Published OnlineFirst April 23, 2019; DOI: 10.1158/0008-5472.CAN-18-2659

Cancer Tumor Biology and Immunology Research

MicroRNA-92 Expression in CD133þ Stem Cells Regulates Immunosuppression in the Tumor Microenvironment via Integrin-Dependent Activation of TGFb Chris Shidal, Narendra P. Singh, Prakash Nagarkatti, and Mitzi Nagarkatti

Abstract

In addition to being refractory to treatment, melanoma enhanced TGFb activation, as evidenced by increased phos- þ cancer stem cells (CSC) are known to suppress host antitumor phorylation of SMAD2. CD133 cells transfected with miR- immunity, the underlying mechanisms of which need further 92a mimic and injected in vivo showed significantly elucidation. In this study, we established a novel role for miR- decreased tumor burden, which was associated with reduced 92 and its associated networks in immunosuppression. immunosuppressive phenotype intratumorally. Using The CSCs were isolated from the B16-F10 murine melanoma cell Cancer Genome Atlas database of patients with melanoma, line based on expression of the putative CSC marker CD133 we also noted a positive correlation between integrin a5 and þ (Prominin-1). CD133 cells were functionally distinct from TGFb1 expression levels and an inverse association between CD133 cells and showed increased proliferation in vitro and miR-92 expression and integrin alpha subunit expression. þ enhanced tumorigenesis in vivo. CD133 CSCs also exhibited a Collectively, this study suggests that a miR-92–driven sig- greater capacity to recruit immunosuppressive cell types naling axis involving integrin activation of TGFb in CSCs þ during tumor formation, including FoxP3 Tregs, mye- promotes enhanced tumorigenesis through induction of loid-derived suppressor cells (MDSC), and M2 macro- intratumoral immunosuppression. phages. Using microarray technology, we identified several þ þ miRs that were significantly downregulated in CD133 cells Significance: CD133 cells play an active role in suppres- compared with CD133 cells, including miR-92. Decreased sing melanoma antitumor immunity by modulating miR-92, expression of miR-92 in CSCs led to higher expression of which increases influx of immunosuppressive cells and TGFb1 target molecules integrin aV and a5 subunits, which, in turn, expression.

Introduction between immune cells and CSCs to determine how specific subpopulations may drive immunosuppression in the tumor Primary have been reported to harbor subpopula- microenvironment (TME). tions of tumor cells with intrinsic self-renewal and proliferative Immunosuppression can be mediated through several immune capacity termed as cancer stem cells (CSC; ref. 1). The CSC theory cell phenotypes including regulatory T cells (Treg), myeloid- may help explain the plastic, chemoresistant, and invasive nature derived suppressor cells (MDSC), and alternative macrophages of refractory melanomas. Several biomarkers have been utilized in (M2; ref. 6). Many cancers, including melanoma, exploit these the identification and isolation of melanoma CSCs including immune cell phenotypes to secrete cytokines and growth factors CD20 (1), aldehyde dehydrogenase (2), CD133 (3), and that create a permissive environment for cancers to proliferate and ABCB5 (4). The murine melanoma cell line B16-F10 was recently eventually metastasize. TGFb is a pleiotropic cytokine with robust shown to contain a distinct subset of cells expressing CD133 that immunosuppressive activity including the ability to repress T-cell had long-term tumorigenic potential and highly expressed the activation and proliferation (7). Part of this immunosuppressive markers Oct4, Nanog, and Sox10 (5). In this study, we þ effect can be carried out by Tregs, which produce abundant TGFb utilized CD133 B16-F10 cells to explore the intricate interactions to modulate immune response to self and foreign antigens (extensively reviewed in ref. 8). However, TGFb is secreted in Department of Pathology, Microbiology and Immunology, University of South an inactive form and must undergo activation to stimulate Carolina School of Medicine, Columbia, South Carolina. downstream signaling cascades through binding of the TGFb receptor (TGFBR; ref. 9). One mechanism for converting latent Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/). TGFb to its active form is through interactions with RGD- recognizing integrins (i.e., integrin av), which associates with Corresponding Author: Mitzi Nagarkatti, University of South Carolina, 6439 latent TGFb-binding (LTBP) and the TGFb prodomain Garners Ferry Road, Building 1 C-23, Columbia, SC 29208. Phone: 803-216-3402; Fax: 803-216-3413; E-mail: [email protected] to free TGFb via mechanical shearing (9). TGFb activation and subsequent signaling through its receptor has been associated Cancer Res 2019;79:3622–35 with immune evasion (10), epithelial-to-mesenchymal transition doi: 10.1158/0008-5472.CAN-18-2659 (EMT; ref. 11), and tumor cell invasion (12); thus, targeting of 2019 American Association for Cancer Research. TGFb signaling in cancer remains a priority (13, 14).

3622 Cancer Res; 79(14) July 15, 2019

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 2019 American Association for Cancer Research. Published OnlineFirst April 23, 2019; DOI: 10.1158/0008-5472.CAN-18-2659

miR-92 Regulates Immunosuppression by Cancer Stem Cells

miRs are small (20–30 nucleotide) noncoding RNAs that Primary tumors generated from subcutaneous injection of B16 generally function to suppress gene expression by targeting the cells or lungs from metastasis-bearing mice intravenously were 30 UTR of mRNAs, several of which have been demonstrated to dissociated using a Tumor Dissociation Kit (Miltenyi Biotec #130- regulate cellular functions pertinent to oncogenesis and tumor 096-730) to dissociate whole-tumor tissues into single-cell sus- progression (15). miR-92, a member of the miR-17-92 cluster, has pensions following the manufacturer's recommended protocol. been reported as both an oncomiR (16, 17) as well as a tumor The resulting cell suspensions were washed with and resuspended suppressor (18, 19) depending on the cancer model. Importantly, in PBS prior to initiating labeling with antibodies. Single-cell miR-92 was shown to regulate expression of integrin a5 in an labeling with fluorophore-conjugated primary antibodies against ovarian cancer model (20). Remarkably, the role of miR-92 in CD45 (BioLegend #103116), CD3 (BioLegend #100306), CD4 melanoma has yet to be explored. (BioLegend #100453), CD8 (BioLegend #100708), NK1.1 (Bio- Our study reveals, for the first time, that miR-92 may regulate an Legend #108748), FOXP3 (BioLegend #126419), IL10 (BioLe- integrin-mediated axis driving TGFb-induced immunosuppres- gend #505031), TGFb (BioLegend #141410), IL17 (BD sion in the TME. Furthermore, this axis may confer a selective #564168), IFNg (BD #563854), CD11b (BioLegend #101222), survival advantage to CSCs present within the heterogeneous F4/80 (BioLegend #123110), CD11c (BioLegend #117334), GR1 tumor population by modulating immunosuppression and (BioLegend #108457), Ly6C (BD #560595), Ly6G (BD exploiting immunosuppressive cell phenotypes such as Tregs and #560603), and CD206 (BioLegend #141723) was performed for M2 macrophage populations present within the TME. These at least 30 minutes on ice, washed with staining buffer, and studies shed on the biological function of CSCs in the context subsequently analyzed on a BD FACSCelesta Flow Cytometer of immune surveillance and also provide a potential therapeutic equipped with BD DIVA software in conjunction with FlowJo target in refractory melanomas in which CSCs may contribute to software. For intracellular labeling against transcription factors patient relapse. and cytokines (i.e., FOXP3), cells were fixed and permeabilized using the True-Nuclear Transcription Factor Buffer Set Kit (BioLegend #444201) following the manufacturer's recommen- Materials and Methods dations. Data were compensated using BD CompBeads (anti- Cell culture and reagents mouse #552843, anti-rat/hamster #552845), labeled with single The B16-F10 cell line was obtained from ATCC. All cell lines antibodies or isotype controls, and analyzed using FlowJo. Span- were grown in DMEM supplemented with 10% heat-inactivated ning-tree Progression Analysis of Density-normalized Events FBS (Atlanta Biologicals), penicillin (100 U/mL, Gibco), (SPADE) V3.0 (22) was used to down-sample and cluster simi- and streptomycin (100 mg/mL, Gibco). Cells were incubated larly labeled populations of cells following compensation and at 37 Cat5%CO2 and subcultured every 72 hours. Routine gating in FlowJo. Compensated FCS 3.0 files were exported and monitoring for Mycoplasma contamination was performed analyzed using the standalone version of SPADE 3.0 using the using the MycoAlert Detection Kit (Lonza #LT07-218). Cells following parameters: Arcsinh transformation ¼ 150, maximum recovered from frozen aliquots were allowed one passage to allowable cells in pooled data ¼ 200,000, outlier density ¼ 1, reach exponential growth phase following recovery before fixed number of remaining cells ¼ 100,000, clustering parameter being used in this study. Cells at passages greater than ten ¼ K-means, and the desired number of clusters ¼ 50. Determina- were not used in the experiments performed in this study. tions for phenotyping each node/cluster was carried out based on þ CD133 and CD133 cells were isolated by FACS and were single-color controls and a representative figure is provided in grown in DMEM/F-12 serum-free media (SFM) containing Supplementary Fig. S2. 1 N-2 Supplement (Gibco #17502-048) 10 ng/mL basic fibroblast growth factor (PeproTech #450-33), and 10 ng/mL MiRNA microarray þ EGF (PeproTech #315-09) in low-cluster 6-well plates Briefly, CD133 and CD133 populations were isolated via (Corning #3471). FACS from the B16-F10 murine melanoma as described above. Total RNA was extracted (Qiagen, miRNeasy #74106) from B16- FACS, flow cytometry, and Spanning-tree Progression Analysis F10 cells sorted from three independent experiments. Each sam- of Density-normalized Events analysis ple was individually analyzed for quantity (NanoDrop 2000, B16-F10 cells were grown as nonadherent oncospheres in Thermo Fisher Scientific) and quality (BioAnalyzer 2100, Agi- SFM as described previously (21). After 7–10 days of culture in lent). For miRNA microarray, aliquots from individual samples þ low-cluster plates, oncospheres were dissociated into single-cell were pooled for each group (n ¼ 3/CD133 / ). All samples used suspensions and labeled using a PE-conjugated CD133 anti- for downstream analysis had an RNA integrity number of at least body [(BioLegend #141204) in 100 mL of staining buffer (2% 8. RNA profiling from samples was performed using the FlashTag FBS/2 mmol/L EDTA in PBS)] at a dilution of 1:100. The Biotin HSR RNA Labeling Kit for GeneChip miRNA Arrays for the appropriate isotype control (BioLegend #400508) was used to Affymetrix GeneChip miRNA 4.0 array platform. Labeled and þ gate the CD133 and CD133 populations. Cells were sorted hybridized chips were scanned on a GeneChip Scanner (Affyme- using a BD FACSAria II into 15-mL conical collection tubes trix) and microarray image data were analyzed using Affymetrix containing approximately 10 mL of ice-cold PBS at 4C. Power Tools. Data analysis and generation of representative Representative histograms demonstrating our gating strategy figures (i.e., scatter plot) were performed using the Transcriptome and postsort purity have been provided (Supplementary Analysis Console (TAC, Affymetrix). MiRNAs with a fold change Fig. S1). After sorting, cells were centrifuged at 300 g for greater than 1.5 or less than 1.5 were considered for further 10 minutes, resuspended in an appropriate amount of PBS, and validation and analysis. Predicted targets and alignment scores for counted by trypan blue exclusion assay on a BioRad TC20 specific miRNAs were generated using online software including Automated Cell Counter before use in subsequent assays. TargetScan Mouse 6.2 and miRDB. Ingenuity Pathway Analysis

www.aacrjournals.org Cancer Res; 79(14) July 15, 2019 3623

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 2019 American Association for Cancer Research. Published OnlineFirst April 23, 2019; DOI: 10.1158/0008-5472.CAN-18-2659

Shidal et al.

(IPA, Qiagen) in combination with MetaCore pathway analysis Microscope (Life Technologies) and images were analyzed using tools (Thomson Reuters) were used to generate potential gene ImageJ Software (NIH, Bethesda, MD). networks associated with significantly altered miRNAs and gen- erate miR-gene interactome pathway maps. In vivo tumor growth models Female C57Bl/6 mice (Jackson #000644) were used at 6–8 qRT-PCR weeks of age. All mice were handled in accordance with the þ CD133 and CD133 B16-F10 cells were isolated by FACS, and American Association for Laboratory Animal Science guidelines total RNA was isolated using miRNeasy Kit (Qiagen), following with the approval of the appropriate Institutional Animal Care the manufacturer's protocol. The expression of indicated mRNA and Use Committees at the University of South Carolina (Colum- and miRNA levels was determined by qRT-PCR. Total RNA was bia, SC; protocol no. 2371). Mice were injected subcutaneously quantitated using a Nanodrop 2000 (Thermo Fisher Scientific). with 1 105 B16-F10 cells in PBS (100 mL). Tumor size was For miRNA expression analysis, cDNA was generated from total monitored three times weekly until animals were sacrificed RNA using miScript II cDNA Synthesis Kit (Qiagen # 218161). because of tumor burden. Tumor volume [V ¼ L W2 Two-step miRNA qRT-PCR were carried out using SsoAdvanced (p/6)] was determined by measuring the greatest linear dimen- SYBR Green Mix (Bio-Rad #1725270) with mouse primers for sions in length (L) and width (W). Snord96a (Qiagen #MS00033733), miR-669a-5p (Qiagen For our experimental metastasis models, 2 105 B16-F10 cells #MS0026222), miR-669l-5p (Qiagen #MS00043337), miR- suspended in 100 mL PBS were injected intravenously into 6- to 466h-5p (Qiagen #MS00012201), and miR-92a-3p (Qiagen 8-week-old, female C57Bl/6 mice via the lateral tail vein. After #MS00005971). Expression levels for miRNAs were normalized approximately 14–16 days, mice were sacrificed. Upon sacrificing to Snord96a. For mRNA expression analysis, cDNA was made the mice, lungs were resected, imaged, dissociated, and labeled from total RNA using miScript II cDNA synthesis . A two-step with antibodies for subsequent flow cytometry analysis. þ amplification with a 60C annealing temperature for qRT-PCR In experiments involving in vivo growth of CD133 -transfected þ was carried out using SsoAdvanced SYBR Green Supermix from cells, mice were injected with CD133 cells transfected with miR- Bio-Rad with mouse primers for IL10, TGFb1, TGFb2, TGFb3, 92a mock (HiPerfect reagent only) or mimic (as described below), Smad2, ITGB1, ITGB3, ITGA5, and ITGAV customized and and tumor volume was measured. On day 15, mice were sacri- ordered from IDT. All PCR experiments used a CFX96 Touch ficed, tumors were dissociated, and labeled with antibody panels Real-Time PCR Detection System (Bio-Rad), and expression levels for various immune phenotypes using flow cytometry. were normalized to b-actin mRNA levels. Fold changes were calculated using the 2 DDCt method. Specific primers sequences Transfection of miR-92a mimics and inhibitors þ are provided in Supplementary Table S1. In brief, CD133 cells (1.5 105/well in 0.5 mL) postsorting were cultured in 24-well plates at 37 C, 5% CO2. The following Immunoblot and densitometry analysis day (24 hours postseeding), transfection was performed follow- Cells were harvested and resuspended in RIPA (150 mmol/L ing the manufacturer's protocol. Seventy-five ng of miR-92a NaCl, 1.0% IGEPAL CA-630, 0.5% sodium deoxycholate, 0.1% mimic, miR-92a inhibitor, or miR-92a mimic þ inhibitor (to a SDS, 50 mmol/L Tris, pH 8.0) buffer (Sigma #20-188) containing final concentration of 10 nmol/L) were diluted in 100 mLof a protease inhibitor cocktail (Sigma #P8340) and PhosStop culture medium without serum. HiPerFect Reagent (4.5 mL; Phosphatase Inhibitor (Roche # 04906845001). concen- Qiagen, #301705) was added to the diluted miR-92a mimic, trations of cell lysates were determined by a Bicinchoninic Acid miR-92a inhibitor, or miR-92a mimic þ inhibitor. The reagents Assay (Thermo Fisher Scientific #23225) and 40–60 mg of total were incubated for 10 minutes at room temperature to allow for protein was loaded per lane on 10% Tris-Gly Gels (Bio-Rad the formation of transfection complexes. The complexes were #4561033), subjected to SDS-PAGE, and transferred to a nitro- added to their respective wells and subsequently mixed by pipet- cellulose membrane using the iBlot System (Invitrogen). Lysates ting to ensure uniform dilution of the transfection complexes. The were probed with antibodies that recognize phosphorylated culture medium was changed after 12–15 hours. Following the SMAD2 (Cell Signaling Technology, #8828S), total SMAD2 (Cell change in medium, cells were incubated for 72 hours at 37C, 5% Signaling Technology, #5678S), b-Actin (Cell Signaling Technol- CO2. The cells were collected 72 hours posttransfection and used ogy, #4970S), Integrin b1 (Cell Signaling Technology, #4749T), for miRNA assays or gene expression. Primer assays and gene Integrin b3 (Cell Signaling Technology, #4749T), Integrin av (Cell expression were determined by RT-PCR and are described in Signaling Technology, #4749T), and Integrin a5 (Cell Signaling Material and Methods. Snord96a (#3150530, Qiagen) was used Technology, #4749T), and GAPDH (Cell Signaling Technology, as an internal control for miR-92a expression and Actin (primer #5174S). Densitometry and image analysis were performed using sequences provided previously) was used to normalize gene a ChemiDoc station equipped with ImageLab Software (Bio-Rad). expression. Densitometry analysis of bands of interest from immunoblots was performed using ImageJ software. Coculture and ELISA Sorted tumor populations were cultured alone or with freshly Oncosphere formation assay isolated whole-splenic cells at a 1:1 ratio (1 106 total cells) in B16-F10–sorted populations were isolated on the basis of 100 mL of serum-free media for 24 hours in cell culture–treated 96- CD133 positivity as described previously. Sorted cells were cul- well plates (Corning #3595). The resulting supernatants were tured in low-adherent 6-well plates (Corning) in SFM at a density centrifuged at 400 g to remove cells and debris, and frozen at of 1 103 cells/mL. Cultures were grown for up to 10 days and 80C until analysis. A free TGFb precoated ELISA kit was used amended with fresh SFM media twice per week. Oncospheres (BioLegend #437707) to determine the concentration of active (>100 mm) were counted and imaged using an EVOS Light TGFb in each sample following the manufacturer's recommended

3624 Cancer Res; 79(14) July 15, 2019 Cancer Research

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 2019 American Association for Cancer Research. Published OnlineFirst April 23, 2019; DOI: 10.1158/0008-5472.CAN-18-2659

miR-92 Regulates Immunosuppression by Cancer Stem Cells

þ protocol. Splenic cells from each well were isolated by centrifu- compared with CD133 cells (Fig. 1D). Not only were CD133 gation at 400 g for 10 minutes and labeled using the fluor- cells capable of generating significantly more floating spheres ophore-conjugated antibodies described previously. Following (106.8 11.6) compared with CD133 cells (41.8 10.7), but þ labeling, cells were washed and resuspended in 500 mL of staining oncospheres from CD133 cells were observed to be much larger buffer for flow cytometric analysis. in diameter (Fig. 1D), suggesting that anchorage-independent þ survival and proliferation was enhanced in the CD133 Analysis of The Cancer Genome Atlas samples population. Expression data from cutaneous melanoma samples contained þ in The Cancer Genome Atlas (TCGA) database were assessed and Tumors initiated by CD133 cells had a more visualized using cBioPortal (23, 24) and UCSC Xena (http://xena. immunosuppressed TME compared with CD133 cells þ ucsc.edu/). Queried included ITGAV, ITGA5, and TGFB1. Following syngeneic transplantation of CD133 or CD133 Correlations based on the Pearson coefficient are represented in B16 tumor cells into C57Bl/6 mice, we allowed palpable tumors each scatter plot, and represent mRNA expression data for each of to grow to approximately 1 cm3 before resecting the tumors. the genes provided. Mutational status for each queried gene is also Following resection, tumors were enzymatically dissociated and provided above each mRNA expression heatmap and explained in labeled to determine the infiltrating immune cell phenotypes in þ detail in the appropriate figure legend. Correlation analysis CD133 and CD133 -initiated tumors. The data from a repre- between miR-92a expression and the genes described above were sentative experiment are shown in Fig. 1E–G and from multiple þ also conducted using Stata 14 (StataCorp, 2015) from cutaneous experiments are summarized in Fig. 1H. CD133 -initiated tumors melanoma datasets containing both miRNA and mRNA expres- were observed to not only grow faster and larger than those sion data available from the TGCA study. Simple linear regression initiated by CD133 cells, but were also observed to have higher analysis was performed to predict expression of integrin alpha abundance of tumor-infiltrating Tregs, granulocytic MDSCs þ þ þ þ subunits as a function of miR-92 expression using Stata. Figures (gMDSC; CD45 CD11b GR1 Ly6G ), and M2 macrophages þ þ þ þ reflecting these analyses are visualized by a scatter plot of the (CD45 CD11b F4/80 CD206 ; Fig. 1E–I). Infiltrating macro- þ þ þ original data, the linear regression line, and the 95% confidence phages, identified as CD45 F4/80 CD11b were reduced within þ þ intervals of the regression line. the TME of CD133 -generated tumors (0.53% of total CD45 þ cells) compared with CD133 tumors (1% of total CD45 cells). þ þ Statistical analysis No changes in pan-T cell (CD45 CD3 ) or pan-MDSC þ þ þ GraphPad Prism 5.0 Software (GraphPad Prism Software, Inc.) (CD45 CD11b GR1 ) populations were observed. Concurrent- was used for all statistical analyses. For all in vitro studies, two- ly, we observed a significant increase in T cells staining positive for þ group comparisons between control and test samples were done TGFb as well as IL17a in tumors initiated by CD133 cells þ by two-tailed Student t test and representative data from three (Fig. 1E). TGFb T cells increased from 13.6% in CD133 tumors þ independent experiments were presented. A one-way ANOVA was to 25.2% in tumors generated by CD133 cells. We also observed þ þ performed on in vitro experiments containing more than one that Helios Tregs were significantly increased in CD133 tumor group, and significance was determined and denoted for each samples increasing from 4.9% in CD133 tumor samples to þ group accordingly. Subcutaneous and experimental metastasis nearly 10% in CD133 tumors. Although we did not observe a in vivo data were analyzed for significance using two-way ANOVA difference in macrophage or MDSC populations, we did find that þ and a two-tailed Student t tests, respectively. For all tests, statistical tumors initiated by CD133 cells had significantly increased significance was assumed when P < 0.05. P values were reported in proportions of infiltrating gMDSC (10.1% of total MDSCs in þ each figure or in their respective figure legends. CD133 group vs. 2.9% of total MDSCs in CD133 group) and þ alternate macrophages (13.4% of total macrophages in CD133 group vs. 8.9% of total macrophages in CD133 group). To Results highlight the changes associated with tumor initiation between þ þ CD133 B16-F10 cells are functionally distinct from CD133 CD133 and CD133 cells, we have included a representative cells, both in vitro and in vivo SPADE tree from pooled tumor samples (Fig. 1I). SPADE analysis þ To study the differential characteristics of CD133 and shows a significant increase in immunosuppressive phenotypes CD133 B16-F10 cells, we injected 2 105 B16-F10 cells from including Tregs and cells staining positive for TGFb or IL10. Taken each phenotype subcutaneously into C57Bl/6 mice. We observed together, these data suggest that more suppressive immune cells þ a 58% increase in mean tumor volume and a 52% increase in infiltrated CD133 tumors allowing for superior tumor growth þ mean wet tumor weight in the CD133 group compared with when compared with CD133 cells. þ CD133 group. Not only did CD133 cells form palpable tumors quicker than CD133 cells, they were also more tumorigenic miRNA microarray identified miR-92 as a regulator of integrin þ (Fig. 1A and B). CD133 cells formed tumors in 6/6 mice, while expression CD133 cells only formed tumors in 4/6 mice (Fig. 1B). Using We used an Affymetrix microarray to screen expression profiles þ þ in vitro functional assays, we observed that CD133 cells had a of several thousand miRNAs in CD133 and CD133 B16-F10 higher propensity to proliferate, form colonies, and generate cells. Analysis of data showed that of the 3,195 miRs screened, anchorage-independent oncospheres when compared with 2,995 miRs were common to these two cell types while 144 miRs þ þ CD133 cells (Fig. 1B–D). In colony-forming assays, CD133 were downregulated and 56 upregulated in CD133 cells when cells generated an average of 42.8 8.4 colonies, while CD133 compared with CD133 cells (Fig. 2A). A comprehensive list of all cells only generated an average of 18.2 3.0 colonies (Fig. 1C). miRNAs with greater than 2-fold change difference between We also observed a significant increase in the ability to form samples is provided (Supplementary Table S2). The microarray þ anchorage-independent oncospheres in CD133 populations also identified miRNAs of the miR-297-669 cluster to be

www.aacrjournals.org Cancer Res; 79(14) July 15, 2019 3625

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 2019 American Association for Cancer Research. Published OnlineFirst April 23, 2019; DOI: 10.1158/0008-5472.CAN-18-2659

Shidal et al.

Figure 1. CSCs are enriched in the CD133þ population and are functionally distinct from their CD133 counterparts. A and B, CD133þ cells form palpable tumors and display elevated growth kinetics in a syngeneic mouse model (n ¼ 6/group). Tumor volumes represent mean tumor volume SEM. Mice (4/6) injected with CD133 cells formed tumors while all (6/6) mice injected with CD133þ cells formed tumors. In vitro colony formation (C) and nonadherent oncosphere formation (D)was significantly increased in CD133-expressing populations. Images depict anchorage-dependent colony growth (C) and anchorage-independent oncosphere growth (D) in SFM media and are representative of data collected from three independent experiments. Statistical significance was determined at P < 0.05 and is denoted by an asterisk (). P values have been provided where appropriate. Dissociation, labeling, and analysis of representative tumor samples initiated by CD133þ and CD133 B16-F10 melanoma cells demonstrated a significant shift in lymphocyte (E), MDSC (F), and macrophage (G) populations. H, Statistical analysis on samples generated from CD133þ- and CD133-initiated tumors identified several significant changes associated with each group. I, SPADE analysis further demonstrated the alterations in immune cell infiltration of the TME between CD133þ cells and CD133 cells. Flow plots and SPADE analysis were generated from representative data collected from two independent in vivo experiments. Significance was determined by Student t test (P < 0.05) and is denoted by an asterisk. , P < 0.05; , P < 0.01; , P < 0.001.

þ downregulated in CD133 cells including miR-669a-5p, miR- 0.92 and 0.90, respectively (Fig. 2D). With this highly predic- 669l-5p, and miR-446h-5p, which was validated by qRT-PCR tive targeting of integrin subunits coupled with the mechanistic (Fig. 2A). The data confirmed that these miRNAs were, in fact, relationship between integrin activation of TGFb,wefurther þ downregulated in CD133 cells with relative expression levels explored the relationship between miR-92a and TGFb in an (normalized to CD133 ) of 0.49, 0.51, 0.57, and 0.49 for miR- in vivo model. 466h, miR-669a, miR-669l, and miR-92a, respectively (Fig. 2A). Because the miR-297-669 cluster is not present in humans (but is Integrin alpha-subunit expression and TGFb signaling through þ conserved in rodents), further analysis of miRs in this cluster were SMAD2 are enhanced in CD133 populations not selected for further characterization. Assessment of miR-92 The increase in TGFb from dissociated primary tumors formed using Metacore (Fig. 2B) and Ingenuity (Fig. 2C) pathway analysis by CSCs, along with mechanical integrin-dependent TGFb acti- tools identified a network of genes associated with melanoma vation, lead us to initially look for miRNAs targeting integrins. progression including CDC42, PTEN, and MAP2K (Fig. 2C) and Microarray analysis in combination with queried miR databases have been identified for clarity by blue circles. With the recent led to the identification of integrins as potential targets of miR-92a discovery that miR-92 could regulate integrin subunit expres- (Fig. 2D); thus, we next utilized qRT-PCR and immunoblot sion (20), we used predictive sequence alignment software to analysis to determine whether or not integrin expression was þ explore potential integrins, which may be targeted by miR-92. significantly increased in CD133 cells when compared with Integrin av and a5 were highly predicted targets of miR-92 with CD133 cells (Fig. 3). Normalized integrin mRNA expression þ weighted context scores of 0.28 and 0.34, and PCT values of was increased in CD133 cells compared with CD133 cells by

3626 Cancer Res; 79(14) July 15, 2019 Cancer Research

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 2019 American Association for Cancer Research. Published OnlineFirst April 23, 2019; DOI: 10.1158/0008-5472.CAN-18-2659

miR-92 Regulates Immunosuppression by Cancer Stem Cells

Figure 2. Microarray analysis of CD133þ B16-F10 cells compared with CD133 cells. Analysis of microarray data demonstrated significant disparities in miR expression between CSC and non-CSC compartments outlined in blue (CD133þ, orange; CD133, green); all miRNAs are shown in the above heatmap and are sorted by fold change (CD133þ relative to CD133 in ascending order). A, Several of these miRNAs were validated using qRT-PCR. miR-92 was identified to target several cancer-associated gene networks using Metacore (B) and Ingenuity (C) pathway analysis tools. D, Using sequence alignment software, miR-92 is highly predicted to target mRNAs for integrin alpha subunits involved in activation of secreted TGFb.

þ 1.5-, 2.5-, 1.3-, 1.6-fold for integrin b1, b3, av, and a5 subunits, tases, metastatic outgrowths initiated by CD133 cells demon- respectively (Fig. 3A). Protein levels of integrin a subunits were strated a significantly higher proportion of immunosuppressive þ also increased as assessed by immunoblot from two independent cell phenotypes including gMDSCs, and TGFb Tregs (Fig. 4B and experiments in which B16 cells were isolated via FACS based on C). In myeloid panels, we observed a significant increase in CD133 expression (Fig. 3B). qRT-PCR analysis of several isoforms gMDSCs with a marginal decrease in mMDSCs; macrophage of TGFb identified TGFb1 as the primary isoform responsible for populations did not exhibit any significant change in lung tissues the disparities in protein expression. No significant difference was from either tumor cell type (Fig. 4C). Spleens from each group observed in TGFb3 between the two groups (Fig. 3A), and TGFb2 showed a significant shift in mMDSCs as well as M1 macrophage was too lowly expressed to amplify using our parameters. All qRT- populations (Supplementary Fig. S4). In the lymphocyte panel, þ þ PCR experiments validating mRNA expression utilized b-actin as a we observed a significant increase in total T cell (CD45 CD3 ) þ þ housekeeping gene. Although qRT-PCR analysis determined a and NK cells (CD45 NK1.1 )infiltration along with a concurrent þ þ significant difference in SMAD2 mRNA expression, we did not decrease in CD8 T cells in the lungs of CD133 cell–initiated observe any difference in protein level expression in our Western mice when compared with mice in the CD133 group (Fig. 4D). A þ blot analysis; however, phosphorylated SMAD2, an indicator of significant increase in the percentage of TGFb Tregs þ þ þ þ þ þ þ TGFb signal activation, was significantly induced in CD133 (CD45 CD3 CD4 FOXP3 ) as well as IL17a CD4 T cells was þ þ tumor cells when compared with CD133 cells (Fig. 3B). Full observed along with a decrease in IFNg CD4 T-cell populations þ images of the exposed membranes have been provided (Supple- in CD133 when compared with CD133 samples (Fig. 4D). The þ mentary Fig. S3). spleens of CD133 -transplanted mice also showed similar results with a significant decrease in IFNg-producing cells as well as a þ CD133 cells generate more TGFb and induce Treg infiltration concurrent increase in Treg populations (Supplementary Fig. S4). in an experimental metastasis model SPADE analysis (Fig. 4E) highlighted the differences in immune þ þ When transplanted via tail vein injection, CD133 cells created cell tumor infiltration between CD133 -initiated lesions (right) larger and more abundant lesions when compared with CD133 and CD133 -initiated lesions (left) in both myeloid (bottom) þ cells (Fig. 4A). CD133 cells generated an average of 41.0 4.5 and lymphocyte (top) panels. Grayscale legends are provided for metastatic nodules compared with 22.6 3.6 nodules from mice each panel to discriminate populations identified for each SPADE þ receiving intravenous injection of CD133 cells. When dissoci- tree. These data suggested that tumors initiated by CD133 tumor ated and labeled for infiltrating immune cells, both phenotypes cells not only generated more immunosuppressive phenotypes were able to induce immune cell infiltration of the pulmonary within the TME, but also potentially resulted in less cytotoxic tissues; however, when compared with CD133 -initiated metas- T-cell response as well.

www.aacrjournals.org Cancer Res; 79(14) July 15, 2019 3627

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 2019 American Association for Cancer Research. Published OnlineFirst April 23, 2019; DOI: 10.1158/0008-5472.CAN-18-2659

Shidal et al.

Figure 3. þ RNA and protein expression analysis of integrin subunits and TGFb-associated signaling molecules. A, Total RNA was isolated from CD133 and CD133 cells and assessed for integrin subunit expression (top) and TGFb signaling through SMAD2 (bottom) by qRT-PCR. mRNA expression levels were normalized to the CD133 cell phenotype. B, Protein level expression for integrin av and a5 subunits as well as SMAD2 phosphorylation was assessed by Western blot analysis. Actin and GAPDH were used as reference proteins for Western blots assessing integrin subunit expression and SMAD signaling, respectively. Band intensities were calculated using ImageJ software and relative densitometric intensity (normalized to Actin/GAPDH) are displayed below the appropriate band. PCR and immunblot analysis data were gathered from two independent experiments in which samples were run in triplicate (qRT-PCR) or in duplicate (immunoblot). Significance was determined by Student t test (P < 0.05) and is denoted by an asterisk. , P < 0.05; , P < 0.01; , P < 0.001.

þ Free/active TGFb was increased in cocultures of CD133 tumor miR-92a mimic, inhibitor, and a combination of mimic and cells and immune cells when compared with cocultures using inhibitor for 48 hours. Gene expression was measured by qRT- CD133 cells PCR for miR-92a, ITGAV, ITGA5, TGFB1, and SMAD2 following þ After isolating CD133 and CD133 cell populations, we used the transfection and isolation of total RNA; all groups were þ a coculture model with splenic cells to determine whether either normalized to CD133 mock-transfected cells as a reference. phenotype was able to readily polarize splenocytes to the immu- When transfected with a miR-92 mimic, we observed an 8.2-fold nosuppressive phenotypes observed in our dissociated tumor increase in miR-92a expression. Relative expression levels of samples. Concurrently, we measured the concentrations of active ITGAV, ITGA5, TGFB1, and SMAD2 mRNA were 0.20, 0.17, TGFb in coculture system as well as in splenocytes and tumors 0.27, and 0.19, respectively (Fig. 6A). Conversely, transfection cells alone. Active TGFb was significantly enhanced in cocultures with an inhibitor of miR-92a decreased miR-92a expression, þ using CD133 and splenic cells (4.9 0.3 and 6.6 1.6 pg/mL at while significantly increasing mRNA levels for ITGAV, ITGA5, 8 and 24 hours, respectively) when compared with cocultures TGFB1, and SMAD2 (2.7, 4.7, 2.8, and 2.4-fold increases, using CD133 cells (2.3 0.5 and 3.9 0.9 pg/mL at 8 and 24 respectively). This phenotype was partially rescued when both þ hours, respectively; Fig. 5A). Interestingly, we did not observe any mimic and inhibitor were transfected into CD133 B16 cells þ differences in active TGFb when CD133 and CD133 cells were returning to baseline mRNA levels of ITGA5, TGFB1, and cultured alone at either time point; although, samples from both SMAD2. These results indicated that miR-92a was involved tumor phenotypes contained significantly more free TGFb than in the expression of genes that regulate TGFb signaling and splenocytes cultured alone (Supplementary Fig. S5). When cells activation (Fig. 6A). isolated from our cocultures were analyzed by flow cytometry, we þ þ þ observed a significant increase in CD11b TGFb cells in cocul- Transfection of CD133 cells with miR-92a mimic suppressed þ tures of CD133 cells when compared with CD133 cocultures tumor initiation and growth through immune alterations in (Fig. 5B–D). No significant changes were observed in T-cell TME populations (Fig. 5C and D). SPADE analysis of flow cytometric To test whether alterations in the expression of miR-92a in þ data is provided to further represent the shift in TGFb-producing CD133 cells would change tumor growth and immune response þ myeloid cells we observed in our samples. in vivo, CD133 cells isolated from the B16 cell line were trans- fected with mock or miR-92 mimic. Transfected cells were next miR-92a regulates integrin and TGFb expression injected subcutaneously into C57Bl/6 mice and allowed to form To directly test whether miR-92a targeted the integrins, B16-F10 tumors over 14 days (Fig. 6B). Upon endpoint of the experiment, cells sorted on positive CD133 expression were transfected with tumors initiated by cells transfected with miR-92 mimic showed a

3628 Cancer Res; 79(14) July 15, 2019 Cancer Research

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 2019 American Association for Cancer Research. Published OnlineFirst April 23, 2019; DOI: 10.1158/0008-5472.CAN-18-2659

miR-92 Regulates Immunosuppression by Cancer Stem Cells

Figure 4. Experimental metastasis model of murine melanoma using CD133-sorted cell populations. A, CD133-sorted B16-F10 cells were injected intravenously into C57Bl/ 6 mice and allowed to colonize the pulmonary tissues. Pulmonary lesions were counted and measured upon resection of lung tissues and plotted as mean number of metastases SEM. CD133þ cells formed larger and more abundant macrometastases when compared with CD133 cells. B, Lungs from tumor-bearing mice were dissociated and labeled using myeloid (top) and lymphocyte (bottom) panels to identify shifts in subsets of T cells, macrophages, and MDSCs from the TME generated by each tumor phenotype. Representative flow plots from lung tissues analyzed by myeloid (C) and lymphocyte (D) panels demonstrated the significant shifts in immune cell phenotypes. E, Downstream analysis using SPADE software depicted changes in the TME between tissues colonized by CD133þ and CD133 melanoma cells. Legend provided is based on these analyses (grayscale). The panels identified significant shifts in immune cell phenotypes in both lymphocyte (top) and myeloid (bottom) panels. Experimental metastasis models were repeated once (n ¼ 5/group). Statistical significance was determined by Student t test and is denoted by an asterisk. P values are provided where appropriate.

þ significantly decreased tumor burden with an average tumor When we analyzed subsets of Th cells (CD4 ), we found that þ volume of 114.4 mm3. This represents a 58% decrease in tumor immunosuppressive phenotypes of FOXP3 Tregs, IL10, and þ volume from the mean tumor volume of the mock-transfected TGFb-producing CD4 cells were significantly decreased þ CD133 cells of 272.8 mm3. Although all mice in both groups following miR-92 mimic transfection (17.8% vs. 4.3%; formed tumors, the significant reduction in tumor volume sug- 34.5% vs. 11.0%; 28.3% vs. 5.7%, respectively). Antitumor gested that miR-92a expression had an inverse association with proinflammatory phenotypes producing IFNg were significant- tumor growth. It was also observed that cells transfected with miR- ly increased from 0.5% to 2.3% when compared with tumor 92 mimic showed a significant delay in tumor formation com- tissues from mock-transfected group. þ pared with CD133 mock-transfected cells. Tumor-bearing mice were sacrificed upon endpoint and Analysis of miRNA and mRNA coexpression data identified tumors were dissociated, labeled with panels of antibodies associations between miR-92a and signaling pathways involved against phenotypic markers for lymphocytes and monocytes, in TGFb signaling and immune response andanalyzedbyflow cytometry (Fig. 6C–F). In tumors initiated To associate our findings in preclinical murine models of by miR-92/mimic, we observed a significant shift in tumor- melanoma to clinical data, we used the publicly available TCGA infiltrating cells with an overall increase in total T cells (4.9% database to authenticate our results in human samples. vs. 1.8%), but a decrease in pan-MDSC (3.0% vs. 1.8%) and We specifically identified samples taken from patients with pan-macrophage (1.7% and 0.5%) populations when com- melanoma and observed a moderate positive correlation pared with tumors initiated by the mock-transfected cells. between integrin a5 and TGFb1 expression levels (Pearson

www.aacrjournals.org Cancer Res; 79(14) July 15, 2019 3629

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 2019 American Association for Cancer Research. Published OnlineFirst April 23, 2019; DOI: 10.1158/0008-5472.CAN-18-2659

Shidal et al.

Figure 5. ELISA for free/active TGFb in a coculture model demonstrated the enhanced ability of CSCs to convert secreted (inactive) TGFb to its active form. A, FACS- sorted CD133þ and CD133 populations were cocultured with splenocytes from C57Bl/6 mice for 8 hours (top) and 24 hours (bottom) at a 1:1 E:T ratio, and the resulting supernatants were analyzed for activated TGFb. B–D, Flow cytometry and subsequent SPADE analysis was conducted on the resulting splenocytes to identify shifts in cell phenotypes after 24-hour coculture. A legend for SPADE analysis has been provided (B). Flow cytometry panels resulting from cocultures using both CD133þ (C) and CD133 (D) are also provided. ELISA and flow cytometric analysis were repeated twice as independent experiments. Statistical significance was determined by Student t test (P < 0.05) and is denoted by an asterisk.

coefficient ¼ 0.66; Fig. 7A and B). Further analysis identified Discussion positive correlations between integrin a and LTBP1 (Pearson 5 The main goal of this study was to investigate whether CD133 coefficient ¼ 0.67) as well as NRP-1 (Pearson coefficient ¼ expression on CSCs would alter the expression of miRNA, which 0.69; Fig. 7B). These associations between groups of genes would target genes involved in the regulation of immune response involved in TGFb activation and signaling in human samples and consequently control tumor growth. On the basis of the validated our preclinical studies in murine models and indicate expression of CD133, this study identified a potential role for that CSCs may, in fact, use these signaling molecules and their miR-92 in regulating immunosuppression by mediating integrin- connected signaling networks to evade immune-mediated dependent TGFb activation. We showed that CSCs based on the tumor ablation. We next identified datasets consisting of both CD133 biomarker are functionally distinct from the bulk tumor miRNA and mRNA expression to explore the relationship population and demonstrate superior tumorigenicity and growth between miR-92a and the genes involved in TGFb activation in vitro and in vivo. These results are in line with previous studies (i.e., integrins). Using these publicly available data, we describing CD133 as a biomarker for CSCs present in melano- observed a moderate inverse association between integrin mas (3, 25), and that these cell populations have intrinsic che- alpha 5 and alpha V subunits, and miR-92a using Spearman moresistant (26), angiogenic (27), and metastatic properties (28). (0.33 and 0.38, respectively) and Pearson correlation þ CD133 B16-F10 melanoma cells had enhanced tumor growth in (0.30 and 0.37, respectively) coefficients. Linear regression subcutaneous tumor growth models and established larger and analysis and the resulting graphics are provided to help visu- more abundant pulmonary lesions in our experimental metasta- alize the relationship described previously (Fig. 7); as miR-92a sis model compared with CD133 cells. Moreover, tumors and expression increased, we saw a subsequent reduction in integrin þ metastatic lesions initiated by CD133 cells were significantly a and a expression (Fig. 7C). Along with the scatter plot of V 5 more immunosuppressed than CD133 -initiated tumors as the original data, the regression line from simple linear regres- assessed by intratumoral abundance of tumor-associated macro- sion analysis, and 95% confidence intervals of the regression phages and Tregs. Indeed, CSCs in several cancer models have line were also provided.

3630 Cancer Res; 79(14) July 15, 2019 Cancer Research

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 2019 American Association for Cancer Research. Published OnlineFirst April 23, 2019; DOI: 10.1158/0008-5472.CAN-18-2659

miR-92 Regulates Immunosuppression by Cancer Stem Cells

Figure 6. miR-92a regulated the integrin/TGFb axis and inhibited tumor growth in vivo. A, B16-F10 cells sorted on positive CD133 expression were transfected with miR- 92a mimic, inhibitor, and a combination of mimic and inhibitor for 48 hours. Gene expression was measured by qRT-PCR for miR-92a, ITGAV, ITGA5, TGFB1, and SMAD2 following the transfection and isolation of total RNA; all groups were normalized to mock-transfected cells as a reference. CD133þ cells isolated from the B16 cell line were transfected with miR-92 mimic or lipid. B, Transfected cells were injected subcutaneously into C57bl/6 mice and allowed to form tumors over 14 days. C–F, Tumor-bearing mice were sacrificed upon endpoint and tumors were dissociated, labeled with panels of antibodies against phenotypic markers for lymphocytes and monocytes, and analyzed by flow cytometry as described previously. Statistical analyses were performed using a Student t test and one-way ANOVA with significance determined at P < 0.05. Statistical significance is denoted in A as follows: a ¼ significant from 2,3,4; b ¼ significant from 1,3,4; c ¼ significant from 1,2,4; d ¼ significant from 1,2,3; e ¼ significant from 2,3. Statistical significance in F is denoted as follows: , P < 0.05; , P < 0.01; , P < 0.001.

been reported to exploit immune cells to create a tumor-tolerant downregulated in CSCs compared with non-CSCs. Interestingly, niche during tumorigenesis and metastasis (29–31). Infiltration it was recently reported that members of this cluster directly of the TME by these immune cell phenotypes resulted in signif- regulated TGFb2 (40). Another miRNA identified by our micro- icantly higher production of TGFb, shifts in MDSC populations array screen, miR-92a, is characterized as an oncomiR in various toward a granulocytic phenotype, and lower IFNg-producing cells cancers (16, 41, 42) and is being employed as a potential serum þ in tumors initiated by CD133 cells when compared with tumors biomarker for certain malignancies (43); however, miR-92 can initiated by CD133 cells. Secretion and activation of TGFb is a also act as a tumor suppressor in other cancers (18, 44, 45). well-studied mechanism of immunosuppression and has been Downregulation of miR-92 in human breast cancers was associ- described in melanomas (32) as well as other models (33). ated with poor prognosis and correlated with stage and disease- Interestingly, blockade of TGFb signaling in a B16-F10 melanoma free survival. Interestingly, the researchers also observed a signif- model led to a T-cell–mediated eradication of tumors (34). Other icant increase in macrophage infiltration; however, the phenotype studies have found that TGFb blockade was sufficient for signif- (M1/M2) of these tumor-infiltrating macrophages was not icantly increasing antitumor immune responses (35), and helps reported (46). Modulation of macrophage populations by B16- promote response to immune checkpoint inhibitors in different F10 cells has been linked to disease progression and metasta- mouse models of melanoma (36, 37). In our model, we observed sis (47), thus miR-92a may have far-reaching effects outside of a significant increase in Tregs, a phenotype that can be induced by regulating TGFb signaling mechanisms. In fact, it was recently TGFb (38, 39), in CSC-initiated tumors when compared with shown that miR-92 was acknowledged to alter miRNA profiles of tumors generated from non-CSCs. These data indicate that tumor induced pluripotent stem cells, an effect that was suggested to be growth and metastasis by B16-F10 cells may be driven by Treg- p53 mediated (48). Additional studies have advocated a role for mediated immunosuppression. miR-92 in neuroblast self-renewal and maintenance (49). Inter- Using microarray technology, we identified several miRs that estingly, a recent report by Huber and colleagues described several þ were downregulated in the CD133 CSC population, which miRNAs that were found in melanoma exosomes and mediated þ targeted TGFb and its associated gene networks. Preliminary data MDSC expansion and differentiation from CD14 mono- indicated that several miRNAs from the miR-297-669 cluster were cytes (50). Additional evidence from a study of exosomes,

www.aacrjournals.org Cancer Res; 79(14) July 15, 2019 3631

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 2019 American Association for Cancer Research. Published OnlineFirst April 23, 2019; DOI: 10.1158/0008-5472.CAN-18-2659

Shidal et al.

Figure 7. TCGA validates the association between integrin a5 and TGFb and miR-92 in human clinical samples of melanoma. A, Using cBioPortal and UCSC Xena, we probed datasets obtained from patients with skin cancer to identify correlations between mRNA level expression of integrin a5 and TGFb1. We identified several clinical specimens in which high expression of TGFb was associated with elevated expression of integrin a5 as highlighted by the rectangles (red, high expression; blue, low expression). B, Using the Pearson and Spearman coefficient, we determined a positive association between the two proteins. We also identified positive associations between integrin a5 and LTBP1 and NRP-1. C, Further analyses of TCGA datasets with available miRNA expression data demonstrated an inverse association between miR-92 expression and integrin expression for alpha-5 and alpha-V subunits. A simple linear regression model was used to predict integrin expression as a function of miR-92a expression and provided 95% confidence intervals for our regression line. Right, Spearman and Pearson correlation was performed using Stata 14 and coefficients have been provided.

which reported that miR-92a can stimulate the proliferation and by an integrin-dependent mechanism (58). qRT-PCR analysis of function of MDSCs (51), indicates that exosome-mediated trans- the integrin b1 and b3 subunits reflected higher mRNA expression þ fer of miRNAs may, in part, function to induce immunosuppres- in CD133 cells compared with CD133 . Interestingly, the sion within the TME. The functional effects of miR-92 seem to be integrin b3 was shown to regulate senescence through induction somewhat ubiquitous as well as tissue- and context-dependent; of the TGFb signaling pathway (59). In addition, we observed thus, more work exploring the functions of miR-92 in melanoma an increase in TGFb1 and activating phosphorylation of the is justified. downstream signaling molecule SMAD2. Our data demonstrate þ þ CD133 cells expressed higher integrin a5 and av (RGD-rec- that CD133 B16-F10 cells highly express components of integrin ognizing subunits) on both a protein and mRNA level when and TGFb-associated signaling cascades, which may, in turn, compared with CD133 cells. Integrin a5 has been reported to be provide a selective survival advantage in the context of immuno- regulated by miR-92 in an ovarian cancer model (20); however, suppression within the TME. Mechanistically, this axis may be whether miR-92 regulates the av subunit has yet to be clarified. regulated through miR-92 modulation of integrin-dependent The integrin av subunit has been explicitly characterized to TGFb activation. heterodimerize with integrin b subunits to form integrins avb3, Cocultures of splenic and tumor cells showed that while no avb5, avb6, and avb8, all of which have been reported to modulate significant change in active TGFb concentrations was observed TGFb activation (52). Integrin a5b1 is the major receptor for between CSCs and non-CSCs cultured alone, cocultures of þ fibronectin (53); fibronectin is required for TGFb activation (54) CD133 tumor and splenic cells resulted in significantly higher and fibronectin matrix assembly (55), suggesting the unique free TGFb when compared with cocultures using CD133 cells. possibility that integrin a5 may also play a role in the liberation Membrane-bound TGFb was significantly increased in tumors þ and activation of TGFb. In endothelial cells, it was shown that resulting from CD133 cells when compared with CD133 cells fibronectin and its receptor (e.g., integrin a5b1) mediated SMAD in subcutaneous and intravenous models of melanoma; however, phosphorylation following exogenous application of TGFb1 and liberated TGFb from each cell type remained unchanged in vitro. BMP-9 (56). Conversely, it was reported that TGFb1 may regulate These data indicated that there are significant interactions integrin a5b1 and integrin signal transduction (57). Further between immune cells and tumor cells that collaborate to produce interactions between TGFb1 and integrin a5b1 were reported in immunosuppression within the TME. In addition, they indicate T cells where TGFb-activated cells were protected from apoptosis that TGFb secretion and activation may involve multiple

3632 Cancer Res; 79(14) July 15, 2019 Cancer Research

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 2019 American Association for Cancer Research. Published OnlineFirst April 23, 2019; DOI: 10.1158/0008-5472.CAN-18-2659

miR-92 Regulates Immunosuppression by Cancer Stem Cells

mechanisms involving several cell phenotypes that are not reca- subunits than non-CSCs. Furthermore, these disparities in expres- pitulated in the culture of cancer cell lines. After 24 hours, splenic sion may have resulted in altered TGFb activation and subsequent cells from cocultures were labeled and analyzed by flow cytometry immunosuppression. Given our initial data, further studies to determine whether any changes in myeloid cell or lymphocyte exploring the miR-92/integrin/TGFb axis are warranted. populations were stimulated by either tumor cell phenotype. While our studies demonstrate a link between miR expression Although we did not see a significant change in regulatory T-cell in tumor cells and their ability to grow in vivo and modulate populations (as in our in vivo models), we did observe a significant immune cell phenotype in TME, it has some limitations and þ increase in TGFb myeloid cell populations in splenic cells opportunities. Use of the B16-F10 cell line, although commonly cocultured with CSCs compared with non-CSCs. It was recently used and easily accessible, may not mimic human disease as described that B16-F10 tumors undergo significant changes in accurately as newer model systems including the Yale University immune cell infiltration depending upon the stage of disease (60). Mouse Model (YUMM; ref. 65). The YUMM systems have fully These studies suggest the intriguing concept that palpable tumor characterized driver mutations, are genomically stable, and syn- formation and initial immunosuppression may be driven by geneic to the C57Bl/6 mouse strain providing a much more immunosuppressive myeloid cell types (i.e., alternate macro- comprehensive understanding of the underlying genetic interac- phages), while late-stage tumor growth and metastasis is con- tions in progressing disease states. Our microarray analysis was trolled by Treg populations. In fact, an increased production of only performed using a sample size of one (i.e., one chip/ immature myeloid cells was observed in patients with cancer (61). experimental group) lending limited interpretability based on In these studies, immature myeloid cells inhibited antigen-stim- our significantly altered miRNAs. Nonetheless, we used micro- ulated T-cell responses, which may help explain the significant array only as a screening tool and validated the expression of miRs þ decrease in CD8 cytotoxic T lymphocytes in our CSC-induced of interest using RT-PCR. Finally, when reviewing TCGA data, only metastasis models, and potentially clarify the disparities in tumor- a small proportion of cases (limitedto cutaneousmelanomas)had associated immunophenotypes between our in vivo models and both RNA and miRNA sequencing data available. Because of the those characterized in our in vitro coculture models. small sample size, a more rigorous study involving a larger sample Finally, we characterized miR-92 to functionally modulate size would be needed to understand the translational impact. expression of integrin subunits as well as mediators of TGFb Nonetheless, our study also forms the basis for further explor- signaling, and correlate these in vitro data with clinical data ing the role of miRs in CSCs and may be of particular interest to derived from the TGCA database. When mimics and inhibitors researchers and clinicians exploring new therapeutics modalities þ of miR-92a were transfected into B16 CD133 cells, the resulting targeting miRs thereby modulating the complex interactions expression profiles supported our original hypothesis that miR-92 between tumor and immune cells in the TME. Future studies controls TGFb-induced immunosuppression. Expression of elucidating miR expression in tumor cells and patient's response ITGA5, ITGAV, TGFB1, and SMAD2 were all significantly affected to immunotherapy or characterizing the degree of immunosup- when expression of miR-92a was altered. Mice receiving B16 cells pression in patient prognosis and treatment, would also be with transfected miR-92a mimic showed significantly reduced relevant clinically. In summary, our study found that melanoma tumor growth compared with mock-transfected B16 cells. CSCs modulate the TME by regulating a miRNA-gene network We wanted to correlate our studies with clinical data from the consisting of miR-92a, integrin a5/av, and TGFb. TCGA database in which thousands of human tumor samples have been characterized. Exploring melanoma samples within the Disclosure of Potential Conflicts of Interest fi database, we identi ed a positive correlation between integrin a5 No potential conflicts of interest were disclosed. and TGFb1, LTBP1, and -1 (NRP-1). LTBP1 targets latent TGFb complexes to the ECM where it is subsequently Authors' Contributions activated (62). NRP-1 was recently shown to modulate TGFb Conception and design: C. Shidal, P. Nagarkatti, M. Nagarkatti signaling in (63). Interestingly, NRP-1 is a bio- Development of methodology: C. Shidal, P. Nagarkatti marker to distinguish natural and inducible Tregs (64), enforcing Acquisition of data (provided animals, acquired and managed patients, the idea that immunosuppression in melanomas might be T-cell– provided facilities, etc.): C. Shidal, P. Nagarkatti, M. Nagarkatti mediated. We also analyzed datasets containing both expression Analysis and interpretation of data (e.g., statistical analysis, biostatistics, of mRNA and miRNAs. When association analyses were per- computational analysis): C. Shidal, N.P. Singh, P. Nagarkatti fi Writing, review, and/or revision of the manuscript: C. Shidal, N.P. Singh, formed, we identi ed a moderate inverse correlation between P. Nagarkatti, M. Nagarkatti integrin subunits and miR-92a as predicted by several databases, Administrative, technical, or material support (i.e., reporting or organizing which aligned the sequences of miR-92a and integrin av and a5. data, constructing databases): C. Shidal, N.P. Singh, M. Nagarkatti We further performed a simple linear regression model to predict Study supervision: P. Nagarkatti, M. Nagarkatti integrin mRNA expression using miR-92a expression as an inde- pendent variable based on the available miRNA and mRNA Acknowledgments expression data. These data clearly showed a reduction in ITGAV This work was supported, in part, by NIH grants P01AT003961, and ITGA5 expression with increased expression of miR-92a. R01AT006888, R01AI123947, R01AI129788, R01MH094755, and Our study categorized miR-92 as a potential tumor suppressor P20GM103641 (to P. Nagarkatti and M. Nagarkatti). in melanoma by modifying TGFb-induced immunosuppression. The costs of publication of this article were defrayed in part by the payment of Disparities in miRNA expression between stem cell–like popula- page charges. This article must therefore be hereby marked advertisement in tions and bulk tumor cells may confer the properties attributed to accordance with 18 U.S.C. Section 1734 solely to indicate this fact. CSCs such as chemoresistance and metastatic outgrowth. While integrins are clearly involved in adhesion, we demonstrated that Received September 11, 2018; revised March 19, 2019; accepted April 18, CSCs express significantly higher levels of RGD-recognizing alpha 2019; published first April 23, 2019.

www.aacrjournals.org Cancer Res; 79(14) July 15, 2019 3633

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 2019 American Association for Cancer Research. Published OnlineFirst April 23, 2019; DOI: 10.1158/0008-5472.CAN-18-2659

Shidal et al.

References 1. Fang D, Nguyen TK, Leishear K, Finko R, Kulp AN, Hotz S, et al. A 25. Dou J, Pan M, Wen P, Li Y, Tang Q, Chu L, et al. Isolation and identification tumorigenic subpopulation with stem cell properties in melanomas. of cancer stem-like cells from murine melanoma cell lines. Cell Mol Cancer Res 2005;65:9328–37. Immunol 2007;4:467–72. 2. Shidal C, Al-Rayyan N, Yaddanapudi K, Davis KR. Lunasin is a novel 26. El-Khattouti A, Selimovic D, Haikel Y, Megahed M, Gomez CR, Hassan M. therapeutic agent for targeting melanoma cancer stem cells. Oncotarget Identification and analysis of CD133(þ) melanoma stem-like cells con- 2016;7:84128–41. ferring resistance to taxol: an insight into the mechanisms of their resistance 3. Monzani E, Facchetti F, Galmozzi E, Corsini E, Benetti A, Cavazzin C, et al. and response. Cancer Lett 2014;343:123–33. Melanoma contains CD133 and ABCG2 positive cells with enhanced 27. Zimmerer RM, Matthiesen P, Kreher F, Kampmann A, Spalthoff S, Jehn P, þ tumourigenic potential. Eur J Cancer 2007;43:935–46. et al. Putative CD133 melanoma cancer stem cells induce initial angio- 4. Schatton T, Murphy GF, Frank NY, Yamaura K, Waaga-Gasser AM, Gasser genesis in vivo. Microvasc Res 2016;104:46–54. M, et al. Identification of cells initiating human melanomas. Nature 2008; 28. Rappa G, Fodstad O, Lorico A. The stem cell-associated antigen CD133 451:345–9. (Prominin-1) is a molecular therapeutic target for metastatic melanoma. 5. Kumar D, Kumar S, Gorain M, Tomar D, Patil HS, Radharani NNV, et al. Stem Cells 2008;26:3008–17. Notch1-MAPK signaling axis regulates CD133(þ) - 29. Shipitsin M, Campbell LL, Argani P, Weremowicz S, Bloushtain-Qimron N, mediated melanoma growth and angiogenesis. J Invest Dermatol 2016; Yao J, et al. Molecular definition of breast tumor heterogeneity. Cancer Cell 136:2462–74. 2007;11:259–73. 6. Lindau D, Gielen P, Kroesen M, Wesseling P, Adema GJ. The immuno- 30. Sultan M, Vidovic D, Paine AS, Huynh TT, Coyle KM, Thomas ML, et al. suppressive tumour network: myeloid-derived suppressor cells, regulatory Epigenetic silencing of TAP1 in Aldefluor(þ) breast cancer stem cells T cells and natural killer T cells. Immunology 2013;138:105–15. contributes to their enhanced immune evasion. Stem Cells 2018;36: 7. Massague J. TGFbeta in cancer. Cell 2008;134:215–30. 641–54. 8. Wan YY, Flavell RA. `Yin-Yang' functions of transforming growth factor- 31. Schatton T, Frank MH. Antitumor immunity and cancer stem cells. Ann N Y beta and T regulatory cells in immune regulation. Immunol Rev 2007;220: Acad Sci 2009;1176:154–69. 199–213. 32. Yang L, Pang Y, Moses HL. TGF-beta and immune cells: an important 9. Shi M, Zhu J, Wang R, Chen X, Mi L, Walz T, et al. Latent TGF-beta structure regulatory axis in the tumor microenvironment and progression. and activation. Nature 2011;474:343–9. Trends Immunol 2010;31:220–7. 10. Tauriello DVF, Palomo-Ponce S, Stork D, Berenguer-Llergo A, Badia- 33. Principe DR, DeCant B, Mascarinas E, Wayne EA, Diaz AM, Akagi N, et al. Ramentol J, Iglesias M, et al. TGFbeta drives immune evasion in genetically TGFbeta signaling in the pancreatic tumor microenvironment promotes reconstituted colon cancer metastasis. Nature 2018;554:538–43. fibrosis and immune evasion to facilitate tumorigenesis. Cancer Res 2016; 11. Pang MF, Georgoudaki AM, Lambut L, Johansson J, Tabor V, Hagikura K, 76:2525–39. et al. TGF-beta1-induced EMT promotes targeted migration of breast cancer 34. Gorelik L, Flavell RA. Immune-mediated eradication of tumors through the cells through the lymphatic system by the activation of CCR7/CCL21- blockade of transforming growth factor-beta signaling in T cells. Nat Med mediated chemotaxis. Oncogene 2016;35:748–60. 2001;7:1118–22. 12. McEarchern JA, Kobie JJ, Mack V, Wu RS, Meade-Tollin L, Arteaga CL, et al. 35. Budhu S, Schaer DA, Li Y, Toledo-Crow R, Panageas K, Yang X, et al. Invasion and metastasis of a mammary tumor involves TGF-beta signaling. Blockade of surface-bound TGF-beta on regulatory T cells abrogates sup- Int J Cancer 2001;91:76–82. pression of effector T cell function in the tumor microenvironment. 13. Colak S, Ten Dijke P. Targeting TGF-beta signaling in cancer. Trends Cancer Sci Signal 2017;10:pii: eaak9702 2017;3:56–71. 36. Mariathasan S, Turley SJ, Nickles D, Castiglioni A, Yuen K, Wang Y, et al. 14. Visan I. Targeting TGF-beta in cancer. Nat Immunol 2018;19:316. TGFbeta attenuates tumour response to PD- blockade by contributing to 15. Hayes J, Peruzzi PP, Lawler S. MicroRNAs in cancer: biomarkers, functions exclusion of T cells. Nature 2018;554:544–8. and therapy. Trends Mol Med 2014;20:460–9. 37. Hanks BA, Holtzhausen A, Evans K, Held M, Blobe GC. Combinatorial 16. Wang H, Ke C, Ma X, Zhao Q, Yang M, Zhang W, et al. MicroRNA-92 TGF-beta signaling blockade and anti-CTLA-4 antibody immunotherapy in promotes invasion and chemoresistance by targeting GSK3beta and acti- a murine BRAF(V600E)-PTEN-/-transgenic model of melanoma. J Clin vating Wnt signaling in bladder cancer cells. Tumour Biol 2016 Nov 9 Oncol 2014;32:3011 [Epub ahead of print]. 38. Nakamura K, Kitani A, Strober W. Cell contact-dependent immunosup- 17. Tsuchida A, Ohno S, Wu W, Borjigin N, Fujita K, Aoki T, et al. miR-92 is a pression by CD4(þ)CD25(þ) regulatory T cells is mediated by cell surface- key oncogenic component of the miR-17–92 cluster in colon cancer. bound transforming growth factor beta. J Exp Med 2001;194:629–44. Cancer Sci 2011;102:2264–71. 39. Shevach EM. Mechanisms of foxp3þ T regulatory cell-mediated suppres- 18. Smith L, Baxter EW, Chambers PA, Green CA, Hanby AM, Hughes TA, et al. sion. Immunity 2009;30:636–45. Down-regulation of miR-92 in breast epithelial cells and in normal but not 40. Becker W, Nagarkatti M, Nagarkatti PS. miR-466a targeting of TGF-beta2 tumour fibroblasts contributes to breast carcinogenesis. PLoS One 2015; contributes to FoxP3(þ) regulatory T cell differentiation in a murine model 10:e0139698. of allogeneic transplantation. Front Immunol 2018;9:688. 19. Shin VY, Siu MT, Liu X, Ng EKO, Kwong A, Chu KM. MiR-92 suppresses 41. Ke TW, Wei PL, Yeh KT, Chen WT, Cheng YW. MiR-92a promotes cell proliferation and induces apoptosis by targeting EP4/Notch1 axis in gastric metastasis of colorectal cancer through PTEN-mediated PI3K/AKT path- cancer. Oncotarget 2018;9:24209–20. way. Ann Surg Oncol 2015;22:2649–55. 20. Ohyagi-Hara C, Sawada K, Kamiura S, Tomita Y, Isobe A, Hashimoto K, 42. Chen MW, Yang ST, Chien MH, Hua KT, Wu CJ, Hsiao SM, et al. The STAT3- et al. miR-92a inhibits peritoneal dissemination of ovarian cancer cells by miRNA-92-Wnt signaling pathway regulates spheroid formation and inhibiting integrin alpha5 expression. Am J Pathol 2013;182:1876–89. malignant progression in ovarian cancer. Cancer Res 2017;77:1955–67. 21. Shidal C, Inaba JI, Yaddanapudi K, Davis KR. The soy-derived peptide 43. Yang X, Zeng Z, Hou Y, Yuan T, Gao C, Jia W, et al. MicroRNA-92a as a Lunasin inhibits invasive potential of melanoma initiating cells. Onco- potential biomarker in diagnosis of colorectal cancer: a systematic review target 2017;8:25525–41. and meta-analysis. PLoS One 2014;9:e88745. 22. Qiu P, Simonds EF, Bendall SC, Gibbs KD Jr., Bruggner RV, Linderman MD, 44. Ottman R, Levy J, Grizzle WE, Chakrabarti R. The other face of miR-17–92a et al. Extracting a cellular hierarchy from high-dimensional cytometry data cluster, exhibiting tumor suppressor effects in prostate cancer. Oncotarget with SPADE. Nat Biotechnol 2011;29:886–91. 2016;7:73739–53. 23. Gao J, Aksoy BA, Dogrusoz U, Dresdner G, Gross B, Sumer SO, et al. 45. Tanaka M, Oikawa K, Takanashi M, Kudo M, Ohyashiki J, Ohyashiki K, Integrative analysis of complex cancer genomics and clinical profiles using et al. Down-regulation of miR-92 in human plasma is a novel marker for the cBioPortal. Sci Signal 2013;6:pl1. acute leukemia patients. PLoS One 2009;4:e5532. 24. Cerami E, Gao J, Dogrusoz U, Gross BE, Sumer SO, Aksoy BA, et al. The cBio 46. Nilsson S, Moller C, Jirstrom K, Lee A, Busch S, Lamb R, et al. Down- cancer genomics portal: an open platform for exploring multidimensional regulation of miR-92a is associated with aggressive breast cancer features cancer genomics data. Cancer Discov 2012;2:401–4. and increased tumour macrophage infiltration. PLoS One 2012;7:e36051.

3634 Cancer Res; 79(14) July 15, 2019 Cancer Research

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 2019 American Association for Cancer Research. Published OnlineFirst April 23, 2019; DOI: 10.1158/0008-5472.CAN-18-2659

miR-92 Regulates Immunosuppression by Cancer Stem Cells

47. Georgoudaki AM, Prokopec KE, Boura VF, Hellqvist E, Sohn S, Ostling J, 57. Cai T, Lei QY, Wang LY, Zha XL. TGF-beta 1 modulated the expression of et al. Reprogramming tumor-associated macrophages by antibody target- alpha 5 beta 1 integrin and integrin-mediated signaling in human hepa- ing inhibits cancer progression and metastasis. Cell Rep 2016;15:2000–11. tocarcinoma cells. Biochem Biophys Res Commun 2000;274:519–25. 48. Neveu P, Kye MJ, Qi S, Buchholz DE, Clegg DO, Sahin M, et al. MicroRNA 58. Rich S, Van Nood N, Lee HM. Role of alpha 5 beta 1 integrin in TGF-beta þ profiling reveals two distinct p53-related human pluripotent stem cell 1-costimulated CD8 T cell growth and apoptosis. J Immunol 1996;157: states. Cell Stem Cell 2010;7:671–81. 2916–23. 49. Yuva-Aydemir Y, Xu XL, Aydemir O, Gascon E, Sayin S, Zhou W, et al. 59. Rapisarda V, Borghesan M, Miguela V, Encheva V, Snijders AP, Lujambio A, Downregulation of the host gene jigr1 by miR-92 is essential for neuroblast et al. regulates cellular senescence by activating the TGF-beta self-renewal in Drosophila. PLoS Genet 2015;11:e1005264. pathway. Cell Rep 2017;18:2480–93. 50. Huber V, Vallacchi V, Fleming V, Hu X, Cova A, Dugo M, et al. Tumor- 60. Salmon H, Idoyaga J, Rahman A, Leboeuf M, Remark R, Jordan S, et al. derived microRNAs induce myeloid suppressor cells and predict immu- Expansion and activation of CD103(þ) dendritic cell progenitors at the notherapy resistance in melanoma. J Clin Invest 2018;128:5505–16. tumor site enhances tumor responses to therapeutic PD-L1 and BRAF 51. Guo X, Qiu W, Wang J, Liu Q, Qian M, Wang S, et al. Glioma exosomes inhibition. Immunity 2016;44:924–38. mediate the expansion and function of myeloid-derived suppressor cells 61. Almand B, Clark JI, Nikitina E, van Beynen J, English NR, Knight SC, et al. through microRNA-29a/Hbp1 and microRNA-92a/Prkar1a pathways. Int J Increased production of immature myeloid cells in cancer patients: a Cancer 2018;144:3111–26. mechanism of immunosuppression in cancer. J Immunol 2001;166: 52. Munger JS, Sheppard D. Cross talk among TGF-beta signaling pathways, 678–89. integrins, and the extracellular matrix. Cold Spring Harb Perspect Biol 62. Tritschler I, Gramatzki D, Capper D, Mittelbronn M, Meyermann R, 2011;3:a005017. Saharinen J, et al. Modulation of TGF-beta activity by latent TGF-beta- 53. Nagae M, Re S, Mihara E, Nogi T, Sugita Y, Takagi J. Crystal structure of binding protein 1 in human malignant glioma cells. Int J Cancer 2009;125: alpha5beta1 integrin ectodomain: atomic details of the fibronectin recep- 530–40. tor. J Cell Biol 2012;197:131–40. 63. Kwiatkowski SC, Guerrero PA, Hirota S, Chen Z, Morales JE, Aghi M, et al. 54. Fontana L, Chen Y, Prijatelj P, Sakai T, Fassler R, Sakai LY, et al. Fibronectin Neuropilin-1 modulates TGFbeta signaling to drive growth is required for integrin alphavbeta6-mediated activation of latent TGF-beta and recurrence after anti-angiogenic therapy. PLoS One 2017;12: complexes containing LTBP-1. Faseb J 2005;19:1798–808. e0185065. 55. Wu C, Bauer JS, Juliano RL, McDonald JA. The alpha 5 beta 1 integrin 64. Yadav M, Louvet C, Davini D, Gardner JM, Martinez-Llordella M, Bailey- fibronectin receptor, but not the alpha 5 cytoplasmic domain, functions in Bucktrout S, et al. Neuropilin-1 distinguishes natural and inducible reg- an early and essential step in fibronectin matrix assembly. J Biol Chem ulatory T cells among regulatory T cell subsets in vivo. J Exp Med 2012;209: 1993;268:21883–8. 1713–22. 56. Tian H, Mythreye K, Golzio C, Katsanis N, Blobe GC. mediates 65. Meeth K, Wang JX, Micevic G, Damsky W, Bosenberg MW. The YUMM fibronectin/alpha5beta1 integrin and TGF-beta pathway crosstalk in endo- lines: a series of congenic mouse melanoma cell lines with defined genetic thelial cells. Embo J 2012;31:3885–900. alterations. Pigment Cell Melanoma Res 2016;29:590–7.

www.aacrjournals.org Cancer Res; 79(14) July 15, 2019 3635

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 2019 American Association for Cancer Research. Published OnlineFirst April 23, 2019; DOI: 10.1158/0008-5472.CAN-18-2659

MicroRNA-92 Expression in CD133+ Melanoma Stem Cells Regulates Immunosuppression in the Tumor Microenvironment via Integrin-Dependent Activation of TGFβ

Chris Shidal, Narendra P. Singh, Prakash Nagarkatti, et al.

Cancer Res 2019;79:3622-3635. Published OnlineFirst April 23, 2019.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-18-2659

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2019/04/23/0008-5472.CAN-18-2659.DC1

Cited articles This article cites 64 articles, 13 of which you can access for free at: http://cancerres.aacrjournals.org/content/79/14/3622.full#ref-list-1

Citing articles This article has been cited by 1 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/79/14/3622.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 Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/79/14/3622. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 2019 American Association for Cancer Research.