Failure to detect HLA-A*6802-restricted CD8⁺ T cells specific for Chlamydia trachomatis antigens in subjects from trachoma-endemic communities

Abstract

The role of HLA-A*6802-restricted CD8⁺ T cells in chlamydial disease was investigated in human ocular infections. Peptides with predicted binding motifs for HLA-A*6802 were synthesized using sequences based on chlamydial antigens, major outer membrane protein (MOMP), macrophage infectivity potentiator (MIP) and heat shock protein (hsp70). Peptides were pooled according to Chlamydia trachomatis protein type and serovar, and were tested in ⁵¹Cr-release cytotoxic T lymphocyte (CTL) and enzyme-linked immunospot (ELISPOT) assays, using peripheral blood mononuclear cells (PBMC) isolated from subjects living in trachoma-endemic communities in The Gambia. Significant CTL activity or interferon-gamma release was not detected in any of the subjects, suggesting either that HLA-A*6802 CD8⁺ T cells may not be important in ocular infections or that the peptides chosen did not represent epitopes.

Introduction

Chlamydia trachomatis is a Gram-negative bacterium, which is the leading cause of infectious blindness, and is also a major cause of genital tract infections world-wide [1,2]. In trachoma, clinical signs may range from a mild asymptomatic inflammation with five or more follicles visible on the tarsal conjunctiva, follicular trachoma, to an intense inflammatory response in which most of the tarsal plate is obscured by capillary congestion. This is mainly seen in young children (0–9 years old), and is usually associated with evidence of ocular C. trachomatis infections. Among older children and adults, conjunctival fibrosis often develops as the follicles resolve. Repeated ocular infection causes scarring of the conjunctiva and the scars are visible as white lines, bands, or sheets. If this scarring is severe, there is inversion of the upper eyelids and eyelashes such that they rub against the eyeball, which is known as trachomatous trichiasis. Eventually, blindness follows corneal abrasion by in-turned lashes. Severe disease and sequelae usually occur in a small proportion of subjects after persistent or repeated infection over many years. The development of sequelae from chlamydial infection is thought to be immunopathologically mediated due to inappropriate responses to chlamydial antigens. Trachoma is endemic in many areas of The Gambia, and although recent data showed that it is declining, it is still responsible for 5–6% of all blindness and the prevalence of active inflammatory trachoma among children aged 0–9 years was 4·9% [3].

Although Chlamydia remains sequestered from the host cell cytoplasm throughout its growth cycle, it has been demonstrated that cell-mediated immune responses are generated during C. trachomatis infections and it is thought that they contribute to protection from infection and resolution of disease. In both animal and human studies, CD8⁺ cytotoxic T lymphocytes (CTL) were shown to be able to lyse Chlamydia-infected cells in vitro [4–8]. CD8⁺ CTL can also produce cytokines, and it has been demonstrated in murine studies using MoAbs against interferon-gamma (IFN-γ) and IFN-γ knockout mice, that IFN-γ production by CTL is required for resolution of C. trachomatis infection [6,9].

Human leucocyte antigens (HLA) are known to play a central role in the control of immune responsiveness and HLA class I alleles present peptide antigens to T cells for immunological reaction. This is thought to account for the association of several HLA alleles with either resistance or susceptibility to disease. The idea that HLA class I-restricted cellular responses may play a role in the immunopathology of chlamydial infections is supported by the fact that the class I allele HLA-A31 has been associated with susceptibility to chlamydial pelvic inflammatory disease in genital infections [10]. Also, HLA-B27 has been associated with reactive arthritis [11], and the role of Chlamydia was emphasized by the association of HLA-B27 with Reiter's syndrome following genital chlamydial infections [12].

We conducted a case-control study designed to investigate genetic and immune factors involved in trachomatous scarring in trachoma-endemic communities in The Gambia. Typing of HLA class I and II alleles showed that no HLA type was associated with protection from trachomatous scarring [13]. We also found low-level CTL responses in HLA-B8 and -B35 children resolving current infection and adults without scarring trachoma, thus suggesting that if HLA class I-restricted CTL are involved in protection, they are not restricted to a few HLA class I alleles [14].

However, results from our case-control study showed that HLA-A28 was significantly more common in case patients with trachomatous scarring than among controls, and this was found to be due to the subtype HLA-A*6802. These results therefore suggested that HLA-A*6802-restricted responses may play a role in the pathogenesis of trachoma. As a follow up to that study, we decided to investigate whether Chlamydia-specific CTL responses are elicited in subjects having the HLA-A*6802 allele, using a reverse immunogenetics approach. In our study, we recruited both cases and controls since class I-restricted CTL responses may be responsible for protection or pathology.

Subjects and methods

Subjects selection and HLA typing

A case control study was carried out in trachoma-endemic communities in The Gambia in which cases with trachomatous scarring and controls matched for age, sex, village of residence and ethnic group were recruited. HLA typing for HLA-A28 and its subtype -A*6802 was carried out by polymerase chain reaction (PCR) on genomic DNA extracted from whole blood using a panel of sequence-specific primers based on a modification of the Oxford Phototyping method [15,16]. After typing, 10 subjects with trachomatous scarring which had progressed to trichiasis in nine instances and 10 unmatched control subjects with no clinical evidence of disease having the HLA-A*6802 alleles were selected for CTL studies. None of the subjects had active infection as determined by clinical signs at the time of blood collection. Four of the subjects were children (ages 10–15 years) and the others were adults (ages 19–75 years). Five laboratory staff not living in trachoma-endemic areas, one having HLA-A*6802 and four without were chosen as controls in the CTL experiments. For ELISPOT assays, frozen peripheral blood mononuclear cells (PBMC) from 11 HLA-A*6802 subjects recruited for the CTL studies (five cases with trachomatous scarring and six controls) were used.

Epitope prediction and peptide synthesis

Peptides were octamers or nonamers designed to cover sequences of C. trachomatis that matched motifs predicted to bind to HLA-A*6802 [17]. The predicted binding peptides had threonine (T) or valine (V) at position 2, and leucine (L) or valine (V) at position 9. Peptides were chosen from sequences of C. trachomatis antigens major outer membrane protein (MOMP; serovars A and B), macrophage infectivity potentiator (MIP; serovar B) and heat shock protein (hsp70; serovar E). Seventeen peptides were synthesized commercially by Genosys Biotechnologies Inc., Cambridge, UK (Table 1). The HLA-B8-restricted influenza (Flu) nucleoprotein peptide ³⁸⁰ELRSRYWAI³⁸⁸ was also commercially synthesized, and used for control experiments which tested whether the stimulation process used for generating CTL was working [18].

Table 1.

Sequences of peptides synthesized based on Chlamydia trachomatis antigens corresponding to the HLA-A*6802 predicted binding motifs

Antigen	Sequence	Amino acid position
MOMPA	PTTRDVAGL	88–96
MOMPA	FTNAAYMAL	129–129
MOMPA	PTIAGKGTV	324–332
MOMPA	VVSSAENEL	332–340
MOMPA	IVDADKYAV*	368–376
MOMPB	PTTTTGNAV	66–74
MOMPB	FTNAACMAL	97–105
MOMPB	PTIAGAGDV	300–308
MOMPB	IVDADKYAV*	344–352
MIPB	RTEDFSLDL	61–69
MIPB	VVKEGTGRV	140–148
MIPB	RVLYIHPDL	203–211
hsp70E	GTTNSCVSV	16–24
hsp70E	GTRTTPSIV	38–44
hsp70E	AVTNFEKTL	63–71
hsp70E	AVFDVEQKL	104–112
hsp70E	LTRAQFEHL	298–306

^*Amino acid residues for HLA-A*6802 are similar in both the MOMPA and MOMPB sequences.

CTL studies

PBMC were isolated from venous blood using Lymphoprep (Nycomed, Birmingham, UK). For some subjects having high PBMC counts, some of the cells were cryopreserved and stored in liquid nitrogen. Fresh PBMC were used for generating effectors for CTL assays, and were stimulated for 1 h with 100 μm of peptides pooled according to C. trachomatis antigen and serovar. PBMC were then cultured without washing in 24-well plates at 2 × 10⁶ cells/ml in 1 ml R10 medium, that is, RPMI 1640 (Sigma, St Louis, MO), supplemented with 10% heat-inactivated fetal bovine serum (FBS; Sigma), 100 U/ml penicillin and 100 μg/ml streptomycin (ICN Biomedicals, Aurora, Ohio, USA), 32 mm sodium bicarbonate (Sigma) and 33 mm HEPES ICN Biomedicals). Seventy-two hours later, Lymphocult T (Human Interleukin-2 growth factor; Biotest, Drieich, Germany) was added to a final concentration of 10 U/ml. Medium exchange with R10 medium containing 10 U/ml IL-2 was done weekly or when the medium changed to yellow. Standard 4 h incubation ⁵¹Chromium (⁵¹Cr) release assays were performed in triplicates on days 7 and 14 of culture, using autologous, HLA-A*6802-matched, or mismatched targets. Epstein–Barr virus-transformed B lymphoblastoid cell lines labelled with ⁵¹Cr (Amersham, Aylesbury, UK) were used for the autologous and HLA-matched targets. Targets from the HLA-A, B ‘negative’ mutant cell line CIR [19] were tested in parallel for each assay as a class I mismatch to confirm class I restriction. Targets were either pulsed with a pool of predicted A*6802 peptides identical to those used to stimulate effector cell population or unpulsed. The supernatants were harvested on glassfibre filter mats (Wallac Oy, Turku, Finland), followed by liquid scintillation counting in 1205 Beta plate scintillation counter (Wallac Oy). Background Cr release was always < 25%. Percentage of lysis was calculated from the formula, 100 × (E − M/T − M), where E is the experimental release, M is the release in the presence of R10 medium, and T is the release in the presence of 5% Triton X-100 detergent. Results were regarded as positive if the lysis of the peptide-pulsed target was > 10% above that of the unpulsed target in at least two separate assays.

ELISPOT assays

A modified ELISPOT assay to detect peptide-specific IFN-γ release was carried out using cryopreserved PBMC obtained from some of the subjects that had been recruited for the CTL studies [20,21]. Briefly, 96-well nitrocellulose plates were coated with a first-layer antibody to IFN-γ (Mabtech, Stockholm, Sweden), then PBMC were placed in the wells at different concentrations in duplicate wells. The concentrations used ranged from 3 × 10⁴ to 2 × 10⁵ depending on the number of viable cells available after resuscitation and 2 h incubation with R10 containing 10% rhIL-2. Cells were pulsed with either no peptide or HLA-A*6802 predicted peptides pools that had been tested in the CTL assays, at a final concentration of 20 μm, and the plate was incubated at 37°C, 5% CO₂ for 16 h. The cells were then washed off and the plate was developed using detector and conjugate antibodies, followed by chromogen. Spots were counted by eye independently by two observers in duplicate wells, and a consensus agreed on by the observers before the average was taken.

Positive control wells contained 10 μg/ml phytohaemagglutinin (PHA; Boehringer Mannheim Biochemica, Mannheim, Germany) to confirm that stimulated cells were able to produce IFN-γ. Negative controls in the ELISPOT assays were wells with PBMC but no peptide, or irrelevant peptide from infectious agents with which the donor was not known to be infected, in this case the HLA-A*6802 HIV glycosaminoglycan peptide (YVDRFFKTL, kindly donated by Dr L. Dorrell, MRC, Fajara). In some instances, wells containing HLA-B8 MOMPA peptides were set up in parallel to confirm HLA class I restriction, since none of the subjects was HLA-B8⁺. Control assays were also carried out using PBMC from a known HLA-B8⁺ subject who had previously demonstrated a strong response to the B8 Flu peptide in CTL assays.

Ethical approval

Informed consent was obtained from all individuals in the study. In the case of children and adolescents, the parents or guardians gave consent. Laboratory staff from the Medical Research Council laboratories, Fajara, and the London School of Hygiene and Tropical Medicine volunteered as normal controls. The study was approved by the ethical committee of the MRC Laboratories, Fajara and the Gambian government.

Results

CTL response to predicted HLA-A*6802 peptides

Peptide-specific CTL responses to chlamydial peptides predicted to bind HLA-A*6802 were assessed in 20 subjects (10 cases with trachomatous scarring and 10 controls) recruited from the case-control study and five laboratory controls. CTL responses were tested at E:T ratios ranging from 10:1 to 50:1 depending on the numbers of viable effectors that grew from each culture.

No CTL activity was detected to the pooled C. trachomatis peptides using either the autologous, HLA-A-matched or CIR targets in any of the subject groups in this study. To confirm that the restimulation method used was capable of demonstrating CTL activity, an experiment was set up for a laboratory control who is a known B8 Flu responder using the HLA-B8-restricted Flu-NP. CTL was elicited when antigen was presented by an autologous target (specific lysis = 32 at E:T ratio of 25:1), but not when it was presented by the CIR target (specific lysis = 3·2 at E:T ratio of 25:1).

Frequency of IFN-γ-producing cells assessed by ELISPOT assays

ELISPOT assays were carried out using frozen PBMC from 11 HLA-A*6802 subjects (five cases with trachomatous scarring and six controls). All the subjects were selected from among the case-control study group that had previously been tested for CTL activity using peptide-stimulated cultures of fresh PBMC and identical HLA-A*6802 predicted peptide pools.

In all cases, no response was elicited to any of the HLA-A*6802 peptide pools, as well as the class I-mismatched peptides and the negative controls. There was however, a strong response to the PHA in the positive control wells.

Control experiments were set up using PBMC from a known HLA-B8⁺ subject who had previously demonstrated a strong response to the B8 Flu peptide in CTL assays. Duplicate wells contained 2 × 10⁴−5 × 10⁵ cells/well. Peptides tested included B8 Flu, B8 MOMPA (irrelevant antigen stimulation), A3 Flu (class I restriction control), B35 MOMPA (irrelevant antigen, mismatched HLA type). In addition, a no peptide negative control, PHA and PPD wells were set up in parallel. Significant responses were observed in all the PHA and PPD wells, and as expected, the number of spot-forming cells increased as the number of cells increased. Significant response to the B8 Flu peptide was observed, but there was no significant response to any of the other peptides tested.

Discussion

The results from the CTL and ELISPOT studies do not support a role for HLA-A*6802 CD8⁺ T cells in ocular chlamydial disease protection or immunopathogenesis.

CTL activity was not found in any of the 20 HLA-A*6802 subjects, irrespective of their case-control status. Seventeen HLA-A*6802 predicted peptide sequences were selected from four chlamydial proteins and commercially synthesized. Although binding assays were not done for these peptides because T2 transfectant cells for HLA-A*6802 are not available, the sequences matched motifs predicted to bind HLA-A*6802 [17]. Experiments were repeated at two time points in most instances and the cultures were kept for 2 weeks to allow sufficient expansion of CTL in vitro. The experimental design employed was capable of demonstrating CTL activity as positive results were obtained when a B8-restricted Flu-NP peptide antigen was presented to PBMC from a known B8 Flu responder.

The subjects recruited in these studies were from trachoma-endemic communities and it is very likely that both cases and controls would have been exposed to Chlamydia since childhood, since 90% of these subjects had IgG antibody to one or more of C. trachomatis serovars tested. Although at the time of the recruitment subjects did not have active disease, we expect them to be in the ‘memory state’ with respect to the bacterial antigens. It is, however, possible that the frequency of antigen-specific CD8⁺ T cells is very low, and therefore expansion of effectors did not produce enough cells for lysis of the targets. CTL secrete IFN-γ, and the ELISPOT assay was designed to detect IFN-γ in the immediate vicinity of the cell from which they are derived. In influenza infection, this technique was shown to be highly sensitive, and in fact more sensitive than the ⁵¹Cr-release cytotoxicity assay for determining low numbers of peptide-specific CTL [20]. This is because unlike the ⁵¹Cr-release assay which can only detect those CTL precursors that proliferate on antigenic stimulation, the ELISPOT method is capable of detecting memory T cells if they are present in sufficient numbers. It is possible that some effectors lack proliferative potential and will therefore be detected by ELISPOT but not by ⁵¹Cr-release assays. The technique was used to determine whether peptide-specific CTL were present in our subjects, but were too few to have been detected by the ⁵¹Cr-release assay. The results of the ELISPOT assay were found to be in total agreement with CTL results, and did not demonstrate HLA-A*6802-restricted CD8⁺ cells to the putative peptide epitopes tested.

Several reasons may account for our failure to detect CD8⁺ T cells. It is possible that the results reflect the genuine picture and they do not exist at all. Therefore, the association of HLA-A*6802 with scarring trachoma is not due to the fact that A*6802-restricted CTL responses play a role in immunopathogenesis, but rather that protective CTL are not elicited in these subjects. Our previous study which demonstrated low-level CTL responses to hsp60 in HLA-B8 and -B35 subjects suggested that protective CTL against C. trachomatis could be elicited [14]. It could also be argued that the peptides selected do not represent chlamydial epitopes, although peptides from the chlamydial protein sequences known to be antigenic targets of humoral immunity were tested. MOMP is the most abundant protein expressed during C. trachomatis infection in vitro, is immunodominant in the humoral immune response, and is known to contain epitopic targets for neutralizing antibody [22]. Peptides were selected from the two serovars of C. trachomatis MOMPA and -B which have been shown to be predominant in trachoma-endemic communities in The Gambia [23]. MIP and hsp70 are also expressed during C. trachomatis infections and neutralizing antibodies have been found against them in vitro [24,25]. With the recent results of the Chlamydia genome project [26], new proteins have been identified which appear to be involved in the immune response, and even though they may not be dominant, functionally they may still be important. Quite recently, HLA class I-restricted CD8⁺ CTL responses were elicited to MOMP peptides in human genital infections with C. trachomatis, in HLA-A2 and -B51 subjects [8]. In these studies, the subjects had recent symptomatic genital infections with C. trachomatis which were confirmed by laboratory diagnosis, but in our study the subjects did not have clinical signs of active disease. It is possible that the memory CTL responses which we were aiming to detect in our experiments were present at low frequency in the peripheral circulation, unlike in Kim's study where the presence of active infection may have resulted in the massive expansion of effector CTL which would then be readily detected. Also, the paper by Kim used a different methodology for inducing CTL activity. CD8⁺ T cells were stimulated with IL-7 at the start of the experiment, and it has been shown in influenza and HIV CTL studies that IL-7 selectively expands the population of peptide-specific CTL and thereby enhances the sensitivity with which CTL can be detected [20]. It is also possible that since our peptides were pooled and not tested individually, there might have been competition or antagonism among the peptides in a pool accounting for our failure to detect T cell activity. Furthermore, we used the predicted peptide groove binding motif to screen potential peptide epitopes, and genuine epitopes can be missed by this method. An alternative way of selecting peptides that might be epitopes would be to synthesize overlapping peptides from one antigen, instead of focusing on one HLA allele as we have done in this study. Another explanation for our failure to detect A*6802-specific CTL in our study could be the fact that the HLA-A*6802 association with trachomatous scarring found previously, was due to chance or to linkage disequilibrium with another non-HLA immune response gene close to the MHC.

Although these experiments were not able to demonstrate HLA-A*6802 CD8⁺ T cell activity, more refined methods of stimulation and detection could be employed to search further. Rather than using peptide antigens, vaccinia recombinants or modified virus Ankara (MVA) recombinants containing whole protein antigens of Chlamydia could be used. The advantage of this approach is that it need not be restricted to certain HLA alleles, but would focus on the antigens that elicit immune responses so that efforts could be directed to identifying specific peptides and epitopes. Currently, a lot of effort is being made to design MVA recombinants since MVA does not replicate and has been shown to be an effective way of generating targeted CTL responses in HIV infections [17]. Identification of immune mechanisms involved in protective immunity and disease pathogenesis is necessary to facilitate development of vaccines to prevent chlamydial infections and disease sequelae.