Microenvironment and Immunology Induction of Immunoregulatory CD271þ Cells by Metastatic Tumor Cells That Express Human Endogenous Retrovirus H

Human endogenous retroviruses (HERV) are associated with many diseases such as autoimmune diseases and cancer. Although the frequent expression of a variety of HERVs in tumor cells has been demonstrated, their functional contributions in cancer are as yet unclear. Intriguingly, HERVs and other retroviruses include an immunosuppressive domain in their transmembrane envelope proteins, but its mechanism of action and cancer relevance are obscure. In this study, we demonstrate that the human endogenous retrovirus HERV-H has a critical role in tumor metastasis and immune escape. We found that expression of herv-hmRNA was elevated in metastatic tumor cells undergoing epithelial-to-mesenchymal transition (EMT) and in primary tumor tissues from advanced colon cancer. The immunosuppressive peptide H17 derived from HERV-H was sufficient to induce EMT in tumor cells that expressed low levels of HERV-H, and it amplified this event within the tumor microenvironment. H17 also stimulated CCL19 expression in tumor cells, which in turn recruited and expanded a population of pluripotent immunoregulatory CD271þ cells, which included mesenchymal stem cells and myeloid-derived suppressor cells. In tumor tissues from patients with advanced colon cancer, we confirmed that CD271þ cells were increased in HERV-HþCCL19þ tumor tissues. Notably, RNAi-mediated change of HERVH or CCL19, or depletion of CD271þ cells, improved immune responses in vitro and in vivo accompanied by tumor regression. Together, our results argued that HERV-H is a critical determinant of immune escape in cancer, suggesting its candidacy as a promising therapeutic target to treat patients with advanced cancer. Cancer Res; 74(5); 1361–70. 2014 AACR. Introduction Human endogenous retroviruses (HERV) are derived from ancestral integration of exogenous retroviruses, and are transmitted vertically through the germ line (1, 2). HERVs are associated with autoimmune diseases such as rheumatoid arthritis, multiple sclerosis, and cancer (1, 2). Increase of various HERV expressions such as HERV-H (3, 4), HERV-K (5), HERV-F (6), HERV-R (7), and HERV-S (8) in human tumor cell lines and tumor tissues of patients has been demonstrated so far. However, the meanings of HERV emerge in cancer, or the functional role of HERVs in tumor cells has been rarely pursued. HERVs and other retroviruses have immunosuppressive properties, for which immunosuppressive domain of the transmembrane envelope protein is responsible (9, 10). Stimulation with a 17-mer peptide derived from this domain upregulates immunosuppressive interleukin (IL)-10 production, and downregulates immunoactivator IL-12 production in peripheral blood mononuclear cells (PBMC; refs. 11, 12). However, the precise immunosuppressive mechanism remains unclear. Also, its effect on the tumor side has never been investigated. We assumed that tumor-derived peptide would necessarily modulate both immune cells and tumor cells within tumor microenvironment. In this study, we investigated the role of HERV-H/env60 (designated HERV-H) expressed in tumor cells, and demonstrate a novel molecular mechanism involved in immunoevasion of cancer. Materials and Methods Cell lines and mice Human cell lines (colon cancer Colo320 and HCT116, pancreatic cancer MIAPaca and Panc1, melanoma Hs294T, and normal epithelial cell ARPE19) were purchased from American Type Culture Collection, and were authenticated by short tandem repeat profiling before experiments. Primary tumor cell lines were established using tumor tissues surgically resected from a patient with stage IV melanoma under the protocol approvedby theKeioUniversity EthicsCommittee. These tumor cells were cultured in 10% fetal calf serum/Dulbecco's Modified Eagle Medium. Female BALB/c nu/nu mice were purchased fromSLC, andweremaintainedunderpathogen-free conditions. These mice were used according to the protocols approved by theAnimalCare andUseCommittee atKeioUniversity School of Medicine (Tokyo, Japan). Authors' Affiliation: Division of Cellular Signalling, Institute for Advanced Medical Research, Keio University School of Medicine, Tokyo, Japan Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/). Corresponding Authors: Chie Kudo-Saito, Division of Cellular Signaling, Institute for Advanced Medical Research, Keio University School of Medicine, 35 Shinanomachi, Shinjuku-ku, Tokyo 160-8582, Japan. Phone: 813-5363-3778; Fax: 81-3-5362-9259; E-mail: [email protected]; andYutaka Kawakami, E-mail: [email protected] doi: 10.1158/0008-5472.CAN-13-1349 2014 American Association for Cancer Research. Cancer Research www.aacrjournals.org 1361 on May 3, 2017. © 2014 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from Gene expression analysis ThemRNA expression in tumor cells or tissues was analyzed by semiquantitative reverse transcription PCR (RT-PCR) using extracted RNAs and paired primers (Supplementary Data) specific for open reading frame (ORF) region within the targeted genes as described before (13). The digital images of the bands were quantified using NIH ImageJ software (http://rsb.info.nih.gov/ij/), and the signal intensity was normalized to gapdh expression. The same experiments were repeatedly conducted two to four times, and all data are compiled and averaged in graphs (mean SD). Functional analysis of tumor cells in vitro and in vivo Tumor cell functions were evaluated as described before (13). Briefly, cell proliferation was assessed by WST1 assay (Takara). Cell invasion was assessed using a Transwell chamber with a Matrigel-coated membrane (BD Biosciences). Cell adhesion was assessed using Fibronectin-coated 96-well plates (BD Biosciences). Chemokines in the tumor supernatant (culture for 1–2 days) were measured using ELISA kits according to the manufacturer's instructions (CCL2; Endogen; CCL19; R&D Systems). To knockdown herv-h, snail, or twist expression, several kinds of siRNAs (3 mg; Invitrogen) targeting a different position of each gene, and its scrambled oligonucleotide as a control were used after making complex with jetPEI (PolyPlus Transfection). For herv-h, four kinds of siRNAs (targeting 5480, 6794, 6965, and 7057) were used. The sequences of these herv-h–specific siRNAs are described in Supplementary Data. Transfection efficacy was validated by RT-PCR as described before (13). For in vivo study, HERV-Hþ tumor cells transfected with lentiviral vector encoding GFP gene (Biogenova) were used. These tumor cells were injected both intravenously (for systemic metastasis; 1 10 cells) and subcutaneously (for draining lymph node analysis; 1 10 cells) into BALB/c nu/nu mice. Three weeks later, lymph node, lung, and spleen were harvested from the mice, and were mechanically dispersed into single-cell suspensions for flow cytometric analysis of GFPþ tumor cells and CD271þ cells. To evaluate effect of a 17-mer peptide derived from the immunosuppressive domain ofHERV-H on tumor cells, HERVH low tumor cells were stimulated withHERV-H454–470 peptide (LQNRRGLDLLTAEKGGL, designated H17) or another site peptide as a control (HERV-H81–97, LQTDWGTSPVSPHLRTS) at 5 mg/mL, or TGF-b at 5 ng/mL (R&D Systems) as an EMT (epithelial-to-mesenchymal transition)-inducer control for 2 to 3 days before assays. These peptides with amidation at Cterminal were synthesized at Invitrogen (purity >98%), and were dissolved in sterile MilliQ water for use. We confirmed by native PAGE that these peptides formed dimers in the preparation for cell culture in comparing with a loading control without peptides (medium only). The fifth glutamin of the HERV-H454-470 peptide was substituted with arginin, because arginin-arginin (RR) has been reported to play a crucial role in the biologic activity of the immunosuppressive domain using carrier-conjugated peptides (12), although we observed no differences of biologic activities between them at least in our study. For signaling pathway analysis, cells were treated with phosphoinositide 3-kinase (PI3K) inhibitor LY294002 or MAP– ERK kinase (MEK) inhibitor PD98059 at 20 mmol/L before assay. To define EMT induction by H17 peptide, H17-treated cells were permeabilized with cytofix/cytoperm solution (BD Pharmingen), and were stained with monoclonal antibodies (mAb) specific for Twist (Abcam), E-cadherin (BD Biosciences), Fibronectin (R&D Systems), CD44 (BD Biosciences), or the appropriate isotype control Ig, followed by staining with immunofluorescence-conjugated secondary antibodies. The molecular expressions were analyzed using a confocal LSM5 Pascal microscope (Carl Zeiss). Stimulation and characterization of human PBMCs Human PBMCs were stimulated with 3-day–cultured tumor supernatant (0.22-mm filtration, 25%), CCL19 (10 ng/mL), and/or H17 peptide (10 mg/mL) for 5 days, and were analyzed by flow cytometry. For tracking CD271þ cell division, CD271þ cells were separated from PBMCs using a MACS system with MicroBeads-conjugated mAbs (Miltenyi Biotec), and were stained with red fluorescent dye PKH26 (Sigma) before stimulation. The immunosuppressive activity of the treatedCD271þ cells was evaluated using the T-cell proliferation system. CD271þ cells (inactivation with mitomycin C, MMC) were cocultured (1:10) with autologous CD3þ T cells in the presence of anti-CD3mAb (1mg/mL) for 5 days, and cell proliferationwas measured by the WST1 assay. The pluripotency of the CD271þ cells was also evaluated as described in Supplementary Data. CD271þ cell depletion studies To evaluate effect of CD271þ cells on dendritic cells (DC), purified CD14þ cells before/after CD271þ cell depletion were cultured with granulocyte macrophage colony–stimulating factor (GM-CSF; 10 ng/mL; PeproTech) in the presence of CCL19 and H17 for 6 days, and were analyzed for CD11cþ HLA-DRþ cells by flow cytometry. To evaluate effect of CD271þ cells on CTLs, bulk PBMCs obtained from HLA-A24þ healthy volunteers were stimulated with a tumor antigen SAGE CTL epitope peptide (LYKPDSNEF; Invitrogen) at 10 mg/mL for 6 days, and the separated CD8þ cells (1 10) were restimulated with the peptide in the presence of IL-2 (100 IU/mL) and antigen-presenting cells (APC; 5 10 of MMC-inactivated autologous PBMCs) before/after depletion of CD271þ cells. After two to three cycles of the restimulation, the CD8þ cells were tested for IFN-g production in response to the peptide (24 hours) using the Cytometric Bead Array Kit (BD Biosciences), and for HLA-A24þSAGEþ Colo320 tumor-killing activity (6 hours) using the Annexin V system (MBL) according to the manufacturer's instructions. Flow cytometric analysis After Fc blocking, cells were stained with immunofluorescence [fluorescein isothiocyanate (FITC), phycoerythrin (PE), or CyChrome]-conjugatedmAbs specific for humanCCR7, CD11b, CD16, and CD271 (Abcam), CD33 (R&D Systems), CD45, CD56, CD80, CD86, HLA-DR, and IDO (AbD Serotec), PDL1 (eBiesciences), or the appropriate isotype control IgG. The immunofluorescence-conjugated secondary antibodies were used if necessary. The mAbs except designated ones were purchased Kudo-Saito et al. Cancer Res; 74(5) March 1, 2014 Cancer Research 1362 on May 3, 2017. © 2014 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from from BD Pharmingen. For intracellular staining, cells were permeabilized with cytofix/cytoperm before staining. The immunofluorescence was analyzed in comparing with the isotype control using a FACSCalibur cytometer (Cellquest software; BD Biosciences). Immunohistochemical analysis of clinical tissues Tissue sections, including normal colon tissues and tumor tissues (primary lesions, its adjacent normal portions and the metastatic lesions), derived from patients with stage I–IV colon were purchased fromMBL and SuperBioChips, and were analyzed for CCL19 and CD271 expressions immunohistochemically. After deparaffinization, tissue sections were treated with 20% ImmunoBlock (DS Pharma) for 30 minutes, 10% goat serum for 30 minutes, and then FITC-conjugated antiCD271 mAb (Abcam), anti-CCL19 mAb (R&D Systems), or isotype IgG as a control at 4 C overnight. PE-conjugated anti-goat IgG secondary antibody was used for CCL19. After 40,6-diamidino-2-phenylindole (DAPI) or hematoxylin staining followed by fixation with ethanol and xylene, immunofluorescence intensity of the mounted sections was automatically measured as pixel counts at two fields under the same level of DAPI pixel using a LSM700 Laser Scanning Microscope (Carl Zeiss), and the data were averaged in graphs. Statistical analysis Significant differences (P < 0.05) were evaluated using the unpaired two-tailed Student t test. The significance of the data was confirmed by analyzing the cumulative data of the repeated experiments using the nonparametric Mann–Whitney U test, when the number of n was small in an experiment. For graphical representation of data, y-axis error bars indicate the SD of the data. Correlation between two gene/molecular expressions in clinical samples was evaluated by the nonparametric Spearman rank test. Results HERV-H is involved in cancer metastasis As previously shown elsewhere (3, 4), herv-h mRNA expressionwas observed in various human cancer cell lines, but not in normal tissues except testis, placenta, and pancreas (Supplementary Fig. S1A). Also, herv-h mRNA was more frequently expressed in tumor lesions comparedwith the adjacent normal portion of the identical patients with colon cancer, particularly of more advanced stage III/IV patients having metastasis (Supplementary Fig. S1B). We assumed that HERV-H expression in tumor cellsmight be associatedwith tumor progression and metastasis. Then, we evaluated herv-h knockdown effect on tumor functions using human HERV-Hþ tumor cell lines (colon cancer Colo320 as the same type of cancers as the clinical tissues used, and pancreatic cancer MIAPaca as another cancer). When HERV-Hþ tumor cells were transfected with herv-h–specific siRNAs, the Matrigel-invasive ability was significantly reduced (P < 0.01), and Fibronectin-adhesive ability was significantly elevated (P < 0.05), although cell proliferation was not influenced (Fig. 1A). This was accompanied by decreased expression of mesenchymal markers Snail, Twist, and Fibronectin, and increased expression of an epithelial marker E-cadherin (P < 0.01 vs. control siRNA; Supplementary Fig. S2A). HERV-H /low cells were not influenced by siRNAherv-h transfection (data not shown). Tumor metastasis to lymph nodes and lung in vivowas also significantly inhibited in immunodeficient mice when siRNA-herv-h was transfected into HERV-Hþ tumor cells injected (P < 0.04; Fig. 1B). When HERV-H /low cells (colon cancer HCT116, pancreatic cancer Panc1, melanoma Hs294T as another cancer, and noncancerous epithelial cell ARPE19) were stimulated with TGF-b, which is a potent inducer of EMT, for 2 days, herv-hmRNA expression significantly increased (P < 0.01; Fig. 1C). Interestingly, no increase of snail and twist was seen under herv-h knockdown even after TGF-b stimulation (Supplementary Fig. S2B). This implies that HERV-H may be required for Snail and Twist induction in tumor cells undergoing TGF-b–induced EMT. By RT-PCR using primers specific for ORF of each HERV envelope protein, we confirmed that mRNA expression of other HERVs (HERV-E, -F, -K, -R, and -W) was not affected by siRNA-herv-h transfection or TGF-b stimulation (data not shown). These results suggest that HERV-H is involved in EMT, which is a cellular event to acquire high motility and invasive ability leading to metastasis (14). HERV-H encodes a transmembrane envelope protein with an immunosuppressive domain of 17 amino acids. A synthetic peptide corresponding to this domain is designated H17. We next evaluated how the H17 peptide would affect tumor invasion. When HERV-H /low cells were stimulated with H17 peptide for 2 days, tumor invasive ability was significantly elevated (P < 0.05 vs. control peptide; Fig. 1D), and Twist, Fibronectin, and CD44 increased, whereas Snail and E-cadherin only slightly changed (Fig. 1E and Supplementary Fig. S2C). This effectwas abrogated by PI3K inhibitor LY294002, but not by MEK inhibitor PD98059. This suggests that PI3K is involved in the signaling pathway, although both MEK/ERK (extracellular signal–regulated kinase) and PI3K/Akt have been reported in the HERV-induced immunosuppressive mechanism (12). Twist has been reported to promote tumor metastasis via PI3K activation (15). These suggest that the H17 peptide has a unique property to induce EMT possibly via the Twist–PI3K pathway in tumor cells. HERV-H is required for CCL19 production in tumor cells Chemokines are closely associated with cancer metastasis, and these expressions in tumor tissues are correlated with a metastasis incidence and a poor prognosis of patients with cancer (16, 17). We next investigated the relationship between chemokines and HERV-H expression in tumor cells. In the siRNA-herv-h–transfectedHERV-Hþ tumor cells, ccl2 and ccl19 expressions were almost completely abrogated, although no impact on other chemokines tested (Supplementary Fig. S2D). However, when HERV-H /low cell lines were stimulated with TGF-b following siRNA transfection, CCL19 induction was significantly inhibited only by siRNA-herv-h, although CCL2 induction was inhibited by either siRNA specific for herv-h, snail, or twist (P < 0.001; Fig. 2A and B). This indicates that HERV-H expression is necessary for CCL19 production in metastatic tumor cells. Interestingly, H17 peptide rather than TGF-b strongly inducedCCL19 production in tumor cells doseRole of HERV-H in Immunoevasion of Cancer www.aacrjournals.org Cancer Res; 74(5) March 1, 2014 1363 on May 3, 2017. © 2014 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from dependently, and this effect was inhibited almost completely by PI3K inhibitor LY294002, and partly by MEK inhibitor PD98059 (Fig. 2C). This suggests that PI3K signaling pathway is also required for the H17-induced CCL19 production in tumor cells. HERV-Hþ tumor cells recruits and expands pluripotent and immunoregulatory CD271þ cells utilizing H17 and CCL19 Next, to examine how CCL19 produced by HERV-Hþ tumor cells would modulate immune responses, we firstly conducted chemotaxis assay using human PBMCs and Transwell chambers. When the migrated cells were analyzed by flow cytometry, CD4þ cells and CD56þ cells were found to be themajority of the cells recruited by CCL19 or HERV-Hþ tumor supernatant (Supplementary Fig. S3). Only CD4CD56 cells decreased when CCL19 or its receptorCCR7 (data not shown)was blocked by the specific mAb, suggesting that CCL19–CCR7 axis is, at least in part, responsible for the migration of the CD4CD56 cells. The phenotype (CD45 /lowCD271þCCR7þ) of these cells was similar to that of mesenchymal stem cells (MSC) reported in human PBMCs and bonemarrow cells (Supplementary Fig. S3C; refs. 18–20). Particularly, one of the potential MSC markers CD271 (21, 22) was distinguishably expressed in the migrant cells. It is inferred that these cells may be one of the MSC subpopulations, although evaluation of the MSC activities was unable due to a small number of cells in the assay. Interestingly, CD271þ cells also significantly increased in number when PBMCs were stimulated with HERV-Hþ tumor supernatant for 5 days (P 1⁄4 0.001 between MIAPaca-treated cells vs. Panc1treated cells), and such increase was not seen using siRNA-hervh–transfected tumor supernatant (Fig. 3A and Supplementary Fig. S4A). This suggests that HERV-H–derived soluble factors could be involved in the mechanism. Indeed, purified CD271þ cells proliferated in response to CCL19 and H17, particularly in the presence of other CD271 cells (Fig. 3B), resulting in significant increase of CD271þ cells in the PBMC culture (P1⁄4 0.019; Fig. 3A). This suggests that a small number ofCD271þ cells are resident in human PBMCs, and are recruited and expanded by CCL19 cooperatively with H17 and other cells within the HERV-Hþ tumor microenvironment. CD271þ cells sorted from the culture with CCL19 and H17 frequently differentiated into mesenchymal lineages such as adipocytes, osteocytes, and chondrocytes (Supplementary Fig. S4B), and Invasion Adhesion Proliferation 3.0