Cellular and Molecular Identity of Tumor- …...Microenvironment and Immunology Cellular and...

Microenvironment and Immunology

Cellular and Molecular Identity of Tumor-Associated Macrophages in GlioblastomaZhihong Chen1,2, Xi Feng2, Cameron J. Herting1, Virginia Alvarez Garcia2,Kai Nie1,2,Winnie W. Pong3, Rikke Rasmussen2, Bhakti Dwivedi4, Sandra Seby4,Susanne A.Wolf5, David H. Gutmann3, and Dolores Hambardzumyan1

Abstract

In glioblastoma (GBM), tumor-associated macrophages(TAM) represent up to one half of the cells of the tumor mass,including both infiltrating macrophages and resident brainmicroglia. In an effort to delineate the temporal and spatialdynamics of TAM composition during gliomagenesis, weused genetically engineered and GL261-induced mouse mod-els in combination with CX3CR1GFP/WT;CCR2RFP/WT doubleknock-in mice. Using this approach, we demonstrated thatCX3CR1LoCCR2Hi monocytes were recruited to theGBM,wherethey transitioned to CX3CR1HiCCR2Lo macrophages andCX3CR1HiCCR2� microglia-like cells. Infiltrating macrophages/monocytes constituted approximately 85% of the total TAMpopulation, with resident microglia accounting for the approxi-

mately 15% remaining. Bone marrow–derived infiltrating macro-phages/monocytes were recruited to the tumor early duringGBM initiation, where they localized preferentially to perivascularareas. In contrast, resident microglia were localized mainlyto peritumoral regions. RNA-sequencing analyses revealed differ-ential gene expression patterns unique to infiltrating and residentcells, suggesting unique functions for each TAMpopulation. Nota-bly, limiting monocyte infiltration via genetic Ccl2 reductionprolonged the survival of tumor-bearing mice. Our findings illu-minate the unique composition and functions of infiltratingand resident myeloid cells in GBM, establishing a rationale totarget infiltrating cells in this neoplasm. Cancer Res; 77(9); 2266–78.�2017 AACR.

IntroductionThe most common histologic type of brain tumor in adults is

the glioma, with themore malignant grades dominating betweenthe fifth and eighth decades of life. Among these clinically aggres-sive and deadly cancers, glioblastoma (GBM;World Health Orga-nization grade IV) is the most frequently encountered. Thesetumors most often arise within the subcortical white matter ofthe cerebral hemispheres, where they appear as poorly demarcat-ed masses associated with increased vascularity and tissue necro-sis. Histologic analysis of GBM reveals highly anaplastic andmitotic tumor cells embedded with a tumor microenvironmentcomposed of brain-resident microglia, infiltrating monocytes/macrophages, reactive astrocytes, endothelial cells, pericytes, neu-ral stem/progenitor cells, and other immune cell infiltrates (1, 2).The tissue macrophage compartment contains both residentmicroglia and bone marrow–derived macrophages, although the

exact composition and tissue origins have been incompletelycharacterized.

The importance of these tumor-associated macrophage(TAM) populations to glioma growth is highlighted by theirlarge numbers within these tumors, comprising as many as30% to 50% of all cells in human GBM (2). In addition,numerous studies focused on experimental murine low-gradeglioma and GBM models have revealed a critical role for theseTAMs in tumor formation and maintenance. Using a neurofi-bromatosis type 1 (NF1) model of low-grade glioma, pharma-cologic or genetic silencing of microglia delays glioma forma-tion and attenuates established tumor growth, suggesting thepresence of growth factors elaborated by these TAMs (3). RNA-sequencing of TAMs in this model revealed that the CCL5chemokine is a major microglia-derived driver of neurofibro-matosis 1 glioma growth (4). In addition, a large number ofelegant studies on murine high-grade glioma have demonstrat-ed that microglia are critical for tumor growth and migration(5). These microglia are recruited by the glioma cells to estab-lish a feed-forward cellular circuit that drives further tumorgrowth (5). In this regard, changing TAM polarization byinhibiting colony-stimulating factor 1 receptor signalingreduced PDGFB-driven murine GBM growth (6); however, thisstrategy failed to show efficacy in clinical trials for recurrentGBM (7).

One reason that targeting TAMs in GBM has not been suc-cessful can be partially attributable to the fact that TAMs are amixed population. Macrophages derived from infiltrating bonemarrow–derived monocytes or resident microglia can be mor-phologically indistinguishable in tissue sections and are notreliably separated using available lineage-specific antibodies.Traditionally, investigators have relied on CD45 antibodies to

1Department of Pediatrics and Aflac Cancer Center of Children's Health Care ofAtlanta, Emory University School of Medicine, Atlanta, Georgia. 2Department ofNeurosciences, Cleveland Clinic Lerner Research Institute, Cleveland, Ohio.3Department of Neurology,Washington University School ofMedicine, St. Louis,Missouri. 4WinshipCancer Institute, EmoryUniversity, Atlanta,Georgia. 5Depart-ment of Cellular Neuroscience, Max-Delbr€uck-Center of Molecular Medicine inthe Helmholtz Association, Berlin, Germany.

Note: Supplementary data for this article are available at Cancer ResearchOnline (http://cancerres.aacrjournals.org/).

Corresponding Author: Dolores Hambardzumyan, Emory University, 1760 Hay-good Drive, E-380, Atlanta, GA 30329. Phone: 347-420-0021; Fax: 404-785-1178; E-mail: [email protected]

doi: 10.1158/0008-5472.CAN-16-2310

�2017 American Association for Cancer Research.

CancerResearch

Cancer Res; 77(9) May 1, 20172266

on December 2, 2020. © 2017 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst February 24, 2017; DOI: 10.1158/0008-5472.CAN-16-2310

http://crossmark.crossref.org/dialog/?doi=10.1158/0008-5472.CAN-16-2310&domain=pdf&date_stamp=2017-4-19

http://cancerres.aacrjournals.org/

distinguish resident microglia (CD45Lo) from macrophages ofhematopoietic origin (CD45Hi) using FACS analysis (8). The useof CD45 to separate these populations was recently challengedby studies using head-protected irradiation chimeras, showingthat microglia can increase CD45 expression to constitute aproportion of the CD45Hi population in gliomas (9). In addi-tion, there is also a significant myeloid cell infiltration in high-grade glioma, postulated to result from total body irradiation(TBI) damage to brain–blood barrier integrity (9, 10). As such, itis currently unclear to what extent GBMmacrophages are derivedfrom infiltrating cells from the blood circulation or what pro-portion of the GBM macrophages are microglia versus mono-cyte/macrophages. Moreover, it is not known where each pop-ulation is physically localized within GBM, which containsspecialized cellular niches (11).

To gain insights into the populations of TAMs that comprisemurine experimental GBM,we leveraged amousemodel inwhichthe Cx3cr1 and Ccr2 genes were replaced with distinct fluorescentproteins as knock-in alleles (Cx3cr1GFP/WT;Ccr2RFP/WT mice).The use of these mice is predicated on the finding that mousemonocytes can be subdivided into Ly6CþCX3CR1IntCCR2þ

inflammatory monocytes and Ly6CLo/�CX3CR1HiCCR2� circu-lating monocytes (12, 13). Combining Cx3cr1GFP/WT;Ccr2RFP/WT

knock-in mice with a genetically engineered mouse model(GEMM) of PDGFB-driven GBM, we show that the majorityof inflammatory monocytes/macrophages express both CCR2and CX3CR1, whereas microglia express only CX3CR1. Multipa-rameter flow cytometry analyses showed that CD45Hi populationconsists of CCR2þCX3CR1þ cells, whereas the CD45Lo popula-tion only contains CX3CR1þ cells. In addition, over 85%of the TAMs within the tumors are bone marrow–derived macro-phages, whereas microglia predominate in the peritumoralareas. In addition, RNA-sequencing analyses revealed differentialgene expression patterns unique to infiltrating and resident cells,suggesting unique functions for eachTAMpopulation. Loss of onecopy of Ccl2, the ligand for CCR2, which recruits monocytes tosites of inflammation, significantly prolonged the survival ofGBM-bearing mice. Our results demonstrate the unique spatialand temporal differential composition, and distinct biologicalfunctions for infiltrating and resident myeloid cells in GBM.

Materials and MethodsMice

Animals were housed in the Cleveland Clinic BiologicalResource Unit or Emory University Division of Animal Resources.Mice undergoing GL261 glioma cell injections were housed at theMax-Delbr€uck-Center of Molecular Medicine in the HelmholtzAssociation. All experimental procedures were approved by theInstitutional Animal Care and Use Committee of the ClevelandClinic (Animal Protocol 2013-1029) and Emory University (pro-tocol no. 2003253) and by the LaGeSo of the Helmholtz Asso-ciation (G0438/12). Mice (6–10 weeks old, except animals thatwent through bonemarrow transplant, which are usually 8 weeksolder) of both genders were used in all experiments. We used ageand sex as criteria to equally distribute mice from differentgenotypes for all the experiments. Gli-Luciferase; Nestin-tv-a;Ink4a-Arf�/�;Ptenfl/fl mice were generated as previously described(1, 14). B6 (Cx3cr1WT/WT;Ccr2WT/WT) andCx3cr1GFP/WT;Ccr2RFP/WT

mice were a gift fromRichard Ransohoff (Biogen, Cambridge,MA;refs. 15, 16).

Cell cultures and transfectionDF-1 cells were purchased from ATCC in 2010. Cells were

grown at 39�C according to instructions from ATCC, expanded topassage 4, and stored in aliquots in liquid nitrogen. Transfectionwith RCAS-PDGFB-HA and RCAS-shRNA p53 were performedusing a Fugene 6 Transfection Kit (no. 11814443001; Roche)according to manufacturer's instructions. Transfected cells areused for injections before they reach passage 25. Cells were testedfor mycoplasma contamination by using a "PlasmoTest" kit(Invivogen). Murine GL261 glioma cells were obtained from theNational Cancer Institute in 2016. They were grown in DMEMwith 10% FCS, 200mmol/L glutamine, 100 U/mL penicillin, and100mg/mL streptomycin (all from Invitrogen). Cells cultured lessthan five passages were used to inject mice to induce glioma.Because of the short history of growth and healthy appearance, nomycoplasma test was done on GL261 cells.

Generation of tumors using RCAS/Ntva systemDonors. Transgenic mice (Gli-Luciferase;Nestin-tv-a,ink4a-Arf�/�

Ptenfl/fl referred as NiG) were anesthetized with intraperitonealinjections of a mixture of ketamine and xylazine (ketamine, 0.1mg/g and xylazine, 0.02 mg/g). One microliter of 4 � 104 cellsuspension containing RCAS-PDGF-B-HA–transfected DF1 cellswas delivered using a 30-gauge needle attached to a Hamiltonsyringe and stereotactic fixation device (Stoelting). Locationsweredetermined according to the brain atlas (17). Cells were injectedinto the right frontal striatum with the following coordinates:bregma AP (anterior/posterior) 1.0 mm, ML (medial/lateral) 1.0mm, and DV (dorsal/ventral) 2.0 mm. The same location wasused to generate tumor inNtva;Ccl2þ/þ andNtva;Ccl2þ/�.Ccl2�/�

micewere obtained from Jackson laboratory (no. 004434; ref. 18)and were further crossed to B6/Ntva mice. Mice were monitoredcarefully and were sacrificed if they displayed lethargy due totumor burden.

Orthotopic glioma generationThe same procedure was used as described above, except that

2.5 � 104 of freshly dissociated tumor cells from RCAS/Ntvadonors were injected into the frontal striatum of recipientanimals.

Generation of murine gliomas using GL261 cell lineOne-microliter cell suspension of 2 � 104 GL261 cells was

delivered using a 30-gauge needle attached to a Hamilton syringe(Hamilton).Coordinates for injectionswere 1mmanterior, 2mmlateral, and 3 mm deep relative to the bregma. Mice were mon-itored daily for the first 2 weeks and twice a day starting from day15 postinjection for symptoms of tumor development (lethargy,hydrocephalus, head tilting). Mice were euthanized 21 dayspostinjection.

Tissue processingAnimals were anesthetized with a ketamine/xylazine mix,

perfused with ice-cold Ringer's solution, and sacrificed. Brainswere removed and processed according to the different applica-tions. For hematoxylin and eosin tumor validation and immu-nohistochemistry staining, brains were fixed in 10% neutralbuffered formalin for 72 hours at room temperature, processedin a tissue processer (Leica TP1050), embedded in paraffin,sectioned (5 mm), and slide mounted. For immunofluorescent

Identity and Gene Profile of Glioblastoma TAM

www.aacrjournals.org Cancer Res; 77(9) May 1, 2017 2267




staining, brains were fixed in 4% PFA overnight at 4�C, immersedin 30% sucrose (dissolved in PBS) for 48 hours at 4�C, embeddedin optimal cutting temperature (OCT, Tissue-Tek) compound,sectioned (8 mm), slide mounted, and stored at �80�C.

Immunofluorescent stainingEight-micron coronal sections were used for frozen sections in

all histologic studies. The following antibodies were used at thestated dilutions: rabbit polyclonal anti-Iba1, 1:100 (Wako PureChemicals); rabbit polyclonal anti-RFP, 1:50 (Rockland), ratmonoclonal anti-CD31, 1:100 (Histonova), and rabbit anti-Ki67,1:100 (Abcam). Secondary antibodies conjugated to differentAlexa-Fluor dyes (488, 555, and 647 nm from Invitrogen) at adilution of 1:500 in PBS/2% BSA were applied. For nuclearcounterstaining, DAPI was used (Sigma). For quantification ofIba1þ, GFPþ, or RFPþ cells, five to ten images (20�) of tumor andperitumoral regions were taken per mouse brain using CD31 orDAPI staining as a reference on an Olympus FV1000 confocalmicroscope. TUNEL assay (Sigma) was performed as previouslydescribed (19). Full brain coronal sections were created by mul-tiple area tiled scanning using the samemicroscope. Cell numberswere counted with FIJI and normalized to a 1 mm2 area.

Flow cytometryBrains were digested in 0.25% Trypsin/EDTA without phenol

red at 37�C for 10 minutes (GBM) or 30 minutes (na€�ve brains,without cerebellum). Digestion was terminated by adding twovolumes of RPMImedium containing 10% FBS. Cells were passedthrough a 40-mm cell strainer, centrifuged, and resuspended in30% Percoll (GE Healthcare) solution, and layered above a 70%Percoll layer (diluted in RPMI medium with 1% FBS). Cells wereseparated by centrifuging at 800 � g for 30 minutes at 4�C. Cellsfrom the 30%/70% Percoll interphase were collected and washedwith FACS buffer (DPBSwith 0.5% BSA) and blockedwith 100 mLof 2� blocking solution (2% FBS, 5% normal rat serum, 5%normal mouse serum, 5% normal rabbit serum, 10 mg/mL2.4g2 anti-FcR, and 0.2% NaN3 in DPBS) on ice for 30 minutes.Cells were then stained on ice for 30 minutes and washed withFACS buffer. For enumeration of blood cell types, blood sampleswere taken via tail vein bleeding, lysed with RBC lysis buffer(BioLegend, cat. no. 420301), washed with FACS buffer, countedwith a hemocytometer, and stained using the same protocoldescribed above. Antibodies used in the study include: CD45-APC,CD11b-PerCP-Cy5.5, Ly6C-PE-Cy7, F4/80-APC-Cy7 (BD Phar-mingen), and Ly6G-V450 (BioLegend). All data were collected ona BD LSR flow cytometer and analyzed using FlowJo 10 software(Tree Star Inc.).

Generation of bone marrow chimerasTwo irradiation regiments were used to ablate host hemato-

poietic system for bonemarrow chimera generations, thefirstwithTBI and the second with head-protected irradiation (HPI), wherethe heads of the mice were protected from the ionizing X-ray. ForTBI, recipient mice 4 to 5 weeks old were irradiated in a ShepherdMark 137Cs irradiator by two doses of irradiation at 600 rads each,with a 4-hour interval in between. For HPI, anesthetized micewere first placed in a custom-made apparatus where their bodieswere accessible byX-raybut their headswereprotectedby a 5-mm-thick lead sheet. The apparatus was then placed in a low-energyPantak Cabinet X-ray irradiator with totaling 950 rads delivered.Bonemarrow cells were collected from femurs and tibias of donor

mice. Single bone marrow cells were suspended in sterile HBSSat 2 � 108 cells/mL. One hundred microliters of cell suspension(2 � 107 cells) was injected in each recipient mouse through theretro-orbital sinus. Recipientmice received 1.2mg/mLgentamicin(Thermo Fisher) through drinking water for 10 days. Eight to10 weeks after the bone marrow transplant, 50 mL of blood werecollected through an incision in the tail vein and placed in 10U ofheparin (Sigma). The erythrocytes were lysed and the leukocyteswere washed in PBS and analyzed by flow cytometry as describedabove to determine the efficiency of chimeric reconstitution. Toquantify the reconstitution efficiency, we took advantage of thefact that all Ly6CHi inflammatory monocytes express CCR2. Aformula thus can be generated: % reconstitution efficiency ¼CCR2RFP cells/Ly6CHi cells � 100%.

RNA-sequencing and data analysesDetailed description can be found in the Supplementary Mate-

rials. Briefly, Na€�vemicroglia, na€�vemonocytes, tumor-associatedmicroglia, and tumor-associated macrophages were isolated byFACS sorting based on CD11b, CD45, and CX3CR1 and CCR2combination, details are described above. Their mRNA wasextracted with RNeasy Plus Mini Kit (Qiagen) per manufacturer'sinstructions. The quality of the mRNA was assessed by Bioana-lyzer (Agilent Technology) and theRINnumberswere above 9.50.Complementary DNA (cDNA) library was established by usingOvation RNA-Seq v2 method (NuGen). Samples were sequencedon the HiSeq 2000 at 2 � 101 bp in the paired ends. IlluminaHiSeq Control Software and Real-Time Analysis were used forsequencing and raw instrument data processing. Gene and iso-form expression levels were calculated using Cufflinks version2.1.1. Transcripts from mitochondrial and ribosomal RNA geneswere masked and not included. The final abundances of genes incells were computed in Fragments Per Kilobase of exonmodel perMillion mapped fragments (FPKM). The general transfer format(GTF) from Ensembl release 67 was used for genome-widetranscriptome quantification of protein-coding genes, annotatedin mm9, including alternative transcript isoform expression esti-mation. To compare gene and transcript expression under twoconditions, the annotatedGTFwas fed to theCuffdiff algorithm inCufflinks to measure the fold change of the coding genes. Finaldifferentially expressed genes (DEG) were listed with theirexpected fragment numbers. To identify overlapping or uniqueDEGs, a minimum Log 2-fold change of 2 and P value � 0.01were used as cut-off value in the pairwise comparisons. A crosscomparison was then applied to identify overlapping andunique DEGs between tumor-associated microglia and macro-phages. The biological processes enriched among the overlap-ping DEGs were searched against a variety of databases usingthe Reactome FIViz plugin (20) in Cytoscape (21). The signif-icance of gene ontology (GO) terms was determined based on Pvalue (�0.001) and FDR (�0.025). The number of uniqueDEGs specific to tumor-associated microglia, TAMs, and com-mon to both in a defined functional category are presented. Theraw RNA-Seq data, along with experimental information isdeposited in the NCBI Sequence Read Archive database underaccession number PRJNA349180.

Quantitative real-time PCRRNA concentrations were measured with a NanoDrop spec-

trophotometer and samples were stored at �80�C. cDNAwas synthesized from total RNA using the SuperScript III

Chen et al.

Cancer Res; 77(9) May 1, 2017 Cancer Research2268




First-Strand Synthesis System (Life Technologies). Quantitativereal-time PCR was performed using a SsoAdvanced UniversalSYBR Green Supermix (Bio-Rad) according to manufacturer'sinstructions. Amplification was performed on a Bio-Rad CFX96Real-time System. Primers are purchased as off-the-shelf pro-ducts from Bio-Rad. HPRT was used as an internal control andthe DDCt method was used to calculate changes in foldexpression.

Statistical analysisGraphs were created using GraphPad Prism 6 (GraphPad

Software Inc.) and were analyzed using an unpaired or pairedparametric two-tailed t-test as appropriate, assuming equalSDs. One-way ANOVA was used in experiments having morethan one group to compare to controls. The details are includedin the figure legends (�) P < 0.05; (��) P < 0.01; (��) P < 0.001;(no asterisk) not significant.

ResultsThe majority of TAMs are infiltrating CD45Hi leukocytes fromthe blood circulation

To characterize TAMs in murine PDGFB-driven GBM, weinitially used standard antibody combinations for flow cyto-metric analysis. For these experiments, tumors were dissectedfrom B6 mice when the mice were terminally ill from tumorburden (survival endpoint), and the total percentages ofCD45þCD11bþ myeloid cells examined (Fig. 1A and B). Rel-ative to na€�ve mice, where the CD45Hi population accounts forabout 1.5 � 0.3% of the total brain myeloid cell population(Fig. 2A), the percentage of CD45Hi cells in the brain tumorswas increased (87.5 � 6.2%, P < 0.01; Fig. 1A). Next, the Ly6C

and Ly6G surface markers were used to subdivide CD45Hi

cells into three distinct populations (Ly6CHiLy6G�monocytes,Ly6CLoLy6G� tissue macrophages, and Ly6CþLy6Gþ neu-trophils; Fig. 1A). Among all of the CD45Hi cells, the majorityexpressed low levels of Ly6C (Ly6CLoLy6G� macrophages). Incontrast, 12.2 � 4.5% of these cells expressed high levelsof Ly6C (Ly6CHiLy6G� inflammatory monocytes), whileonly 4.4 � 2.6% expressed both markers (Ly6CþLy6Gþ

neutrophils; Fig. 1A). CD45Lo resident microglia expressed noor very low levels of Ly6C, consistent with previous findings(22). Interestingly, when the expression of F4/80, a maturephagocytic cell marker, was examined, we found that theCD45HiLy6CLo population exhibited the highest levels ofF4/80 (mean fluorescent intensity or MFI at 2.3 � 1.4 � 103, P< 0.001), indicating that these cells had differentiated into tissuemacrophages (Fig. 1B). In contrast, Ly6CHi inflammatory mono-cytes maintained low levels of F4/80 (MFI¼ 0.83� 0.68� 103),confirming that they recently infiltrated from the blood circula-tion into the tumor.Microglial expression of F4/80 (MFI¼ 0.77�0.29� 103) was comparable with that of the Ly6CHi cells, but wasmuch lower than that observed in the Ly6CLo cells. Neutrophilexpression of F4/80 was undetectable (Fig. 1B).

Cx3cr1GFP/WT;Ccr2RFP/WT knock-in mice distinguish residentmicroglia from bone marrow–derived macrophages

Although the above FACS marker combinations are useful fordistinguishing resident microglia from bone marrow–derivedmonocyte/macrophages, they are not suitable for identifyingthese cells in histologic sections (23). To define the monocytepopulations in murine GBM tissues using genetic strategies,we took advantage of Cx3cr1GFP/WT;Ccr2RFP/WTmice, which havethesefluorescent protein transgenes inserted into the endogenous

Figure 1.

The majority of TAMs are infiltratingCD45Hi leukocytes from the bloodcirculation. A, Representative dotplots gated on the CD11bþCD45þ cellsfrom tumors generated in B6 mice.The total population of CD11bþCD45þ

cells is considered to be 100%, withCD11bþCD45Hi (blood-derivedmonocytes and macrophages) andCD11þCD45Lo (resident brainmicroglia) populations gatedseparately. Their expression of Ly6Cand Ly6G were analyzed and thepercentage of each subpopulationquantified. B, Quantification of F4/80intensity for all the subpopulations;N ¼ 5; �� , P < 0.001 by ANOVA.






loci to mimic normal CX3CR1 and CCR2 expression. In theseexperiments, we leveraged the complementary expressionintensities of CX3CR1 and CCR2 to distinguish inflammatorymonocytes (CX3CR1LoCCR2Hi cells with high Ly6C levels),patrolling monocytes (CX3CR1HiCCR2Lo cells with low Ly6Clevels; ref. 24), and resident brain microglia (CX3CR1HiCCR2�

cells) in murine GBM.We first demonstrated that a single copy replacement of Cx3cr1

or Ccr2 gene by GFP or RFP did not affect tumorigenesis, asdetermined by mouse survival and glioma penetrance (Supple-mentary Table S1). Consistent with previously published reports(25), na€�ve brains from Cx3cr1GFP/WT;Ccr2RFP/WT mice containedonly CX3CR1HiCCR2� resident brain microglia (Fig. 2A). Inaddition, there were no differences observed in the percentagesof the CD45Hi infiltrating cells when tumors from Cx3cr1GFP/WT;Ccr2RFP/WTmicewere comparedwith those generated inwild-typeB6 mice (Fig. 2B). Second, three distinct, but related monocytepopulations, were observed in the murine brain tumors as strat-

ified by their CX3CR1-GFP or CCR2-RFP expression intensities.The CX3CR1LoCCR2Hi (31.93 � 13.16% of total CD11bþ cells)and CX3CR1HiCCR2Lo (13.81 � 8.43%) populations demon-strated polarizing Ly6C expression (Fig. 2B). Surprisingly, adominant double-positive population (36.05 � 8.13%) wasfound (gated in yellow, Fig. 2B). These double-positive cellsexpressed intermediate/low levels of Ly6C, but high levels ofF4/80 (not shown), and correspond to the CD45HiLy6CLoF4/80Hi population in tumors generated in wild-type B6 mice(Fig. 1). However, the presence of CCR2 demonstrates thatthese cells are bone marrow–derived, rather than residentmicroglia that upregulated CD45 expression. Finally, the smallnumbers (9.83 � 1.42%) of CD45Lo microglia expressed abun-dant CX3CR1, as indicated by strong GFP signal with low tonegative Ly6C expression (Fig. 2B). Collectively, these datasupport the use of Cx3cr1GFP/WT and Ccr2RFP/WT knock-in miceto study the temporal and spatial dynamics of TAM infiltrationand function in murine GBM in vivo.

Figure 2.

Cx3cr1GFP/WT;Ccr2RFP/WT knock-inmice distinguish resident microgliafrom bone marrow–derivedmacrophages. Representative dotplots gated on CD11bþCD45þ cellsfrom na€�ve (A) and tumors (B)generated in Cx3cr1GFP/WT;Ccr2RFP/WT

mice. A, Magenta and blue circlesdelineate the CD11bþCD45Hi

(blood-derived monocytes andmacrophages) and CD11þCD45Lo

(resident brain microglia) populations.Their CX3CR1-GFP and CCR2-RFPprofiles are shown. B, TAMs wereidentified by their CD11b and CD45expression, and the CD45Hi andCD45Lo cells are further gated on RFP(CCR2) and GFP (CX3CR1) positivity.The CD45Hi population can bestratified into three related but distinctpopulations, whose Ly6C expression isexamined. The CD45low population(microglia) expressed high level ofCX3CR1-GFP, but little to noCCR2-RFP. Quantification of thesubpopulations of CD45Hi or CD45Lo

myeloid cells are described in theResults section. n ¼ 3 for na€�ve miceand n ¼ 5 tumor-bearing mice.

Chen et al.





Inflammatory monocytes infiltrate to GBM from the bloodcirculation

In terminally ill mice with GBM, the number of bonemarrow–derived cells from the circulation represents the great-est portion of the TAMs. To determine whether the number ofcirculating bone marrow–monocytes correlates with diseaseprogression, we analyzed whole blood of B6 and Cx3cr1GFP/WT;Ccr2RFP/WT mice 20 days posttumor cell transplantation (notumors at this time point) and at death. Quantification of thenumbers of monocytes and neutrophils show that there is asignificant decrease in Ly6CHi blood monocytes from tumor-bearing mice at end stage compared with 20 days posttumor celltransplantation (Supplementary Fig. S1). Identical results wereobserved in wild-type B6 and Cx3cr1GFP/WT;Ccr2RFP/WT mice. Thetotal CD11bþ myeloid population (data not shown) and neu-trophil numbers did not differ as a function of tumor develop-ment (Supplementary Fig. S1). Together, these data suggest thatdecreased number of blood monocytes is associated with theirincreased infiltration into GBM.

Bone marrow–derived monocyte/macrophages predominatewithin the GBM parenchyma, while microglia reside at thetumor periphery

The GBM microenvironment is compartmentalized intoanatomically-distinct regions, referred to as "tumor niches,"which contain various populations of stromal cells as well asglioma stem cells, the most resistant population to radio- andchemotherapy (11). In this regard, it is important to distinguishdifferences in the spatial localization of bone marrow–derivedmacrophages relative to resident microglia. To define the distri-bution of these TAM populations, tumor-bearing Cx3cr1GFP/WT;

Ccr2RFP/WTmouse brainswere immunostainedwithCD31 (endo-thelial cells) and RFP antibodies (due to the weakened signal oftheRFP reporter followingfixation) coupledwith the endogenousGFP reporter (less affected by fixation). Most of the GFPþ cellsinside of the tumor parenchyma were also positive for RFPexpression. Interestingly, these GFPþRFPþ double-positive bonemarrow–derived cells formed clusters around blood vessels(arrow, Fig. 3A). In agreement with our flow cytometry data,single GFPþ (RFP�) cells were observed in much lower numbersthan that of the double-positive cells (69.9� 7.9 vs. 216.6� 16.9,P < 0.01) inside the tumor proper (Fig. 3B).

In contrast, the myeloid cells in the peritumoral regionswere frequently only GFPþ, in particular outside of the tumorproper (dotted line demarcates the tumor). GFPþRFPþ double-positive cells were mostly observed within the tumor paren-chyma (Fig. 3A).

We also investigated the spatial distribution of these myeloidcells in anothermurine GBMmodel by implanting GL261 cells inCx3cr1GFP/WT;Ccr2RFP/WTmice (26). The results were similarbetween the two models with respect to the TAM cellular distri-bution within the tumors (Fig. 3C and D), where the majority ofTAMs within the tumor were derived from the bone marrow. Inthe peritumoral regions, the number of single GFPþ microglia inthe GL261model is similar to that observed in the PDGFB-drivenglioma (Fig. 3D). In addition, the number of RFPþGFPþ macro-phages was higher in GL261 model, although this difference didnot reach statistical significance (173 � 36.8 vs. 98.7 � 30.9, P ¼0.056 by t test).

To demonstrate that these double-positive cells infiltrated fromthe blood circulation, we generated tumors in Cx3cr1GFP/GFP;Ccr2RFP/WTmice.Wehavepreviously shown that tumors generated

Figure 3.

Bone marrow–derived monocyte/macrophages were predominantwithin the GBM. A, Brain tumorsections fromCx3cr1GFP/WT;Ccr2RFP/WT

mice were visualized with GFPreporter (green), staining with anti-RFP (red), and anti-CD31 (gray)antibodies, and counterstained withDAPI (blue). Representative imagesdemonstrate that the majority ofGFPþRFPþ cells are localized inperivascular areas (yellow arrow)within the tumor, whereas in theperitumoral areas, the majoritymyeloid cells are singleGFPþ cells. B, Quantification ofGFPþRFPþ double positive orGFPþ single positive cells. C,Immunofluorescence of GFP, RFP, andCD31 in and around GL261-inducedtumors. D, Quantification ofGFPþRFPþ double positive or GFPþ

single positive cells in GL261-inducedtumors. Scale bar, 50 mm. �� , P < 0.01by t test; n ¼ 3 for each group. Dottedlines mark tumor margin.






in this strain have significant infiltration of Ly6CHi inflammatorymonocytes from the blood circulation (27). Histologic examina-tion of the tumors generated in Cx3cr1GFP/GFP;Ccr2RFP/WT micerevealed the presence of a GFPþRFPþ population in the perivas-cular areas of the tumors, whereas single GFPþ cells were found inthe peritumoral locations (Supplementary Fig. S2).

Bone marrow–derived cells infiltrate early duringGBM formation

It is not known when bone marrow–derived cells begin toinfiltrate the tumor during GBM development. To gain insightsinto the temporal course of events, we transplanted primarytumor cells isolated from RCAS/Ntva donors into Cx3cr1GFP/WT;Ccr2RFP/WTmice, and euthanized the animals at 25 days postinjec-

tionwhenno tumor-induced neurologic symptomswere apparent(median survival, 56 days; Supplementary Table S1).Wewere ableto obtain tumors of various sizes in approximately half of themice,reflecting the spectrum of stages of tumor progression (Fig. 4A).Even in the smallest tumors generated (which are invisible on thesurface of the brain by gross examination; #1 in Fig. 4A), theCX3CR1-GFP/CCR2-RFP double positive cells inside the tumormass are evident and numerous (#1 in Fig. 4B). This distributionwas also observed in tumors of increasing size (Fig. 4B). When wequantified the number of these cells as a function of tumor size(Fig. 4C), we found no difference between small and large tumors,suggesting that bone marrow–derived cells infiltrate the tumormass early during GBMdevelopment, andmaintain their presencewith continued tumor growth. In na€�ve mice, RFPþ cells were not

Figure 4.

Bone marrow–derived cells infiltrateGBM early during initial tumorformation stage.A,Gross examinationof brains dissected from Cx3cr1GFP/WT;Ccr2RFP/WT mice euthanized 25 daysafter transplantation of tumorsisolated form RCAS/Ntva donors. B,Immunofluorescence of coronal brainsections containing tumors atdifferent stages as manifested bysmall to large sizes. Scale bar, 1 mm. C,Quantification of GFPþRFPþ doublepositive or GFPþ single positive cellsinside the tumors. One-way ANOVAtest does not detect statisticalsignificance between tumors ofvarious sizes. D, Representativeimmunofluorescent sections of large(top) and small (bottom) tumorsdissected from GL261 model.Scale bar, 100 mm.

Chen et al.





observed in the brain parenchyma, but only occasionally in themeninges (Supplementary Fig. S3).

Unlike PDGFB-driven GBM, multiple tumors of varioussizes/stages form in a single brain following the injection ofGL261 glioma cells. Similarly, we found that irrespective of thetumor sizes, CX3CR1-GFP/CCR2-RFP double positive cellsdominate the tumor, particularly in the perivascular spaces,whereas single GPFþ cells largely reside in the periphery of thetumor mass (Fig. 4D).

Bone marrow chimerism confirms that the majority TAMs areinfiltrating monocyte/macrophages

To further confirm the relative ratios of bone marrow–derivedcells to resident microglia within the GBM tumor mass, wegenerated bone marrow chimeras where bone marrow cells

derived from Cx3cr1GFP/WT;Ccr2RFP/WTmice were transplantedinto na€�ve wild-type B6 mice. In conjunction with Iba1 IHC,these chimeric mice allow us to unambiguously distinguishdonor-derived TAM as Iba1þGFPþ, while host microglia asIba1þGFP�. The procedure used to generate these bone mar-row–chimeric tumor-bearing mice is illustrated in Fig. 5A. Weused two irradiation regimens to ablate the host immune system,TBI, and head protected irradiation (HPI), because both regi-ments have inherent strengths and weaknesses. Surprisingly,when we examined the reconstitution rate of the chimeras treatedby HPI, we found that only approximately 50% of the leukocytesderived from the donor marrows (Supplementary Fig. S4).Because this low reconstitution rate yields a mixed populationof GFPþ and GFP� monocytes, precluding the chimeras frombeing used for meaningful evaluation, cellular quantification of

Figure 5.

Bonemarrow chimerism confirms thatthe majority TAMs are infiltratingmonocyte/macrophages. A,Schematic illustration of thegeneration ofCx3cr1GFP/WT;Ccr2RFP/WT

bonemarrowchimeras inCx3cr1WT/WT;Ccr2WT/WT (B6)mice followed by GBMimplantation. B, Representativeimages of the DAPI staining for normalbrain, tumor, and peritumoral regionsof glioma-bearing chimeric mice. C,Immunofluorescent staining for Iba1(magenta) and endogenous GFP(green) in normal brain tissue ortumors generated in chimeric mice.D, Quantification of Iba1þ and GFPþ

cells within the tumor or in theperitumoral regions. Scale bar,50 mm. �� , P < 0.01; n ¼ 5 each group.






tissue sections obtained from HPI-treated chimeric mice was notperformed.

We first examined the reconstitution efficiency of the bonemarrow following TBI treatment, and found that over 98% of theleukocytes are bone marrow–derived cells at the time of tumorimplantation (Supplementary Fig. S4). At death, tumors weredissected and analyzed by IHC. Tumor parenchyma and peritu-moral areas were identified by nuclei densities, as visualized byDAPI staining (Fig. 5B), and were used as regions-of-interest toquantify the numbers of Iba1þGFPþ (bone marrow–derived) andIba1þGFP� (resident microglia; Fig. 5C). In nontumor-bearingna€�ve chimeric mice, all of the Iba1þ cells were GFP negative,indicating their host origin. However, within the tumor parenchy-ma, the majority (82.8 � 3.8%) of the Iba1þ cells also expressedGFP, indicative of a bone marrow origin (Fig. 5D), in agreementwith flow cytometry analysis (Figs. 1 and 2). In striking contrast,only 45.3� 2.0% of the Iba1þ cells in the peritumoral regionwerebone marrow–derived, while the majority were resident microglia(Iba1þGFP�) with ramified morphologies (Fig. 5C and D). Occa-sionally, Iba1�GFPþ cells were observed in the tumor (Fig. 5C),which likely represent rare CX3CR1-expressing T cells (28).

RNA sequencing analysis identifies unique geneexpression profiles in infiltrating bone marrow–derivedcells versus microglia

Given their differential localizations, we sought to identify thegene expression profiles of these distinct populations. For thesestudies, bone marrow–derived or resident myeloid cells weresorted from both na€�ve and tumor-bearing Cx3cr1GFP/WT;Ccr2RFP/WT mice according to red (Ccr2) and green (Cx3cr1)fluorescence in combination with CD45 and CD11b expression,and their transcriptome profiles analyzed by RNA-Seq. We foundunique gene expression profiles from each of these four differentpopulations (na€�vemicroglia, na€�vemonocytes, tumor-associatedmicroglia, and tumor-associated macrophages) by pairwise com-parisons (Fig. 6A). Not surprisingly, substantial differences wereobserved between na€�ve microglia and na€�ve monocytes (1,964upregulated and 1,144 downregulated genes; Supplementary Fig.S5), consistentwith their distinctive origins (13, 29, 30).Whenwespecifically explored the DEGs between tumor-associated micro-glia and bone marrow–derived TAMs, we found that 530 geneswere upregulated in both populations, but 938 and 1,083 distinctgenes were upregulated in tumor-associated microglia or TAMs,respectively (Fig. 6B). Functional queries of these gene sets usingReactome FIViz Network analysis identified "cellular migration"as most enriched in TAMs, whereas genes associated with "proin-flammatory cytokines" and "metabolism" were enriched intumor-associated microglia (Fig. 6C). Pathways related to "cellproliferation" were significantly enriched in both populations(Fig. 6C). A complete list of these genes andpathways are includedin Supplementary Tables S2–S4. We selected several deferentiallyregulated genes in tumor-associated microglia compared to na€�vemicroglia (Fig. 6D), TAMs compared with na€�ve monocytes(Fig. 6E), and TAMs compared with tumor-associated microglia(Fig. 6F), and confirmed their RNA expression levels by qRT-PCR.

Heterozygous loss of Ccl2 prolongs the survival ofGBM-bearing mice

Previous studies had established that the CCL2/CCR2 axis isessential for monocyte migration into the inflamed CNS (31). Inthe setting ofmurine GBM, we have shown that neoplastic cells in

GBM express high levels of CCL2 (a.k.a. MCP-1), which contri-butes to the directional infiltration of CCR2Hi inflammatorymonocytes into the tumor (27). When we queried the humanTCGA database forCCL2 expression and divided the patients intohigh and low CCL2 cohorts, we found that the patients with lowtumoral CCL2 expression survived significantly longer than thosewith high tumoral CCL2 expression (Fig. 7A). Because CCR2Hi

bone marrow–derived cells account for over 85% of the TAMpopulation in mouse GBM (Fig. 2B), and supported by ourobservations in human GBM patients, we hypothesized thatheterozygous loss of the Ccl2 gene would reduce CCR2Hi inflam-matory monocyte infiltration and prolong the survival of GBM-bearing mice. To test this hypothesis,Ntvamice were intercrossedwith Ccl2�/� mice to generate Ntva;Ccl2þ/� double transgenicmice for RCAS-mediated PDGFB-driven tumor induction (Fig.7B). In agreement with a previous study (32), genetic Ccl2reduction extended the survival time of tumor-bearing Ntva;Ccl2þ/�mice relative toNtva;Ccl2þ/þ littermate controls (mediansurvival days are 69 and 55 respectively, P < 0.05; Fig. 7C). Toinvestigate the mechanism underlying this prolonged survivalconferred by Ccl2 reduction, we examined TAM accumulation,vasculature morphology and cell proliferation by histology (Sup-plementary Fig. S6A). Therewas a trend towards reduced numbersof Iba1þ macrophages in the tumors of Ntva;Ccl2þ/� mice com-pared with Ntva;Ccl2þ/þ mice (P ¼ 0.136; Supplementary Fig.S6B), in line with the previous study showing that myeloid-derived suppressor cells were significantly reduced in Ccl2�/�

mice in SB transposon-mediated de novo murine glioma model(32). However, there was no difference observed in total vesselarea or average vessel size between these two genotypes (Supple-mentary Fig. S6B). We also examined the number of proliferatingcells by phosphor-histone 3 and Ki67 staining, but did notobserve any difference in these terminal stage tumors (Supple-mentary Figs. S7 and S8). In addition, we determined whetherthere is difference in cell death betweenCcl2þ/� versusWTmice byTUNEL staining.We foundnodifference in cell death in the tumorproper (Supplementary Fig. S9). To differentiate the source ofCCL2 of being produced either by cancer cells or the stromal cells,we transplanted WT PDGFB-driven primary murine tumors intoCcl2þ/þ (WT) orCcl2þ/�mice. Kaplan–Meier analyses indicate nosurvival advantage when Ccl2 gene expression is diminished instromal cells (Supplementary Fig. S10), suggesting that the cancercells are the major producers of CCL2. We have previouslydemonstrated that increased levels of CCL2 expression is associ-ated with increased infiltration of monocytes and correlates withshorter survival of tumor-bearing mice (27). Together, these dataindicate that decreased level of CCL2 in neoplastic cells prolongssurvival time of tumor-bearing mice.

DiscussionMacrophages account for 30% to 50% of the tumor mass in

GBM (2, 8). However, whether these TAMs arise from peripheralmonocytes or CNS resident microglia remains unclear. Here, bytaking advantage of Cx3cr1GFP/WT;Ccr2RFP/WT double transgenicknock-in mice, we demonstrated that infiltrating macrophagesconstitute approximately 85% of the total TAM population, withresident microglia accounting for the remaining approximately15% of TAMs. Bone marrow–derived infiltrating cells preferen-tially localize to perivascular areas within the tumor proper,whereas the smaller population of CCR2� resident microglia are

Chen et al.





Figure 6.

RNA sequencing analysis identifies unique gene profiles between infiltrating macrophages and resident microglia. A, Venn diagram showing the numberof DEGs revealed by pairwise comparison (Log2 fold change � 2, P < 0.01) between na€�ve microglia (MG), tumor-associated microglia, na€�ve monocytes (MO), orTAMs (Mf). B, Venn diagram showing the number of DEGs (P � 0.01) between tumor-associated MG or Mf. C, Reactome FI Network analysis identifieddeferential biological functions specifically enriched in tumor-associated microglia, TAM, or in both populations. Numbers indicate significantly upregulated genescontributing to that specific biological function. qRT-PCR confirmed upregulation of selected molecules identified by RNAseq in either tumor-associatedmicroglia compared to na€�ve microglia (D), TAMs compared to na€�ve monocytes (E), and TAMs compared to tumor-associated microglia (F). � , P < 0.05;�� , P < 0.01; �� , P < 0.001 by t test; N ¼ 3 independent samples each group.






mainly localized to the peritumoral regions. By using RNA-sequencing analyses, we discovered differential gene expressionpatterns unique to infiltrating and resident cells, suggestingunique functions for each TAMpopulation related to their patho-biological attributes in tumor development. Given the abundanceof CCR2Himonocytes within the tumor, we demonstrate that lossof single copy of Ccl2 from both tumor and stromal cells pro-longed survival of tumor-bearing mice.

The importance of developing methods to distinguish micro-glia fromhematopoieticmyeloid cells becomes increasingly clear,with an ever-increasing number of studies revealing differentialroles of these monocytic cells in various CNS diseases (33, 34).Differential functions ofmicroglial cells canbepartially attributedto their unique origin from yolk sac progenitors (29) and the factthat they can maintain themselves by virtue of longevity and self-renewal (29, 35, 36). In this regard, resident microglia represent adistinct population of myeloid cells. Monocytes, however, orig-inate from hematopoietic stem cells, with series of progenitorsdifferentiating within the bonemarrow and ultimately released tothe blood circulation to colonize certain peripheral organs underboth normal and inflammatory conditions (37).

In the context of GBM, approaches to distinguishing microgliafrom invading monocytes have traditionally relied on the use ofCD45 antibodies to separate resident microglia (CD45Lo) frommacrophages of hematopoietic origin (CD45Hi) by FACS analysis(8). Analysis of human glioma samples has revealed that theCD45Hi population is greater than the CD45Lo population, sug-gesting that gliomas contain more recruited monocytes thanmicroglia (10). This concept was recently challenged by a studyusing irradiation mouse chimeras, which demonstrated that themajority of TAMs are intrinsicmicroglia, and that thesemicroglialcells upregulate their CD45 expression to constitute a significantproportion of the CD45Hi population in gliomas (9). The dis-crepancy between our results and those of Muller and colleaguesmaybe explainedby thedifferences in genetic compositions of theanimal models used, as was initially observed and reported byBadie and Schartner (8). It is also likely that the presence ofmicroglia in TAMswas overestimated byMuller and colleagues, asthey analyzed tumor-bearing hemispheres rather than the micro-dissected tumor proper. In this respect, tumor-bearing hemi-spheres contain large numbers of microglia that reside outsideof the tumor mass (Figs. 3–5). Although both microglia andinfiltrating monocytes can upregulate their CD45 and CD11bexpressions at the height of tumor development, there remains aclear demarcation in CD45 and CD11b intensity between thesetwo populations, as visualized by flow cytometry (Fig. 2B).

Similarly, another study used single staining with antibodiesagainst either CX3CR1 or CCR2 to conclude that the majorityof TAMs were monocyte-derived (CX3CR1�CCR2þ) macro-phages (38). However, the prominent CX3CR1 and CCR2 dou-ble-positive population was not considered. It should be notedthat CX3CR1 is expressed by both blood monocytes and micro-glia, and can be increased during monocyte differentiation intomacrophages. These findings argue against the use of CX3CR1alone as a microglia-specific marker either in the na€�ve mouse(12, 13) or in the context of glioma (27). In addition, no widelyavailable and well-validated antibodies to CX3CR1 or CCR2 existto discriminate microglia, monocytes, and monocyte-derivedmacrophages. Approaches, like those taken to identify Tmem119(transmembrane protein 119), may reveal potential antibodiesthat selectively distinguish microglia from bone marrow–derivedmonocytes (39). Together, the discrepant results obtained fromthe use of bone marrow chimeras and standardly employed cellsurface antibodies highlight the urgent need to reevaluate thesepublished conclusions inmyeloid cell distribution inGBMand toperform lineage-tracing experiments using reporter mice thataccurately distinguish microglia from monocytes/macrophagesin GBM.

UsingCx3cr1GFP/WT;Ccr2RFP/WT double knock-inmice, we dem-onstrated that during tumor development, bonemarrow–derivedinfiltrating cells dominate the TAM landscape (Figs. 2 and 4). Thefact that they preferentially locate to perivascular regions suggeststhat they extravasate through blood vessels within the tumor. Incontrast, resident microglia do not migrate easily through braintissue (40) to the interior of the tumor and therefore are largelyobserved in the peritumoral spaces. It is also interesting to notethat the intensity of both CX3CR1-GFP and CCR2-RFP variesconsiderably among the TAMs (Fig. 2). It is well documented thatCCR2Hi cells are monocytes newly arrived at the site of inflam-mation (41). Once homed to inflamed tissues, these cells grad-ually downregulate their CCR2 expression as they differentiateinto macrophages (41). Here, we demonstrated that the TAMsexhibit a broad range of CCR2-RFP intensities, indicating acontinuous transformation of these cells from infiltrating mono-cytes to mature macrophages. Remarkably, the intensityof CX3CR1-GFP in these cells also varied, and inversely correlatedto CCR2-RFP expression (Fig. 2). This dynamic transition ofthe surface markers (Supplementary Video S1) indicates thatbone marrow–derived macrophages are highly plastic and thatthese cells evolve to maturation in situ following extravasation(24), although our data cannot entirely exclude the possibilitythat the small CD45HiCCR2�CX3CR1Hi population infiltrated

Figure 7.

Heterogeneous loss of CCL2, the CCR2ligand, extends survival of GBM-bearingmice. A, Survival curve of GBM patientsstratified according to their CCL2expression queried from the TCGAdatabank. B, Diagram illustrating thegeneration of tumors in Ccl2þ/� miceusing RCAS/Ntva system. C, Survivalcurve of tumor-bearing Ntvaþ mice. n isdenoted in the figures. � , P < 0.05 byMantel–Cox test.

Chen et al.





directly from the blood circulation. Additional linage tracingstudies will be necessary to establish the ontology of theseCD45HiCCR2�CX3CR1Hi cells in GBM.

Mature macrophages derived in situ from infiltrating mono-cytes play important roles in various inflammatory diseases andtherefore stand as suitable targets for therapeutic interventions(42). Our gene profiling experiments revealed that pathwaysinvolved in cell migration were significantly and specificallyenriched in bone marrow–derived TAMs (Fig. 6). CCL2, amember of the MCP (monocyte chemoattractant protein) che-mokine family, plays an important role in mediating monocytemigration through its receptor CCR2. It is interesting to notethat low CCL2 expression in human GBM is associated withsignificantly prolonged patient survival (Fig. 7). Together, thesefindings raise the question as to whether reducing monocyteinfiltration by targeting CCL2–CCR2 axis is a viable option fortreating murine GBM. To address this question, we showed thatgenetically interrupting the CCL2–CCR2 axis prolonged thesurvival of GBM-bearing mice, in agreement with previouspharmacologic studies (43, 44). However, in contrast to thepromising preclinical studies, neutralizing mAbs against CCL2administered to patients with metastatic, solid tumors did notproduce favorable outcomes. Meta-analysis of the data fromthese clinical trials indicated that initial CCL2 inhibition mayhave unexpectedly caused subsequent increases in circulatingCCL2 levels, possibly due to a compensatory feedback loop(45). Lack of therapeutic benefits from inhibiting this axisindicates that CCL2–CCR2 interaction represents a complexsignaling network that is not well understood. Since the reduc-tion of CCL2 did not result in statistically significant decrease ofTAM infiltration, it also implies that other MCP family che-mokines likely function in synergy with CCL2 to recruit mono-cytes into GBM. In this regard, MCP-3 (CCL-7) had been shownto play a critical role in recruiting bone marrow–derived mono-cytes to sites of inflammation (46).

Although previous studies have revealed that bone marrow–derived cells penetrate experimental murine GBM tumors, to ourknowledge, this is the first report to provide conclusive evidencedemonstrating that tumor-infiltrating bone marrow cells arerecruited early during GBM development in at least two indepen-dent murine models (Fig. 4). CCR2-RFPþ cells are found inclusters around blood vessels and spread across the entire tumormass at all stages of tumorigenesis. This novel observation impliesthat bone marrow–derived TAMs interact with neoplastic cells inthe perivascular niche to promote tumor growth from the initialstage of tumor development. In light of this notion, it is plausiblethat the prolonged survival of the Ntva;Ccl2þ/� mice reflectsdelayed tumor initiation as a result of diminished initial TAMinfiltration due to Ccl2 reduction in tumor cells (Fig. 7).

In summary, we establish that the majority of GBM-associatedmacrophages are bonemarrow–derived infiltratingmyeloid cells.These cells are recruited early during tumor formation, preferen-tially localize to the perivascular niche, and contribute to tumordevelopment. Reducing their infiltration by geneticCcl2 (MCP-1)modulation significantly prolongs the survival of GBM-bearingmice. Future studies investigating members of MCP family che-mokines have potential to elucidate the contributions of thesestromal cells to tumor maintenance and may one day yieldeffective stroma-directed therapies.

Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed.

Authors' ContributionsConception and design:Z. Chen, X. Feng, D.H.Gutmann, D. HambardzumyanDevelopment of methodology: Z. Chen, X. Feng, D. HambardzumyanAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): Z. Chen, X. Feng, C.J. Herting, V. Alvarez Garcia,K. Nie, W.W. Pong, S.A. Wolf, D. HambardzumyanAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): Z. Chen, X. Feng, C.J. Herting, V. Alvarez Garcia,K. Nie, R. Rasmussen, B. Dwivedi, S. Seby, D.H. Gutmann, D. HambardzumyanWriting, review, and/or revisionof themanuscript:Z.Chen,X. Feng,C.J.Herting,V. Alvarez Garcia, W.W. Pong, B. Dwivedi, D.H. Gutmann, D. HambardzumyanAdministrative, technical, or material support (i.e., reporting or organizingdata, constructing databases): Z. ChenStudy supervision: D. Hambardzumyan

AcknowledgmentsThe authors are grateful toDr. ChristopherNelson for editorial assistance.We

thank the FlowCytometryCores at theCCF and theCHOA for technical services.We acknowledge the excellent imaging assistance from Dr. Neil Anthony, andwe thank Ms. Jennifer Powers for cell sorting. We are grateful to the Biostatisticsand Bioinformatics services at Washington University and Emory University.The authors also thank Dr. Richard Ransohoff for providing the Cx3cr1GFP/WT;Ccr2RFP/WT mice and Dr. Helmut Kettenmann for providing GL261 tumorsamples.

Grant SupportThis research project was supported in part by the Emory University

Integrated Cellular Imaging Microscopy Core of the EmoryþChildren'sPediatric Research Center. This work was also supported by NCIgrants U01CA160882 (D. Hambardzumyan and D.H. Gutmann) andR01CA195692 (D.H. Gutmann).

The costs of publication of this articlewere defrayed inpart by the payment ofpage charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Received August 24, 2016; revised October 4, 2016; accepted February 21,2017; published OnlineFirst February 24, 2017.

References1. Becher OJ, Hambardzumyan D, Fomchenko EI, Momota H, Mainwar-

ing L, Bleau AM, et al. Gli activity correlates with tumor grade inplatelet-derived growth factor-induced gliomas. Cancer Res 2008;68:2241–9.

2. Charles NA, Holland EC, Gilbertson R, Glass R, Kettenmann H. The braintumor microenvironment. Glia 2012;60:502–14.

3. Simmons GW, PongWW, Emnett RJ, White CR, Gianino SM, Rodriguez FJ,et al. Neurofibromatosis-1 heterozygosity increases microglia in a spatiallyand temporally restricted pattern relevant tomouse optic glioma formationand growth. J Neuropathol Exp Neurol 2011;70:51–62.

4. Solga AC, Pong WW, Kim KY, Cimino PJ, Toonen JA, Walker J, et al. RNAsequencing of tumor-associated microglia reveals Ccl5 as a stromal che-mokine critical for neurofibromatosis-1 glioma growth. Neoplasia2015;17:776–88.

5. Hambardzumyan D, Gutmann DH, Kettenmann H. The role of microgliaand macrophages in glioma maintenance and progression. Nat Neurosci2016;19:20–7.

6. Pyonteck SM, Akkari L, Schuhmacher AJ, Bowman RL, Sevenich L, QuailDF, et al. CSF-1R inhibition alters macrophage polarization and blocksglioma progression. Nat Med 2013;19:1264–72.






7. Butowski N, Colman H, De Groot JF, Omuro AM, Nayak L, Wen PY, et al.Orally administered colony stimulating factor 1 receptor inhibitorPLX3397 in recurrent glioblastoma: an Ivy Foundation Early Phase ClinicalTrials Consortium phase II study. Neuro Oncol 2016;18:557–64.

8. Badie B, Schartner JM. Flow cytometric characterization of tumor-associ-ated macrophages in experimental gliomas. Neurosurgery 2000;46:957–61.

9. Muller A, Brandenburg S, Turkowski K, Muller S, Vajkoczy P. Residentmicroglia, and not peripheral macrophages, are the main source of braintumor mononuclear cells. Int J Cancer 2015;137:278–88.

10. Parney IF, Waldron JS, Parsa AT. Flow cytometry and in vitro analysis ofhuman glioma-associated macrophages. Laboratory investigation. J Neu-rosurg 2009;110:572–82.

11. Hambardzumyan D, Bergers G. Glioblastoma: defining tumor niches.Trends Cancer 2015;1:252–65.

12. Geissmann F, Jung S, Littman DR. Blood monocytes consist of twoprincipal subsets with distinct migratory properties. Immunity 2003;19:71–82.

13. Yona S, Kim KW,Wolf Y, Mildner A, Varol D, Breker M, et al. Fate mappingreveals origins and dynamics of monocytes and tissue macrophages underhomeostasis. Immunity 2013;38:79–91.

14. Hambardzumyan D, Amankulor NM, Helmy KY, Becher OJ, Holland EC.Modeling adult gliomas using RCAS/t-va technology. Transl Oncol 2009;2:89–95.

15. Jung S, Aliberti J, Graemmel P, Sunshine MJ, Kreutzberg GW, Sher A, et al.Analysis of fractalkine receptor CX(3)CR1 function by targeted deletionand green fluorescent protein reporter gene insertion. Mol Cell Biol2000;20:4106–14.

16. Cook DN, Chen SC, Sullivan LM, Manfra DJ, Wiekowski MT, Prosser DM,et al. Generation and analysis of mice lacking the chemokine fractalkine.Mol Cell Biol 2001;21:3159–65.

17. Franklin KBJ, Paxinos G. The mouse brain in steriotaxic coordinates.Cambridge: Academic Press; 1997.

18. Lu B, Rutledge BJ, Gu L, Fiorillo J, Lukacs NW, Kunkel SL, et al. Abnor-malities in monocyte recruitment and cytokine expression in monocytechemoattractant protein 1-deficient mice. J Exp Med 1998;187:601–8.

19. Chen Z, Jalabi W, Hu W, Park HJ, Gale JT, Kidd GJ, et al. Microglialdisplacement of inhibitory synapses provides neuroprotection in the adultbrain. Nat Commun 2014;5:4486.

20. WuG, Feng X, Stein L. A human functional protein interaction network andits application to cancer data analysis. Genome Biol 2010;11:R53.

21. Smoot ME, Ono K, Ruscheinski J, Wang PL, Ideker T. Cytoscape 2.8: newfeatures for data integration and network visualization. Bioinformatics2011;27:431–2.

22. Mildner A, Schmidt H, Nitsche M, Merkler D, Hanisch UK, Mack M, et al.Microglia in the adult brain arise from Ly-6ChiCCR2þ monocytes onlyunder defined host conditions. Nat Neurosci 2007;10:1544–53.

23. Carson MJ, Bilousova TV, Puntambekar SS, Melchior B, Doose JM, EthellIM. A rose by any other name? The potential consequences of microglialheterogeneity during CNS health and disease. Neurotherapeutics2007;4:571–9.

24. Dal-Secco D, Wang J, Zeng Z, Kolaczkowska E, Wong CH, Petri B, et al. Adynamic spectrumofmonocytes arising from the in situ reprogramming ofCCR2þmonocytes at a site of sterile injury. J Exp Med 2015;212:447–56.

25. Mizutani M, Pino PA, Saederup N, Charo IF, Ransohoff RM, Cardona AE.The fractalkine receptor but not CCR2 is present on microglia fromembryonic development throughout adulthood. J Immunol 2012;188:29–36.

26. Ausman JI, Shapiro WR, Rall DP. Studies on the chemotherapy of exper-imental brain tumors: development of an experimental model. Cancer Res1970;30:2394–400.

27. Feng X, Szulzewsky F, Yerevanian A, Chen Z, Heinzmann D, RasmussenRD, et al. Loss of CX3CR1 increases accumulation of inflammatory mono-cytes and promotes gliomagenesis. Oncotarget 2015;6:15077–94.

28. Cardona AE, Pioro EP, Sasse ME, Kostenko V, Cardona SM, Dijkstra IM,et al. Control of microglial neurotoxicity by the fractalkine receptor. NatNeurosci 2006;9:917–24.

29. Ginhoux F, Greter M, Leboeuf M, Nandi S, See P, Gokhan S, et al. Fatemapping analysis reveals that adultmicroglia derive fromprimitivemacro-phages. Science 2010;330:841–5.

30. Butovsky O, Jedrychowski MP, Moore CS, Cialic R, Lanser AJ, Gabriely G,et al. Identification of a unique TGF-beta-dependent molecular and func-tional signature in microglia. Nat Neurosci 2014;17:131–43.

31. Boring L, Gosling J, Chensue SW, Kunkel SL, Farese RVJr, Broxmeyer HE,et al. Impaired monocyte migration and reduced type 1 (Th1) cytokineresponses in C-C chemokine receptor 2 knockout mice. J Clin Invest1997;100:2552–61.

32. Fujita M, Kohanbash G, Fellows-Mayle W, Hamilton RL, Komohara Y,Decker SA, et al. COX-2 blockade suppresses gliomagenesis by inhibitingmyeloid-derived suppressor cells. Cancer Res 2011;71:2664–74.

33. Ajami B, Bennett JL, Krieger C, McNagny KM, Rossi FM. Infiltratingmonocytes trigger EAE progression, but do not contribute to the residentmicroglia pool. Nat Neurosci 2011;14:1142–9.

34. Shemer A, Jung S. Differential roles of resident microglia and infiltratingmonocytes in murine CNS autoimmunity. Semin Immunopathol 2015;37:613–23.

35. Ajami B, Bennett JL, Krieger C, Tetzlaff W, Rossi FM. Local self-renewal cansustain CNS microglia maintenance and function throughout adult life.Nat Neurosci 2007;10:1538–43.

36. Elmore MR, Najafi AR, Koike MA, Dagher NN, Spangenberg EE, Rice RA,et al. Colony-stimulating factor 1 receptor signaling is necessary formicroglia viability, unmasking a microglia progenitor cell in the adultbrain. Neuron 2014;82:380–97.

37. Shi C, Pamer EG. Monocyte recruitment during infection and inflamma-tion. Nat Rev Immunol 2011;11:762–74.

38. Zhou W, Ke SQ, Huang Z, Flavahan W, Fang X, Paul J, et al. Periostinsecreted by glioblastoma stem cells recruits M2 tumour-associated macro-phages and promotes malignant growth. Nat Cell Biol 2015;17:170–82.

39. Bennett ML, Bennett FC, Liddelow SA, Ajami B, Zamanian JL, Fernhoff NB,et al. New tools for studyingmicroglia in themouse and human CNS. ProcNatl Acad Sci U S A 2016;113:E1738–46.

40. DavalosD,Grutzendler J, YangG,Kim JV, ZuoY, Jung S, et al. ATPmediatesrapid microglial response to local brain injury invivo. Nat Neurosci2005;8:752–8.

41. Gordon S, Taylor PR. Monocyte and macrophage heterogeneity. Nat RevImmunol 2005;5:953–64.

42. Getts DR, Terry RL, Getts MT, Deffrasnes C, Muller M, van Vreden C, et al.Therapeutic inflammatory monocyte modulation using immune-modify-ing microparticles. Sci Transl Med 2014;6:219ra7.

43. Zhu X, Fujita M, Snyder LA, Okada H. Systemic delivery of neutralizingantibody targeting CCL2 for glioma therapy. J Neurooncol 2011;104:83–92.

44. Chang AL, Miska J, Wainwright DA, Dey M, Rivetta CV, Yu D, et al. CCL2produced by the gliomamicroenvironment is essential for the recruitmentof regulatory T cells and myeloid-derived suppressor cells. Cancer Res2016;76:5671–82.

45. Lim SY, Yuzhalin AE, Gordon-Weeks AN, Muschel RJ. Targeting the CCL2-CCR2 signaling axis in cancer metastasis. Oncotarget 2016;7:28697–710.

46. Tsou CL, Peters W, Si Y, Slaymaker S, Aslanian AM, Weisberg SP, et al.Critical roles for CCR2 and MCP-3 in monocyte mobilization from bonemarrow and recruitment to inflammatory sites. J Clin Invest 2007;117:902–9.


Chen et al.




2017;77:2266-2278. Published OnlineFirst February 24, 2017.Cancer Res Zhihong Chen, Xi Feng, Cameron J. Herting, et al. in GlioblastomaCellular and Molecular Identity of Tumor-Associated Macrophages

Updated version

10.1158/0008-5472.CAN-16-2310doi:

Access the most recent version of this article at:

Material

Supplementary

http://cancerres.aacrjournals.org/content/suppl/2017/02/24/0008-5472.CAN-16-2310.DC1

Access the most recent supplemental material at:

Cited articles

http://cancerres.aacrjournals.org/content/77/9/2266.full#ref-list-1

This article cites 45 articles, 12 of which you can access for free at:

Citing articles

http://cancerres.aacrjournals.org/content/77/9/2266.full#related-urls

This article has been cited by 20 HighWire-hosted articles. Access the articles at:

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

Subscriptions

Reprints and

[email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/77/9/2266To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/lookup/doi/10.1158/0008-5472.CAN-16-2310

http://cancerres.aacrjournals.org/content/suppl/2017/02/24/0008-5472.CAN-16-2310.DC1

http://cancerres.aacrjournals.org/content/77/9/2266.full#ref-list-1

http://cancerres.aacrjournals.org/content/77/9/2266.full#related-urls

http://cancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://cancerres.aacrjournals.org/content/77/9/2266


Cellular and Molecular Identity of Tumor- …...Microenvironment and Immunology Cellular and...

Documents

Transcript of Cellular and Molecular Identity of Tumor- …...Microenvironment and Immunology Cellular and...