The Tumor Microenvironment Shapes Lineage, Transcriptional, and Functional Diversity of Inﬁltrating Myeloid Cells

Published OnlineFirst March 31, 2014; DOI: 10.1158/2326-6066.CIR-13-0209

Cancer Research Article Immunology Research

Kutlu G. Elpek1, Viviana Cremasco1, Hua Shen4,5, Christopher J. Harvey1, Kai W. Wucherpfennig1, Daniel R. Goldstein4,5, Paul A. Monach2, and Shannon J. Turley1,3

Abstract Myeloid cells play important regulatory roles within the tumor environment by directly promoting tumor progression and modulating the function of tumor-infiltrating lymphocytes, and as such, they represent a potential therapeutic target for the treatment of cancer. Although distinct subsets of tumor-associated myeloid cells have been identified, a broader analysis of the complete myeloid cell landscape within individual tumors and also across different tumor types has been lacking. By establishing the developmental and transcriptomic signatures of infiltrating myeloid cells from multiple primary tumors, we found that tumor-associated macrophages (TAM) and tumor-associated neutrophils (TAN), while present within all tumors analyzed, exhibited strikingly different frequencies, gene expression profiles, and functions across cancer types. We also evaluated the impact of anatomic location and circulating factors on the myeloid cell composition of tumors. The makeup of the myeloid compartment was determined by the tumor microenvironment rather than the anatomic location of tumor development or tumor-derived circulating factors. Protumorigenic and hypoxia-associated genes were enriched in TAMs and TANs compared with splenic myeloid-derived suppressor cells. Although all TANs had an altered expression pattern of secretory effector molecules, in each tumor type they exhibited a unique cytokine, chemokine, and associated receptor expression profile. One such molecule, haptoglobin, was uniquely expressed by 4T1 TANs and identified as a possible diagnostic biomarker for tumors characterized by the accumulation of myeloid cells. Thus, we have identified considerable cancer-specific diversity in the lineage, gene expression, and function of tumor-infiltrating myeloid cells. Cancer Immunol Res; 2(7); 655–67. 2014 AACR.

Introduction how they are recruited to and expand within growing tumors The tumor microenvironment contains a multiplicity of and metastases are of utmost importance to developing novel stromal cells of hematopoietic and nonhematopoietic develop- therapies and improving existing ones against cancer. mental origin, such as T cells, B cells, natural killer (NK) cells, Tumor growth is associated with the accumulation of a myeloid cells, fibroblasts, pericytes, adipocytes, and endothelial variety of myeloid cell types (2). Common myeloid cell pro- cells, which collectively shape the disease course (1–4). Although genitors in the bone marrow can give rise to myeloid cells with fi fi immunosuppressive potential, often referred to as myeloid- speci c roles have been identi ed for discrete stromal subsets, þ derived suppressor cells (MDSC). Monocyte-like CD11b factors controlling their recruitment, expansion, and function in low þ hi different tumors remain enigmatic. Therefore, a more complete Gr1 and granulocyte/neutrophil-like CD11b Gr1 subsets characterization of these subsets and a better understanding of of MDSCs have been reported to accumulate in the spleen, liver, blood, and bone marrow during tumor progression. Within tumors, myeloid cells with similar phenotypes are Authors' Affiliations: 1Department of Cancer Immunology and AIDS, referred to as tumor-associated macrophages (TAM) or neu- 2 Dana-Farber Cancer Institute; Boston University School of Medicine; trophils (TAN), possibly reflecting a more differentiated iden- 3Department of Microbiology and Immunobiology, Harvard Medical School, Boston, Massachusetts; Departments of 4Internal Medicine and tity. In vitro studies have shown that differentiation of bone 5Immunobiology, Yale School of Medicine, New Haven, Connecticut marrow progenitors into MDSCs requires a combination of Note: Supplementary data for this article are available at Cancer Immu- cytokines, particularly interleukin (IL)-6 and granulocyte col- nology Research Online (http://cancerimmunolres.aacrjournals.org/). ony-stimulating factor (G-CSF) or granulocyte macrophage Current address for K.G. Elpek: Jounce Therapeutics, Inc., Cambridge, MA colony-stimulating factor (GM-CSF), and the transcriptional 02138. regulator CCAAT/enhancer-binding protein (C/EBP; ref. 5). Corresponding Author: Shannon J. Turley, Dana-Farber Cancer Institute, Although splenic MDSCs are considered a reservoir for tumor- 44 Binney Street, D1440a, Boston, MA 02115. Phone: 617-632-4990; Fax: infiltrating myeloid cells (6), the exact relationship between 617-582-7999; E-mail: [email protected] these cells remains elusive. Accumulating evidence indicates doi: 10.1158/2326-6066.CIR-13-0209 that MDSCs, whether in the spleen or in the tumor, have direct 2014 American Association for Cancer Research. suppressive effects on cytotoxic leukocytes. In addition to

www.aacrjournals.org 655

Elpek et al.

þ MDSCs, conventional and plasmacytoid dendritic cells (DC) PtenL/ ) were kindly provided by Dr. DePinho [while at the may exert immunoregulatory effects in tumors (2) using a Dana-Farber Cancer Institute (Boston, MA); currently at the variety of mediators such as indoleamine 2,3-dioxygenase MD Anderson Cancer Center (Houston, TX); ref. 11]. (IDO), inducible nitric oxide synthase (iNOS), and arginase I to suppress T-cell proliferation, cytotoxicity, and effector Cell lines and tumor models cytokine production. B16, EL4, 4T1, A20, and CT26 cells were obtained from Given that myeloid cells with protumorigenic and immu- American Type Culture Collection. Pan02 cells were obtained nomodulatory functions have been observed in multiple ani- from the NCI Developmental Therapeutics Program reposito- mal tumor models and in patients with cancer, they represent ry. The 4T07 cells were obtained from Dr. J. Lieberman (Har- important targets for immunotherapy. Efforts are under way to vard University, Cambridge, MA); they were expanded, ali- identify myeloid-focused strategies. Approved chemotherapy quoted, and banked in liquid nitrogen. No additional authen- agents, such as gemcitabine (7), 5-fluorouracil (8), and suni- tication was performed on these cells. The Her2 cell line was tinib (9), can eliminate or prevent the accumulation of MDSCs, derived from a spontaneous breast tumor in a Balb/c Her-2/ especially in lymphoid organs, and retard tumor progression. neu–transgenic mouse that was found to lose Her-2/neu Likewise, agents that block myeloid recruitment to tumors, expression in vitro. Transgenic rat Her-2/neu sequence was such as CSF1R inhibitors (10), hold clinical promise. However, inserted into a pMFG retroviral vector to obtain a stably to improve current strategies and identify new universal overexpressing cell line. Cell lines were maintained in Dulbec- targets for therapeutic intervention, it is essential to under- co's Modified Eagle Medium (B16 and EL4) or RPMI (4T1, Her2, stand how each myeloid cell subset from one tumor relates to A20, Pan02, CT26, and 4T07) supplemented with 10% FBS and the same population in other tumor types. 1 penicillin/streptomycin. In this study, we analyzed myeloid subsets in multiple For subcutaneous tumors, 1 to 2.5 105 cells were injected murine tumors to study how phenotype, frequency, and tran- into the flanks of na€ve mice (4T1, Her2, CT26, and 4T07 into scriptional profiles relate within different tumors, using triple- Balb/c; B16 and Pan02 into C57BL/6; A20 into Balb/cxCD45.1; þ negative 4T1 breast cancer, Her2 breast cancer, and B16 and EL4 into C57BL/6xCD45.1), and tumor growth was moni- melanoma as models. Strikingly, each tumor type contained tored using calipers until tumor size reached 0.5 to 1 cm diameter. a distinct myeloid cell landscape, with TAMs, TANs, and DCs For transgenic models, tumors were monitored and analyzed at represented in all tumors but at markedly different ratios, different time points based on tumor growth (RIP-Tag2 at 11–12 while systemic MDSC accumulation was exquisitely tumor- weeks, and PDA and Her2 transgenic at 11–24 weeks). specific. Our data suggest that tumor type, rather than anatomic location, dictates myeloid composition in the tumor Cell preparation and flow cytometry lesion. Furthermore, although each subset exhibits similar Spleens and tumors were processed by gentle mechanical transcriptional signatures associated with its identity in dif- disruption with forceps followed by enzymatic digestion using ferent tumors, our study demonstrates that functional differ- 0.2 mg/mL collagenase P (Roche), 0.8 mg/mL dispase (Invitro- ences exist across myeloid subsets from different tumors. In gen), and 0.1 mg/ML DNase I (Sigma). Tissues were incubated in particular, our data suggest that haptoglobin may represent a digestion medium at 37C for 30 minutes; released cells were biomarker for tumors characterized by systemic accumulation collected and filtered, and the remaining tissue was further pro- of myeloid cells. Thus, our study provides important insights cessed. Blood was collected into tubes containing 2 mmol/L into the identity and functional characteristics of tumor-asso- EDTA, and bone marrow cells were isolated by flushing media ciated myeloid subsets. Perhaps more importantly, our data through tibias and femurs. Red blood cells were lysed using support that in-depth transcriptomic analysis of tumor-infil- ammonium–chloride–potassium solution. Cells were resus- trating myeloid cells may reveal novel therapeutically attrac- pended in fluorescence-activated cell sorting (FACS) buffer tive targets for cancer. containing 1% FBS and 2 mmol/L EDTA. Cells were stained in FACS buffer containing FcR-blocking Materials and Methods antibody (2.4G2) with combinations of fluorochrome-conju- Ethics statement gated or biotinylated antibodies against CD11b (M1/70), All animal work has been carried out in accordance with U.S. CD11c (N418), CD45 (30.F11), Gr1 (RB6-8C5), MHC class II NIH guidelines. This study is reviewed and approved by the (M5/114.15.2), CD80 (16-10A1), CD86 (GL-1), CD40 (HM40-3), Dana-Farber Cancer Institute Animal Care and Use Committee CD26 (H194-112), CD103 (2E7), cKit (2B8), Flt3 (A2F10), or (protocol IDs: 04-025 and 07-038). isotype controls purchased from BioLegend and BD Bios- ciences. Cells were analyzed using BD FACSAria II or BD Mice FACSCalibur (BD Biosciences), and FlowJo Software (TreeStar, þ Six-week-old, sex-matched C57BL/6, Balb/c, or CD45.1 (B6. Inc.). For cell sorting from spleens, myeloid cells were enriched þ þ þ SJL-PtprcaPepcb/BoyJ and CBy.SJL(B6)-Ptprca/J) mice were by depletion of CD3e , CD19 , and CD49b cells using bio- purchased from The Jackson Laboratory. F1 (Balb/cxC57BL/ tinylated antibodies and anti-biotin beads (MACS; Miltenyi þ 6) mice were bred in-house. RIP1-Tag2 mice were obtained Biotec). In tumor samples, CD45 cells were enriched by from Mouse Model of Human Cancer Consortium [National positive selection using a biotinylated antibody and anti-biotin Cancer Institute (NCI)]. Mice with spontaneous invasive pan- beads (EasySep; STEMCELL Technologies). After enrichment, creatic ductal adenocarcinoma (PDA; Pdx1-Cre LSL-KrasG12D cells were stained with fluorochrome-conjugated antibodies

656 Cancer Immunol Res; 2(7) July 2014 Cancer Immunology Research

Heterogeneity of the Tumor Myeloid Cell Landscape

and propidium iodide (Sigma) to exclude dead cells. Cells were Elpek and S.J. Turley). Principal component analyses were con- sorted using a BD FACSAria II equipped with a 100-mm nozzle ducted using the PopulationDistances module created by Imm- running at 20 psi. After an initial sort to verify purity, a second Gen (probes with mean expression value >120 selected, dataset sort was performed to collect myeloid cell populations of 95% log2 transformed, probes with top 15% variability selected). to 100% purity directly into TRIzol reagent (Invitrogen). Sam- ple size for each population was as follows: CD11b /int Cytokine arrays þ þ þ þ þ Gr1 CD11c MHCII (N ¼ 4), CD11b Gr1lowCD11c MHCII Cell culture supernatants from 4T1, Her2, and B16 cells were þ þ (N ¼ 3), CD11b Gr1lowCD11c MHCII (N ¼ 3), and CD11b collected in two separate batches when cells were approxi- þ Gr1hiCD11c MHCII (N ¼ 2) from B16 tumors; CD11b mately 80% confluent and analyzed with Mouse Cytokine þ þ þ Gr1lowCD11c MHCII (N ¼ 3), CD11b Gr1lowCD11c MHCII Antibody Array 3 (RayBiotech, Inc.) according to the manu- þ (N ¼ 3), and CD11b Gr1hiCD11c MHCII (N ¼ 2) from facturer's protocol. Spot densities were measured using ImageJ þ þ þ 4T1 tumors; CD11b Gr1lowCD11c MHCII (N ¼ 3) and (NIH), and relative amounts were calculated on the basis of þ CD11b Gr1hiCD11c MHCII (N ¼ 3) from Her2 tumors; positive and negative controls. Only cytokines and chemokines þ þ CD11b Gr1lowCD11c MHCII (N ¼ 2) and CD11b with a positive signal are shown. Gr1hiCD11c MHCII (N ¼ 3) from spleens of 4T1 tumor- bearing mice. ELISA for haptoglobin Serum was obtained from na€ve or tumor-bearing mice. RNA isolation and microarray analysis Human sera were obtained from patients with metastatic Total RNA was prepared from TRIzol using chloroform ex- breast carcinoma upon enrollment into Institutional Review traction according to the manufacturer's protocol, and 100 ng Board (IRB)–approved vaccine trials at the Dana-Farber Can- of RNA from each sample was used for amplification, labeling, cer Institute. Informed consent was obtained for each partic- and hybridization by Expression Analysis, Inc. Mouse Gene ST ipant about usage of collected samples for research purposes. 1.0 chips (Affymetrix) were used for microarray analysis. All data Samples were collected by routine procedures before begin- are Minimum Information About a Microarray Experiment ning study treatment. Haptoglobin levels were quantified using (MIAME)-compliant, and the raw data generated as part of Mouse or Human Haptoglobin ELISA (ICL, Inc.) according to the Immunological Genome Project (ImmGen) have been the manufacturer's protocol. Concentrations were calculated deposited in a MIAME-compliant database [National Center using a four-parameter logistics curve. for Biotechnology Information (NCBI) Gene Expression Omni- bus (GEO) data repository; record no: GSE15907 and GSE37448]. Statistical analysis for nonmicroarray data Various modules included in GenePattern platform (Broad Data are expressed as mean SD and were analyzed using Institute, Cambridge, MA) were used for data analysis. Addi- one-tailed, unpaired Student t test. ELISAs were analyzed using < tional myeloid cell populations from healthy tissues analyzed ANOVA and Tukey test. P 0.05 was considered statistically fi within the same microchip batch of ImmGen were used as a signi cant. comparison [F4/80 macrophages from peritoneal cavity (PC; þ þ n ¼ 2), Ly6C DCs from bone marrow (n ¼ 2), CD24 Sirpa Results and Discussion þ þ DCs from bone marrow (n ¼ 3), CD24 Sirpa DCs from bone Identification and enumeration of tumor-associated þ marrow (n ¼ 3), CD4 DC from spleen (n ¼ 1), CD11c CD103 myeloid cells reveal tumor-specific diversity DC from small intestine (n ¼ 1), neutrophil from bone marrow Thus far, it remains uncertain how tumor origin, localization, (C57BL/6; n ¼ 2), and neutrophil from bone marrow (Balb.c; n ¼ and history influence the myeloid cell landscape of different 2)]. Raw data were normalized with ExpressionFileCreator mod- solid malignancies. To address these questions, we initially ule using the Robust Microarray Average (RMA) method. The sought to compare the composition of myeloid cell infiltrates MultiPlot module was used for dataset comparisons, including among 4T1 and Her2, two distinct murine tumors sharing the fold-change analysis and statistical filtering. In each analysis, we same mammary tissue origin. Tumors were established subcu- preprocessed the dataset for populations included in specific taneously in the flanks of Balb/c mice and analyzed when they analyses. In fold-change analyses, an arbitrary cutoff value of 2 reached different tumor sizes. We characterized multiple mye- was used, in combination with coefficient of variation (CV) < 0.5 loid cell populations isolated from each tumor type on the basis for replicates. The t test P < 0.05 were considered statistically of myeloid-specific surface markers CD11b, Gr1, and CD11c by significant for each probe. After filtering, we selected all probes cytofluorimetry. In both types of tumors and throughout tumor þ with a mean expression above 100 in at least one of the popula- history, the CD45 hematopoietic compartment was dominat- þ tions analyzed, which indicates actual expression with 95% ed by CD11b cells (>75%; Fig. 1A and Supplementary Fig. S1A). þ confidence level. For hierarchical clustering, Pearson correlation The intratumoral CD11b cells comprised three subpopula- was used with datasets following log2 transformation, row cen- tions based on Gr1 and CD11c expression (Fig. 1A). In 4T1 tering, and row normalization in HierarchicalClustering module. tumors, the hematopoietic compartment contained approxi- þ þ Heatmaps were constructed using HeatmapViewer module. mately 20% CD11b Gr1hiCD11c cells and >45% CD11b Pathway enrichments were analyzed by Ingenuity Pathway Anal- Gr1low;with>60% of the latter expressing CD11c. Compared ysis. DC- and macrophage-associated genes were determined with 4T1, Her2 tumors were infiltrated by a relatively small þ previously (12, 13), and a similar analysis was performed to population of CD11b Gr1hiCD11c cells (3%), while the majority þ þ identify neutrophil-associated genes (unpublished data; K.G. (>95%) of myeloid cells were CD11b Gr1lowCD11c .Wealso

www.aacrjournals.org Cancer Immunol Res; 2(7) July 2014 657

Elpek et al.

A E + + low + hi – – DC SI CD11b Gr1 CD11b Gr1 B16 Tmr CD11b Gr1 + CD45 + DC BM + B16 Tmr 4T1 Tmr Her2 Tmr B16 Tmr 4T1 Tmr 4T1 Spl – + – + + – 4T1 Tmr 4T1 Spl Her2 Tmr B16 Tmr DC BM DC BM + – 12 hi hi hi hi 5 60 10 82 15 Gr1 Gr1 Gr1 Gr1 62 66 18 96 4 CD103 CD11c 34 – – + + + + Mac PC Sirpa Sirpa

9 PDCA1 hi – + CD11c CD11c CD11c CD11c CD11c CD11c DC Spl 104 + +

40 low low low low low low 10 21 6 CD24 CD11b Neutro B6 BM Neutro Balbc BM CD11b CD4 Gr1 CD11b Gr1 Gr1 Ly6C CD24 Gr1 Gr1 Gr1 4T1 103 F4/80 CD11b CD11b CD11b

20 5 Btla 3 Flt3 0 Kmo P2ry10 15 0 0 0 0 1 2 3 4 Hmgn3 100 101 102 103 104 100 101 102 103 104 10 10 10 10 10 0102 103 104 105 Bri3bp Napsa 40 Ccr7 5 Klri1 105 95 8 6 82 60 5 92 94 H2−Eb2 4 30 Gpr68 104 Traf1 3 40 Gpr82 3 20 Kit Her2 103 2 Gpr68 Zbtb46 20 10 1 Slamf7 0 Rab30 Dpp4 0 0 CD11b 13 0 0 1 2 3 4 0 1 2 3 4 Jak2 100 101 102 103 104 10 10 10 10 10 10 10 10 10 10 0102 103 104 105 Bcl11a Haao Gr1 CD11c Ap1s3 Spint2 Tbc1d8 Gpr114 Pvrl1 Gpr132 B Ass1 + hi – – Runx3 CD11b+Gr1low CD11b+Gr1low CD11b Gr1 CD11b Gr1 Anpep H2−Q6 + – CD11c– CD11c+ Ciita CD11c CD11c H2−DMb2

100 100 H2−Aa 100 100 H2−Ab1 86 H2−Eb 80 30 80 18 80 80 83 Cd164 A930039A15Rik 60 60 60 60 Tmem195 4T1 1810011H11Rik 40 40 40 40 Pecr Comt1

20 20 20 20 Pon3 Mr1 Fcgr1 0 0 0 0 2 3 4 5 2 3 4 5 2 3 4 5 0102 103 104 105 Camk1 010 10 10 10 010 10 10 10 010 10 10 10 Myo7a Mertk 100 100 100 100 Abca1 Cd14 80 92 80 96 80 4 80 96 Pla2g15 Plod1

60 60 Ctsl 60 60 Tpp1 Her2 Sepp1 40 40 40 40 Blvrb Glul 20 20 20 20 Tspan14 Tlr7 0 0 0 0 Lamp2 2 3 4 5 2 3 4 5 2 3 4 5 010 10 10 10 010 10 10 10 010 10 10 10 0102 103 104 105 Tcn2 Pld3 MHCII Slc48a1 Pla2g4a Nln Pcyox1 C Tbxas1 Naïve 4T1 Her2 Akr1b10 4 10 104 4 Amica1 2.9 45 10 Cnn2 3.7 1.7 Pstpip1 3 1.2 10 1.6 103 3 Cbfa2t3 10 Adam19 Gm5068 2 10 2 10 102 Fgl2 Spleen Amica1 Stfa2l1 + 1 10 1 CD45 10 101 G0s2 Fam122a Gm5416 0 CD11b 10 100 0 0 1 2 3 4 10 Spatc1 10 10 10 10 10 0 1 2 3 4 10 10 10 10 10 100 101 102 103 104 Prok2 Arsg CD11c Fert2 4 4 104 10 10 Fgd4 Sqrdl Ptplad2 3 3 103 10 10 Tom1 B430306N03Rik Ctsd + 2 2 CD11b 102 10 10 Fcgr3 15 85 20 80 25 75 Clec5a – Tlr4 1 1 CD11c 101 10 10 Slc16a3 Slfn1 Csf3r 0 0 0 CD11b 10 10 10 Slfn4 0 1 2 3 4 0 1 2 3 4 100 101 102 103 104 10 10 10 10 10 10 10 10 10 10 Hdc Retnlg Gr1 Sgms2 Cd300lf Grina Arg2 Pdlim7 D Dok3 Gr1lowCD11c+ Her2 Tmr Slc27a4 low + Tirap Gr1 CD11c 4T1 Tmr Klhl2 Gr1lowCD11c+ B16 Tmr Mboat7 low – Pxn Gr1 CD11c B16 Tmr Pfkfb4 Gr1lowCD11c– 4T1 Tmr Ppp1r3d low – Phospho1 Gr1 CD11c 4T1 Spl Sfxn5 F4/80hi Mac PC A530064D06Rik + + 2010002M12Rik Ly6C PDCA1 DC BM Signature gene Dhrs9 CD11b–CD11c+ B16 Tmr Gm9949 – + 4732465J04Rik CD24 Sirpa DC BM Dendritic cell Fam123a CD24+Sirpa– DC BM Krt86 + Macrophage Rlf CD4 DC Spl Chi3l1 CD11b–CD103+ DC SI Neutrophil Mgam + hi Asprv1 CD11b Gr1 B16 Tmr Il1f9 CD11b+Gr1hi Her2 Tmr Mrgpra2a + hi Slc2a3 CD11b Gr1 4T1 Spl 3.27 13.50 Abtb1 CD11b+Gr1hi 4T1 Tmr Expression (log ) Ankrd22 2 Ceacam10 Neutro Balbc BM Msl1 Neutro B6 BM Slc22a15 Ccpg1 Slc35a5 A530023O14Rik Foxd4 E430024C06Rik

658 Cancer Immunol Res; 2(7) July 2014 Cancer Immunology Research

Heterogeneity of the Tumor Myeloid Cell Landscape

A B ** 100 30 ** *** *** ** 80 *** in tumor + 20 60

in tumor *** 4T1 Figure 2. Abundance of myeloid cell + *** 40 Her2 subsets across different tumors. A, 10

*** * within CD45

* + CD45

%Subsets within 20 Whisker box plots summarizing the *** B16 relative frequency of myeloid

0 %CD3 0 subsets among hematopoietic – CD11c+ Other/ 4T1 Her2 B16 cells within 4T1, Her2, and B16 CD11c TANs DCs þ TAMs TAMs nonmyeloid tumors. B, frequency of CD3 T cells within hematopoietic cells in C 100 4T1, Her2, and B16 tumors. C, frequency of myeloid cells in 4T07, 80 Pan02, EL4, CT26, PDA, RIP-Tag2, 60 and A20 tumors (n ¼ 3–7). in tumor

+ , P < 0.05; , P < 001; , 40 P < 0.001.

CD45 20 %Subsets within

0 CD11c– CD11c+ TANs DCs Other/ TAMs TAMs nonmyeloid

4T07 Pan02 EL4 CT26 PDA RIP-Tag2 A20

þ found a Gr1 CD11c population among CD11b cells, which lymphoid and nonlymphoid tissues of healthy animals. Den- comprised <1% of all hematopoietic cells in both 4T1 and Her2 dogram analysis based on gene clustering depicted similarities þ þ tumors. Further analysis showed that CD11b Gr1lowCD11c and between neutrophils and tumor-derived Gr1hi cells, macro- þ þ CD11b Gr1 CD11c cells in the tumor were largely MHCII , pahges and Gr1low cells (regardless of CD11c expression), and þ þ þ whereas CD11b Gr1lowCD11c and CD11b Gr1hiCD11c cells conventional DCs and CD11b Gr1 CD11c cells (Fig. 1D). were MHCII (Fig. 1B). A marked accumulation of splenic Hierarchical clustering of 139 probes corresponding to signa- þ þ CD11b Gr1low and CD11b Gr1hi cells (Fig. 1C) was observed ture genes of DCs, macrophages (12, 13), and neutrophils in 4T1-bearing mice, whereas the splenic numbers in mice (unpublished data; K.G. Elpek and S.J. Turley) demonstrated with Her2 tumors were similar to na€ve mice. These data that tumor-inﬁltrating myeloid cell populations are indeed þ suggest that each tumor type may contain a distinct myeloid CD11c and CD11c TAMs, TANs, and DCs (Fig. 1E and cell inﬁltrate. Supplementary Table S1).

Lineage determination of tumor-infiltrating myeloid cell Striking heterogeneity in myeloid cell infiltrates across subsets based on transcriptional profiling different tumor types We next sought to determine the lineage of each tumor- Among the three tumor types analyzed, Her2 tumors were þ infiltrating myeloid cell population using transcriptomic anal- uniquely characterized by the presence of a large CD11c ysis. To this end, myeloid subsets from 4T1 and Her2 breast TAM population, whereas 4T1 tumors contained a relatively carcinomas and B16 melanoma were sorted to high purity, large TAN population (Fig. 2A). In B16 tumors, myeloid cells according to the ImmGen standard operating procedure. comprised only 40% of tumor-infiltrating leukocytes com- Transcriptional profiling data were generated on Affymetrix pared with >75% in 4T1 and Her2 tumors (Fig. 2A). The ST1.0 microarrays adhering to profiling and quality control aforementioned differences in myeloid abundance also pipelines of ImmGen (14). We used hierarchical clustering of reflected variation in the abundance of other leukocytes approximately 13.5 K probes, after excluding those with infiltrating tumors, including T, B, NK, NKT cells, and pDCs. þ CV < 0.5 and mean expression values <120 to compare popula- For example, B16 tumors were enriched in CD3 T cells tions of tumor-infiltrating myeloid cells with representative compared with 4T1 and Her2 tumors, consistent with a more myeloid cell subsets (DCs, macrophages, and neutrophils) from immunogenic environment in B16 melanoma (Fig. 2B). To

Figure 1. Multiple myeloid cell subsets are present within tumors of different origin. A, analysis of myeloid cell subsets within subcutaneous 4T1 and Her2 tumors by ﬂow cytometry. Hematopoietic cells were analyzed for the expression of CD11b and Gr1, and gated populations in the dot plots were þ þ further analyzed for CD11c expression. B, MHCII expression by indicated subsets. C, accumulation of CD11b Gr1 cells in the spleens of 4T1 tumor-bearing mice (top) and Gr1 expression (bottom). Numbers indicate percentage of cells for each gate or region. D, dendrogram analysis based on hierarchical clustering of tumor-associated and steady-state myeloid cells (BM, bone marrow; SI, small intestine; Spl, spleen; Tmr, tumor). Sample sizes are indicated in Materials and Methods. E, hierarchical clustering of the same populations based on a list of 139 genes associated with DCs, macrophages, or neutrophils.

www.aacrjournals.org Cancer Immunol Res; 2(7) July 2014 659

Elpek et al.

further ascertain the degree of myeloid heterogeneity, we tumor in mammary glands revealed that both lesions con- þ extended our analysis to five additional transplanted tained high frequencies of CD11c TAMs (50%–72%) and tumors, including 4T07 breast cancer, CT26 colon carcino- relatively low frequencies of TANs (<5%) among hematopoietic ma, EL4 T lymphoma, A20 B lymphoma, and Pan02 PDA, and cells (Fig. 3A and Supplementary Fig. S2B). A strong similarity three autochthonous mouse models, including RIP-Tag2 was observed between the myeloid cell composition of sub- (insulinoma), Her-2/neu (mammary carcinoma), and cutaneous 4T1 tumors and spontaneous metastatic lesions in Pdx1-Cre LSL-KrasG12D PtenL/L (PDA) mice. The same the peritoneal cavity, occurring within 2 to 3 weeks after tumor four major populations were present within each tumor challenge. Similar to primary 4T1 tumors, these metastases are þ model but at significantly different frequencies across the enriched with TAMs (50%, of which >60% are CD11c ) and tumors analyzed (Fig. 2A–C). TANs (15%–21%) among hematopoietic cells (Fig. 3A and In all tumors, DCs were present at very low frequencies, with Supplementary Fig. S2C). Furthermore, these results were the highest levels in B16, EL4, and RIP-Tag2 tumors (4%– corroborated in a B16 metastatic model. When B16 cells are 10%; Fig. 2). Given that DCs account for only a small percentage injected intravenously, tumor nodules grow in lungs and of hematopoietic cells within the tumors, and that the major kidneys within 2 to 3 weeks and contain a similar myeloid subset expressing CD11c in tumors is TAMs, our study also landscape to that of subcutaneous tumors: approximately 30% suggests that CD11c alone is an unreliable marker for DCs in TAMs (45% expressing CD11c) and 4% to 9% TANs among neoplastic lesions. Further analysis comparing DCs in B16 hematopoietic cells (Fig. 3A and Supplementary Fig. S2D). tumors with TAMs indicated that while TAMs expressed a Altogether, these data provide evidence that tumor type rather variety of endocytic and regulatory receptors, including Pilrb, than its anatomic location shapes the myeloid infiltrate. Sirpa, Lilrb4, and Clec5a (Supplementary Fig. S1B), tumor DCs Next, we tested the possible dominant effect of the micro- expressed signature genes such as Dpp4, Flt3, and Kit, as well as environment in dictating myeloid cell composition by analyz- CD8/CD103 DC-associated genes (13), including Itgae, Batf3, ing myeloid cells in mice with two tumors growing simulta- þ Tlr3, Ifi205, and Xcr1, identifying them as CD11b CD103 neously. For this purpose, we compared Her2 and 4T1 or Her2 tissue-resident DCs. Flow cytometric analysis confirmed that and B16 tumors, as each of these tumors has distinct myeloid DCs were the myeloid cells expressing high levels of CD26 infiltrates (Figs. 1 and 2). Balb/c mice were inoculated with (Dpp4), Flt3, cKit, and CD103 (Itgae) protein in these tumors Her2 tumors under one flank and simultaneously with 4T1 (Supplementary Fig. S1C; and data not shown). On the basis of tumors under the contralateral flank (Fig. 3B and Supplemen- these distinctions, DCs may be considered as a reference tary Fig. S2E). Similarly, we established Her2 tumors in the population in future studies to identify unique targets on presence of B16 tumors on the opposite flank in F1 (Balb/c x TAMs and TANs to modulate their tumor-promoting func- C57BL/6) mice (Fig. 3C and Supplementary Fig. S2F). In both tions. For example, DCs lack the surface receptor Clec5a (MDL- models, we did not observe a difference in growth rate of either 1), whereas expression in TAMs, TANs, and neutrophils is tumor when compared with tumors grown alone (data not relatively high (>18-fold higher than in DCs), suggesting that shown). Likewise, the myeloid content of each tumor, either it may represent an attractive target for antibody-mediated alone or in the presence of a second tumor in the same host was þ þ depletion of CD11b Gr1 cells in tumors (15). similar (Fig. 3B and C, left and Supplementary Fig. S2E and Differences in Gr1low and Gr1hi cells across several diverse S2F), even though there was pronounced systemic accumula- tumors were not restricted to the tumor lesion itself. For tion of MDSCs associated with 4T1 tumor growth (Fig. 3B and example, a substantial increase in MDSCs was observed in C, right). Overall, these results suggest that the signals attract- the spleen and blood of 4T1 tumor-bearing mice throughout ing specific myeloid cells into tumors are driven by cancer cells tumor progression but not in mice bearing other tumor types in the primary lesion, regardless of anatomic location or (Supplementary Fig. S1D and S1E). presence of distant signals from other microenvironments.

Differences in myeloid cell composition arise from Tumor-infiltrating myeloid cells exhibit a hypoxia- specificity of the tumor microenvironment associated signature compared with their splenic The myeloid composition of different tumors is likely shaped counterparts by a combination of tumor-specified growth factors. Indeed, As shown in Fig. 1, MDSCs accumulate in the spleens of 4T1 4T1, Her2, and B16 cell lines produce unique combinations of tumor-bearing mice at high frequencies. As splenic MDSCs cytokines and chemokines, suggesting that each tumor may may serve as reservoirs for tumor-infiltrating cells (6), it is create a distinctive microenvironment (Supplementary Fig. important to understand how they relate to their counterparts S2A). These differences may also explain why MDSCs prefer- within tumors. Thus, we compared the gene expression profiles þ þ entially accumulate in certain tumors. For example, high levels of CD11b Gr1 cells from the spleens and tumors of 4T1 of G-CSF and LIX (CXCL5) expression by 4T1 cells may tumor-bearing mice. As shown in Fig. 4A, 263 probes were þ contribute to systemic accumulation of MDSCs in vivo (16, upregulated in TAMs (both CD11c and CD11c subsets) 17). To assess whether specific tumor microenvironments within 4T1 tumors compared with Gr1low MDSCs from spleens, directly influence the myeloid cell content, we analyzed mye- whereas 152 probes were enriched in splenic cells. In addition, loid cells within the same tumor type growing at different 201 probes were enriched in TANs compared with splenic Gr1hi anatomic locations. Comparison of myeloid cell composition MDSCs, while 647 probes were enriched in Gr1hi MDSCs (Fig. in subcutaneous Her2 tumors and the parental spontaneous 4B). Next, we compared expression of probes upregulated in

660 Cancer Immunol Res; 2(7) July 2014 Cancer Immunology Research

Heterogeneity of the Tumor Myeloid Cell Landscape

A Her2 4T1 B16 Primary Mammary gland Primary Peritoneal met Primary Lung met Kidney met (s.c.) (transgenic) (s.c.) (spont.) (s.c.) (i.v.) (i.v.) 4 4 4 10 10 4 4 4 4 10 10 10 10 10 59 3 3 3 10 10 3 3 3 3 10 10 10 10 10 7 3 2 6 12 10 2 2 2 2 28 4 3.8 69 10 10 10 44 10 2 2 CD45+ 60 21 10 31 10 33 1 1 1 10 10 1 1 1 1 10 10 10 10 10

0 10 0 0 0 0 10 10 10 10 100 0 0 1 2 3 4 10 0 1 2 3 4 0 1 2 3 4 10 10 10 10 10 10 10 10 10 10 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 Gr1 Gr1 Gr1

60 50 50 25 25 25 56 44 61 39 62 38 120 29 15 40 20 4 96 98 40 20 40 2 75 71 90 30 15 30 10 15

20 60 20 10 10 20 5 TAMs 10 30 10 5 5

0 0 0 0 CD11b 0 0 0 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 100 101 102 103 104 10 0 10 1 10 2 10 3 10 4 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 CD11c CD11c CD11c

B Single tumor Double tumor Spleen Her2 4T1 Her2 4T1

4 4 4 10 104 10 10 *

3 *** 3 3 10 103 10 10 3.7 2 2 N.S. 2 + 10 71 102 10 11.4 10 21 17.3 + CD45 58 77.4 58 40 N.S. 1 1 1 10 101 10 10 30 0 0 0 10 100 10 10 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 10 10 10 10 10 100 101 102 103 104 10 10 10 10 10 10 10 10 10 10 Gr1 20 10 100 80 94 40 6 60 64 93 74 36 80 7 26 %Cells in CD45 0 60 30

40 60 Her2 4T1 Double

40 20 40

TAMs 20 20 10 low hi 20 Gr1 MDSC Gr1 MDSC

CD11b 0 0 0 0 0 1 2 3 4 0 1 2 3 4 10 10 10 10 10 100 101 102 103 104 100 101 102 103 104 10 10 10 10 10 CD11c

C Single tumor Double tumor Spleen Her2 B16 Her2 B16

4 4 10 10 104 104 72 N.S. 3 3 10 10 103 103 2 2.3 1.3 10 + 2 2 32 2 CD45+ 10 10 10 83 102 33 8 N.S.

1 1 10 10 101 101 6

0 0 10 10 100 100 0 1 2 3 4 0 1 2 3 4 10 10 10 10 10 10 10 10 10 10 100 101 102 103 104 100 101 102 103 104 4 Gr1 2

80 150 96 46 49 %Cells in CD45 0 4 60 54 5 95 51 30 Her2 B16 Double 60 100 40 20 40 low hi TAMs 20 50 Gr1 MDSC Gr1 MDSC 10 20

CD11b 0 0 0 0 0 1 2 3 4 0 1 2 3 4 10 10 10 10 10 100 101 102 103 104 100 101 102 103 104 10 10 10 10 10 CD11c

Figure 3. Tumor type dictates the composition of myeloid cell subsets within tumors. A, myeloid cells within subcutaneous and mammary gland of transgenic Her2 tumors (n ¼ 4–7); within primary subcutaneous and spontaneous peritoneal metastatic 4T1 tumors (n ¼ 3); and within subcutaneous B16 tumors or lung and kidney metastases (n ¼ 3–4). B, left, analysis of myeloid subsets within tumors from Balb/c mice with single (Her2 or 4T1; n ¼ 5) or double (Her2 and 4T1 on opposite flanks; n ¼ 8) tumors. Right, frequency of myeloid cells in the spleens of tumor-bearing mice. C, left, analysis of myeloid cell subsets within tumors from F1 (Balb/c C57BL/6) mice with single (Her2 or B16; n ¼ 3) or double (Her2 and B16 on opposite flanks; n ¼ 3) tumors. Right, frequency of myeloid cells in the spleens of tumor-bearing mice. Numbers indicate percentage of cells for each gate or region. N.S., not statistically significant; , P < 0.05; , P < 0.001.

www.aacrjournals.org Cancer Immunol Res; 2(7) July 2014 661

Elpek et al.

A C D Hypoxia-associated genes

263 Arg1 Tgm2 MDSC) DC SI + DC BM

Nos2 Ptps2 + low DC BM DC BM + Vegfa – Havcr2 Spp1 Adam8 Cd274 Mmp12 CD103 TAM B16 TAM 4T1 TAM Her2 TAM 4T1 TAM B16 TAM B16 TAM 4T1 TAM Her2 TAM 4T1 TAM B16 Cxcl16 – + + + – – Mac PC Sirpa Sirpa 10 Slc2a1 – – + + + PDCA1 hi – + MDSC 4T1 DC Spl + MDSC 4T1 MDSC 4T1 +

Abca1 MDSC 4T1 low hi Hif1a Mmp14 low hi CD11c Gr1 TAN 4T1 TAN B16 TAN Her2 CD11c CD11c CD4 CD24 CD11b CD24 Ly6C CD11c Neutro B6 Neutro Balbc Gr1 F4/80 DC B16 Tmr CD11c Gr1 Gr1 CD11c CD11c CD11c TAN 4T1 TAN B16 TAN Her2 Serpinb9 CD11c CD11c Kdr

TAM vs Gr1 TAM Mgl2

– Car9 Fold change 1 Pdk1 Cdh1 Ccnd1 152 Pdgfb 0.2 Igf2 Igfbp2 0.1 110 Tgfa Serpine1 (4T1 CD11c Fold change Flt1 + low Cxcl12 (4T1 CD11c TAM vs Gr1 MDSC) Mmp2 Met Cdkn1a B Pfkfb3 Egln3 0 Nos2 10 Slc2a1

) Arg1 –1 10 10 Vegfa Vegfa Hif1a –2 Slc2a1 Hk2 10 Arg1 Hk1 Egln3 Arg1 Gys1 10–3 Ldha Niacr1 Pgk1 –4 Tnfrsf26 Pkm2 10 Cstb Cd274 Eno1 Aldoa –5 Cxcl10 10 Tgm2 Mmp9 Slc2a3 –6 Kdm5b t test P value (log 10 647 Ccng2 201 Cxcr4 10–7 Plaur Crebbp Il1a Crebbp 0.2 10 1 3.87 13.68 4.74 13.54 hi Fold change (4T1 TAN vs Gr1 MDSC) Expression (log2) Expression (log2)

þ Figure 4. Tumor-infiltrating myeloid cells express immunomodulatory genes. A, comparison of gene expression profiles of CD11c and CD11c TAMs to Gr1low MDSCs from 4T1 tumor-bearing mice based on 2-fold change (fold-change vs. fold-change plot). B, comparison of gene expression profiles of TANs and Gr1hi MDSCs from 4T1 tumor-bearing mice (fold-change vs. t test P value plot). C, heatmap generated by hierarchical clustering using probes identified in A and B for all tumor myeloid subsets. D, heatmap showing the expression of hypoxia-related genes by tumor-associated and steady-state myeloid cells. Selected genes associated with the populations are indicated on the plots.

4T1 tumors with that of myeloid cells originated in other tumor of aggressive tumor cells. Expression analysis of Hif-1a–reg- models. Hierarchical clustering indicated that TAMs, TANs, ulated hypoxia-associated genes (19) revealed that these genes and splenic MDSC populations clustered separately, confirm- are indeed enriched in TAMs and TANs compared with splenic ing differences between tumor and splenic myeloid cells. MDSCs and steady-state populations, indicating reprogram- Notably, similar profiles were obtained from equivalent popu- ming in the tumor environment (Fig. 4D and Supplementary lations originated from different tumor types (Fig. 4C), sup- Table S2). Hypoxia and Hif-1a have been implicated in pro- porting that transcriptomic analysis may illuminate universal longed survival and reduced oxidative burst in neutrophils features of tumor myeloid cells. Indeed, this analysis identified (20, 21), and as such may also contribute to the accumulation of several genes that highlight the protumorigenic roles of tumor- these short-lived cells. infiltrating myeloid cells. For example, genes involved in extracellular matrix remodeling (Mmp12, Mmp13, Mmp14, Neutrophils in the tumor environment are Adam8, and Tgm2), immunomodulation (Arg1, Nos2, Cd274, protumorigenic and exhibit an altered functional profile Ptgs2, and Spp1), and hypoxia regulation (Vegfa, Hif1a, and Our finding that the myeloid cell landscape of each tumor Slc2a1) were markedly upregulated in TAMs and TANs com- type is distinctive raises the possibility that different tumor pared with splenic MDSCs. Interestingly, expression of these microenvironments impart unique functional attributes in the þ genes was similar between CD11b CD103 tumor DCs and myeloid cells therein. Although TAMs in different tumors have DCs from tumor-free mice. This result suggests that hypoxic been extensively characterized by functional and transcriptional responses are not uniform within the same microenvironment profiling approaches (22–26), our understanding of TANs is and may be tightly regulated among tumor-infiltrating leuko- limited. Recently, Fridlender and colleagues compared steady- cytes. Immunosuppression and tissue remodeling are critical state neutrophils to splenic MDSCs and TANs in a mesotheli- functions for promoting tumor progression, and hypoxia, oma model by transcriptional profiling and identified striking through hypoxia-inducible factor-1a (Hif-1a; ref. 18), can differences (27). However, this study focused only on a single promote transcriptional changes that facilitate the selection tumor and it is not known if such characteristics of TANs are

662 Cancer Immunol Res; 2(7) July 2014 Cancer Immunology Research

Heterogeneity of the Tumor Myeloid Cell Landscape

A 214 170

10 10

1 1 Fold change Fold

Fold change Fold 0.1 0.1 MDSC vs. naïve neutrophil) naïve MDSC vs. hi (B16 TAN vs. naïve neutrophil) naïve vs. TAN (B16 0.1 110 0.1 110

(4T1 Gr1 Fold change (4T1 TAN vs. naïve neutrophil) Fold change (Her2 TAN vs. naïve neutrophil) Figure 5. Expression of protumorigenic genes by TANs. A, B C comparison of gene expression IL10 signaling hi profiles of TANs and Gr1 MDSCs Hepatic fibrosis/hepatic stellate from 4T1 tumor-bearing mice to cell activation MDSC 4T1 naïve neutrophils (left, fold-change hi LXR/RXR activation Neutro B6 Neutro Balb TAN 4T1 TAN Her2 TAN B16 Gr1 vs. fold-change, 2-fold change). Hepatic cholestasis Expression of these probes was further analyzed in fold-change iNOS signaling versus fold-change plots VDR/RXR activation comparing TANs from B16 and Her2 tumors with naïve neutrophils IL6 signaling (right); 170 of the 214 probes PPAR signaling passed the CV < 0.5 filter. B, AHR signaling heatmap showing the expression Role of macrophages/fibroblasts/ of the 170 probes identified in A endothelial cells in RA across neutrophils. C, top 20 Glucocorticoid receptor signaling canonical pathways enriched for TANs and Gr1hi MDSCs compared MIF regulation of innate immunity with naïve neutrophils analyzed by Eumelanin biosynthesis Ingenuity Pathway Analysis. ROS, LPS/IL1–mediated inhibition reactive oxygen species. of RXR function Acute phase response signaling

Molecular mechanisms of cancer

PPARα/RXRα activation Production of NO and ROS in macrophages IL12 signaling and production in macrophages HGF signaling

0123456789 Ingenuity canonical pathway 4.23 13.73 Top 20 –log10(P ) Expression (log2)

common across different tumors. By comparing TANs from 4T1, tor-activated receptor (PPAR) signaling (Fig. 5C). Collectively, B16, and Her2 tumors to Gr1hi MDSCs from 4T1 tumor-bearing our data support that tumor-associated myeloid cells exhibit a mice and neutrophils from the bone marrow of na€ve mice (Fig. protumorigenic transcriptional program. 4), we identified a conserved expression pattern of immuno- To further interrogate molecular programs in TANs across modulatory and hypoxia-related genes across these tumors. We different tumors, a side-by-side comparison of TANs from also compared genes highly expressed in both Gr1hi MDSCs and 4T1, Her2, and B16 tumors was performed. We identified 43 TANs from 4T1 tumors to their expression in steady-state probes enriched in 4T1 TANs, 61 in Her2 TANs and 93 in B16 neutrophils (Fig. 5A, left) and found 214 probes selectively TANs (Fig. 6A). Genes involved in cell trafficking or differ- enriched in 4T1-associated neutrophils. Interestingly, many of entiation, such as Cxcr1, Foxo3,andSlc28a2, were enriched these genes were similarly enriched in TANs from B16 and Her2 2- to 5-fold in 4T1 TANs and may regulate the marked tumors compared with steady-state neutrophils (170 of 214 accumulation of TANs in this cancer type (28). Given that passed CV < 0.5 filter; Fig. 5A, right). Heatmap analysis also our data point to a direct, tumor cell–specificinvolvementin indicated that these probes were highly expressed by all TANs the recruitment and expansion of TANs, we analyzed the and Gr1hi MDSCs compared with steady-state neutrophils (Fig. expression levels of cytokines, cytokine receptors, chemo- 5B and Supplementary Table S3). Importantly, these genes are kines, and chemokine receptors by TANs, as these may involved in pathways critical for tumor growth and immune account for the tumor–myeloid cell cross-talk regulating responses, such as iNOS, IL-10, IL-6, and peroxisome prolifera- cell infiltration (Fig. 6B and Supplementary Table S4).

www.aacrjournals.org Cancer Immunol Res; 2(7) July 2014 663

Elpek et al.

A B Cytokine Chemokine Cytokine Chemokine 43 receptor receptor 61 10 Slc28a2 Cxcr1 Foxo3 Irf47 Irgm2 Ccgn2 Trim12a Irf1 Gbp1 TAN 4T1 TAN B16 TAN Her2 TAN 4T1 TAN B16 TAN Her2 TAN 4T1 TAN B16 TAN Her2 Ccl8 Ltb4r1 TAN 4T1 TAN B16 TAN Her2 1 Il2ra Ccr2 Ifna1 Ccl12 Il2rb Ccr5 Ifna9 Ccl2 Cd36 Il7r Cx3cr1 Ifnz Ccl5 Ly6a Ccl7 Il2rg Ccr1 Flt3l Ccl4 Arsb Ccr10 Ifna7 Thbs1 Itgb5 Il10rb Ccl7 0.1 Ccr2 Plxdc2 Il12rb1 Ccr7 Il1b Ccl9 Xcr1 Axl Il9r Ifna5 Ccl22 93 Il17rb Cxcr3 Il17a Ccl8 Cxcr4 Il31 Fold change (4T1 vs. Her2 TANs) Her2 change (4T1 vs. Fold 0.1 110 Il11ra1 Ccl3 Il6ra Cxcr1 Il13 Ccl6 Fold change (4T1 vs. B16 TANs) Il17ra Cxcr2 Il25 Ccl27a Il13ra1 Ltb4r1 Il16 Cxcl2 Il28ra Ltb Cxcl3 Il15ra 5.91 12.47 Ifna2 Ccl1 C Expression (log ) Mif Il27ra 2 Cx3cl1 Il12a Ccl17 5.53 11.34 Il15 Cxcl1 Expression (log ) Il18 2 Cxcl14 Il1f9 Cxcl16 Il17c

MDSC 4T1 Xcl1

hi Csf3 Ifna14 Cxcl10 Cxcl9 Gr1 Neutro B6 Neutro Balb TAN 4T1 TAN Her2 TAN B16 Il23a Tn f Ccl19 Lifr Il12b Ccl11 Emr1 Il1a Cxcl17 Vcam1 Tgfb1 Ccl25 Il18 Cxcl11 Pla1a 5.73 13.37 Ccrl2 Expression (log ) Ccl2 2 5.67 13.68

Il1rn D Expression (log2) Mt2 Mt1 Ccl3 Cxcl10 Nfkbia 4T1 B16 Icam1 TAN Ifi47 7 TAN Cebpb Il1b b Birc3 0.4 Her2 Irgm2 TAN Stat2 0.2 Stat3 Il1rap PC3 0 Steap4 (5.9%) B6 0.4 Stat1 −0.2 Il6ra MDSC Balb/c 0.2 Orm2 PC1 Lcn2 −0.4 0 Traf2 (66.4%) Nfkb1 0.4 0.2 –0.2 0 5.77 13.87 –0.2 Expression (log2) PC2 (14.6%)

Figure 6. Tumor type–specific transcriptional profiles of TANs. A, fold-change versus fold-change plot comparing 4T1 TANs with TANs in Her2 and B16, based on 2-fold change. Selected genes associated with the populations are indicated on the plots. B, chemokine, chemokine receptor, cytokine, and cytokine receptor expression profiles of TANs. C, PPARa-associated gene expression profiles of neutrophils. D, principal component analysis of neutrophil populations.

Strikingly, we identified different expression profiles for each triene B4. We further analyzed the expression of PPARa TAN population, which may reflect the differences observed target genes associated with inflammation (31) in neutrophil in the frequencies of these populations in each tumor. In subsets (Fig. 6C and Supplementary Table S5). Although particular, TANs from 4T1 tumors expressed significantly many of these genes are expressed at higher levels in TANs greater levels of Ltb4r1 (BLT1), a leukotriene B4 chemotactic compared with Gr1hi MDSCs and steady-state neutrophils, factor for neutrophils (29, 30). Interestingly, we also found neutrophils from 4T1-bearing mice expressed higher levels enrichment of genes in the anti-inflammatory PPARa sig- of Stat1, Stat2,andStat3, among other genes. Stat3 is a naling pathway (Fig. 5D), which is also a target for leuko- mediator of CXCR2 expression in neutrophils upon G-CSF

664 Cancer Immunol Res; 2(7) July 2014 Cancer Immunology Research

Heterogeneity of the Tumor Myeloid Cell Landscape

stimulation (32). In agreement with this, G-CSF is highly splenic Gr1hi MDSCsfrom4T1tumor-bearingmice,andin expressed by 4T1 cells (Supplementary Fig. S2A) and Cxcr2 is TANs from 4T1, Her2, and B16 tumors. Heatmap analysis expressed at higher levels by 4T1 TANs (Fig. 6B). Overall, and hierarchical clustering indicated that all TANs shared these signaling pathways may incite the marked accumula- an altered expression profile compared with na€ve granulo- tion of TANs and Gr1hi MDSCs in the 4T1 model. The cytes (Fig. 7A and Supplementary Table S6). Interestingly, relationship between neutrophil populations from different Gr1hi MDSCs expressed the majority of the effector mole- tumors was also confirmed by the principal component cules. Among the molecules associated with neutrophil analysis to highlight functional differences. Neutrophils from granules is the acute-phase responder haptoglobin (Hp;ref. na€ve mice clustered separately and more closely to MDSCs 34). Haptoglobin is known to be produced in the liver; on PC2 (Fig. 6D). On the other hand, while TANs clustered however, recent studies indicate that it is also expressed separately from neutrophils and MDSCs, variability on PC1 by neutrophils and packaged into secretory granules (35). reflected tumor-to-tumor differences among TANs. Analysis of haptoglobin expression by tumor-associated myeloid cells revealed that haptoglobin is not expressed by þ Haptoglobin levels are associated with myeloid cell DCs or CD11c TAMs in tumors, and its level is moderate in accumulation CD11c TAMs and in TANs from B16 and Her2 tumors (Fig. During acute inflammatory responses, neutrophils exocy- 7B). However, haptoglobin expression is high in na€ve gran- tose effector molecules stored in specific granules (33). ulocytes, Gr1hi MDSCs, and TANs from 4T1 tumors. Accord- Recently, Fridlender and colleagues showed that some of ingly, serum haptoglobin levels were elevated in mice bear- these molecules are actually downregulated in TANs com- ing advanced 4T1 tumors but not in Her2 or B16 tumor- paredwiththoseinsteady-state neutrophils and Gr1hi bearing hosts (Fig. 7C). Indeed, serum haptoglobin levels MDSCs (27). To understand how this critical function is correlatedwiththefrequencyofGr1hi MDSCs in the blood of influenced by different tumors, we compared the expression 4T1 tumor-bearing mice (Fig. 7D). This select pattern sug- of these effector molecules in steady-state neutrophils, in gests that haptoglobin expression may be associated with

A .313.90 5.43 Expression (log Neutro B6 Neutro Balb TAN Her2 TAN 4T1 TAN B16 hi 2

) Gr1 MDSC 4T1 Hp Lcn2 Fpr1 Mmp9 Mmp8 Itgam Itgb2 Mmp25 Camp Ltf Cyba Lyz1 Lyz2 Cd14 Mmp2 Fcgr3 Slc11a1 Bpi Prtn3 Ctsg Elane Mpo

Figure 7. Altered expression proﬁle B D of secretory granule molecules in 8,000 100 TANs. A, expression of granular effector molecules by neutrophils. 6,000

B, mean expression value of in blood hi 50 2 haptoglobin (Hp, HP) across 4,000 R = 0.71 neutrophils and tumor-associated 2,000 %Gr1 myeloid cells. C, serum haptoglobin levels in mice with Mean expression value 0 0 small (S), medium (M), large (L) 0 100200300400500 tumors and naïve mice. D, Serum HP concentration (μg/mL) DC B16 TAN 4T1 correlation between the frequency TAN B16 TAM 4T1 TAM 4T1 TAN Her2 TAM B16 TAM B16 Neutro B6 - - + + TAM Her2

hi + MDSC 4T1 MDSC 4T1 Neutro Balb

of Gr1 cells in blood and the hi low

serum haptoglobin concentration. Gr1 CD11c CD11c Gr1 CD11c CD11c E, haptoglobin levels in serum from CD11c healthy donors (n ¼ 4) and patients C E with breast cancer (n ¼ 10). ** 800 7,000 , P < 0.05; , P < 0.01. ** * 6,000 600 5,000 4,000 400 3,000

200 2,000 1,000 0 0 Serum HP concentration ( μ g/mL) Serum HP concentration ( μ g/mL) S MLS MLL Healthy Breast cancer Naive 4T1 Her2 B16 donor patient

www.aacrjournals.org Cancer Immunol Res; 2(7) July 2014 665

Elpek et al.

the considerable myelopoiesis observed in the 4T1 setting Disclosure of Potential Conflicts of Interest K.W. Wucherpfennig reports serving as a consultant/advisory board member (Supplementary Fig. S1E). Although further studies are fl fi for Novartis and Ipsum. No potential con icts of interest were disclosed by the required to con rm a direct relationship between neutro- other authors. phils and haptoglobin levels, we detected elevated serum haptoglobin in patients with breast cancer (Fig. 7E), con- Authors' Contributions sistent with reports for patients with glioblastoma and Conception and design: K.G. Elpek, S.J. Turley ovarian cancer (36, 37). Thus, haptoglobin may serve as a Development of methodology: K.G. Elpek, K.W. Wucherpfennig, S.J. Turley diagnostic marker indicative of cancers with myeloid cell Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K.G. Elpek, V. Cremasco, C.J. Harvey, D.R. Goldstein, accumulation. P.A. Monach, S.J. Turley Tumors require a sustained influx of myeloid cells to Analysis and interpretation of data (e.g., statistical analysis, biostatistics, support angiogenesis, remodel peritumoral stroma, and computational analysis): K.G. Elpek, V. Cremasco, C.J. Harvey, K.W. Wucherp- fennig, P.A. Monach, S.J. Turley suppress antitumor immunity. Therefore, tumor-associated Writing, review, and/or revision of the manuscript: K.G. Elpek, V. myeloid cells represent potential targets to improve immu- Cremasco, D.R. Goldstein, P.A. Monach, S.J. Turley Administrative, technical, or material support (i.e., reporting or orga- notherapy as well as chemotherapy against cancer. Although nizing data, constructing databases): H. Shen there are limited strategies available, identifying novel targets for elimination or functional differentiation of these cells requires a deeper understanding of their development, Acknowledgments The authors thank the Immunological Genome Project team for technical trafficking, and phenotypic and functional characteristics. help and discussions; Drs. Dobrin Draganov and Glenn Dranoff for the Her2 cell Moreover, to ascertain whether a broader therapeutic appli- line; and Dr. Ronald DePinho for PDA mice. cation is feasible, it is critical to determine whether key fi molecular attributes are conserved among speci c myeloid Grant Support populations from tumors of different origins. A better This study was supported by research funding from NIH T32 grant to K.G. understanding of the functional differences among discrete Elpek, NIH T32 CA 070083-15 and Postdoctoral Fellowship from the Cancer Research Institute to V. Cremasco, NIH grants (AG028082, AG033049, AI098108, myeloid subsets in tumor lesions, as well as secondary and AI101918) and an Established Investigator Award (0940006N) from Amer- lymphoid organs, may lay the groundwork for development ican Heart Association to D.R. Goldstein, NIH grants (ROI DK074500, PO1 AI045757, and R21 CA182598), American Cancer Society Research Scholar Grant, of novel therapeutics. In this regard, our study provides and Claudia Adams Barr Award for Innovative Cancer Research to S.J. Turley. fundamental insights into functional differences of tumor- The costs of publication of this article were defrayed in part by the payment of infiltrating myeloid cells, and identifies potential molecular page charges. This article must therefore be hereby marked advertisement in pathways that could be investigated for therapies aimed at accordance with 18 U.S.C. Section 1734 solely to indicate this fact. fi reducing in ltration and/or reprogramming function of Received November 30, 2013; revised February 19, 2014; accepted March 12, tumor myeloid cells. 2014; published OnlineFirst March 31, 2014.

References 1. Bussolati B, Grange C, Camussi G. Tumor exploits alternative strat- 11. Ying H, Elpek KG, Vinjamoori A, Zimmerman SM, Chu GC, Yan H, et al. egies to achieve vascularization. FASEB J 2011;25:2874–82. PTEN is a major tumor suppressor in pancreatic ductal adenocarci- 2. Gabrilovich DI, Ostrand-Rosenberg S, Bronte V. Coordinated regula- noma and regulates an NF-kappaB-cytokine network. Cancer Discov tion of myeloid cells by tumours. Nat Rev Immunol 2012;12:253–68. 2011;1:158–69. 3. Rahir G, Moser M. Tumor microenvironment and lymphocyte infiltra- 12. Gautier EL, Shay T, Miller J, Greter M, Jakubzick C, Ivanov S, et al. tion. Cancer Immunol Immunother 2012;61:751–9. Gene-expression profiles and transcriptional regulatory pathways that 4. Tlsty TD, Coussens LM. Tumor stroma and regulation of cancer underlie the identity and diversity of mouse tissue macrophages. Nat development. Annu Rev Pathol 2006;1:119–50. Immunol 2010;13:1118–28. 5. Marigo I, Bosio E, Solito S, Mesa C, Fernandez A, Dolcetti L, et al. 13. Miller JC, Brown BD, Shay T, Gautier EL, Jojic V, Cohain A, et al. Tumor-induced tolerance and immune suppression depend on the C/ Deciphering the transcriptional network of the dendritic cell lineage. EBPbeta transcription factor. Immunity 2010;32:790–802. Nat Immunol 2012;13:888–99. 6. Cortez-Retamozo V, Etzrodt M, Newton A, Rauch PJ, Chudnovskiy A, 14. Heng TS, Painter MW. The Immunological Genome Project: networks Berger C, et al. Origins of tumor-associated macrophages and neu- of gene expression in immune cells. Nat Immunol 2008;9:1091–4. trophils. Proc Natl Acad Sci U S A 2012;109:2491–6. 15. Cheung R, Shen F, Phillips JH, McGeachy MJ, Cua DJ, Heyworth PG, 7. Le HK, Graham L, Cha E, Morales JK, Manjili MH, Bear HD. Gemci- et al. Activation of MDL-1 (CLEC5A) on immature myeloid cells triggers tabine directly inhibits myeloid derived suppressor cells in BALB/c lethal shock in mice. J Clin Invest 2011;121:4446–61. mice bearing 4T1 mammary carcinoma and augments expansion of T 16. Toh B, Wang X, Keeble J, Sim WJ, Khoo K, Wong WC, et al. Mesen- cells from tumor-bearing mice. Int Immunopharmacol 2009;9:900–9. chymal transition and dissemination of cancer cells is driven by 8. Vincent J, Mignot G, Chalmin F, Ladoire S, Bruchard M, Chevriaux A, myeloid-derived suppressor cells infiltrating the primary tumor. PLoS et al. 5-Fluorouracil selectively kills tumor-associated myeloid-derived Biol 2011;9:e1001162. suppressor cells resulting in enhanced T cell-dependent antitumor 17. Waight JD, Hu Q, Miller A, Liu S, Abrams SI. Tumor-derived G-CSF immunity. Cancer Res 2010;70:3052–61. facilitates neoplastic growth through a granulocytic myeloid- 9. Ko JS, Zea AH, Rini BI, Ireland JL, Elson P, Cohen P, et al. Sunitinib derived suppressor cell-dependent mechanism. PLoS ONE 2011;6: mediates reversal of myeloid-derived suppressor cell accumulation in e27690. renal cell carcinoma patients. Clin Cancer Res 2009;15:2148–57. 18. Shay JE, Celeste Simon M. Hypoxia-inducible factors: crosstalk 10. DeNardo DG, Brennan DJ, Rexhepaj E, Ruffell B, Shiao SL, Madden between inflammation and metabolism. Semin Cell Dev Biol 2012;23: SF, et al. Leukocyte complexity predicts breast cancer survival and 389–94. functionally regulates response to chemotherapy. Cancer Discov 19. Favaro E, Lord S, Harris AL, Buffa FM. Gene expression and hypoxia in 2011;1:54–67. breast cancer. Genome Med 2011;3:55.

666 Cancer Immunol Res; 2(7) July 2014 Cancer Immunology Research

Heterogeneity of the Tumor Myeloid Cell Landscape

20. McGovern NN, Cowburn AS, Porter L, Walmsley SR, Summers C, expressed in rheumatoid blood and synovial tissue. Ann Rheum Dis Thompson AA, et al. Hypoxia selectively inhibits respiratory burst 2010;69:755–60. activity and killing of Staphylococcus aureus in human neutrophils. 29. Kim ND, Chou RC, Seung E, Tager AM, Luster AD. A unique require- J Immunol 2011;186:453–63. ment for the leukotriene B4 receptor BLT1 for neutrophil recruitment in 21. Walmsley SR, Print C, Farahi N, Peyssonnaux C, Johnson RS, Cramer inflammatory arthritis. J Exp Med 2006;203:829–35. T, et al. Hypoxia-induced neutrophil survival is mediated by HIF- 30. Saiwai H, Ohkawa Y, Yamada H, Kumamaru H, Harada A, Okano H, 1alpha–dependent NF-kappaB activity. J Exp Med 2005;201:105–15. et al. The LTB4-BLT1 axis mediates neutrophil infiltration and sec- 22. Duff MD, Mestre J, Maddali S, Yan ZP, Stapleton P, Daly JM. Analysis ondary injury in experimental spinal cord injury. Am J Pathol of gene expression in the tumor-associated macrophage. J Surg Res 2010;176:2352–66. 2007;142:119–28. 31. Rakhshandehroo M, Knoch B, Muller M, Kersten S. Peroxisome pro- 23. Ojalvo LS, King W, Cox D, Pollard JW. High-density gene expression liferator-activated receptor alpha target genes. PPAR Res 2010;2010: analysis of tumor-associated macrophages from mouse mammary pii:612089. tumors. Am J Pathol 2009;174:1048–64. 32. Nguyen-Jackson H, Panopoulos AD, Zhang H, Li HS, Watowich SS. 24. Ojalvo LS, Whittaker CA, Condeelis JS, Pollard JW. Gene expression STAT3 controls the neutrophil migratory response to CXCR2 ligands analysis of macrophages that facilitate tumor invasion supports a role by direct activation of G-CSF–induced CXCR2 expression and via for Wnt-signaling in mediating their activity in primary mammary modulation of CXCR2 signal transduction. Blood 2010;115:3354–63. tumors. J Immunol 2010;184:702–12. 33. Borregaard N. Neutrophils, from marrow to microbes. Immunity 25. Pucci F, Venneri MA, Biziato D, Nonis A, Moi D, Sica A, et al. A 2010;33:657–70. distinguishing gene signature shared by tumor-infiltrating Tie2- 34. Wang Y, Kinzie E, Berger FG, Lim SK, Baumann H. Haptoglobin, an expressing monocytes, blood "resident" monocytes, and embryonic inflammation-inducible plasma protein. Redox Rep 2001;6:379–85. macrophages suggests common functions and developmental rela- 35. Theilgaard-Monch K, Jacobsen LC, Nielsen MJ, Rasmussen T, Udby tionships. Blood 2009;114:901–14. L, Gharib M, et al. Haptoglobin is synthesized during granulocyte 26. Stearman RS, Dwyer-Nield L, Grady MC, Malkinson AM, Geraci MW. A differentiation, stored in specific granules, and released by neutrophils macrophage gene expression signature defines a field effect in the lung in response to activation. Blood 2006;108:353–61. tumor microenvironment. Cancer Res 2008;68:34–43. 36. Kumar DM, Thota B, Shinde SV, Prasanna KV, Hegde AS, Arivazhagan 27. Fridlender ZG, Sun J, Mishalian I, Singhal S, Cheng G, Kapoor V, et al. A, et al. Proteomic identification of haptoglobin alpha2 as a glioblas- Transcriptomic analysis comparing tumor-associated neutrophils with toma serum biomarker: implications in cancer cell migration and tumor granulocytic myeloid-derived suppressor cells and normal neutrophils. growth. J Proteome Res 2010;9:5557–67. PLoS ONE 2012;7:e31524. 37. Zhao C, Annamalai L, Guo C, Kothandaraman N, Koh SC, Zhang H, 28. Turrel-Davin F, Tournadre A, Pachot A, Arnaud B, Cazalis MA, Mougin et al. Circulating haptoglobin is an independent prognostic factor in the B, et al. FoxO3a involved in neutrophil and T cell survival is over- sera of patients with epithelial ovarian cancer. Neoplasia 2007;9:1–7.

www.aacrjournals.org Cancer Immunol Res; 2(7) July 2014 667

The Tumor Microenvironment Shapes Lineage, Transcriptional, and Functional Diversity of Infiltrating Myeloid Cells

Kutlu G. Elpek, Viviana Cremasco, Hua Shen, et al.

Cancer Immunol Res 2014;2:655-667. Published OnlineFirst March 31, 2014.

Updated version Access the most recent version of this article at: doi:10.1158/2326-6066.CIR-13-0209

Supplementary Access the most recent supplemental material at: Material http://cancerimmunolres.aacrjournals.org/content/suppl/2014/04/03/2326-6066.CIR-13-0209.DC1

Cited articles This article cites 37 articles, 14 of which you can access for free at: http://cancerimmunolres.aacrjournals.org/content/2/7/655.full#ref-list-1

Citing articles This article has been cited by 7 HighWire-hosted articles. Access the articles at: http://cancerimmunolres.aacrjournals.org/content/2/7/655.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 Subscriptions at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerimmunolres.aacrjournals.org/content/2/7/655. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.