Identification of Claudin 6-specific HLA class I- and HLA class II-restricted T cell receptors for cellular immunotherapy in ovarian cancer.

Adoptive cell therapy (ACT) is one of promising immunotherapies for cancer patients by providing a large amount of cancer antigen-specific effector T cells that can be manufactured rapidly by ex vivo gene engineering. To provide antigen-specificity to patients' autologous T cells in a short-term culture, T-cell receptors (TCRs) or chimeric antigen receptors (CARs) are transduced to bulk T cells. Because of intra- and inter-tumoral heterogeneity in tumor antigen expression, a repertoire of TCR or CAR genes targeting a wide range of tumor antigens are required for a broad and effective treatment by ACT. Here, we characterized immunogenicity of claudin 6 (CLDN6) in ovarian cancer patients and identified specific TCR genes from CD8+ and CD4+ T cells. CLDN6 protein was frequently expressed on EpCAM+ ovarian cancer cells but not CD45+ lymphocytes in tumor ascites of ovarian cancer patients. Spontaneous CLDN6-specific CD4+ and CD8+ T-cell response was detected in peripheral blood mononuclear cells (PBMCs) from 1 out of 17 ovarian cancer patients. HLA-A*02:01 (A2) and DR*04:04 (DR4)-restricted TCR genes were isolated from CLDN6-specific CD8+ and CD4+ T cells, respectively. T cells that were engineered with A2-restricted TCR gene recognized and killed A2+CLDN6+ cancer cells. DR4-restricted TCR-transduced T cells directly recognized DR4+CLDN6+-overexpressed cancer cells. Our results demonstrate that these CLDN6-specific TCR genes are useful as therapeutic genes for ACT to patients with ovarian and other solid tumors expressing CLDN6.

Introduction

Adoptive cell therapy (ACT) using tumor antigen-specific T cells is a powerful strategy for the treatment of cancer patients as infusion of a large number of anti-tumor effector T cells provides immediate tumor-debulking effects. Furthermore, a subset of infused T cells is expected to differentiate into memory T cells to provide long-term immunosurveillance and potentially prevent recurrence of disease.[1,2] Recent gene engineering techniques to transduce target antigen-specific chimeric antigen receptor (CAR) or T-cell receptor (TCR) enable rapid generation of a large number of autologous therapeutic T cells with potent anti-tumor activity.[3-6] In our previous studies, we have cloned and characterized NY-ESO-1-specific and MHC class I- and class II-restricted TCR genes that are currently tested in a phase I clinical trial (ClinicalTrials.gov Identifier: NCT03691376).[7,8] Although the safety and efficacy of ACT using NY-ESO-1-specific TCR-transduced cells in cancer patients has been demonstrated,[9,10] eligible patients for the therapy are limited by NY-ESO-1 expression (30–80% of solid tumors)[11-13] and patients’ HLA types. In addition, NY-ESO-1 frequently shows heterogeneous expression within a tumor,[14,15] which allows antigen-negative tumor variants to escape immune attack. To develop broadly applicable and more efficient TCR gene-engineered T-cell therapy for cancer patients, identification of a panel of off-the-shelf TCR genes for shared cancer antigens or shared neoantigens is essential.

Claudin 6 (CLDN6) is a cell surface membrane protein expressed on multiple solid tumor tissues such as ovarian cancer, testicular cancer, and endometrial cancer, while the expression is not observed on normal adult tissues at the transcription and protein levels.[16-20] Thus, CLDN6 is considered to be a promising target for antibody-based cancer immunotherapies. CLDN6-specific antibodies have been engineered as bi-specific antibodies and cytotoxic-drug conjugates.[21,22] In addition, Reinhard et al. demonstrated the efficacy of CLDN6-specific CAR-T cells in mouse models,[16] and a phase I/IIa clinical trial using the CAR-T cell is ongoing. Because of the tumor-specific cell surface expression, CLDN6 is also a potential target for other immunotherapies such as cancer vaccines. Although the expression of CLDN6 on solid tumors has been broadly investigated, its immunogenicity in cancer patients has not been evaluated. Here, we investigated spontaneously induced T-cell response against CLDN6 in ovarian cancer patients. By characterizing CLDN6-specific CD8+ and CD4+ T cells, we have identified novel CLDN6-derived short peptides that bind on HLA-A*02:01 (A2) and -DRB1*04:04 (DR4). Furthermore, we have cloned HLA-A2 and -DR4-restricted CLDN6-specific TCR genes and characterized cancer cell recognition of TCR gene-transduced T cells for the development of TCR-engineered T-cell therapy.

Materials and methods

Biospecimen processing

Blood and malignant ascites were obtained from ovarian cancer patients under an approved protocol (I 215512) from the institutional review board at Roswell Park Comprehensive Cancer Center (Roswell Park). Patients signed a consent form to approve the use of their specimens for research. Most of specimens that were used in this study were collected at the time of debulking surgery prior to chemotherapy from patients with high-grade serous ovarian cancer. Healthy donor blood samples were obtained from the blood donor center at Roswell Park. Peripheral blood mononuclear cells (PBMCs) and ascites mononuclear cells were isolated using lymphocyte separation media (Corning), cryopreserved in 10% DMSO/90% FBS, and stored in a liquid nitrogen freezer.

ELISPOT assay

PBMC were thawed and 5–10 × 105 cells were cultured with a pool of synthetic overlapping peptides for CLDN6 (1 µg/ml, JPT Peptide Technologies), or a pool of 27 peptides of Cytomegalovirus, Epstein-Barr virus, Influenza virus, and Tetanus toxin (CEFT; 0.5 µM/each, GenScript) in RPMI1640 medium (Corning) supplemented with 10% human AB serum (Gemini), 1× MEM non-essential amino acid (Corning), 100 U/ml penicillin (Corning), 100 µg/ml streptomycin (Corning), 10 U/ml IL-2 (Roche), and 10 ng/ml IL-7 (R&D Systems) in 96-well round-bottom plates. The cells were expanded every 3–4 days. After 13–15 d of culture, cells were harvested and suspended in X-VIVO 15 media (Lonza). Fifty thousand cells were seeded in ELISPOT plate (MultiScreen-HA filter plate, Millipore Sigma) that were pre-coated with anti-IFN-γ antibody (clone 1-D1K, Mabtech) in the presence or absence of the respective peptide mix. The plate was incubated in a 5% CO2 incubator for 20–24 h, followed by development with biotinylated detection anti-IFN-γ antibody (clone 7-B6-1, Mabtech), streptavidin-alkaline phosphatase (ALP) (Mabtech), and 5-bromo-4-chloro-3ʹ-indolyphospate p-toluidine salt/nitro-blue tetrazolium chloride (BCIP/NBT) substrate (Sigma-Aldrich). The numbers of spots were counted using an ImmunoSpot S6 Core analyzer (CTL). The response was considered as positive when the number of spots is ≥50 and twice higher than background.[23]

CLDN6 expression

Mononuclear cells from ascites were thawed and stained with Fixable Viability Stain 700 (BD Horizon) followed by antibodies against CD45 (clone H130, BioLegend), CD326 (EpCAM; clone 9C4, BioLegend), and CLDN6 (clone 342927, R&D Systems). PA-1 ovarian cancer cell line was obtained from the American Type Culture Collection. An epithelial ovarian cancer (EOC) cell line was established from solid tumor single cell suspension. The tumor cells were injected intraperitoneally into NOD/SCID/IL-2Rγnull (NSG) mice under the institutional IACUC approved protocol (1346 M) at Roswell Park. When abdominal circumference was extended, ascites was harvested and the mononuclear cells were cultured in RPMI1640 medium supplemented with 20% FBS, 100 U/ml penicillin and 100 µg/ml streptomycin. EpCAM expression was confirmed by flow cytometry. Cancer cell lines were cultured in the presence or absence of 1,000 U/ml human IFN-γ recombinant protein (PeproTech) for 2 days. The cells were harvested using 0.25% trypsin-EDTA solution (Corning) and stained with antibodies for HLA-A2 (clone BB7.2, BioLegend) and CLDN6. Cells were acquired by the BD Fortessa flow cytometer and analyzed using FlowJo software.

Generation of Epstein-Barr virus (EBV)-transformed B cell lines

PBMC were first stained with biotinylated anti-CD4 (clone OKT4) and anti-CD8 (clone HIT8a) antibodies (BioLegend) and CD4+ and CD8+ T cells were depleted using Dynabeads Biotin Binder (Thermo Fisher Scientific). One million T-cell-depleted PBMCs were cultured with a culture supernatant of B95-8 cell line in the presence of 2 µg/ml cyclosporin A (Sigma-Aldrich) in RPMI1640 medium supplemented with 10% FBS, 100 U/ml penicillin, and 100 µg/ml streptomycin.[24] HLA types of EBV-transformed B (EBV-B) cells were determined by the Immune Analysis Shared Resource (IASR) at Roswell Park.

Generation of CLDN6-specific TCR gene-transduced T cells

CLDN6-specific T cells were isolated as previously described using the IFN-γ secretion assay kit (Miltenyi Biotech).[25] The patient P01 EBV-B cells (1 × 106 cells) were labeled with CFSE and incubated with 2 µM CLDN6 pooled peptides overnight in a CO2 incubator. The patient P01 T cells (2 × 106 cells) were incubated with peptide-pulsed CFSE-labeled EBV-B cells in 0.5 ml of X-VIVO medium in a 5 ml round-bottom tube for 4 h. After the incubation period, cells were labeled with an IFN-γ catch reagent, further incubated for 45 min, and stained by anti-IFN-γ detection and anti-CD4 antibodies (clone RPA-T4, BioLegend) according to the manufacturer’s instruction. CFSE-negative IFN-γ+CD4+ and IFN-γ+CD4− cells were sorted using the BD FACSAria cell sorter (BD Biosciences). One thousand cells were dissolved in TRI Reagent solution (Invitrogen) for RNA extraction. Total RNA was used for RT-PCR amplification of TCR α- and β-chain genes that were assembled to generate TCR-expressing plasmid vectors as previously described.[26] Culture supernatants from high-titer retrovirus producing PG13 clones were harvested and infected to healthy donor PBMC. Transduction efficiency was determined by flow cytometry using anti-Vβ8 (for CD8-TCR; clone JR2, BioLegend) and anti-Vβ7.1 (for CD4-TCR; clone ZOE, Beckman Coulter) antibodies.

T-cell functional analysis

CLDN6-specific CD8+ T-cell line was established by expanding the sorted CLDN6-specific T cells with 10 µg/ml phytohemagglutinin in the presence of 30 Gy γ-irradiated healthy donor PBMC, 10 U/ml IL-2, and 10 ng/ml IL-7. For intracellular cytokine staining, T cells were stimulated with peptide-pulsed or -unpulsed P01-derived or HLA-matched EBV-B cells (Table 1) for 6 h in the presence of 5 µg/ml monensin (Sigma-Aldrich). In some experiments, anti-CD107a (clone H4A3, BD Biosciences) and anti-CD107b (clone H4B4, BD Biosciences) antibodies along with 5 µg/ml Brefeldin A (Sigma-Aldrich) were added during incubation. Cells were stained for CD3 (clone OKT3, BioLegend), CD4 (clone OKT4, BioLegend), CD8 (clone RPA-T8, BioLegend), Vβ7.1, and/or Vβ8, followed by fixation with 2% formaldehyde and permeabilization with Permeabilization Medium B (Thermo Fisher) for intracellular staining for IFN-γ (clone B27, BD Biosciences), TNF-α (clone MAb11, BD Biosciences), GM-CSF (clone BVD2-21 C11, BD Biosciences), and/or IL-2 (clone MQ1-17H12, BioLegend). To determine cytokine levels in a culture supernatant, T cells (100,000 cells/well) were cocultured with target cells (50,000 cells/well) for 24 h. To define T-cell epitope, an independent set of 21 CLDN6 20-mer peptides with 10 amino acid overlapping each (EZBiolab) was used (Supplemental Table 1). Cancer cells were cultured for 2 d with 1,000 U/ml IFN-γ (PeproTech) and washed thoroughly before the co-culture with T cells. The culture supernatant was harvested, and IFN-γ level was determined using an ELISA kit (eBioscience) according to the manufacturer’s instruction.

Table 1.

HLA types of patient and EBV-B cells. For each gene, HLA types matched with P01 is shown as bold and underlined. Blank (-) indicates homozygous expression

Sample	A	B	C	DRB1	DQB1	DPB1
P01	*02*2402	*27-	*01-	*0404*1401	*0302*0503	*0201*0401
EBVB-1	*01*02	*0702*0801	*07-	*1102*1401	*0301*0503	*0201-
EBVB-2	*0101-	*0801*5701	*0602*0701	*0301*0701	*0201*0303	*0401-
EBVB-3	*02*2301	*27*44	*02*04	*04*07	*0201-	*0401-
EBVB-4	*0201*0301	*0702*4405	*0202*0702	*0404*1101	*0301*0302	*0401-
EBVB-5	*02*0301	*15-	*03-	*0401*1101	*0301*0302	*0401*0402
EBVB-6	*02*03	*40-	*02*03	*1101*1301	*03*06	*0201*0401

Tetramer analysis

HLA-A2-binding epitope of CLDN6 between CLDN6131–140 was predicted using SYFPEITHI database.[27] TLIPVCWTA (CLDN6132–140) peptide was synthesized by EZBiolab with 82% purity. An HLA-A2/peptide monomer was generated with the Flex-T HLA-A*02:01 monomer UVX reagent followed by tetramerization using PE-streptavidin according to the manufacturer’s instruction (BioLegend). Optimum staining concentration of the A2/CLDN6132–140 tetramer was determined by titration. CD8-TCR- or mock-transduced T cells were first treated with 50 nM dasatinib (Sigma-Aldrich) for 30 min at 37°C and then stained with the tetramer reagent for 30 min on ice. Tetramer-stained cells was further stained with purified anti-PE (clone PE001, BioLegend) and PE goat anti-mouse Ig (multiple adsorption, BD Biosciences) antibodies.[28]

Retroviral expression of HLA genes

Coding region of HLA-A/B/C genes were amplified by RT-PCR using a tumor tissue as a template and cloned into a murine stem cell virus (MSCV)-based retroviral expression vector. HLA types of cloned genes were determined by Sanger sequencing at the Genomics Shared Resource at Roswell Park. High titer retroviral vectors were produced from PG13 packaging cell lines and were used to transduce K562 and PA-1 cells.

Generation of ovarian cancer cell lines overexpressing CLDN6 and HLA-A*02:01 or HLA-DRB1*04:04

Coding regions of CLDN6, HLA-A2, and -DR4 were amplified by RT-PCR from cDNA of a tumor tissue. CLDN6 gene was genetically fused to the puromycin-resistant gene via P2A translational skipping sequence and DR4 was fused to human CIITA gene[7] via T2A sequence. CLDN6-Puro and A2 or DR4-CIITA gene were cloned in a piggyBac transposon plasmid, as shown in Supplemental Fig S6. These piggyBac transposon plasmids were electroporated into CLDN6− ovarian cancer cell line, A2780, together with a hyperactive piggyBac transposase (hyPBase)[29]-expressing plasmid using the Nucleofector 4D instrument. Two days after electroporation, cells were cultured in the presence of 5 µg/ml puromycin for 10 days. Because a part of cells that were electroporated with the CLDN6 and DR4-expressing plasmid did not express cell surface HLA-class II, HLA-DR+ A2780 was isolated using anti-DR antibody (clone L243, BioLegend) and a pan mouse IgG kit (Invitrogen) according to the manufacturer’s instruction before co-culturing with TCR gene-transduced T cells.

Cytotoxicity assay

Cytotoxicity was determined by a calcein-AM release assay as previously described.[8] Briefly, the parental and A2+CLDN6+ A2780 were labeled with 5 μM calcein-AM (Sigma-Aldrich) and mixed with CD8-TCR- or mock-transduced T cells for 4 h at E/T ratio, 50, 25, or 12.5. Fluorescence release in cell-free supernatants was measured by the Synergy H1 microplate reader (BioTek) with excitation 485 and emission 528. To induce the maximum release, target cells were treated with 2% Triton-X, whereas the spontaneous release was determined from supernatants of target cell alone. Cytotoxicity was calculated using the following formula: % cytotoxicity = 100 × [(experimental release – spontaneous release)/(maximum release – spontaneous release)]. To evaluate the induction of apoptotic cell death of cancer cells by CD8-TCR-transduced T cells, the parental and A2+CLDN6+ A2780 (5 × 105 cells) were cocultured with T cells (1 × 106 cells) for 18 h. Cells were harvested with 0.25% trypsin–EDTA solution and stained with anti-CD3 antibody (clone UCHT1, BioLegend) followed by annexin V and PI according to the manufacturer’s instruction (BioLegend). Percentages of Annexin V+ and PI+ cells were analyzed for CD3− cells.

Statistical analysis

The statistical difference of CLDN6 expression on EpCAM+ and CD45+ cells was analyzed by GraphPad Prism software using the two-tailed paired t-test. The correlation between %EpCAM+ and CLDN6+ cells was analyzed by GraphPad Prism software using a simple linear regression analysis.

Results

Expression of CLDN6 on EpCAM+ cancer cells in ascites of ovarian cancer patients

To investigate the cell surface CLDN6 expression on primary ovarian cancer cells, mononuclear cells from ascites of 32 ovarian cancer patients were stained with antibodies for CLDN6, EpCAM, and CD45 and analyzed by flow cytometry (Figure 1). In most ascites specimens, CD45+ cells were highly infiltrated, whereas percentages of EpCAM+ cells were relatively low (Supplemental Fig S1). CLDN6 expression on EpCAM+ and CD45+ cells was determined using a Fluorescence Minus One (FMO) control (Figure 1a). CLDN6 was significantly but variably expressed on EpCAM+ cells but not on CD45+ cells (Figure 1b). The mean frequency of CLDN6 expression on EpCAM+ cells was 30% (range 0.2–80.0%) while it was 0.17% on CD45+ cells (0–3.0%). CLDN6 expression on EpCAM−CD45− cells was negligible (data not shown). There was no correlation between percentages of EpCAM+ cells and CLDN6+ expression levels (Figure 1c). The lack of expression in normal tissues,[16] but frequent and tumor-specific expression suggest that CLDN6 is a promising target for immunotherapy for ovarian cancer patients.

Figure 1.

CLDN6 expression on EpCAM+ cancer cells and CD45+ lymphocytes in ascites fluids of ovarian cancer patients. (a) Flow cytometry analysis of ascites mononuclear cells. CLDN6 expression was analyzed on EpCAM+ or CD45+ cells. Gating of CLDN6 was determined based on FMO controls. (b) Percentages of CLDN6 expression on EpCAM+ or CD45+ cells from 32 patients were plotted with mean ± standard deviation. (c) A relationship between CLDN6 expression levels and percentages of EpCAM+ cells in ascites was analyzed by a linear regression analysis.

Detection of spontaneous T-cell responses against CLDN6 in ovarian cancer patients

To investigate spontaneously induced T-cell response against CLDN6 in ovarian cancer patients, CLDN6-specific response was tested following in vitro stimulation of PBMC with a pool of overlapping peptides. One (P01) out of 17 patients showed a specific response against CLDN6 based on the criteria for a positive response (the number of spots is ≥50 and twice higher than background) (Figure 2a). For patient P01, we found that both CD8+ and CD4+ T cells were reactive to CLDN6 pooled peptides by intracellular cytokine staining (Figure 2b).

Figure 2.

Analysis of CLDN6-specific T-cell response in ovarian cancer patients. (a) CLDN6- or CEFT-specific T-cell responses from 17 patients (P01–P17) were determined by ELISPOT assays. PBMC were cultured with CLDN6 or CEFT peptides for 13–15 d and IFN-γ spots in the presence (pep) or absence (-) of the respective peptide were enumerated. (b) IFN-γ and TNF-α production from CD8+ or CD4+ T cells from P01 against CLDN6 peptide-pulsed or -unpulsed autologous P01 EBV-B cells was analyzed by intracellular cytokine staining.

Generation of CLDN6-specific T cells by TCR gene engineering

To clone TCR genes, total RNA was extracted from CLDN6-specific CD4+ and CD8+ T cells purified by sorting IFN-γ-producing cells from the patient P01 (Figure 2b). The coding regions of TCR α and β chain genes that were amplified by RT-PCR from sorted CD8+ or CD4+ T cells were assembled as an expression cassette in a retroviral plasmid vector by our TCR-expressing retroviral vector construction method.[26] DNA fingerprinting of TCR inserts in randomly selected plasmid clones by restriction enzyme digestion showed that the majority of plasmid clones contained the same TCR inserts, indicating oligoclonal CD4+ and CD8+ T-cell responses dominated by a single clone against CLDN6 (Supplemental Fig S2). Retroviral particles produced from PG13 virus-producing cells were used to transduce TCR genes to T cells derived from healthy donor PBMC. According to the Sanger sequencing data of dominant TCR-expressing plasmids (Supplemental Data 1), we identified TCR Vβ subtype of CLDN6-specific CD8+ T cells and CD4+ T cells as Vβ8 and Vβ7.1, respectively. TCR-transduced T cells were stained with those TCR Vβ antibodies and the transduction efficiency was determined to be >85% (Figure 3a). Reactivity of TCR-transduced T cells to CLDN6 pooled peptides was confirmed by intracellular cytokine staining (Figure 3b and Figure 3c). While both CD8+ and CD8− (i.e. CD4+) T cells transduced with TCR derived from CD4+ T cells (CD4-TCR) strongly produced IFN-γ against peptides, CD8− T cells transduced with CD8-TCR showed a weak response, indicating that CD8-TCR requires CD8 co-ligation for the recognition (Figure 3c). Transduction efficiency and reactivity of TCR-transduced T cells were similar in three healthy donor-derived PBMCs (Supplemental Fig S3).

Figure 3.

Establishment of CLDN6-specific TCR gene-engineered T cells. Healthy donor PBMC were retrovirally transduced with TCR genes from CLDN6-specific CD8+ T cells (CD8-TCR) or CD4+ T cells (CD4-TCR) derived from the patient P01 (Figure 2b). (a) Transduction efficiency of TCR- or mock-transduced T cells were analyzed by TCR-Vβ8 or -Vβ7.1 antibody. (b, c) IFN-γ production against CLDN6 pooled peptide-pulsed or -unpulsed P01 EBV-B cells were analyzed by intracellular cytokine staining. IFN-γ production from Vβ8 or Vβ7.1 positive/negative cells (b), and CD8 positive/negative cells (c) is shown.

Identification of epitope and HLA restriction for CLDN6-specific CD8+ and CD4+ T cells

By utilizing TCR-transduced T cells, we characterized a peptide region recognized by CLDN6-specific T cells. By determining reactivity of CD8-TCR- and CD4-TCR-transduced T cells against individual peptides covering the whole CLDN6 protein, the epitope for CD4-TCR was determined to be in the N-terminal CLDN61–20 peptide in a signal peptide region, while the epitope for CD8-TCR is in an overlapping region of peptides CLDN6121–140 and CLDN6131–150, as CD8-TCR-transduced T cells similarly recognized these peptides (Figure 4a). The HLA restriction element of CD8-TCR was determined by using K562 cells that were transduced with the patient P01’s HLA class I. HLA class I types of the patient P01 were determined to be HLA-A2, -A*24:02, -B*27, and -Cw*01 (Table 1). HLA-negative K562 cells were retrovirally transduced with these HLA genes, pulsed with CLDN6 peptides, and co-cultured with CD8-TCR-transduced T cells. CD8-TCR was found to be restricted to A2 as it showed strong reactivity against peptides only when pulsed on A2-transduced K562 cells (Figure 4b). Based on A2 and peptide region recognized by CD8-TCR, A2-binding epitope of CLDN6 was predicted as TLIPVCWTA (CLDN6132–140) using the SYFPEITHI algorithm.[27] TCR binding of the predicted epitope was demonstrated by A2/CLDN6132–140 tetramer staining (Figure 4c). Using the CLDN6132–140 peptide, we tested TCR avidity against the peptide and found that CD8-TCR-transduced T cells efficiently recognized 1 nM peptide (Figure 4d). The HLA restriction element of CD4-TCR was identified as HLA-DR*04:04 based on the peptide reactivity using a panel of partially HLA-matched EBV-B cells as antigen-presenting cells (Figure 4e and Table 1). The HLA restriction of CD4-TCR was confirmed against CLDN61–20-pulsed DR4-overexpressing A2780 cell line (Supplemental Fig S7).

Figure 4.

Characterization of CLDN6-specific TCR gene-engineered T cells. (a) TCR gene-engineered T cells were cocultured with 20-mer CLDN6 overlapping peptides-pulsed or -unpulsed P01 EBV-B cells for 24 h. IFN-γ level in the culture supernatant was measured by ELISA. (b) Cytokine production from CD8-TCR-engineered CD8+ T cells against CLDN6 peptide-pulsed K562 cells transduced with indicated HLA genes was analyzed by intracellular cytokine staining. (c) CD8-TCR- or mock-transduced T cells were stained with or without HLA-A2/CLDN6132–140 tetramer along with anti-Vβ8 antibody. (d) IFN-γ production on CD8-TCR-transduced CD8+ T cells against different concentration of CLDN6132–140 was analyzed by intracellular cytokine staining. Error bars indicate standard deviation of technical duplicates. (e) IFN-γ production on CD4-TCR-transduced CD4+ T cells stimulated with CLDN61–20-pulsed or -unpulsed partially HLA-matched EBV-B cells (Table 1) was analyzed by intracellular cytokine staining.

Cytokine production from CD4-TCR- and CD8-TCR-transduced CD4+ and CD8+ T cells

The major role of CD4+ T cells is enhancing anti-tumor immunity by producing cytokines. Because CD4-TCR provided function to both CD4+ and CD4− T cells, we further characterized cytokine-producing patterns for CD4+ and CD8+ T cells (Figure 5 and Supplemental Fig S4). CD4-TCR-transduced T cells were stimulated with CLDN61-20-pulsed P01 EBV-B cells and analyzed TNF-α, GM-CSF and IL-2 as well as IFN-γ gated on Vβ7.1+CD4+ or Vβ7.1+CD8+ T cells. As shown in Supplemental Fig S4, cytokines were produced from Vβ7.1+ T cells on both CD4+ and CD8+ T cells. While 60% of CD4+ T cells produced more than three cytokines, only 9% of CD8+ T cells showed multi-cytokine production (Figure 5a). CD8-TCR-transduced T cells were also analyzed for multi-cytokine production (Figure 5b). CD8+ T cells produced not only IFN-γ but also TNF-α, GM-CSF, and IL-2 similar to CD4+ T cells transduced with CD4-TCR. This result suggests that CD4 or CD8 co-ligation is required to produce multiple cytokines.

Figure 5.

Multi-cytokine production from TCR-transduced CD4+ or CD8+ T cells. (a) CD4-TCR-transduced T cells was stimulated with a CLDN61–20 peptide pulsed on the patient P01-derived EBV-B cells and cytokine production was analyzed on Vβ7.1+CD4+ or Vβ7.1+CD8+ T cells. (b) CD8-TCR-transduced T cells was stimulated with a CLDN6132–140 peptide pulsed on the patient P01-derived EBV-B cells and cytokine production was analyzed on Vβ8+CD4+ or Vβ8+CD8+ T cells. Pie charts indicate percentages of cells producing 4/3/2/1/0 cytokines.

Cancer cell recognition by CD8-TCR-transduced T cells

We next assessed the reactivity of CD8-TCR-transduced T cells against CLDN6+ ovarian cancer cells. To compare the reactivity of TCR-transduced T cells with that of naturally occurring CLDN6-specific CD8+ T cells, a part of sorted CLDN6-specific CD8+ T cells was polyclonally expanded. As expected, CLDN6-specific CD8+ T-cell line expressed TCR Vβ8 and showed strong reactivity against CLDN6 pooled peptide-pulsed autologous EBV-B cells (Supplemental Fig S5A and S5B).

We investigated expression of CLDN6 and HLA-A2 on the patient P01-derived EOC cell line and an established ovarian cancer cell line PA-1 that was reported to highly express CLDN6 by flow cytometry.[21] As PA-1 cells is A2 negative, A2 was retrovirally transduced. We found that both ovarian cancer cell lines express CLDN6 (Figure 6a). Although the patient P01 was A2+, the EOC cell line showed marginal cell surface A2 expression (Figure 6a). To upregulate HLA expression, the cancer cells were treated with IFN-γ. IFN-γ treatment upregulated A2 on the EOC and A2-transduced PA-1 cells but did not affect the intensity of CLDN6 expression (Figure 6a). When T cells were cocultured with these cancer cells, the P01 CD8+ T-cell line recognized EOC cells only after treatment with IFN-γ (Figure 6b). CD8-TCR-transduced T cells recognized IFN-γ-treated EOC cells with similar efficiency to the P01 CD8+ T-cell line (Figure 6c). CD8-TCR-transduced T cells also showed strong reactivity against A2-transduced and IFN-γ-treated PA-1 cells. The results indicate that CD8-TCR transduced T cells can recognize cancer cells expressing CLDN6 and A2.

Figure 6.

Cancer cell recognition by CLDN6-specific TCR gene-engineered T cells. (a) CLDN6 and HLA-A2 expression on ovarian cancer cells with or without IFN-γ treatment was determined by flow cytometry. Expression was shown as quadrant gating based on the unstained control (left) and histogram including unstained controls (right). PA-1/A2: PA-1 transduced with HLA-A2. (b, c) CLDN6-specific CD8+ T-cell line (b), or CD8-TCR- or mock-transduced T cells (c) were cocultured with or without IFN-γ-treated or -untreated cancer cells and IFN-γ level in the culture supernatant was measured by ELISA. Error bars indicate standard deviation of technical duplicates.

Recognition of CLDN6-overexpressing cancer cells by CD4-TCR and CD8-TCR-transduced T cells

To determine if CD4-TCR-transduced T cells directly recognize CLDN6-expressing cancer cells, we generated A2780 cells co-expressing CLDN6, HLA-DR4, and CIITA genes (DR4/CLDN6). For comparison, A2780 cells were similarly engineered with CLDN6 and HLA-A2 genes (A2/CLDN6) (Figure 7a and Supplemental Fig S6). CD8-TCR-transduced T cells strongly recognized A2/CLDN6-overexpressing cancer cells without IFN-γ treatment (Figure 7b). CD4-TCR-transduced T cells moderately but specifically recognized DR4/CLDN6-overexpressing, but not A2/CLDN6-overexpressing, cancer cells. Because cytotoxic activity by CD8+ T cells play a critical role in tumor regression by T-cell therapy, we tested cytotoxicity of CD8-TCR-transduced T cells. In a 4-h calcein-AM release assay, CD8-TCR-transduced T cells indicated specific cytotoxicity against A2/CLDN6-overexpressing cancer cells (Figure 7c). When CD8-TCR-transduced T cells were co-cultured with cancer cells for 18 h, apoptotic cell death was induced in cancer cells (Figure 7d). These results demonstrate that both epitopes that were recognized by CD4-TCR and CD8-TCR are naturally processed and presented on DR4 and A2, respectively, of CLDN6-expressing cancer cells.

Figure 7.

Recognition of CLDN6-overexpressing cancer cells by CD8-TCR- and CD4-TCR-transduced T cells. (a) Expression of CLDN6, HLA-A2, and HLA-class II on the parental, A2/CLDN6-overexpressing, and DR4/CLDN6-overexpressing A2780 ovarian cancer cell line was determined by flow cytometry. (b) IFN-γ production from CD8-TCR or CD4-TCR-transduced T cells against indicated A2780 cell lines was determined by intracellular cytokine staining. (c) Cytotoxicity of CD8-TCR or mock-transduced T cells against indicated A2780 cell lines was determined by a 4-h calcein-AM release assay. (d) Induction of apoptotic cell death in cancer cells by CD8-TCR-transduced T cells was determined by annexin V and PI staining after 18-h coculture.

Discussion

Adoptive transfer of tumor antigen-specific T cells is a promising strategy to eradicate malignant cells. Although a large number of antigen-specific T cells can be rapidly generated by genetic engineering of patient-derived T cells using TCR or CAR genes, this technology is limited by target antigens.[30,31] To prevent on-target off-tumor toxicity, target antigens should not be expressed on essential normal tissues. As of now, NY-ESO-1 is one of the few promising antigenic targets for TCR-engineered T cells in patients with solid tumors, and ACT using NY-ESO-1-specific TCR-engineered T cells has shown efficacy and safety in clinical trials.[5,32,33] However, because of heterogeneity in NY-ESO-1 expression and HLA types, eligible patients for the therapy are limited. Extensive efforts are ongoing to identify additional TCR/CAR targets that could provide ACT for a broad range of patients. In this study, we identified TCR genes specific to CLDN6, which is a potential target antigen with frequent expression in cancer patients.

CLDN6 is an oncofetal protein expressed on ES cells associated with epithelialization as well as solid cancer cells including ovarian cancer.[18,19,21] Our flow cytometry analysis of ascites cells demonstrated that the majority (94%, 30 out of 32 samples) of patients’ EpCAM+ cancer cells express CLDN6 at varied levels. Expression was low (<10%) in 13%, moderate (10–40%) in 53% and strong (>40%) in 33% of 30 samples. CLDN6 expression was observed in 69.4% of ovarian carcinomas and in 34.6% of ovarian serous adenomas by IHC.[18] Different sensitivities of IHC and flow cytometry can explain higher expression rate in this study. Moreover, CLDN6 expression is reported at a frequency of 7–100% in other solid tumors such as lung,[34] endometrial,[17] gastric,[35-37] and testicular cancers.[20]

Despite the frequent expression, spontaneous T-cell response against CLDN6 in peripheral blood was detected in only 1 out of 17 (6%) ovarian cancer patients. A patient-derived ovarian cancer cell line from the patient (P01) who showed spontaneous anti-CLDN6 T-cell response expressed significant level of CLDN6 (Figure 6). It is possible that high level of CLDN6 expression is required to induce spontaneous T-cell response. However, because CLDN6 expression levels of ascitic cells or tumor tissues from patients including P01 who provided PBMC for T-cell analyses were not investigated in this study, further analyses are required to determine the correlation between T-cell response and CLDN6 expression level. In addition, to define the immunogenicity of CLDN6 and anti-tumor role of CLDN6-specific immune response in cancer patients, analyzing more samples including tumor-infiltrating T cells is needed. In our previous study, we were able to detect NY-ESO-1-specific and neoantigen-specific T cells frequently in the tumor microenvironment.[38,39]

Limited induction of T-cell response against CLDN6 in cancer patients potentially indicates low immunogenicity of the antigen, which can be overcome by ACT of genetically engineered CLDN6-specific T cells. As CLDN6 is expressed on the cell surface of cancer cells, it is a promising target for antibody-based and CAR-T therapies.[16,21] An early phase clinical trial testing CLDN6-specific CAR-T cells for solid tumors is ongoing [ClinicalTrials.gov Identifier: NCT04503278]. Once safety and efficacy are demonstrated, CLDN6-specific CAR-T cell could be a promising therapy for cancer patients with solid tumors including ovarian cancer because of the frequent antigen expression. In comparison with CAR-T cells, TCR-engineered T cells have several potential advantages including (i) lower immunogenicity of TCR transgenes; (ii) general safety profile because it is derived from naturally occurring T cells responded to tumor or presented by antigen-presenting cells (APCs) in patients; and (iii) reactivation of memory T-cell response through APC,[40,41] which are critical for long-term in vivo maintenance of engineered T cells for immunosurveillance against recurrent tumors.

A strength of the present study is that we successfully identified TCR genes specific to CLDN6 from both CD8+ and CD4+ T cells and demonstrated specific reactivity to CLDN6+ cancer cells by TCR-transduced T cells. We also identified HLA-A2-binding epitope as CLDN6132–140 and CD8-TCR-transduced T cells were stained with tetramer. These CLDN6-specific TCR genes are considered to be safe because they spontaneously occurred and expanded in a patient. In addition, the CD8-TCR exhibited specific cytotoxic activity and is restricted by HLA-A2, which is the most frequent HLA class I allele in the Caucasian population; thus, this TCR could be further developed as off-the-shelf therapeutic TCR gene for clinical use. However, CD8-TCR in this study was found to be CD8-dependent because only TCR-transduced CD8+ T cells but not CD4+ T cells showed the reactivity against a peptide. T cells with higher affinity TCR could be identified in patients[8] or affinity enhancement by gene modification[42] although careful assessment of off-target toxicity is very important before testing in clinical trials. In contrast to CD8-TCR, CD4-TCR is considered as high avidity because the response was observed in a CD4 coreceptor-independent manner. The frequency of HLA-DR*04:04 is about 6% of the Caucasian population in the US (Allele Frequency Net Database). Moreover, because CD4-TCR showed direct recognition of DR4+CLDN6+ cancer cells, CD4-TCR-transduced T cells are expected to enhance CD8+ T-cell functions.[7] We also demonstrated that CD4 or CD8 co-ligation is required to obtain multi-cytokine producing ability, which will be considered to co-transduction of those molecules along with TCR genes for cell manufacturing.[43]

In summary, we for the first time demonstrated naturally occurring CD8+ and CD4+ T-cell responses against CLDN6 in cancer patients and successful generation of CLDN6-specific TCR-transduced T cells. Together with high expression rate of CLDN6 on cancer cells, our findings suggest that CLDN6 is a promising target for TCR-engineered T-cell therapy for ovarian cancer patients.

Click here for additional data file.