The Immunosuppressive Surface Ligand CD200 Augments the Metastatic Capacity of Squamous Cell Carcinoma

Published OnlineFirst March 23, 2010; DOI: 10.1158/0008-5472.CAN-09-4380

Tumor and Stem Cell Biology Cancer Research The Immunosuppressive Surface Ligand CD200 Augments the Metastatic Capacity of Squamous Cell Carcinoma

Magda Stumpfova1, Desirée Ratner2, Edward B. Desciak2, Yehuda D. Eliezri2, and David M. Owens1,2

Abstract CD200 (OX-2) is a cell surface glycoprotein that imparts immune privileges by suppressing alloimmune and autoimmune responses through its receptor, CD200R, expressed primarily on myeloid cells. The ability of CD200 to suppress myeloid cell activation is critical for maintaining normal tissue homeostasis but may also enhance the survival of migratory neoplastic cells. We show that CD200 expression is largely absent in well- differentiated primary squamous cell carcinoma (SCC) of the skin, but is highly induced in SCC metastases to the lymph node and other solid tissues. CD200 does not influence the proliferative or invasive capacity of SCC cells or their ability to reconstitute primary skin tumors. However, loss of CD200 impairs the ability of SCC cells to metastasize and seed secondary tumors, indicating that the survival of CD200+ SCC cells may depend on their ability to interact with CD200R+ immune cells. The predominant population of CD200R+ stromal cells was CD11b+Gr-1+ myeloid-derived suppressor cells, which release elevated levels of granulocyte colony-stimulating factor and granulocyte macrophage colony-stimulating factor when in the presence of SCC cells in a CD200- dependent manner. Collectively, our findings implicate CD200 as a hallmark of SCC metastasis and suggest that the ability of CD200+ SCC keratinocytes to directly engage and modulate CD200R+ myeloid-derived suppressor cells is essential to metastatic survival. Cancer Res; 70(7); 2962–72. ©2010 AACR.

Introduction vascular endothelia (15), skin (16), and lymphoid cells (12). Mice lacking CD200 are normal in appearance but exhibit Solid tumor metastasis is a complex multistep process, elevated numbers of activated CD11b+ macrophages and during which neoplastic cells must invade tissue, enter and granulocytes (11). As a result, CD200-null animals display navigate via the circulatory system, and extravasate to colo- chronic central nervous system inflammation, early onset of ex- nize a distant foreign tissue (1, 2), and the success of this pro- perimental autoimmune encephalomyelitis, and increased sus- cess is highly dependent on modulating the immune system ceptibility to autoimmunity in experimental autoimmune (3, 4). Some well-known examples of tumor-associated im- uveoretinitis due to a failure to adequately inactivate CD200R1+ mune modulation include loss of tumor antigens, alterations macrophages in response to injury (11, 17). Therefore, during in HLA class I antigens (5, 6) and proinflammatory (Th1) and homeostasis, CD200 maintains an appropriate level of acti- anti-inflammatory (Th2) T helper cell–derived cytokine pro- vated macrophages to preserve tissue integrity; whereas dur- duction (7–9), decreased antigen-presenting cell activity, ing injury-induced heightened inflammatory states, CD200 secretion of proapoptotic factors, and recruitment of im- also provides immune privilege to CD200+ tissues to escape munosuppressive cells, all of which may disable antitumor macrophage-mediated tissue damage (16, 18). CD200R, a effector T cells (10). transmembrane glycoprotein, is expressed mostly in myeloid CD200 is a widely expressed cell surface glycoprotein that cells (13, 19) and smaller subsets of lymphoid-derived cells (13, acts as a potent suppressor of CD200R-expressing immune 20). There are five described CD200R isoforms, CD200R1–R5, cells (11–13). The immunosuppressive capacity of CD200 is of which CD200R1 is reported to be the major mediator of crucial for maintaining homeostasis in a number of tissues, CD200 immunosuppressive signaling (13, 21). including the central and peripheral nervous system (14), Several lines of evidence suggest that the immunosuppressive capacity of CD200 may also stimulate cancer development. Authors' Affiliations: Departments of 1Pathology and 2Dermatology, CD200 is upregulated in chronic lymphocytic leukemia (22), Columbia University, College of Physicians and Surgeons, New York, multiple myeloma (23), and acute myeloid leukemia (24). Fur- New York thermore, CD200-expressing melanoma (25) and ovarian can- Note: Supplementary data for this article are available at Cancer cer cells downregulate Th1 cytokine production in coculture Research Online (http://cancerres.aacrjournals.org/). with mixed leukocytes (26), indicating that CD200-mediated Corresponding Author: David M. Owens, Herbert Irving Comprehen- sive Cancer Center, Columbia University, 1130 St. Nicholas Avenue, immune suppression may also be featured in tumor progres- Room 321A, New York, NY 10032. Phone: 212-851-4544; Fax: 212-851- sion and/or metastasis. This idea is supported in a model of 4540; E-mail: [email protected]. B cell lymphocytic leukemia, in which ectopic CD200 expres- doi: 10.1158/0008-5472.CAN-09-4380 sion inhibited the ability of lymphocytes to clear tumor cells, + ©2010 American Association for Cancer Research. whereas CD200-negative tumor cells were rejected by CD200R

2962 Cancer Res; 70(7) April 1, 2010

CD200 in SCC Metastasis

peripheral mononuclear blood cells (25). Finally, CD200 ex- factor (Invitrogen), 5 μg/mL of insulin (Sigma), 0.5 μg/mL of − pression has also been observed in a tumorigenic keratinocyte hydrocortisone (MP Biomedicals), 10 10 mol/L of cholera line isolated from carcinogen-treated mouse skin (16); how- enterotoxin (ICN), and 1× penicillin-streptomycin (Invitro- ever, the significance of CD200 in epithelial tumor metastasis gen). All cell lines used are listed in Supplementary Table S1. remains to be determined. Cell passages 1 to 20 were used in experiments. We have identified CD200 induction as a hallmark of me- Generation of stable CD200 knockdown cell lines. To ab- tastatic squamous cell carcinoma (SCC). CD200 does not in- late CD200 expression in Lung Met cells, the pSM2c retroviral fluence primary tumor development but promotes the vector encoding a short hairpin RNA against murine CD200 survival of metastatic SCC cells and their ability to seed sec- (pRS-CD200, clone V2MM_8018; Open Biosystems) or control ondary tumors. Finally, CD200R1+ myeloid-derived suppressor (pRS-NS) was transfected into ΦNX-Eco (ecotrophic) packaging cells (MDSC) were implicated as the primary myeloid immune cells to produce virus to subsequently transduce AM12 (am- cell type potentially responsible for supporting the survival of phitrophic) producer cells as previously described (27, 30). CD200+ metastatic SCC cells. To transduce Lung Met cells, primary cultures were maintained with mitotically arrested AM12 packaging cells for Materials and Methods 48 h, after which transduced Lung Met cells were selected in 2 μg/mL of puromycin (Sigma). CD200 knockdown was Antibodies. Antibodies were used against human CD200, visualized by indirect immunofluorescence on glass coverslips mouse CD200, CD3ε,CD86,α6 integrin-FITC (BD Biosciences); and quantified by flow cytometry analysis. NK1.1-488, c-kit, CD123, MHC class II, CD11b-FITC, CD11b-PE, Flow cytometry. All flow cytometry experiments were Gr-1-FITC, Gr-1-PE, CD11c (BioLegend); Langerin, Foxp3 performed on freshly isolated tumor cell suspensions or (eBioscience); Keratin 14 (Covance); CD200R1 (R&D Systems); pRS-CD200–transduced, pRS-NS–transduced, or untransduced CD4-APC (Abcam); granulocyte colony-stimulating factor Lung Met 1, 2, and 4 cultured keratinocytes (see Supplementary (G-CSF; Santa Cruz Biotechnology); anti-rabbit horseradish Materials and Methods). peroxidase (Jackson ImmunoResearch); species-specific In vitro studies. To measure proliferation, pRS-NS or pRS- AlexaFluor-488, -594, -647, and -680 (Invitrogen); and human CD200 Lung Met 1, 2, or 4 cells were plated into 12-well plates α6 integrin (kindly provided by Fiona Watt, Cambridge Re- at 50,000 cells/well and cell counts were performed using a search Institute, Cambridge, United Kingdom). hemacytometer in 48-h increments for a period of up to Two-stage skin carcinogenesis. Murine skin tumors were 2 wk. Each time point was counted in triplicate for each group induced by a standard two-stage chemical carcinogenesis and average cell numbers were statistically compared using a protocol in FVB mice as previously described (ref. 27; see Student's t test (P < 0.05) between pRS-NS and pRS-CD200 Supplementary Materials and Methods). All mice were groups for each Lung Met cell line. housed and sacrificed according to Institutional Animal Care For invasion assays, pRS-NS or pRS-CD200 Lung Met 1, 2, and Use Committee and NIH guidelines for experimental end or 4 cells were plated at 2 × 105 cells per well in Matrigel in- points including tumor size and morbidity. vasion chambers according to the protocols of the manufac- Tissue harvesting. Skin tumors were harvested from 7,12- turer (BD Biosciences). After 48 h, membranes were stained dimethylbenz(a)anthracene (DMBA)/12-O-tetradecanoyl- with 1% toluidine blue, mounted and microscopically evalu- phorbol-13-acetate–treated mice upon sacrifice and residual ated for the number of invasive cells on five randomly chosen normal skin on the periphery of the tumors was removed fields of view per insert. Normal mouse epidermal keratino- using surgical scissors. Mice were also examined internally, cytes were analyzed as a control. The average number of cells and enlarged lymph nodes and lung tumor specimens were per membrane was statistically compared using the Student's extracted with forceps and cleaned from connective tissue. t test (P < 0.05) between pRS-NS and pRS-CD200 groups for De-identified human cutaneous SCC specimens were obtained each Lung Met cell line (n = 3 inserts per group). from the Dermatology Department and human metastatic In vivo studies. For skin tumor reconstitution, pRS-NS SCC specimens from the Herbert Irving Comprehensive or pRS-CD200 Lung Met 1, 2, or 4 cells were resuspended Cancer Center Tissue Bank at Columbia University under in HBSS solution and 1.0 × 105 (n = 3 mice per cell group) institutional review board approval. Normal skin and tumor or 5.0 × 105 (n = 3 mice per cell group) cells were implanted specimens were embedded in optimal cutting tempera- onto the dorsal fascia of recipient mice in silicon grafting ture compound and cryopreserved or fixed in 10% neutral chambers as previously described (ref. 28; see Supplementary buffered formalin and paraffin-embedded. Materials and Methods). To measure metastasis, pRS-NS or Cell culture. Primary keratinocyte cultures were estab- pRS-CD200 Lung Met 1, 2, or 4 cells were resuspended in lished from normal skin, benign papillomas, primary cuta- HBSS and 2 × 105 cells were injected into the tail vein of fully neous SCC, or metastatic SCC to the lymph node or lung immunocompetent female FVB mice (n = 9 mice per cell as previously described (28, 29). All cells were plated on col- group). Lungs were analyzed at 7 or 35 d postinjection for lagen and fibronectin-coated dishes in the presence of a mito- the presence of metastatic SCC (n =3–6 mice per time point; tically arrested fibroblast feeder layer (28) at 32°C in Williams' see Supplementary Materials and Methods). Medium E (Invitrogen) supplemented with 8 μg/mL of trans- Cytokine array. pRS-NS or pRS-CD200 Lung Met cells (8 × ferrin (Invitrogen), 409 mg/L of glutamine (Invitrogen), 20% 105) were cocultured with fluorescence-activated cell sorted fetal bovine serum (Hyclone), 10 ng/mL of epidermal growth (FACS) CD200R1+ immune cells (1.6 × 105) freshly isolated

www.aacrjournals.org Cancer Res; 70(7) April 1, 2010 2963

Stumpfova et al.

from DMBA/12-O-tetradecanoylphorbol-13-acetate–induced n = 8; Fig. 1B). Interestingly, expression of CD200 within skin tumors. Lung Met/immune cell cocultures were incu- PD SCC was consistently localized in tumor keratinocytes in- bated at 32°C for 24 h in complete Williams' Medium E, after vading the underlying muscle fascia (Fig. 1B). Importantly, which supernatants were collected, filtered, and hybridized virtually all (LN Met: average 88%, n = 7; Lung Met: average to the Mouse Inflammation Antibody Array 1 according to 93%, n = 3) keratinocytes isolated from metastatic SCC speci- the protocols of the manufacturer (RayBiotech). Each exper- mens expressed CD200 (Fig. 1B). imental group was run in duplicate and controls included A similar expression pattern for CD200 was observed in supernatants harvested from Lung Met or CD200R1+ immune human cutaneous and metastatic SCC. CD200 expression cells cultured alone. For densitometry analysis, individual was rarely observed in human WD SCC (1 of 20, 5%) but cytokine values were corrected by subtraction from the aver- was more frequent in human PD SCC (2 of 7, 29%; Fig. 2; age background reading for each membrane followed by nor- Table 1). Notably, 85% (11 of 13) of human metastatic SCC malization to the average positive control signal (n = 6 positive to the lymph node and 100% (7 of 7) of metastatic SCC seeding control readings per membrane). Average-corrected cytokine a variety of other secondary organs exhibited CD200 induction levels (n = 2 cytokine readings per membrane) were statis- (Fig. 2; Table 1). No CD200 expression was detected in normal tically compared between all of the following groups: human epidermis (Fig. 2). These observations confirm that CD200R1+ cells alone, pRS-NS Lung Met cells alone, pRS- CD200 induction is a hallmark for murine and human meta- CD200 Lung Met cells alone, CD200R1+/pRS-NS Lung Met static SCC and indicate that the CD200+ pool of SCC keratino- cultures, and CD200R1+/pRS-CD200 Lung Met cultures cytes rapidly expands during late stage carcinogenesis. (Student's t test; P < 0.05). CD200 does not influence SCC keratinocyte proliferation or invasion. Established cultures of LN and Lung Met SCC Results keratinocytes maintained their CD200 expression status, whereas no CD200 expression was observed in K14+ keratino- CD200 is induced in metastatic SCC. To determine a role cytes isolated from skin, papilloma, or SCC (Fig. 3A and B). The for CD200 in skin carcinogenesis, we analyzed CD200 expres- molecular fingerprint of DMBA-induced cutaneous SCC is the sion in primary and metastatic SCC induced by classic two- transversion mutation in the 61st codon of H-ras (34, 35). We stage chemical carcinogenesis in FVB mouse skin (27, 31). In detected the H-ras 61st codon mutation in two of the three response to topical DMBA/12-O-tetradecanoylphorbol-13- (67%) Lung Met specimens (Lung Met 2 and 4), confirming acetate treatment, FVB mice develop cutaneous SCC that metas- that these tumors were derived from DMBA-induced skin tu- tasize to the lymph node and lung (ref. 27; Fig. 1A), a model that is mors (Supplementary Fig. S2). analogous to human SCC pathogenesis (32, 33). CD200 was pres- To measure the effects of CD200 on proliferation, short ent in the hair follicle bulge but not in the epidermis of normal hairpin RNA–transduced (pRS-NS or pRS-CD200) Lung Met murine skin (ref. 16; Fig. 1A) and was also observed in the dermal 4 cells (Fig. 3B) were plated at equal densities and counted papilla of telogen and anagen hair follicles (Fig. 1A; Supplemen- every 48 hours for a period of 2 weeks, by which time, all cells tary Fig. S1). No CD200 expression was detected in benign pap- had reached confluency. No significant difference in log phase illomas (0 of 9) or well-differentiated (WD) SCC of the skin (0 of growth was observed between pRS-NS (doubling rate = 1.6 × 5 5 24); however, CD200 induction was first observed in the invasive 10 ) and pRS-CD200 (doubling rate = 1.6 × 10 ) Lung Met 4 front of poorly differentiated (PD) SCC (Fig. 1A) and CD200 was cells (Fig. 3C). Similar results were obtained for Lung Met 1 highly induced in metastatic SCC in the lymph node (14 of 14, and 2 cultures (data not shown). pRS-NS and pRS-CD200 Lung 100%) and lung (3 of 3, 100%; Table 1). Keratin K14 staining Met 4 cells were also evaluated for their capacity to invade distinguished tumor keratinocytes from surrounding stroma through Matrigel. Although Lung Met cells were 6-fold to (Fig. 1A). These results indicate that induction of CD200 strong- 7-fold more invasive than normal mouse epidermal keratino- ly correlates with late stage skin carcinogenesis and raises the cytes (data not shown), there was no significant difference in possibility that CD200 may play a role in SCC metastasis. the invasive capacity between pRS-NS and pRS-CD200 Lung Next, we quantified the number of CD200+ cells from freshly Met 4 cells (Fig. 3D). Similar results were obtained for Lung isolated primary and metastatic SCC that coexpress the kera- Met 1 and 2 cultures (data not shown). Collectively, these re- tinocyte surface receptor α6 integrin by flow cytometry anal- sults indicate that CD200 does not influence the proliferation ysis. Little to no α6+/CD200+ cells were observed in WD SCC or invasion of metastatic SCC cells. (average 3.9%, n = 5; Fig. 1B); however, a substantial population CD200 does not influence primary cutaneous SCC for- of α6+/CD200+ cells was detected in PD SCC (average 68%, mation. To assess the role of CD200 in primary cutaneous

Figure 1. CD200 is induced in murine cutaneous and metastatic SCC. A, gross morphology (top row) and H&E staining (second row), CD200 immunofluorescence (third row), and keratin K14 immunofluorescence (bottom row) of murine normal skin, papilloma, WD SCC, PD SCC, LN Met, and Lung Met. Arrowheads point to metastatic SCC nodules in lung (top row). Dashed lines on H&E sections segregate metastatic SCC (kc) from lymph node or lung stroma (st). For immunofluorescence, nuclei were delineated with 4′,6-diamidino-2-phenylindole (DAPI; blue), and the basal layer of normal epidermis (IFE) is outlined with a dashed line. White arrowheads point to CD200−/K14+ epidermal keratinocytes in normal skin, or CD200+/K14+ metastatic keratinocytes in Lung Met. B, FACS pseudo-color dot plot showing α6 integrin (X-axis) and CD200 (Y-axis) expression or K14 (X-axis) and CD200 (Y-axis) expression (Lung Met only). H&E-stained sections correspond to portions of the same tumors analyzed by FACS. Inset, CD200 immunofluorescence (green) in PD SCC. Bars, 100 μm. Abbreviations: dm, dermis; dp, dermal papilla; hf, hair follicle; hg, hair germ; alv, alveolar.

2964 Cancer Res; 70(7) April 1, 2010 Cancer Research

CD200 in SCC Metastasis

www.aacrjournals.org Cancer Res; 70(7) April 1, 2010 2965

Stumpfova et al.

lung. Harvested lung tissues were analyzed by FACS analysis Table 1. CD200 expression in human and for CD200 and K14 (K14 is not normally expressed in lung murine SCC tissue) expressing metastatic SCC keratinocytes. At 7 days

+ postinjection, low levels of both pRS-NS (average, 0.76%) Tumor type Tumor CD200 (%) and pRS-CD200 (average, 0.48%) Lung Met cells were detected number (n) in total lung cell suspensions but these levels were not statistically different (P = 0.11; n = 3 mice per group; Fig. 4C). How- Human + Cutaneous tumors (n = 27) ever, the percentage of K14 Lung Met cells increased 4-fold in – n WD SCC 20 1 (5) pRS-NS injected mice ( = 3 mice), whereas the percentage of + ∼ – PD SCC 7 2 (29) K14 cells actually decreased 2-fold in pRS-CD200 injected P n Metastatic SCC (n = 20) mice ( = 0.0001; = 6 mice) by 35 days (Fig. 4C). Consequently, Lymph node (head-neck) 13 11 (85) the metastatic capacity of pRS-NS Lung Met cells was 8-fold Other secondary site* 7 7 (100) to 9-fold higher compared with pRS-CD200 Lung Met cells (P = 0.0001; Fig. 4C). Similar results were obtained for all three Mouse Lung Met cell lines (data not shown). These data indicate that Cutaneous tumors (n = 47) CD200 promotes SCC metastasis by supporting the survival of Papilloma 9 0 metastatic cells seeded in secondary tissues. WD SCC 24 0 CD200R-expressing immune cells are present in cutaneous and metastatic SCC. To determine whether CD200+ SCC PD SCC 14 14 (100) + Metastatic SCC (n = 17) cells may interact with CD200R immune cells during skin Lymph node 14 14 (100) carcinogenesis, DMBA-induced skin and metastatic tumors Other secondary site* 3 3 (100) were analyzed for the presence of CD200R1-expressing cells. CD200R1+ cells were present in the stroma of benign papil- *Other secondary sites include liver, brain, bone, abdomen, lomas (data not shown), WD SCC, LN Met, and Lung Met chest wall, larynx, and oral mucosa in human (one specimen (Fig. 5A and Supplementary Fig. S3A) and were tightly asso- per site) and lung in mouse (all three). ciated with SCC keratinocytes suggestive of direct cell-cell contact (Fig. 5A). Little to no CD200R1+ cells were found to coexpress NK1.1, CD3, or c-kit, thereby excluding a natural killer cell, lymphoid, or mast cell phenotype (Supplemen- SCC formation, pRS-NS or pRS-CD200 Lung Met 4 cells were tary Fig. S3C). In addition, no significant labeling for CD123, surgically implanted onto the dorsal fascia of fully immuno- Langerin, CD86, or CD11c was observed in the CD200R1+ popu- competent FVB mice and evaluated for their ability to recon- lation by FACS analysis (data not shown). However, a predomi- stitute cutaneous SCC (Fig. 4A). FVB mice are an inbred strain nant population of CD200R1+ cells was found to coexpress the and highly syngeneic, thereby circumventing problems related myeloid lineage marker CD11b (average, 87.9% of CD200R1+ to host-graft compatibility. Within 5 to 7 weeks postgrafting, cells coexpress CD11b; n =5;Fig.5B).Furthermore,most both pRS-NS and pRS-CD200 Lung Met 4 cells (1 × 105 or 5 × of the CD200R1+/CD11b+ cell population also expressed Gr-1 105 cells/graft) were capable of regenerating cutaneous SCC (average, 74.5% of CD200R1+/CD11b+ cells coexpress Gr-1; (Fig. 4A and B). There was no difference in SCC size as deter- Fig. 5B). Finally, the remaining population of CD200R1+ cells mined by tumor diameter (data not shown) or weight between coexpressed the MHC class II surface molecule (average, pRS-NS and pRS-CD200 Lung Met 4 grafts (Fig. 4B). Similar 8.4% of CD200R1+ cells coexpressed MHC class II; Fig. 5B), results were obtained in SCC grafts obtained from Lung Met potentially marking a population of Langerhans cells (36). 1 and 2 cells (data not shown). These data indicate that CD200 The coexpression of CD200R1 and CD11b in stromal SCC cells may not play a role in primary SCC formation, an idea consis- identified by FACS analysis was confirmed by immunofluores- tent with the onset of CD200 induction in late stage PD SCC. cence labeling in tissue sections of WD SCC (Fig. 5B and Sup- Loss of CD200 diminishes SCC metastatic potential. To plementary Fig. S3B) and LN Met (Supplementary Fig. S3B). determine whether CD200 may be essential for SCC metasta- Immunosuppressive regulatory T-cells are also implicated sis, pRS-NS or pRS-CD200 Lung Met 4 keratinocytes were in cutaneous SCC progression (37); however, we could not assessed for their ability to seed secondary tumors in the correlate any distribution of Foxp3+ cells with the presence lung following tail vein injections in FVB mice. The tail vein or absence of CD200 (Supplementary Fig. S4A). The identifica- assay is a well-established model of experimental metastasis tion of Foxp3+ regulatory T cells was confirmed by CD4 cola- that tests the ability of tumor cells to cross the alveolar- beling (data not shown). Collectively, these studies identify the capillary barrier and form tumors in the lung. Under these major CD200R1+ stromal cell present in primary skin SCC as conditions, a 5-week incubation was the latest allowable time CD11b+/Gr-1+ MDSCs. point for harvesting tissues as the morbidity related to lung Metastatic SCC cells modulate MDSC activity via metastasis was typically too severe by 6 weeks postinjection CD200–CD200R interaction. To gain insight into a potential (data not shown). Upon sacrifice, animals were subjected to mechanistic link between the CD200–CD200R interaction gross inspection to detect the presence of secondary tumors and SCC metastasis, pRS-NS or pRS-CD200 Lung Met kerati- in internal organs; however, SCC were only observed in the nocytes were cocultured with equal amounts of CD200R1+

2966 Cancer Res; 70(7) April 1, 2010 Cancer Research

CD200 in SCC Metastasis

MDSCs extracted from DMBA-induced skin tumors, after CD200 has recently been established as a marker of a stem which the levels of 40 secreted proinflammatory cytokines cell population residing in the bulge region of the hair follicle were analyzed on Inflammation Antibody Arrays. We observed (39) and is thought to be one of a number of immunosuppres- a substantial increase in the release of 2 out of 40 cytokines, sive molecules expressed by hair follicle stem cells that protect granulocyte macrophage colony-stimulating factor (GM-CSF), against immune destruction, thereby preserving the regenera- and G-CSF, in cocultures consisting of pRS-NS Lung Met tive capacity of hair appendages (40). Interestingly, the hair (CD200+) cells and CD200R1+ MDSCs but not in cocultures follicle bulge has recently been shown to be a site of origin of pRS-CD200 Lung Met (CD200 knockdown) cells and for cutaneous SCC stem cells (41) and some normal pheno- CD200R1+ MDSCs or any of the single Lung Met or MDSC cul- typic markers of bulge keratinocyte stem cells, such as CD34 tures (Supplementary Fig. S4B). In addition, the net produc- expression, are preserved in cutaneous SCC (41). Along these tion values of GM-CSF and G-CSF in pRS-NS Lung Met/ lines, it is feasible that CD200 expression in SCC cells may CD200R1+ MDSC cocultures compared with pRS-CD200 Lung mark a population of cancer stem cells. However, we were un- Met/CD200R1+ MDSC cocultures were 4-fold (P = 0.049) and able to detect CD200+ keratinocytes in papillomas, the benign 20-fold (P = 0.002) higher, respectively (Fig. 5C). These results precursor to SCC. Although we did detect a small popula- show a functional role for the interaction between CD200 tion of CD200+ keratinocytes in WD SCC by FACS analysis, ligand in metastatic SCC cells and CD200R1 in MDSCs. Finally, we cannot rule out the possibility of contaminating normal immunofluorescence staining confirmed the presence of cells contributing to this population. The results of our tumor G-CSF–expressing cells in the stroma of CD200+ cutaneous reconstitution studies would suggest that, if CD200 does label murine PD SCC (Fig. 5D) in the same area in which MDSCs SCC stem cells, it is not required for their development in the are localized (Fig. 5A). G-CSF–positive infiltrative cells were skin. Moreover, the uniform expression of CD200 in metastatic detected in 67% (8 of 12) SCC analyzed (Fig. 5D); however, SCC is not entirely consistent with the notion that cancer stem the presence of G-CSF was not exclusive to PD SCC. cells are rare cells that maintain differentiated and phenotyp- ically distinct progeny in tumors. Discussion Using a mouse multistage model of SCC metastasis, we identified the predominant CD200R1+ immune cell species in- Our study links CD200 induction to the invasive and meta- filtrating the stroma of cutaneous and metastatic SCC as + + static capacity of both human and murine SCC, and establishes CD11b /Gr-1 MDSCs. In the mouse, MDSCs have been well + + CD200 as a potential criterion in the diagnosis of metastatic described as highly immunosuppressive CD11b /Gr-1 cells cutaneous SCC. Interestingly, CD200 does not directly alter the that commonly infiltrate solid tumors or other sites of inflam- phenotype of metastatic SCC cells but depends on interaction mation (42, 43). To our knowledge, this is the first report of with CD200R1+ immune cells to promote metastasis. A similar CD200R expression in MDSCs infiltrating any tumor type, model has been reported in leukemia cells that exhibit increased and therefore, provides a new dynamic to the behavior of surface levels of CD47, an immunoglobulin-like ligand that binds tumor-promoting MDSCs. In a normal tissue microenviron- its receptor, SIRPα, expressed on macrophages. The CD47– ment, ligation to CD200 would suppress the activity of + SIRPα interaction blocks macrophage phagocytosis of leuke- CD200R mature myeloid cells. However, in a tumor microen- mia cells, thereby stimulating metastasis (38). vironment, activation of CD200R in MDSCs may stimulate

Figure 2. CD200 is induced in human metastatic SCC. Normal human skin (top row), cutaneous (middle row), and metastatic (bottom row) SCC were H&E counterstained or probed with antibodies against α6 integrin (red) or CD200 (green). DAPI counterstain was conducted to visualize nuclei. Arrowheads delineate the basal layer of the epidermis (epid; H&E) or point to positive surface detection of α6 integrin and CD200 ligand (LN Met) or the absence of CD200 (skin; WD SCC). Dashed lines segregate CD200-positive tumor from CD200-negative stroma (LN Met).

www.aacrjournals.org Cancer Res; 70(7) April 1, 2010 2967

Stumpfova et al.

Figure 3. CD200 does not influence SCC keratinocyte proliferation or invasion. A, CD200, DAPI, and K14 immunofluorescence in normal mouse epidermal keratinocytes (NEK), papilloma (Pap), WD SCC, and LN Met–cultured keratinocytes. Arrowhead points to positive detection of CD200 (LN Met). All three images for each column represent the same field of view. B, top: CD200 and K14 immunofluorescence in untransduced and pRS-CD200–transduced LN Met cells. Arrows point to the presence (untransduced) or loss (pRS-CD200) of CD200 expression. Dotted line delineates the boundary of a pRS-CD200 keratinocyte colony. Both images for untransduced and pRS-CD200 cells represent the same field of view. FACS histograms showing CD200 surface levels in Lung Met 1, 2, and 4 keratinocytes (red line), IgG control for each plot is outlined in blue, or Lung Met 4 cells transduced with pRS-NS and pRS-CD200 short hairpin RNA. Bars designate gate set for FACS sorting into α6+/CD200+ (pRS-NS) and α6+/CD200− (pRS-CD200) cell populations. C, average counts of Lung Met 4 cells transduced with pRS-NS (•) or pRS-CD200 (▵) in 48-h increments (n = 3 per time point). Log phase of the growth (days 6–12) was used to determine the slope of each growth curve. D, bar graph showing the average number of cells per insert as a measure of the invasive capacity of Lung Met 4 cells transduced with pRS-NS or pRS-CD200 (n = 3 inserts per group).

2968 Cancer Res; 70(7) April 1, 2010 Cancer Research

CD200 in SCC Metastasis

Figure 4. Loss of CD200 impairs SCC metastatic potential. A, gross images of reconstituted SCC grafts from pRS-NS or pRS-CD200 Lung Met 4 cells. Bottom, H&E or keratin K14 (red) staining in graft sections. Bar, 100 μm. B, bar graph illustrating similar average weights of SCC grafts reconstituted with 1 × 105 pRS-NS (n = 2), 1 × 105 pRS-CD200 (n = 3), 5 × 105 pRS-NS (n = 3), or 5 × 105 pRS-CD200 (n = 2) Lung Met 4 cells. C, bar graph showing the percentage of K14+ metastatic keratinocytes present in the lung at 7 or 35 d following tail vein injection. Vehicle control-HBSS–injected mice. *, statistically significant difference (P = 0.0001); **, statistically significant difference (P = 0.0001). D, H&E and keratin K14 immunofluorescence (red) in lung sections confirm the presence of metastatic SCC in pRS-NS Lung Met 4–injected mice but not in HBSS control mice. the ability of these cells to promote tumorigenesis by immune antitumor immune function in mice and humans (45). Although suppression or other means. This concept is supported in part it is conceivable that the modulation of MDSC behavior via by our in vitro data demonstrating that the general effects of the CD200–CD200R interaction may establish a highly immuno- CD200+ tumor cells engaging CD200R1+ MDSCs are stimulato- suppressive, prometastatic milieu through G-CSF and GM-CSF ry rather than suppressive. Interestingly, MDSCs are present in production, the mechanistic significance of these proinflamma- the stroma of benign papillomas and WD SCC well before the tory cytokines remains to be determined. Along these lines, a onset of CD200 induction. This would also suggest multi- more comprehensive delineation of MDSC behavior resulting ple roles for MDSCs in carcinogenesis in that the previously from CD200R activation will be the focus of future studies. reported immunosuppressive mechanisms of MDSCs (42, The pattern of CD200 induction was preferentially local- 43), and perhaps CD200R-independent, may support the early ized to the invading front of tumor cells in PD SCC, indicating development and malignant conversion of neoplastic cells in that CD200 may play roles in local tumor invasion as well as the primary tumor site. Whereas in later stages of carcinogen- metastasis and that the proximity of neoplastic cells to tumor esis, CD200+ SCC cells use MDSCs in a CD200R-dependent stromal-derived factors might be a key factor to CD200 induc- manner to gain immune privilege and seed secondary tumors. tion. In fact, both IFNγ and tumor necrosis factor-α could in- We observed that MDSCs are stimulated to release elevated duce CD200 expression (46). Therefore, it is reasonable to levels of G-CSF and GM-CSF upon interaction with metastatic assume that CD200 may be predominantly important for pri- SCC cells in a CD200-dependent manner. A number of studies mary and metastatic tumor cells as they invade certain micro- have reported that G-CSF administration could enhance tu- environments and, in as much, might not be fundamental for mor angiogenesis, growth, and malignant progression in vivo the metastatic process. However, we found CD200 expression (44), whereas GM-CSF has been shown primarily to stimulate present in human metastatic SCC specimens isolated from at

www.aacrjournals.org Cancer Res; 70(7) April 1, 2010 2969

Stumpfova et al.

Figure 5. CD200R1+ stromal cells are CD11b+/Gr-1+ MDSCs. A, CD200R1 immunofluorescence in WD SCC, LN Met, and Lung Met (white arrows). DAPI counterstain was conducted to visualize nuclei. Dashed line demarcates SCC keratinocytes in LN Met. B, FACS analysis of CD200R1, CD11b, and Gr-1 or MHC class II expression. The total population of CD200R1+ cells was gated and subsequently analyzed for expression of CD11b and Gr-1 or MHC class II. The percentage of the total CD200R1 pool for a single experiment is shown. Far right, murine SCC were stained with antibodies against CD200R1 (red) and CD11b (green) and counterstained with DAPI (blue). Arrows point to CD200R1+/CD11b+ (yellow) stromal MDSCs. C, left: bar graph showing the densitometric units representing GM-CSF and G-CSF levels in pRS-NS Lung Met/CD200R1+ cocultures versus pRS-CD200 Lung Met/CD200R1+ cocultures. *, statistically significant difference (GM-CSF, P = 0.049; G-CSF, P = 0.002). Right, H&E staining (top left) and G-CSF (red) and CD200 (green) immunofluorescence in murine PD SCC. Nuclei were delineated with DAPI (blue). Bars, 50 μm.

2970 Cancer Res; 70(7) April 1, 2010 Cancer Research

CD200 in SCC Metastasis

least eight different body sites, suggesting that CD200 was not Disclosure of Potential Conflicts of Interest selectively regulated by certain tumor microenvironments but might be a general feature of the metastatic SCC cell. Moreover, No potential conflicts of interest were disclosed. we did not observe CD200 in human or murine WD SCC, although these are highly invasive lesions, and CD200 levels were Acknowledgments maintained in metastatic SCC cells in culture, suggesting that the α regulation of CD200 is a permanent fixture of the late-invasive/ We are grateful to Fiona Watt for providing the human 6 antibody and especially to Kristie Gordon for her assistance with flow cytometry. metastatic cell. Finally, the observation of enhanced early survival of CD200+ cells following tail vein injection, but preceding tu- − mor implantation, compared with CD200 cells strongly favors a Grant Support key role for CD200 in metastasis. In conclusion, the onset of NIH R01CA114014 and R21CA131897 grants. CD200 expression in PD SCC indicates roles for CD200 in tumor The costs of publication of this article were defrayed in part by the payment invasion and metastasis, whereas the mechanisms that underlie of page charges. This article must therefore be hereby marked advertisement in the induction of CD200, whether they involve a combination of accordance with 18 U.S.C. Section 1734 solely to indicate this fact. – tumor stromal derived factors and genetic/epigenetic changes Received 12/01/2009; revised 01/21/2010; accepted 01/29/2010; published within invading SCC cells, remain to be determined. OnlineFirst 03/23/2010.

References 1. Coussens LM, Werb Z. Inflammation and cancer. Nature 2002;420: 17. Copland DA, Calder CJ, Raveney BJ, et al. Monoclonal antibody- 860–7. mediated CD200 receptor signaling suppresses macrophage activa- 2. Chambers AF, Groom AC, MacDonald IC. Dissemination and growth tion and tissue damage in experimental autoimmune uveoretinitis. of cancer cells in metastatic sites. Nat Rev Cancer 2002;2:563–72. Am J Pathol 2007;171:580–8. 3. de Visser KE, Eichten A, Coussens LM. Paradoxical roles of the im- 18. Broderick C, Hoek RM, Forrester JV, Liversidge J, Sedgwick JD, Dick mune system during cancer development. Nat Rev Cancer 2006;6: AD. Constitutive retinal CD200 expression regulates resident microglia 24–37. and activation state of inflammatory cells during experimental auto- 4. Zou W. Immunosuppressive networks in the tumour environment and immune uveoretinitis. Am J Pathol 2002;161:1669–77. their therapeutic relevance. Nat Rev Cancer 2005;5:263–74. 19. Wright GJ, Cherwinski H, Foster-Cuevas M, et al. Characterization of 5. French LE, Tschopp J. Defective death receptor signaling as a cause the CD200 receptor family in mice and humans and their interactions of tumor immune escape. Semin Cancer Biol 2002;12:51–5. with CD200. J Immunol 2003;171:3034–46. 6. Algarra I, Garcia-Lora A, Cabrera T, Ruiz-Cabello F, Garrido F. The 20. Fallarino F, Asselin-Paturel C, Vacca C, et al. Murine plasmacytoid selection of tumor variants with altered expression of classical and dendritic cells initiate the immunosuppressive pathway of tryptophan nonclassical MHC class I molecules: implications for tumor immune catabolism in response to CD200 receptor engagement. J Immunol escape. Cancer Immunol Immunother 2004;53:904–10. 2004;173:3748–54. 7. Toutirais O, Chartier P, Dubois D, et al. Constitutive expression of 21. Boudakov I, Liu J, Fan N, Gulay P, Wong K, Gorczynski RM. Mice TGF-β1, interleukin-6 and interleukin-8 by tumor cells as a major lacking CD200R1 show absence of suppression of lipopolysaccharide- component of immune escape in human ovarian carcinoma. Eur induced tumor necrosis factor-α and mixed leukocyte culture responses Cytokine Netw 2003;14:246–55. by CD200. Transplantation 2007;84:251–7. 8. Somasundaram R, Jacob L, Swoboda R, et al. Inhibition of cytolytic 22. Kretz-Rommel A, Qin F, Dakappagari N, et al. CD200 expression on T lymphocyte proliferation by autologous CD4+/CD25+ regulatory tumor cells suppresses antitumor immunity: new approaches to can- T cells in a colorectal carcinoma patient is mediated by transforming cer immunotherapy. J Immunol 2007;178:5595–605. growth factor-β. Cancer Res 2002;62:5267–72. 23. Moreaux J, Hose D, Reme T, et al. CD200 is a new prognostic factor 9. Shurin MR, Lu L, Kalinski P, Stewart-Akers AM, Lotze MT. Th1/Th2 in multiple myeloma. Blood 2006;108:4194–7. balance in cancer, transplantation and pregnancy. Springer Semin 24. Tonks A, Hills R, White P, et al. CD200 as a prognostic factor in acute Immunopathol 1999;21:339–59. myeloid leukaemia. Leukemia 2007;21:566–8. 10. Rabinovich GA, Gabrilovich D, Sotomayor EM. Immunosuppressive 25. Petermann KB, Rozenberg GI, Zedek D, et al. CD200 is induced by strategies that are mediated by tumor cells. Annu Rev Immunol 2007; ERK and is a potential therapeutic target in melanoma. J Clin Invest 25:267–96. 2007;117:3922–9. 11. Hoek RM, Ruuls SR, Murphy CA, et al. Down-regulation of the mac- 26. Siva A, Xin H, Qin F, Oltean D, Bowdish KS, Kretz-Rommel A. Im- rophage lineage through interaction with OX2 (CD200). Science mune modulation by melanoma and ovarian tumor cells through ex- 2000;290:1768–71. pression of the immunosuppressive molecule CD200. Cancer 12. Wright GJ, Puklavec MJ, Willis AC, et al. Lymphoid/neuronal cell sur- Immunol Immunother 2008;57:987–96. face OX2 glycoprotein recognizes a novel receptor on macrophages 27. Owens DM, Romero RM, Gardner C, Watt FM. Suprabasal α6β4 in- implicated in the control of their function. Immunity 2000;13:233–42. tegrin expression in epidermis results in enhanced tumorigenesis 13. Gorczynski R, Chen Z, Kai Y, Lee L, Wong S, Marsden PA. CD200 is and disruption of TGFβ signaling. J Cell Sci 2003;116:3783–91. a ligand for all members of the CD200R family of immunoregulatory 28. Jensen UB, Yan X, Triel C, Woo SH, Christensen R, Owens DM. A molecules. J Immunol 2004;172:7744–9. distinct population of clonogenic and multipotent murine follicular 14. Meuth SG, Simon OJ, Grimm A, et al. CNS inflammation and neuro- keratinocytes residing in the upper isthmus. J Cell Sci 2008;121: nal degeneration is aggravated by impaired CD200–200R-mediated 609–17. macrophage silencing. J Neuroimmunol 2008;194:62–9. 29. Wu WY, Morris RJ. Method for the harvest and assay of in vitro clo- 15. Ko YC, Chien HF, Jiang-Shieh YF, et al. Endothelial CD200 is hete- nogenic keratinocytes stem cells from mice. Methods Mol Biol 2005; rogeneously distributed, regulated and involved in immune cell- 289:79–86. endothelium interactions. J Anat 2009;214:183–95. 30. Levy L, Broad S, Zhu AJ, et al. Optimized retroviral infection of hu- 16. Rosenblum MD, Olasz EB, Yancey KB, et al. Expression of CD200 on man epidermal keratinocytes: long-term expression of transduced epithelial cells of the murine hair follicle: a role in tissue-specific im- integrin gene following grafting on to SCID mice. Gene Ther 1998; mune tolerance? J Invest Dermatol 2004;123:880–7. 5:913–22.

www.aacrjournals.org Cancer Res; 70(7) April 1, 2010 2971

Stumpfova et al.

31. Owens DM, Wei SJ, Smart RC. A multihit, multistage model of chem- tion of stem cell-enriched human hair follicle bulge cells. J Clin Invest ical carcinogenesis. Carcinogenesis 1999;20:1837–44. 2006;116:249–60. 32. Quon H, Liu FF, Cummings BJ. Potential molecular prognostic mar- 40. Cotsarelis G. Epithelial stem cells: a folliculocentric view. J Invest kers in head and neck squamous cell carcinomas. Head Neck 2001; Dermatol 2006;126:1459–68. 23:147–59. 41. Malanchi I, Peinado H, Kassen D, et al. Cutaneous cancer stem cell 33. Green CL, Khavari PA. Targets for molecular therapy of skin cancer. maintenance is dependent on β-catenin signalling. Nature 2008;452: Semin Cancer Biol 2004;14:63–9. 650–3. 34. Quintanilla M, Brown K, Ramsden M, Balmain A. Carcinogen-specific 42. Gabrilovich DI, Nagaraj S. Myeloid-derived suppressor cells as reg- mutation and amplification of Ha-ras during mouse skin carcinogen- ulators of the immune system. Nat Rev Immunol 2009;9:162–74. esis. Nature 1986;322:78–80. 43. Murdoch C, Muthana M, Coffelt SB, Lewis CE. The role of myeloid 35. Bizub D, Wood AW, Skalka AM. Mutagenesis of the Ha-ras onco- cells in the promotion of tumour angiogenesis. Nat Rev Cancer 2008; gene in mouse skin tumors induced by polycyclic aromatic hydrocar- 8:618–31. bons. Proc Natl Acad Sci U S A 1986;83:6048–52. 44. Okazaki T, Ebihara S, Asada M, Kanda A, Sasaki H, Yamaya M. 36. Rosenblum MD, Woodliff JE, Madsen NA, McOlash LJ, Keller MR, Granulocyte colony-stimulating factor promotes tumor angiogene- Truitt RL. Characterization of CD200-receptor expression in the mu- sis via increasing circulating endothelial progenitor cells and Gr1 rine epidermis. J Invest Dermatol 2005;125:1130–8. +CD11b+ cells in cancer animal models. Int Immunol 2006;18:1–9. 37. Clark RA, Huang SJ, Murphy GF, et al. Human squamous cell carci- 45. Eubank TD, Roberts RD, Khan M, et al. Granulocyte macrophage nomas evade the immune response by down-regulation of vascular colony-stimulating factor inhibits breast cancer growth and metastasis E-selectin and recruitment of regulatory T cells. J Exp Med 2008;205: by invoking an anti-angiogenic program in tumor-educated macro- 2221–34. phages. Cancer Res 2009;69:2133–40. 38. Jaiswal S, Jamieson CH, Pang WW, et al. CD47 is upregulated on 46. Chen Z, Marsden PA, Gorczynski RM. Role of a distal enhancer circulating hematopoietic stem cells and leukemia cells to avoid in the transcriptional responsiveness of the human CD200 gene to phagocytosis. Cell 2009;138:271–85. interferon-γ and tumor necrosis factor-α. Mol Immunol 2009;46: 39. Ohyama M, Terunuma A, Tock CL, et al. Characterization and isola- 1951–63.

2972 Cancer Res; 70(7) April 1, 2010 Cancer Research

The Immunosuppressive Surface Ligand CD200 Augments the Metastatic Capacity of Squamous Cell Carcinoma

Magda Stumpfova, Desirée Ratner, Edward B. Desciak, et al.

Cancer Res 2010;70:2962-2972. Published OnlineFirst March 23, 2010.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-09-4380

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2010/03/23/0008-5472.CAN-09-4380.DC1

Cited articles This article cites 46 articles, 12 of which you can access for free at: http://cancerres.aacrjournals.org/content/70/7/2962.full#ref-list-1

Citing articles This article has been cited by 8 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/70/7/2962.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/70/7/2962. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.