Introduction
Cancer of the cervix uteri is the fourth most common cancer among women worldwide, with over 500,000 new cases and 200,000 deaths each year [
1]. It is induced by persistent infection with oncogenic types of HPV, of which high-risk type 16 accounts for over 50 % of the cases [
2]. Whereas early-stage cervical cancer displays a low recurrence risk after treatment (15 %), this is up to 70 % for more advanced stages [
3‐
6]. Metastatic cervical cancer is prototypically chemotherapy resistant and requires better treatments.
Adoptive transfer of tumor-specific T cells is a promising immunotherapeutic treatment modality for patients with advanced or recurrent cancer. Infusions of ex vivo expanded TILs resulted in tumor regressions of lymphoma, Hodgkin’s disease and melanoma [
7‐
9]. Good clinical responses following ACT are related to recognition of tumor-specific antigens [
10,
11] by both CD4+ and CD8+ T cells. Since (metastatic) cervical cancer cells express the viral E6 and E7 oncoproteins, these are good tumor-specific antigens to be targeted by ACT. Indeed, strong spontaneous and vaccine-induced T cell responses to HPV16 E6 and E7 are protective and able to induce regression of premalignant lesions [
12‐
14], whereas ACT with TIL containing HPV-specific T cells of metastatic cervical cancer patients resulted in an objective clinical response in three of the nine treated patients [
15].
Our studies on the systemic and local HPV16-specific T cell response in cervical cancer patients revealed that these responses are detectable in only 30 % of the blood [
16] and 35 % of the tumor specimens [
17] of these patients, indicating that blood or tumor tissue may not function as a consistent source for HPV-specific T cells. However, the TDLN of patients with cervical cancer more frequently (50–83 %) contain HPV-specific T cells [
17,
18] reactive to many epitopes [
18], suggesting that TDLN may form better starting material to obtain cells for ACT. To investigate this, we expanded and characterized HPV-specific CD4+ and CD8+ T cells from lymph node mononuclear cells (LNMC) in patients with HPV16-induced cervical cancer under clinical grade (i.e., GMP) conditions and found that this reproducibly led to the expansion of polyclonal HPV16-specific type 1 T cell populations.
Materials and methods
Subjects
Eleven women with histologically proven early-stage HPV16-positive cervical cancer enrolled in the CIRCLE study, which investigates cellular immunity against HPV16+ anogenital lesions, were analyzed for the presence of HPV16+-specific T cells in their TDLN. HPV status was tested using HPV16-specific primers on DNA isolated from surgical resection specimens [
19]. The patients were staged according to International Federation of Gynecology and Obstetrics (FIGO) and treated by radical hysterectomy and bilateral pelvic lymphadenectomy. The mean age was 44 years (median 42; range 28–65 years). Nine patients were staged FIGO IB1, one patient FIGO IB2 (C809) and one patient FIGO IIA (C800); all patients had squamous cell carcinoma. An additional 8 HPV16+ cervical cancer patients were incorporated of whom the blood samples were analyzed. Six patients were staged FIGO IB1, one patient IIA (C417) and one patient IA1 (C436). The study design was approved by the Medical Ethical Committee of the Leiden University Medical Center (LUMC), Leiden, The Netherlands.
Isolation of peripheral blood mononuclear cells
Venous blood samples (54 mL) were drawn, and peripheral blood mononuclear cells (PBMC) were isolated using Ficoll density gradient centrifugation and cryopreserved as described earlier [
14].
Isolation and stimulation of lymph node-derived T cells
TDLN, suspected for tumor cell infiltration, were derived from the pelvic region and sampled during standard lymphadenectomy. The lymph nodes were cut into pieces, incubated for 1 h with collagenase IV (200 IU/ml, Sigma, St Louis, USA) and DNAse (50 mg/ml, Sigma) at 37 °C, after which a single cell suspension of the lymph node mononuclear cells (LNMC) was made using a cell strainer (BD, Erebodemgem, Belgium). LNMC samples were cryopreserved in 90 % FCS (Life Technologies, Bleiswijk, The Netherlands)/10 % DMSO (WAK-Chemie Medical, Steinbach, Germany) using a freezing machine and stored in the vapor phase of the liquid nitrogen. Alternatively, the lymph node was cut in half, and one side was scraped with a scalpel to obtain a layer of single cells of the LNMC, which were then cryopreserved. Only in two cases, patient C727 and patient C796, the resected lymph node used for the study was found to be tumor cell negative upon pathological examination.
For the test, LNMC were thawed, washed and seeded at a density of 0.5–2 × 10
6 cells per well of a 24-well plate in 1 ml of IMDM (Lonza, Breda, The Netherlands) enriched with 10 % pooled human AB serum (Sanquin Bloodbank, Dordrecht, The Netherlands), penicillin (100 U/ml), streptomycin (100 µg/ml) and 2 mM
l-glutamin (all from Lonza) and stimulated with HPV16 E6 and E7 clinical grade long peptide pools (5 µg/ml) in the presence of either recombinant human IL-2 (Aldesleukin, Novartis, Arnhem, The Netherlands) at 3 different concentrations (75, 150 and 300 U/ml) or a combination of 10 % T cell growth factor (TCGF; Zeptometric, Buffalo, USA) and 5 ng/ml recombinant human IL-15 (PeproTech, Hamburg, Germany), for a total of 22 days. The IL-2 and TCGF/IL-15 were added when culture medium was refreshed. Based on the results obtained, following LNMC were stimulated with GMP-grade materials only (C331, C427, C711-II, C707, C726-I and C726-II, and C727-I and C727-II), in which 150 U/ml IL-2 was added three times a week. In order to test the secondary expansion capacity [
20,
21], 22-day cultured LNMC from C331, C427, C711 and C707 were re-stimulated using antigen-loaded autologous monocytes for another 3 weeks and then tested for responsiveness by proliferation, cytokine production and multiparameter flow cytometry. The expansion capacity was expressed as the fold change, which is the total number of viable cells at the end of the culture period (day 22) divided by the starting number of viable cells (both in case of the first and second expansion period). The remaining LNMC bulk culture cells were cryopreserved in 90 % FCS/10 % DMSO.
Antigens
To stimulate LNMC, clinical grade HPV16 E6 and E7 long (25–35-mer) overlapping peptides (HPV16 E6: 9 peptides, HPV16 E7: 4 peptides) were synthesized in the GMP facility of the LUMC [
13,
22]. To analyze the response of LNMC, a set of overlapping 22-mer peptides spanning the entire HPV16 E6 and E7 amino acid sequence as well as recombinant E6 and E7 proteins was synthesized [
23]. The peptides were dissolved as described earlier [
24].
Analysis of T cell specificity and cytokine secretion
Expanded LNMC (50,000 cells/well) were stimulated with HPV16 E6 and E7 (22-mer) peptides (5 µg/ml) or protein (10 µg/ml)-pulsed autologous APC (i.e., B-LCL or monocytes) in triplicate wells in a 3-day proliferation assay. Medium only was taken along as a negative control, and PHA (0.5 µg/ml final concentration; HA16, Murex BioTech, Kent, UK) served as a positive control. After 48 h, 50 µl supernatant of each of the triplicate wells was pooled and stored at −20 °C. During the last 16 h of culture, 50 µl of 10 µCi/ml [
3H]-thymidine (0.5 µCi/well; PerkinElmer, Groningen, The Netherlands) was added to measure proliferation as counts per minute (CPM) [
24]. Stimulation indexes (SI) above 3, calculated as the average CPM of the triple-stimulated cell conditions divided by the average CPM of the three control wells consisting of LNMC with unloaded APC, were considered to be positive responses.
Antigen-specific production of cytokines in the supernatants of the proliferation assays was measured by ELISA (IFNγ and IL-10, according to the manufacturer Sanquin, Amsterdam, The Netherlands) and/or by the human Th1/Th2 cytometric bead array (BD Pharmingen) [
14,
23]. Antigen-specific cytokine production was defined by a cytokine concentration above the cutoff value of the test (as defined by the manufacturer) and at least twice the concentration of the medium control (unloaded APC). The phenotype and poly-functionality of the T cell reactivity was examined by multiparameter flow cytometry as described previously [
18]. Non-stimulated and PHA-stimulated LNMC served as negative and positive control, respectively. Responses were considered positive when the percentage of HPV-stimulated activated (CD154+ and/or CD137+) T cells and/or the percentage of cytokine producing (IL-2 and/or IFNγ) T cells was at least 3 times the medium control.
Isolation of T cell clones
T cell clones from the TDLN of patient C331 were isolated using limited dilution as described earlier [
17]. Briefly, CD4+ and CD8+ T cells were isolated using Dynabeads (Invitrogen, Carlsbad, CA, USA) or magnetic cell sorting (MACS; and detaching beads, Miltenyi Biotec, Bergisch Gladbach, Germany), respectively. Then, the cells were plated at 0.3 cells/well in IMDM (with 5 % human AB serum, 5 % FCS and 100 U/ml IL-2) and cultured in the presence of irradiated and overnight with HPV16 E6/E7 peptide-loaded autologous B-LCL as well as feeder cells (pooled PBMCs). Re-stimulation was done every 10 (CD8+) or 14 days (CD4+), and fresh medium with IL-2 was added weekly. The phenotype and TCRVβ usage [
18] of T cell clones were tested by flow cytometry. Antigen specificity was determined by proliferation and cytokine production.
Phenotyping of bulk cultures
The LNMC bulk cultures at day 22 were phenotyped by using three sets of markers and analyzed by flow cytometry. The cultured T cells were screened for the presence of regulatory T cells as described previously [
23], the memory subtypes [
25] and expression of inhibitory markers [
26]. The memory set contains CD3-Pacific Blue (Clone UCHT1; DAKO), CD4-PE-CF594 (Clone RPA-T4; BD), CD8-APC-Cy7 (Clone SK1, BD), CD25-Brilliant Violet 605 (Clone 2A3, BD), CD27-Horizon V500 (Clone M-T271, BD), CD28-FITC (Clone CD28.2, BD), CD45RA-PerCP-Cy5.5 (Clone HI100, Biolegend), CD45RO-PE (Clone UCHL1, BD), CD62L-Alexa Fluor 700 (Clone DREG-56, Biolegend), CD95-PE-Cy7 (Clone DX2, Biolegend) and CCR7-Alexa Fluor 647 (Clone 3D12, BD). The inhibitory set consists of the same antibodies for CD3, CD4 and CD8 and additionally CD152-PE-Cy5 (Clone BN13, BD), CD279-Brilliant Violet 605 (Clone EH12.2H7, Biolegend) and TIM3-PE (Clone F38-2E2, Biolegend). The cryopreserved bulk LNMC were thawed, and after dead cell staining (yellow fixable live/dead staining kit, Invitrogen) for 20 min at room temperature (RT) in the dark, the cells (2 × 10
6) were washed in PBS supplemented with 0.5 % bovine serum albumin (BSA), incubated for 10 min at RT in PBS/0.5 % BSA/10 %FCS (in the dark) to prevent nonspecific antibody binding and centrifuged, and the cell pellet was re-suspended in the antibody mixtures as described above and incubated for 20 min on ice (in the dark). Then, the cells were washed twice with PBS/0.5 % BSA and finally re-suspended in 1 % paraformaldehyde (LUMC). The acquisition on the Fortessa (BD) was performed immediately after the staining was finished. Analysis was performed by using DIVA software (version 6.2). An example of the gating strategy for the memory and inhibitory panel is provided in Supplementary Fig. 1. Following the frequency analysis, the data were subjected to a Multi-experimental viewer 4 (MeV4.0) software containing an algorithm for hierarchical clustering using Euclidean distance and complete linkage.
Discussion
In this explorative study, we found that isolation and expansion of HPV16-specific LNMC derived from TDLN of patients with HPV16-induced cervical cancer are feasible and result in the generation of a polyclonal HPV-specific T cell response in all eleven tested patients. After stimulation with GMP-grade E6 and E7 peptides and IL-2, the LNMC expanded ~36-fold. The expansion of HPV16-specific CD4+ T cell was found in all nine patients tested in detail, and in three patients, also HPV16-specific CD8+ T cells were detected. The bias toward CD4+ T cell reactivity against HPV-derived epitopes is not likely a result of the culture method used here, but more a reflection of what is generally found in the spontaneous T cell response to HPV in cervical cancer [
17,
18,
29‐
31], as well as among TILs from patients with head and neck cancer [
32]. The T helper type 1 (Th1) cytokine IFNγ was produced in all LNMC cultures and in some cases also the Th2 cytokines IL-10 and IL-5. Importantly, the procedure was reproducible as complete repeats of the stimulation procedures under research and under full GMP conditions showed similar results when compared to the first runs. Promising results already have been obtained in a pilot study in colorectal cancer patients, as immunotherapy or in adjuvant setting, using TDLN-expanded T cells for ACT [
33,
34].
The HPV peptide-stimulated LNMC cultures predominantly contained HPV16-specific CD4+ T cells, producing IFNγ and/or IL-5. HPV16-specific T cells with this mixed cytokine profile were also found in antigen-experienced healthy individuals [
35] and in patients with a complete regression of their HPV16-induced high-grade vulvar lesion after therapeutic vaccination [
13,
14,
23], indicating that the responding LNMC cells acquired an appropriate cytokine profile during the stimulation procedure. The outgrowth of tumor-specific T cells of only a CD4+ phenotype should not pose a problem for their use in ACT. Although successes have been achieved with ACT products containing merely tumor-directed CD8+ T cells [
36,
37], there are indications that CD4+ T cells can help or can do the job. Substantial or complete tumor regressions have been achieved by ACT of T cells consisting only or mainly of CD4+ T cells [
7,
38‐
40]. Transferred CD4+ T cells can contribute to antigen spreading [
38], enhance the recruitment of CD8+ T cells to the tumor as well as sustain their effector function [
41], reduce CD8+ T cell exhaustion [
42], switch tumor-induced M2 macrophages to activated M1-like macrophages [
43] and kill tumor cells via direct and indirect mechanisms [
44,
45]. Thus, the infusion of tumor-specific Th1 cells may have great clinical benefit by altering the tumor micromilieu into a preferred type 1 cytokine-associated immune contexture [
46].
In a number of ACT studies in metastatic melanoma, the tumor-specific T cells are derived from PBMC [
27,
47,
48]. Based on our previous study on HPV-specific T cell responses in almost 100 patients with cervical cancer, we predicted that the PBMC of patients with cervical cancer would be an inferior source for HPV16-specific T cells than TDLN. We stimulated the PBMCs of eight different HPV16+ cervical cancer patients and detected a secondary proliferative response with production of IFNγ in four patients, three of which was expected based on the earlier conducted LST. The stimulated and expanded PBMC of two additional patients produced only IFNγ when stimulated with HPV16 antigens, suggesting that these cells lost their capacity of secondary peptide-specific expansion. These data suggest that PBMCs can be used as a source for successful expansion of HPV-specific T cells in about 50 % of the cases, and the highest success rate (100 %), however, is obtained when TDLN are used as source.
The relatively low numbers of obtained T cells after ex vivo expansion may be a limitation in ACT studies. Success rates of 50–70 % have been described in ACT studies in patients with metastatic melanoma [
8,
37,
49]. In these studies, the functional properties, including the expansion rate and the presence of Th1 cytokines, and the total number of the transfused T cells were shown to correlate with clinical success. Most melanoma patients in the first ACT studies received over 10
11 cells. In this study, we were able to expand 1–5 × 10
6 LNMC to 36–180 × 10
6 T cells per patient, respectively. It is not likely that the amount of starting material will be more in a clinical setting. Lymph node material will have to be collected during histological biopsy, and in our experience, on average 12 × 10
6 (range 0–72 × 10
6; median 5.6 × 10
6;
n = 48) LNMC are obtained when TDLN isolated during pelvic lymphadenectomy are scraped (unpublished data). There is evidence that ACT with lower numbers of T cells also results in clinical benefit. A recent study in melanoma patients showed that transfusions of on average 5 × 10
8 T cells lead to clinical responses in half of the patients [
27]. Moreover, a single transfusion of 0.37–2.2 × 10
5 cytomegalovirus-specific CD8+ T cells/kg body weight after allogeneic stem cell transplantation leads to viral clearance in acute leukemia patients [
50]. As the expanded numbers of LNMC using the current protocol will be lower than transferred to melanoma patients but the tumor antigens are known, there is the option to increase the effects of ACT by vaccination following T cell transfer [
46]. We recently showed that vaccination induced in vivo massive (<1000-fold) clonal expansions and regression of tumors in a mouse melanoma model [
51]. Notably, in HPV-driven cancers, the tumor-specific antigens are always the same. Hence, the use of an off-the-shelf vaccine, such as the HPV16 synthetic long peptide vaccine with proven strong immunogenicity in end-stage cervical carcinoma patients [
22,
26,
52,
53], following ACT may lead to massive numbers of HPV-specific polyclonal T cells in patients with HPV16-induced cancer.
Taken together, our results show that all TDLN cell cultures isolated from patients with HPV16-induced cervical cancer are a rich source of polyclonal HPV-specific Th1 and Th2 cells, which after proper ex vivo stimulation can be expanded under full GMP conditions. These HPV-specific T cells may be used in future ACT studies in patients with metastatic HPV16-induced cancer.