Patients (n = 19) have been enrolled from the Department of Gynecology and Obstetrics, Faculty Hospital Brno. The study was approved by the local institutional ethics committee of the Faculty of Medicine, Masaryk University. The study was performed in accordance with the Declaration of Helsinki. Written informed consents were obtained from all patients. Endometriosis was assessed according to the revised American Fertility Society (r-AFS) classification during laparoscopy. Patients received no hormonal therapy for a minimum of 3 months prior to laparoscopic surgery.

The patient characteristics are shown in Table 1 .

Peripheral blood (PB) and peritoneal fluid (PF) samples were obtained from endometriosis patients and were processed within 2 h of collection. PF samples were taken during the planned surgery from lower pelvis cavity by fine needle suction from cavum Douglasi at the opening phase of diagnostic laparoscopy prior the surgical procedure as less invasive technique for more patients than tissue biopsy. At the same time, it allows to obtain sufficient volumes of biological material for subsequent analysis. Buffy coats from age- and sex-matched healthy volunteers (n = 33) were collected at the Transfusion and Tissue Bank, Faculty Hospital Brno. Peripheral blood mononuclear cells (PBMCs) were isolated from blood by density gradient centrifugation using Lymphoprep (Stem Cell Technologies) following the manufacturer’s recommendations. Plasma samples were collected and stored at −80°C.

Endometrial carcinoma cell lines KLE (ATCC^® CRL1622™) and RL95-2 (ATCC^® CRL1671™) were purchased from ATCC (American Type Culture Collection, VA, USA). Ishikawa cell line was purchased from Sigma-Aldrich (St. Louis, MO, USA). KLE and RL95-2 cell lines were maintained in Dulbecco’s modified Eagle’s medium (DMEM/F12) supplemented with 10% fetal bovine serum (FBS) and 2% penicillin/streptomycin (all Thermo Fisher Scientific); RL95-2 cells with the addition of 5ug/ml of insulin (Sigma-Aldrich, MO, USA). The Ishikawa cell line was maintained in MEM supplemented with 5% FBS and 2% penicillin/streptomycin.

In addition, human tumor cell lines including myeloma (U266, EJM) and chronic myeloid leukemia (LAMA-84, KYO-1) were purchased from DSMZ (German Collection of Microorganisms and Cell Cultures GmbH, Germany). Renal (A-498), prostate (DU-145), breast adenocarcinoma (MCF-7) and histocytic lymphoma (U937), chronic myeloid leukemia (K562), acute monocytic leukemia (THP-1) and glioblastoma (U87 MG) cells were purchased from ATCC. Glioblastoma cells (U-373 MG and U251 MG) were purchased from The European Collection of Authenticated Cell Culture (ECACC). Myeloma cell lines (OPM-2, LP-1, KMS-11) were a kind gift from Dr. Krejci (Institute of Biology, Masaryk University Brno). Cells lines (U266, LAMA-84, KYO-1, K562, U-937, THP-1, EJM) were cultured in RPMI-1640 containing 10% FBS, 2mM L-glutamine, and 2% penicillin/streptomycin. Cell lines (A-498, DU-145, HT-29, U-87 MG, MCF-7) were cultured in modified Eagle’s medium (MEM) (Sigma Aldrich) with 10% FBS, 2 mM l-glutamine, and 2% penicillin/streptomycin. The MCF-7 cells were supplemented with nonessential amino acids (NEAA, Sigma Aldrich). Cell lines (U251 MG, U-373 MG) were cultured in DMEM/F12 with 10% FBS. All cells were grown at 37°C in 5% CO₂ atmosphere up to 70%–80% confluence; adherent cells were harvested by using gentle dissociation solution TrypLE (Gibco, Thermo Fisher Scientific) and counted by using Trypan blue exclusion.

The EphA2 inhibitor ALW-II-41-27 was purchased from MedChem Express (Monmouth Junction, NJ, USA). It was dissolved in sterile DMSO at 10 mM stock concentration and solution stored in aliquots at −20°C.

The cell phenotype was assessed by fluorescence-activated cell sorting (FACS) by using staining with the monoclonal antibodies MICA, MICB, CD112, CD155, B7-H6 (R&D Systems, clone 875001), and EphA2 (R&D Systems, clone 371805). The tumor cell lines were harvested, washed with cold phosphate-buffered saline (PBS, Sigma) containing 2% FBS, and incubated for 30 min on ice with fluorescently labelled monoclonal antibodies. Gamma-delta T cells were identified in freshly isolated PBMCs labelled with CD3 (Thermo Fisher Scientific, clone SK7), Vδ1 TCR (Thermo Fisher Scientific, clone TS8.2), Vδ2 TCR (BD Pharmingen, clone B6) or Vδ2 TCR (Sony, clone B6). CD27 (BD Pharmingen, clone M-T271), and CD45RA (Exbio, clone MEM-56) were used for immunophenotyping. Samples were washed and acquired using FACSCanto^® (BD Biosciences) and data analyzed using FlowJo^® software (FlowJo, Ashland, OR, USA). Forward and side scatter gating were used to discriminate live cells from dead cells and γδ T cells were derived from SSC vs. FSC-gated bulk PBMCs with doublet exclusion (FSC-A vs. FCS-H). To determine the placement of the gates, appropriate fluorescence minus one (FMO) and unstained controls were used.

Fresh γδ T-cell populations were sorted by positive selection using anti-TCR Vδ1 (Beckman Coulter, clone R9.12) or anti-TCR Vδ2 (BD Pharmingen, clone B6) monoclonal antibodies and magnetic microbeads (Miltenyi Biotec, Germany) according to manufacturer’s instruction. The cell purity was routinely greater at 97%.

Freshly sorted Vδ1 or Vδ2 γδ T lymphocytes were incubated with tumor target cells at indicated effector:target (E:T) 5:1 and 10:1 ratios in duplicates for 4 h co-culture at 37°C as described previously (31). Briefly, tumor target cells were washed in Hank’s buffered saline solution (HBSS, Invitrogen Life Technologies) to remove FBS and culture media. Cells were resuspended in diluent C (Sigma) and labeled with PKH67 fluorescent dye (Sigma). To-Pro-3 iodide (1 μM in PBS) (Invitrogen Life Technologies) was added immediately prior to the acquisition on the flow cytometer. At least 10,000 target cells were acquired after gating out the green fluorescence of PKH67 dye and the proportion of To-Pro-3 iodide positive cells. Background target cell death was determined from the cells incubated in the absence of effector cells. In the blocking experiments, the EphA2 antibody (R&D Systems, clone 371805) and ALW-II-41-27 EphA2 inhibitor (10 µM, 1 µM) or DMSO as a control were added to tumor cultures prior the cytotoxicity assays.

Total RNA has been extracted from tumor cell lines using RNeasy Mini kit (Qiagen) according to manufacturer’s instruction. RNA was eluted in RNAse-free water and stored in −80°C. Complementary DNA (cDNA) has been synthesized using 20 ng/µl total RNA that has been reverse transcribed using High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems). The glyceraldehyde 3-phosphate dehydrogenase (GAPDH) housekeeping gene has been used as an internal control by quantitative real-time polymerase chain reaction (real-time qPCR). cDNAs were amplified using TaqMan^® Gene Expression Assay (ID : Hs01072272_ml, Applied Biosystems). Samples were analyzed on StepOne™ Real-Time PCR Systems (Applied Biosystems).

The EPHA2 gene knockout was performed with the EPHA2 CRISPR gRNA + Cas9 in Lenti-particles (supplied from antibodies-online GmbH) and used closely following manufacturer’s instructions. Briefly, vector pLenti-U6-sgRNA-SFFV-Cas9-2A-Puro (product number ABIN5252263) was used to generate EPHA2 knockouts in human endometrial cancer line RL95-2, renal carcinoma cell line A-498, and colon carcinoma cell line HT-29. After infection, positive clones were selected by 3.5 μg/ml puromycin, and the single clones were transferred separately into 48-well plates and further passaged. The EPHA2 knockouts were confirmed by flow cytometry after antibody staining (anti-EPHA2, R&D Systems, clone 371805).

Data analyses were performed using GraphPad Prism5 software (GraphPad Software Inc., La Jolla, CA). The Student’s t-test was used to determine significant differences between groups. Differences between sample groups were evaluated with the nonparametric Mann-Whitney U test. p < 0.05 values were considered to be significant. Data are expressed as mean ± standard deviation (SD).

First, we aimed to determine the two major populations of γδ T cells (Vδ1 and Vδ2 subsets, respectively) in peripheral blood (PB) samples from patients with endometriosis (n = 19) and compared the frequencies with age- and sex-matched healthy donors (HD, n = 33). Flow cytometric analysis of peripheral blood mononuclear cells (PBMCs) where proportion of Vδ1 and Vδ2 γδ T cells among leukocyte gate followed by the percentage of CD3 lymphocytes is shown in Figure 1A . Immunophenotyping of Vδ1 ( Figure 1B ) and Vδ2 ( Figure 1C ) using the CD27 and CD45RA antibodies to determine the naïve/memory/effector memory and TEMRA phenotypes was analyzed, and representative flow plots are shown. We found significantly low percentages of Vδ1 T cells in PB (p = 0.008) (median 0.5%, range 0.1%–2.4%) in endometriosis patients compared with HD (median 0.9%, range 0.1%–3.8%), as shown in Figure 2A , whereas Vδ2 T cells showed no difference between the endometriosis patients (median 1.5%, range 0.2%–7.9%) and healthy controls (median 2.4%, range 0.3%–11.9%), as shown in Figure 2B . Next, the absolute counts of Vδ1 and Vδ2 γδ T cells in PB in patients compared with HD were determined in Figures 2C, D , respectively. We found dramatically reduced Vδ1 T-cell absolute counts in patients PB (p = 0.0002) (median 2.23 cells/µl, range 0.14–14.01) and HD controls (median 13.3 cells/µl, range 0.11–242.1). No differences in Vδ2 T-cell counts were observed between the patients PB (median 9.8 cells/µl, range 0.12–118.7) and HD controls (median 17.3 cells/µl, range 0.11–198).

Second, we analyzed γδ T-cell infiltration in patient’ peritoneal fluid (PF) samples and compared the counts with paired PB. We found most patients with dramatically reduced Vδ1 T cells in PF (median 0.1%, range 0%–2.8%) compared with PB samples (median 0.5%, range 0.1–2.4%) (p = 0.001) in Figure 2E . Similarly, no significant differences were identified for circulating Vδ2 T cells (median 1.5%, range 0.2%–7.9%) compared with PF samples (median 2.0%, range 0%–11.5%), in Figure 2F . These results show for the first time the presence of Vδ1 and Vδ2 γδ T cells in peritoneal fluid in patients with endometriosis.

Third, we found most Vδ1 T cells of naïve (CD27+CD45RA+) and TEMRA (CD27-CD45RA+) phenotype in patients PB shown in Figure 3A . Peritoneal fluid samples showed majority of Vδ1 T cells at the memory stage of differentiation (CD27+CD45RA-) in Figure 3B . Vδ2 T cells in patients PB and PF samples were predominantly of memory phenotypes in Figures 3C, D .

We analyzed the cytotoxic function of Vδ1 and Vδ2 γδ T cells freshly sorted from healthy donors against endometrial tumor cell lines including Ishikawa, KLE, and RL95-2. We determined the 4-h killing reactivity shown as percentages of specific lysis of Vδ1 and Vδ2 γδ T cells at 5:1 and 10:1 E:T ratio. All of the tested γδ T lymphocytes efficiently killed the tumor targets. First, the summary of Vδ1 T cell-mediated killing (n = 4) at 5:1 E:T against KLE (mean 29.8%, SD 4.2%), RL95-2 (mean 28.4%, SD 12.7%) and Ishikawa (n = 3, mean 25.1%, SD 3.8%) in Figure 4 is shown. Importantly, significant antiendometrial reactivity of Vδ1 T cells was detected at 10:1 E:T against KLE (mean 34.8%, SD 1.3%), RL95-2 (35.9%, SD 14.2%), and Ishikawa (mean 37.8%, SD 4.3%).

Second, summary of Vδ2 T-cell-mediated killing against KLE, RL95-2, and Ishikawa ( Figure 4) is shown. At low E:T ratio of 5:1, the specific lysis was detected against KLE (n = 5, mean 44.4%, SD 10.3%), RL95-2 (n = 4, mean 26.0%, SD 8.6%), and Ishikawa (n = 4, mean 34.4%, SD 6.4%). Prominent ability of Vδ2 γδ T cells to recognize and kill endometrial tumor targets was observed at 10:1 E:T against KLE (mean 49.8%, SD 11.1%), RL95-2 (mean 33.2%, SD 8.8%) and Ishikawa (mean 40.7%, SD 10%). Altogether, the endometrial tumor killing was comparable for γδ T-cell subsets isolated from different donors and was reproducible between the assays for all cell lines.

To elucidate possible mechanisms involved in γδ T cell cytotoxicity, we evaluated several molecules typically involved in γδ T-cell killing including the MICA and MICB as ligands for the NKG2D receptor; CD112 and CD155 as ligands for the DNAM-1 receptor; and ligand B7-H6 for the NKp30 receptor. We analyzed the surface expression of MICA, MICB, CD112, CD155, and B7-H6 on target endometrial tumor cell lines including Ishikawa, KLE and RL95-2 by flow cytometry. Variable expression of these markers is shown in Figure 5A . Furthermore, we analyzed the expression of the EphA2 receptor, which is known to be overexpressed in many human malignancies, including endometrial carcinoma. We showed activation and high expression of the EphA2 receptor on KLE and RL95-2 endometrial tumor cell lines but only weak expression on Ishikawa cells in Figure 5A . In addition, we determined the EphA2 RNA expression by the real-time qPCR in a panel of tumor cell lines as fold gene expression in Figure 5B . We found the highest EphA2 expression in solid tumors including prostate (DU-145), colon (HT-29), and renal (A-498) carcinoma cell lines in contrast to mostly negative hematological cell lines.

Next, we aimed to determine whether the EphA2 is involved in γδ T-cell killing. In the blocking experiments, we first preincubated the target cells with the EphA2 antibody prior to 4-h cytotoxicity assays and then analyzed the specific lysis of KLE ( Figure 6A ) and RL95-2 ( Figure 6B ) target cells in the presence/absence of the EphA2. Interestingly, all Vδ1 T cells isolated from healthy donors killed efficiently both tumor cell lines at 5:1 and 10:1 E:T ratios; however, the cytotoxicity was reduced when the EphA2 receptor was blocked. The inhibitory effect of anti-EphA2 on tumor cell killing was determined for KLE cells in the range 14%–40% (median 25%) and for RL95 cells in the range of 15%–40% (median 26%). Together, these results suggested that EphA2 was recognized by cytotoxic Vδ1 T cells in the tumor killing.

To validate these findings, we then tested the in vitro effects of the EPHA2 small-molecule inhibitor ALW-II-41-27 on KLE endometrial tumor cells. We incubated the KLE target cells with ALW-II-41-27 inhibitor at 1 and 10 µM concentrations and showed the specific lysis significantly reduced at both 5:1 and 10:1 E:T ratios in the range 50%–80% (median 71%) ( Figure 6C ). Importantly, the inhibition of cytotoxicity by Vδ1 T cells was shown as dose dependent. Next, we used the EphA2-negative Ishikawa cell line and determined the specific lysis at 10:1 E:T in the presence of ALW-II-41-27 (1 µM). No significant reduction of tumor killing was observed ( Figure 6D ). Of note, the pharmacological effect of inhibitor ALW-II-41-27 on cell viability in drug treatment sample relative to a DMSO control group was determined independently prior the killing assays and no increase of the spontaneous cell lysis was detected after 4h (data not shown). In addition, the inhibitor ALW-II-41-27 showed no change in the surface expression of EphA2, B7-H6 and stress ligands in 4-h cytotoxicity incubation (data not shown).

Finally, to confirm the inhibition of the EPHA2 resulting in significantly decreased in vitro tumor cell death, we generated EPHA2 knockouts (KO) in endometrial cell line RL95-2 by the CRISPR/Cas9 method. The loss of EPHA2 significantly reduced specific lysis by 35%–90% (median 45%) by Vδ1 T cells in knockout versus wild type (WT) at 5:1 E:T ratio is shown in Figure 7A . Next, we observed the Vδ1 T-cell cytotoxicity inhibition of RL95-2 WT versus KO cells and also in the addition of ALW-II-41-27 inhibitor (1 µM) at 10:1 E:T in Figure 7B .

To further verify these results, we generated EPHA2 knockouts in A-498 (renal) and HT-29 (colon) tumor cell lines which had previously showed the highest EphA2 expression in Figure 5B . The summary of Vδ1 T cell cytotoxicity results is shown for A-498 in Figure 7C and HT-29 in Figure 7D . The A-498 KO cells showed significant protection from specific lysis mediated by Vδ1 T cells compared with WT cells at 10:1 ratio between 40% and 77% (median 61%). Similarly, HT-29 KO cells presented significant reduction of tumor killing than WT cells at 10:1 ratio between 42% and 75% (median 44%).

In summary, we evaluated Vδ1 γδ T-cell cytotoxicity against tumor cells and found consistently that EphA2 expressed on cancer cells show susceptibility to cell lysis by tumor-reactive Vδ1 T cells.