Surgery followed by Persistence of High-Grade Squamous Intraepithelial Lesions Is Associated with the Induction of a Dysfunctional HPV16-Specific T-Cell Response

Cervical cancer is preceded by well-defined stages of changes in the epithelium known as cervical intraepithelial neoplasia (CIN) or squamous intraepithelial lesions (SIL), which are caused by persistent infection with human papillomavirus (HPV). Changes that affect more than one-third of the epithelium are diagnosed as CIN 2 or 3 or high-grade SIL (HSIL). These HSIL have a high chance of progressing to cancer if left untreated (1). Most HSIL are associated with the presence of a high-risk HPV type, particularly HPV16 (2). The HPV genome encodes two oncoproteins, E6 and E7, which are constitutively expressed in high-grade cervical lesions and cancer because they are required for the onset and maintenance of the malignant cellular phenotype (3).

The key role of the adaptive cellular immune system in the protection against HPV-induced lesions is indicated by the high incidence of persistent HPV infections and subsequent HPV-related malignancies in immunosuppressed individuals (4) as well as by the fact that only a fraction of infected subjects develop progressing epithelial lesions or cancer (5). Because HPV proteins are foreign to the body, one would expect the immune system to mount a response against these antigens when expressed in the cervical epithelium. Indeed, HPV16 E6-, E7-, and E2-specific Th1- and Th2-type CD4⁺ T-cell responses were frequently detected in peripheral blood mononuclear cell (PBMC) cultures of healthy individuals (6–8) and both HPV16-specific CD4⁺ and CD8⁺ T cells are able to migrate on antigenic challenge, (9) showing that successful defense against HPV16 infection is commonly associated with the induction of a systemic effector T-cell response against these viral antigens.

The local microenvironment in a HSIL is associated with an increase in interleukin (IL)-10 production and a decrease in proinflammatory cytokines (10–13), which poses a harsh milieu for the immune system and is likely to affect both systemic and local immune responses. Indeed, patients with HSIL show evidence of nonspecific suppression of type 1 T-helper cell cytokine production (11, 14), but how this affects the priming and the character of a HPV-specific immune response is only slowly elucidated. Thus far, the presence of HPV-specific immunity has been studied either in several really small cohorts of patients with HSIL (7, 15–21) or in somewhat larger cohorts that focused on a highly restricted set of antigenic peptides (22) and/or studies using only one single immunologic variable (23–26) to decide whether a response is present. Unfortunately, huge differences between these studies in design and outcomes complicate the development of a unifying picture on how HPV-specific immunity and HSIL coevolve.

So-called therapeutic vaccines are being developed for the treatment of individuals who contracted a high-risk type of HPV and were unable to control the viral infection as shown by the presence of a HPV-induced (pre)malignant lesion (27–31). Several therapeutic vaccines have been tested in patients with cervical and noncervical high-grade genital lesions but with modest success (32–38). Because these therapeutic vaccines aim at reinstating an effective T-cell response against HPV16 E6 and/or E7, it will become very important not only to know the presence of preexisting HPV16-specific T-cell immunity in patients with HSIL but also to understand their functionality, as this may bear effect on vaccine efficacy (39, 40).

Here, we have studied the presence and function of spontaneously induced HPV16-specific T cells in a large group of patients with HPV16+ HSIL. HPV16-specific proliferative T-cell responses were detected in less than half of the patients and the majority of the responses were not associated with the production of IFN-γ or other proinflammatory cytokines. Notably, HPV16-specific regulatory T cells could be isolated from HSIL tissue. The induction of these HPV-specific T-cell responses is most likely to be the product of surgical treatment with recurrence or persistence of disease.

Patient inclusion and sample collection. Patients visiting the colposcopy clinic at the Department of Gynaecology of the Leiden University Medical Centre or the Haga Teaching Hospital in The Hague were recruited in the CIRCLE study, which investigates cellular immunity against HPV16+ cervical lesions. The study design was approved by the Medical Ethical Committees of the Leiden University Medical Center and the Haga Teaching Hospital. Patients were eligible for the current study if they had a histologically proven CIN at the time of diagnostic colposcopy or loop electrical excision procedure (LEEP). Notably, patients with chronic HSIL were more motivated to participate in this study; hence, this group of patients is somewhat overrepresented in our cohort. Informed consent was obtained from all patients. Blood (70 mL) was drawn on the day before LEEP. Serum was obtained and PBMC were isolated from heparinized blood samples by Ficoll (Sigma) density centrifugation for the analysis of HPV-specific T-cell reactivity. In several cases, tissue from the lesion was obtained for research purposes.

HPV typing. Patients with a HSIL were typed for HPV on paraffin-embedded sections of biopsies using three general HPV primer sets (CPI/II [1], MY 9/11[2], and GP 5+/6+ 3) followed by sequencing. Sequencing results were analyzed by the National Center for Biotechnology Information BLAST program. As controls, β-globin PCR and a blank sample were included (41–43).

Antigens. A set of 22 amino acid long peptides, overlapping by 12 amino acids and indicated by the first and last amino acids in the sequence of the E6 and E7 protein of HPV16 (e.g., E6.1-22 and the last peptides E6.137-158 and E7.77-98), were used for the screening of T-cell responses. The peptides were mixed into four pools of E6 peptides and two pools of E7 peptides (E6.1-E6.4, E7.1, and E7.2). These pools consisted of four 22-mer peptides. Notably, peptide pools E6.3 and E6.4 both contained peptide E6.111-132, whereas peptide pool E7.2 harbored the last 5 peptides of HPV16 E7. The peptides were synthesized and dissolved as described previously (44). Memory response mix consisted of tetanus toxoid (0.75 limus flocculentius/mL; Netherlands Vaccine Institute), sonicated Mycobacterium tuberculosis (5 μg/mL; kind gift from Dr. P. Klatser, Royal Tropical Institute), and Candida (0.015%; HAL Allergenen Lab). The response to memory response mix was used as positive control in the assays (8) to confirm the capacity of the antigen-presenting cells that are present in PBMC to process and present antigens to memory T cells.

Proliferative capacity of HPV16-specific T cells by lymphocyte stimulation test. The capacity of T cells to proliferate on stimulation with the antigen was determined by short-time proliferation assay as described earlier (6, 8). Briefly, freshly isolated PBMC (1.5 × 10⁵) were seeded into 8 replicate wells of a 96-well U-bottomed plate (Costar) to which the indicated peptide pools were added at a final concentration of 10 μg/mL. Medium without antigen served as background control and memory response mix was taken along as a positive control. The test was conducted in IMDM (BioWhittaker) containing 10% autologous serum. On day 6, supernatant was harvested for cytokine analysis; subsequently, the cells were pulsed with 0.5 μCi [³H]thymidine (Perkin-Elmer) per well and incubated for an additional 18 h. Then, the cells were harvested onto filters (Wallac) using the Micro-cell Skatron harvester (Skatron Instruments) and counted on the 1205 Betaplate counter (Wallac). The average and SD of the 8 medium-only control wells were calculated and the cutoff was defined as this average plus 3 × SD. The stimulation index was calculated as the average of tested 8 wells divided by the average of the medium control 8 wells. A positive proliferative response was defined as a stimulation index of at least 3 and the counts of at least 6 of the 8 wells must be above the cutoff value (7).

Cytokine analysis. The supernatants isolated on day 6 of the proliferation assay were subjected to a Th1/Th2 inflammation cytokine bead array kit (BD Biosciences). In this array, the levels of IFN-γ, tumor necrosis factor-α, IL-10, IL-5, IL-4, and IL-2 were determined. According to manufacturer's instructions, the proposed detection limit was 20 pg/mL. However, for IFN-γ the cutoff value was set to 100 pg/mL because the standard curve showed linearity starting at a concentration of 100 pg/mL. Positive antigen-specific cytokine production was defined as a cytokine concentration above the cutoff value and more than two times the concentration of the medium control (7).

Culture of CIN-infiltrating lymphocytes. CIN-infiltrating lymphocytes (CIL) were isolated and cultured as described previously (45). Briefly, CIL cultures were expanded using a mix of irradiated autologous EBV-transformed B-cell lines (B-LCL) and 5 μg/mL cognate peptide in IMDM supplemented with 10% human AB serum (PAA Laboratories), 10% T-cell growth factor (Zeptometrix), and 5 ng/mL recombinant human IL-15 (Peprotech).

Analysis of T-cell specificity. T-cell cultures (25,000-50,000 cells per well) were stimulated with autologous monocytes or irradiated autologous B-LCL pulsed with their cognate peptide (ID2 HPV16 E7.71-92; ID23 HPV16 E7.51-72; 5 μg/mL; ref. 46) and protein (10 μg/mL) in triplicate wells in a 3-day proliferation assay. After 48 h, the supernatant was harvested and stored at -20°C for cytokine analysis. Antigen-specific IFN γ and IL-10 production was measured by ELISA as described earlier (19).

Detection of CD4⁺CD25⁺Foxp3⁺ T cells. HPV16-specific CIL lines were stained 3 weeks after their last antigen-specific activation in vitro first for surface markers CD25 (anti-CD25 FITC; clone M-A251; BD Pharmingen) and CD4 (anti-CD4-APC; clone RPA-T4; BD Pharmingen) before the cells were fixed and permeabilized. Blocking was done with 2% normal rat serum followed by the addition of anti-human Foxp3 (PCH101; eBiosciences) antibody or rat isotype IgG2a control. Then, the cells were washed and analyzed by flow cytometry. A previously isolated HPV16-specific CD4⁺CD25⁺Foxp3⁺ regulatory T-cell clone (C148.31) was used as a positive control and a HPV16-specific CD4⁺CD25⁺Foxp3- T-cell clone (C271.9; ref. 46) was used as a negative control. The fluorescence intensity of these two control clones was used to set the gates for the other samples in which the CD25⁺Foxp3⁺ expression of the stimulated polyclonal T-cell populations was analyzed.

HPV16-specific T-cell suppression assay. T-cell suppression assays were done as described previously (46). Briefly, the CIL lines were cocultured with allogenic CD4⁺CD25⁻ responder cells in the presence of 1 μg/mL agonistic anti-CD3 (OKT-3; Ortho Biotech) and APC mixture of five different B-LCL cell lines. Suppression of the responder cells was analyzed on proliferation and IFN-γ production as described previously (46). HPV16 antigen-dependent suppression was measured using a flow cytometry-based proliferation assay. Responder cells were labeled with CFSE and cocultured with PKH-26-labeled CIL lines at a 1:1 ratio. The responder cells were stimulated with a pool of five allogenic B-LCL; the HPV-specific CIL lines were stimulated with 5 μg/mL cognate peptide and autologous B-LCL in the presence of IL-2 (300 IU/mL). After 4 days of culture, the allospecific proliferation of responder T cells was analyzed by flow cytometry. HPV-specific CIL lines were treated with 50 μg/mL mitomycin C (Kyowa) for 1 h followed by irradiation (2,000 rad) to prevent proliferation but not effector function.

Statistical analyses. To evaluate the effect of a previous treatment for high-grade CIN and persistence of the lesion afterwards on HPV-specific immunity, the patients were divided into two groups. The group of patients with a persistent lesion after treatment consisted of patients who had already undergone a surgical treatment for a HSIL after which the lesion persisted at least for 8 months (range, 8-72 months) as indicated by the detection of Pap 3a or higher in follow-up smears or a HSIL at follow-up colposcopy for which these patients all had to undergo a second surgical treatment at the time that blood was drawn for the detection of HPV-specific immunity. Patients without a persistent infection where defined as patients who had no prior treatment before the drawing of blood for the immunologic assay. These groups were then subdivided into patients who did show a HPV-specific immune response or in whom no specific immune response was detected and analyzed by a two-sided Fisher's exact test.

Patients and HPV distribution. During a period of 5 years, a total of 74 patients with a HSIL were included in this study, 16 patients who were diagnosed with a CIN 2 and 58 patients with a CIN 3 (Table 1). The median age was 38 years (range, 24-68 years). HPV typing revealed that 60 (81%) patients were HPV16+ and 2 (3%) were HPV18+. In 11 (15%) patients, another HPV type was found (HPV45, HPV33, and HPV31) and 1 patient was HPV- (Table 1).

Patients with HSIL fail to induce a strong HPV16 E6- and E7-specific T-cell response. From 38 HSIL patients (Table 2), freshly isolated PBMC were stimulated with peptides derived from HPV16 proteins E6 and E7 as well as with a mix of common recall antigens (memory response mix) in a short-term proliferation assay (Fig. 1A). We have shown previously that this assay is geared toward the detection of CD4⁺ T-cell responses (7, 8, 14, 22). HPV16-specific T-cell responses were detected in 15 (39%) of the 38 patients (Table 3). In 6 cases, proliferation was detected against E6, in 3 cases to E7, and in 6 cases to both E6 and E7 (Table 3). Analysis of the supernatants of these T-cell cultures for the presence of types 1 and 2 cytokines revealed the secretion of the Th1 cytokine IFN-γ in 6 of 15 (40%) patients with a proliferative T-cell response (Fig. 1B; Table 3). Occasionally, low levels of tumor necrosis factor-α and IL-5 were produced by the HPV16-specific responding cells. Of all the proliferative responses measured (n = 31) in these patients, only 9 (<30%) were associated with the production of a proinflammatory cytokine (Table 3).

HPV-specific T-cell responses are correlated with the persistence/recurrence of HSIL. The majority of HSIL lesions are precancers that are destined to persist (1) and for that reason are surgically removed. As a small percentage of HPV16+ HSIL may spontaneously regress after a biopsy (47), we hypothesized that the combination of HPV antigens and an invasive (surgical) treatment may deliver sufficient antigenic stimulation and danger signals to activate a HPV16-specific T-cell response. Because several of our patients (Table 2) had already been surgically treated in the near past for HSIL, the patients were divided into two groups and analyzed with respect to the absence and presence of HPV16-specific immunity. Group 1 (n = 28) consisted of those patients who were treated for the first time for a HSIL at the time of testing for HPV-specific immunity, whereas group 2 (n = 10) consisted of patients with HSIL that persisted/recurred after a first destructive treatment and who were treated again at the time that their HPV-specific immune status was assessed. The latter group displayed a significantly higher response rate to HPV16. The percentage of positive systemic responses in the group with persistent/recurrent lesions (group 2) was 90% versus 21% in group 1 (P = 0.0002; Fig. 2). Altogether, our results suggest that a systemic immune response, most often of a non-Th1/Th2 type, develops against HPV16 E6 and/or E7 and is often induced if the lesion persists or recurs after initial surgical treatment of HSIL. For a high number of patients, clinical follow-up data were available (Table 2). The potential clinical effect of the immunologic response was analyzed by dividing the subjects according to treatment and subdivision according to immune status and either (a) the presence/absence of a Pap 3b or CIN 2-3 score once within 12 months or, more stringently, (b) the presence/absence of a Pap 3a score twice within 12 months. Neither this nor grouping only according to immune status and clinical status in follow-up revealed any significant relationship.

HPV16 E7-specific regulatory T cells infiltrate cervical HSIL lesions. From 7 patients, we were able to receive a small piece of cervical tissue for research purposes to characterize HPV16-specific T cells present in the HSIL lesion. As part of a larger study, we established CIL cultures from all 7 cases (45). In 3 of these cultures, T cells responding to HPV16 E7 were detected. The percentages of CD3⁺ T cells in CIL after 3 weeks of culture did not differ between HPV16-specific and nonspecific CIL (54 ± 16% versus 52 ± 31%, respectively). From two of these CIL cultures (ID2 and ID23), sufficient numbers of T cells were obtained for a more in-depth analysis. On stimulation with their cognate peptide, both CIL cultures produced IFN-γ (Fig. 3A). In addition, the culture of ID2 produced low levels of IL-10 and a small population of the CD4⁺ CIL population coexpressed CD25 and FoxP3 as measured by flow cytometry (Fig. 3A and B). As we had observed that the presence of HPV16-specific T cells with such a phenotype in cervical carcinoma represented regulatory T cells with the capacity to suppress immune responses (46), we assessed the suppressive capacity of the HPV-specific CIL cultures in a classic suppression assay (Fig. 3C). The CIL culture of patient ID2 was able to suppress both proliferation and IFN-γ production of CD4⁺CD25⁻ responder T cells, whereas the CIL culture of patient ID23 did not contain such suppressive capacity. To prove that the HPV16-specific T cells were responsible for this suppressive effect, a CFSE-based suppression assay was done in which the suppressive action of the CIL culture in the presence and absence of HPV16 peptide was tested (46). As expected, the CIL culture of patient ID2 almost completely suppressed the proliferation of the responder cells (85% reduction in proliferation) dependent on stimulation with HPV peptide, whereas the CIL culture of patient ID23 did not show substantial suppression (32% reduction in proliferation of responder cells) when stimulated with its cognate HPV peptide (Fig. 3D). These data show that HPV-specific regulatory T cells not only are present in patients with cervical carcinoma but also can already develop earlier during the malignant transformation of a persistently HPV16-infected cervix.

In this study, we show that a systemic proliferative T-cell response against HPV16 is often absent in the majority of patients who visit the clinic for treatment of a HPV16+ HSIL lesion. In several cases, HPV16-specific T-cell reactivity can be detected in the form of proliferative responses, which are not associated with HPV16-specific secretion of the proinflammatory Th1 or Th2 signature cytokines. Importantly, HPV16-specific T-cell reactivity is predominantly detected in patients returning to the clinic for repetitive treatment of a HPV16+ HSIL lesion because of persistence or recurrence of the lesion after initial destructive treatment. This was not the case in patients visiting the clinic for a first treatment of their HPV16 HSIL lesion (Fig. 2). The observation that these responses lack a clear proinflammatory signature indicates that this type of immune activation should not be regarded as beneficial. These responses rather are a reflection of the fact that most of the subjects, although in principle able to mount a HPV16-specific T-cell response, fail to develop a cellular immune response that is associated with protection against HPV-induced lesions (7, 18, 22, 25). This notion is sustained by our data, which show, for the first time, that a population of HPV16 E7-specific regulatory T cells can arise and infiltrate a HPV16+ premalignant lesion (Fig. 3).

In a previous study, in which the immune response to HPV16 E6 and E7 was studied in patients with cervical carcinoma, we were able to detect a HPV16-specific proliferative response in about half of all patients but only in 1 of the 8 tested subjects with a HPV16+ HSIL (7). Similar observations regarding the low frequency of responders were made in small cohort studies done by others (18, 22). The reason why fewer patients with HSIL mounted an immune response to E6 and E7 in comparison with cervical cancer patients at that time remained unclear. Our current study of a relatively large group of 38 HPV16+ HSIL patients confirms that the majority of HSIL patients fail to mount a HPV16 E6/E7-specific T-cell response. The absence of a HPV16-specific immune response in patients with HSIL and the presence of such responses in about half of the patients with cancer previously led us to hypothesize that the long-term presence of the HPV16 E6 and E7 antigens in a developing tumor may eventually trigger the induction of a CD4⁺ T-cell response (7). Here, we identified a larger group of HPV16+ HSIL patients, which was able to mount a HPV16 E6/E7-specific immune response. Importantly, most patients within this group suffered from a persistent/recurrent HSIL lesion and had already been treated by destructive treatment before this immunologic analysis was done. This suggests that the HPV16 E6/E7-specific immune response detectable in the circulation of cervical cancer patients has developed as part of long-term exposure to the HPV16 antigens in the persistent/recurrent HSIL in combination with the danger signals delivered by the previous invasive treatment and is less likely to develop in patients who have failed to induce a detectable response before their treatment and show no sign(s) of a persistent/recurrent HSIL lesion afterwards. This notion is sustained by our observation that the cytokine profile of the HPV16-specific immune response in persistent/recurrent HSIL patients is similar to that of what we observed in cervical cancer patients (7) as well as the fact that the majority of established high-grade CIN lesions will evolve toward cervical carcinoma when left untreated (1). As such, the group of patients with recurrent/persistent HSIL may reflect our “missing link” with respect to the absence of HPV16 E6/E7-specific immune responses in first time diagnosed HSIL patients and the presence of these responses in patients with cervical cancer. As yet, it is unclear whether either the previously given destructive treatment, the persistence of the lesion, and as such HPV antigens, or the combination of both is responsible for the activation of the HPV16-specific proliferative response. Notably, a previous study in HSIL patients showed that activation of HPV16 E7-specific immunity shortly after a local invasive procedure occurred only in 2 of 18 patients (26). In that study, immunity was measured by IFN-γ ELISPOT, whereas we showed that most of the HPV16-specific proliferative responses detected in our group of patients were not associated with the production of IFN-γ (Fig. 1; Table 3). The former study thus may have underestimated the response rate after local invasive procedures. Interestingly, a history of a long period (15-51 months) of persistent HPV16 infection is also associated with the activation of HPV16 E7-specific immunity, albeit that these responses are weak considering the fact that an indirect measurement consisting of a highly sensitive cellular bioassay for IL-2 production was needed to detect these responses (23). Taken together, these studies build a strong case supporting the idea that local danger signals and long-term exposure to sufficient amounts of antigen are key in the development of the weak and dysfunctional immune responses observed in HSIL patients.

The number of circulating, regulatory T cells as defined by CD4⁺CD25^high T cells (48) or CD4⁺CD25^high CTLA4⁺ T cells (49) is increased in patients with HSIL when compared with healthy controls. In addition, immunohistochemical analysis of HSIL lesions revealed not only that HSIL lesions represent immunosuppressive environments (13) but also that immune cells possessing a suppressive phenotype, as defined by CD25⁺TGF-β⁺ and CD4⁺TGF-β⁺ immune cells, may infiltrate such HSIL lesions (12). However, the specificity of these circulating and HSIL-infiltrating regulatory T cells was never determined. Previously, we showed that the CD4⁺ subset of T cells infiltrating cervical carcinomas consisted of HPV16-specific regulatory T cells able to suppress proliferation and cytokine production of responder cells (46). Similar functional analyses of HPV16-specific T-cell populations infiltrating HSIL showed not only that indeed fully functional regulatory T cells can infiltrate a premalignant cervical lesion but also that these regulatory T cells exerted their action on recognition of their cognate HPV16-specific E7 antigen (Fig. 3). Of note, because this was found in only one of the two patients tested, follow-up studies are required to determine the significance of this observation. Interestingly, a higher number of circulating CD4⁺CD25^high CTLA4⁺ regulatory T cells coincides with the presence of HPV16-specific T cells in the blood of patients with HSIL (49), indicating that such responses may coevolve. Recently, we have developed an assay to measure the percentage of HPV16-specific regulatory T cells in the circulation of patients with cervical cancer (27). Similar analyses in patients with HSIL may reveal to which extent the detected HPV16-specific responses lacking a clear-cut proinflammatory signature (Fig. 2; Table 3) may actually represent HPV16-specific regulatory T cells. The lack of PBMC precluded such an analysis in the current group of patients, but this question will be addressed in a new study.

The detection of HPV16-specific T cells with a non-Th1/Th2 cytokine and even immunosuppressive signature in patients with HSIL bears implications to immunotherapeutic vaccine approaches aiming at reinforcing HPV-specific immunity to attack HPV-induced lesions. Recently, we showed that such a vaccine also activates/boosts an unwanted preexisting HPV16-specific T-cell repertoire in cervical cancer patients (27), suggesting that strategies to overrule or eliminate the responses of these subsets of T cells in cancer patients should be considered for immunotherapeutic strategies against HPV-induced cervical lesions. In view of this, it is important to realize that the patients with HSIL who failed to induce a detectable dysfunctional T-cell response may actually be naive to the HPV antigens and as such may mount a more functional T-cell response after therapeutic vaccination.

No potential conflicts of interest were disclosed.