University of Groningen Preclinical evaluation and molecular imaging of HER family dynamics to guide cancer therapy Kol, Klaas Jan-Derk

Preclinical evaluation and molecular imaging of HER family dynamics to guide cancer therapy

Kol, Klaas Jan-Derk

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.

Document Version

Publisher's PDF, also known as Version of record

Publication date: 2019

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):

Kol, K. J-D. (2019). Preclinical evaluation and molecular imaging of HER family dynamics to guide cancer therapy. Rijksuniversiteit Groningen.

Copyright

Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

Take-down policy

If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.

CHAPTER

MAPK pathway activity plays a key role in

programmed death ligand-1 expression of

EGFR wild-type lung adenocarcinoma cells

Arjan Kol1*

Thijs S. Stutvoet2*

Elisabeth G.E. de Vries2

Marco de Bruyn1

Rudolf S.N. Fehrmann2

Anton G.T. Terwisscha van Scheltinga3

Steven de Jong2

1_{Departments of Obstetrics and Gynecology,} 2_{Medical Oncology, Cancer Research Center Groningen,}

University of Groningen, University Medical Center Groningen, Groningen, The Netherlands.

3 _{Department of Clinical Pharmacy and Pharmacology,}

University of Groningen, University Medical Center Groningen, Groningen, The Netherlands.

*Both authors contributed equally.

Submitted.

ABSTRACT

Immune checkpoint inhibitors targeting programmed cell death protein 1 (PD-1) and programmed death-ligand 1 (PD-L1) have improved the survival of epidermal growth factor receptor (EGFR) wild-type non-small cell lung cancer (NSCLC), the largest group of NSCLC patients. Still, many patients do not respond to these inhibitors. PD-L1 expression, one of the factors that can influence efficacy of immune checkpoint inhibitors, is very dynamic. Here, we studied the regulation of PD-L1 expression in EGFR wild-type NSCLC by epidermal growth factor (EGF) and interferon gamma (IFNγ). Analysis of RNA sequencing data of EGFR wild-type NSCLC tumors revealed that inferred IFNγ and mitogen activated protein kinase (MAPK) signaling correlated with PD-L1 gene expression in lung adenocarcinoma. In EGFR wild-type lung adenocarcinoma cell lines, stimulation with EGF or IFNγ strongly increased PD-L1 mRNA, protein, and membrane expression, which was further enhanced by combining EGF and IFNγ. Similarly, tumor cell PD-L1 membrane expression increased after coculture of activated peripheral blood mononuclear cells (PBMCs). Inhibition of the MAPK pathway, using EGFR-inhibitors cetuximab and erlotinib or mitogen-activated protein kinase kinase 1 and 2 (MEK1/2) inhibitor selumetinib, prevented EGF and IFNγ-induced PD-L1 mRNA, protein, and membrane upregulation, but had no effect on IFNγ-induced MHC-I upregulation. Interestingly, while IFNγ increases transcriptional activity of PD-L1, MAPK signaling acted both through increased transcription and stabilization of PD-L1 mRNA. In conclusion, MAPK signaling plays a key role in EGF and IFNγ-induced PD-L1 expression in EGFR wild-type adenocarcinoma and may present a target to improve efficacy of immunotherapy.

6

INTRODUCTION

After years of limited progress in the treatment of advanced non-small cell lung cancer (NSCLC), a major leap forward has been made with the introduction of programmed cell death protein 1 (PD-1)/ programmed death-ligand 1 (PD-L1) targeting immune checkpoint inhibitors. These have greatly improved overall survival of patients with advanced NSCLC, especially in patients with wild-type epidermal growth factor receptor (EGFR) accounting for 70% of NSCLC (1,2). Patients with PD-L1 positive tumors generally respond better to PD-1 targeted immune checkpoint inhibition, however, discrepancies between observed PD-L1 expression and benefit from treatment often occur (3). This is highlighted by the fact that even in a preselected patient population with >50% PD-L1 positive tumor cells, only 45-55% of patients respond to therapy (4). The limited value of tumor PD-L1 expression as a biomarker may be caused by the highly dynamic expression of PD-L1 due to the influence of multiple factors (5). The best characterized inducer of PD-L1 expression in NSCLC is the pro-inflammatory cytokine interferon γ (IFNγ), which is secreted by T cells (6,7). PD-L1 expressed on tumor cells in return binds to PD-1 on T cells, disrupting T cell function and thereby preventing an effective tumor immune response (8). Constitutive activation of growth factor receptors, such as EGFR, are known inducers of PD-L1 expression in NSCLC cells as well. In EGFR mutant cell lines, EGFR-induced activation of the phosphatidylinositol 3-kinase-mammalian target of rapamycin- (PI3K-mTOR), janus kinase/signal transducer and activator of transcription- (JAK/STAT), and mitogen-activated protein kinase (MAPK) pathway are the main drivers of PD-L1 expression (9–12).

Interestingly, EGFR wild-type NSCLC tumors have higher levels of PD-L1 expression and tumor infiltrating lymphocytes compared to EGFR mutant NSCLC (1,2). However, there is only limited data about the regulation of PD-L1 expression in EGFR wild-type NSCLC (13–15). Better understanding of PD-L1 regulation may provide a rationale to combine immune checkpoint inhibitors with other targeted agents. In the present study, we aimed to identify pathways regulating PD-L1 expression in EGFR wild-type NSCLC by using RNA sequencing data from The Cancer Genome Atlas (TCGA) lung adenocarcinoma and squamous cell lung carcinoma data sets. We functionally validated our findings using EGFR wild-type lung adenocarcinoma cell lines and cocultures with peripheral blood mononuclear cells (PBMCs). Our results indicate that growth factor-dependent MAPK signaling plays an important role in EGF and IFNγ and induced PD-L1 expression of EGFR wild-type lung adenocarcinoma.

MATERIALS AND METHODS TCGA data retrieval and analysis

cell carcinoma (16,17) were obtained from the cBioportal (18) for Cancer Genomics on October 14th 2017. We selected the 197 adenocarcinoma and 172 squamous cell carcinoma samples that were known to be EGFR wild-type and for which RNA sequencing data was available. Data was analyzed and visualized using R (available from https:// www.r-project.org/) and the R studio interface 1.1.453 (available from https://www. rstudio.com/) and ggplot2 package for R 3.5.1 (available from http://ggplot2.tidyverse. org). TCGA RNA sequencing data was normalized in two steps. Each sample was first log10-transformed, and next Z-score normalized by subtracting the mean expression of each gene and dividing by the standard deviation. Next, MAPK, and PI3K/AKT pathway activation were inferred according to the methods of previously described gene signatures for Rat Sarcoma- (RAS), mitogen-activated protein kinase kinase- (MEK), and PI3K-activation (19–21). The original papers that describe the signatures devise a list of genes that are positively, or negatively correlated with pathway activation. Inferred pathway activities were calculated as the unweighted average expression of positively correlated genes, minus the unweighted average of negatively correlated genes. The correlation of PD-L1 RNA expression with these signature scores was calculated using Spearman correlation. For parts of the EGFR and IFNγ pathway where no previously described signatures were available, individual genes known to represent pathway activation were chosen.

Cell culture

The human NSCLC cell lines HCC827, H292, A549, H358 and H460 were obtained from the American Type Culture Collection (ATCC). H322 was obtained from Sigma-Aldrich. All cell lines are from the adenocarcinoma histological subtype, except H292 which is an adenocarcinoma-like mucoepidermoid carcinoma. Cells were quarantined until screening for microbial contamination and mycoplasma was performed and proven to be negative. Cells were tested and authenticated using short tandem repeat (STR) profiling. Cells were grown in Roswell Park Memorial Institute (RPMI) medium with 10% fetal calf serum (FCS), with 2 mM glutamine added for H322 cells. All cells were incubated in a

humidified atmosphere with 5% CO₂ and at 37°C.

Antibodies and treatments

For flow cytometry, mouse anti-PD-L1 (clone 29E.2A3, BioLegend), MHC-I (Clone W6/32, BioLegend), and secondary antibodies against mouse IgG (polyclonal goat anti-mouse PE, SouthernBiotec) were used. For Western blotting, membranes were incubated with 1:250 EGFR (#2232), 1:500 pEGFR (#3777), and 1:1000 PD-L1 (#13684), pERK1/2 (#9106), ERK1/2 (#9102), pSTAT1Ser727 (#8826), STAT1 (#9172), pSTAT3Tyr705 (#9145), STAT3 (#12640), pAKTs473 (#9271), pAKTthr308 (#9275), pS6Ser235/236 (#2211), S6 (#2217) antibodies (Cell Signaling Technology), 0.4 µg/mL CMTM6 antibody (HPA026980,

6

Atlas Antibodies), 1:1000 GAPDH antibody (128915, Abcam), 1:10000 β-actin antibody (#69100, MP Biochemicals) and secondary HRP-anti-mouse or HRP-anti-rabbit antibodies at 1:1500 (Dako). Detection was performed using Lumi-Light Western blotting substrate (Roche Diagnostics Nederland B.V.). Cells were treated with EGF (R&D systems), HGF (R&D systems), IFNγ (R&D Systems), erlotinib (LC Laboratories), cetuximab (Merck KGaA), selumetinib (AZD6244, Axon Medchem), XL147 (LC Laboratories), everolimus (Selleckchem), BMS911543 (Selleckchem), and actinomycin D (Sigma-Aldrich).

siRNA-transfection

Cells were transiently transfected with small interfering RNAs (siRNA) targeting STAT3 (Eurogentec), or a negative control siRNA (12935300, Invitrogen) using oligofectamine (11252011, Invitrogen) in Opti-MEM (51985, Invitrogen) according to the manufacturer’s instructions. Twenty-four hours after transfection, cells were treated with indicated ligands and treatments. After a total of 48 hours STAT3 knockdown efficiency and proteins of interest were analyzed by Western blotting or flow cytometry. All experiments were performed in triplicate.

Flow cytometric analysis

Cells were harvested using trypsin and kept on ice during the entire protocol. After harvesting, cells were kept in phosphate buffered saline with 2% FCS until analysis. Cells were incubated with anti-PD-L1 antibodies at 10 µg/mL for 45 minutes. Bound antibody was detected by incubating cells with goat anti-mouse IgG at a 1:50 dilution for 45 minutes. Measurements were performed on a BD Accuri C6 flow cytometer (BD Biosciences). Data analysis was performed with FlowJo v10 (Tree Star) and surface receptor expression was expressed as mean fluorescence intensity (MFI). Measurements were corrected for background fluorescence and unspecific binding of the secondary antibody. Unless stated otherwise, all experiments were performed in triplicate. Western blot

Lysates from cells were made using mammalian protein extraction reagent with protease and phosphatase inhibitors diluted 1:100 (Thermo Fisher Scientific). Proteins were separated using SDS-PAGE. Target proteins were stained with the earlier mentioned antibodies. Images were captured using a digital imaging system (Bio-Rad). β-actin and GAPDH were used as loading control.

Viability assays

For the viability assays H292 (8000 cells/well), H358 (20000 cells/well), A549 (2000 cells/well), H322 (10000 cells/well) and H460 (2000 cells/well) cells were plated in 96 well plates in their respective media and after 6 hours erlotinib or selumetinib were

added in concentrations ranging from 0.01 to 10 µM. After 96 hours treatment, cells were fixated using 3.7% formaldehyde and stained using crystal violet. The unbound dye was removed by washing with water. Bound crystal violet was dissolved using 10% ethanoic acid and absorption was measured at a wavelength of 590 nm. Cell survival was calculated as percentage of untreated control. All proliferation assays were performed three times in triplicate.

RNA sample collection and qRT-PCR

Total RNA was extracted using Trizol reagent (Invitrogen) and possible DNA contamination was removed using TURBO DNase ambion (Life technologies, AM2238). Next, RNA was reverse transcribed to cDNA with M-MLV reverse transcriptase (Thermo Fisher Scientific, 28025013). Real-time PCR was performed using IQ SYBR Green Supermix (Bio-Rad, 1708886) according to manufacturer’s instructions. The following primers were used: PD-L1 forward 5’-CAATGTGACCAGCACACTGAGAA-3’, reverse 5’- GGCATAATAAGATGGCTCCCAGAA-3’; GAPDH forward 5’-CCCACTCCTCCACCTTTGAC-3’, reverse 5’-CCACCACCCTGTTGCTGTAG-3’. The relative gene expression was calculated using the double delta CT method and GAPDH as loading control (22). All qPCR experiments were performed three times in duplicate.

Coculture experiments

Human PBMCs were isolated from whole blood by Ficoll-Paque density centrifugation (Ficoll-Paque PLUS, GE Healthcare Life Sciences) from peripheral blood donated by healthy volunteers. The acquired PBMCs were activated for 72 hours using human T-activator CD3/28 beads (Thermo Fisher Scientific) and 100 IU/mL IL-2 (Proleukin, Novartis) in the presence of tumor cells. Separately, tumor cells were seeded into

96-well plates at a density of 1 x 104 _{cells/well for 48 hours. Then, the pre-activated PBMCs}

were added into the coculture system at a 5:1 ratio of PBMCs to tumor cells. During coculture, cells were treated with EGF (20 ng/mL), erlotinib (10 µM) and selumetinib (10 µM). After 24 hours of coculture cell-free supernatant was collected for IFNγ analysis by enzyme-linked immunosorbent assay (ELISA, Sino Biological). Cells were harvested for flow cytometric measurement of PD-L1 membrane expression. In separate experiments, tumor cells were cultured in cell-free supernatant from activated PBMCs. After 24 hours PD-L1 membrane expression was determined using flow cytometry.

Statistics

Cell line experiments were assessed for differences with unpaired two-tailed Student’s t-test or two-way ANOVA followed by Bonferroni post-hoc or Dunnett’s test. Experiments were performed at least three times. Results are represented as means ± SD. A P-value < 0.05 was considered statistically significant. Statistical analyses were performed using

6

GraphPad Prism software (version 6.0 GraphPad software). RESULTS

MAPK pathway activation correlates with PD-L1 gene expression in lung adenocarcinoma To study which EGFR-related signaling pathways regulate PD-L1 expression in EGFR wild-type NSCLC, we analyzed RNA sequencing data of 197 EGFR wild-type lung adenocarcinoma and 172 EGFR wild-type squamous cell lung carcinoma samples from TCGA. Activating KRAS mutations were present in 75 of the lung adenocarcinoma and 2 of the squamous cell carcinoma samples. Activation status of the MAPK pathway was inferred using a signature for either RAS- or MEK-activation, and one signature was used to infer the activation status of the PI3K/AKT pathway, as previously described (19–21). MAPK pathway activation scores were higher in KRAS mutant lung adenocarcinomas (Fig. S1A). In addition, there was a moderate correlation between RAS and MEK activation

-2 -1 0 1 2 -1 0 1 2 -2 -1 0 1 2 -1 _{SQcc_wt$Dry} -1 0 1 -2 -1 0 1 2 -1 0 1

Figure 1

IRF-1 expression (Z-score) Adenocarcinoma rs = .52 p < 1.1 x 10-14 Adenocarcinoma 2 1 0 -1 -2 PD -L 1 ex pr es sio n (Z -s co re ) -1 0 1 Adenocarcinoma

KRAS wild-type AdenocarcinomaKRAS wild-type Adenocarcinoma 2 1 0 1 2 -1 0 1 -2 1 0 1 2 -1 0 1 -2 1 0 1 2 - 0 1 -1 PD -L 1 ex pr es sio n (Z -s co re ) 2 1 0 -1 -2 -1 2 1 0 -1 -2 -1 0

1 2 1 0 -1 -2 -1 0 1 2 -2 2 1 0 -1 -2 -1 0 1 2 -2 PD -L 1 ex pr es sio n (Z -s co re )

RAS-ac�va�on score MEK-ac�va�on score RAS-ac�va�on score MEK-ac�va�on score

RAS-ac�va�on score MEK-ac�va�on score -1 0 1 PD -L 1 ex pr es sio n (Z -s co re ) STAT1 expression (Z-score) Adenocarcinoma rs = .37 p = 7.6 x 10-7 r s = .36 p = 4.8 x 10-7 rs = .48 p = 8.2 x 10-5 r s = .47 p = 1.7 x 10-4 rs = .61 p < 1.1 x 10-14 C A B LUSC LUSC rs = -.03 p = .67 rp = .48s = .05 Figure 1:

MAPK-activation and IFNγ signaling correlate with PD-L1 expression in EGFR wild-type lung adenocarcinoma tumors. RNA sequencing data from all EGFR wild-type samples was collected from the TCGA lung adenocarcinoma and squamous cell lung carcinoma datasets. RNA sequencing data was Z-score normalized

after 10_{log transformation. (A) In all samples the correlation between PD-L1 mRNA expression and}

RAS-activation score or MEK-RAS-activation score was calculated using spearman correlation. (B) In the EGFR wild-type, KRAS wild-type adenocarcinoma subset the correlation between PD-L1 mRNA expression and RAS-activation score or MEK-activation score was calculated using spearman correlation. (C) In all samples the correlation of STAT1 and IRF-1 mRNA expression with PD-L1 mRNA expression was calculated using Spearman correlation. LUSC = squamous cell lung carcinoma.

Figure 2

Con EGF IFNγ

EGF + IFN γ H322 H292 A549 H358 H460 KRAS wild-type KRAS mutant PD-L1 (MFI) 3.0 x 105 2.0 x 105 1.0 x 105 0 H292 H358 A B C Time (h) PD-L1 pERK1/2 ERK1/2 pSTAT1 STAT1 pSTAT3 STAT3 pS6 S6 Ac�n

EGF IFNγ EGF + IFNγ

0 1 24 4872 1 2448 72 1 24 48 72

EGF IFNγ EGF + IFNγ

0 1 24 4872 1 24 48 72 1 24 48 72 Time (h) PD-L1 pERK1/2 ERK1/2 pSTAT1 STAT1 pSTAT3 STAT3 pS6 S6 Ac�n

Con EGF IFNγ

EGF + IFN γ

Con EGF IFNγ

EGF + IFN γ

Con EGF IFNγ

EGF + IFN γ

Con EGF IFNγ

EGF + IFN γ Unspecific control Untreated EGF IFNγ EGF + IFNγ H322 # Cells Fluorescence intensity A549 H460 100 ₁₀2 ₁₀4 ₁₀6 ₁₀0 ₁₀2 ₁₀4 ₁₀6 100 ₁₀2 ₁₀4 ₁₀6 ₁₀0 ₁₀2 ₁₀4 ₁₀6₁₀0 ₁₀2 ₁₀4 ₁₀6 200 400 600 0 800 H292 H358 E ** ** ** * PD-L1 (MFI) 3.0 x 105 2.0 x 105 1.0 x 105 0 4.0 x 105 5.0 x 105 *** *** *** ** *** Con EGFIFNγ EGF + IFN γ Con EGFIFNγ EGF + IFN γ Con EGFIFNγ EGF + IFN γ Con EGFIFNγ EGF + IFN γ PD-L1 mRN A (f ol d chang e)150 100 50 0 D

Con EGF IFNγ

EGF + IFN γ

Con EGF IFNγ

EGF + IFN γ 24 hours 72 hours H292 H358 H292 H358 - 60 kDa - 45 kDa - 43 kDa - 43 kDa - 90 kDa - 90 kDa - 80 kDa - 82 kDa - 32 kDa - 32 kDa - 42 kDa - 60 kDa - 45 kDa - 43 kDa - 43 kDa - 90 kDa - 90 kDa - 80 kDa - 82 kDa - 32 kDa - 32 kDa - 42 kDa Figure 2:

EGF and IFNγ induce PD-L1 in NSCLC cell lines. (A, B) A panel of EGFR wild-type NSCLC cell lines was treated with 20 ng/mL EGF, 20 ng/mL IFNγ, or both. After 24 hours, PD-L1 membrane expression was measured using flow cytometry. (Student’s t test * P < 0.05, ** P < 0.01 compared to IFNγ, n = 4). (C) EGF and IFN were incubated up to 72 hours in H292 and H358. PD-L1 membrane expression was measured using flow cytometry. (Two-way ANOVA with Bonferroni multiple comparisons method ** P< 0.01, *** P < 0.001 compared to 24 hour stimulation) (D) H292 and H358 were treated with 20 ng/mL EGF or IFNγ, or both. After 24 hours PD-L1 mRNA levels were measured using RT-qPCR. Data was analyzed using the double delta CT method and GAPDH as a loading control. (E) H292 and H358 were treated with 20 ng/mL EGF, 20 ng/mL IFNγ, or both. After 1, 24, 48, and 72 hours, protein levels were measured using Western blotting. Actin was used as loading control. Con = untreated control.

6

scores in both subtypes (adenocarcinoma: rs = 0.50, p < 2.2 x 10-16_{; squamous cell}

carcinoma rs = 0.55, p < 2.2 x 10-16_{, Fig. S1B). Interestingly, RAS and MEK activation scores}

correlated with PD-L1 mRNA levels in EGFR wild-type adenocarcinomas but not in EGFR wild-type squamous cell carcinomas (Fig. 1A, supplementary table 1). Subset analysis showed that these correlations were strongest in EGFR wild-type, KRAS wild-type lung adenocarcinomas (Fig. 1B). In contrast, PI3K pathway activation and STAT3 mRNA did not correlate with PD-L1 mRNA levels. In both NSCLC subtypes, gene expression of STAT1 and interferon regulatory factor 1 (IRF-1), important IFNγ-responsive transcriptional enhancers of PD-L1, correlated with PD-L1 mRNA expression (Fig. 1C, S1C) (6). This analysis suggests that activation of the MAPK and IFNγ pathway is related to increased PD-L1 expression in EGFR wild-type lung adenocarcinomas.

EGF increases IFNγ-induced PD-L1 expression in EGFR wild-type NSCLC cells

A panel of EGFR wild-type lung adenocarcinoma cell lines, including a KRAS wild-type (H322), 3 KRAS mutant (A549, H358, and H460), and a KRAS wild-type adenocarcinoma-like mucoepidermoid carcinoma cell line (H292) (23), was selected to further investigate the relation between MAPK pathway activation, IFNγ pathway activation and PD-L1 expression. PD-L1 membrane expression was observed in all cell lines, irrespective of KRAS mutation status (Fig. S2A). The highest levels were found in H292, H358 and H460 cells. Levels were comparable to PD-L1 membrane levels of HCC827 EGFR mutant NSCLC cells (Fig. S2A). We wondered whether EGF, a known activator of the MAPK pathway via EGFR, and IFNγ would increase PD-L1 expression in our panel. Treatment with EGF or IFNγ for 24 hours using a physiologically relevant concentration (20 ng/mL) (24,25) increased PD-L1 membrane expression in both KRAS wild-type and KRAS mutant cells (Fig. 2A,B). Moreover, exposure of cells to EGF combined with IFNγ resulted in a further increase in PD-L1 expression compared to IFNγ alone. This response was time-dependent, since prolonged incubation up to 72 hours further enhanced PD-L1 expression in H292 and H358 (Fig. 2C).

To gain insight in the underlying mechanism of the increase in PD-L1 surface expression, we determined PD-L1 mRNA and protein levels. The EGF- and IFNγ-induced increase in surface expression was reflected in a strong induction of PD-L1 mRNA and total protein levels (Fig. 2D,E). EGF stimulated the MAPK and PI3K pathway, as signified by increased levels of phosphorylated extracellular signal-related kinases 1 and 2 (pERK1/2) and phosphorylated ribosomal S6 protein (pS6), respectively (Fig. 2E, S2B). IFNγ strongly increased STAT1 and phosphorylated STAT1 (pSTAT1) levels for up to 72 hours. Taken together, these results indicate that EGF and IFNγ treatment results in activation of the MAPK, PI3K/AKT, and STAT1 pathway, and a concurrent increase in PD-L1 mRNA, protein and membrane levels.

EGFR inhibition prevents EGF- and IFNγ-induced PD-L1 upregulation

To analyze the regulation of PD-L1 membrane expression by EGFR and STAT1 signaling, H292 and H358 cells were treated with EGF and IFNγ in the presence of anti-EGFR monoclonal antibody cetuximab or EGFR small molecule inhibitor erlotinib. Interestingly, cetuximab and erlotinib prevented not only EGF-induced but also IFNγ-induced upregulation of PD-L1 mRNA levels, which was reflected in PD-L1 membrane and total protein expression levels (Fig. 3A,B, S3A). Data from multiple experiments showed that erlotinib slightly reduced basal PD-L1 membrane levels in H358 but not H292 cells (Fig. S3B). Erlotinib had a similar effect on EGF- and IFNγ-induced PD-L1 membrane expression in two other cell lines but not, as expected, in the erlotinib-resistant H460 cell line (Fig. S4A). EGFR inhibition effectively reduced EGF-dependent MAPK and PI3K/AKT signaling, and modestly decreased IFNγ-induced upregulation of (p)STAT1 expression (Fig. 3B). These results indicate that EGF- and IFNγ-induced PD-L1 mRNA, protein and membrane expression levels are dependent on EGFR-mediated signaling.

0 1.0×105 2.0×105

3.0×105 Control Cetuximab Erlo�nib

A

B

Figure 3

2.0 x 105 1.0 x 105 PD-L1 (MFI) 3.0 x 105 2.0 x 105 1.0 x 105 0

Control Cetuximab Erlo�nib

H292

CEGF IFNE+I CEGF IFNE+I CEGF IFNE+I CEGF IFNE+I CEGF IFNE+I CEGF IFNE+I

H358 PD-L1 EGFR pEGFR pERK1/2 ERK1/2 pSTAT1 STAT1 pSTAT3 STAT3 pS6 S6 Ac�n PD-L1 EGFR pEGFR pERK1/2 ERK1/2 pSTAT1 STAT1 pSTAT3 STAT3 pS6 S6 Ac�n 3.0 x 105 PD-L1 (MFI) 0 ** _*** *** *** *** *** ** *** ** *** *** Figure 3:

Erlotinib and cetuximab decrease EGF- and IFNγ-induced PD-L1 expression. H292 and H358 cells were treated with 20 μg/mL cetuximab or 10 μM erlotinib with and without co-treatment with 20 ng/mL EGF, 20 ng/mL IFNγ, or both. (A) After 24 hours, PD-L1 membrane expression was measured using flow cytometry. (Two-way ANOVA with Dunnett’s multiple comparisons test ** P < 0.01, *** P < 0.001 compared to untreated control). (B) After 24 hours cellular protein levels were measured using Western blotting. Actin was used as loading control. Data from a representative experiment are shown. C = untreated control, E + I = EGF + IFNγ.

6

4 .0×1 05 1 .0×1 05 2 .0×1 05 3 .0×1 05 4 .0×1 05 Control EGF Figure 4

IFNγ EGF + IFNγ

H292 H358

Control EGF IFNγ EGF + IFNγ

PD-L1 (MFI) 4.0 x 105 3.0 x 105 2.0 x 105 1.0 x 105 0 PD-L1 (MFI) 4.0 x 105 3.0 x 105 2.0 x 105 1.0 x 105

C XL Ev Selu C XL Ev Selu C XL Ev Selu C XL Ev Selu 0 C XL Ev Selu C XL Ev Selu C XL Ev Selu C XL Ev Selu PD-L1 pERK1/2 ERK1/2 pSTAT1 STAT1 pSTAT3 STAT3 pAKTs473 pAKTT308 AKT pS6 S6 Ac�n A B *** *** H292 H358 5.0 x 105 4.0 x 105 3.0 x 105 2.0 x 105 1.0 x 105 0 PD -L1 (MFI) Control Erlo�nib Selume�nib 5.0 x 105 4.0 x 105 3.0 x 105 2.0 x 105 1.0 x 105 0 PD -L1 (MFI) * *** *** * * ** ** *** _{** **}*** *** * C HGF EGF IFNγ + + + + + + + + -+ -+ -+ -+ + + + + -+ -+ -+ -+ + + + + -HGF EGF IFNγ + + + + + + + + -+ -+ -+ -+ + + + + -+ -+ -+ -+ + + + + -1.0 x 106 2.5 x 106 2.0 x 106 5.0 x 105 0 MHC-I (MFI) 1.5 x 106

Con EGF IFNγ EGF + IFN γ _Con EGF IFNγ EGF + IFN γ H358 H292 Erlotinib Selumetinib Control ** ns Lorem ipsum PD-L1 pERK1/2 ERK1/2 pSTAT1 STAT1 pSTAT3 STAT3 pAKTs473 pAKTT308 AKT pS6 S6 Ac�n ns * ns ns * *** D Figure 4:

Selumetinib effectively decreases tumor cell PD-L1 expression induced by growth factors and IFNγ. H292 and H358 cells were treated with 10 μM XL147, 10 μM everolimus, or 10 μM selumetinib, with and without co-treatment with 20 ng/mL EGF, 20 ng/mL IFNγ, or both. (A) After 24 hours, PD-L1 membrane expression was measured using flow cytometry. (Two-way ANOVA with Dunnett’s multiple comparisons test ns = not significant, * P < 0.05, *** P < 0.001 compared to ligand-stimulated control) (B) After 24 hours, cellular protein levels were measured using Western blotting. Actin was used as loading control. Data are from a representative experiment. Data from a representative experiment. N = 2. (C) H292 and H358 cells were treated with 10 μM erlotinib or 10 μM selumetinib, with and without co-treatment with 20 ng/mL EGF, 20 ng/mL HGF, 20 ng/ mL IFNγ, or a combination. After 24 hours, PD-L1 tumor cell membrane expression was measured using flow cytometry. (Two-way ANOVA with Dunnett’s multiple comparisons test * P < 0.05, ** P < 0.01, *** P < 0.001 compared to ligand-stimulated control). (D) Cells were treated with 10 μM selumetinib or 10 μM erlotinib with and without co-treatment with 20 ng/mL EGF, 20 ng/mL IFNγ, or both. After 24 hours, MHC-I expression was measured using flow cytometry (Student’s t test ** P < 0.01). ns = not significant, C = untreated control, XL = XL147, Ev = everolimus, Selu = selumetinib.

MAPK pathway inhibition prevents PD-L1 expression induction by EGF and IFNγ Next, we assessed the involvement of MAPK and PI3K/AKT signaling in PD-L1 upregulation. Inhibition of PI3K using XL147, or mTORC1 using everolimus had no effect on the induction of PD-L1 membrane expression by EGF and IFNγ (Fig. 4A), although both drugs effectively inhibited their respective target proteins, as indicated by reduced phosphorylated protein kinase B (pAKT) and pS6 levels (Fig. 4B). XL147 partially suppressed EGF- and IFNγ-induced upregulation of PD-L1 total protein (Fig. 4B). Selumetinib, an inhibitor of MEK1/2, suppressed induction of PD-L1 mRNA by EGF and IFNγ in H292 and H358, and diminished the induction of protein and membrane expression levels (Fig. 4A,B, S3A). Selumetinib effectively inhibited MEK1/2 activity, as reflected in the reduction in pERK1/2 levels, and had a moderate effect on pSTAT1 levels. The effect of selumetinib on PD-L1 membrane expression was confirmed in additional cell lines (Fig. S4A). Moreover, selumetinib decreased basal PD-L1 membrane expression of H292 and H358 cells (Fig. S3B). We used a wide range of selumetinib and erlotinib concentrations to determine if the reduction in PD-L1 expression is only observed at high drug concentrations. After 24 hours even the lowest concentration (0.1 µM) selumetinib strongly reduced pERK levels, used as a read-out for MAPK pathway activity, as well as PD-L1 protein and membrane expression levels of H292 and H358 cells both in control and EGF- and IFNγ-stimulated cells (Fig. S3C). Treatment of the cells with this selumetinib concentration for 96 hours resulted in a growth reduction of 30-50% (Fig. S3D), indicating that PD-L1 expression can be manipulated with a MAPK activity inhibitor using non-cytostatic concentrations. In line with these results, selumetinib had a similar moderate effect on growth in the other three cell lines (Fig. S4B). Similar results were observed with erlotinib. Concluding, MAPK pathway inhibition suppresses EGF- and IFNγ-induced PD-L1 mRNA, protein and membrane expression at non-cytostatic concentrations.

Hepatocyte growth factor induces PD-L1 surface expression via the MAPK pathway We investigated whether activation of the MAPK pathway via another growth factor receptor, hepatocyte growth factor receptor (cMET), has a similar effect on PD-L1 expression as MAPK activation by EGFR. Overexpression of cMET and its ligand HGF occur frequently in lung adenocarcinoma tumors (26). Moreover, upon binding of hepatocyte growth factor (HGF), cMET is known to activate PI3K-mTOR, MAPK, and JAK/ STAT pathways, similar to EGFR (27). As expected, HGF enhanced PD-L1 expression and augmented IFNγ-induced PD-L1 expression in the cMET-positive (data not shown) H292 and H358 cell lines (Fig. 4C). Combining HGF and EGF had no additional effect on PD-L1 membrane expression compared to single EGF or HGF treatment. Also in this case, selumetinib effectively prevented HGF-induced effects on PD-L1 expression levels in both cell lines, while erlotinib only showed efficacy in H292 cells. Taken together, these

6

results demonstrate that, irrespective of the upstream growth factor receptor, MAPK pathway activation is essential to raise PD-L1 membrane expression. This also implies that direct targeting of MAPK pathway is preferred to decrease PD-L1 expression. MAPK pathway inhibition does not interfere with IFNγ-induced MHC-I upregulation Antigen presentation at the tumor cell membrane in the context of major histocompatibility complex class I (MHC-I) is important for tumor cell recognition by the

Figure 5

A PD-L1 mRN A (f ol d chang e) IFNγ ActD EGF Selume�nib + -+ + + + + + + + + -H292 H358 2.5 x 105 2.0 x 105 1.5 x 105 1.0 x 105 0.5 x 105 0 PD -L1 (MFI)

Con EGF IFNγ E+I Con EGF IFNγ E+I

H292 H358 2.5 x 105 2.0 x 105 1.5 x 105 1.0 x 105 0.5 x 105 0 PD -L1 (MFI) B Control BMS911543

Con EGF IFNγ E+I Con EGF IFNγ E+I

Control BMS911543 PD-L1 pSTAT1 GAPDH PD-L1 pSTAT1 GAPDH PD-L1 mRN A (f ol d chang e) IFNγ ActD EGF Selume�nib + -+ + + + + + + + 1.0 0.8 0.6 0.4 0.2 0 1.0 0.8 0.6 0.4 0.2 0 ** *** ** ** *** ** * +- ++ - - - -- -- +- +- ++ Figure 5:

MAPK signaling increases PD-L1 mRNA stability. (A) Cells were treated with BMS911543 in the presence of 20 ng/mL EGF and IFNγ. After 24 hours PD-L1 membrane expression was measured using flow cytometry and cellular protein levels were measured using Western blotting. Blot from a representative experiment. N = 2. (B) H292 and H358 were treated with IFNγ for 24 hours. After washing, cells were treated with 5 µg/mL actinomycin for 10 minutes, after which 20 ng/mL IFNγ, 20 ng/mL EGF, or 10µM selumetinib were added for 80 minutes. RNA was harvested and PD-L1 mRNA levels were measured using RT-qPCR. Data was analyzed using the double delta CT method and GAPDH as a loading control. (Two-way ANOVA with Tukey test * P < 0.05, ** P < 0.01, *** P < 0.001). Con = untreated control, E+I = EGF + IFNγ, ActD = actinomycin D.

Chapter 6

C

H292 H358 PD-L1 (MFI) 3.0 x 105 2.0 x 105 1.0 x 105 0 3.0 x 106 2.0 x 106 1.0 x 106 0 MHC-I (MFI) H292 H358 *** * *** _*** ** ** *** *** ** ** *** ****** *** * *** * ** *** * ** * **

B

EGF + ++ + + + + -- -PBMC - R A A A A A A Erlo�nib Selume�nib - -+ ++ + + + + -- -- R A A A A A A -EGF PBMC Erlo�nib Selume�nib + + + + + + + -- -- R A A A A A A -+ ++ + + + + -- -- R A A A A A A

-Figure 6

A

c-MET HGF EGF EGFR MAPK signaling Wild-type RAS Degrada�on Transcrip�on Cell membrane STAT IFNγR1/2 JAK JAK ST AT P Nucleus PD-L1 Protein PD-L1 mRNA IFNγ P P P P P P STAT ST AT P STATP PSTAT KRAS muta�on * Figure 6:

Erlotinib and selumetinib prevent PBMC-induced PD-L1, but not MHC-I expression in NSCLC cells. (A) H292 and H358 cells were cocultured with 72 hour pre-activated PBMCs from healthy volunteers at a ratio of 5 PBMCs per tumor cell. During coculture cells were treated with 20 ng/mL EGF, and 10 μM erlotinib or 10 μM selumetinib. After 24 hours, PD-L1 membrane expression (Two-way ANOVA with Bonferroni’s multiple comparisons method ** P < 0.01, *** P < 0.01 compared to control + pre-activated PBMCs). (B) and MHC-I membrane expression were measured using flow cytometry (Two-way ANOVA * P < 0.05, ** P < 0.01 compared to control). (C) Proposed model for the role of IFNγ and MAPK signaling in PD-L1 regulation of lung adenocarcinoma. IFNγ derived from tumor infiltrating immune cells induces transcription of PD-L1 in tumor cells through activation of JAK/STAT-signaling. PD-L1 mRNA is translated into PD-L1 protein, which is transported to the cell membrane. Growth receptor- and KRAS mutation-induced MAPK signaling increases STAT signaling, potentially adding to transcriptional activity. Also, MAPK signaling increases stability of PD-L1 mRNA, resulting in increased mRNA and protein levels, and subsequently increasing PD-L1 membrane expression. R = resting PBMCs, A = activated PBMCs.

C

H292 H358

PD-L1 (MFI)

3.0 x 105 2.0 x 105 1.0 x 105 0 3.0 x 106 2.0 x 106 1.0 x 106 0

MHC-I (MFI)

H292 H358 *** * *** _*** ** ** *** *** ** ** *** ****** *** * *** * ** *** * ** * **

B

EGF

+

+

+

+

+

+

+

--

-PBMC

-

R A A A A A A Erlo�nib Selume�nib

-

--

+

+

+

+

+

+

+

--

--

R A A A A A A

-EGF PBMC Erlo�nib Selume�nib

+

+

+

+

+

+

+

--

--

R A A A A A A

--

+

+

+

+

+

+

+

--

--

R A A A A A A

-Figure 6

A

c-MET HGF EGF EGFR

MAPK signaling

Wild-type RAS Degrada�on Transcrip�on Cell membrane STAT IFNγR1/2 JAK JAK ST AT P Nucleus PD-L1 Protein PD-L1 mRNA IFNγ P P P P P P STAT ST AT P STAT P P STAT KRAS muta�on *

MAPK pathway activity plays a key role in PD-L1 expression of EGFR wild-type lung adenocarcinoma cells

6

C

H292 H358

PD-L1 (MFI)

3.0 x 105 2.0 x 105 1.0 x 105 0 3.0 x 106 2.0 x 106 1.0 x 106 0

MHC-I (MFI)

H292 H358 *** * *** _*** ** ** *** *** ** ** *** ****** *** * *** * ** *** * ** * **

B

EGF

+

+

+

+

+

+

+

--

-PBMC

-

R A A A A A A Erlo�nib Selume�nib

-

--

+

+

+

+

+

+

+

--

--

R A A A A A A

-EGF PBMC Erlo�nib Selume�nib

+

+

+

+

+

+

+

--

--

R A A A A A A

--

+

+

+

+

+

+

+

--

--

R A A A A A A

-Figure 6

A

c-MET HGF EGF EGFR

MAPK signaling

Wild-type RAS Degrada�on Transcrip�on Cell membrane STAT IFNγR1/2 JAK JAK ST AT P Nucleus PD-L1 Protein PD-L1 mRNA IFNγ P P P P P P STAT ST AT P STAT P P STAT KRAS muta�on *

immune system and induction of an anti-tumoral immune response (28). While IFNγ is an important inducer of MHC-I membrane expression, EGFR and MEK1/2 signaling have been described as suppressors (29–31). Hence, we studied the effect of EGF and IFNγ on MHC-I membrane expression in the presence of EGFR and MEK1/2 blockade. IFNγ increased MHC-I membrane expression in 4 out of 5 cell lines (Fig. 4D, S4C). EGF had no effect on MHC-I membrane expression. Treatment with erlotinib or selumetinib did not counteract IFNγ-induced upregulation of MHC-I membrane expression, suggesting that MHC-I-mediated tumor cell recognition by immune cells will not be impaired by these drugs.

MAPK signaling increases stability of PD-L1 mRNA

Because both MHC-I and PD-L1 transcription are induced by STAT1-signaling, we wondered why only PD-L1 expression is influenced by MAPK signaling (5,32). Suppression of STAT1 signaling using JAK2 inhibitor BMS911543 prevented IFNγ-induced PD-L1 and MHC-I expression, but not EGF-induced PD-L1 expression (Fig. 5A, S5A). Also, inhibition of STAT3 using an siRNA had no influence on PD-L1 regulation by MAPK signaling (Fig. S5B). KRAS mutations were recently shown to be involved in posttranscriptional regulation of basal PD-L1 levels through modulation of PD-L1 mRNA stability (33). To study whether MAPK signaling controls stability of IFNγ-induced PD-L1 mRNA, KRAS wild-type and mutant cells were pretreated with IFNγ followed by the addition of the transcriptional blocker actinomycin D (34). PD-L1 mRNA levels had halved 90 minutes after start of the transcriptional inhibition (Fig. 5B). Interestingly, degradation of PD-L1 mRNA in the presence of actinomycin D was counteracted by EGF-induced activation of MAPK signaling. Accordingly, inhibition of MAPK signaling with selumetinib accelerated PD-L1 mRNA degradation and decreased the PD-L1 stabilization by EGF. These results show that MAPK signaling influences stability of PD-L1 mRNA, contributing to regulation of PD-L1 protein and membrane expression.

MAPK pathway inhibition decreases PBMC-induced PD-L1 surface expression

To study the relation between immune cell activation and PD-L1 expression of tumor cells, we performed cocultures of PBMCs and NSCLC cells. After 24 hours of coculture, membrane expression of PD-L1 and MHC-I were strongly induced in tumor cells (Fig. 6A, B). PD-L1 expression further increased in the presence of EGF. Conditioned medium from activated PBMCs also strongly induced PD-L1 membrane expression of NSCLC cells, indicating the involvement of secreted soluble factors (Fig. S6A). A causal factor can be IFNγ, which levels reached up to 30 ng/mL in conditioned medium of activated PBMCs (Fig. S6B). These levels are higher than we had used to stimulate these NSCLC cells in the aforementioned experiments. Treatment with erlotinib or selumetinib during these cocultures decreased tumor cell PD-L1 upregulation, irrespective of the presence of EGF,

but had no effect on MHC-I upregulation in tumor cells (Fig. 6A,B). Our results indicate that MAPK pathway inhibition can reduce tumor cell PD-L1 expression in a more complex system, without interfering with MHC-I induction in tumor cells, potentially improving immunogenicity of these cells.

DISCUSSION

In this study we reveal a correlation between MAPK pathway activation and PD-L1 gene expression in lung adenocarcinomas but not in squamous cell lung carcinomas using TCGA RNA sequencing data from EGFR wild-type NSCLC tumors. Subsequently, we demonstrate the importance of MAPK signaling in the upregulation of PD-L1 by growth factors and IFNγ in EGFR wild-type lung adenocarcinoma cell lines, which is mediated by MAPK-dependent regulation of PD-L1 mRNA stability (Fig. 6C). Inhibition of the MAPK pathway prevents PD-L1 upregulation, whereas it does not interfere with IFNγ-induced MHC-I upregulation. Taken together, these results indicate that MAPK pathway inhibition may improve tumor cell immunogenicity of EGFR wild-type lung adenocarcinomas, comprising over 80% of all lung adenocarcinoma tumors in the Western World (35). In the present study, a key role for MAPK pathway activity in the induction of PD-L1 expression is demonstrated. Our TCGA analysis suggests that a connection between MAPK activity and PD-L1 expression is primarily present in EGFR wild-type lung adenocarcinoma. Interestingly, in this NSCLC subtype PD-L1 RNA expression correlates with poor prognosis and second line PD-1 blocking is effective, contrary to findings in squamous cell lung carcinoma (36,37). The correlation between inferred MAPK signaling and PD-L1 mRNA expression in lung adenocarcinoma may result from higher dependency on the MAPK pathway in this subtype due to genetic alteration patterns and frequent overexpression of EGFR and cMET compared to squamous cell lung carcinoma (38). Our results demonstrate a relation between IFNγ signaling, growth factor-induced MAPK signaling and PD-L1 expression in vitro. Moreover, we show that MAPK signaling acts on the stability of IFNγ-induced PD-L1 mRNA. We provide evidence that the massive induction of PD-L1 mRNA by EGF and IFNγ in KRAS wild-type and mutant cells can be prevented by MAPK pathway inhibition, which causes a reduction in PD-L1 mRNA stability. In line with these findings, MAPK pathway activation increases stability of PD-L1 mRNA. This post-transcriptional mechanism has recently been described for basal PD-L1 expression in KRAS mutant cell lines, including NSCLC (33). The strong involvement of MAPK activity in PD-L1 mRNA stability might explain why MAPK inhibition does not affect MHC-I expression, although PD-L1 and MHC-I are both transcriptional targets of STAT1 (5,32). Since MAPK signaling also modestly influences pSTAT1 levels, its involvement in PD-L1 transcription cannot be excluded (Fig. 6C). At the protein level, PD-L1 expression can be affected by several mechanisms such as glycosylation, ubiquitination, and

6

stabilization at the cell membrane (13,14,39,40). However, we found no direct effect of EGF or IFNγ on CKLF-like MARVEL transmembrane domain containing protein 6 (CMTM6) protein levels in H292 or H358 cells (data not shown). Interestingly, western blotting of PD-L1 consistently demonstrated a strong band between 45 and 60 kDa, far above the expected molecular weight of 32 kDa, in H292 and H358. This suggests high PD-L1 glycosylation in these cells, which is known to be essential for its interaction with PD-1 (41). Importantly, expression of this supposedly glycosylated PD-L1 is dependent on MAPK pathway activity.

Our in vitro experiments demonstrated that both EGFR and MEK1/2 inhibitors decrease EGF and IFNy induced PD-L1 expression, potentially increasing immunogenicity of lung adenocarcinoma cells. Nevertheless, because the MAPK pathway is downstream of a plethora of growth factor receptors, downstream inhibition with MEK1/2 inhibitors may be more effective to modulate PD-L1 expression than inhibition of specific growth factor receptors. This is supported by our finding that MEK1/2 inhibition, but not EGFR inhibition, prevented HGF-induced PD-L1 expression and by earlier findings in renal cell carcinoma (42). In vitro we observed PD-L1 downregulation at non-cytostatic selumetinib concentrations, indicating an immunomodulatory effect may already be present at lower doses than previously utilized in patients (43). In other cancer types the immunomodulatory role of MAPK signaling is also increasingly being recognized (44). Multiple studies using in vivo colon cancer models showed that MEK inhibition potentiates the anti-tumor immune response by preventing T cell apoptosis and decreasing levels of myeloid suppressor cells and regulatory T cells. This resulted in sustained tumor regression when combined with PD-L1, PD-1, or cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) blocking treatment (45–47). In two studies in pretreated NSCLC patients, adding selumetinib to either docetaxel or erlotinib did not improve survival, but resulted in an increased level of circulating T cells and decreased level of regulatory T cells, indicating that selumetinib had an immune-modulating effect (48,49). In EGFR wild-type NSCLC patients with PD-L1 expression and high mutational load, combining inhibitors of PD-1 and CTLA-4 resulted in a trend towards better progression free survival compared to targeting PD-1 alone (50). Combining EGFR, cMET or MEK1/2 inhibition with checkpoint inhibition of for example PD-1/PD-L1, or CTLA-4 might improve treatment efficacy, especially in EGFR wild-type lung adenocarcinoma, which is more immunogenic than EGFR mutant adenocarcinoma (1,2). These combination strategies are currently being tested in NSCLC patients (NCT03600701, NCT03299088).

In conclusion, our results show the importance of growth factor induced MAPK pathway signaling in PD-L1 expression in EGFR wild-type lung adenocarcinoma. This provides a rationale to explore the combination of selumetinib with immune checkpoint inhibitors

in EGFR wild-type lung adenocarcinoma to improve immunogenicity of cancer cells by decreasing PD-L1 expression.

ACKNOWLEDGEMENTS

This work is supported by a POINTING grant of the Dutch Cancer Society to E.G.E. de Vries. T.S. Stutvoet is supported by a fellowship of the Junior Scientific Master Class (JSM) of the University of Groningen.

 

REFERENCES

1. Dong ZY, Zhang JT, Liu SY, Su J, Zhang C, Xie Z, Zhou Q, Tu HY, Xu CR, Yan LX, Li YF, Zhong WZ, et al. EGFR mutation correlates with uninflamed phenotype and weak immunogenicity, causing impaired response to PD-1 blockade in non-small cell lung cancer. Oncoimmunology 2017;6.

2. Zhang M, Li G, Wang Y, Wang Y, Zhao S, Haihong P, Zhao H, Wang Y. PD-L1 expression in lung cancer and its correlation with driver mutations: a meta-analysis. Sci Rep 2017;7:10255.

3. Sacher AG, Gandhi L. Biomarkers for the clinical use of PD-1/PD-L1 inhibitors in non-small-cell lung cancer: A review. JAMA Oncol 2016;2:1217–22.

4. Ferrara R, Mezquita L, Besse B. Progress in the management of advanced thoracic malignancies in 2017. J Thorac Oncol 2018;13:301–22.

5. Sun C, Mezzadra R, Schumacher TN. Regulation and function of the PD-L1 checkpoint. Immunity 2018;48:434–52.

6. Lee S-J, Jang B-C, Lee S-W, Yang Y-I, Suh S-I, Park Y-M, Oh S, Shin J-G, Yao S, Chen L, Choi I-H. Interferon regulatory factor-1 is prerequisite to the constitutive expression and IFN-gamma-induced upregulation of B7-H1 (CD274). FEBS Lett 2006;580:755– 62.

7. Schalper KA, Carvajal-Hausdorf D, McLaughlin J, Altan M, Velcheti V, Gaule P, Sanmamed MF, Chen L, Herbst RS, Rimm DL. Differential expression and significance of PD-L1, IDO-1, and B7-H4 in human lung cancer. Clin Cancer Res 2017;23:370–8. 8. Califano R, Kerr K, Morgan RD, Lo Russo G, Garassino M, Morgillo F, Rossi A. Immune

checkpoint blockade: a new era for non-small cell lung cancer. Curr Oncol Rep 2016;18:59.

9. Chen N, Fang W, Zhan J, Hong S, Tang Y, Kang S, Zhang Y, He X, Zhou T, Qin T, Huang Y, Yi X, et al. Upregulation of PD-L1 by EGFR activation mediates the immune escape in EGFR-driven NSCLC: implication for optional immune targeted therapy for NSCLC patients with EGFR mutation. J Thorac Oncol 2015;10:910–23.

10. Ota K, Azuma K, Kawahara A, Hattori S, Iwama E, Tanizaki J, Harada T, Matsumoto K, Takayama K, Takamori S, Kage M, Hoshino T, et al. Induction of PD-L1 expression by the EML4-ALK oncoprotein and downstream signaling pathways in non-small cell

6

lung cancer. Clin Cancer Res 2015;21:4014–21.

11. Concha-Benavente F, Srivastava RM, Trivedi S, Lei Y, Chandran U, Seethala RR, Freeman GJ, Ferris RL. Identification of the cell-intrinsic and -extrinsic pathways downstream of EGFR and IFNγ that induce PD-L1 expression in head and neck cancer. Cancer Res 2016;76:1031–43.

12. Sumimoto H, Takano A, Teramoto K, Daigo Y. RAS-mitogen-activated protein kinase signal is required for enhanced PD-L1 expression in human lung cancers. PLoS One 2016;11:e0166626.

13. Li C-W, Lim S-O, Xia W, Lee H-H, Chan L-C, Kuo C-W, Khoo K-H, Chang S-S, Cha J-H, Kim T, Hsu JL, Wu Y, et al. Glycosylation and stabilization of programmed death ligand-1 suppresses T-cell activity. Nat Commun 2016;7:12632.

14. Horita H, Law A, Hong S, Middleton K. Identifying regulatory posttranslational modifications of PD-L1: a focus on monoubiquitination. Neoplasia 2017;19:346–53. 15. Lastwika KJ, Wilson W, Li QK, Norris J, Xu H, Ghazarian SR, Kitagawa H, Kawabata S, Taube JM, Yao S, Liu LN, Gills JJ, et al. Control of PD-L1 expression by oncogenic activation of the AKT-mTOR pathway in non-small cell lung cancer. Cancer Res 2016;76:227–38.

16. Imielinski M, Berger AH, Hammerman PS, Hernandez B, Pugh TJ, Hodis E, Cho J, Suh J, Capelletti M, Sivachenko A, Sougnez C, Auclair D, et al. Mapping the hallmarks of lung adenocarcinoma with massively parallel sequencing. Cell 2012;150:1107–20. 17. Network TCGAR. Comprehensive genomic characterization of squamous cell lung

cancers. Nature 2014;489:519–25.

18. Cerami E, Gao J, Dogrusoz U, Gross BE, Sumer SO, Aksoy BA, Jacobsen A, Byrne CJ, Heuer ML, Larsson E, Antipin Y, Reva B, et al. The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics. Cancer Discov 2012;2:401–4.

19. Loboda A, Nebozhyn M, Klinghoffer R, Frazier J, Chastain M, Arthur W, Roberts B, Zhang T, Chenard M, Haines B, Andersen J, Nagashima K, et al. A gene expression signature of RAS pathway dependence predicts response to PI3K and RAS pathway inhibitors and expands the population of RAS pathway activated tumors. BMC Med Genomics 2010;3:26.

20. Dry JR, Pavey S, Pratilas CA, Harbron C, Runswick S, Hodgson D, Chresta C, McCormack R, Byrne N, Cockerill M, Graham A, Beran G, et al. Transcriptional pathway signatures predict MEK addiction and response to selumetinib (AZD6244). Cancer Res 2010;70:2264–73.

21. Creighton CJ, Fu X, Hennessy BT, Casa AJ, Zhang Y, Gonzalez-Angulo AM, Lluch A, Gray JW, Brown PH, Hilsenbeck SG, Osborne CK, Mills GB, et al. Proteomic and transcriptomic profiling reveals a link between the PI3K pathway and lower estrogen-receptor (ER) levels and activity in ER+ breast cancer. Breast Cancer Res

2010;12:R40.

22. Livak KJ, Schmittgen TD. Analysis of relative gene expression data using real-time quantitative PCR and the 2-ΔΔCT method. Methods 2001;25:402–8.

23. Wang R, Li Y, Hu E, Kong F, Wang J, Liu J, Shao Q. S100A7 promotes lung adenocarcinoma to squamous carcinoma transdifferentiation, and its expression is differentially regulated by the Hippo-YAP pathway in lung cancer cells. 2017;8:24804–14. 24. Sigismund S, Woelk T, Puri C, Maspero E, Tacchetti C, Transidico P, Di Fiore PP, Polo

S. Clathrin-independent endocytosis of ubiquitinated cargos. Proc Natl Acad Sci U S A 2005;102:2760–5.

25. Qin A, Coffey DG, Warren EH, Ramnath N. Mechanisms of immune evasion and current status of checkpoint inhibitors in non-small cell lung cancer. Cancer Med 2016;5:2567–78.

26. Matsumura Y, Umemura S, Ishii G, Tsuta K, Matsumoto S, Aokage K, Hishida T, Yoshida J, Ohe Y, Suzuki H, Ochiai A, Goto K, et al. Expression profiling of receptor tyrosine kinases in high-grade neuroendocrine carcinoma of the lung: a comparative analysis with adenocarcinoma and squamous cell carcinoma. J Cancer Res Clin Oncol 2015;141:2159–70.

27. Gelsomino F, Facchinetti F, Haspinger ER, Garassino MC, Trusolino L, De Braud F, Tiseo M. Targeting the MET gene for the treatment of non-small-cell lung cancer. Crit Rev Oncol Hematol 2014;89:284–99.

28. Leone P, Shin E-C, Perosa F, Vacca A, Dammacco F, Racanelli V. MHC class I antigen processing and presenting machinery: organization, function, and defects in tumor cells. J Natl Cancer Inst 2013;105:1172–87.

29. Brea EJ, Oh CY, Manchado E, Budhu S, Gejman RS, Mo G, Mondello P, Han JE, Jarvis CA, Ulmert D, Xiang Q, Chang AY, et al. Kinase regulation of human MHC class I molecule expression on cancer cells. Cancer Immunol Res 2016;4:936–47.

30. Mimura K, Kua L-F, Shiraishi K, Kee Siang L, Shabbir A, Komachi M, Suzuki Y, Nakano T, Yong W-P, So J, Kono K. Inhibition of mitogen-activated protein kinase pathway can induce upregulation of human leukocyte antigen class I without PD-L1-upregulation in contrast to interferon-γ treatment. Cancer Sci 2014;105:1236–44.

31. Im JS, Herrmann AC, Bernatchez C, Haymaker C, Molldrem JJ, Hong WK, Perez-Soler R. Immune-modulation by epidermal growth factor receptor inhibitors: implication on anti-tumor immunity in lung cancer. PLoS One 2016;11:e0160004.

32. Zhou F. Molecular mechanisms of IFN-gamma to up-regulate MHC class I antigen processing and presentation. Int Rev Immunol 2009;28:239–60.

33. Coelho MA, de Carné Trécesson S, Rana S, Zecchin D, Moore C, Molina-Arcas M, East P, Spencer-Dene B, Nye E, Barnouin K, Snijders AP, Lai WS, et al. Oncogenic RAS signaling promotes tumor immunoresistance by stabilizing PD-L1 mRNA. Immunity 2017;47:1083–1099.e6.

6

34. Bensaude O. Inhibiting eukaryotic transcription: Which compound to choose? How to evaluate its activity? Transcription 2011;2:103–8.

35. Midha A, Dearden S, McCormack R. EGFR mutation incidence in non-small-cell lung cancer of adenocarcinoma histology: A systematic review and global map by ethnicity (mutMapII). Am J Cancer Res 2015;5:2892–911.

36. Yang H, Chen H, Luo S, Li L, Zhou S, Shen R, Lin H, Xie X. The correlation between programmed death-ligand 1 expression and driver gene mutations in NSCLC. Oncotarget 2017;8:23517–28.

37. Herbst RS, Baas P, Kim DW, Felip E, Pérez-Gracia JL, Han JY, Molina J, Kim JH, Arvis CD, Ahn MJ, Majem M, Fidler MJ, et al. Pembrolizumab versus docetaxel for previously treated, PD-L1-positive, advanced non-small-cell lung cancer (KEYNOTE-010): A randomised controlled trial. Lancet 2016;387:1540–50.

38. Pikor LA, Ramnarine VR, Lam S, Lam WL. Genetic alterations defining NSCLC subtypes and their therapeutic implications. Lung Cancer 2013;82:179–89.

39. Mezzadra R, Sun C, Jae LT, Gomez-Eerland R, De Vries E, Wu W, Logtenberg MEW, Slagter M, Rozeman EA, Hofland I, Broeks A, Horlings HM, et al. Identification of CMTM6 and CMTM4 as PD-L1 protein regulators. Nature 2017;549:106–10. 40. Burr ML, Sparbier CE, Chan YC, Williamson JC, Woods K, Beavis PA, Lam EYN,

Henderson MA, Bell CC, Stolzenburg S, Gilan O, Bloor S, et al. CMTM6 maintains the expression of PD-L1 and regulates anti-tumour immunity. Nature 2017;549:101–5. 41. Li CW, Lim S, Chung EM, Kim YS, Park AH, Yao J, Cha JH, Xia W, Chan LC, Kim T, Chang

S-S, Lee H-H, et al. Eradication of triple-negative breast cancer cells by targeting glycosylated PD-L1. Cancer Cell 2018;33:187–201.e10.

42. Balan M, Mier y Teran E, Waaga-Gasser AM, Gasser M, Choueiri TK, Freeman G, Pal S. Novel roles of c-Met in the survival of renal cancer cells through the regulation of HO-1 and PD-L1 expression. J Biol Chem 2015;290:8110–20.

43. Kim C, Giaccone G. MEK inhibitors under development for treatment of non-small-cell lung cancer. Expert Opin Investig Drugs 2018;27:17–30.

44. Bedognetti D, Roelands J, Decock J, Wang E, Hendrickx W. The MAPK hypothesis: immune-regulatory effects of MAPK-pathway genetic dysregulations and implications for breast cancer immunotherapy. Emerg Top Life Sci 2017;1:429–45. 45. Liu L, Mayes PA, Eastman S, Shi H, Yadavilli S, Zhang T, Yang J, Seestaller-Wehr

L, Zhang S-Y, Hopson C, Tsvetkov L, Jing J, et al. The BRAF and MEK inhibitors dabrafenib and trametinib: effects on immune function and in combination with immunomodulatory antibodies targeting PD-1, PD-L1, and CTLA-4. Clin Cancer Res 2015;21:1639–51.

46. Poon E, Mullins S, Watkins A, Williams GS, Koopmann J-O, Di Genova G, Cumberbatch M, Veldman-Jones M, Grosskurth SE, Sah V, Schuller A, Reimer C, et al. The MEK inhibitor selumetinib complements CTLA-4 blockade by reprogramming the tumor

immune microenvironment. J Immunother cancer 2017;5:63.

47. Ebert PJR, Cheung J, Yang Y, McNamara E, Hong R, Moskalenko M, Gould SE, Maecker H, Irving BA, Kim JM, Belvin M, Mellman I. MAP kinase inhibition promotes T cell and anti-tumor activity in combination with PD-L1 checkpoint blockade. Immunity 2016;44:609–21.

48. Jänne PA, van den Heuvel MM, Barlesi F, Cobo M, Mazieres J, Crinò L, Orlov S, Blackhall F, Wolf J, Garrido P, Poltoratskiy A, Mariani G, et al. Selumetinib plus docetaxel compared with docetaxel alone and progression-free survival in patients with KRAS-mutant advanced non-small cell lung cancer: the SELECT-1 randomized clinical trial. JAMA 2017;317:1844–53.

49. Carter CA, Rajan A, Keen C, Szabo E, Khozin S, Thomas A, Brzezniak C, Guha U, Doyle LA, Steinberg SM, Xi L, Raffeld M, et al. Selumetinib with and without erlotinib in KRAS mutant and KRAS wild-type advanced nonsmall-cell lung cancer. Ann Oncol 2016;27:693–9.

50. Hellmann MD, Ciuleanu T-E, Pluzanski A, Lee JS, Otterson GA, Audigier-Valette C, Minenza E, Linardou H, Burgers S, Salman P, Borghaei H, Ramalingam SS, et al. Nivolumab plus ipilimumab in lung cancer with a high tumor mutational burden. N Engl J Med 2018;

SUPPLEMENTAL TABLE AND FIGURES

Supplementary table 1: Correlation between inferred MAPK and PI3K pathway activity, IFNγ

signaling gene expression, and PD-L1 gene expression in NSCLC subtypes

Adenocarcinoma Squamous cell carcinoma

Total

N = 197 KRASwt N = 123 KRASmt N = 74 N = 172 Total KRASwt N = 170

RAS-score rs p 4.1 x 10.38 -8 _{3.5 x 10}.48 -8 .14 _ns -.032 _ns -.04 _ns MEK-score rs p 5.7 x 10.40 -9 _{7.2 x 10}.47 -8 .27 _ns .05 _ns .04 _ns PI3K-score rs p .07 ns .02 ns .18 ns .10 ns .11 ns STAT1 rs p <2.2 x 10.57 -16 _{5.5 x 10}.54 -11 _{<2.2 x 10}.68 -16 _{4.1 x 10}.41 -8 _{1.2 x 10}.40 -7 STAT3 rs p -.20 ns -.20 ns -.17 ns .07 ns .07 ns IRF-1 rs p <2.2 x 10.52 -16 _{<2.2 x 10}.56 -16 _{3.6 x 10}.45 -3 _{1.9 x 10}.31 -3 _{2.9 x 10}.30 -3 wt = wild-type; mt = mutant; rs = Spearman’s rho; ns = not significant

6

p < 0.001

Wild type Mutated KRAS status 2 1 0 -1 -2 p < 0.001

Wild type Mutated 2 1 0 -1 -2 KRAS status 2 1 0 -1 -2 -3 1 0 1 2 - -2 1 0 -1 -2 -3 1 0 1 2 -

-A

B

rs = .50 p < 2.2 x 10-16 r s = .55 p < 2.2 x 10-16 Adenocarcinoma LUSC Adenocarcinoma Adenocarcinoma RAS-activation s core MEK -ac tivation score RAS-activation s core RAS-activation s core

MEK-activation score MEK-activation score

rs = .40 p = 1.4 x 10-6 2 1 0 -1 -2 -1 0 1 2 -2 STAT1 expression (Z-score) LUSC 2 1 0 -1 -2 -1 0 1 2 -2 IRF-1 expression (Z-score) LUSC rs = .31 p = 1.2 x 10-3 PD-L1 expression (Z-score) PD-L1 expression (Z-score)

C

Figure S1:

MAPK pathway activation scores. All EGFR wild-type samples with RNA sequencing data were collected from the TCGA adenocarcinoma and squamous cell carcinoma datasets. (A) In the selected lung adenocarcinoma samples the difference between MAPK pathway activation in KRAS wild-type and mutant tumors was determined using a RAS-activation score and MEK-activation score. Statistical analysis was performed using Wilcoxon rank-sum. (B) In the selected lung adenocarcinoma and squamous cell carcinoma samples the correlation between a RAS-activation and MEK-activation score was calculated using spearman correlation (19,20). (C) In the selected squamous cell carcinoma samples the correlation of STAT1 and IRF-1 mRNA expression with PD-L1 mRNA expression was calculated using Spearman correlation. LUSC = squamous cell lung carcinoma.

A

B H292 H358

EGF IFN EGF + IFN

Time (min): 0 5 15 60 5 15 60 5 15 60

EGF IFN EGF + IFN

0 5 15 60 5 15 60 5 15 60 PD-L1 pEGFR pERK1/2 pSTAT1 pSTAT3 Actin 400 800 0 500 400 300 200 100 100 ₁₀2 ₁₀4 ₁₀6 ₁₀0 ₁₀2 ₁₀4 ₁₀6 ₁₀0 ₁₀2 ₁₀4 ₁₀6 100 ₁₀2 ₁₀4 ₁₀6 600 200 200 400 600 100 200 300 0 0 0 0 0 200 400 600 100 ₁₀2 ₁₀4 ₁₀6 200 400 600 800 100 ₁₀2 ₁₀4 ₁₀6 Fluorescence intensity # Cells H322 H292 A549 H358 H460 HCC827 0 PD-L1 Figure S2:

EGF and IFNγ increase PD-L1 expression in EGFR wild-type NSCLC cell lines. (A) PD-L1 membrane expression of NSCLC cells was measured during the exponential growth phase using flow cytometry. (B) H292 and H358 cells were treated with 20 ng/mL EGF, 20 ng/mL IFNγ, or both. After 5, 15, and 60 minutes, protein levels were measured using Western blotting. Actin was used as a loading control.

6

0 1 .0×1 05 2 .0×1 05 3 .0×1 05 4 .0×1 05 5 .0×1 05 PD -L 1 (M FI ) 5.0 x 105 4.0 x 105 3.0 x 105 2.0 x 105 1.0 x 105 0 PD -L1 (MFI)

Control EGF + IFN

Conc (µM): 0 0.1 1 10 0.1 1 10 0 0.1 1 10 0.1 1 10 5.0 x 105 4.0 x 105 3.0 x 105 2.0 x 105 1.0 x 105 0 PD -L1 (MFI) Conc (µM): 0 0.1 1 10 0.1 1 10 0 0.1 1 10 0.1 1 10 PD-L1 pEGFR pERK1/2 pSTAT1 pSTAT3 pS6

Control H358 EGF + IFN H292 150 100 50 0 + + + + + -PD-L1 mRNA (f old change ) H292 H358 A C 0 0.01 0.05 0.1 0.5 1 5 10 Erlo Selu 0.01 0.1 1 10 0 25 50 75 100 0 Ce ll vi ab ility (%) Ce ll vi ab ility (%) Selume nib Erlo nib H292 H358 Selume nib Erlo nib 0 25 50 75 100 0.01 0.1 1 10 0 0 0.01 0.05 0.1 0.5 1 5 10 Erlo Selu D PD-L1 pEGFR pERK1/2 pSTAT1 pSTAT3 pS6 Conc (µM) Conc (µM) Conc (µM) Conc (µM) H292 H358 10 x 105 7.5 x 104 5.0 x 104 2.5 x 104 0 PD -L1 (MFI) + + + + + -* * *** B Contr ol Contr ol Figure S3:

(A) H292 and H358 were treated with 20 ng/mL EGF and IFNγ, and 10 µM erlotinib or selumetinib. After 24 hours PD-L1 mRNA levels were measured using RT-qPCR. Data was analyzed using the double delta CT method and GAPDH as a loading control. (B) H292 and H358 cells were treated with 10 µM erlotinib or selumetinib for 24 hours. After 24 hours PD-L1 membrane expression was measured using flow cytometry. (Student’s T test compared to control * P < 0.05, *** P < 0.001, n = 8) (C) H292 and H358 cells were treated with varying concentration of erlotinib or selumetinib with and without co-treatment with 20 ng/mL EGF and IFNγ. After 24 hours, PD-L1 membrane expression was measured using flow cytometry and protein levels were measured using Western blotting. Actin was used as a loading control. (D) Cells were plated in 96-well plates and treated with different concentrations of erlotinib or selumetinib for 96 hours. Cell density was measured using crystal violet absorption. Pictures are from a representative experiment. Conc = concentration, Erlo = erlotinib, Selu = selumetinib.

0 1 .0×1 05 2 .0×1 05 3 .0×1 05 2.0 x 105 1.0 x 105 3.0 x 105

ConEGF ConEGF ConEGF ConEGF Con EGF ConEGF ConEGF Con EGF ConEGF

H322 A549 H460 0

A

Control Control Control

B

PD -L1 (MFI) 0 0.01 0.05 0.1 0.5 1 5 10 0 0.01 0.05 0.1 0.5 1 5 10 0 0.01 0.05 0.1 0.5 1 5 10 Conc (µM) Erlo Selu *** *** ** ** ** * ****** 0.01 0.1 1 10 0 25 50 75 100 0 Cell viability (%) H322 0.01 0.1 1 10 0 A549 0.01 0.1 1 10 0 Selu Erlo H460

C

1.0 x 106 2.5 x 106 2.0 x 106 5.0 x 105 0 MHC-I (MFI) 1.5 x 106 Con EGF _IFN H322 A549 H460 Erlotinib Selumetinib Control *** *** *

Con EGF _IFN Con EGF IFN

Conc (µM)

Figure S4:

Effects of erlotinib and selumetinib on PD-L1 and MHC-I expression and proliferation in a panel of NSCLC cell lines. (A) Cells were treated with 10 μM selumetinib or 10 μM erlotinib with and without co-treatment with 20 ng/mL EGF, 20 ng/mL IFNγ, or both. After 24 hours, PD-L1 membrane expression was measured using flow cytometry. (Two-way ANOVA * P < 0.05, ** P < 0.01, *** P < 0.001 compared to ligand-treated control) (B) Cells were plated in 96-well plates and treated with different concentrations of erlotinib or selumetinib for 96 hours. Cell density was measured using crystal violet absorption. Pictures are from a representative experiment. (C) Cells were treated with 20 ng/mL EGF, 20 ng/mL IFNγ, or both. After 24 hours MHC-I expression was measured using flow cytometry. (Student’s T test, * P < 0.05, *** P < 0.001). Conc = concentration, Erlo = erlotinib, Selu = selumetinib.

6

H358 H292 EGF + IFN 4.0 x 105 3.0 x 105 2.0 x 105 1.0 x 105 0 PD -L1 (MFI) C C _siC siST AT 3 Erlo Selu si + erlo si + selu EGF + IFN 4.0 x 105 3.0 x 105 2.0 x 105 1.0 x 105 0 PD -L1 (MFI ) pSTAT3 PD-L1 pMAPK GAPDH pSTAT3 PD-L1 pMAPK GAPDH C C _siC siST AT 3 Erlo Selu si + erlo si + selu

A

B

H358 H292 10 x 105 8.0 x 105 6.0 x 105 4.0 x 105 2.0 x 105 0 Con Con *** * ns ns MHC-I (MFI) 10 x 105 8.0 x 105 6.0 x 105 4.0 x 105 2.0 x 105 0 Con Con MHC-I (MFI) Control BMS911543 Control BMS911543 Figure S5:

(A) Cells were treated with BMS911543 in the presence of 20 ng/mL IFNγ. After 24 hours PD-L1 membrane expression was measured using flow cytometry. (Student’s T test compared to control: ns = not significant, * P < 0.05, *** P < 0.001) (B) Cells were transfected with siRNAs targeting STAT3. After 24 hours cells were treated with the indicated ligands and treatments. After a total of 48 hours, PD-L1 membrane expression was measured using flow cytometry. PD-L1, pSTAT3, pERK1/2, and GAPDH protein levels were detected using Western blotting. Blots are from a representative experiment. N = 3. C = untreated control, siC = non-targeting siRNA control, siSTAT3 = STAT3 siRNA, Erlo = erlotinib, Selu = selumetinib, si = STAT3 sRiNA.

50 40 30 20 10 0

ControlPBMCs Control PBMCs

No signal 3.0 x 105 2.0 x 105 1.0 x 105 0 + coculture

medium + coculturemedium

ControlEGFControlEGFControlEGFControlEGF

PD-L1 (MFI)

H292 H358

A

B

Figure S6:

Conditioned medium from activated PBMCs contains IFNγ and increases PD-L1 membrane expression. (A) H292 and H358 were cultured in supernatant from PBMCs activated in coculture, with or without EGF. After 24 hours, PD-L1 expression was measured using flow cytometry. (B) IFNγ concentration was measured in medium after coculture using ELISA.