The major CD8 T cell effector memory subset in the normal and Chlamydia trachomatis-infected human endocervix is low in perforin

Consistent with our previous findings with cytobrush-retrieved endocervical T cells, immunohistological staining for CD3 and CD8 T cell infiltrates in six C. trachomatis- negative and four C. trachomatis-positive banked endocervical tissues demonstrated that the number of CD3⁺ and CD8⁺ T cells is significantly higher in C. trachomatis-positive than in C. trachomatis-negative endocervical tissues (p <0.01 for both comparisons) (Figure 1). Interestingly, when we stained serial sections of endocervical tissues for CD8 and perforin, we observed that the majority of CD8 positive cells were perforin negative (Figure 2). Quantitative analysis of perforin-expressing CD8 T cells, however, was limited by the immunohistological single staining technique’s inability to differentiate these cells from other immune cells that also express perforin, such as eosinophils, basophils, mast cells and NK cells. Thus, while our immunohistochemical data provided significant insights on the make-up of T cell infiltrates and suggested that CD8 T cells expressed limited amounts of perforin, we recognized that a multiparameter flow cytometric analysis would be needed to support and extend these observations.

To further analyze the endocervical immune cell repertoire, we performed multiparameter flow cytometric analyses to determine the mononuclear leukocyte types in isolated peripheral blood mononuclear cells (PBMC) and cytobrush-retrieved endocervical cells from 15 C. trachomatis- infected and 6 uninfected young women attending the Delgado Clinic. The gating strategy utilized in this study is shown in a representative analysis flow chart shown in Figure 3. Utilizing this strategy, we identified the lymphocyte population based on forward (FSC) and side-scatter (SSC) profile, from which the NK cell, CD4 T cell, and CD8 T cell subpopulations were delineated. We then assessed the perforin and granzyme expression in each of these cell types from paired samples obtained from the blood and endocervix. NK cells and CD4 T cells were utilized as cellular controls, while granzyme B was utilized as an intracellular staining control. Perforin and granzyme B expression of the different immune cells from the blood and endocervix of a representative C. trachomatis-negative and C. trachomatis-positive women is shown in Figure 4. Consistent with previous reports, we observed that perforin expression is relatively low in CD4 T cells, and high in NK cells [14, 15].

We then further delineated the CD8 T cell population into memory subsets based on the expression of CD45RA and CCR7 surface markers, as described by the groups of Sallusto and Lanzavecchia (Figure 5A). Representative distributions of CD8 T cells in blood and endocervix of C. trachomatis-negative and C. trachomatis-positive women are shown in Figure 5B. Analyses of blood and endocervical CD8 T cell subtype percent- distribution from 16 women using repeated ANOVA indicated a significant tissue site interaction (p<0.0001; with Greenhouse-Geisser correction). Further, a paired comparison of mean percentage CD8 T cell subtypes in blood and endocervix, using t- test with confirmatory analysis using the Wilcoxon Signed Rank test, indicated a significantly higher percentage of T_EM cells in the endocervix (p<0.0001), with a significantly lower percentage of naïve cells (p<0.0001). Taken together these data indicate that, while the different CD8 T cell subtypes are relatively distributed in the blood, the endocervix is primarily populated by effector memory (T_EM) CD8 T cells (Figure 5C and 5D).

Based on the observation that the endocervix is primarily populated by CD8 T_EM cells, we focused our subsequent analyses on this subpopulation. Here, the cytolytic potential of CD8 T_EM cells was assessed by analyzing flow cytometry data for the percentage of perforin expressing CD8 T_EM cells, using difference scores (blood – endocervix) and non-parametric analysis. Based on the Wilcoxon Rank Sum test, a significant difference in the percentage of perforin-expressing CD8 T_EM cells from blood and endocervix were observed in both C. trachomatis-negative (p<0.05) and in C. trachomatis-positive (p<0.0001) women as shown in Figures 6A and 6C respectively. Furthermore, pair-wise comparison of the percentage of perforin+ T_EM in the blood and endocervix of individual samples demonstrated that in both C. trachomatis-negative (Figure 6B) and C. trachomatis-positive patients (Figure 6D) there was a general trend for the endocervical T_EM to have lower percentage of perforin positive cells compared to T_EM derived from the blood. However, further analyses, with repeated measures ANOVA, using C. trachomatis infection status as a grouping factor, showed that although there were differences in blood vs. endocervix, no interaction with C. trachomatis-infection status was observed. These data show that despite the significant increase in CD8 T cells infiltrating the endocervix during C. trachomatis-infection, the relative low perforin level of the CD8 T T_EM cell population is still observed.

Our ex vivo data suggested that endocervical CD8 T T_EM cells have a low perforin content. Therefore, we investigated one of the factors that could influence this phenotype. We hypothesized that IFN gamma could be one of the mediating factors that drives the decrease in perforin content of endocervical CD8 T_EM cells based on the following observations from previous studies: 1) IFN gamma levels in the female genital tract are elevated during the secretory stage of menstruation [16] and during decidualization in successful pregnancies [17]; 2) Animal models of C. trachomatis infections have demonstrated that T cells with the capacity to secrete IFN gamma migrate to the site of Chlamydia infection [18]; 3) High levels of IFN gamma are found in the FGT of C. trachomatis-infected patients [19]; and 4) IFN gamma can induce the expression of indoleamine-2,3-dioxygenase (IDO) in epithelial cells [20]; and IDO, a tryptophan catabolic enzyme, could downmodulate perforin expression in CD8 T cells [21]. Therefore, based on these previous findings, we developed a PBL-HeLa 229 cervical epithelial cell co-culture model to investigate whether the exposure of epithelial cells to IFN gamma has an effect on perforin expression by CD8 T_EM cells. To do this, PBLs were cultured alone or with a monolayer of HeLa 229 cells and with or without 600 U/mL of IFN gamma (Figure 7A). After 38 hours in culture, the PBLs were then stained to identify the different CD8 T cell subpopulations and intracellular perforin and granzyme expression. The gating strategy to delineate the different CD8 T cell subpopulation is shown in Figure 7B. This gating allowed us to identify the CD8 T_EM cell subpopulation from the PBLs, and to assess the perforin expression of these cells in the absence or presence of IFN gamma.

We found that the level of perforin expression by the control CD8 T_EM cells from PBLs cultured in the absence of HeLa 229 cells did not shift in the presence of IFN gamma (Figure 7C – left panel). However, when we co-cultured PBLs with HeLa cells, the addition of IFN gamma in the culture medium resulted in the negative shift in the fluorescence intensity of perforin, suggesting a downmodulation of the expression of this protein (Figure 7C – right panel). Therefore, our in vitro data suggest that IFN gamma could be potential mediator of the downmodulation of perforin expression in CD8 T_EM cells.

Archived endocervical tissue paraffin blocks were utilized to assess the perforin and granzyme expression of endocervical CD8 T cell infiltrates in situ. Collection of endocervical tissues was approved by Louisiana State University Health Sciences Center Institutional Review Board. The endocervical tissues were derived from biopsies or discarded hysterectomy specimen. All specimens were screened for the presence of inclusions by chlamydial LPS staining. A confirmatory test of chlamydial positivity was performed on 3 of the 4 chlamydial-LPS positive tissues by ompA genotyping as previously described [30]. Endocervical tissues positive for C. trachomatis infection (n=4) and endocervical tissues negative for C. trachomatis infection (n=6) were analyzed by immunohistochemical staining for CD3, CD8 and perforin expression. In each experiment, sections of cell pellet consisting of peripheral blood mononuclear cells buffered with HeLa 229 cells were used as a positive control, while staining with isotype-matched irrelevant antibodies were used as negative controls. Immunohistochemical staining was undertaken by first cutting 4 μM sections of tissue which were immediately fixed on glass slides. Paraffin wax was then removed by baking the slides at 65°C for 1 hour, after which the slides were soaked in Antigen Retrieval solution (Biocare) and heated in a de-cloaking chamber (Biocare) for 20 minutes. The processed sections of tissues on the slides were then blocked using Background Sniper blocking reagent (Biocare) to prevent non-specific background staining for 1 hour at room temperature. Individual tissue sections were then reacted with the following primary antibodies: Anti-CD3 (Dakocytomation), anti-CD8 (Biocare), anti-perforin (Biolegend) or isotype-matched control (BD Biosciences) solution and incubated for 1 hour. After several washes with phosphate buffered saline (PBS), the tissue sections were incubated with appropriate secondary-antibodies conjugated to alkaline phosphatase. The slides were then washed with PBS to remove unbound antibodies, and Vulcan Fast Red chromogen system (Biocare) was added and allowed to develop. After a substrate reaction was observed in positive controls, the slides were soaked in distilled water and counterstained with hematoxylin. Tissues were then dehydrated in a series of methanol and fixed on the slide by covering with coverslip with permanent mounting medium, VectaMount (Vector Laboratories). The differences in number of CD3 and CD8 T cell infiltrates were assessed using Student’s t-test after checking the assumptions of normality and homogeneity of variance, followed by confirmatory analyses with Wilcoxon Rank Sum test.

To assess of the cytolytic potential of CD8 T cells in the blood and endocervix, we recruited participants into a study through the Delgado Clinic, New Orleans, Louisiana. Written informed consent as approved by the Louisiana State University Health Sciences Center Institutional Review Board was obtained from each participant. A total of 21 paired blood and endocervical cell samples were collected from women 18–30 years old attending the clinic who were enrolled based on satisfying at least one of the following criteria: (1) clinical evidence of cervicitis, (2) having a male partner who was tested positive for C. trachomatis infection (CT+) and (3) a positive test result for C. trachomatis infection by nucleic acid amplification test (NAAT) 3 to 30 days prior to enrolment. The participants provided information on demographics, history of previous sexually transmitted infections, antibiotic usage, and sexual behaviour. The participants were also examined by a nurse practitioner for mucupurulent cervical and vaginal discharge, bleeding and presence of clue cells. Urine samples were obtained to test for Neisseria gonorrhoea and C. trachomatis infection by NAAT. The C. trachomatis positive population were defined by a positive C. trachomatis NAAT result and confirmed with a C. trachomatis positive culture. Additional vaginal swabs were taken to assess for Trichomonas vaginalis, yeast infection, and bacterial vaginosis as previously described [15]. HIV positive patients were excluded from the study.

To analyze the immune cell repertoire in the endocervix of women enrolled in the study, parallel samples of whole blood (40 mL) and endocervical cells collected using two cytobrushes were performed as previously described. Peripheral blood mononuclear cells (PBMC) were isolated by conventional ficoll differential centrifugation, and endocervical cells were isolated from the cytobrushes following the previously established protocol in our laboratory [30].

To analyze the cytolytic potential of systemic and endocervical CD8 T cells, polychromatic (6-color) flow cytometry was performed. After processing the PBMC and endocervical cells, aliquots of these cells were made. One aliquot of cells was stained with fluorescently labeled antibodies against surface markers: CD56-APC (BD Pharmingen), CD3-PerCP Cy5.5 (BD Biosciences), CD4-Alexa Fluor 700 (Biolegend), CD8-APC-Cy7 (BD Pharmingen). Second aliquot for Isotype controls was stained with isotype-matched fluorescent antibodies: APC IgG1k Isotype (BD Biosciences), PerCP Cy5.5 IgGk isotype (BD Biosciences), Alexa Flour 700 IgG1k isotype (Biolegend), APC-Cy7 IgG1k Isotype (BD Biosciences). The cells were then permeabilized using Cytofix/Cytoperm (BD Pharmingen), a formulation of paraformaldehyde and saponin, to fix and permeabilize the cells. Following permeabilization, the cells were stained for intracellular proteins using labeled antibodies against Perforin-FITC (Abcam) and Granzyme B-PE (Cell Sciences). The second isotype aliquot was stained with FITC IgG2b Isotype (BD Biosciences) and PE IgG1 Isotype (BD Biosciences). CD4 T cells and NK cells were also analyzed to serve as cellular controls. Granzyme was utilized as intracellular staining control. Using the polychromatic flow cytometry approach, the different immune cells were simultaneously identified and cytolytic potentials measured. The percentage of cells positive for perforin was determined by gating the cells based on the threshold fluorescence of the FITC IgG2b Isotype control.

To further analyze CD8 T cell subpopulations, a third aliquot of cell preparation were stained for CD3-PerCP Cy5.5 and CD8-APC Cy7 in conjunction with CCR7-APC (R&D Systems) and CD45 RA-Alexa Fluor 700 (AbD Serotec), perforin-FITC (Abcam) and granzyme B-PE (CellSciences). This is to categorize the different subpopulations of CD8 T cells following the classification by Sallusto et al.[8, 9]. Further phenotyping of CD8 T cells allowed the identification of the different CD8 T cell subpopulation for a more extensive analysis of memory CD8 T cell population.

HeLa 229 cervical epithelial cells were obtained from the ATCC, and cultured in complete RPMI 1640 (cRPMI) culture medium with 10% human AB serum (Sigma); supplemented with 200 mM glutamine (Invitrogen) and 0.25% (wt/vol) glucose (Sigma). HeLa cells were seeded in a 12-well culture plate and cultivated for 24 to 36 hours until about 80% confluence was achieved. Peripheral blood mononuclear cells (PBMCs) were isolated from whole blood as described above. The isolated PBMCs were then placed in 6-well culture dish and incubated at 5% CO2 at 37°C for 4 hours to allow monocytes to adhere to the plastic. The non-adherent-PBLs were then collected, washed and counted using a heamocytometer. The PBLs were then placed on cell culture transwell insert on laid on top of a well on a 12 well plate without or with HeLa cells. The cells were cultured for 38 hours in the absence or presence of 600 U/mL IFN gamma in cRPMI. The cells were harvested, and the viability of the cells were assessed by trypan blue staining to confirm that ~95% of the cells are viable. The cells were then stained with monoclonal antibodies against CD3, CD8, CD45RA, CCR7, perforin and granzyme, and flow cytometric analyses were performed as described above.

Data were summarized using means (+ standard deviations), medians, and percentages as appropriate. Normality assumptions were assessed using the Shapiro-Wilks test, and where they were violated, parametric analyses were confirmed with analogous nonparametric methods. Differences between levels of CD3 and CD8 cells in C. trachomatis-positive and C. trachomatis-negative tissues were tested using Student’s t-test (Wilcoxon Rank Sum test). Paired blood and endocervical specimens were assessed usi`ng paired t-test (Wilcoxon Signed Rank test), and repeated measures ANOVA (Signed Rank and Rank Sum tests) was used to determine if blood versus endocervical differences varied with C. trachomatis infection status (interaction). SAS statistical software (version 9.2) was used for all analyses.