EGFR Cooperates with EGFRvIII to Recruit Macrophages in ... · Tumor Biology and Immunology EGFR...

Tumor Biology and Immunology

EGFR Cooperates with EGFRvIII to RecruitMacrophages in GlioblastomaZhenyi An1, Christiane B. Knobbe-Thomsen2, Xiaohua Wan1, Qi Wen Fan1,Guido Reifenberger2, and William A.Weiss1,3,4

Abstract

Amplification of the EGFR gene and its truncationmutant EGFRvIII are hallmarks of glioblastoma. Althoughcoexpression of EGFR and EGFRvIII confers a growthadvantage, how EGFR and EGFRvIII influence the tumormicroenvironment remains incompletely understood.Here, we show that EGFR and EGFRvIII cooperate toinduce macrophage infiltration via upregulation of thechemokine CCL2. EGFRvIII was significantly enrichedin glioblastoma patient samples with high CCL2, andknockout of CCL2 in tumors coexpressing EGFR and

EGFRvIII led to decreased infiltration of macrophages.KRAS was a critical signaling intermediate for EGFR-and EGFRvIII-induced expression of CCL2. Our resultsillustrate how EGFR and EGFRvIII direct the microenvi-ronment in glioblastoma.

Significance: Full-length EGFR and truncated EGFRvIIIwork through KRAS to upregulate the chemokine CCL2and drive macrophage infiltration in glioblastoma. Cancer Res;78(24); 6785–94. �2018 AACR.

IntroductionGlioblastoma is the most common primary brain tumor and

among themost aggressive of all cancers (1). Average survival afterdiagnosis is only 1 year, even in the setting of extensive chemo-therapy and radiotherapy (1). Amplification of the EGFR isidentified in more than half of tumors. One third of EGFR-amplified tumors also coamplify EGFRvIII, a truncation mutantof EGFR (2). Amplification of EGFRvIII always occurs in tumorsthat also amplify EGFR, suggesting that coamplification of thosemolecules confers a growth advantage (3). Our lab previouslyshowed that (1) EGFRvIII and EGFR cooperated to promoteglioblastoma growth both in vitro and in vivo, (2) EGFRvIII wasa substrate of EGFR, and (3) phosphorylation of EGFRvIII byEGFR was critical for tumorigenesis and growth (3). EGFR andEGFRvIII cooperated to promote tumor growth more dramati-cally in vivo than in vitro, suggesting that tumor-extrinsic factorscontribute to tumor growth (3).

The tumor microenvironment is a critical factor in initiationand progression of glioblastoma (4, 5). EGFRvIII potentiates theexpression of IL6, IL8, and LIF, indicating that EGFRvIII signals to

the gliomamicroenvironment through upregulating specific cyto-kines (6–8). Interestingly, EGFRvIII promotes the growth ofcells expressing wild-type EGFR via a paracrine IL6 and/or LIF-dependent manner (8, 9). However, how EGFR and EGFRvIIIconverge on cytokines to regulate immune cell infiltrationremains incompletely understood.

Cytokines contribute to infiltration of immune cells includingtumor-associated macrophages (TAM). Macrophages representup to half of the glioblastoma tumor mass, with the C-C chemo-kine CCL2 important for macrophage infiltration (10, 11). Levelsof CCL2 are elevated in glioblastoma and drive progression bypromoting immunosuppression (12, 13). In this study, we dem-onstrate that EGFR and EGFRvIII cooperate through KRAS, toupregulate CCL2, promoting infiltration of macrophages.

Materials and MethodsCell culture

U87 cells expressing control vector, EGFR, EGFRvIII, andEGFR/EGFRvIII were described previously (3). Cells were main-tained inDMEM supplemented with 10%FBS and 1� penicillin–streptomycin. A172 cells were purchased in 2017 (ATCC) andauthenticated by the ATCC using short tandem repeat (STR)profiling. Cells were certifiedMycoplasma-free by the ATCC. Cellswere transducedwith EGFRandEGFRvIII lentiviruses as describedpreviously (3). A172 cells expressing control vector, EGFR, EGFR-vIII, and EGFR/EGFRvIII were maintained in DMEM supplemen-ted with 10% FBS and 1� penicillin–streptomycin. The cells werepassaged less than 15 times after thawing. GBM6 and GBM39were from Dr. Jann Sarkaria at Mayo Clinic (Rochester, MN). Thexenografted tumors were dissociated as described (14). The cellswere grown in DMEM:F12 with 20 ng/mL EGF, 20 ng/mL FGF,1xB27 supplement, and 1� penicillin–streptomycin. For differ-entiation, the cellswere grown inDMEMsupplementedwith 10%FBS and 1� penicillin–streptomycin. The cells were passaged lessthan 5 times after thawing. For cytokine-related assays, cells were

1Department ofNeurology,University of California, SanFrancisco, San Francisco,California. 2Department of Neuropathology, Heinrich Heine University,D€usseldorf, Germany. 3Helen Diller Family Comprehensive Cancer Center,University of California, San Francisco, San Francisco, California. 4Departmentsof Pediatrics and Neurological Surgery, University of California, San Francisco,San Francisco, California.

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

Corrected online September 10, 2019.

CorrespondingAuthor:WilliamA.Weiss, University of California, San Francisco,1450 3rd St, San Francisco, CA 94158. Phone: 415-502-1694; Fax: 415-476-0133;E-mail: [email protected]

doi: 10.1158/0008-5472.CAN-17-3551

�2018 American Association for Cancer Research.

CancerResearch

www.aacrjournals.org 6785

on July 25, 2020. © 2018 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst November 6, 2018; DOI: 10.1158/0008-5472.CAN-17-3551





http://crossmark.crossref.org/dialog/?doi=10.1158/0008-5472.CAN-17-3551&domain=pdf&date_stamp=2019-9-9

http://cancerres.aacrjournals.org/



grown for 24 hours in DMEM:F12 with 20 ng/mL EGF, 20 ng/mLFGF, 1xB27 supplement, and 1� penicillin–streptomycin.MV-4-11 cells were purchased in 2015 (ATCC) and authenti-cated by the ATCC using STR profiling. Cells were certifiedMycoplasma-free by the ATCC. These cells were cultivated inIMDM supplemented with 10% FBS at 37�C and 5% CO2. Thecells were passaged less than 5 times after thawing. Mycoplasmastatus was monitored monthly in the lab using the HEK-Bluedetection kit (InvivoGen).

Transwell macrophage migration assayThe assaywas carriedout usingQCMChemotaxis 5mm24-Well

Cell Migration Assay Kit (Millipore) following themanufacturer'sinstructions. Specifically, 1�106macrophage cells (MV-4-11) perwell were plated on transwell insert in 250 mL IMDMwithout FBS.Bottom chamber was filled with conditioned media (filteredthrough 0.45 mm filters) from equal numbers of indicated cells.After 4 hours, cells in the bottom chambers were lysed, and datawere obtained by reading fluorescence 480/520nm as in themanufacturer's protocol. Results were normalized to medium-only controls.

CRISPR knockout or siRNA knockdown of indicated genesTo generate CCL2 or KRAS CRISPR knockout, a pool of 3

different gRNA plasmids were used (Santa Cruz Biotechnology).The sequences for the CCL2 gRNAs were: 50-30: TACCTGGCT-GAGCGAGCCCT; 50-30: TTTGGGTTTGCTTGTCCAGG; 50-30:TCAGGATTCCATGGACCACC. The sequences for the KRASgRNAs were: 50-30: TCTCGACACAGCAGGTCAAG; 50-30: CAAT-GAGGGACCAGTACATG; 50-30: TTCTCGAACTAATGTATAGA.The corresponding HDR plasmids were cotransfected followingthe manufacturer's instructions (Santa Cruz Biotechnology). Thecontrol siRNAandKRAS siRNAwere purchased fromDharmacon.

In vivo studiesIndicated cell lines were injected into the brain of 4- to 6-week-

old female BALB/C nu/nu mice. Specifically, mice were anesthe-tized using ketamine and xylazine. Note that 3 � 105 cells wereinjected intracranially at 2 mm anterior and 1.5 mm lateral of theright hemisphere relative to bregma at a depth of 3 mm. Eachgroup contains 10 mice. All mouse experiments were performedfollowing protocols approved by UCSF's Institutional AnimalCare and Use Committee.

IHC stainingExcised tumors from mice were fixed with 10% neutral-buff-

ered formalin for 24 hours and changed to 70% ethanol. Tumorswere paraffin-embedded and sectioned by the Brain TumorResearch Center at UCSF. For IHC, slides were deparaffinized,and antigen retrieval was performed using a pressure cooker. TheMouse on Mouse (M.O.M.) basic kit (Vector laboratories) wasused for masking endogenous mouse antigen. The VECTASTAINABC reagent (Vector laboratories) was used for signal detection.The CD45 antibody (05-1410) was purchased from EMD Milli-pore and was used at a concentration of 1:50. The CD38(ab216343), CD68 (ab955), and TMEM119 (ab209064) anti-bodies were purchased from Abcam and were used at a concen-tration of 1:100. TheCD206 antibody (AF2535SP)was purchasedfrom R&D Systems and was used at a concentration of 1:50.Images were taken using a Zeiss AxioImagerM1microscope. Cellswere quantified using Image J.

Immunofluorescent stainingFormalin-fixed paraffin sections from primary grade IV

human glioblastoma tumor tissues were deparaffinized inxylene and rehydrated over a graded ethanol series, followedby treatment in 10 mmol/L citrate buffer at pH 6.0 for 20minutes in a steamer for antigen retrieval. Tissue sections wereequilibrated in TBST and then blocked for 10 minutes withUltravision Protein Block (Thermo Scientific) and incubatedovernight at 4�C with primary antibodies against EGFR (mousemonoclonal DAK-H1-WT, Dako, 1:50), EGFRvIII (rabbit poly-clonal 6549, Celldex, 1:100), and CD45 (rat monoclonalSM1744P, Acris, 1:100), diluted in Dako REAL Antibody Dil-uent. Three washing steps in TBST were followed by incubationwith secondary antibodies at room temperature for 1 hourdiluted in Dako REAL Antibody Diluent (anti-mouse AF488,anti-rabbit AF594, anti-rat AF647, Thermo Fisher, 1:1,000).Three washing steps in TBST were followed by incubation ofthe slides with DAPI diluted in PBS to a final concentration of1 mg/mL for 5 minutes at room temperature. Following repeat-ed washing, stained sections were mounted in ProLong Goldantifade reagent (Thermo Fisher) and sealed using nail polish.Fluorescent images were taken using a Leica TCS SP8 Stedmicroscope with either a 100x oil immersion objective or a40x water immersion objective. Taken images were furtherprocessed using the Leica Las X software, Adobe Photoshop,and Adobe InDesign.

Real-time PCRRNA was extracted using the Quick-RNA MiniPrep Kit accord-

ing to the manufacturer's instructions (ZYMO Research). cDNAwas produced using the iScript cDNA synthesis Kit (Biorad). Real-time PCR was performed using SYBR FAST ABI Prism qPCR kit(KAPA Biosystems) on an AB7900HT real-time PCR machine(Applied Biosystems). The primers for CCL2 were as follows:50-30 CAGCCAGATGCAATCAATGCC; 50-30 TGGAATCCTGAA-CCCACTTCT. IL1B: 50-30 AATCTGTACCTGTCCTGCGTGTT; 50-30 TGGGTAATTTTTGGGATCTACACTCT. Actin: 50-30 CATG-TACGTTGCTATCCAGGC; 50-30 CTCCTTAATGTCACGCACGAT.CXCR4: 50-30 GATCAGCATCGACTCCTTCA; 50-30 GGCTCCAAG-GAAAGCATAGA. CXCL12: 50-30 ATGCCCATGCCGATTCTTCG;50-30 GCCGGGCTACAATCTGAAGG. LIF: 50-30 CAGTGCCAA-TGCCCTCTTTAT; 50-30 GGCCACATAGCTTGTCCAGG. CD90/THY1: 50-30 CGCTCTCCTGCTAACAGTCTT; 50-30 CAGGCT-GAACTCGTACTGGA. SOX1: 50-30 GCGGAAAGCGTTTTCTTG;50-30 TAATCTGACTTCTCCTCCC.

Western blottingCellswere lysedusingRIPAbufferwith cOmpleteMini protease

inhibitor cocktail (Roche) and PhosSTOP (Roche). The IkB(9242) and p-IkB (9246) antibodies were purchased from CellSignaling Technology and used at a concentration of 1:500. TheActin antibody (sc-1615) was purchased from Santa Cruz Bio-technology and used at a concentration of 1:1,000. The KRASantibody (SAB2101297) was purchased from Sigma and used at aconcentration of 1:500.

Cell surface EGFR/EGFRvIII staining and FACS sortingCells were dissociated using Accutase (Innovative Cell Tech-

nologies), washed with cell staining buffer with sodium azide(BioLegend), and stained with an Alexa Fluor 488 anti-humanEGFR antibody (AY13, BioLegend) for 20 minutes. Cells were

An et al.

Cancer Res; 78(24) December 15, 2018 Cancer Research6786




sorted using a FACSAria3 sorter (BD) at the UCSF Laboratory forCell Analysis core facility. Approximately 20,000 cells were sortedfor each population.

Cytokine arrayCytokine levels in conditioned medium from equal number of

indicated cell lines were measured using Human Cytokine ArrayPanel A (R&D Systems #ARY005) following the manufacturer'sinstructions.

RNA sequencingRNA was extracted using Quick-RNAMiniPrep kit according to

the manufacturer's instructions (ZYMO Research). RNA sequenc-ing (RNA-seq) was performed with Illumina HiSeq4000 at Uni-versity of California, Berkeley. Results were analyzed usingGalaxy(https://usegalaxy.org/). The raw data were deposited to GEO(accession number: GSE117838). Heat map was generatedusing GENE-E. KEGG pathway analysis was performed usingCPDB (http://cpdb.molgen.mpg.de/). Gene set enrichment wasperformed using GSEA oncogenic signatures (http://software.broadinstitute.org/gsea/index.jsp).

The Cancer Genome Atlas data analysisEGFRvIII data were obtained from ref. 2. CCL2 expression

levels were downloaded from http://gliovis.bioinfo.cnio.es/ [TheCancer Genome Atlas (TCGA), platform HG-133A]. Three caseswith unknown EGFR status and one case with EGFRvIII and EGFReuploid were excluded from analysis.

Statistical analysisUnless otherwise noted, one-way ANOVA was performed, and

P < 0.05 was considered statistically significant.

ResultsEGFRvIII cooperates with EGFR to induce infiltrationof CD45þ cells

To further clarify how EGFR and EGFRvIII promote tumorgrowth, we tested whether EGFR and EGFRvIII influencedinfiltration of immune cells. We analyzed EGFRvIII-positivesections from primary human glioblastoma, for expression ofEGFR, EGFRvIII, and CD45, a pan hematopoietic lineage mark-er (Fig. 1A). Cells double-positive for EGFR and EGFRvIII weresurrounded by CD45-positive cells (Fig. 1A). Glioblastomatumors are heavily infiltrated by immune cells, with infiltratingmacrophages representing up to half of the tumor mass (11).To test whether inhibition of EGFR and EGFRvIII decreasedmacrophage attraction, we analyzed short-term cultures fromtwo primary patient-derived xenograft lines (GBM6 andGBM39—both EGFR and EGFRvIII coamplified; ref. 15), trea-ted with DMSO or the EGFR inhibitor lapatinib. Conditionedmedia from equal numbers of DMSO-treated GBM6 andGBM39 cells attracted more CD45þ macrophages than mediafrom lapatinib-treated GBM6 cells (Fig. 1B and C).

To test the function of EGFR and EGFRvIII in attracting macro-phages in an isogenic system, we used our published cell lines,which stably express EGFRandEGFRvIII singly or together inU87:MG cells (3). Conditioned media from cells coexpressing EGFRand EGFRvIII attracted significantly more CD45þ macrophagesthan conditioned media from cells expressing EGFR or EGFRvIIIalone (Fig. 1D). Intracranial xenografted tumors coexpressingEGFR and EGFRvIII also showed significantly more CD45þ cells,

compared with tumors expressing either EGFR or EGFRvIII alone(Fig. 1E and F). Together, these results show that coexpression ofEGFR and EGFRvIII increases infiltration of immune cells inglioblastoma tumors.

Infiltration of both M1 and M2 macrophages in response toEGFR and EGFRvIII

TAMs can polarize towards M1 proinflammatory and anti-inflammatory M2 phenotypes, respectively (16). However, M1andM2macrophagesmay coexist (17), depending on the balanceof activating and inhibitory signals and their interplay with thetumor microenvironment (17). We stained intracranial xeno-grafted tumors with M1 macrophage marker CD38 and M2macrophage marker CD206. Tumors coexpressing EGFR andEGFRvIII showed significantly more CD38þ cells and CD206þ

cells, compared with tumors expressing either EGFR or EGFRvIIIsingly (Fig. 2A–C). Thus, in these xenografted tumors, coexpres-sion of EGFR and EGFRvIII increases infiltration of both M1 andM2 macrophages.

EGFRvIII cooperates with EGFR to induce infiltration ofmacrophages and microglia

To separately assess recruitment of macrophages from recruit-ment of brain-resident microglia, we injected tumor cells intoflank regions of nudemice, a region free of microglia. Establishedtumors coexpressing EGFR and EGFRvIII showed significantlymore infiltration of CD68þ monocyte lineage cells, comparedwith tumors expressing either EGFR or EGFRvIII alone (Fig. 2D).Microglia, a brain-resident macrophage population, constitutes5% to 10% of total brain cells (18). We therefore also analyzedintracranial xenografts, staining with TMEM119, a marker ofmicroglia (19). Tumors coexpressing EGFR and EGFRvIII alsoshowed significantly more TMEM119þ cells, compared withtumors expressing either EGFR or EGFRvIII alone (Fig. 2E).Quantification further validated the results (Fig. 2F and G).Together, these results suggest that EGFR and EGFRvIII cooperateto attract both peripheral macrophages and microglia intotumors.

Inductionof cytokines and chemokines in response toEGFRvIIIand EGFR

Next, we performed RNA-seq to determine whether coexpres-sion of EGFR and EGFRvIII influenced the abundance of cyto-kines, to promote infiltration of CD45þ cells. Glioblastoma cellscoexpressing EGFR and EGFRvIII upregulated the transcription ofa large group of cytokines and chemokines, compared with cellsexpressing vector, EGFR, or EGFRvIII alone. Top hits included IL6and IL8, previously reported as regulated by EGFRvIII (20), andnew candidates including IL1RN, IL33, IL11, and CCL2 (Fig. 3A).Pathway enrichment analysis demonstrated that genes incytokine–cytokine receptor interaction and chemokine signalingpathways were upregulated cooperatively by EGFR and EGFRvIII(Fig. 3B). Using real-time PCR, we verified that EGFR andEGFRvIII upregulated transcription of IL1B, CCL2, LIF, CXCR4,and CXCL12 (Fig. 3C and D; Supplementary Fig. S1A).

We verified these results in a second-line, A172 glioblastomacells, demonstrating that coexpression of EGFR and EGFRvIII alsoled to increased transcription of CCL2 and IL1B (SupplementaryFig. S1B).Next, we ruled out the possibility that increased receptordensity on the surface of cells coexpressing EGFR and EGFRvIIIaccounted for induction of CCL2. We demonstrated that cells

EGFR and EGFRvIII Converge on the Tumor Microenvironment

www.aacrjournals.org Cancer Res; 78(24) December 15, 2018 6787



https://usegalaxy.org/

http://cpdb.molgen.mpg.de/

http://software.broadinstitute.org/gsea/index.jsp

http://software.broadinstitute.org/gsea/index.jsp

http://gliovis.bioinfo.cnio.es/


expressing EGFR or EGFRvIII alone, at cell surface receptor densitycomparable with that of cells coexpressing EGFR and EGFRvIII,failed to induce significant amounts of CCL2. We labeled cellswith Alexa Fluor 488–conjugated EGFR antibody (recognizesboth EGFR and EGFRvIII), and sorted cells with comparableEGFR/EGFRvIII levels (Supplementary Fig. S1C). For each cellline, an EGFR/EGFRvIII-high and an EGFR/EGFRvIII-low popu-lation were sorted (�10-fold difference in signal intensity). With-in each sorted population, CCL2 mRNA expression was analyzedby real-time PCR (Supplementary Fig. S1C and S1D). Cellsexpressing EGFR or EGFRvIII alone, at comparable cell surfacereceptor level as cells coexpressing EGFR and EGFRvIII, expressedsignificantly less CCL2 (Supplementary Fig. S1D). Cells with a

lower cell surface EGFR/EGFRvIII density expressed significantlyless CCL2 than those cells with a higher cell surface EGFR/EGFRvIII density (Supplementary Fig. S1D). These results indi-cated that EGFR and EGFRvIII cooperate to induce CCL2 expres-sion, and that receptor density affected the induction level.

We went onto analyze expression of proteins using cytokinearrays. Consistent with the RNA-seq data, coexpression of EGFRand EGFRvIII also upregulated IL1B, CCL2, and other cytokine atthe protein level (Fig. 3E–G; Supplementary Fig. S2). In addition,we treated short-term cultures of GBM6 and GBM39 (EGFR/EGFRvIII coamplified) with lapatinib. This treatment led to sig-nificantly less production of CCL2 compared with control cells(Fig. 3H and I). Collectively, these data suggest that EGFR and

Figure 1.

Coexpression of EGFR and EGFRvIII leads to infiltration of CD45þ cells. A, Cells double-positive for EGFR and EGFRvIII colocalize with CD45þ cells in sections fromprimary EGFR/EGFRvIIIþ human glioblastoma. B and C, Transwell migration assay showing attraction of macrophages (macrophage line, MV-4-11) in response toconditioned media from GBM6 and GBM39 cells treated with DMSO or 5 mmol/L lapatinib for 24 hours. D, Transwell migration assay showing attraction ofmacrophages (macrophage line, MV-4-11) in response to conditioned media from U87 cells coexpressing EGFR and EGFRvIII. E, Hematoxylin and eosin (H&E)and IHC staining validating increased numbers of CD45-positive cells in intracranial xenografts driven by EGFR and EGFRvIII, and decreased numbers ofCD45-positive cells in xenografts deleted for CCL2 and KRAS. F, Quantification. Macrophage attraction levels were normalized to the control cell line. Resultswere reproduced in three independent experiments with triplicate samples. �� , P < 0.01; �� , P < 0.001. Scale bar, 50 mM.

An et al.





EGFRvIII cooperate to drive expression of cytokines andchemokines.

EGFR and EGFRvIII signal through CCL2 to recruitmacrophages

CCL2, among the most prominent chemokines for macro-phage attraction, was induced by EGFR and EGFRvIII (Fig. 3).Incubation of conditioned media from cells coexpressingEGFR and EGFRvIII, with a CCL2-neutralizing antibody, sig-nificantly reduced macrophage attraction (Fig. 4A). Condi-tioned media from cells coexpressing EGFR and EGFRvIII, after

CRISPR/Cas9-mediated deletion of CCL2, also led to decreasedmacrophage infiltration (Fig. 4B; Supplementary Fig. S3A).Knockout of CCL2 in intracranial xenografted tumors coexpres-sing EGFR and EGFRvIII led to significant decreases in theabundance of infiltrating CD45þ cells (Fig. 1E and F) andmodestly enhanced survival (Fig. 4C). When we sacrificedanimals at end point, tumors were localized to forebrains,where we injected the tumor cells. To extend this result, weanalyzed genomic data from human glioblastoma. In TCGAdatasets, EGFRvIII amplification was enriched significantly insamples with high expression of CCL2 mRNA (Fig. 4D).

Figure 2.

Coexpression of EGFR and EGFRvIII leads to infiltration of M1 and M2 macrophages and microglia. A, IHC staining validating increased numbers of CD38-and CD206-positive cells in intracranial xenografts driven by EGFR and EGFRvIII. B and C, Quantification. D, IHC staining validating increased numbersof CD68-positive cells in flank xenografts driven by EGFR and EGFRvIII. E, IHC staining validating increased numbers of TMEM119-positive cells in intracranialxenografts driven by EGFR and EGFRvIII. F and G, Quantification.�� , P < 0.01; �� , P < 0.001. Scale bar, 50 mM.






Figure 3.

Coexpression of EGFR and EGFRvIII induces multiple cytokines. A, Heat map of mRNA levels for top 40 genes (listed from top to bottom according to expressionlevels) induced by EGFR and EGFRvIII. B, Pathway enrichment of genes induced by EGFR and EGFRvIII. C and D, Real-time PCR of IL1B and CCL2 from U87 cellsexpressing indicated genes. E–G, Cytokine array (IL1B and CCL2 highlighted in red boxes) and quantification confirming that EGFR and EGFRvIII cooperateto drive expression of IL1B andCCL2 in U87 cells expressing indicated constructs.H and I,Real-time PCR of CCL2 fromGBM6 andGBM39 cells treatedwith DMSOor 5mmol/L lapatinib for 24 hours. Real-time PCR results were normalized to the control cell line. Resultswere reproduced in at least three independent experiments withtriplicate samples. �� , P < 0.001.

An et al.





EGFR and EGFRvIII signal through KRAS to upregulateexpression of cytokines

To determine how EGFR and EGFRvIII activate cytokinesand chemokines, we looked for oncogenic signatures in genesets enriched by EGFR and EGFRvIII (GSEA analysis). The topfive gene sets were all related to KRAS (Fig. 5A). Knockout ofKRAS in cells coexpressing EGFR and EGFRvIII decreasedexpression of cytokines and chemokines including CCL2(Fig. 5B and C; Supplementary Fig. S3B). Consistently, con-ditioned media from cells coexpressing EGFR and EGFRvIII,and knocked out for KRAS, attracted fewer macrophages thanconditioned media from the same cells with intact KRAS

(Fig. 5D). Conditioned media from cells transfected with KRASsiRNA attracted significantly fewer macrophages comparedwith conditioned media from cells transfected with controlsiRNA (Supplementary Fig. S3C and S3D). In response totreatment with CCL2-neutralizing antibody, conditionedmedia from KRAS siRNA-transfected cells attracted fewermacrophages compared with conditioned media from controlsiRNA-transfected cells (Supplementary Fig. S3D). Theseresults indicate that although CCL2 is a major chemokine forEGFR/EGFRvIII/KRAS-induced macrophage attraction, otherKRAS-dependent cytokines also play minor roles. Intracranialtumors coexpressing EGFR and EGFRvIII and knocked out for

Figure 4.

EGFR and EGFRvIII induce macrophage attraction through upregulating CCL2. A, Depletion of CCL2 in conditioned media from U87 cells expressing EGFR andEGFRvIII decreased attraction of MV-4-11 (antibody concentration, 1 mg/mL). B, Knockout of CCL2 decreased attraction of MV-4-11. C, Kaplan–Meier curve ofnude mice bearing intracranial xenografts of indicated cell lines. D, CCL2 expression levels were plotted according to EGFRvIII status in glioblastoma patientsamples (TCGA). Red dots, 20 samples with highest expression of CCL2 mRNA. Blue dots, 20 samples with lowest expression of CCL2 mRNA. Macrophageattraction levels were normalized to the control cell line. Results were reproduced in three independent experiments with triplicate samples. �� , P < 0.001.






KRAS also showed decreased infiltration by CD45þ cells (Fig.1E and F). These data indicate that EGFR and EGFRvIII signalthrough KRAS to upregulate expression of CCL2, and topromote infiltration by TAMs.

TheNF-kBpathway is an important pathway regulating inflam-mation. We tested IkB phosphorylation, a commonly usedreadout for NF-kB activation. Under the same condition as ourRNA-seq and cytokine array, we failed to observe increasedphosphorylation of IkB (Supplementary Fig. S4). This result

suggests that other pathways might be important in regulatingEGFR/EGFRvIII/KRAS-induced cytokine expression.

Our RNA-seq results show that coexpression of EGFR andEGFRvIII upregulated CCL2, as well as stem cell markers such asCD90/THY1 and SOX1 (Supplementary Fig. S5A; refs. 21, 22).Expression of these stem cell markers was dependent onKRAS (Supplementary Fig. S5A). We also cultured GBM6 andGBM39 in stem cell medium or differentiated them with DMEMand FBS. Consistently, GBM6 and GBM39 expressed less CCL2,

Figure 5.

EGFR and EGFRvIII signal through KRAS to induce cytokine expression and macrophage attraction. A, Gene set enrichment analysis of genes upregulated byEGFR and EGFRvIII. B, Heat map of top 40 genes induced by EGFR and EGFRvIII in KRAS knockout U87 cells coexpressing EGFR and EGFRvIII. C, Real-timePCR. D, Knockout of KRAS decreased attraction of MV-4-11. Real-time PCR and macrophage attraction results were normalized to the control cell line. Resultswere reproduced in three independent experiments with triplicate samples. �� , P < 0.001.

An et al.





CD90/THY1, and SOX1 in differentiated media comparing withstem cell media (Supplementary Fig. S5B). These observationssuggest that EGFR and EGFRvIII drive expression of CCL2 pref-erentially in glioblastoma stem cells.

DiscussionAmplification of EGFR and EGFRvIII represents hallmark genet-

ic changes in glioblastoma. Amplification of EGFRvIII occurs onlyin tumors that amplify EGFR, suggesting that those twomoleculescould cooperate to promote tumor growth. Based on this coam-plification, we and others have previously studied EGFR andEGFRvIII in human cancers, to assess cooperation (3, 8, 9,23, 24). We previously demonstrated coexpression of EGFR andEGFRvIII proteins in rare individual cells within glioblastoma, inall 11 of 11 primary human glioblastoma samples analyzed (3).To address whether this coexpression was coupled to cooperationbetween EGFR and EGFRvIII, we went on to show that EGFR andEGFRvIII cooperated in transformation in vitro and in vivo, thatEGFR phosphorylated EGFRvIII, and that these two kinases con-verged on STAT3 (3). These published data illustrate clearly thatEGFR and EGFRvIII cooperated through a cell-intrinsic mecha-nism. In this article, we demonstrate that this cooperation extendsto cell-extrinsic signaling. We show that EGFR and EGFRvIIIpromote cytokine expression in vitro, leading to in vivo macro-phage attraction through the KRAS pathway. These observationsprovide new insights that cooperation of those twomolecules canmodulate the tumor microenvironment.

Glioblastoma tumors are heavily infiltrated by immune cellsincluding macrophages, myeloid-derived suppressor cells, andregulatory T cells (25), among which, infiltrating macrophagescan represent half of the tumor mass (11). Here, we provideevidence that EGFR and EGFRvIII cooperatively upregulate CCL2and promote infiltration of macrophages. Although CCL2 is amajor chemokine recruiting macrophages, knockout of CCL2 incells coexpressing EGFR and EGFRvIII did not completely blockEGFR- and EGFRvIII-induced macrophage attraction in vitro,indicating that other chemokines also contribute to macrophagerecruitment. Our RNA-seq results suggest CCL8 as an additionalcandidate macrophage-recruiting chemokine (26), as CCL8 wasalso upregulated in cells coexpressing EGFR and EGFRvIII.

Within a glioblastoma tumor, the majority of cells expresspredominantly EGFR or EGFRvIII, with only rare cells coexpres-sing both EGFR and EGFRvIII (3). These observations suggest thatEGFR and EGFRvIII may signal in glioma stem cells, with subse-quent mitoses leading them to segregate independently to moredifferentiated daughter cells. Our results are consistent withobservations that unlike their differentiated progeny, glioma stemcells show an increased capacity for active chemoattraction andrecruitment of macrophages, through upregulation of cytokines(4). Our results suggest a possible link between KRAS-mediatedcytokine expression and glioblastoma stemness.

Macrophages can be divided into tumor-suppressive andtumor-promoting classes (27). CCL2 can polarize macrophages

toward a tumor-promoting, immunosuppressive phenotype(28). Our results showed that coexpression of EGFR and EGFRvIIIpromoted infiltration of both tumor-promoting (M2) and tumor-suppressive macrophages (M1). This may be one reason whyknockout of CCL2 in cells coexpressing EGFR and EGFRvIII onlymodestly affected survival. Another issue is that macrophagesinduced by EGFR and EGFRvIII cannot in-turn influence theabundance or cellular activity of lymphocytes, which are absentin the nude mouse models used in this study. Notably, it wasreported that B cells and T cells in the immune system,whichnudemice lack, drove macrophage polarization to M2 (29, 30). Ourhuman cell-based model systems pose real limitations (31).Although immunocompetent mouse models are better thannude mice for studying the tumor microenvironment, we areunaware of immunocompetentmodels for glioblastoma inwhichEGFR and EGFRvIII are clear drivers. Such a model, which we areactively trying to develop, is essential to further delineate func-tionally how EGFR and EGFRvIII modulate the microenviron-ment in glioblastoma.

Disclosure of Potential Conflicts of InterestG. Reifenberger is a consultant/advisory board member for AbbVie. W.A.

Weiss is co-founder of StemSynergy Therapeutics. No potential conflicts ofinterest were disclosed by the other authors.

Authors' ContributionsConception and design: Z. An, W.A. WeissDevelopment of methodology: Z. An, C.B. Knobbe-ThomsenAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): Z. An, C.B. Knobbe-Thomsen, X.Wan, G. ReifenbergerAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): Z. An, C.B. Knobbe-Thomsen, Q.W. FanWriting, review, and/or revision of the manuscript: Z. An, W.A. WeissAdministrative, technical, or material support (i.e., reporting or organizingdata, constructing databases): Z. An, C.B. Knobbe-Thomsen, Q.W. FanStudy supervision: W.A. Weiss

AcknowledgmentsZ. An received support from Alex's Lemonade Stand Foundation (CA-

0091339), American Brain Tumor Association in honor of Juliana Schaferand Team Hope (BRF1800011), NIH T32CA108462, and Program forBreakthrough Biomedical Research, which is partially funded by the SandlerFoundation (7028233). C.B. Knobbe-Thomsen was supported by the Ger-man Cancer Aid (DKH #110663) and by the BMBF (#01ZX1401B). Q.W. Fanwas supported by NIH grant R01CA221969. W.A. Weiss was supported byNIH grants R01CA221969, R01NS091620, P50CA097257, U01CA217864,and P30CA82103; the Samuel G. Waxman Cancer Research Foundation(A128431); and the Evelyn and Mattie Anderson Chair. We thank Drs.Hideho Okada and Erin Simonds for useful discussions, Dr. Miller Huangand Lauren McHenry for critical review, and the UCSF Brain Tumor Core forhelp on intracranial injection.

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 November 16, 2017; revised July 31, 2018; accepted October 24,2018; published first November 6, 2018.

References1. Cloughesy TF, Cavenee WK, Mischel PS. Glioblastoma: from molec-

ular pathology to targeted treatment. Ann Rev Pathol 2014;9:1–25.

2. Brennan CW, Verhaak RG, McKenna A, Campos B, Noushmehr H,Salama SR, et al. The somatic genomic landscape of glioblastoma. Cell2013;155:462–77.

3. Fan QW, Cheng CK, Gustafson WC, Charron E, Zipper P, Wong RA, et al.EGFR phosphorylates tumor-derived EGFRvIII driving STAT3/5 and pro-gression in glioblastoma. Cancer Cell 2013;24:438–49.

4. Audia A, Conroy S, Glass R, Bhat KPL. The impact of the tumor microen-vironment on the properties of glioma stem-like cells. Front Oncol2017;7:143.






5. Mondal A, Kumari Singh D, Panda S, Shiras A. Extracellular vesicles asmodulators of tumor microenvironment and disease progression in glio-ma. Front Oncol 2017;7:144.

6. Gurgis FM, Yeung YT, Tang MX, Heng B, Buckland M, Ammit AJ, et al. Thep38-MK2-HuR pathway potentiates EGFRvIII-IL-1beta-driven IL-6 secre-tion in glioblastoma cells. Oncogene 2015;34:2934–42.

7. Bonavia R, Inda MM, Vandenberg S, Cheng SY, Nagane M, Hadwiger P,et al. EGFRvIII promotes glioma angiogenesis and growth through the NF-kappaB, interleukin-8 pathway. Oncogene 2012;31:4054–66.

8. Inda MM, Bonavia R, Mukasa A, Narita Y, Sah DW, Vandenberg S, et al.Tumor heterogeneity is an active process maintained by a mutant EGFR-induced cytokine circuit in glioblastoma.GenesDevelop 2010;24:1731–45.

9. Zanca C, Villa GR, Benitez JA, Thorne AH, Koga T, D'Antonio M, et al.Glioblastoma cellular cross-talk converges on NF-kappaB to attenuateEGFR inhibitor sensitivity. Genes Develop 2017;31:1212–27.

10. Leung SY, Wong MP, Chung LP, Chan AS, Yuen ST. Monocyte chemoat-tractant protein-1 expression andmacrophage infiltration in gliomas. ActaNeuropathol (Berl) 1997;93:518–27.

11. Chen Z, Feng X, Herting CJ, Garcia VA, Nie K, PongWW, et al. Cellular andmolecular identity of tumor-associated macrophages in glioblastoma.Cancer Res 2017;77:2266–78.

12. 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.

13. Platten M, Kretz A, Naumann U, Aulwurm S, Egashira K, Isenmann S, et al.Monocyte chemoattractant protein-1 increases microglial infiltration andaggressiveness of gliomas. Ann Neurol 2003;54:388–92.

14. Carlson BL, Pokorny JL, Schroeder MA, Sarkaria JN. Establishment,maintenance and in vitro and in vivo applications of primary humanglioblastoma multiforme (GBM) xenograft models for translationalbiology studies and drug discovery. Curr Protoc Pharmacol 2011;Chapter 14:Unit 14 16.

15. Johnson H, Del Rosario AM, Bryson BD, Schroeder MA, Sarkaria JN, WhiteFM. Molecular characterization of EGFR and EGFRvIII signaling networksin human glioblastoma tumor xenografts. Mol Cell Proteomics2012;11:1724–40.

16. Yang L, Zhang Y. Tumor-associated macrophages: from basic research toclinical application. J Hematol Oncol 2017;10:58.

17. Martinez FO, Gordon S. The M1 and M2 paradigm of macrophageactivation: time for reassessment. F1000prime Rep 2014;6:13.

18. Li Q, Barres BA. Microglia and macrophages in brain homeostasis anddisease. Nat Rev Immunol 2018;18:225–42.

19. Bennett ML, Bennett FC, Liddelow SA, Ajami B, Zamanian JL, Fernhoff NB,et al. New tools for studying microglia in the mouse and human CNS.PNAS 2016;113:E1738–46.

20. An Z, Aksoy O, Zheng T, Fan QW, Weiss WA. Epidermal growth factorreceptor and EGFRvIII in glioblastoma: signaling pathways and targetedtherapies. Oncogene 2018;37:1561–75.

21. ShaikhMV, KalaM, Nivsarkar M. CD90 a potential cancer stem cell markerand a therapeutic target. Cancer Biomark 2016;16:301–7.

22. VenereM,Han YG, Bell R, Song JS, Alvarez-Buylla A, BlellochR. Sox1marksan activated neural stem/progenitor cell in the hippocampus. Develop-ment 2012;139:3938–49.

23. Feng H, Hu B, Jarzynka MJ, Li Y, Keezer S, Johns TG, et al. Phosphorylationof dedicator of cytokinesis 1 (Dock180) at tyrosine residue Y722 by Srcfamily kinases mediates EGFRvIII-driven glioblastoma tumorigenesis.PNAS 2012;109:3018–23.

24. Li L,Chakraborty S, YangCR,HatanpaaKJ, CipherDJ, PuliyappadambaVT,et al. An EGFRwild type-EGFRvIII-HB-EGF feed-forward loop regulates theactivation of EGFRvIII. Oncogene 2014;33:4253–64.

25. Sampson JH,Maus MV, June CH. Immunotherapy for brain tumors. J ClinOncol 2017;35:2450–56.

26. Huma ZE, Sanchez J, Lim HD, Bridgford JL, Huang C, Parker BJ, et al.Key determinants of selective binding and activation by the monocytechemoattractant proteins at the chemokine receptor CCR2. Sci Signal2017;10.

27. Sica A, Mantovani A. Macrophage plasticity and polarization: in vivoveritas. J Clin Invest 2012;122:787–95.

28. Sierra-Filardi E, Nieto C, Dominguez-Soto A, Barroso R, Sanchez-Mateos P,Puig-Kroger A, et al. CCL2 shapes macrophage polarization by GM-CSFand M-CSF: identification of CCL2/CCR2-dependent gene expressionprofile. J Immunol 2014;192:3858–67.

29. Wong SC, Puaux AL, Chittezhath M, Shalova I, Kajiji TS, Wang X, et al.Macrophage polarization to a unique phenotype driven by B cells.Eur J Immunol 2010;40:2296–307.

30. Chan T, Pek EA, Huth K, Ashkar AA. CD4(þ) T-cells are important inregulating macrophage polarization in C57BL/6 wild-type mice. CellImmunol 2011;266:180–6.

31. Allen M, Bjerke M, Edlund H, Nelander S, Westermark B. Origin of theU87MG glioma cell line: good news and bad news. Sci Transl Med2016;8:354re3.


An et al.




Correction

Correction: EGFR Cooperates with EGFRvIIIto Recruit Macrophages in GlioblastomaZhenyi An, Christiane B. Knobbe-Thomsen, Xiaohua Wan,Qi Wen Fan, Guido Reifenberger, and William A.Weiss

In the original version of this article (1), two panels were mistakenly labeled asFig. 1E. The error has been corrected in the latest online HTML and PDF versions ofthe article. The authors regret this error.

Reference1. An Z, Knobbe-Thomsen CB, Wan X, Fan QW, Reifenberger G, Weiss WA. EGFR cooperates with

EGFRvIII to recruit macrophages in glioblastoma. Cancer Res 2018;78:6785–94.

Published first November 1, 2019.Cancer Res 2019;79:5681doi: 10.1158/0008-5472.CAN-19-2800�2019 American Association for Cancer Research.

CancerResearch

www.aacrjournals.org 5681



2018;78:6785-6794. Published OnlineFirst November 6, 2018.Cancer Res Zhenyi An, Christiane B. Knobbe-Thomsen, Xiaohua Wan, et al. GlioblastomaEGFR Cooperates with EGFRvIII to Recruit Macrophages in

Updated version

10.1158/0008-5472.CAN-17-3551doi:

Access the most recent version of this article at:

Material

Supplementary

http://cancerres.aacrjournals.org/content/suppl/2019/02/06/0008-5472.CAN-17-3551.DC1

Access the most recent supplemental material at:

Cited articles

http://cancerres.aacrjournals.org/content/78/24/6785.full#ref-list-1

This article cites 29 articles, 10 of which you can access for free at:

Citing articles

http://cancerres.aacrjournals.org/content/78/24/6785.full#related-urls

This article has been cited by 2 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/78/24/6785To 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-17-3551

http://cancerres.aacrjournals.org/content/suppl/2019/02/06/0008-5472.CAN-17-3551.DC1

http://cancerres.aacrjournals.org/content/78/24/6785.full#ref-list-1

http://cancerres.aacrjournals.org/content/78/24/6785.full#related-urls

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

mailto:[email protected]

http://cancerres.aacrjournals.org/content/78/24/6785


EGFR Cooperates with EGFRvIII to Recruit Macrophages in ... · Tumor Biology and Immunology EGFR...

Documents

Transcript of EGFR Cooperates with EGFRvIII to Recruit Macrophages in ... · Tumor Biology and Immunology EGFR...