains (average P cross value, Յ0.18), followed by 74% identity for VD4 (P cross Յ 0.42), 76% for VD2 (P cross Յ 0.48), and 94% for VD3 (P cross Յ 0.89). Therefore, these reactive regions with high levels of sequence polymorphism have the potential for use in the design of peptide antigens that discriminate among antibodies against different C. trachomatis serovariants.
Reactivities of C. trachomatis-and C. pneumoniae-specific peptides with individual sera. To further characterize peptide antigens by testing with individual rather than pooled sera, the most highly ranked 38 C. trachomatis and 18 C. pneumoniae peptides (Table 1) were tested with a panel of 48 individual serum samples from women with C. trachomatis infection (Fig. 6). Results showed that the newly identified host-dependent epitopes had reaction strengths and frequencies similar to those seen with the host-independent epitopes identified previously with mouse sera (Fig. 6; see also trachomatis MIF titer-dependent IgG reactivity of C. trachomatis but not C. pneumoniae peptide antigens. Thirty-eight C. trachomatis and 8 C. pneumoniae peptide human host-dependent peptide antigens (Fig. 4) were probed with C. trachomatis-positive sera pooled by MIF titer (Fig. 3). Signals are shown in Log 2 RLU IgG signal bars on the scale as described for Fig. 2. The signals of the C. trachomatis peptides correlated strongly with anti-C. trachomatis MIF titers (R ϭ 0.79; P Ͻ 10 Ϫ6 ), but those of the C. pneumoniae peptides did not (P Ͼ 0.61).   Only the actual chlamydial sequence of peptide antigens is shown, without the N-terminal biotin or the serine-glycine-serine-glycine spacer that is attached to each peptide (39). e Reactivity (ϩ) or lack of reactivity (Ϫ) with C. trachomatis-specific or C. pneumoniae-specific hyperimmune mouse sera is shown (39).
f Average percent amino acid sequence identity with 22 strains representing all major clades of C. trachomatis (Table S2) is shown. Sequences with identities below 40% typically cannot be aligned correctly, and the probability of peptide cross-reactivity is less than 1% (39).
g Average percent amino acid sequence identity with 6 major strains of C. pneumoniae (Table S2) is shown.
h Maximum probability of cross-reactivity (39) with C. muridarum-or C. suis-specific sera based on sequence identity (Table S2) is shown.
i Maximum probability of cross-reactivity (39) with sera specific for the remaining 8 Chlamydia spp. (Table S2) is shown. Peptide Set-3

Correlation of C. trachomatis and C. pneumoniae peptide reactivities and species-specific MIF titers.
Reactivity of the 48 individual sera against C. trachomatis peptide antigens (peptide sets 1 to 3 in Fig. 6) correlated significantly with the anti-C. trachomatis MIF titers of the individual sera (R ϭ 0.56 to 0.33; P ϭ 10 Ϫ4 to 0.02; Table 2). However, the reactivities of C. trachomatis peptide set 4 (Hsp60, pGP3; Fig. 6) did not correlate with anti-C. trachomatis MIF titers (P ϭ 0.42; Table 2). As expected, the signal intensities of all of the sets of C. trachomatis peptides (sets 1 to 4) failed to correlate with anti-C. pneumoniae MIF titers. Similarly, signals of C. pneumoniae peptides (set 5) did not correlate with anti-C. trachomatis MIF titers (P ϭ 0.66). Surprisingly, the reactivities of C. pneumoniae peptides (set 5) did not correlate significantly with anti-C. pneumoniae MIF titers (P ϭ 0.39). This may have been due to strong human serum reactivities with the non-OmpA proteins of C. pneumoniae but weak reactivities with OmpA antigens (major component of MIF antigens). Overall, the results in Table 2 indicate that C. trachomatis, but not C. pneumoniae, peptide reactivities correlated with the respective MIF titers (Fig. 6). As expected, the ELISA signal intensities of the C. trachomatis peptides in set 1 (Fig. 6) showed strong correlation with other C. trachomatis peptides (set 2) (R ϭ 0.81; P ϭ 10 Ϫ6 ; Table 2). In contrast, the reactivities of the C. pneumoniae peptides (set 5) showed only weak correlation with the C. trachomatis peptides in set 1 (R ϭ 0.42; P ϭ 0.003) ( Table 2).
Reactivities of serovariant peptides from C. trachomatis OmpA variable domain 1 (VD1) to VD4. As shown in Fig. 6, the OmpA VD1 peptide of C. trachomatis serovar D/UW-3/CX showed IgG1 reactivity with about 44% of the individual sera (21 of 48 sera) and the VD2, VD3, and VD4 peptides with 38%, 50%, and 79% of the sera, respectively. Testing of 5 additional serovariant peptides from OmpA VD1 increased the frequency of individual serum IgG1 reactivity by 20% (65% with six OmpA VD1 serovariant peptides compared to 44% with a single serovar D VD1 peptide). The use of 3 additional serovariant peptides from OmpA VD4 increased the reactivity frequency by only 4% (83% for four VD4 serovariant peptides compared to 79% with one serovar D VD4 peptide). Reactivities of short-lived IgG3 were almost identical to those of long-lived IgG1, except for distinct differences in some individuals (Fig. 6). These exceptions suggest a potential to distinguish current/recent infections indicated by the presence of IgG3 antibodies from past infections indicated by the presence of IgG1 antibodies. Interestingly, the six serovariant peptides from C. trachomatis OmpA VD1 generated infrequent but distinct signals (Fig. 6). In contrast, the OmpA VD4 peptides showed frequent and common reactivities. Combined with sequence polymorphism data ( Table 1; see also Table S2), this result suggests that the use of several serovariant peptides from the highly polymorphic OmpA VD1 region may identify C. trachomatis serovarspecific antibodies and that assaying just a few variant peptides from OmpA VD3 and VD4 may allow detection of all anti-C. trachomatis antibodies.
C. pneumoniae OmpA and non-OmpA peptide reactivities. In contrast to the highly reactive C. trachomatis OmpA peptide antigens (Fig. 6), C. pneumoniae OmpA peptides 53 to 56 showed rare and weak reactivity that was much weaker and less frequent than the reactivity seen with dominant peptides 39 to 47 from other C. pneumoniae proteins. All four OmpA variable-domain peptides and the non-OmpA peptides (IncA, PmpD, PmpG/I, CT618, PmpG, Cpn0525, and YscC) (Fig. 6) were highly conserved across all strains of C. pneumoniae (95% to 100% sequence identity) (P cross , Յ0.91 to 0.95) ( Table 1; see also Table S2). Several of those C. pneumoniae non-OmpA peptides (IncA, PmpD, PmpG/I, and CT618; Fig. 6) showed frequent detection of anti-C. pneumoniae antibodies. Pearson's correlation coefficient (R) and P values were calculated in linear regression analyses. The correlations are based on the mean reactivity of a peptide set with each of 48 individual sera (Fig. 6). -, absence of significant correlation or self-comparison.

DISCUSSION
In this study, we first reconfirmed the specificities of 60 peptide antigens from the 11 Chlamydia spp. (39), using Chlamydia monospecies-specific hyperimmune mouse sera ( Fig. 1 and Table S1). Subsequent testing of these peptides with the human C. trachomatis-positive and -negative sera confirmed the C. trachomatis-specific reactivity of these peptides with human sera and established the mouse/human hostindependent nature of the chlamydial B cell epitopes (Fig. 1, 2, and 3). Next, by screening large numbers of additional C. trachomatis candidate peptide antigens with human sera (Table 1), we identified novel human host-dependent B cell epitope peptide antigens of C. trachomatis that had not previously reacted with mouse anti-C. trachomatis hyperimmune sera (Fig. 4 and 5). Additionally, using the previously identified mouse species-specific peptide antigens (39), we confirmed the presence of both anti-C. pneumoniae and anti-C. trachomatis antibodies in these human sera (Fig. 6). Again, by screening a large number of additional C. pneumoniae candidate peptide antigens with the human sera (Table 1), we identified novel human host-dependent C. pneumoniae B cell epitope peptide antigens (Fig. 4 and 5).
These experiments conclusively established the concept of human host-dependent antibody responses against certain immunodominant B cell epitopes of C. trachomatis and C. pneumoniae. Overall, the identification of many human host-dependent B cell epitopes from two natural human chlamydial pathogens confirmed the concept that the range of antibody responses to chlamydiae in the natural host species is wider than in the experimental murine host (40).
Compared to the frequent reactivities of C. trachomatis and C. pneumoniae peptide antigens (Fig. 2 to 5), only few antigens from C. abortus, C. felis, C. pecorum, and C. psittaci produced ELISA signals, and the signals that those few produced were very weak ( Fig. 2 and 3), suggesting sporadic exposure of the human hosts to those animal Chlamydia spp. Additionally, all 10 peptide antigens from C. muridarum were nonreactive with the study sera ( Fig. 2 and 3), despite the high anti-C. trachomatis antibody levels in the C. trachomatis-positive serum pool. This indicates that these C. muridarum peptide antigens do not cross-react with anti-C. trachomatis antibodies, despite minimal evolutionary divergence between these two chlamydial species. Similarly, reactivity of peptides of C. suis, another species closely related to C. trachomatis, was essentially absent compared to that seen with C. trachomatis ( Fig. 2 and 3). These results clearly indicate that these peptides are highly species-specific antigens (Fig. 2).
With the completion of this study, we report now an expanded set of peptide antigens that may be valuable in serodiagnosis of human chlamydial infections (Table 1). These peptide antigens are sufficiently divergent to eliminate virtually any cross-reactivity among antibodies against other Chlamydia spp. (see Table S2 in the supplemental material). Derived from the most strongly reactive regions of immunodominant proteins identified in the chlamydial proteome (40)(41)(42)(43), they produce very strong ELISA signals under conditions of detection with anti-human antibody conjugates (Fig. 6). The distinguishing characteristic of these peptide antigens is that they react highly specifically with their cognate antibodies and can therefore be used in assays for species-and even serovar-specific detection of antibodies against C. trachomatis and C. pneumoniae. This fills a distinctive void among Chlamydia antibody assays for effective species/type-specific detection of antichlamydial antibodies (13)(14)(15)(16)(17)(18)(19)(20)(44)(45)(46)(47)(48).
A clear absence of reactivity to certain peptides in individual serum samples ("hole in antibody repertoire"), but strong reactivity to others (Fig. 6), points to the molecular uncertainty principle of the antibody response. We attribute this to the stochastic "roll of the dice" nature of immunoglobulin gene recombination (49), which may have failed to create even a single B cell receptor molecule that bound to a B cell epitope, preventing epitope-specific B cell expansion and antibody affinity maturation. These data strongly argue that combined results from multiple peptide antigens, preferably from several proteins, will be required for accurate anti-C. trachomatis antibody detec-tion that approximates the quantitative accuracy of methods using complex multiepitope antigens.
C. trachomatis-seropositive individuals show holes in the antibody repertoire not only against individual peptide antigens but also against whole-protein antigens (40)(41)(42)(43)(50)(51)(52)(53). Thus, the molecular uncertainty principle of the antibody response applies also to proteins. The use of single antigens in serology therefore inherently compromises sensitivity as well as quantitative accuracy. These disadvantageous performance characteristics of the use of single protein or peptide antigens may inherently impede their suggested use in programs for control and monitoring for trachoma, sexually transmitted infections (STI), or community-acquired pneumonia (52)(53)(54)(55)(56).
Interestingly, C. pneumoniae OmpA peptides were strongly reactive with anti-C. pneumoniae mouse sera (Fig. 1), but not with human sera (Fig. 6), while C. trachomatis OmpA peptides reacted strongly with anti-C. trachomatis mouse sera as well as human sera ( Fig. 1 and 6). In agreement with these findings, Campbell et al. reported that rabbit anti-C. pneumoniae immune sera recognized C. pneumoniae OmpA poorly but that C. trachomatis immune sera dominantly recognized the respective OmpA proteins (57). These results may explain the finding that anti-C. trachomatis but not anti-C. pneumoniae MIF titers correlated significantly with the corresponding speciesspecific peptide reactivities (Table 2), since OmpA is the dominant MIF antigen. In agreement with the previous findings (29,57), these results suggest that the C. trachomatis but not the C. pneumoniae OmpA peptides would be suitable for sensitive antibody detection in serological assays.
Synthetic OmpA peptides have been tested before for Chlamydia species-specific serology, and OmpA peptide-based assays provided specificity but lacked sensitivity (58)(59)(60)(61). However, those studies used short, spacerless peptide antigens, resulting in low sensitivity that was most likely due to weak antibody binding to the short peptides (62) and/or to steric hindrance of antibody binding (63). Recently, several groups have attempted to identify non-OmpA peptide antigens for C. trachomatis and C. pneumoniae serology and to use levels of antibody against 1 to 3 of these peptides as markers for disease (64,65). However, those assays had low sensitivity, possibly because the peptides do not represent immunodominant B cell epitopes. Therefore, given that only a fraction of Chlamydia-infected individuals respond, those assays are inherently of limited use as disease markers. And even if the epitopes were immunodominant, use of multiple antigens is required for sensitive detection of a wide repertoire of anti-C. trachomatis antibodies, as results determined by us and others (40,41) indicate (Fig. 6).
These synthetic peptide antigens can be readily commercialized in the standard ELISA format as well as in the microarray format (63,66,67), and such assays may become widely available tools for chlamydial serology. In a recent study (66), we detected species-and serovar-specific antibodies against Chlamydia spp. by a peptide microarray that used several of the peptide antigens described here ( Fig. 1; see also Table 1). With the possibility of spotting 196 to 784 peptides on a single chip (67), such microarrays offer unprecedented opportunities for Chlamydia species serology by reducing the need for labor-intensive multipeptide ELISAs. Averaged microarray signals of multiple peptide antigens may reliably quantify the overall antichlamydial antibody response, and specific seroreactivity patterns of individual peptides may be associated with a risk of sequelae following chlamydial infection. In conclusion, these peptide antigens will improve C. trachomatis and C. pneumoniae serology and will thereby help to more accurately diagnose infections and sequelae and aid in C. trachomatis surveillance and control programs (30-34, 50, 59, 68).

MATERIALS AND METHODS
Mouse sera. Chlamydia monospecies-specific mouse sera for each of the 11 Chlamydia spp. were used to confirm the specificity of peptide antigens for detection of species-specific antibody. Nine of such hyperimmune sera had been raised in our previous study (39), but additional C. avium-and C. gallinaceaspecific sera were raised in the current study. Briefly, EB suspensions of C. avium strain 10DC88 and C. gallinacea strain 08-1274/3 were intranasally inoculated in A/J mice as described previously (39). Sera of 9 to 50 animals infected with C. avium or C. gallinacea were collected into a monospecies-specific serum pool. All animal experimental protocols were approved by the Institutional Animal Care and Use Committees at Auburn University.
Human sera. Sera were collected from 125 women with C. trachomatis infection confirmed by NAAT (69) and from 18 healthy, low-risk women who were never diagnosed with C. trachomatis infection. Chlamydia trachomatis MIF titers (A to I, K, and L1 to L3 antigens) were determined for 108 of the 125 C. trachomatis NAAT-positive women at the University of Washington, and 106 were also tested for C. pneumoniae (TW183 antigen) (21)(22)(23)(24)26). Serum samples from 18 healthy women at low risk for C. trachomatis exposure were tested for antibodies against C. trachomatis by an ELISA using a C. trachomatis EB antigen (GenWay Biotech, Inc., San Diego, CA). Seventeen anti-C. trachomatis EB antibody-negative sera were included in this study as C. trachomatis-negative-control sera, and a single positive serum sample was excluded. These samples from the 17 women at low risk for C. trachomatis infection were not tested by MIF. The chlamydial antibody status of all human sera was also characterized by the reactivity with mouse serum-reactive peptide antigens of all 11 chlamydial species.
Among the 125 C. trachomatis-positive women, 74 were African American, 28 were Caucasian, and 23 were Hispanic, Asian, or mixed race. The 17 C. trachomatis-negative healthy women were all Caucasian but were of similar age to the C. trachomatis-positive serum donors. The study protocol was approved by both the Institutional Review Boards for Human Research of the University of Pittsburgh and the University of North Carolina. All participants provided written informed consent at the time of enrollment.
Pooling of human sera. Aliquots of serum from the 125 women diagnosed with C. trachomatis were pooled and are referred to here as the C. trachomatis-positive pooled sera. Serum aliquots from the 17 women never diagnosed with C. trachomatis were similarly pooled and are referred to here as the C. trachomatis-negative pooled sera. The C. trachomatis MIF titer for 108 C. trachomatis-infected women was known at study onset, and these 108 samples were used to prepare four serum subpools ranked by C. trachomatis MIF titer. The serum subpools were prepared by mixing equal proportions of 19, 21, 33, and 35 serum samples from women with C. trachomatis MIF titers of 1:4 to 1:8, 1:16 to 1:32, 1:64 to 1:128, and 1:128 to 1:1,048, respectively. To identify additional peptide antigens that showed host-dependent reactivity with human sera, we used the human serum pool or subpools of 125 C. trachomatis-positive and 17 C. trachomatis-negative women to screen a large panel of 271 C. trachomatis and 153 C. pneumoniae peptides that had been nonreactive with mouse sera.
Peptide antigens. The B cell epitope prediction approach has been described in detail in preceding studies (39,62). In brief, we first identified and ranked 72 immunodominant proteins among all chlamydial proteomes on the basis of published data (40)(41)(42)(43). On the basis of alignments of each of these 72 individual proteins, suitable regions for identification of genus-, species-, and serovar-specific epitope candidates were further subjected to in silico B cell epitope analyses (62). The majority of the peptide antigens were selected from polymorphic protein regions for species-specific reactivity and smaller numbers from highly polymorphic regions (serovar-specific reactivity) or conserved regions (genusspecific reactivity). Conserved B cell epitopes that corresponded to a given Chlamydia species but were divergent from those of other Chlamydia spp. were selected for analyses of the species specificity of the peptide antigens. Divergent B cell epitopes within a Chlamydia species were selected for serovar-specific reactivity.
Peptide antigens were chemically synthesized with N-terminal biotin followed by a serine-glycineserine-glycine spacer and captured on streptavidin-coated white microtiter plates (Fisher Scientific, Roskilde, Denmark). For antigen optimization of some B cell epitope regions, they were scanned by the use of 16-to-40-mer peptides overlapping by 4 to 12 amino acids (aa), and peptides of different lengths corresponding to regions showing peak reactivity were tested to maximize reactivity.
Chemiluminescent detection was chosen because of the signal linearity over 4 to 5 orders of magnitude, and light emission was measured as relative light units (RLU) per second. Background signals were determined for wells without peptide antigen and wells coated with nonspecific peptide antigens (39). The differences between all background signals and all peptide signals were analyzed by one-tailed paired Student's t test. This analysis established that peptide signal correction by subtraction of the background mean value plus 3 standard deviations allowed unequivocal identification of reactive peptides. The polyclonal signal value was expressed as the base value ϫ 10 3 for comparative displays, since polyclonal anti-human Ig conjugates produced higher chemiluminescent signals than monoclonal Ig conjugates.

ACKNOWLEDGMENTS
This work was supported by grant 103 from the Auburn University Molecular Diagnostics Laboratory to K.S.R. and by National Institute of Allergy and Infectious Diseases grants U19 AI084024 and R01 AI119164.