identiﬁcation of the cell-intrinsic and -extrinsic ... · fernando concha-benavente1, raghvendra...

Microenvironment and Immunology

Identification of the Cell-Intrinsic and -ExtrinsicPathways Downstream of EGFR and IFNg ThatInduce PD-L1 Expression in Head andNeck CancerFernando Concha-Benavente1, Raghvendra M. Srivastava2, Sumita Trivedi2, Yu Lei3,Uma Chandran4, Raja R. Seethala5, Gordon J. Freeman6, and Robert L. Ferris1,2,7

Abstract

Many cancer types, including head and neck cancers (HNC),express programmeddeath ligand 1 (PD-L1). Interaction betweenPD-L1 and its receptor, programmed death 1 (PD-1), inhibits thefunction of activated T cells and results in an immunosuppressivemicroenvironment, but the stimuli that induce PD-L1 expressionare not well characterized. Interferon gamma (IFNg) and theepidermal growth factor receptor (EGFR) utilize Janus kinase 2(JAK2) as a common signaling node to transmit tumor cell–mediated extrinsic or intrinsic signals, respectively. In this study,we investigated the mechanism by which these factors upregulatePD-L1 expression in HNC cells in the context of JAK/STAT

pathway activation, Th1 inflammation, andHPV status.We foundthat wild-type, overexpressed EGFR significantly correlated withJAK2 and PD-L1 expression in a large cohort of HNC specimens.Furthermore, PD-L1 expression was induced in an EGFR- andJAK2/STAT1-dependent manner, and specific JAK2 inhibitionprevented PD-L1 upregulation in tumor cells and enhancedtheir immunogenicity. Collectively, our findings suggest a novelrole for JAK2/STAT1 in EGFR-mediated immune evasion, andtherapies targeting this signaling axis may be beneficial toblock PD-L1 upregulation found in a large subset of HNC tumors.Cancer Res; 76(5); 1031–43. �2015 AACR.

IntroductionCancer immunoediting implies that lymphocytes successfully

suppress tumor growth (1). However, tumor cell immune escapecan eventually occur by downregulating HLA class I antigenprocessing (2) or by providing inhibitory signals (3, 4). Pro-grammed death-1 (PD-1) is an immune checkpoint receptorexpressed by tumor infiltrating lymphocytes (TIL), which limitsthe function of activated T lymphocytes (3, 5). Its cognate ligand,programmed death ligand-1 (PD-L1) is expressed in many typesof cancers (5). Indeed, trials targeting the PD-L1/PD-1 pathwaywith blocking antibodies show encouraging results where tumorPD-L1 expression enriches for clinical responders (6–10). Theincidence of HPVþ tumors is rapidly increasing (11) and thesetumors are more responsive to oncologic therapy, which may be

in part immune mediated (12–14). A previous report suggestedthat PD-L1 expression contributed to immune resistance inHPVþ

HNC (15). In contrast, PD-1þ CD8þ T cells with an activatedphenotype may be a favorable prognostic biomarker in HPVþ

patients (16). Therefore, the stimuli and signaling pathways thatinduce PD-L1 expression in HNC may permit more effectivetherapeutic approaches.

PD-L1 expression in tumor cellsmay be regulated by twomajormechanisms. First, an "extrinsic" mechanismwhere an antitumorcellular immune response driven by natural killer cells (NK) andCD8þ TIL produce IFNg , which in turn may induce PD-L1expression on tumor cells. Second, an "intrinsic"mechanismmayexist in which constitutive oncogenic signaling pathways withinthe tumor cell lead to PD-L1 overexpression. In glioblastoma,PTEN deletion promotes PI3K–AKT-mediated PD-L1 overexpres-sion (17), while EGFR-mutant lung cancer cells have been asso-ciated with PD-L1 overexpression (18, 19). In contrast to lungcancer, in the setting of HNC, EGFRmutations are extremely rare,whereas wild-type EGFR is overexpressed in approximately 80%to 90% of tumors (20). We hypothesized that targeting signalingmolecules involved in PD-L1 expression in HNC cells mightsynergize with current anti-EGFR antibody targeted immu-notherapies such as cetuximab, that are known to activate NKcells and cytotoxic T lymphocytes (CTL; ref. 21). However,because activated NK and T cells secrete IFNg , a known stimulusfor PD-L1 expression, understanding the complex signaling path-ways regulating PD-L1 is crucial. Because extrinsic (IFNg-medi-ated) and intrinsic (EGFR-mediated) mechanisms may cooperateto promote PD-L1 upregulation, we investigated signaling path-ways that mediate both IFNg and EGFR induced PD-L1 upregula-tion in tumor cells. These findings have relevance for immuno-therapeutic combinations with cetuximab, which can both blockEGFR signaling and stimulate IFNg secretion via activation of NKcells and CTL (21–23).

1Department of Immunology, University of Pittsburgh, Pittsburgh,Pennsylvania. 2Department of Otolaryngology, University of Pitts-burgh, Pittsburgh, Pennsylvania. 3Department of Periodontics andOral Medicine, School of Dentistry and Department of Otolaryngolo-gy–Head and Neck Surgery, School of Medicine. University of Michi-gan, Ann Arbor, Michigan. 4Department of Biomedical Informatics,University of Pittsburgh, Pittsburgh, Pennsylvania. 5Department ofPathology, University of Pittsburgh, Pittsburgh, Pennsylvania.6Department of Medical Oncology, Harvard Medical School, DanaFarber Cancer Institute, Boston, Massachusetts. 7Cancer ImmunologyProgram, University of Pittsburgh Cancer Institute, Pittsburgh,Pennsylvania.

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

Corresponding Author: Robert L. Ferris, Hillman Cancer Center ResearchPavilion, 5117 Centre Avenue, Room 2.26b, Pittsburgh, PA 15213. Phone: 412-623-0327; Fax: 412-623-4840; E-mail: [email protected]

doi: 10.1158/0008-5472.CAN-15-2001

�2015 American Association for Cancer Research.

CancerResearch

www.aacrjournals.org 1031

on March 4, 2016. © 2016 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst December 16, 2015; DOI: 10.1158/0008-5472.CAN-15-2001

http://cancerres.aacrjournals.org/

Materials and MethodsTumor cell lines

HPV� HNC cell lines (JHU020, JHU022, JHU029, PCI13) andHPVþ (SCC2, SCC47, SCC90, 93VU) were cultured in IMDMcomplete media (Mediatech). JHU020, JHU022, and JHU029were a gift from Dr. James Rocco in January of 2006 (Ohio StateUniversity). SCC90 and PCI13were isolated frompatients treatedat the University of Pittsburgh through the explant/culture meth-od, authenticated and validated using STTR profiling and HLAgenotyping (24, 25). SCC2 and SCC47 were a kind gift from Dr.Thomas Carey inDecember of 2005 (University ofMichigan) and93VUagift fromDr.HenningBier inOctober of 2013 (TechnischeUniversitat Munchen, Germany). HNC lines were tested every 6months and were free of Mycoplasma infection.

Antibodies and treatmentsThe anti-PD-L1 mAb (clone 405.9A11) was previously validat-

ed (26) by Dr. Gordon J. Freeman (Dana-Farber Cancer Institute,Boston, MA), anti-pJAK2 (Y1007 and Y1008; Abcam) were usedfor IHC staining. The anti-PD-L1-PE (BD Pharmingen) anti-HLA-ABC-FITC mAb (clone w6/32, eBioscience), IgG1-PE isotypecontrol, pSTAT1 Tyr701-PE, STAT1-PE and STAT3-PE (BD Bios-ciences), phospho-AKT (Thre308)–PE and primary anti-p44/42MAPK(Erk1/2) and phospho-p44/42 MAPK(Erk1/2) (Thr202/Tyr204) and secondary anti-rabbit-PE antibodies (Cell SignalingTechnology). WB antibodies included rabbit anti-human JAK2,pJAK2 (y1007/1008), total AKT, pAKT, pERK and mouse anti-human b-actin (Cell Signaling Technology). IFNg (R&D Systems)wasused at 10 IU/mL. IFNa2a (PBL Interferon Source)wasused at1,000 IU/mL. The JAK2 inhibitor BMS-911543 was characterizedpreviously (25), provided by Bristol-Myers Squibb and used at 10mmol/L. JAK1/3 inhibitor (ZM39923; Tocris Bioscience) wascharacterized previously (27, 28) and was used at 10 mmol/L.Wortmannin (Cell Signaling Technology) was used at 1 mmol/L.PI3Ka110 subunit inhibitor (BYL-719) was used at 1 mmol/L andMEK1/2 inhibitor (PD0325901) was used at 5 mmol/L (TocrisBioscience).

Flow cytometry analysisCell viability was determined by Zombie Aqua staining (Biole-

gend; ref. 29), then incubated with fluorophore-conjugated anti-bodies at 1:10 dilution for 15 minutes at 4�C, then washed twiceand resuspended in 2% PFA solution until analysis. Intracellularflow cytometry was performed (30) and analyzed using an LSRFortessa cytometer (BD Biosciences), and FlowJo version 10 soft-ware. Median fluorescence intensity (MFI) fold change was calcu-lated by normalizing after subtracting the isotype control MFI.

Western blottingWestern blotting was performed as described previously (2)

using the indicated antibodies.

siRNA knockdownCellswere transfected as described elsewhere (2). After 48hours

of transfection, cells were incubated �IFNg (10 IU/mL) or EGF(10 ng/mL) for 48 hours, then harvested and analyzed by flowcytometry. siRNA STAT1: 5-CUACGAACAUGACCCUAUTT-3(s)and 5-AUAGGGUCAUGUUCGUAGGTG-3(as) siRNA STAT3: 5-GCCUCAAGAUUGACCUAGATT-3(s) and 5-UCUAGGUCAAU-CUUGAGGCCT-3(as) and siRNA nontargeting 5-AGUACAG-

CAAACGAUACG Gtt-3 control: (s) and 5-CCGUAUCGUUUG-CUGUACUtt-3(as).

Quantitative PCRqPCR was performed as described previously (31). PCR probes

for PD-L1 (Hs01125301_m1) and GUSB (Hs99999908_m1;Applied Biosystems for TaqMan Gene Expression Assay). GUSBwas amplified as an internal control. Relative expression of thetarget genes to GUSB control gene was calculated using the DCT

method: relative expression ¼ 2�DCT, where DCT ¼ CT (target

gene) � CT (GUSB).

Chromatin immunoprecipitation assayCells were serum starved for 18 hours prior to incubation with

IFNg (10 IU/mL) for 30 minutes at 37�C or sequentially withcetuximab (10 mg/mL) for 30 minutes at 37�C. Then chromatinimmunoprecipitation (ChIP) assay was performed as describedpreviously (2) using the Ez-ChIP kit (Millipore). Purified DNAwas used in quantitative RT-PCR using the EpiTect ChIP qPCR(Qiagen) SYBR-Green Master Mix method using a primer for PD-L1 promoter NM_014143.2 (�)16Kb. qPCR amplification datawere normalized and analyzed as a percentage of input (32) andexpressed as relative enrichment to percentage of input.

Immunohistochemistry protocolThe University of Pittsburgh IRB #99-069 approved the use of

clinical samples and written informed consent was obtained.Slides were deparaffinized and rehydrated. Antigen retrieval wasperformed using Diva Retrieval (Biocare Medical) and a decloak-ing chamber at 124�C, 3 minutes, and cooled for 10 minutes.Slides were placed on an Autostainer Plus (Dako) using a TBSTrinse buffer (Dako) and stained using: 3% H2O2 (ThermoFisherScientific) for 5 minutes, CAS Block (Invitrogen) for 10 minutes,the primary antibody for PD-L1 (clone 405.9A11; ref. 33) and thepJAK2 (Y1007-1008; clone E132) were used per instructions. Thesecondary consisted of Envision Dual Link þ (Dako) polymerfor 30 minutes, rinsed, then a TBST holding rinse was applied for5 minutes. The substrate used was 3,3, Diaminobenzidine þ(Dako) for 7 minutes and counterstained with hematoxylin.PD-L1 and pJAK2 staining were quantified by positive pixel countv9 algorithm (Aperio). A head and neck pathologist blinded toclinical patient data examined tumor sections. Scoring was deter-mined by the percentage of tumor stained for PD-L1 or pJAK2,respectively. Tumors with <5% tumor cell–positive staining wereconsidered negative.

Cellular cytotoxicity assaysCytotoxicity was determined by a 51Cr release assay as

described before (34). Tumor targets and NK cells were coincu-bated with cetuximab or IgG1 isotype (10 mg/mL) for 4 hours.Controls for spontaneous and maximal lysis were included.Specific lysis ¼ (experimental lysis � spontaneous lysis)/(exper-imental lysis � maximal lysis) � 100.

The Cancer Genome Atlas data retrieval and analysisRNAseq data from queried genes were downloaded from the

UCSC cancer genomics browser (https://genome-cancer.ucsc.edu). TheHNC gene expression profile from 500HNC specimenswas measured experimentally (35). The RSEM units to quantitateRNAseq expression data were described previously (36). Correla-tions fromTheCancerGenomeAtlas (TCGA)datawere calculated

Concha-Benavente et al.

Cancer Res; 76(5) March 1, 2016 Cancer Research1032




using Pearson r test and linear regression curve fits were graphedusing GraphPad PRISM software v6.

IPA Ingenuity pathway software analysisSoftware was accessed via the University of Pittsburgh license.

Path Explorer tool available in IPA Ingenuity was utilized forassociating EGFR and IFNg pathways with PD-L1 expression.

ResultsPD-L1 protein expression is higher in HPVþ HNC tumorspecimens

IHC stainingof tumor specimens (n¼134) revealed that 59.7%of HNC patients express detectable PD-L1 on the tumor cellsurface, as determined by >5% positive tumor cells (Fig. 1A).Furthermore, when segregated by HPV status (n ¼ 127, 63 HPV�

and64HPVþ),wenoted that PD-L1 expressionwasmore frequent

in HPVþ specimens (70% vs. 43.3%, respectively, Fig. 1B) and thepercentage of PD-L1 expression was also significantly higher in inHPVþ tumors (Fig. 1C). Interestingly, PD-L1 expressionwasmoreintense on the cell membrane than in the cytoplasm and washeterogeneously expressedwithin themicroenvironment, generallyforming clusters of PD-L1þ tumor cellswithahigher intensity at theclusterperiphery (Fig. 1DandE). To study the stimuli andpathwaysby which PD-L1 is upregulated in vitro, we analyzed a panel ofHPVþ and HPV� HNC cell lines for PD-L1 expression, which wasexpressed variably (Fig. 1F) similar to patient tumors by IHC.

HPVþ tumor specimens show higher Th1-type expressionprofile

We then analyzed PD-L1 expression in a large cohort of HNCspecimens for which gene expression TCGA repository data wereavailable (35). Because PD-L1 expression has been linked with that

C D

A

4X 20X

7,0006,0005,0004,0003,0002,0001,0000

Isotype controlJHU020JHU022JHU029

PCI13SCC2

SCC47SCC90

93VU

PD-L1-MFI

HPV+

HPV-

F

B

40.3%59.7%

ALL

PD-L1 NegativeTotal = 134

PD-L1 Positive

**

30%

70%

HPV +HPV –

HPV +HPV –

Total = 127PD-L1 NegativePD-L1 Positive

E

% T

umor

PD

L1+

100

90

80

70

60

50

40

30

20

10

0

57.7% 43.3%

Figure 1.PD-L1 protein expression is higher inHPVþ tumor specimens. A, PD-L1protein expression in HNC tumorspecimens (IHC, n ¼ 134). Tumorswere considered positive for PD-L1when higher than 5% tumor stainingthreshold. 59.7% of tumors were PD-L1þ. B, PD-L1 expression in HPV� andHPVþ tumor specimens using thesame criteria as in A. 70%HPVþ versus43.3% HPV� specimens were PD-L1þ

C. percent PD-L1þ tumor area in HPV�

and HPVþ specimens. Dotted line, 5%tumor positive; solid lines, median.(Mann–Whitney, �� , P < 0.001.)D, representative high intensity,100% PD-L1þ, HPVþ tumor. E,representative low intensity, 50%PD-L1þ HPV� tumor. Insets on the leftrepresent magnification (�20) on theright. F. HNC cells expressheterogeneous levels of PD-L1.

EGFR/IFNg-Mediated PD-L1 Expression Is JAK2–STAT1 Dependent

www.aacrjournals.org Cancer Res; 76(5) March 1, 2016 1033




A

E

B

PD-L

1

PD-1

CD

8A

IFN

G

JAK

1

JAK

2

STA

T1

EGFR

PI3K

CA

TOR

C1

4EB

P1

MA

PK1

0

20

40

60

80

100

RN

A e

xpre

ssio

n (R

SEM

uni

ts) HPV –

HPV +

**** ** *

806040200 80604020070

80

90

100

110

PD-L1 (RSEM units) PD-L1 (RSEM units)

EGFR

(RSE

M u

nits

)

HPV NegativePearson r = 0.2424

P = 0.0249 (*)Y = 0.2847*X + 11.73

EGFR

(RSE

M u

nits

)

C

D

8060402000

10

20

30

40

PD-L1 (RSEM units)

IFN

g (R

SEM

uni

ts)

HPV NegativePearson r = 0.5853

P = 0.0001 (***)Y = 0.4157*X + 1.540

8060402000

20

40

60

PD-L1 (RSEM units)

IFN

g (R

SEM

uni

ts)

HPV PositivePearson r = 0.8621

P = 0.0001 (***)Y = 0.7008*X + 3.944

HPV PositivePearson r = 0.4995

P = 0.0090 (**)Y = 0.4500*X – 1.464

0 20 40 60 80 0 20 40 60 800

20

40

60

80

100

PD-L1 (RSEM units) PD-L1 (RSEM units)

JAK

2 (R

SEM

uni

ts)

HPV Negative HPV PositivePearson r = 0.7051 Pearson r = 0.7716

P = 0.0001 (***) P = 0.0001 (***)Y = 0.6899*X + 38.13Y = 0.7051*X + 26.14

JAK

2 (R

SEM

uni

ts)

F

150

100

50

0

150

100

50

0

1 2 3

PD-L1

pJAK2

Figure 2.HPVþ specimens show higher expression of a Th1-type RNA expression profile and PD-L1 expression correlates with that of JAK2, EGFR, and IFNg regardless ofHPV status. A, heatmapofRNAseq expression expressed asRSEMunits (as described inMaterials andMethods)of PD-L1, PD-1, CD8A, CD8B, IFNg , JAK2, JAK1, STAT1,EGFR, PIK3CA, TORC1, 4EBP1, and MAPK1 (66 HPV� and 22 HPVþ) from TCGA database (35); red boxes emphasize higher expression of a Th1 profile in HPVþ

specimens and higher EGFR expression in HPV� counterparts (yellow, 10-fold higher; turquoise, 10-fold lower relative expression change over black).B, HPVþ tumor specimens show significantly higher expression of a Th1-type expression profile: PD-1, CD8A, IFNg , and JAK2. (Continued on the following page.)






ofCD8and IFNg (15,23),we investigated theTh1mRNAexpressionprofile of HPVþ versus HPV� specimens. Pooled data from 88HNCspecimens were plotted using a heat map, segregated by HPV status(Fig. 2A, redboxesdepict higher expression inHPVþ tumors). ATh1-type expression profile (PD-1, CD8A, CD8B, IFNG, and JAK2) wassignificantly higher inHPVþ thanHPV� tumors (Fig. 2B), suggestingthat activated immune effector cells readily infiltrate HPVþ tumors,which may be important for PD-L1 induction due to this source ofIFNg . Importantly, JAK2 expression (but not JAK1) was also higherinHPVþ tumors (Fig. 2B). Therefore, JAK2was associatedwith a Th1profile and with PD-L1 expression, particularly in HPVþ tumors.

PD-L1 expression correlates with that of JAK2, EGFR, IFNg, anda Th1 profile regardless of HPV status

Given that JAK2 is a common signaling molecule downstreamof the EGFR and IFNg pathways, we found that PD-L1 and JAK2mRNA expression were highly correlated (n¼ 500) and persistedwhen the cohort was segregated by HPV status (Fig. 2C). In orderto further assess the relationship between JAK2 and PD-L1 in vivoat the protein level, we determined phospho-JAK2 and PD-L1using IHC from adjacent sections of HNC specimens (n ¼ 23).Corroborating our previous findings, PD-L1 was predominantlyexpressed on the tumor cell membrane while phospho-JAK2exhibited strong nuclear staining, with occasional weak-moderatecytoplasmic staining. PD-L1–positive tumor islands were foundto be strongly positive for phospho-JAK2 (Fig. 2D). Furthermore,we also found a significant correlation between EGFR and PD-L1expression, which was somewhat weaker in HPV� tumors (Fig.2E). Likewise, PD-L1 expression was highly correlated with a Th1-type expression profile (IFNg , CD8A, andPD-1) regardless ofHPVstatus (Fig. 2F and Table 1). In addition, correlation of EGFR andCD8or JAK2was only significant inHPVþ tumors, given that theirexpression level was higher than in HPV� tumors. However, thisfinding did not preclude the fact that JAK2 could also be impor-tant for PD-L1 expression in HPV� tumors given that they werestrongly correlated regardless of HPV status.

STAT1, but not STAT3, PIK3CA, orMAPK1, expression is higherin tumor tissue and strongly correlates with PD-L1, EGFR, andIFNg regardless of HPV status

Because PD-L1 expression strongly correlated with a Th1-typeexpression profile in the tumor microenvironment, we hypoth-

esized that PD-L1maydependonSTAT1 activation, a knownTh1-type transcription factor. Indeed, STAT1 emerged as one of thehighly predicted transcription factors binding to the PD-L1 pro-moter region and common to EGFR and IFNg pathways whenutilizing previously validated software for transcription factorbinding prediction (MATCH) and pathway exploration (Ingenu-ity IPA; ref. 37). Given that previous reports presented STAT3,PI3K, and MAPK as possibly involved in PD-L1 expression inother tumor types, we included these in our investigation. Wepooled RNAseq data collected from46paired specimens of tumorversus matched normal mucosa and found that STAT1 (but notSTAT3, PIK3CA, or MAPK1) was significantly upregulated intumor tissue (Fig. 3A). Furthermore, STAT1 expression highlycorrelated with that of PD-L1 in TCGA (n ¼ 500), which waspreserved when segregated by HPV status (Fig. 3B). Concordantwith TCGA data, we found that STAT1 protein was widelyexpressed in HNC tumor tissues, and that PD-L1–positive tumorislands were also strongly positive for total STAT1 staining (Fig.3C, circled areas highlight colocalization, 100� inset). Interest-ingly, STAT1 expression also showed strong correlation with thatof EGFR (Fig. 3D). As expected, STAT1 tumor expression was alsostrongly correlated with that of IFNg (Fig. 3E). Notably, STAT3and PI3K pathway components (AKT1, TORC1, or 4EBP1)showed no correlation with PD-L1 in HPVþ tumors and onlyweakly in the HPV� HNC. Likewise, MAPK1 was not correlatedwith PD-L1 expression in either cohort (Table 2). Overall, ourfindings suggest that the JAK2/STAT1 pathway may serve as animportant commonmediator for both EGFR- and IFNg-mediatedPD-L1 expression in HNC tumors, regardless of HPV status.

IFNg-mediated PD-L1 upregulation is JAK2/STAT1 dependentBased on our TCGA analysis and previous reports linking

IFNg with PD-L1 expression at the mRNA level, we investigatedthe signaling pathway by which IFNg upregulates PD-L1expression in vitro. Indeed, a panel of HPVþ and HPV� HNCcell lines upregulated PD-L1 expression after IFNg treatment(Fig. 4A). Given that IFNg-mediated PD-L1 upregulation waslinked with PI3K pathway activation (17), we tested whetherwortmannin (pan-PI3K inhibitor) or BYL-719 (PI3Ka110 sub-unit specific inhibitor) could prevent IFNg-mediated PD-L1upregulation. PI3K pathway inhibition did not induce PD-L1downregulation, under conditions in which these inhibitors

Table 1. Correlation of PD-L1, EGFR, and IFNg with a Th1 profile in HPV-negative and HPV-positive tumors

HPV negative HPV positiveCorrelation (XY) Pearson r P Pearson r P

PD-L1 vs. PD-1 0.5937 <0.0001 (��) 0.7538 <0.0001 (��)PD-L1 vs. CD8A 0.5157 <0.0001 (��) 0.8363 <0.0001 (��)EGFR vs. CD8A �0.01368 0.4566 (ns) 0.4705 0.0136 (�)EGFR vs. JAK2 0.1853 0.0681 (ns) 0.438 0.0207 (�)IFNG vs. CD8A 0.7747 <0.0001 (��) 0.9396 <0.0001 (��)IFNG vs. JAK2 0.7264 <0.0001 (��) 0.7391 <0.0001 (��)NOTE: PD-L1 mRNA expression highly correlated with that of PD-1 and CD8A regardless of HPV status. In addition, EGFR expression correlated with that ofCD8A or JAK2 in HPVþ but not HPV� tumors. As expected, IFNg showed a strong correlation with that of CD8A and JAK2 regardless of HPV status. �, P < 0.05;�� , P < 0.0001.Abbreviation: ns, nonsignificant.

(Continued.) EGFR expression is significantly higher in HPV� tumor specimens (Mann–Whitney; � , P < 0.05; �� , P < 0.001). C, PD-L1 expression significantlycorrelated with JAK2 in both HPV� and HPVþ specimens (Pearson r and linear regression curve fit; �� , P < 0.0001). D, PD-L1 and pJAK2 IHC staining in adjacentsections and matching areas of HNC specimens. PD-L1 (top) was predominantly expressed on the tumor cell membrane. Phospho-JAK2 (bottom) exhibits strongnuclear staining with occasional weak–moderate cytoplasmic staining. PD-L1–positive tumor islands are also diffusely strongly positive for phospho-JAK2(three representative specimens out of 23). E, PD-L1 mRNA expression significantly correlated with EGFR in both HPV� and HPVþ specimens (Pearson r and linearregression curve fit; � , P < 0.05; �� , P < 0.001). F, PD-L1 mRNA expression significantly correlated with IFNg regardless of HPV status (Pearson r and linear regressioncurve fit; ��, P < 0.0001). Sixty-six HPV� and 22 HPVþ tumor specimens were collected from the TCGA database.






AC

D

0 20 40 60 8085

90

95

100

105

110

PD-L1 (RSEM units)

EGFR (RSEM units)

IFNg (RSEM units) IFNg (RSEM units)

EGFR (RSEM units)

PD-L1 (RSEM units)

STA

T1 (R

SEM

uni

ts)

STA

T1 (R

SEM

uni

ts)

STA

T1 (R

SEM

uni

ts)

STA

T1 (R

SEM

uni

ts)

STA

T1 (R

SEM

uni

ts)

HPV Negative HPV Positive

HPV PositiveHPV Negative

HPV PositiveHPV Negative

Pearson r = 0.4349P = 0.0003 (***)

Y = 0.1418*X + 91.46

0 20 40 60 8080

90

100

110

120

STA

T1 (R

SEM

uni

ts)

Pearson r = 0.5793P = 0.0047 (**)

Y = 0.1715*X + 90.85

B

E

20 40 60 80 100 12080

85

90

95

100

105Pearson r = 0.3023

P = 0.0068 (**)Y = 0.1157*X + 86.37

0 20 40 60 80 100 12070

80

90

100

110Pearson r = 0.7644

P < 0.0001 (***)Y = 0.2039*X + 80.23

0 10 20 30 40 5080

90

100

110Pearson r = 0.6663

P = 0.0007 (**)Y = 0.2426*X + 91.85

0 10 20 30 4085

90

95

100

105

110Pearson r = 0.6057

P < 0.0001 (***)Y = 0.2780*X + 92.86

STAT1 STAT3 PIK3CA MAPK10

3,000

6,000

9,000

12,000

15,000

RSE

M u

nits

Matched controlTumor

* * *

4X

20X

100X

i

ii

i

ii

Figure 3.STAT1, but not STAT3, PI3KCA, or MAPK1, expression is higher in tumor tissue and strongly correlates with that of PD-L1, EGFR, and IFNg regardless of HPVstatus. A, expression of STAT1, but not STAT3, PIK3CA, or MAPK1, was significantly higher in tumor specimens when compared with matched controlmucosa (TCGA, 46 HNC tumor specimens and matched controls, Mann–Whitney; �� , P < 0.0001). B, STAT1 expression is strongly correlated with PD-L1regardless of HPV status (Pearson r and linear regression curve fit; �� , P < 0.001; �� , P < 0.0001). C, PD-L1þ tumor islands are also strongly positive for STAT1protein in HNC specimens. Representative section of a HNC specimen costained for PD-L1 (brown chromogen) and STAT1 (red chromogen). Insetsrepresent magnification of the tumor area; yellow circles, colocalization. D and E, STAT1 expression correlated with that of EGFR and IFNg regardless of HPVstatus. (Pearson r and linear regression curve fit; �� , P < 0.001; �� , P < 0.0001).






effectively prevented AKT phosphorylation (SupplementaryFig. S1A–S1D). Because IFNg signals via JAK1 and JAK2, weutilized a clinical grade, selective JAK2 inhibitor BMS-911345(JAK2i), which was previously characterized (38), finding anabrogation of IFNg-mediated PD-L1 upregulation in all cell linestested, both at the mRNA and protein level (Fig. 4A and B;Supplementary Fig. S1E and S1F). Interestingly, specific JAK1/3inhibition (JAK1/3i) did not show a significant downregulationof IFNg-mediated PD-L1 protein expression (Fig. 4C). We thenused IFNa, which signals via JAK1 and TYK2, but not JAK2. IFNatreatment did not upregulate PD-L1 expression (Fig. 4D) but stillinduced pSTAT1 upregulation, although to a lesser extent thanIFNg , in all cell lines tested. Moreover, JAK2 inhibition did notaffect HLA-ABC upregulation (Supplementary Fig. S2A andS2B), which suggests that the kinetics of IFNa-induced pSTAT1binding to the PD-L1 promoter differ for HLA-ABC.

In order to determine whether the IFNg-mediated PD-L1upregulation was solely STAT1 dependent, we silenced eachtranscription factor using siRNA technology (80%–90% knock-down efficiency; Supplementary Fig. S3A). STAT1 but notSTAT3 knockdown potently impaired IFNg-mediated upregu-lation of PD-L1 (Fig. 4E). Moreover, ChIP assays documentedthat pSTAT1 but not pSTAT3 binds to the promoter region ofPD-L1 after IFNg treatment (Fig. 4F). Interestingly, cetuximab-mediated EGFR blockade downregulated IFNg-induced pSTAT1binding to the PD-L1 promoter and significantly downregu-lated the IFNg-mediated PD-L1 upregulation at the mRNA andprotein level, respectively (Fig. 4G and H and SupplementaryFig. S3B).

EGFR-mediated PD-L1 upregulation is JAK2/STAT1 dependentBecause EGFR strongly correlated with PD-L1 expression in

TCGA specimens and a previous report showed that EGFR acti-vating mutations induce PD-L1 in lung cancers (18), we hypoth-esized thatwild type EGFR, overexpressed in 80% to90%ofHNC,may promote PD-L1 upregulation. EGF treatment induced upre-gulation of PD-L1 protein in 7 of 8HNC lines studied, though to alesser extent than that induced by IFNg (Fig. 5A). This effect wasalso seen at the mRNA level (Fig. 5B). Although EGFR activatesmultiple downstream pathways, including the PI3K, MAPK, andJAK/STAT pathway, TCGA analysis yieldedweak if any correlationbetween PD-L1 and PIK3CA or MAPK1 (Table 2). However, astrong correlation was observed with that of JAK2 and STAT1(Figs. 2C and 3B, respectively). Given that JAK2 serves as acommon signaling molecule for both IFNg and EGFR pathways,we investigated whether EGF-mediated PD-L1 upregulation wasJAK2 and/or STAT1 dependent. Indeed, basal expression of PD-L1in HNC cell lines was downregulated by JAK2 but not JAK1/3inhibition (Fig. 5C). Furthermore, EGF induced JAK2 phosphor-ylation (Supplementary Fig. S4A) and upregulation of basal

PD-L1 expression (Fig. 5D). Additionally, specific JAK2, but notJAK1/3, inhibition prevented EGF induced PD-L1 upregulation(Fig. 5D and Supplementary Fig. S4B). Interestingly, the EGF-mediated PD-L1 upregulation was higher in cell lines with ahigher EGFR expression (JHU029 and JHU022 vs. 93VU andSCC90). Likewise, JAK2 inhibition more strongly downregulatedbasal and EGF-mediated PD-L1 expression in the EGFRhigh celllines (Fig. 5D; JHU022 and JHU029).

Because EGFR activates PI3K and MAPK pathways, we testedwhether these mediated PD-L1 upregulation after EGFR stim-ulation. We found that neither wortmannin-mediated PI3Kinhibition nor MEK1/2-mediated MAPK inhibition preventedEGF-induced PD-L1 upregulation (Supplementary Fig. S4C andS4D). However, these inhibitors effectively suppressed AKT andERK phosphorylation, respectively (Supplementary Fig. S4Eand S4F). In light of this result and the positive correlationfound between EGFR and STAT1, we hypothesized that EGFmay be activating STAT1 phosphorylation, mediated by JAK2.Indeed, EGF induced STAT1 (tyrosine701) phosphorylationreaching its maximum peak at 24 hours, while total STAT1levels remained stable (Fig. 5E). Furthermore, siRNA-targetedSTAT1 knockdown efficiently suppressed total STAT1 levels,(Supplementary Fig. S5) as well as significantly abrogating EGF-induced PD-L1 upregulation (Fig. 5F).

JAK2 inhibition prevents tumor PD-L1 expression andenhances cetuximab-mediated NK cell cytotoxicity

Because JAK2 represents a key player in PD-L1 upregulationin both EGFR (intrinsic) and IFNg (extrinsic) pathways in vitro,we tested whether JAK2 inhibition enhanced NK-mediatedkilling via antibody-dependent cell cytotoxicity (ADCC; ref. 21)against PD-L1þ HNC cells. When NK cells were coculturedwith HNC targets and cetuximab, activated NK cells upregu-lated tumor PD-L1 expression in an IFNg-dependent fashion(Fig. 6A, open bars). As a control, the EGFR specific mAbpanitumumab (IgG2 isotype), which does not bind to CD16on NK cells, did not induce PD-L1 upregulation, most likelybecause of a lack of NK cell activation and IFNg secretion (39).Importantly, the IFNg-mediated PD-L1 upregulation on HNCcells was prevented when they were pretreated with the JAK2inhibitor (left panel, closed bars), but not with a JAK1/3 specificinhibitor (right panel, closed bars). We therefore tested thehypothesis that NK cells would more efficiently lyse JAK2inhibitor pretreated tumor cells, in the setting of reducedPD-L1 expression. Indeed, NK cells showed approximately25% higher specific lysis of HNC cells pretreated with the JAK2inhibitor (Fig. 6B). Overall, these results confirm that JAK2 is animportant regulator of PD-L1 expression in HNC tumor cells,and its inhibition reverses PD-L1–mediated tumor cell escapefrom cetuximab-mediated ADCC.

Table 2. Correlation of PD-L1 and STAT3, PI3K, and MAPK pathway components in HPV-negative and HPV-positive tumors

HPV negative HPV positiveCorrelation (XY) Pearson r P Pearson r P

PD-L1 vs. STAT3 0.2503 0.0427 (�) 0.3867 0.0754 (ns)PD-L1 vs. AKT1 �0.2048 0.099 (ns) 0.00568 0.98 (ns)PD-L1 vs. TORC1 �0.2743 0.0258 (�) �0.1973 0.3787 (ns)PD-L1 vs. 4EBP1 �0.2488 0.044 (�) �0.2206 0.3238 (ns)PD-L1 vs. MAPK1 �0.1659 0.1832 (ns) 0.07862 0.728 (ns)

NOTE: PD-L1 mRNA expression did not show a strong correlation with the STAT3, PI3K, and MAPK pathways regardless of HPV status (TCGA, 66 HPV� and 22 HPVþ

tumor specimens). � , P < 0.05.Abbreviation: ns, nonsignificant.






A B

C D

JHU02

0

JHU02

2

JHU02

9

SCC47

SCC9093

VU

PD-L

1 - M

FI (f

old

chan

ge)

PD-L

1 - M

FI(fo

ld c

hang

e / v

ehic

le)

PD-L

1 - M

FI (f

old

chan

ge)

E

Rel

ativ

e en

richm

ent t

o %

Inpu

tF

93VUSCC90JHU029JHU0220

2

4

6

8204060

PD-L

1 m

RN

A (f

old

chan

ge)

VehicleIFNγIFNγ + JAK2

* ***** ***PD

-L1

mR

NA

fold

cha

nge)

G H

SCC90PCI13JHU0290

3

6

9

12

PD-L

1 - M

FI (f

old

chan

ge)

Vehicle

MockpSTAT1pSTAT3

nsnsnsPD

-L1

- MFI

(fol

d ch

ange

)

14VehicleJAK2iIFNγIFNγ + JAK2i

JAK2i

IFNγ

IFNγ + JAK2iIFNγ + JAK1/3i

Ctrl siRNA

STAT1 siRNACtrl siRNA + IFNγ

VehicleIFNγIFNγ + Cetuximab Vehicle

Untreated

0.08

0.06

0.04

0.02

0.00IFNg IFNg + Cet

30’

IFNγ

IFNγ + CetuximabCetuximab

STAT1 siRNA + IFNγSTAT3 siRNASTAT3 siRNA + IFNγ

IFNαIFNα + JAK2iIFNγ

JAK1/3i

HPV- HPV+

12

10

8

6

4

2

0

8

7

6

5 nsns

ns

ns43

2

1

0JHU022 JHU029 SCC90 93VU

JHU029

8

6

4

2

0PCI13 SCC90 93VU

JHU029

60 6

4

2

0

40

2010

86420

SCC9093VU JHU029 SCC9093VU






DiscussionGiven the known importance of HPV infection in the etiology

ofHNC, several studies have sought to correlate PD-L1 expressionwith HPV status. Recent reports have shown higher PD-L1 expres-sion in HPVþ tumors (15, 16, 40, 41). Lyford-Pike and colleagues(15) reported only 29%PD-L1 positivity among a small cohort (n¼ 9) of HPV�HNCpatients, while Malm and colleagues reported80% (41), making the association of PD-L1 expression with HPVstatus controversial. In our large series of 134 patients, themajority (approximately 60%) was PD-L1þ. Most importantly,we found HPVþ tumors to be more frequently PD-L1 positive(70%, n ¼ 64) and have a significant higher percent area andintensity of PD-L1. Importantly, a previous study showed that PD-L1 colocalized with CD3 in 56% of tumors while 44% showeda diffuse pattern with no colocalization noted (41), raisingthe question of how PD-L1 is regulated in those tumors. Thus,PD-L1 expression could be "extrinsically" induced by IFNgsecreting TILs (where colocalization was found), particularly inHPVþ tumors and "intrinsically" induced via endogenous EGFRsignaling (where no colocalization was found), particularly inHPV� tumors.

Because HPVþ tumors showed higher PD-L1 protein expres-sion in vivo and to extend findings reported previously, we tookadvantage of the RNAseq expression data available in TCGA (n¼ 500). HPVþ tumors have significantly higher expression of aTh1-type profile including CD8A, PD-1, IFNG, and JAK2 (Fig.2A and B). Our findings are concordant with those of a previousreport where HPVþ HNC tumors showed more PD-1þ CD8þ T-cell infiltration, which correlated with favorable clinical out-come (16). These data suggest that PD-1 expressing cells arebiologically relevant and may play a crucial role in HPVþ

disease and PD-L1 induction. Importantly, we report thatPD-L1 expression highly correlated with that of JAK2 at themRNA (TCGA) and protein level in vivo (IHC). Additionally, wefound that pJAK2 staining was significantly higher in HPVþ

than HPV� specimens (data not shown). Interestingly, PD-L1was also strongly correlated with a Th1-type profile regardlessof HPV status, suggesting that the PD-L1/PD-1 axis representsan important mechanism of immune evasion in both HPV-negative and -positive tumors, such that HPV� tumors may relymore on a tumor-intrinsic (EGFR-driven) PD-L1 expression,while HPVþ tumors rely more on a tumor extrinsic IFNg-mediated Th1-like response.

STAT1 was upregulated in HNC tumors when compared withpaired autologous normal mucosa and that was significantlycorrelated with PD-L1 expression regardless of HPV status. Nota-bly, components of other signaling pathways such as PI3K andMAPK, which have been previously associatedwith PD-L1 expres-sion in other types of cancer (17, 28), did not show significantcorrelation with PD-L1 expression in HPVþ tumors or induce PD-L1. Indeed, the unique biology, mutational landscape, and pre-dominant signaling pathways in HNC may explain the discrep-ancy in PD-L1 expression. It has been recently reported that PTENloss-of-function mutations are frequent in glioblastoma (31.9%of specimens, TCGA data; ref. 42). Furthermore, PD-L1 wasupregulated after PTEN loss/PI3K activation in glioblastomalines, suggesting that gliomas may rely more on this signalingpathway (17). Likewise, in the setting of non–small cell lungcancer (NSCLC), PD-L1 protein was upregulated after EGFR/RAS/MAPK pathway–activating mutations. Indeed, HRAS and EGFRmutations inNSCLC are farmore frequent than inHNC (2%–5%;refs. 43, 44). Therefore, mutant EGFR may induce a strongerMAPK pathway activation than wild-type EGFR. On the otherhand, PTEN or PIK3CA mutations are rather infrequent in HNC,7% and 8%, respectively (44). Hence, in the setting of HNC, theintrinsic oncogenic signaling mostly depends on overexpressedwild-type EGFR stimulation and presents as a unique feature ofthis type of cancer, in which the JAK/STAT3 oncogenic pathway isbest characterized (45). Importantly, in our series, STAT3 showedno significant correlation with PD-L1 expression in HPVþ tumorsand only a weak correlation in HPV� tumors, most likely becauseof higher EGFR expression in these tumors versus HPVþ ones.

Concordant with other types of cancer, IFNg induced PD-L1upregulation in all of the HNC cell lines tested in our study;however, its upregulation was not PI3K dependent as reported forglioma, lymphoma, or lung cancer (28, 46, 47). We believe this isthe first report in which specific JAK2 inhibition completelyabrogates IFNg-mediated PD-L1 upregulation at the mRNA andprotein level. Interestingly, IFNa, which does not signal via JAK2,did not upregulate PD-L1 expression, confirming the specific roleof JAK2 upregulating PD-L1. Likewise, we should emphasize thatIFNa not only induces STAT1 phosphorylation but also STAT2,and complexes with IRF9 in the transcription factor assemblycascade, forming the ISGF3 transcription complex, where IRF9 isthe main DNA binding domain (48). In contrast, IFNg mainlyinduces the formation of pSTAT1 dimers that directly bind to thepromoter region of the target gene, which explains why IFNamay

Figure 4.IFNg-mediated PD-L1 upregulation is JAK2/STAT1 dependent. A, JAK2 inhibitor BMS-911345 (JAK2i) abrogates IFNg-mediated PD-L1 protein upregulationregardless of HPV status. HNSCC cells were treated with vehicle control, JAK2i (10 mmol/L), IFNg (10 IU/mL), or the combination for 48 hours, harvested, and PD-L1expression was determined by flow cytometry. B, JAK2i abrogates IFNg-mediated PD-L1 mRNA upregulation. Cell lines were treated with vehicle control, IFNg(10 IU/mL) or JAK2i (10 mmol/L), or the combination for 24 hours; PD-L1 mRNA was determined by qPCR and expressed as fold change over vehicle. C, JAK1/3inhibitiondid not prevent IFNg-mediatedPD-L1 upregulation. Cell lineswere treatedwith JAK1/3i (10mmol/L), JAK2i (10mmol/L), IFNg (10 IU/mL), or the combinationfor 48 hours and PD-L1 expression was determined by flow cytometry (ANOVA; ns, not significant). D, IFNa did not upregulate PD-L1 expression. Cells wereincubated with IFNa (1000 IU/mL), JAK2i (10 mmol/L), and the combination for 48 hours. IFNg (10 IU/mL) was used as a positive control. PD-L1 expression wasdetermined by flow cytometry (ANOVA; ns, not significant). E, IFNg-mediated PD-L1 upregulation is abrogated when STAT1, but not STAT3, is silenced. Cells wereincubatedwith STAT1 siRNA, STAT3 siRNA, or control siRNA (10nmol/L) for 48hours, thenwere left untreatedor treatedwith IFNg (10 IU/mL) for additional 48hours,harvested, and PD-L1 expression was determined by flow cytometry. F, pSTAT1 but not pSTAT3 binds to the PD-L1 promoter region after IFNg treatment asdetermined by the ChIP assay. Cells were either left untreated or treatedwith IFNg (10 IU/mL) or IFNg þ cetuximab for 30minutes, ChIP assay showed enrichment ofpSTAT1 in the PD-L1 promoter (black bars). PD-L1 enrichment was calculated as a percentage of input DNA (refer to Materials and Methods; ANOVA; � , P < 0.05;�� , P < 0.01). G, cetuximab-mediated EGFR blockade downregulated IFNg-mediated PD-L1 mRNA upregulation. Cell lines were either treated with vehicle, IFNg (10IU/mL), cetuximab (10mg/mL), or IFNg þ cetuximab for 24 hours, harvested, andmRNAwas quantified by qPCR and expressed as fold change over vehicle (ANOVA;� , P < 0.05; ��, P < 0.0001). H, cetuximab-mediated EGFR blockade downregulated the IFNg-mediated PD-L1 protein upregulation. Cell lines were treatedas in G for 48 hours, harvested, and PD-L1 protein expression was determined by flow cytometry (ANOVA; �� , P < 0.0001).






C

BA

D

E

48 h24 h4 h2 h1 hUntreated0.0

0.4

0.8

1.2

Tota

l STA

T1M

FI (f

old

chan

ge)

JHU022SCC9093VU

Ctrl siRNACtrl siRNA + EGFSTAT1 siRNASTAT1 siRNA + EGF

pSTA

T1(Y

701)

MFI

(fol

d ch

ange

)

PD-L

1 - M

FI (f

old

chan

ge)F



JHU022

JHU020

PD-L1 - MFI (fold change)0 3 6 9 12

JHU029

PCI13

SCC2

SCC47

SCC90

93VU

0

2

4

6

81030

7050

Untreated 1 h0.5

1.0

1.5

2.0

2 h 4 h 24 h 48 h

93VUJHU029JHU022

0.0

0.5

1.0

1.5 JAK2iJAK1/3iVehicle

UntreatedEGFIFNγ

VehicleEGFIFNγ

****

PD-L

1 m

RN

A (f

old

chan

ge)


1

2

3

4

5

0

1

2

3

4

5

PD-L

1 - M

FI (f

old

chan

ge)

VehicleEGFEGF + JAK1/3iEGF + JAK2i

***

**

***

*** ns

PD-L

1 - M

FI (f

old

chan

ge)

Figure 5.EGFR-mediated PD-L1 upregulation is JAK2/STAT1 dependent. A, EGF upregulates PD-L1 protein expression. Cells were either left untreated or treated with EGF(10 ng/mL) for 48 hours, IFNg (10 IU/mL) was used as a positive control. Cells were harvested and PD-L1 surface expression was determined by flow cytometry.B, EGF treatment upregulates PD-L1 mRNA expression. Cells were treated as in A for 24 hours, harvested, and PD-L1 mRNA expression was determined byqPCR and expressed as fold change over vehicle control (ANOVA; � , P < 0.05; �� , P < 0.01; ��, P < 0.001). C, JAK2 but not JAK1/3 inhibition downregulates baselineexpression of PD-L1. Cellswere treatedwith JAK1/3 inhibitor (JAK1/3i, 10mmol/L) or JAK2i (10 mmol/L) for 48 hours andPD-L1 expression levelwasdeterminedbyflowcytometry (ANOVA, � ,P<0.05; �� ,P<0.001).D, JAK2butnot JAK1/3 inhibitionpreventsEGF-mediatedPD-L1upregulation. Cellswere treatedwith JAK1/3i (10mmol/L)or JAK2i (10 mmol/L) for 48 hours and PD-L1 expression level was determined by flow cytometry (ANOVA; ns, nonsignificant; �� , P < 0.001; �� , P < 0.0001).E, EGF induces pSTAT1y701 upregulation. Cells were treated with EGF (10 ng/mL) for 1, 2, 4, 24, and 48 hours, harvested, fixed and permeabilized, and pSTAT1y701or total STAT1 expression was determined by ICF. F, STAT1 silencing prevents EGF-induced PD-L1 upregulation. Cells were treated with control siRNA orSTAT1 siRNA (10 nmol/L) and EGF (10 ng/mL) for 48 hours, harvested, and PD-L1 expression was determined by flow cytometry (ANOVA; �� , P < 0.001).






not upregulate PD-L1, although it still upregulates pSTAT1. Inaddition, and corroborating our TCGA findings, in vitro knock-down experiments show that STAT1 but not STAT3 mediatesIFNg-induced PD-L1 upregulation, further supported by ChIPassay, providing additional evidence that pSTAT1 but not pSTAT3binds to the PD-L1 promoter region as early as 30 minutes afterIFNg treatment. Our findings complement that of a previousreport performed in lung cancer that shows IRF-1 binding to thePD-L1 promoter after IFNg treatment (49). Interestingly, we arethe first to report that cetuximab-mediated EGFR blockade sig-nificantly downregulates IFNg-induced PD-L1 expression, sug-gesting cross-talk between the IFNg and EGFR pathways in reg-ulating PD-L1 expression mediated through STAT1 modulation.

EGFR and PD-L1 showed a higher correlation in HPVþ tumorsthat corresponds to the higher correlation seen with CD8A, JAK2,and STAT1 as well. These otherwise counterintuitive results maybe explained by the fact that PD-L1 expression might be moredependent on the strength of EGFR/JAK2 pathway activation thanEGFR higher expression in HPV� tumors, which may induce anincreased STAT1 activation and induction of PD-L1 expression.Alternatively, other immune cells infiltrating the tumor microen-vironmentmay also express PD-L1, such as dendritic cells, macro-phages,monocytes, B cells, as well as tumor-associated fibroblasts

and stromal cells (26, 50). PD-L1 expression on these cells mayconfound the strength of correlation with that of EGFR, given thatthe expression of the latter is mostly on tumor cells, given that theTCGA RNAseq values represent unfractionated tumor.

Thus, JAK2/STAT1 signaling is a major common regulator forPD-L1 transcription driven by IFNg and EGFR pathways. BecauseEGFRmutations are very rare inHNC(2%; ref. 43), EGFRpathwayoveractivation, rather than activating mutations, is more impor-tant for PD-L1 upregulation inHNC.We are the first to report thatthat wild-type EGFR pathway induces PD-L1 upregulation at themRNA and protein level, and that specific JAK2 inhibition sig-nificantly downregulated baseline and EGF-induced PD-L1 upre-gulation, though not completely, suggesting that other alternativepathways also contribute to PD-L1 expression in HNC. Notewor-thy, JAK2 inhibition more effectively downregulated basal andEGF-mediated PD-L1 expression in EGFRhigh-expressing cell lines(JHU029 and JHU022). In addition, we are the first to reportthat EGFR stimulation induces phosphorylation of STAT1, whichin turn mediates PD-L1 upregulation, because its silencingcompletely abrogated this effect, EGFR and JAK2 inhibition maysynergize downregulating the "intrinsic" PD-L1 expression. Mostimportant, however, is the speculation of potential added benefitof JAK2 inhibition with the simultaneous blockade of the

Figure 6.JAK2 inhibition prevents NKmediated PD-L1 upregulation on tumor cells and enhances cetuximab-mediated NK cell cytotoxicity. A, PD-L1 expression is upregulatedon tumor cellswhen coculturedwith cetuximab-activatedNK cells in IFNg-dependent fashion (left, open bars). JAK2i pretreatment of tumor targets prevented PD-L1upregulation (left, closed bars). In contrast, JAK1/3 inhibition did not prevent PD-L1 upregulation under the same conditions (right, closed bars). Tumor target cellswere incubated in media alone or JAK2i (10 mmol/L) supplemented media for 48 hours, then cocultured with NK cells for 24 hours either alone or in the presence ofcetuximab (10 mg/mL), panitumumab (10 mg/mL), or cetuximabþanti-IFNg–blocking antibody (50 mg/mL), harvested, and PD-L1 expression on tumor cells wasdetermined by flow cytometry. Data representative of two independent experimentswith similar results. B, higher NK-cetuximabmediated lysis of JAK2i pretreatedtargets (closed black bars, 5:1 and 20:1 effector:target ratio). Tumor cells were pretreated with JAK2i (10 mmol/L) for 48 hours, and then labeled with 51Cr andcocultured with purified NK cells plus media, IgG1 control (10 mg/mL), or cetuximab (10 mg/mL) for 4 hours (ANOVA; � , P < 0.05; �� , P < 0.0001).






"extrinsic" IFNg-mediatedPD-L1upregulation,which seems tobemore important in HPVþ tumors. An added benefit of usingcombined JAK2 inhibition and cetuximab-mediated EGFR block-ade would be that the JAK2-mediated downregulation of PD-L1expression on tumor cells would ultimately enhance the effectorproperties of PD-1þ NK cells activated in the tumor microenvi-ronment. Indeed, we show that JAK2 inhibition in tumor targetsenhances cetuximab-mediated ADCC to a significant extent,therefore reversing PD-L1-mediated immunoescape of tumorcells toNK cell killing.Moreover, JAK2 inhibitionmight synergizewith mAbs targeting PD-1 and/or CTLA-4 by enhancing ADCC ofPD-1þCTLA4þ lymphoid or myeloid suppressor immune infil-trates in the tumor microenvironment as well as reducing PD-L1expression.

Disclosure of Potential Conflicts of InterestG.J. Freeman has ownership interest (including patents) in Merck, Bristol-

Myers Squibb, Roche, EMD Serono, Amplimmune, Boehringer Ingelheim, andNovartis and is a consultant/advisory board member for Novartis, Roche,Bristol-Myers Squibb, Eli Lilly, and Surface Oncology. R.L Ferris reports receiv-ing commercial research grant from Bristol-Myers Squibb and AZ/Meddimuneand is a consultant/advisory board member for Merck, Celgene, Bristol-MyersSquibb, andAZ/Meddimune.Nopotential conflicts of interestwere disclosedbythe other authors.

Authors' ContributionsConception and design: F. Concha-Benavente, R.L. FerrisDevelopment of methodology: F. Concha-Benavente, R.M. Srivastava,G.J. FreemanAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): F. Concha-Benavente, S. Trivedi, Y. LeiAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): F. Concha-Benavente, R.M. Srivastava, Y. Lei,U. Chandran, R.R. Seethala, R.L. FerrisWriting, review, and/or revision of the manuscript: F. Concha-Benavente,R.M. Srivastava, S. Trivedi, R.L. FerrisAdministrative, technical, or material support (i.e., reporting or organizingdata, constructing databases): F. Concha-Benavente, G.J. Freeman, R.L. FerrisStudy supervision: R.L. Ferris

Grant SupportThis study was supported by National Institute of Health grants 1R01

CA 206517-01, R01 DE019727, P50 CA097190, CA110249 and Universityof Pittsburgh Cancer Center Support Grant P30CA047904, P50CA101942(G.J. Freeman), and DE024173 (Yu Lei).

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 July 27, 2015; revised October 16, 2015; accepted November 5,2015; published OnlineFirst December 16, 2015.

References1. Dunn GP, Old LJ, Schreiber RD. The three Es of cancer immunoediting.

Annu Rev Immunol 2004;22:329–60.2. Leibowitz MS, Andrade Filho PA, Ferrone S, Ferris RL. Deficiency of

activated STAT1 in head and neck cancer cells mediates TAP1-dependentescape from cytotoxic T lymphocytes. Cancer Immunol Immunother2011;60:525–35.

3. Jie HB, Gildener-Leapman N, Li J, Srivastava RM, Gibson SP, Whiteside TL,et al. Intratumoral regulatory T cells upregulate immunosuppressivemole-cules in head and neck cancer patients. Br J Cancer 2013;109:2629–35.

4. Schreiber RD, Old LJ, Smyth MJ. Cancer immunoediting: integratingimmunity's roles in cancer suppression and promotion. Science 2011;331:1565–70.

5. Pardoll DM. The blockade of immune checkpoints in cancer immuno-therapy. Nat Rev Cancer 2012;12:252–64.

6. Topalian SL,Hodi FS, Brahmer JR,Gettinger SN, SmithDC,McDermottDF,et al. Safety, activity, and immune correlates of anti-PD-1 antibody incancer. N Engl J Med 2012;366:2443–54.

7. Brahmer JR, Tykodi SS, ChowLQ,HwuWJ, Topalian SL,HwuP, et al. Safetyand activity of anti-PD-L1 antibody in patients with advanced cancer.N Engl J Med 2012;366:2455–65.

8. Topalian SL, Drake CG, Pardoll DM. Targeting the PD-1/B7-H1(PD-L1)pathway to activate anti-tumor immunity. Curr Opin Immunol 2012;24:207–12.

9. Taube JM, Klein A, Brahmer JR, Xu H, Pan X, Kim JH, et al. Association ofPD-1, PD-1 ligands, and other features of the tumor immune microenvi-ronment with response to anti-PD-1 therapy. Clin Cancer Res 2014;20:5064–74.

10. Seiwert TY, Burtness B,Weiss J, Gluck I, Eder JP, Pai SI, et al. A phase Ib studyofMK-3475 in patients with human papillomavirus (HPV)-associated andnon-HPV-associated head andneck (H/N) cancer. ASCOMeeting Abstracts2014;32:6011.

11. Ramqvist T, Dalianis T. Oropharyngeal cancer epidemic and humanpapillomavirus. Emerg Infect Dis 2010;16:1671–7.

12. Chung CH, Gillison ML. Human papillomavirus in head and neck cancer:its role in pathogenesis and clinical implications. Clin Cancer Res2009;15:6758–62.

13. Fischer CA, Zlobec I, Green E, Probst S, Storck C, Lugli A, et al. Is theimproved prognosis of p16 positive oropharyngeal squamous cellcarcinoma dependent of the treatment modality? Int J Cancer2010;126:1256–62.

14. Spanos WC, Nowicki P, Lee DW, Hoover A, Hostager B, Gupta A, et al.Immune response during therapy with cisplatin or radiation for humanpapillomavirus-related head andneck cancer. ArchOtolaryngolHeadNeckSurg 2009;135:1137–46.

15. Lyford-Pike S, Peng S, Young GD, Taube JM, Westra WH, Akpeng B, et al.Evidence for a role of the PD-1:PD-L1 pathway in immune resistance ofHPV-associated head and neck squamous cell carcinoma. Cancer Res2013;73:1733–41.

16. Badoual C, Hans S, Merillon N, Van Ryswick C, Ravel P, Benhamouda N,et al. PD-1-expressing tumor-infiltrating T cells are a favorable prognosticbiomarker in HPV-associated head and neck cancer. Cancer Res 2013;73:128–38.

17. Parsa AT, Waldron JS, Panner A, Crane CA, Parney IF, Barry JJ, et al. Loss oftumor suppressor PTEN function increases B7-H1 expression and immu-noresistance in glioma. Nat Med 2007;13:84–8.

18. Azuma K, Ota K, Kawahara A, Hattori S, Iwama E, Harada T, et al.Association of PD-L1 overexpression with activating EGFR mutations insurgically resected non–small-cell lung cancer. Ann Oncol 2014;25:1935–40.

19. Akbay EA, Koyama S, Carretero J, Altabef A, Tchaicha JH, Christensen CL,et al. Activation of the PD-1 pathway contributes to immune escape inEGFR-driven lung tumors. Cancer Discov 2013;3:1355–63.

20. Maiti GP, Mondal P, Mukherjee N, Ghosh A, Ghosh S, Dey S, et al.Overexpression of EGFR in head and neck squamous cell carcinoma isassociated with inactivation of SH3GL2 and CDC25A genes. PLoS One2013;8:e63440.

21. Lee SC, Srivastava RM, Lopez-Albaitero A, Ferrone S, Ferris RL. Naturalkiller (NK): dendritic cell (DC) cross talk induced by therapeutic mono-clonal antibody triggers tumor antigen-specific T cell immunity. ImmunolRes 2011;50:248–54.

22. Srivastava RM, Lee SC, Andrade Filho PA, Lord CA, Jie HB, Davidson HC,et al. Cetuximab-activated natural killer and dendritic cells collaborate totrigger tumor antigen-specific T-cell immunity in head and neck cancerpatients. Clin Cancer Res 2013;19:1858–72.

23. Li J, Jie HB, Lei Y, Gildener-Leapman N, Trivedi S, Green T, et al.PD-1/SHP-2 inhibits Tc1/Th1 phenotypic responses and the activa-tion of T cells in the tumor microenvironment. Cancer Res 2015;75:508–18.

24. ZhaoM, SanoD, Pickering CR, Jasser SA,Henderson YC, ClaymanGL, et al.Assembly and initial characterization of a panel of 85 genomically






validated cell lines fromdiverse head and neck tumor sites. Clin Cancer Res2011;17:7248–64.

25. Heo DS, Snyderman C, Gollin SM, Pan S, Walker E, Deka R, et al.Biology, cytogenetics, and sensitivity to immunological effector cells ofnew head and neck squamous cell carcinoma lines. Cancer Res 1989;49:5167–75.

26. Curiel TJ, Wei S, DongH, Alvarez X, Cheng P, Mottram P, et al. Blockade ofB7-H1 improves myeloid dendritic cell-mediated antitumor immunity.Nat Med 2003;9:562–7.

27. Chikamatsu K, Sakakura K, Toyoda M, Takahashi K, Yamamoto T,Masuyama K. Immunosuppressive activity of CD14þ HLA-DR- cells insquamous cell carcinoma of the head and neck. Cancer Sci 2012;103:976–83.

28. Han SJ, Ahn BJ, Waldron JS, Yang I, Fang S, Crane CA, et al. Gammainterferon-mediated superinduction of B7-H1 in PTEN-deficient glioblas-toma: a paradoxical mechanism of immune evasion. Neuroreport2009;20:1597–602.

29. Schmitt NC, Trivedi S, Ferris RL. STAT1 activation is enhanced by cisplatinand variably affected by EGFR inhibition inHNSCC cells. Mol Cancer Ther2015;14:2103–11.

30. Krutzik PO, Nolan GP. Intracellular phospho-protein staining techniquesfor flow cytometry: monitoring single cell signaling events. Cytometry A2003;55:61–70.

31. Leibowitz MS, Srivastava RM, Andrade Filho PA, Egloff AM, Wang L,Seethala RR, et al. SHP2 is overexpressed and inhibits pSTAT1-mediatedAPM component expression, T-cell attracting chemokine secretion, andCTL recognition in head and neck cancer cells. Clin Cancer Res 2013;19:798–808.

32. Haring M, Offermann S, Danker T, Horst I, Peterhansel C, Stam M.Chromatin immunoprecipitation: optimization, quantitative analysis anddata normalization. Plant Methods 2007;3:11.

33. Chen BJ, Chapuy B, Ouyang J, Sun HH, Roemer MG, Xu ML, et al. PD-L1expression is characteristic of a subset of aggressive B-cell lymphomas andvirus-associated malignancies. Clin Cancer Res 2013;19:3462–73.

34. Jie HB, Schuler PJ, Lee SC, Srivastava RM, Argiris A, Ferrone S, et al. CTLA-4(þ) regulatory T cells increased in cetuximab-treated head and neck cancerpatients suppress NK cell cytotoxicity and correlate with poor prognosis.Cancer Res 2015;75:2200–10.

35. CancerGenomeAtlasN.Comprehensive genomic characterizationof headand neck squamous cell carcinomas. Nature 2015;517:576–82.

36. Li B, Dewey CN. RSEM: accurate transcript quantification from RNA-Seqdata with or without a reference genome. BMC Bioinformatics 2011;12:323.

37. Kel AE, Gossling E, Reuter I, Cheremushkin E, Kel-Margoulis OV, Win-gender E. MATCH: a tool for searching transcription factor binding sites inDNA sequences. Nucleic Acids Res 2003;31:3576–9.

38. Purandare AV, McDevitt TM, Wan H, You D, Penhallow B, Han X, et al.Characterization of BMS-911543, a functionally selective small-moleculeinhibitor of JAK2. Leukemia 2012;26:280–8.

39. Lopez-Albaitero A, Ferris RL. Immune activation by epidermal growthfactor receptor specific monoclonal antibody therapy for head and neckcancer. Arch Otolaryngol Head Neck Surg 2007;133:1277–81.

40. Ukpo OC, Thorstad WL, Lewis JS Jr. B7-H1 expression model for immuneevasion in human papillomavirus-related oropharyngeal squamous cellcarcinoma. Head Neck Pathol 2013;7:113–21.

41. Malm IJ, Bruno TC, Fu J, Zeng Q, Taube JM, Westra W, et al. Expressionprofile and in vitro blockade of programmed death-1 in human papillo-mavirus-negative head and neck squamous cell carcinoma. Head Neck2015;37:1088–95.

42. Ramkissoon SH, Bi WL, Schumacher SE, Ramkissoon LA, Haidar S, KnoffD, et al. Clinical implementation of integrated whole-genome copy num-ber and mutation profiling for glioblastoma. Neuro Oncol 2015;17:1344–55.

43. McBride SM, Rothenberg SM, Faquin WC, Chan AW, Clark JR, Ellisen LW,et al.Mutation frequency in 15 common cancer genes in high-risk head andneck squamous cell carcinoma. Head Neck 2014;36:1181–8.

44. Stransky N, Egloff AM, Tward AD, Kostic AD, Cibulskis K, Sivachenko A,et al. Themutational landscape of head andneck squamous cell carcinoma.Science 2011;333:1157–60.

45. Sen M, Pollock NI, Black J, DeGrave KA, Wheeler S, Freilino ML, et al. JAKkinase inhibition abrogates STAT3 activation andhead andneck squamouscell carcinoma tumor growth. Neoplasia 2015;17:256–64.

46. YamamotoR,NishikoriM, TashimaM, Sakai T, Ichinohe T, Takaori-KondoA, et al. B7-H1 expression is regulated by MEK/ERK signaling pathway inanaplastic large cell lymphoma and Hodgkin lymphoma. Cancer Sci2009;100:2093–100.

47. Ota K, Azuma K, Kawahara A, Hattori S, Iwama E, Tanizaki J, et al.Induction of PD-L1 expression by the EML4-ALK oncoprotein and down-stream signaling pathways in non-small cell lung cancer. Clin Cancer Res2015;21:4014–21.

48. Sadzak I, Schiff M, Gattermeier I, Glinitzer R, Sauer I, Saalmuller A, et al.Recruitment of Stat1 to chromatin is required for interferon-induced serinephosphorylation of Stat1 transactivation domain. Proc Natl Acad Sci U S A2008;105:8944–9.

49. Lee SJ, Jang BC, Lee SW, Yang YI, Suh SI, Park YM, et al. Interferonregulatory factor-1 is prerequisite to the constitutive expression andIFN-gamma-induced upregulation of B7-H1 (CD274). FEBS Lett2006;580:755–62.

50. Freeman GJ, Long AJ, Iwai Y, Bourque K, Chernova T, Nishimura H, et al.Engagement of the PD-1 immunoinhibitory receptor by a novel B7 familymember leads to negative regulation of lymphocyte activation. J Exp Med2000;192:1027–34.






2016;76:1031-1043. Published OnlineFirst December 16, 2015.Cancer Res Fernando Concha-Benavente, Raghvendra M. Srivastava, Sumita Trivedi, et al. Head and Neck Cancer

That Induce PD-L1 Expression inγDownstream of EGFR and IFNIdentification of the Cell-Intrinsic and -Extrinsic Pathways

Updated version

10.1158/0008-5472.CAN-15-2001doi:

Access the most recent version of this article at:

Material

Supplementary

http://cancerres.aacrjournals.org/content/suppl/2015/12/10/0008-5472.CAN-15-2001.DC1.html

Access the most recent supplemental material at:

Cited articles

http://cancerres.aacrjournals.org/content/76/5/1031.full.html#ref-list-1

This article cites 50 articles, 21 of which you can access for free 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

[email protected]

To request permission to re-use all or part of this article, contact the AACR Publications Department at



http://cancerres.aacrjournals.org/lookup/doi/10.1158/0008-5472.CAN-15-2001

http://cancerres.aacrjournals.org/content/suppl/2015/12/10/0008-5472.CAN-15-2001.DC1.html

http://cancerres.aacrjournals.org/content/76/5/1031.full.html#ref-list-1

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

mailto:[email protected]

mailto:[email protected]


identiﬁcation of the cell-intrinsic and -extrinsic ... · fernando concha-benavente1, raghvendra...

Documents