EGFR induces expression of IRF-1 via STAT1 and STAT3 activation leading

to growth arrest of human cancer cells

Peter Andersen1, Mikkel Wandahl Pedersen1, Anders Woetmann2, Mette Villingshøj1, Marie-Th�er�ese Stockhausen1,Niels Ødum2 and Hans Skovgaard Poulsen1*

1Department of Radiation Biology, The Finsen Center, Copenhagen University Hospital, Copenhagen, Denmark2Institute of Molecular Biology, Department of Immunology and Institute of Medical Microbiology and Immunology,University of Copenhagen, Copenhagen N, Denmark

Recently, we reported that epidermal growth factor receptor(EGFR) induce expression of a module of genes known to be in-ducible by interferons and particularly interferon-c. Here weshow that the module is tightly regulated by EGFR in the 2 humancancer cell lines that overexpress EGFR, A431 and HN5. Themodule of genes included the tumor suppressor IRF-1, which wasused as a prototypical member to further investigate the regula-tion and function of the module. Ligand-activated EGFR induceexpression of IRF-1 via phosphorylation of STAT1 and STAT3. Incontrast, cells expressing the constitutively active cancer specificreceptor EGFRvIII are unable to mediate phosphorylation ofthese STATs and thereby incapable of inducing IRF-1. We alsodemonstrate that IRF-1 is expressed in an EGF dose-dependentmanner, which correlates with inhibition of cell proliferation, andthat the regulation of IRF-1 is partially dependent on intracellularSrc family kinase activity. Treatment with the dual specific Abl/c-Src kinase inhibitor AZD0530 significantly reduces the growth in-hibitory effect of high EGF concentrations, signifying that EGFRinduced IRF-1 is responsible for the observed growth inhibition.In addition, we show that media from these EGF treated cancercells upregulate the activation marker CD69 on both B-cells andT-cells in peripheral blood. Taken together, these results suggestthat cells acquiring sustained high activity of oncogenes such asEGFR are able to activate genes, whose products mediate growtharrest and activate immune effector cells, and which potentiallycould be involved in alerting the immune system in vivo leading toelimination of the transformed cells.' 2007 Wiley-Liss, Inc.

Key words: EGFR; EGFRvIII; IRF-1; STAT; Src; signaling; inter-ferons; cancer

The epidermal growth factor receptor (EGFR) belongs to the re-ceptor tyrosine kinases (RTKs), which regulate critical cellularactivities such as proliferation, motility, survival and apoptosis.EGFR is the best-characterized member of the ErbB family and itsexpression is frequently associated with human cancers.1,2 The re-ceptor often holds enhanced or altered signaling capacity due tooverexpression, autocrine growth factor loops and/or mutations inthe EGFR gene, where the most common mutation is the constitu-tively activated type III mutation (EGFRvIII).3–5 These alterationsoften result in cellular phenotypes characterized by uncontrolledgrowth, increased proliferation, resistance to apoptosis and abilityto invade surrounding tissue.

Using a global approach, we have previously reported thatligand-activated EGFR induce expression of a module of genesknown to be inducible by interferons (IFNs) and particularly IFN-g.6 Recent studies have reported similar ‘‘IFN signatures’’ inhuman cancers, including those of lymphoblastic, breast and ovar-ian origin.7 Similar IFN signatures have been reported to beinduced by a mutant version of the platelet derived growth factorreceptor (PDGFR)8 and by the fusion oncogene BCR/ABL1,9 sug-gesting that the products of these IFN-like signatures may have abiological function in neoplastic processes. Very recently ‘‘IFNsignatures’’ have been shown to correlate positively with the typeand location of tumor-infiltrating immune cells. A higher numberof tumor infiltrating immune cells was found to be a better predi-cator of patient survival.10 To investigate the functional role ofthis module of genes downstream of EGFR in human cancer, thetranscription factor IFN regulatory factor 1 (IRF-1) was used as a

prototypical member of the module. IRF-1 regulates cellular andimmune response against pathogens and a major regulator of bothcell growth and apoptosis.11–15

We show that the mRNA levels of IRF-1, suppressor of cyto-kine signaling 3 (SOCS-3), interleukin 6 (IL-6) and guanylatebinding protein-2 (GBP-2) are induced in human cancer cells thatoverexpress EGFR. Regulation of IRF-1 was also detected inmammalian cell lines stably overexpressing EGFR, but not in cellsexpressing the constitutive active deletion variant EGFRvIII. Fur-thermore, it is shown that expression of IRF-1 is EGF-dose de-pendent, dependent on Src family kinase activity and mediated byactivation of STAT1 and STAT3. We also provide evidence thatthe inhibition of cell proliferation observed at high EGF levelsmost likely are mediated by upregulation of the tumor suppressorIRF-1 and that these cancer cells produce products that activatesimmune effector cells.

Material and methods

Materials

Recombinant human EGF was from Calbiochem (Germany)and alexa-488-labeled EGF was from Invitrogen. Phospho-EGFR(pY845) STAT1, STAT3, phospho-STAT1 (pY701), phospho-STAT3 (pY705) antibodies were from Cell Signaling Technology(Germany). Phospho-EGFR (pY1173) was from Upstate Biotech-nology (USA). EGFR antibody was from Fitzgerald Industries(USA). Antibodies against IRF-1 and Tubulin were from SantaCruz Biotechnology, CA. HRP-conjugated secondary antibodieswere purchased from, DAKO Cytomation (Denmark). FITC antihuman CD19 (4G7), PerCP anti human CD3 (SK7), APC, humanCD69 and Annexin V-conjugated Phycoerythrin (PE) were fromBecton Dickinson. AG1478, U73122 and SU6656 were from Cal-biochem (CA). LY294002 was from Bioscource (CA). U0126 wasfrom Promega (Germany). SB203580 and Wortmannin were fromUpstate Cell Signaling Solution (NY). Iressa and AZD0530 werekindly provided by AstraZeneca (Denmark).

Cell cultures

The cell lines NR6, NR6wtEGFR, NR6W and NR6M havebeen described previously.6 Briefly, NR6wtEGFR and NR6Wexpress low and high levels of EGFR, respectively. NR6M expressEGFRvIII, whereas NR6 lacks expression of these receptors. Thehuman epidermoid carcinoma cell line A431 was obtained fromthe American Type Culture Collection and the head and neck car-cinoma cell line HN5 was kindly provided by Dr. Jiri Bartek,Department of Cell Cycle and Cancer, Danish Cancer Society, Co-penhagen, Denmark. All cells were maintained in DMEM (Invi-

Grant sponsor: Danish Cancer Society.*Correspondence to: Department of Radiation Biology, The Finsen

Center, Section 6321, Copenhagen University Hospital, Blegdamsvej 9,DK-2100 Copenhagen, Denmark. Fax:145-35456301.E-mail: hans.skovgaard.poulsen@rh.regionh.dkReceived 21 February 2007; Accepted after revision 6 July 2007DOI 10.1002/ijc.23109Published online 4 October 2007 in Wiley InterScience (www.interscience.

wiley.com).

Int. J. Cancer: 122, 342–349 (2008)' 2007 Wiley-Liss, Inc.

Publication of the International Union Against Cancer

trogen, Denmark) supplemented with 10% fetal calf serum (FCS),50 U/ml penicillin and 50 lg/ml streptomycin.

Immunoblot analyses

Five microgram whole cell lysate was resolved by SDS-PAGEand electroblotted onto nitrocellulose membranes. The membraneswere incubated with primary antibodies in 5% nonfat milk over-night at 4�C, and secondary antibodies for 1 hr at room tempera-ture. The chemiluminescence detection method (ECL) was usedfor all western blot experiments. Band intensities were determinedusing the public domain Java image-processing program ImageJ(http://rsb.info.nih.gov/ij).

Quantitative real-time RT-PCR

Total RNA was extracted using Trizol reagent (Invitrogen, Den-mark). The cDNA synthesis and the real-time RT-PCR reactionwere carried out using the SuperScript III Platinum Two-StepqRT-PCR Kit with SYBR Green (Invitrogen, Denmark) with thefollowing IRF-1 primer sets (annealing temp (Ta) 57.7�C): IRF-1forward primer (50-CGAATCGCTCCTGCAGCAGA-30) and IRF-1 reverse primer (50-GCCCAGCTCCGGAACAAACA-30), IL-6primer sets (Ta 55�C): IL-6 forward primer (50-GAGAA-GATTCCAAAGATGTAGCC-30) and IL-6 reverse primer (50-CCAGATTGGAAGCATCCATC-30), GBP-2 primer sets (Ta 47�C):GBP-2 forward primer (50-CGTTGGAGGGGTAAAGTAAAA30)and GBP-2 reverse primer (50-CCCTTTAGTGTTATCAATGAGGC-30), SOCS-3 primer sets (Ta 63�C): SOCS-3 forwardprimer (50-CCTTTGTGGACTTCACGGCC-30) and SOCS-3reverse primer (50-TTGCTGTGGGTGACCATGGC-30) and thehousekeeping primers sets (RPL13A) (Ta 65�C): RPL13A forwardprimer (50-CATCGTGGCTAAACAGGTACT-30) and RPL13Areverse primer (50-GCACGACCTTGAGGGCAGCA-30). All pri-mers were from DNA Technology (Denmark). The PCR was initi-ated with 95�C for 10 min and the PCR cycling conditions were asfollows: 95�C for 20 sec, the above mentioned specific Ta for 20sec and 72�C for 20 sec for 40 cycles. The qRT-PCR values werenormalized using a specific cDNA standard curve obtained usingknown amounts of cDNA. IRF-1, IL-6, GBP-2 and SOCS-3 valueswere then normalized to RPL13A.

Precipitation assay

50 Biotin conjugated oligonucleotides were designed accordingto the published GAS containing promoter sequence of IRF-116

(Taq Copenhagen A/S, Denmark): Sense sequence (Biotin-50-ACAGC-CTGATTTCCCCGAAATGAC-30) and the antisensesequence (Biotin-50-GTCATTTCGGGGAAAT-CAGGCTGT-30).The oligonucleotides were mixed in equal molar proportions togenerate a double stranded biotin-conjugated probe. One milli-gram cell lysate was precleared and mixed with the probe over-night. The biotin-probe:protein-complexes were precipitated withAvidin–agarose beads (Kem-En-Tec A/S, Denmark), collected bycentrifugation and washed 5 times in RIPA lysisbuffer. Biotin-probe:protein-complexes were analyzed by Western blot analysis.

MTT assay

For the EGF concentration profile cells were seeded in 96-wellplates (1,000 cells/well), and incubated overnight in medium con-taining 0.5% FCS. Cells were then treated with EGF for 72 hrbefore addition of 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetra-zolium bromide (MTT) (Sigma-Aldrich, Denmark). For the IRF-1profile cells were transfected with the IRF-1 encoding plasmid(pCMVBL/IRF-1) kindly provided by Dr. John Hiscott, LadyDavis Institute, Montreal, Canada. Control cells were transfectedwith an empty vector (pCMVBL). After transfection, cells weretrypsinized and seeded in 96-well plates at a concentration of1,000 cells/well in medium containing 10% FCS for 48 hr beforeaddition of MTT.

Flow cytometry assays

Proliferation/apoptosis. Cells were grown to 80% confluencein media containing 10% FCS and stained with 7.5 lM carboxy-fluorescein diacetate succinimidyl ester (CFSE) (Invitrogen,Denmark) in PBS for 15 min. CFSE was removed and replacedwith normal media (10% FCS) for 30 min. Cells were trypsinizedand cultured in T25 flasks (200,000 cells/flask) in media contain-ing 0.5% FCS overnight. Cells were either left untreated, treatedwith 0.1 nM EGF or treated with 10 nM EGF. Before flow cytom-etry, cells were trypsinized, collected by centrifugation and keptin suspension in media containing 1 mM EDTA. Cells werestained with Annexin V-conjugated PE 30 min prior to flowcytometry analysis.

PBMC stimulation: A431 cells in DMEM 1 0.5% FCS were ei-ther grown in the presence 10 nM EGF or left untreated for 72 hr.Media were then harvested and spun at 300g for 15 min to removeloose cells. Peripheral blood mononuclear cells (PBMC’s) wereisolated from Buffy coats using standard Ficoll-Hypaque tech-nique. One million PBMC’s were transferred to T75 flasks con-taining 15 ml of media from the EGF treated or untreated cellsand incubated for 48 hr. About 50,000 PBMC’s from each treat-ment group was incubated with 10 ll of the following antibodies:FITC mouse anti human CD19 (BD Biosciences), PerCP mouseanti human CD3 (BD Biosciences) and APC mouse anti humanCD69 (BD Biosciences). Cells were gated as T-cells (CD3 posi-tive) or B-cells (CD19 positive) and the number of CD69 positivecells in each gate determined.

Cell morphology and counting

Cells were cultured (200,000 cells/T25 flask) in normal mediaovernight and subsequently media was replaced with media con-taining 0.5% FCS and incubated overnight, before EGF treatment.Pictures of cell morphology were taken at the indicated timesusing a Nikon Eclipse TS100 microscope. Afterwards cells weretrypsinized followed by addition of a trypan blue:PBS solution(1:5) for 5 min. The number of viable and dead cells was countedusing an Olympys CK2 microscope.

Immunoflourescence

HN5 cells were seeded in duplicate in 96-well image plates inDMEM 1 0.5% FCS. Cells were then added 10 nM of Alexa-488-labeled EGF (Invitrogen) and allowed to incubate for different pe-riod of time from 30 min to 48 hr. LysoTracker Red DND-99(Invitrogen) was added to wells at a final concentration of 75 nM1 hr prior to fixation. Cells were then fixed in 4% formaldehydefor 30 min, washed 3 times in wash buffer (PBS with 0.1%Tween-20). Images were taken using a Leika fluorescence micro-scope at 4003 magnification.

Results

Activated EGFR induces expression of IRF-1, SOCS-3,IL-6 and GBP-2

Recently, using a model system consisting of 4 murine fibro-blast cell lines either not expressing or stably expressing humanEGFR or EGFRvIII we found that activated EGFR, but not EGFR-vIII induces expression of a module of genes known to be induci-ble by IFNs.6 (Fig. 1a) To confirm and extend these results tohuman cancer cells, EGFR induced expression of selected genesfrom the module (IRF-1, SOCS-3, IL-6 and GBP-2) were analyzedin the human cancer cell lines A431 and HN5 known to overex-press EGFR. Cells were treated with 10 nM EGF for up to 24 hrand the mRNA levels estimated by quantitative real-time PCR.Both cell lines show a rapid increase in mRNA levels, whichreaches a maximum induction after 1–2 hr in HN5 cells (Fig. 1b),clearly demonstrating that these genes are regulated by EGFR.SOCS-3 and IRF-1 show a maximum fold induction of �120- and�40-fold, respectively, whereas both IL-6 and GBP-2 show a �6-to 7-fold induction. A similar expression pattern was observed in

343EGFR MEDIATED IRF-1 EXPRESSION THROUGH STAT ACTIVATION

FIGURE 1 – (a) The module of interferon-associated genes (Reproduced with permission from Pedersen et al., J Cell Biochem, 2005, 96, 412–427, ' Wiley Interscience). (b) IRF-1, SOCS-3, IL-6, GBP-2 mRNA levels were detected in Trizol extracted RNA samples by quantitative real-time PCR analyzes. Bars indicate standard error of mean (SEM) from 3 independent experiments. (c) Western blot analysis showing levels of IRF-1 in HN5 cells stimulated with 10 nM EGF for 1 hr (1) or left untreated (2). (d) Western blot analyzes showing levels of IRF-1 in HN5, A431 andNR6 cells stimulated with 10 nM EGF for 48 hr and NR6M released from 5 lM AG1478 inhibition for 48 hr at the indicated periods. (e) Relativeincreases in IRF-1 protein levels quantified by measuring band intensities detected by Western blot analyzes, where 1 of 3 independent experimentsare shown in d. (f) Localization of Alexa-488 labeled EGF in HN5 cells after incubation for the different periods of time and western blot analysisshowing EGFR levels to the different indicated periods. (g) Colocalization of Alexa-488 labeled-EGF and Lysotracker Red in HN5 after 4 hr ofEGF stimulation. Student’s t-test was performed and statistically changes is indicated in the figures by 1 asterisk * (p < 0.05).

A431, but with a maximum fold induction after 4 hr (data notshown). To investigate the expression of the module in moredetail, IRF-1 was chosen as a prototypical member of the moduleas it is a well-known transcriptional activator of various IFN genesand possibly genes in the module.17 The cell lines A431, HN5,NR6, NR6wtEGFR, NR6W and NR6M was used to investigatethe expression of IRF-1 in response to EGF. Cells were treatedwith 10 nM EGF for 1 hr and IRF-1 protein levels confirmed thatonly EGFR expressing cells are able to activate IRF-1 expressionin response to EGF (Fig. 1c). Notably, a high basal level of IRF-1was detected in the A431 cell line. Interestingly, A431 cells areknown to endogenously produce TGF-a and it has been reportedthat EGFR is activated in an ‘‘autocrine’’ manner,18 which maypartly explain the higher background levels of IRF-1 observed inthis cell line.

We then analyzed the time-course of IRF-1 protein expressionfor up to 48 hr. A431, HN5 and NR6 cells were stimulated withEGF, whereas NR6M cells were pretreated with the specificEGFR/EGFRvIII tyrosine kinase inhibitor AG1478, and thenreleased from inhibition, to simulate activation of EGFRvIII.Analogous to the mRNA profiles, HN5 cells show a higher foldincrease of IRF-1 protein levels compared to A431 (Figs. 1d and1e). The protein levels are almost completely attenuated after 48hr, indicating that IRF-1 is transiently regulated. EGF treatment ofthe parental NR6 cells, which lacks endogenous EGFR expression,as well as releasing NR6M cells from AG1478, had no effect onIRF-1 expression. To investigate whether the nature of the geneexpression was due to downregulation and degradation of EGFR,HN5 cell were added 10 nM Alexa-488-EGF and incubated fordifferent periods of time. The fate of the ligand and EGFR wasfollowed visually and EGFR levels were detected by western blotanalysis (Fig. 1f). It is clear from the results that EGF leads to arapid surface staining followed by a rapid internalization and deg-radation of the ligand-receptor complex (Fig. 1f). After 48 hr verylittle fluorescent signal is left although the media is saturated withlabeled EGF, indicating that the receptor has been degraded. Fur-thermore, the ligand-receptor complex is found in lysozomes after4 hr as shown by colocalization of the Alexa-488-EGF signal andLyso-Tracker Red (Fig. 1g).

These results demonstrate that IRF-1, SOCS-3, IL-6 and GBP-2are directly regulated by activated EGFR in human cancer celllines overexpressing EGFR and that EGFRvIII is unable to induceIRF-1 expression.

IRF-1 expression is regulated by STAT1 and STAT3

Previously, we found that activation of the IFN-associated mod-ule correlated with phosphorylation of STAT1 and STAT3.6 Tovalidate the role of STAT1 and STAT3, we investigated the phos-phorylation patterns of these STATs as well as of EGFR parallelwith the expression of IRF-1 in HN5 and A431 cells treated with10 nM EGF for up to 24 hr (Fig. 2a). The results demonstrate thatEGFR induced expression of IRF-1 correlates with phosphoryla-tion of STAT1, STAT3 and EGFR. Notably, the expression ofIRF-1 as well as the phosphorylation pattern of STAT1 andSTAT3 was very similar between the 2 cell lines. Analogousresults were seen in NR6wtEGFR and NR6W, however, both NR6and NR6M were unable to induce phosphorylation of neitherSTAT1, STAT3 nor expression of IRF-1 (data not shown). Fur-thermore, EGF appears to upregulate STAT1 protein levels after 8hr, whereas STAT3 levels seem to be constant in the time framestudied.

Previous studies have demonstrated that STAT1, and to a lesserextend STAT3, are able to activate transcription of IRF-1 inresponse to IFN-g treatment.19 According to the published pro-moter sequence,16 IRF-1 contains a conserved IFN-g activationsite (GAS) within the promoter. Furthermore, IFN-g inducedexpression of IRF-1 is completely abolished in STAT12/2 cells.20

On the basis of the observed correlation between IRF-1 expressionand phosphorylation of STAT1 and STAT3, we hypothesized that

EGFR activated STAT1 and/or STAT3 would bind to the GASsite within the IRF-1 promoter and thereby induce its expression.To demonstrate such a mechanism, the ability of STAT1 andSTAT3 to bind a biotin-conjugated dsDNA probe containing theGAS consensus sequence from the IRF-1 promoter in response toEGF was investigated. Cells were treated with 10 nM EGF for 15min or left untreated and whole cell lysates from A431 and HN5cells were precipitated with the biotin-conjugated probe andSTAT1 and STAT3 levels were detected by western blot analysis.(Fig. 2b). The results demonstrate that EGF increases binding of

FIGURE 2 – (a) Western blot analysis showing levels of IRF-1,EGFR, STAT1 and STAT3 and phosphorylation status of EGFR,STAT1 and STAT3 in HN5 and A431 treated with 10 nM EGF for 24hr, tubulin is used as loading control. (b) Biotin-conjugated GASsequence precipitation using avidin–agarose beads and whole celllysates from HN5 and A431 cells left untreated or stimulated 15 minwith 10 nM EGF. (c) Western blot analysis showing IRF-1 and EGFRlevels as well as phosphorylation status of Y845- and Y1173EGFR inHN5 cells treated with 5 lM AG1478; 1 lM Iressa; 20 lM U0126;0.1 lM Wortmannin; 5 lM LY294002; 0.1 lM U-73122; 10 lMSB203580; 15 lM SU6656, 0.1 lM AZD0530 and/or with 10 nMEGF for 1 hr. Tubulin levels confirm equal protein loading.

345EGFR MEDIATED IRF-1 EXPRESSION THROUGH STAT ACTIVATION

both STAT1 and STAT3 to the GAS site from the IRF-1 promoterin both cell lines. In summary, these results demonstrate that EGFtreatment induce phosphorylation of STAT1 and STAT3, whichthen bind to the GAS site within the IRF-1 promoter and therebyrapidly induce transcription of IRF-1.

IRF-1 expression is dependent on Src kinase activity

To dissect if other signaling molecules downstream of EGFRwere involved in the induction of IRF-1 expression, we treatedHN5 cells with a panel of small molecular weight inhibitors andmeasured the levels of IRF-1 in response to EGF. Cells weretreated with inhibitors for 2 hr prior to EGF treatment. Detectionof EGFR tyrosine phosphorylation at 2 different sites (pY845 andpY1173) show that treatment with the 2 specific EGFR tyrosinekinase inhibitors AG1478 and Iressa inhibits phosphorylation ofEGFR and demonstrates that IRF-1 expression is dependent onEGFR tyrosine kinase activity (Fig. 2c). The inhibitors U0126(MEK inhibitor); Wortmannin and LY294002 (PI3-K inhibitors);U73122 (PLC-g inhibitor); SB203580 (p38 inhibitor) had noeffect on IRF-1 expression. However, the two specific Src kinaseinhibitors SU6656 and AZD0530 resulted in a marked decrease inIRF-1 expression. Importantly, these inhibitors did not reducephosphorylation of Y1173- and Y845EGFR, demonstrating that atthese concentrations they have no effect on the EGFR tyrosine ki-nase. Thus, IRF-1 expression is dependent on both EGFR tyrosinekinase activity as well as Src family kinase activity. Similar resultwas observed in the A431 cell line (data not shown).

IRF-1 expression is EGF-dose dependent and correlates withreduced cell viability

Next we analyzed the effect of various EGF concentrations onexpression of IRF-1. HN5 cells were treated with EGF concentra-tions ranging from 0.01 nM to 25 nM and IRF-1 levels weredetected by immunoblotting (Fig. 3a). The results show that EGFinduces expression of IRF-1 in a dose-dependent manner. Theincrease in IRF-1 levels correlated with an increase in phosphoryl-ation of both STAT1 and STAT3, once more signifying the rolefor these STATs in regulation of IRF-1.

It is well known that cells expressing high levels of EGFRrespond to low concentrations of EGF by increased proliferationwhereas high concentrations of EGF reduces cell growth.6,21 Thisobservation was verified in our study (Fig. 3a), where treatmentwith low EGF concentrations (0.01–1 nM) increased the numberof metabolically active cells (viable cells), while higher concentra-tions of EGF (2.5–25 nM) markedly reduced the number. Thereduction in the number of viable cells correlates with IRF-1expression, suggesting that IRF-1 may be involved in growth inhi-bition in these cells.

To further confirm a role for IRF-1 in EGF mediated tumor cellproliferation IRF-1 was transiently overexpressed in HN5 cells (Fig.3b). A significant reduction in cell viability was observed in cellstreated with the IRF-1 encoding plasmid (pCMVBL/IRF-1) and dem-onstrate that elevated IRF-1 expression reduce cell viability. Addition-ally, the dual specific Abl/c-Src kinase inhibitor AZD0530 which wefound in Figure 2c tend to decrease IRF-1 levels was also found toincrease the number of viable cells when exposed to high levels ofEGF (10 nM) (Fig. 3c). An observation that suggest that high EGFconcentrations activates a Src dependent pathway downstream ofEGFR and that these gene products, presumably IRF-1, is involved inthe EGF induced decrease in cell proliferation.

EGF inhibit cell proliferation

Next we sought to clarify if the observed reduction in cell viabil-ity mediated by exposure to high levels of EGF was due to increasedapoptosis, necrosis or growth arrest. CFSE stained HN5 and A431cells were left untreated or treated with either low (0.1 nM) or high(10 nM) concentrations of EGF and the proliferation was measuredafter 0, 24, 48 and 72 hr (Fig. 4a). Apoptosis was measured 24 hrpost EGF exposure by staining cells with annexin V conjugated PEas described in Material and Methods. The results show that highconcentrations of EGF (10 nM) significantly reduce proliferation ofboth HN5 and A431 cells, but has no significantly effect on the num-ber of apoptotic cells (Fig. 4a and Table I). As expected low concen-trations of EGF (0.1 nM) increases proliferation in both cell lines(Fig. 4a). We visually investigated the effect of EGF for up to 72 hrby light microscopy and by counting viable and dead cells using try-pan blue exclusion (Figs. 4b and 4c). About 10 nM EGF signifi-cantly reduced the number of viable cells compared to the untreated

FIGURE 3 – (a) Western blot analysis showing levels of IRF-1 and phosphorylation status of STAT1, STAT3 and EGFR in HN5 cells treatedwith the indicated EGF concentrations for 1 hr. Tubulin was used as loading control. MTT assay using HN5 cells treated with the indicated EGFconcentrations for 72 hr. Bars indicate SEM. (b) MTT assay using HN5 cells transfected with either pCMVBL/IRF-1 (IRF-) or empty vector(control) for 48 hr. (c) MTT assay using HN5 cells treated with 10 nM EGF or with 10 nM EGF 1 0.1 lM AZD0530. Student’s t-test was per-formed on the data and statistically changes in cell viability is indicated in the figures by 3 asterisks *** (p < 0.001).

346 ANDERSEN ET AL.

control cells. Furthermore, the fraction of dead cells remains con-stant throughout the investigated time period, further demonstratingthat high EGF concentrations do not induce apoptosis in these cells.Remarkably, after 72 hr the cells treated with high EGF concentra-tions exhibited an altered morphology with more cell protrusionsand less cell–cell adhesion (more spreading) (Fig. 4c). This is a phe-notype that is typical of highly motile cells and indeed using woundhealing assays and we found that 10 nM EGF increased HN5 cellmotility, but with a delayed response of 36–72 hr after EGF treat-ment (data not shown).

EGF treated cancer cells activate immune effector cells

To get an indication of the in vivo relevance of the interferon-like response we investigated if media from EGF stimulated cells

were able to activate immune cells, isolated from peripheral blood,as compared to media from unstimulated A431 cells. WhenPBMC’s were incubated 72 hr with media from untreated or 10nM EGF treated A431 cells as described in Material and Methodsthe activation marker CD69 increased on both T- and B-cells. Thenumber of CD69 positive T-cells increased 7.9% whereas thenumber of CD69 positive B cells increased 36% (Fig. 4d) as com-pared to control treatment. These results indicate that cytokines,released from EGF stimulated A431 cells, are able to activateimmune effector cells.

Discussion

EGFR regulates expression of IRF-1, IL-6, GBP-2 and SOCS-3

We found that the 2 human cancer cell lines A431 and HN5induce expression of 4 selected genes (IRF-1, IL-6, GBP-2 andSOCS-3) from the IFN-associated module of genes in response toEGF treatment (Fig. 1b) showing that the module is tightly regu-lated by EGFR in these cancer cell lines. In both A431 and HN5,the expression of the module of genes is transiently induced inresponse to EGF treatment, peaking at 2 hr post stimulation inHN5 (Fig. 1b) and 4 hr in A431 (data not shown). This transientnature is presumably caused by internalization and degradation ofEGFR (Figs. 1f and 1g). Interestingly, SOCS-1 and SOCS-3 areamong the upregulated genes, and are both known to negativelyregulate STAT activation by EGFR,22 suggesting that SOCS-1

FIGURE 4 – (a) Proliferation index of CFSE stained HN5 and A431 cells left untreated or treated with either 0.1 nM EGF, 10 nM EGF or cellsleft untreated up to 72 hr and measured by flow cytometrial analysis in 5 independent experiments. (b) HN5 cells treated with 10 nM EGF or leftuntreated for up to 72 hr. A similar growth response was observed in the A431 cell line. (c) Cell counting from three independent experimentsof Trypan Blue stained HN5 cells treated with 10 nM EGF or left untreated for up to 72 hr. (d) Percent increase in CD69 positive B- and T-cellsgrown in media from EGF stimulated A431 cells as compared to CD69 positive B- and T-cells grown in media from unstimulated A431 cells.Student’s t-test was performed and statistically changes is indicated in the figures by 1 asterisk * (p < 0.03), 2 asterisks ** (p < 0.003) and 3asterisks *** (p < 0.001).

TABLE I – ANNEXIN V POSITIVE CELLS [%]

HN5 A431

Untreated 15.6 6 6.6 8.0 6 3.00.1 nM EGF 9.6 6 3.8 8.1 6 4.210 nM EGF 12.8 6 5.8 7.6 6 3.7

Apoptosis in annexin V conjugated PE stained HN5 and A431 cellstreated with either 0.1 nM EGF, 10 nM EGF or cells left untreated upto 72 h and measured by flowcytometry analysis in 4 independentexperiments.

347EGFR MEDIATED IRF-1 EXPRESSION THROUGH STAT ACTIVATION

and SOCS-3 also may be involved in limiting the expression ofthe module by decreasing the activation of STAT1 and STAT3 byEGFR.

We decided to focus our investigation of the IFN module onthe transcription factor IRF-1, since IRF-1 is a major regulatorof immune responses due to its ability to activate transcriptionof genes that primarily are regulated by IFNs.12,13 Furthermore,IRF-1 is a well-documented tumor suppressor gene.11,14,15 Onlyligand-activated EGFR was able to induce phosphorylation ofSTAT1 and STAT3 and expression of IRF-1, whereas cellsdevoid of EGFR (NR6) or cells expressing the naturally occur-ring constitutively active variant EGFRvIII (NR6M) failed toinduce STAT phosphorylation6 as well as expression of IRF-1(Figs. 1c and 1d). These effects may be explained by EGFR-vIII’s inability to bind EGF and thereby reach sufficient thresh-old levels of phosphorylation, which are needed to induce acti-vation of both STAT1 and STAT3 (Fig. 3a). We have previouslydemonstrated that both NR6M and also the EGFRvIII express-ing glioblastoma cell line U87MGvIII holds relatively low lev-els of EGFRvIII phosphorylation compared to EGF treatedEGFR expressing cells, and that neither of these cells are able toinduce activation of STAT1 and STAT3.6 Interestingly, ithas been reported that the EGFR-associated protein Grb2 regu-lates EGFR induced STAT3 activation negatively in an EGFconcentration dependent manner,23 suggesting that STAT3 onlycan be activated by EGFR when the amount of availableGrb2 molecules are low in the cytosol, which is caused by Grb2binding to activated EGFR. Thus, when EGFR is activated incells that overexpress EGFR it is likely that Grb2 becomesdepleted, increasing the probability of EGFR mediated STATactivation.

EGFR induced IRF-1 expression is Src dependent and regulatedby STAT1 and STAT3

Expression of IRF-1 correlated with phosphorylation of STAT1and STAT3 suggesting that EGFR induced IRF-1 expression ismediated through STAT activity. This was further supported bythe observation that both STAT1 and STAT3 bind to the knownSTAT-binding site (GAS) within the IRF-1 promoter in responseto EGF treatment (Fig. 2b). STAT1 has previously been associatedwith IRF-1 expression as STAT12/2 cells are unable to induceIRF-1 in response to IFN-g.20 Moreover, it has been demonstratedthat STAT1 and to a lower degree STAT3 activates transcriptionof GAS containing promoters,18 suggesting that both are activatorsof EGFR induced IRF-1 expression. Expression of IRF-1 was fur-ther found to be dependent on EGFR and Src tyrosine kinase ac-tivity as shown by the decreased expression in response to the spe-cific EGFR inhibitors AG1478 and Iressa, and the SRC familyinhibitors SU6656 and AZD0530. Furthermore, EGFR inducedactivation of STATs have previously been demonstrated to be par-tially dependent on Src activity,24 further signifying the regulationof IRF-1 by STAT1 and STAT3.

EGF induces IRF-1 and decrease cell viability

Another observation was that low concentrations of EGF (0.01–1nM) increased the number of viable cells while higher concentra-tions (2.5–25 nM) decreased the number of viable cells (Fig. 3a).Interestingly, high IRF-1 levels correlated with a decreased numberof viable cells both when IRF-1 was induced by EGFR (Figs. 3a and3b) and when IRF-1 was introduced by an IRF-1 encoding expres-sion plasmid (Fig. 3c). This correlation suggests that IRF-1 may beinvolved in limiting cell proliferation in response to high concentra-tions of EGF. Additionally, this support the role of IRF-1 as a nega-tive regulator of cell growth.11,14,15 Small interfering RNA (siRNA)is a powerful method to analyze the effect of a certain gene of inter-est. However, this method to block IRF-1 was not used, since siRNAactivates an IFN response when introduced to cells.25,26 Instead weshow that inhibition of Src activity and secondarily IRF-1 expression

significantly reduces the EGF induced growth inhibition (Fig. 3c),suggesting that IRF-1 is involved in the EGF induced growth inhibi-tion. This may also partly explain the growth inhibition in responseto EGF observed in other studies.6,21,27,28 About 10 nM EGF did notinduce any significantly changes in apoptosis between untreatedcells and cells treated with 0.1 nM EGF (Table I), but significantlyreduced proliferation of cells compared to untreated and 0.1 nMEGF treated cells, indicating that the negative effect on cell growthis due to cell cycle arrest (Figs. 4a and 4c), whereas low concentra-tion of EGF (0.1 nM) clearly stimulated cell proliferation in bothcell lines (Fig. 4a).

Role of EGFR induced IRF-1 and IFN-associated genes

IRF-1 recognizes and binds to the DNA elements IRF enhancers(IRF-Es)29 and/or IFN-stimulated response elements (ISREs),16-

motifs that appears within many IFN-stimulated genes, includinggenes from the module,30 suggesting that IRF-1 may be involvedin activation of these genes. Ligand-activated EGFR has previ-ously been reported to increase activity of ISRE containing pro-moters,27 an attribute that may well be explained by the inducedexpression of IRF-1 demonstrated in our study. It has beenreported that IRF-1 cooperate with the tumor suppressor p53–acooperation necessary for some cells to undergo apoptosis.31 Inter-estingly, both HN5 and A431 cell lines are known to carry muta-tions in the gene encoding p53,32,33 suggesting that IRF-1 may beunable to induce apoptosis because of the lack of functional p53.A number of studies have reported IFN-associated genes or ‘‘IFNsignatures’’ associated with human cancers.7–10 One recent studyrevealed that IFN-associated genes, including IRF-1, IL-6, SOCS-1 and CCL-10, were induced in human glioblastomas by activatedEGFR and not EGFRvIII,34 indicating that the regulation of theseIFN-associated genes not is cell type specific. A general under-standing is that cytotoxic T-cells and NK cells can eliminate ma-lignant cells by the recognition of tumor-associated antigens.Interestingly, IRF-1 is a key regulator of the cell surface expres-sion of major histocompatibility complex (MHC) class I and classII molecules,35,36 whose expression are crucial in recognition oftransformed cells. Other genes from the module have also beendemonstrated to play critical roles during the immune response.IL-6, which is frequently found associated with human cancers,regulates both T- and B-cell growth and differentiation.37 The che-mokine ligand 7 (CCL-7L)/MCP-3 has been demonstrated toattract and activate macrophages during inflammation and metas-tasis, and to reduce tumorigenicity through activation of naturalkiller (NK) cells and T lymphocytes.38 GBP-2 is a member of theIFN-g induced family of GTPases and has been demonstrated toreduce viral replication in NIH3T3 cells.39 Thus, this IFN-signa-ture is comprised of both internal and external effectors, which areinvolved in the immune response.

Relevance in vivo

We found that media from EGF stimulated A431 were able toincrease the number of activated T- and B cells compared to cellsincubated with media from untreated cells (Fig. 4d), indicatingthat products from the EGFR induced interferon-signature areinvolved in regulation of the immune system in vivo. Activated B-Cells, T-Cells, NK-Cells etc. are often found in neoplasticlesions,10,40 suggesting that immune cells may be recruited andactivated by cytokines released from the cancer cells themselves.Thus, a feasible explanation could be that these IFN-signaturesmay be part of a conserved mechanism, which is activated inresponse to altered signaling and that the products (cytokines,interferons, etc.) may alert the immune system leading toimproved recognition and elimination of transformed cells in vivo.

Conclusion

This study shows that cells acquiring sustained high activityof EGFR are able to activate a module of IFN-associated genes

348 ANDERSEN ET AL.

including IRF-1 through activation of STAT1 and STAT3.We provide evidence that one of the products, IRF-1,mediates growth arrest in vitro and could be responsible forthe inhibition of cell proliferation seen when cells are stimu-lated with high concentrations of EGF. Our results supportthe role of IRF-1 as a negative regulator of cell growth. The

relevance of this finding in vivo however needs to be investi-gated further.

Acknowledgement

Thanks to Dr. Jesper Kastrup for help with the FACS analysisof activated B- and T-cells.

References

1. Alroy I, Yarden Y. The ErbB signaling network in embryogenesis andoncogenesis: signal diversification through combinatorial ligand-re-ceptor interactions. FEBS Lett 1997;410:83–6

2. Normanno N, De Luca A, Bianco C, Strizzi L, Mancino M, Maiello MR,Carotenuto A, De Feo G, Caponigro F, Salomon DS. Epidermal growthfactor receptor (EGFR) signaling in cancer. Gene 2006;366:2–16.

3. Prenzel N, Fischer OM, Streit S, Hart S, Ullrich A. The epidermalgrowth factor receptor family as a central element for cellular signaltransduction and diversification. Endocrinol Relat Cancer 2001;8:11–31.

4. Pedersen MW, Meltorn M, Damstrup L, Poulsen HS. The type IIIepidermal growth factor receptor mutation. Biological significance andpotential target for anti-cancer therapy. Ann Oncol 2001;12:745–60.

5. Cohen DW, Simak R, Fair WR, Melamed J, Scher HI, Cordon-Cardo C.Expression of transforming growth factor-alpha and the epidermal growthfactor receptor in human prostate tissues. J Urol 1994;152:2120–4

6. Pedersen MW, Pedersen N, Damstrup L, Villingshøj M, Sønder SU,Rieneck K, Bovin LF, Spang-Thomsen M, Poulsen HS. Analysis ofthe epidermal growth factor receptor specific transcriptome: effect ofexpression level and an activating mutation. J Cell Biochem 2005;96:412–27.

7. Einav U, Tabach Y, Getz G, Yitzhaky A, Ozbek U, Amariglio N,Izraeli S, Rechavi G, Domany E. Gene expression analysis reveals astrong signature of an interferon-induced pathway in childhood lym-phoblastic leukemia as well as in breast and ovarian cancer. Oncogene2005;24:6367–75.

8. Fambrough D, McClure K, Kazlauskas A, Lander ES. Diverse signalingpathways activated by growth factor receptors induce broadly overlap-ping, rather than independent, sets of genes. Cell 1999;97: 727–41.

9. Hakansson P, Segal D, Lassen C, Gullberg U, Morse HC, III, FioretosT, Meltzer PS. Identification of genes differentially regulated by theP210 BCR/ABL1 fusion oncogene using cDNA microarrays. ExpHematol 2004;32:476–82.

10. Galon J, Costes A, Sanchez-Cabo F, Kirilovsky A, Mlecnik B,Lagorce-Pages C, Tosolini M, Camus M, Berger A, Wind P, Zinzin-dohoue F, Bruneval P, et al. Type, density, and location of immunecells within human colorectal tumors predict clinical outcome. Sci-ence 2006;313:1960–4.

11. Bouker KB, Skaar TC, Riggins RB, Harburger DS, Fernandez DR,Zwart A, Wang A, Clarke R. Interferon regulatory factor-1 (IRF-1)exhibits tumor suppressor activities in breast cancer associated withcaspase activation and induction of apoptosis. Carcinogenesis2005;26:1527–35.

12. Matsuyama T, Kimura T, Kitagawa M, Pfeffer K, Kawakami T, Wata-nabe N, Kundig TM, Amakawa R, Kishihara K, Wakeham A. Targeteddisruption of IRF-1 or IRF-2 results in abnormal type I IFN gene induc-tion and aberrant lymphocyte development. Cell 1993;75:83–97.

13. Harada H, Willison K, Sakakibara J, Miyamoto M, Fujita T, Tanigu-chi T. Absence of the type I IFN system in EC cells: transcriptionalactivator (IRF-1) and repressor (IRF-2) genes are developmentallyregulated. Cell 1990;63:303–12.

14. Kroger A, Dallugge A, Kirchhoff S, Hauser H. IRF-1 reverts thetransformed phenotype of oncogenically transformed cells in vitroand in vivo. Oncogene 2003;22:1045–56.

15. Nozawa H, Oda E, Nakao K, Ishihara M, Ueda S, Yokochi T, Ogasa-wara K, Nakatsuru Y, Shimizu S, Ohira Y, Hioki K, Aizawa S, et al.Loss of transcription factor IRF-1 affects tumor susceptibility in micecarrying the Ha-ras transgene or nullizygosity for p53. Genes Dev1999;13:1240–5.

16. Sims SH, Cha Y, Romine MF, Gao PQ, Gottlieb K, Deisseroth AB. Anovel interferon-inducible domain: structural and functional analysisof the human interferon regulatory factor 1 gene promoter. Mol CellBiol 1993;13:690–702.

17. Henderson YC, Chou M, Deisseroth AB. Interferon regulatory factor1 induces the expression of the interferon-stimulated genes. Br J Hae-matol 1997;96:566–75.

18. Van de Vijver MJ, Kumar R, Mendelsohn J. Ligand-induced activa-tion of A431 cell epidermal growth factor receptors occurs primarilyby an autocrine pathway that acts upon receptors on the surface ratherthan intracellularly. J Biol Chem 1991;266:7503–8.

19. Lamb P, Seidel HM, Haslam J, Milocco L, Kessler LV, Stein RB,Rosen J. STAT protein complexes activated by interferon-g andgp130 signaling molecules differ in their sequence preferences andtranscriptional induction properties. Nucl Acids Res 1995;23: 3283–9.

20. Durbin JE, Hackenmiller R, Simon MC, Levy DE. Targeted disrup-tion of the mouse Stat1 gene results in compromised innate immunityto viral disease. Cell 1996;84:443–50.

21. Kwok TT, Sutherland RM. Differences in EGF related radiosensitisa-tion of human squamous carcinoma cells with high and low numbersof EGF receptors. Br J Cancer 1991;64:251–4.

22. Xia L, Wang L, Chung AS, Ivanov SS, Ling MY, Dragoi AM, PlattA, Gilmer TM, Fu XY, Chin YE. Identification of both positive andnegative domains within the epidermal growth factor receptor COOH-terminal region for signal transducer and activator of transcription(STAT) activation. J Biol Chem 2002;277:30716–23.

23. Zhang T, Ma J, Cao X. Grb2 regulates Stat3 activation negatively inepidermal growth factor signalling. Biochem J 2003;376:457–64.

24. Xi S, Zhang Q, Dyer KF, Lerner EC, Smithgall TE, Gooding WE,Kamens J, Grandis JR. Src kinases mediate STAT growth pathways insquamous cell carcinoma of the head and neck. J Biol Chem 2003;278:31574–83.

25. Sledz CA, Holko M, de Veer MJ, Silverman RH, Williams BR. Acti-vation of the interferon system by short-interfering RNAs. Nat CellBiol 2003;5:834–9.

26. Bridge AJ, Pebernard S, Ducraux A, Nicoulaz AL, Iggo R. Inductionof an interferon response by RNAi vectors in mammalian cells. NatGenet 2003;34:263–4.

27. Kirchhoff S, Hauser H. Cooperative activity between HER oncogenesand the tumor suppressor IRF-1 results in apoptosis. Oncogene1999;18:3725–36.

28. Zhao X, Dai W, Zhu H, Zhang Y, Cao L, Ye Q, Lei P, Shen G. Epi-dermal growth factor (EGF) induces apoptosis in a transfected cellline expressing EGF receptor on its membrane. Cell Biol Int2006;30:653–8.

29. Tanaka N, Kawakami T, Taniguchi T. Recognition DNA sequencesof interferon regulatory factor 1 (IRF-1) and IRF-2, regulators of cellgrowth and the interferon system. Mol Cell Biol 1993;13:4531–8.

30. Harroch S, Revel M, Chebath J. Interleukin-6 signaling via four tran-scription factors binding palindromic enhancers of different genes.J Biol Chem 1994;269:26191–5.

31. Tanaka N, Ishihara M, Lamphier MS, Nozawa H, Matsuyama T, MakTW, Aizawa S, Tokino T, Oren M, Taniguchi T. Cooperation of thetumour suppressors IRF-1 and p53 in response to DNA damage. Nature1996;382:816–18.

32. Gusterson BA, Anbazhagan R, Warren W, Midgely C, Lane DP, O’HareM, Stamps A, Carter R, Jayatilake H. Expression of p53 in premalignantand malignant squamous epithelium. Oncogene 1991;6:1785–9.

33. Thompson-Hehir J, Davies MP, Green JA, Halliwell N, Joyce KA,Salisbury J, Sibson DR, Vergote I, Walker C. Novel polymerase chainreaction approach for full-coding p53 mutation detection in microdis-sected archival tumors. Diagn Mol Pathol 2000;9:110–19.

34. Ramnarain DB, Park S, Lee DY, Hatanpaa KJ, Scoggin SO, Otu H,Libermann TA, Raisanen JM, Ashfaq R, Wong ET, Wu J, Elliott R, etal. Differential gene expression analysis reveals generation of an auto-crine loop by a mutant epidermal growth factor receptor in gliomacells. Cancer Res 2006;66:867–74.

35. Hobart M, Ramassar V, Goes N, Urmson J, Halloran PF. IFN regula-tory factor-1 plays a central role in the regulation of the expression ofclass I and II MHC genes in vivo. J Immunol 1997;158:4260–9.

36. Chang CH, Hammer J, Loh JE, Fodor WL, Flavell RA. The activationof major histocompatibility complex class I genes by interferon regu-latory factor-1 (IRF-1). Immunogenetics 1992;35:378–84.

37. Horn F, Henze C, Heidrich K. Interleukin-6 signal transduction andlymphocyte function. Immunobiology 2000;202:151–67.

38. Wetzel K, Struyf S, Van Damme J, Kayser T, Vecchi A, Sozzani S, Rom-melaere J, Cornelis JJ, Dinsart C. MCP-3 (CCL7) delivered by parvovirusMVMp reduces tumorigenicity of mouse melanoma cells through activa-tion of T lymphocytes and NK cells. Int J Cancer 2007;120:1364–71.

39. Carter CC, Gorbacheva VY, Vestal DJ. Inhibition of VSV and EMCVreplication by the interferon-induced GTPase, mGBP-2: differentialrequirement for wild-type GTP binding domain. Arch Virol 2005;150:1213–20.

40. Giatromanolaki A, Koukourakis MI, Ritis K, Mimidis K, Sivridis E.Interferon regulatory factor-1 (IRF-1) suppression and derepression dur-ing endometrial tumorigenesis and cancer progression. Cytokine 2004;26:164–8.

349EGFR MEDIATED IRF-1 EXPRESSION THROUGH STAT ACTIVATION