Breast and Ovarian Carcinoma Overexpress HLA-G, a Neglected Cancer Immunosuppressive Protein

Purpose: HLA-G binds to the inhibitory receptors of uterine NK cells and plays an important role in protection of fetal cells from maternal NK lysis. HLA-G also mediates tumor escape, but the immunosuppressive role is often neglected. These studies have focused on the examination of HLA-G expression in human breast and ovarian carcinoma and HLA-G immunosuppressive role in NK cytolysis. Methods: We examined HLA-G expression in breast and ovarian carcinoma cell lines by real time PCR, ELISA and immunofluorescent staining, and in frozen breast and ovarian carcinoma tissues by immunohistochemistry (IHC). We treated the breast cancer cell lines with anti-human HLA-G antibody or progesterone. Then, NK cytolysis was measured by using MTT assay. Results: We find breast and ovarian cancer cell lines increase the expression of HLA-G mRNA and protein, compared to normal cells. IHC shows that 100% of frozen breast and ovarian carcinoma tissues overexpress HLA-G protein. HLA-G IHC scores of breast and ovarian carcinoma are significantly higher than normal breast and ovarian tissues, respectively (both p < 0.01). Blocking HLA-G of the breast cancer cells by the antibody increases NK cytolysis. Progesterone upregulates HLA-G mRNA and protein of human breast cancer cell lines. The increased HLA-G expression by progesterone suppresses the NK cytolysis. Conclusion: Human breast and ovarian carcinoma overexpress HLA-G immunosuppressive molecules. Blocking HLA-G protein by antibody improves the cytolysis of NK cells against human breast cancer cell lines. In contrast, upregulation of HLA-G expression by progesterone impairs NK cytolytic function. Thus, HLA-G is a new immune checkpoint protein and potential cancer immunotherapeutic target.


Introduction
Human leukocyte antigen (HLA) molecules are essential for the immune recognition and subsequent immunosurveillance. Human immune system uses the HLAs to distinguish own protein from foreign proteins. HLA molecules bind to peptide fragments derived from pathogens or cancer and display them on the cell surface for recognition by the proper T cells. Immune system including T cells, B cells and macrophages is further activated. The consequences are that invaded pathogens and diseases are eliminated [1]. Impaired HLA class I (HLA-I) expression on cell surface is an early and frequent event of carcinogenesis [2] [3]. Total or partial loss of classical HLA class I expression has been reported in various human cancers [4] [5] [6]. Another HLA-mediated immune escape is that cancer cells overexpress non-classical HLA-I molecules such as HLA-G and HLA-E, which function as inhibitor ligands for immune-competent cells [7] [8].
HLA-G was first reported to be restrictedly expressed at the maternal fetal interface on cytotrophoblasts. HLA-G is a ligand for inhibitory receptors of uterine natural killer (NK) cells and associated with maternal-fetal tolerance [9]. In physiology, HLA-G also expressed in the immune-privileged tissues such as cornea, thymic medulla, pancreatic islets, erythroblasts and mesenchymal stem cells [10] [11] [12]. In pathological condition, HLA-G express is found in cancer, inflammatory, autoimmune disease and viral infection [13] [14] [15]. HLA-G gene has alternative splicing of the primary transcript to generate seven different isoforms, 4 membrane bound (HLA-G1 to G4) and 3 soluble (HLA-G5 to G7), all isoforms have a negative regulation on immune cells including NK cells, cytotoxic T lymphocytes (CTLs) and antigen-presenting cells (APCs), by binding to specific receptors [16] [17] [18]. The HLA-G promotor region has regulatory elements such as heat shock, progesterone and hypoxia-responsive elements, and unidentified responsive elements for IL-10, glucocorticoid and other transcription factors [19] [20] [21]. Progesterone is an immunomodulatory steroid hormone secreted both by corpus luteum and placenta and contributes to the immunosuppressive environment of the maternal-fetal interface [22]. The underlying mechanism is primarily mediated by progesterone-binding to an alternative progesterone response element (PRE) in the HLA-G promotor to induce HLA-G expression [19] [20]. Immunohistochemistry (IHC) has shown that many cancers overexpress HLA-G protein in cancer tissues. Lin et al reviewed HLA-G IHC in paraffin embedding tissues of thirty types of tumors and showed that cancers had 30% -75% HLA-G expression depending on cancer types. HLA-G expression was correlated with clinical advanced disease stage, metastasis and worse prognosis, indi- cating that HLA-G could promote tumor immune escape [13]. Paraffin embedding is thought to better preserve morphological details but can mask epitopes of antigen. The positive rates of HLA-G expression might be underestimated in paraffin embedding cancer tissues. In comparison to conventional checkpoint proteins such as PD1/PDL, CTLA4/B7, etc., the immunosuppressive role of HLA-G in cancer has been neglected [23]. In the present study, we examined HLA-G expression of frozen breast and ovary carcinoma tissues by IHC. We found 100% breast and ovary carcinoma tissues overexpressed HLA-G protein.
HLA-G expression of normal and cancer cell lines was also studied by real time polymerase chain reaction (qPCR), enzyme-linked immunosorbent assay (ELISA) and immunofluorescence. HLA-G mRNA and protein were increased in cancer cell lines. To determine the immune inhibitory role of HLA-G in cancer, we blocked HLA-G protein of breast cancer cell lines by anti-human HLA-G monoclonal antibody and induced HLA-G expression by progesterone, and found the specific antibody improved the NK cytolysis to the breast cancer cells and the upregulation of HLA-G expression impaired the cytolytic function of NK cells.

Cell Culture
Human immortalized, untransformed mammary epithelial line MCF-12A was obtained from American Type Culture Collection (ATCC, Rockville, MD, USA). MCF-12A cells were cultured in a 1:1 mixture of Ham's F12 medium and DMEM containing 0.1 µg/mL cholera enterotoxin, 10 µg/mL insulin, 0.5 µg/mL hydrocortisone, 20 µg/mL epidermal growth factor, and 5% horse serum (Sigma Chemical Co., St. Louis, MO, USA). Human primary dermal fibroblasts, breast carcinoma cell lines MCF-7, T47D and MDA-MB-231, and ovarian carcinoma cell lines OVCAR-8 and TOV-112D, and NK-92MI cells were obtained from ATCC. MCF-7 and T47D are estrogen receptor (ER) and progesterone receptor (PR) positive, but MDA-MB-231 is ER and PR negative. Fibroblasts and all cancer cell lines were grown in alpha-MEM supplemented with 10% fetal bovine serum (GIBCO Invitrogen, Carlsbad, CA, USA). NK-92MI is an interleukin 2 (IL-2) independent NK cells derived from the NK-92 cell line by transfection with human IL-2 cDNA. The cell line is cytotoxic against a wide range of malignant cells [24]. NK-92MI cells were grown in alpha-MEM with 0.2 mM inositol, 0.02 mM folic acid, 0.1 mM 2-mercaptoethanol, 12.5% horse serum (Sigma Aldrich, St. Louis, MO, USA) and 12.5% fetal bovine serum. All cell lines were maintained in the media without supplement of any antibiotics.

qPCR
We used qPCR to measure mRNA expression of HLA-G. The detailed protocol was published previously [25]. In brief, total RNA was isolated by PurLink RNA Kit (Invitrogen, Carlsbad, CA, USA). cDNA was synthesized by the High Capacity RNA-to-cDNA kit (Applied Biosystems, Grand Island, NY, USA). Gene ex-  Cell homogenates and total RNA were obtained. Protein concentration of homogenates was measured by BCA protein assay (Thermo Scientific, Rockford, IL, USA). Levels of HLA-G mRNA were determined by qPCR. HLA-G protein in homogenates was measured by HLA-G ELISA kit (Biomatik, Delaware, USA). The procedure was referred to the protocol of kit provided by manufactory. HLA-G concentration (pg/ml) was calibrated by total protein concentration (mg/ml). The final HLA-G concentrations were presented as pg/mg protein [HLA-G concentration by ELISA (pg/ml) ÷ total protein concentration (mg/ml)]. All the experiments were carried out in triple wells and repeated four times independently.

Immunofluorescent Staining
10,000 cells were cultured on 35 mm glass bottom cell culture dishes (Stellar Scientific, Owings Mills, MD, USA) overnight. At next day. The cells were fixed 15 minutes by 4% paraformaldehyde solution in phosphate-buffered saline (PBS). The staining procedure was as follows: washing 3 times with PBS, blocking 1 hour at room temperature (RT) with 5% goat serum PBS, washing 2 times, incubating cells with 1:200 rabbit anti-human HLA-G polyclonal antibody (Sigma Aldrich, St Louis, MO, USA) at 4˚C overnight, washing 4 times, incubating cells with 1:500 goat anti-rabbit IgG (H + L)-Alexa 488 (Invitrogen, Carlsbad, CA, USA) for 60 minutes at RT, and washing 4 times. Antibody diluent without HLA-G antibody was added into the dishes as blank control. The cells were observed by Olympus IX83 fluorescent microscope (Tokyo, Japan). Five pictures were taken in different areas of each dish. All pictures were taken with 1 second exposure. Fluorescent intensity was determined by software Cellsense 1.16. All the experiments were repeated three times independently.

IHC of Frozen Tissues
Frozen slides of human breast and ovarian normal and carcinoma tissues were purchased from BioChain (IRB# IORG0006917, Newark, CA, USA). Slides were fixed with pre-chilled acetone for 15 minutes, dried completely for 30 minutes at RT under airflow, and rinsed 3 times with PBS. HLA-G staining procedure was referred from the protocol of HRP/DAB (ABC) Detection IHC kit (Abcam, Cambridge, MA, USA). The procedure was briefly as follows: blocking non-specific antibody binding and endogenous peroxidases with 5% goat serum PBS and hydrogen peroxide blocking reagent respectively, washing with PBS, incubating tissues with 1:200 rabbit anti-human HLA-G polyclonal antibody at 4˚C overnight, washing with PBS, incubating slides with 1:2000 goat anti-rabbit IgG (H + L)-biotin (Invitrogen, Carlsbad, CA, USA) at RT for 60 minutes, washing with PBS, incubating with strapavidin-perioxidase complex for 10 minutes at RT, washing with PBS, and developing stain with DAB reagent for 3 minutes at RT. Blank controls were covered by antibody diluent without rabbit anti-human HLA-G. The slides were counterstained with hematoxylin, dehydrated and covered with coverslips. HLA-G staining was evaluated by Klein et al semi-quantitative scoring system [26] [27] (Table 1).

Statistical Analysis
Data were analyzed by t-test and presented as mean ± standard deviation (SD) (N, sample size). Findings were considered significant at p < 0.05.

Progesterone Upregulates HLA-G mRNA and Protein of Normal Breast and Cancer Cell Lines
Progesterone binds to progesterone receptor (PR) and glucocorticoid receptor (GR). The affinity of progesterone to PR is much higher than to GR [29].

Breast and Ovarian Carcinoma Tissues Overexpress HLA-G Protein
HLA-G protein is detected by IHC in 15 frozen normal breast and 44 breast carcinoma tissues. Except the pathological diagnosis, other patient clinical information has not been released by the tissue suppliers. All breast carcinoma tissues (100%) have positive HLA-G brown staining in the cytoplasm. Breast carcinoma increases HLA-G expression ( Figure 5). HLA-G staining scores of breast carcinoma tissues are significantly higher than that of normal breast tissues (p < 0.01) (  Figure 6).

Blocking HLA-G Protein Increase Natural Killer Cytolysis
The

Upregulated HLA-G of Breast Cancer Cells by Progesterone Inhibits the Cytolysis of NK-92MI
We pre-treated MCF-7 and T47D with 0.64 µM progesterone, and MDA-MB-231 with 3.2 µM progesterone for 24 hours. We tested the cytolysis of NK-92MI to these 3 cell lines. Progesterone 0.64 µM and 3.2 µM don't alter the growth of breast cancer cells. Table 5 shows pre-treatment with progesterone inhibits the cytolysis of NK-92MI to the breast cancer cell lines, compared to controls without pre-treatment (all p < 0.05, paired t-test). The pre-treatment with the mix-ture of progesterone and 10 µg/ml anti-HLA-G antibody doesn't significantly increase or decrease the NK cytolysis to the 3 breast cancer cell lines (all p < 0.05, unpaired t-test, compared to the non-treated cells, N = 12). The blocking HLA-G protein by 10 µg/ml anti-HLA-G can neutralize the inhibitory effect of progesterone on NK cytolysis (Figure 8). These suggest the increased HLA-G by progesterone inhibits the cytolytic function of natural killer cells to breast cancer cells.

Discussion
The whereas HLA-G-expressing tumors grew. Blocking of HLA-G by a specific neutralizing antibody prevented the growth of HLA-G-expressing tumors, providing the proof of concept for new antitumor therapeutic strategy [35]. In the present study, we confirm breast and ovarian carcinoma cell lines overexpress HLA-G mRNA and proteins, compared to human primary fibroblasts and normal mammary epithelia (Table 2, Figure 1 and Figure 2). In frozen tissues, 100% of human breast and ovarian carcinoma overexpress HLA-G. HLA-G is significantly upregulated in breast and ovarian carcinoma tissues, compared to normal breast and ovarian tissues, respectively (Table 3 and Figure 5). Lin et al reviewed HLA-G immunohistochemistry studies in paraffin-embedding tissues of thirty types of tumors. There were 30% -75% HLA-G expression depended on cancer types [13]. The rate of HLA-G expression was apparently underestimated in paraffin-embedding tissues. The process of paraffin-embedding of tissues might cause the loss of HLA-G antigenicity or mask the epitopes of HLA-G.
Breast carcinoma is associated with progesterone. In a case-cohort study of postmenopausal women, elevated circulating progesterone levels were associated with a 16% increase in the risk of breast cancer [36]. The underlying mechanism might include upregulation of HLA-G by progesterone on normal mammary epithelia. Progesterone binds to an alternative progesterone response element (PRE) in the HLA-G promotor to induce HLA-G expression [19] [20]. We find progesterone increases the HLA-G expression of normal mammary epithelial MCF-12A and 3 breast cancer cell lines (Figure 4 and Figure 5). The upregulation of HLA-G expression may lead to immune suppression which promotes breast tumorigenesis and cancer prognosis. Progesterone hormone therapy has been used to treat symptoms of menopause and some types of cancer [37] [38]. Our results in the present study suggest that caution should be taken to observe the effect of progesterone on HLA-G and host anti-tumor immunity when we give progesterone hormone therapy for cancer patients.
NK cells play major roles in first-line innate immunity against viral infections, tumorigenesis, and tumor growth and progression. NK cells possess a combination of activating and inhibitory receptors. In humans, major activating receptors involved in target cell killing are the natural cytotoxicity receptors (NCRs) and NKG2D. Activating receptors recognize ligands that are overexpressed or expressed de novo upon cell stress, viral infection, or tumor transformation. The HLA-I-specific inhibitory receptors constitute a fail-safe mechanism to avoid unwanted NK-mediated damage to healthy cells [39]. HLA-G binds to the inhibitory receptors on NK cells to inhibit NK cytolysis in human pregnancy and cancers [9] [30] [40]. We find the blocking HLA-G on breast cancer cells improves anti-tumor function of NK-92MI. In contrast, increased HLA-G expression of breast cancer cells by progesterone impeded the tumor lysis of NK-92MI.
At least in vitro, these data provide the proof of principle of blocking HLA-G for new antitumor therapeutic strategy. On the other side, blockade of inhibitory receptors, the HLA-G ligands, on NK and T cells is another new cancer immunotherapeutic strategy. It has been reported that blocking and downregulating ILT2, an HLA-G ligand, by Lenalidomide restore NK cell function in chronic lymphocytic leukemia [41].