Mucosal and systemic immune responses to plasmid protein pgp3 in patients with genital and ocular Chlamydia trachomatis infection

Abstract

The circulating and cervical B cell responses to Chlamydia trachomatis plasmid protein pgp3 were characterized in children and adults with ocular or genital chlamydial infection using the enzyme-linked immunospot assay (ELISPOT) and ELISA. No pgp3-specific ASCs were detected in healthy controls, but predominantly IgA ASCs were detected in UK adults with uncomplicated cervicitis or urethritis (P = 0·03, 0·019). In patients with extragenital complications or pelvic inflammatory disease a mixed response with more IgG and IgM ASCs was evident, suggesting a breach of mucosal immune compartmentalization with more extensive infection. In women with chlamydial cervicitis, ASCs secreting predominantly IgA, but also IgG, to pgp3 were present in cervix at presentation, with a frequency 30–50 times higher than blood. Cervical ASC numbers, especially IgG, fell markedly six weeks after antibiotic treatment. We detected principally IgA pgp3-specific antibody secreting cells (ASCs) in children resident in a Gambian endemic area, with a trend towards suppression of IgA responses during intense trachomatous inflammation (P = 0·06), as previously reported for other chlamydial antigens, and in keeping with the findings in genital disease. These data provide a rationale for further studies of immune responses to pgp3 in humans and animal models of chlamydia-induced disease, and its potential use in diagnostic assays and protective immunization strategies.

INTRODUCTION

The plasmid encoded protein pgp3 of Chlamydia trachomatis was initially identified by analysis of the 7·5 kb C. trachomatis common plasmid (pCT) which is thought to be present in the majority of strains and clinical isolates [1,2]. pgp3 has been demonstrated within chlamydial inclusions in infected cells by immunofluorescence [3] and there is evidence to suggest that it may be a membrane associated protein [3,4]. As such pgp3 would be a target for immune responses and therefore may be a useful antigen to induce protective immunity through immunization, or in diagnostic assays based on serology.

In fact, although the function of pgp3 remains unknown, immune responses to pgp3 have been demonstrated by serology in patients with genital chlamydial disease. In a study employing five recombinant antigens (pgp3, C. trachomatis major outer membrane protein – MOMP, outer membrane protein 2 – OMP2, Chlamydia specific LPS and heat shock protein 60 – hsp60) in serum ELISA, pgp3 was found to have the highest specificity (89%), positive predictive value and agreement with the other four antigens employed [5]. When combined with MOMP the assay resulted in 79% sensitivity and 82% specificity [5]. The high specificity of an immune response to pgp3 seen in that study confirmed previous findings by these authors using immunoblotting, microimmunofluorescence and ELISA [6]. We too found serum IgG pgp3 antibody responses in the majority of subjects who were seropositive for C. trachomatis by microimmunofluorescence, and had clinical evidence of genital tract infection; but not in healthy subjects, or subjects who had only serum Chlamydia pneumoniae antibodies [4]. Thus pgp3 appears to be an antigen specifically exposed to the immune system during human genital C. trachomatis infection.

Studies based on serum antibody have the problem of prolonged persistence of IgG after resolution of infection, and do not easily permit temporal analysis of transient immune responses during acute infections. In contrast, the enzyme-linked immunospot (ELISPOT) assay which detects spontaneous antibody secreting cells (ASCs) has the advantage of characterizing temporal humoral immune events. It has been shown in human and animal studies of infection and immunization that ASC responses are tightly regulated and occur only transiently after antigenic stimulation [7–10]. We have previously employed ELISPOT to characterize the immune responses to the membrane associated antigen MOMP, heat shock proteins and whole elementary bodies (EBs) of C. trachomatis in adults and children with ocular C. trachomatis infection (trachoma) [11]. We observed ASC and serological responses to all three antigens and a polarization of the ASC response during the most intense form of trachoma [11]. The purpose of the current study was to determine whether pgp3 responses occurred during ocular C. trachomatis infection (trachoma), and to characterize the nature of the response in both ocular and genital disease, both in the circulation and at the mucosa, during different clinical presentations.

MATERIALS AND METHODS

UK subjects

Study subjects consisted of men and women attending the department of Genito-Urinary Medicine St. George's Hospital with symptoms and signs suggestive of chlamydial genital infections. Genital infection with Neisseria gonorrhoea was excluded by Gram stain, microscopy and culture. Blood samples, urine and swabs were obtained from study subjects at presentation. Swabs (from the cervix in women and the urethra in men) were taken and processed at St. George's Hospital routine diagnostic laboratories using the Enzyme Immunoassay (EIA) kit (Microtrack II, Syva UK, Maidenhead, UK) and positive results were confirmed using Direct Immunofluorescence Assay (DIF) kit (Microtrack Syva UK). Separate swabs were taken from a subgroup of patients for analysis by polymerase chain reaction. The swabs were transported and stored in phosphate buffered saline (PBS) at − 70°C until used. When required for PCR testing the samples were thawed and vortexed. The solution was aspirated and centrifuged at 9500 × g for 30 min and DNA extracted from the pelleted cellular material. Chlamydial DNA was detected using the method and primers described previously [12]. All subjects received a standard seven day course of doxycycline. Follow-up blood, urine and genital swabs were obtained from a subgroup at 2 and 6 weeks after commencement of treatment. In a further group of women, cervical biopsies were taken at presentation and at six weeks follow up. All subjects provided written informed consent, and the study was approved by the St George's Ethical Committee.

Gambian subjects

Trachoma study subjects consisted of children and adults resident in the Gambian villages of Jali and Berending, an area endemic for trachoma as described previously [12,13]. The study was approved by the Gambia Government/Medical Research Council joint Ethics Committee. Subjects were examined and trachoma eye signs graded clinically according to WHO criteria by a trained ophthalmic assistant, and checked by an experienced observer (RLB). Sixty children (mean age 8·9 years) were divided into three groups: 19 with no signs of active ocular disease (NS), 36 with trachomatous inflammation of follicular grade (TF), and 5 with trachomatous inflammation of intense grade (TI). Seventeen age and sex matched pairs of adults with no signs of active trachoma (ANS) or the presence of trachomatous conjunctival scarring (ATS) were also studied. None of the adults with ATS had signs of active disease.

Preparation of cells and ELISPOT assay for ASC responses to pgp3

Peripheral blood mononuclear cells (PBMCs) were obtained by Ficoll-‘Hypaque’ (Sigma, Poole, UK) discontinuous gradient centrifugation of 5–10 ml heparinized whole blood obtained by antecubital fossa venesection. Cervical biopsies were obtained from cervical squamo-columnar junction by colposcopic punch biopsy, and transported in complete medium (CM) consisting of RPMI 1640 medium with 2 mM Glutamine, 0·01% sodium pyruvate, 100 IU Penicillin and 100 µg/ml streptomycin. The method described by Moscicki et al. [14] was used to separate mononuclear cells from tissue with some modification. The specimen was washed in CM and incubated in 1 mM DL-dithriothreitol (Sigma) for 20 min at room temperature and washed to rid the sample of excess mucus. The specimen was cut into fragments of 0·5 mm³ using sterile scalpel and incubated in 2 ml of CM with 0·2% collagenase (type III) and 0·04% Dnase (type I) at 37°C, agitating the sample with a pipette frequently until the tissue began to fray and fragment but no longer than 45 min The tissue was then sieved through a fine 0·1 µm plastic cell filter (Falcon) and spun at 500 × g for 7 min at 4°C. A single cervical biopsy yielded around 10⁷ cervical mononuclear cells (CMCs), >95% viable in each case. PBMCs and CMCs were resuspended and washed twice in phosphate buffered saline.

ELISPOT assay [15] as described previously [16] and adapted from [9] was used to detect antigen-specific antibody secreting cells (ASCs) in peripheral blood. pgp3 protein used in ELISPOT and ELISA was obtained as described in [4] as a recombinant stable 28 kD protein by over expression of ORF3 in Escherichia coli and purification of the product from periplasmic extracts under nondenaturing conditions. Briefly, polystyrene 25-well plates (Sterilin, Teddington, UK) were coated with chlamydial pgp3 antigen at a concentration of 10 µg/ml in carbonate buffer, pH 9·6, at 4°C for 12 h then blocked with 1% w/v Hammarsten casein (BDH, Poole, UK) in PBS pH 7·4. After washing, cells were counted in a haemocytometer chamber, resuspended in RPMI 1640 with no added serum or antibiotics, and 10⁶ cells per well incubated at 37°C with 5% CO₂ for 18 h. Cells were removed by vigorous washing with PBS/0·05%‘Tween-20’ and plates incubated for 2 h at 37°C with goat antihuman γ-chain or goat antihuman α-chain diluted 1 : 500, or goat antihuman µ-chain diluted 1 : 250. After washing in PBS-‘Tween-20’, wells received antigoat IgG-alkaline phosphatase conjugate diluted 1 : 250. All antibodies were from Sigma Chemical and diluted in 1% w/v casein/0·05%‘Tween-20’/PBS pH 7·4. IgA subclass-specific ASCs were detected using mouse antihuman IgA₁ or IgA₂ monoclonal antibodies (SeraLab, Crawley Down, UK) diluted 1 : 100, followed by rabbit antimouse IgG alkaline phosphatase conjugate (Sigma) diluted 1 : 250. Bound antibody was detected using 1 mg/ml 5-bromo-4-chloro-3-indolylphosphate (BCIP) substrate (Sigma) in 2% molten agarose/AMP buffer (Sigma). Spots representing one antibody secreting cell were enumerated after 24 h at room temperature by low power magnification. On polystyrene/polycarbonate plates each spot with the characteristic size and appearance of an ELISPOT is counted as significant, with control wells without antigen used to confirm specificity. Results are expressed as number of ASCs per 2 × 10⁶ PBMCs or CMCs assayed, and means are for the whole group including subjects with no detectable ASCs.

In a pilot study three groups of volunteers at St. George's Hospital were initially used in the development of the assay: 13 asymptomatic healthy UK subjects (mean age 34·0 years, range 25–49), 9 patients with active Salmonella infection (to identify possible cross reaction with gram negative lipopolysaccharide) and 6 patients with previous exposure to C. pneumoniae as evidenced by high serum levels of antibodies to C.pneumonia detected by MIF. No ASCs could be detected in any of these subjects, reconfirming the highly specific nature of the immune response to pgp3 [5].

ELISA assay for serum antibody responses to pgp3

The serum IgA and IgG antipgp3 specific antibody response was determined by direct enzyme linked immunosorbant assay (ELISA). Costar high-binding 96 well EIA plates were coated overnight at 4°C with 100 µl of pgp3 at a concentration of 6 µg/ml in carbonate buffer pH 9·6, then blocked with 200 µl of 1% casein in carbonate buffer pH 9·6. In subsequent stages wells received 100 µl with reagents diluted in 1% w/v casein/0·05%‘Tween-20’/PBS pH 7·4. All sera were assayed in duplicate in 4–8 doubling dilutions from 1 : 4, or 1 : 20 if high activity was detected. Each plate assayed included replicate dilutions of a reference serum with known activity. After 150 min at 37°C plates were washed and goat antihuman γ-chain-alkaline phosphatase conjugate or goat antihuman α-chain-alkaline phosphatase conjugate (Sigma) diluted 1 : 500 was added. After 150 min at 37°C plates were washed and 1 mg/ml p-nitrophenyl phosphate (Sigma) in 1 M diethanolamine/0·5 mM MgCl₂·6H₂0 buffer, pH 9·8. Plates were read at 405 nm when the reference samples reached an OD of 1·0. The antichlamydial activity of test sera, expressed in nominal IgG or IgA ELISA units, was determined relative to the reference serum on the same plate by parallel line regression analysis [17] as described previously [9]. Thus the activity of each test serum is relative to the IgG and IgA reference samples independently, and direct comparisons between IgG and IgA activity cannot be made. However, IgA levels are roughly 5–10 times less than IgG in this assay [9].

Statistical methods

Student's t-test (paired for age- and sex-matched adults with trachoma, unpaired for children with trachoma and adults with genital disease) was used to compare mean number of antigen- and isotype-specific ASCs between clinical groups, and paired t-test within groups. The chi-square test was used to compare numbers of subjects demonstrating an ASC response between groups, or Fisher's exact test where expected numbers were less than 5. Two tailed P-value less than 0·05 was taken as significant.

RESULTS

Immune responses to pgp3 in UK adults with genital C. trachomatis infection

Circulating ASC and serological responses in different clinical presentations.

Fifty subjects with uncomplicated genital infection (23 women with cervical chlamydial disease (CCD) and 27 men with nongonococal urethritis (NGU)), and 12 subjects were studied with chronic disease or extragenital complications (4 women with pelvic inflammatory disease (PID) and 3 men with extragenital disease in the form of Reiter's syndrome (RS), together with 5 men with chronic nongonococal urethritis (CNGU)). All had either detectable ASCs to one or more chlamydial antigens (pgp3, MOMP and EBs), or chlamydial antigen detected by PCR or DIF direct immunofluorescence in genital swabs at presentation. Subjects with no detectable response to any of the recombinant chlamydial antigens and negative PCR/DIF were excluded from the analysis, as this is associated with a high negative predictive value for chlamydial infection [5,6]. One hundred and thirteen subjects were recruited to the larger study of genital chalmydial infection, for which the pgp3 responses reported here was a subset. Of these, 70 had detectable ASCs to a chlamydial antigen, of which 32 were positive and 38 negative to either EIA, DIF or PCR of genital swab. Of the 43 without detectable ASCs to a chlamydial antigen, 11 were positive and 32 negative in the antigen detection assays. This expected discrepancy between antibody and antigen detection highlights the differential kinetics of organism shedding and the immune response. The distribution of positive and negative results between the different assays for the groups reported here is shown in Table 1.

Table 1.

Distribution of positive and negative results between study groups for each assay

Group	Number in group	PCR positive	EIA/DIF positive	ASC positive for any chlamydial antigen	ASC positive for pgp3	pgp3 ELISA positive
NGU	27	7	7	27	22	26
CCD	23	8	11	15	12	21
RS	3	0	0	3	3	3
PID	4	1	1	4	2	4
CNGU	5	0	1	5	5	5

The IgA antipgp3 ASC response was significantly higher than IgG and IgM in uncomplicated infections (NGU and CCD groups; P= 0·03 and 0·019, respectively), compatible with the localization of disease to a mucosal compartment (Fig. 1a). The IgA predominance was also significant within the CCD group alone (P = 0·05 and 0·03). In contrast, there were increased IgG and IgM ASC responses in the groups with extragenital complications (RS and PID) when compared to IgA. In the CNGU group ASC responses of all three isotypes were present but at low frequencies. Serum IgG and IgA pgp3 antibody was low or undetectable in the healthy controls (Fig. 1b), in keeping with the specificity of serological responses to pgp3 reported elsewhere [4–6]. Serum IgA responses in patients with active chlamydial infection were seen in all groups, the highest level in men with Reiter's syndrome. Serum IgG levels were highest in the PID group and lowest in the healthy controls and CNGU group. Although the serology broadly followed the trend of the ASC responses, none of the differences between the groups were significant. Unlike the ELISPOT it is not possible to directly compare the magnitude of the IgG and IgA responses.

Fig. 1.

ASC and serum antibody responses in genital infection. (a) mean pgp3-specific IgA (□), IgG (▪) and IgM () ASCs in blood per 2 × 10⁶ PBMCs for each clinical group. Figures above columns indicate number studied. *P≤ 0·05 for difference from IgA. Panel (b): mean antipgp3 IgA (□) and IgG (▪) in relative ELISA units for each clinical group, and healthy volunteers. Error bars indicate SEM. Clinical Groups: NGU, nongonococcal urethritis; CCD, cervical chlamydial disease; RS, Reiter's syndrome; PID, pelvic inflammatory disease; CNGU, chronic nongonococcal urethritis; HC, healthy controls.

Cervical ASC responses.

In a separate study IgA and IgG ASC in blood and uterine cervix tissue biopsies were measured simultaneously at presentation and six weeks later, after a 7-day course of doxycycline, in 10 women with chlamydial cervicitis confirmed by PCR. The initial cervical response (Fig. 2a) was around 30–50 times higher than the response measured in blood (Fig. 2b), and IgA responses in both compartments were higher than IgG. Responses in both compartments fell rapidly over time, but even after treatment IgA responses were still significant in the cervix at six weeks. IgM responses were not tested as the low number of CMCs extracted prompted us to focus on IgG and IgA.

Fig. 2.

Cervical and blood ASC responses in chlamydial cervicitis. Panel (a) mean pgp3-specific IgA (○) and IgG (•) ASCs in cervix per 2 × 10⁶ CMCs at presentation and six weeks later. Panel (b) mean pgp3-specific IgA (□) and IgG (▪) ASCs in blood per 2 × 10⁶ PBMCs at presentation and six weeks later. Error bars indicate SEM.

Immune responses to pgp3 in Gambian subjects with ocular infection

ASC responses in children.

ASCs secreting antipgp3 IgG and IgA antibody could be detected in children without signs of trachoma, and in those with mild and intense inflammatory disease (Fig. 3a). IgA was the predominant isotype, and the number of IgA ASCs to pgp3 and percentage of children manifesting an ASC response was lowest in the TI group. This pattern of response was identical to that observed in our previous study for MOMP and whole EBs [11]. However, in this study the difference in the magnitude of the IgG and IgA ASC responses did not quite reach statistical significance (P = 0·06 for TF group). There were no differences in the mean IgG ASC response or percentage of children with an IgG ASC response between the groups. IgM ASCs were tested in 20 children (16 with TF and 4 NS controls) but none were detected.

Fig. 3.

ASC and serum antibody responses in ocular infection. (a) Mean pgp3-specific IgA (□) and IgG (▪) ASCs in blood per 2 × 10⁶ PBMCs for each clinical group of children and adults. Values above columns are percentages of subjects in each group with a detectable ASC response to pgp3. (b) Anti-pgp3 IgA (□) and IgG (▪) in ELISA units for each clinical group of children and adults. Error bars indicate SEM. Clinical Groups: Children: NS, no ocular signs; TF, trachomatous inflammation, follicular grade; TI, trachomatous inflammation, intense grade; Adults: ANS, no ocular signs; ATS, trachomatous scarring.

ASC responses in adults.

Both groups of Gambian adults had a high percentage of subjects with detectable IgA ASCs, and the mean number of IgA ASCs against pgp3 was higher in the ATS scarred group than in the ANS group with normal eyes (Fig. 3a), but this difference was not statistically significant. Interestingly, only low levels of IgG ASCs were detected in adults, and there was no difference in the mean number between the two groups. These findings are as observed previously for MOMP [11]. IgM ASCs were not tested.

The serum IgA and IgG responses to pgp3 determined by ELISA are shown in Fig. 3b, and were as observed previously for MOMP [11]. Overall, the serum IgA and IgG response to pgp3 was higher in adults, but there were no significant differences between any of the groups.

DISCUSSION

Infection with Chlamydia trachomatis induces adaptive immune protection against rechallenge in human and murine models of infection (reviewed in [18]). From various murine models of gene-knockout and genital infection, IL-12 dependent, CD4+ TH1, IFN-γ mediated immunity is crucial in clearance of primary genital tract infection, complemented by B cell immunity in clearance of secondary infection [18]. Various vaccine strategies are being developed, including DNA vaccination [19], but antigenic variation in MOMP has led to disappointment with MOMP-based vaccines, and it is likely that multiple antigen vaccines, or even live attenuated bacteria may be required to induce optimum protective responses [18,20]. This has led to the search for other vaccine antigen candidates [20].

Although detection of C. trachomatis DNA in swabs or urine is the gold-standard for diagnosis, serology remains an important diagnostic tool in resource limited situations. Several studies have established pgp3 as immunologically active during human genital C. trachomatis infection [4–6]. Whether pCT, and specifically pgp3 has a specific role in human infections is still unclear, but a serological response to it appears to be highly specific to C. trachomatis. If pgp3 is actively presented to the immune system during human infection, it may be a suitable candidate for diagnosis or inducing protective immunity through immunization. In this study we report for the first time the response to pgp3 in ocular disease, and characterize the antibody isotype of the response in different clinical presentations of infection, and the mucosal B cell response in the cervix.

ELISPOT assays have the advantage over serology in that they give a picture of B-cellular events within a few weeks or days of antigen presentation, avoiding a ‘carry-over’ from more distant infections. Furthermore, mucosal derived B cells can be identified in the peripheral circulation during the brief period in which they traffic before homing back to mucosal sites. This enables the indirect assessment of immune events at inaccessible sites such as the genital tract. Finally by quantitating the frequency of ASCs of different isotype, a direct comparison between the isotypes can be made. We have shown previously that the ELISPOT can detect B cells trafficking in response to antigen presentation during human chlamydial ocular infection [11], and that there was a down-regulation of IgA ASCs associated with the form of disease characterized by high levels of TH1 cytokines [21]. There is generally little correlation between ASC responses and serum antibody during mucosal infection [7] or immunization [22]; and ASC responses tend to correlate best with mucosal secretion antibody levels [22] indicating a mucosal source and destination of these trafficking plasmablasts. This is further supported by the mucosal rather than systemic phenotype of plasmablasts induced by mucosal immune stimulation [23]. The ability to enumerate relative ratios of ASCs with a specific phenotype (e.g. class and subclass of secreted antibody, or homing molecule expression such as alpha 4 beta 7 integrin or L-selectin) has enabled the study of immune responses to infection or immunization in some detail, including in humans. This can be correlated with the known negative effect of TH1 and positive effect of TH2 cytokines on the IgA antibody response [24,25]: for example systemic immunization or oral immunization of humans correlated with a differential IgG and IgA ASC ratio and tissue homing receptor expression detected using ELISPOT [23]. A differential IgG ASC subclass profile and suppression of IgA ASCs has also been used to differentiate localized and chronic mucosal diseases such as gingivitis, and its association with pro-inflammatory cytokines [26]. Thus the application of single cell assays to characterize B and T cell interactions during different immune activation in man is a convenient mucosal immunology tool where inductive and effector sites may be inaccessible, that has been recently reviewed [27].

In this study we demonstrate that pgp3 is presented to the immune system during ocular infection, and that the number of ASCs and ratio of pgp3-specific IgA and IgG ASCs was very similar to those reported previously for the major chlamydial antigen MOMP [11]. Numbers of anti-MOMP IgM ASCs seen previously were very low [11], and we did not detect any antipgp3 IgM ASCs. That ASCs can be detected in children with apparently normal eyes is in keeping with our previous findings [11], and indicates the high frequency of infections in these endemic areas, or possibly nasopharyngeal carriage.

The IgA ASC response to MOMP was suppressed in children with intense disease [11] suggesting a polarization of the immune response to this intracellular pathogen to a T helper-1 (TH1) type, in keeping with reports of TH1 cytokines in the tears of patients with trachoma [21] and the crucial nature of TH1 responses in clearing infection [18]. We also observed a predominant antipgp3 IgA response in children with normal eyes and follicular grade trachoma, and reduced IgA ASC responses to pgp3 in children with intense disease. However the difference did not reach statistical significance, possibly due to the smaller number of subjects for whom pgp3 data was available.

The observation that adults with trachomatous scarring had a detectable IgA ASC response to pgp3, even though they did not have evidence of active ocular infection, is as reported previously for MOMP and hsp60 [11], and is in keeping with our observation that serum IgA antibody to pgp3 was high in both adult groups. There is evidence of persistent or recurrent infection in adults with trachomatous scarring [28], which may account for these findings, as with the children in the NS group.

We were unable to detect pgp3-specific ASCs in asymptomatic healthy UK subjects, or in patients with active enteric Salmonella gut infection (who could potentially have had responses to cross-reacting LPS), or subjects with high titre Chlamydia pneumoniae antibody. This confirms the apparent specificity of antipgp3 responses with C. trachomatis infection [4–6]. In contrast, we were able to detect ASCs in men and women with different clinical presentations of chlamydial genital infection. In uncomplicated chlamydial infection (cervicitis or NGU) IgA was the predominant isotype of ASCs secreting antibody to pgp3. This is in keeping with the mucosal nature of the response expected from these sites [29]. We were able to study the mucosal ASC response directly by cervical biopsy of women presenting with chlamydial cervicitis, confirmed by PCR. Although we observed both IgA and IgG ASCs secreting antibody to pgp3, IgA was the predominant isotype. Cervical ASCs were 30–50 times higher than in the blood at the same time, reflecting the plasma cell pool that have homed to this site. At follow-up six weeks after initiation of antibiotic treatment, all women were PCR negative for chlamydia. The IgG response in blood and cervix had almost completely disappeared, but while there appeared to be mechanisms operating in the cervix to promote and maintain the IgA response, the number of ASCs was much lower at six weeks, in keeping with the short-lived nature of plasma cells. Studies of normal human cervix by immunohistochemistry have observed higher numbers of IgA than IgG plasma cells [30], but in one study using ELISPOT analysis of cervix there were higher total numbers of IgG ASCs [31]. In that study plasma cells were excluded by Ficoll gradient separation. In our study Ficoll separation was not employed, and all antigen specific ASCs at all stages of maturity from plasmablasts to plasma cells were therefore included.

In contrast to the ‘mucosal’ response to cervical infection, with extragenital manifestations (such as Reiter's syndrome) or more extensive involvement of the female genital tract (as in PID), there was a marked change in the pattern of response, with relatively more IgG and IgM ASCs and fewer IgA ASCs. This may have arisen because the response breached mucosal compartmentalization and induced activation of systemic immunity, or from a shift towards a TH1 type response. The latter is analogous to the suppression of IgA ASCs during intense trachomatous inflammation. The induction of IgG may lead to complement binding and a more intense inflammatory response, which may account for the symptomatology and possibly the immunopathological sequelae associated with PID and Reiter's syndrome. The direct dissemination of C. trachomatis to systemic sites such as the joints, as reported in Reiter's [32] may also account for the activation of systemic responses. The patients with extragenital complications also had the highest IgA antibody levels to pgp3, comparable to that seen in children with TI and adults in the trachoma endemic region. These data exemplify the complex interaction between site of infection, chronicity of infection and possibly dose of infectious agent on the pattern of the immune response observed. In marked contrast, the patients with chronic NGU had very low levels of IgA antibody and few ASCs of all isotypes. This either suggests that active infection plays a minor role in this condition, that responses to pgp3 are not significantly involved in the immunopathology, or that B cell responses are down-regulated generally in this presentation.

As well as providing a further insight into the immune regulation of human chlamydial infection, these data prompt further that studies in humans and animal models of chlamydia-induced disease to elucidate the role of pgp3 as a potential diagnostic and vaccine antigen.