Supplementary Table 1A, B, C, D and E

Supplementary methods and materials

Objectives: Immune checkpoint inhibitors (ICIs) have provided a breakthrough in the treatment of non-small cell lung cancer (NSCLC) patients, but only some patients benefit substantively. Identifying definitive predictive biomarkers could overcome this limitation. Materials and methods: We selected 146 metastatic NSCLC patients treated with anti-PD-(L)1. Immunohistochemistry of HLA-I, PD-L1 and CD73 was performed in 122 tumor biopsies at diagnosis. The association with patients, tumor parameters, and the predictive value to ICI treatment were determined. Results: In our cohort, 42 %, 25 %, and 21 % of the tumors exhibited high levels of HLA-I, PD-L1, and CD73, respectively. Lung adenocarcinomas displayed elevated CD73 levels, compared with lung squamous cell carcinomas (P = 0.026). High PD-L1 was significantly correlated with high levels of HLA-I (P = 0.005) and of CD73 (P = 0.025). Patients with high-level HLA-I tumors exhibited more favorable clinical outcomes following ICI, with a median overall survival of 30.7 months (95 % confidence interval [CI]: 18.3 months-not reached), compared with 18.2 months (95 % CI: 12.4-25.2 months) in patients with low-level HLA-I tumors (P = 0.016). The median progression-free survival (PFS) for patients with high-level HLA-I tumors was 18.5 months (95 % CI: 11.1-57.1 months), longer than patients with low-level HLA-I tumors, whose median PFS was 9.2 months (95 % CI: 7.2-11.9 months) (P = 0.006). In a multivariable analysis, high-level HLA-I was independently associated with lower risk of progression to ICI (HR = 0.46, 95 % CI 0.24-0.87; P = 0.018). Conclusions: High-level HLA-I were associated with better clinical outcomes to ICI in our cohort of NSCLC patients. Therefore, further investigations are warranted to refine this biomarker and validate its efficacy in prospective and larger set of patients.

PDF file 236K, Table S1. List of the 121 lung cancer cell lines screened for MAX alterations. Information about the histopathology of each cell line, and the presence of alterations at MYC and BRG1 is also included. Grey boxes indicate that no information is available. Table S2. List of genes that are up-regulated or down-regulated upon MAX reconstitution. The values represent the n-fold change in the level of gene expression of each of the lung cancer cell lines infected with the wild type MAX relative to the controls (?). Table S3. List of genes that are up-regulated or down-regulated upon depletion of BRG1. The values represent the n-fold change in the level of gene expression of each of the lung cancer cell lines infected with the shBRG1 relative to the controls ?. Table S4. List of the cell lines included in Figure 5. The information about alterations at the indicated genes was obtained from different sources, as indicated. For data extracted from databases we applied the following criteria to define a mutation: i) mutations at tumor suppressor genes (BRG1, SMARCB1, MAX, ARID1A, PRBM1, and MGA) should be homozygous and predictive of truncated proteins, ii) for amplification at the MYC family of oncogenes, only very high levels of gene amplification have been considered to be positive. Other genes, related to MYC/MAX or to the SWI/SNF complex, have also been searched for alterations (i.e., ARID1B, ARID2, MXI, MXDs) but either no alterations were reported in the databases or the changes did not fulfill our selection criteria. CCLE, Cancer Cell Line Encyclopedia (Broad-Novartis Cancer Cell Line Encyclopedia; website, http://www.broadinstitute.org/ccle/). COSMIC, Catalogue of Somatic Mutations in Cancer (Trust Sanger Institute's Cancer Cell Line Project; website, http://cancer.sanger.ac.uk/)

PDF file 422K, Fig. S1. Radio charts of the multiplex ligation-dependent probe amplification (MLPA), depicting intragenic deletions at the indicated lung cancer cell lines and lung primary tumor. The five exons of the MAX gene and the names and the relative location of the different probes used in the MLPA are also shown. The ratio charts depicting a the analysis of a normal DNA, an intragenic deletion of exons from 3 to 5 for the Lu134 cell line and deletion of the entire coding region for the Lu165 cell line are included. The radio chart for the tumorgraft from the same individual as the Lu134 cells is also depicted, ruling out the possibility of it being a cellculture artifact. Fig. S2. A, Cell proliferation, measured using MTT assays, was significantly greater in the cells carrying the empty vector (? control cells) than in those expressing ectopic MAX, from the parental H1417 cells. Lines represent the number of viable cells relative to the total number of cells at 0 h. Error bars, standard deviation *p<0.001. B, Western blot of total lysates of MAX in H1299tr-BRG1wt and H1299tr-BRG1mut cells, treated with (1 ng/μl) to allow BRG1 expression, depicts the ectopic expression of MAX in a hormone-free (HF) environment, or in the presence of glucocorticoids (GC) at the indicated concentrations. In the left panel, cells transfected with the 5'UTR-MAX constructs exhibit decreased levels of ectopic MAX in the BRG1mut cells compared with BRG1wt cells. On the right, cells transfected with the MAX construct lacking 5'UTR show similar levels of ectopic MAX regardless of the status of BRG1. C, Western blots illustrate the reduction of MAX upon depletion of BRG1 in the indicated lung cancer cells carrying amplification of MYC in an HF environment and at the indicated hours after treatment with GC (5 μM). TUBULIN is included as a loading control. Fig. S3. Enriched gene ontology (GO) classifications (p<0.05 for all categories shown) among genes up-regulated in the MAX gene-expression profile (from Table S2). Enrichment (represented by several GO categories) is significant for genes annotated as being involved in neural development and in glucose/carbohydrate metabolism

The incorporation of immunotherapy (IT) with immune-checkpoint inhibitors (ICIs) into clinical practice has represented a major breakthrough in non-small cell lung cancer (NSCLC) treatment, particularly in cases where the cancer has no druggable genetic alterations. The human histocompatibility complex (HLA-1) is essential for antigen presentation capability and immune response. Here we evaluate HLA-1 and other immune-related markers as potential predictive factors of response to ICI in NSCLC. We evaluated the immunophenotype in a cohort of 140 metastatic NSCLC patients who received treatment with ICI based regimens for metastatic setting at ICO Badalona from 2014 to 2019. We profiled the expression levels by immunohistochemistry (IHC) of HLA-1, and other immune-related markers including CD73, CD8, and PD-L1 (Ventana SP263) from formalin-fixed paraffin-embedded (FFPE) human tissue samples. We evaluated the response and clinical outcomes to ICI. The Chi-Square test for categorical variables and Kaplan Meier method for survival analysis were performed. In our cohort of 140 patients: 86% males and 14% females, 63% were lung adenocarcinomas (LuAD) and 37% squamous cell carcinoma (SCC). They received IT treatment as a 1st line (29%), 2nd (46%), and 3rd or further lines (25%). PD-L1≥50% was present in 25% of cases. 67 patients were evaluable for HLA-1 at the moment of the analysis. Our work reveals that ~45% of NSCLC in our cohort express low staining levels of HLA-1 (down regulation or total absence) compared to normal/high staining (55%). Those patients present worse clinical outcomes: mPFS to IT 9.1 (6.5–20.2) vs 21 (13.9-NR) months (p-value 0.028), respectively. We also report that HLA-1 is co-expressed with PD-L1 (p <0.005), regardless of histological subtype. Down-regulation of HLA-1 expression is a mechanism of immune-evasion and affects a subset of NSCLC, which abrogates the response to ICI. HLA-1 IHC is an emerging immunomarker in NSCLC and predictor of response to ICI. In addition, we observed that HLA-1 is co-expressed with PD-L1 and represents a surrogate marker of immune-inflamed phenotype which might predict better outcomes to PD(L)-1 blockade.

PDF file 48K, Description of the methodology for: -Lung tumor specimens and cancer cell lines. -Screening for MAX gene alterations: Sanger direct sequencing and Multiplex ligation dependent probe amplification (MLPA). -Expression vectors and lentiviral production. -Chromatin immunoprecipitation (ChIP) assays. -Antibodies and western blots. -Treatments and MTT assays. -Microarray global gene expression analysis. -Statistical and bioinformatic analysis

Supplementary Table 1. Characteristics of tumors and patients included in the analysis; Supplementary Table 2. List of antibodies and treatments used; Supplementary Table 3. Sequence of primers used for real-time quantitative PCR and size of amplified; Supplementary Table 4. Alterations at the indicated genes in lung cancer cell lines. cell carcinomas; SCLC, small Supplementary Table S5: RNA-sequencing report; Supplementary Table S6: List of genes from the MET-Activated (MET-Sign) and Interferon Treatment (IFN-

Supplementary Fig. 1. A, Western blot showing ectopic and transient expression of the LKB1 protein in the indicated cell lines; Supplementary Fig. 2. A, Profile of JAK2 mutations in NSCLC primary tumors, from the cBioPortal for Cancer Genomics (www.cbioportal.org); Supplementary Fig. 3. Western blot depicting the levels of the indicated proteins and cell lines, upon treatment with IFNγ, at a concentration of 30 ng/μL for 6 h. ACTIN and TUBULIN, protein-loading controls; Supplementary Fig. 4. Enriched gene ontology classifications (P<0.05) for the up-regulated genes in the MET-Sign and IFN-Sign.

Legends to supplementary figures from 1 to 4

Supplementary Figure S1. Cell viability and colony formation assays of the indicated cell lines. Supplementary Figure S2. Multidimensional scaling plots Supplementary Figure S3. Cell viability assays and western blots

Elucidating the adaptive mechanisms that prevent host immune response in cancer will help predict efficacy of anti-programmed death-1 (PD1)/L1 therapies. Here, we study the cell-intrinsic response of lung cancer (LC) to interferon-γ (IFNγ), a cytokine that promotes immunoresponse and modulates programmed death-ligand 1 (PD-L1) levels. We report complete refractoriness to IFNγ in a subset of LCs as a result of JAK2 or IFNGR1 inactivation. A submaximal response affects another subset that shows constitutive low levels of IFNγ-stimulated genes (IγSGs) coupled with decreased H3K27ac (histone 3 acetylation at lysine 27) deposition and promoter hypermethylation and reduced IFN regulatory factor 1 (IRF1) recruitment to the DNA on IFNγ stimulation. Most of these are neuroendocrine small cell LCs (SCLCs) with oncogenic MYC/MYCL1/MYCN. The oncogenic activation of MYC in SCLC cells downregulates JAK2 and impairs IγSGs stimulation by IFNγ. MYC amplification tends to associate with a worse response to anti-PD1/L1 therapies. Hence alterations affecting the JAK/STAT pathway and MYC activation prevent stimulation by IFNγ and may predict anti-PD1/L1 efficacy in LC.

Identifying molecular oncogenic drivers is crucial for precision oncology. Genetic rearrangements, including gene fusions and gene amplification, involving and activating receptor tyrosine kinases (RTKs) are recurrent in solid tumors, particularly in non-small cell lung cancer. Advances in the tools to detect these alterations have deepened our understanding of the underlying biology and tumor characteristics and have prompted the development of novel inhibitors targeting activated RTKs. Nowadays, druggable oncogenic rearrangements are found in around 15% of lung adenocarcinomas. However, taken separately, each of these alterations has a low prevalence, which poses a challenge to their diagnosis. The identification and characterization of novel targetable oncogenic rearrangements in lung cancer continue to expand, as shown by the recent discovery of the CLIP1-LTK fusion found in 0.4% of lung adenocarcinomas. While tyrosine kinase inhibitors that block the activity of RTKs have represented a breakthrough in the therapeutic landscape by improving the prognosis of this disease, prolonged treatment inevitably leads to the development of acquired resistance. Here, we review the oncogenic fusions and gene amplifications involving RTK in lung cancer. We address the genetic and molecular structure of oncogenic RTKs and the methods to diagnose them, emphasizing the role of next-generation sequencing technologies. Furthermore, we discuss the therapeutic implications of the different tyrosine kinase inhibitors, including the current clinical trials and the mechanisms responsible for acquired resistance. Finally, we provide an overview of the use of liquid biopsies to monitor the course of the disease.

Genetic inactivation in specific SWI/SNF complex members induce sensitivity to the inhibition of KDM6A/UTX and KDM6B/JMJD3 in cancer. Despite the genetic inactivation of the tumor suppressor SMARCA4 is frequently found in cancer, there are no therapies that effectively target SMARCA4-deficient tumors. In our previous work, we reported that unlike the cells with activated MYC oncogene, cells with SMARCA4 inactivation are refractory to the histone deacetylase inhibitor, SAHA, leading to the aberrant accumulation of H3K27me3. In this context, SMARCA4 deficient cells showed an impaired transactivation and significantly reduced levels of the histone H3K27me3 specific demethylases, KDM6A/UTX and KDM6B/JMJD3, and their inhibition compromises cell viability specifically in SMARCA4 mutant cells. Furthermore, the in vivo administration of KDM6s inhibitor (GSK-J4) to mice orthotopically implanted with SMARCA4 mutant lung cancer cell lines or primary tumors of small cell carcinoma of the ovary, hypercalcemic type (SCCOHT), showed a strong anti-tumor effect, highlighting the vulnerability of the SMARCA4 deficient tumors to KDM6s inhibition as a biomarker that could be exploited for treating SMARCA4-mutant cancer patients (Romero OA, Vilarrubi A, et al. SMARCA4 deficient tumors are vulnerable to KDM6A/UTX and KDM6B/JMJD3 blockade. Nat Comm 12, 4319, 2021). Considering our previous observations, and due to the biological role of SMARCA4 as a core component of the SWI/SNF chromatin-remodeling complex, whose members are genetically inactivated in approximately 20% of all human cancers, we wanted to determine the potential response to KDM6s inhibitors in a context of genetic alterations in genes encoding for the other subunits of this remodeling complex. For this propose, we have integrated state of the art technology like genome-wide chromatin modification analysis (ChIP-seq) and transcriptome analysis (RNA-seq), using human cancer cell lines and preclinical models of different cancer types with genetic inactivation at different SWI/SNF-complex members, including mouse models such as orthoxenografts, to design a personalized epigenetic treatment based on the genetic background. Our results showed that, like SMARCA4 deficient cells, specific mutations at some, but not all, SWI/SNF components induce refractoriness to SAHA, aberrant increase of H3K27me3 mark and sensitizes cancer cells to KDM6 inhibitor accompanied with significantly reduced levels of KDM6A and KDM6B expression. These results suggest a strong functional relationship between KDM6A/6B activity and the SWI/SNF-complex, in the control of gene expression and cancer development. These results will be of great value for the stratification of tumors according to their genetic background for tailored treatments, opening the possibility to use SWI/SNF mutations as potential biomarkers for personalized epigenetic-based therapeutics in cancer. Citation Format: Andrea Vilarrubi, Fernando Setien, Eva Pros, Pedro P. Medina, Antonio Gomez, Alberto Villanueva, Octavio A. Romero, Montse Sanchez-Cespedes. SWI/SNF inactivation vulnerability [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 3277.

Background There is no effective therapy for patients with malignant pleural mesothelioma (MPM) who progressed to platinum-based chemotherapy and immunotherapy. Methods We aimed to investigate the antitumor activity of CDK4/6 inhibitors using in vitro and in vivo preclinical models of MPM. Results Based on publicly available transcriptomic data of MPM, patients with CDK4 or CDK6 overexpression had shorter overall survival. Treatment with abemaciclib or palbociclib at 100 nM significantly decreased cell proliferation in all cell models evaluated. Both CDK4/6 inhibitors significantly induced G1 cell cycle arrest, thereby increasing cell senescence and increased the expression of interferon signalling pathway and tumour antigen presentation process in culture models of MPM. In vivo preclinical studies showed that palbociclib significantly reduced tumour growth and prolonged overall survival using distinct xenograft models of MPM implanted in athymic mice. Conclusions Treatment of MPM with CDK4/6 inhibitors decreased cell proliferation, mainly by promoting cell cycle arrest at G1 and by induction of cell senescence. Our preclinical studies provide evidence for evaluating CDK4/6 inhibitors in the clinic for the treatment of MPM.

Despite the genetic inactivation of SMARCA4, a core component of the SWI/SNF-complex commonly found in cancer, there are no therapies that effectively target SMARCA4-deficient tumours. Here, we show that, unlike the cells with activated MYC oncogene, cells with SMARCA4 inactivation are refractory to the histone deacetylase inhibitor, SAHA, leading to the aberrant accumulation of H3K27me3. SMARCA4-mutant cells also show an impaired transactivation and significantly reduced levels of the histone demethylases KDM6A/UTX and KDM6B/JMJD3, and a strong dependency on these histone demethylases, so that its inhibition compromises cell viability. Administering the KDM6 inhibitor GSK-J4 to mice orthotopically implanted with SMARCA4-mutant lung cancer cells or primary small cell carcinoma of the ovary, hypercalcaemic type (SCCOHT), had strong anti-tumour effects. In this work we highlight the vulnerability of KDM6 inhibitors as a characteristic that could be exploited for treating SMARCA4-mutant cancer patients.

Abstract Background The capability of tumors to avoid immune surveillance has emerged as therapeutically approachable, especially through the blockade of immune checkpoints (ICB) such as PD-L1/PD-1. We previously reported that B2M inactivating mutations in lung cancer (LC) impair immunorecognition through the disruption of the MHC-I complex. On the other hand, the refractoriness to IFNγ has also emerged as a mechanism that promotes tumor's immunoescape. Our purpose is to identify novel gene alterations that contribute to immunoescape in LC. Methods A panel of 44 NSCLC cell lines (35 commercially available and 9 ex vivo, derived from LC patients with metastatic pleural effusions, PE) were tested for response to IFNγ. Whole exome sequencing and RNA-sequencing was performed in the PE or gathered from public databases (Sanger and Cancer Cell Line Encyclopaedia) in the commercially available LC cells. The response to IFNγ was evaluated by determining the activation of various downstream targets, including an increase in the expression of CD274 (PD-L1) or IRF1 among others, using WB or RT-qPCR. We evaluated other parameters, such as the levels and location of the HLA-1/B2M proteins (HLA-I complex) by immunofluorescence, in selected LC cell lines. Results We could distinguish three different categories in the response to IFNγ; good responders, 75% (readily increase the levels of all downstream targets); non responders, 9% (did not increase the levels of any of the targets tested) or partial responders, 16% (were able to up-regulate all the targets tested but CD274/PD-L1, upon IFNγ exposure). By immunofluorescence, we observed that none of the non responders were able to increase HLA-1/B2M proteins in the cell surface, after IFNγ exposure. We identified biallelic inactivating alterations at JAK2 in three non responders (intragenic homozygous deletion, p.R426X and p.S507X in the NCI-H2126, NCI-H1993 and NCI-H1573, respectively). These mutations were concurrent with genetic alterations of main oncodrivers in LC. In the remaining non responder (PE-1), we identified a truncating mutation in a candidate gene. We overexpressed the wild type version of this gene and the response to IFNγ was restored. A mutational screening of this gene and immunostaining information for the encoded protein is currently ongoing in a cohort of 100 primary LC samples. Conclusions Alterations in molecules that are related with immunorecognition (B2M and HLA-I) are associated to immunotolerance, in LC. Here, we report that genetic alterations in factors that mediate the response to IFNγ are also involved in mediating immunotolerance in LC. Functional characterization of alterations in these genes is crucial to understand the mechanisms that contribute to tumor's immunoescape. Loss of function mutations in our candidate genes are likely to facilitate tumor growth by enabling immune tolerance and may affect the response to ICB. Citation Format: Juan Jose Alburquerque-Bejar, Maria Saigi, Eva Pros, Octavio Romero, Montse Sanchez-Cespedes. Searching for genetic alterations that drive the capability of evading tumor immunosurveillance in lung cancer [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 998.

Background: The etiology and the molecular basis of lung adenocarcinomas (LuADs) in nonsmokers are currently unknown. Furthermore, the scarcity of available primary cultures continues to hamper our biological understanding of non-smoking-related lung adenocarcinomas (NSK-LuADs). Patients and methods: Weestablished patient-derived cancer cell (PDC) cultures frommetastaticNSK-LuADs, including two pairs of matched EGFR-mutant PDCs before and after resistance to tyrosine kinase inhibitors (TKIs), and then performed whole-exome and RNA sequencing to delineate their genomic architecture. For validation, we analyzed independent cohorts of primary LuADs. Results: In addition to known non-smoker-associated alterations (e.g. RET, ALK, EGFR, and ERBB2), we discovered novel fusions and recurrentlymutated genes, including ATF7IP, a regulator of gene expression, that was inactivated in 5% of primary LuADcases. We also found germline mutations at dominant familiar-cancer genes, highlighting the importance of genetic predisposition in the origin of a subset of NSK-LuADs. Furthermore, there was an over-representation of inactivating alterations at RB1, mostly through complex intragenic rearrangements, in treatment-naive EGFR-mutant LuADs. Three EGFR-mutant and one EGFR-wild-type tumors acquired resistance to EGFR-TKIs and chemotherapy, respectively, and histology on re-biopsies revealed the development of small-cell lung cancer/squamous cell carcinoma (SCLC/LuSCC) transformation. These features were consistent with RB1 inactivation and acquired EGFR-T790M mutation or FGFR3-TACC3 fusion in EGFR-mutant tumors. Conclusions: We found recurrent alterations in LuADs that deserve further exploration. Our work also demonstrates that a subset of NSK-LuADs arises within cancer-predisposition syndromes. The preferential occurrence of RB1 inactivation, via complex rearrangements, found in EGFR-mutant tumors appears to favor SCLC/LuSCC transformation under growth inhibition pressures. Thus RB1 inactivation may predict the risk of LuAD transformation to a more aggressive type of lung cancer, and may need to be considered as a part of the clinical management of NSK-LuADs patients.

Gene fusions in lung adenocarcinoma (LuAD) involving tyrosine kinase receptors such as ROS1, ALK or RET are recurrent oncogenic drivers (∼10%), enriched in light or never-smokers. Some of them represent emerging and predictive biomarkers for targeted therapies. Here we report the fusions detected in a cohort of metastatic LuAD patients with low tobacco exposure (never or former-smokers). Patient-derived cancer cell lines (PDC) were successfully established from malignant pleural effusions from 11 patients diagnosed with LuAD. We assessed the genetic and molecular profile by whole-exome sequencing (WES) and RNA sequencing (RNA-seq) in each cell line. Patients' characteristics: median age, 58 (39-86); 9 were female. Eight of eleven were never-smokers and three, former-smokers. Seven patients were treatment naïve when pleural effusion samples were collected. A cytological examination of pleural fluid was performed by a lung pathologist and all samples were positive for malignant cells. Known driver mutations in lung primary tumours included one ALK translocation detected by FISH and three EGFR Del19 mutations by targeted sequencing. The three EGFR-mutant LuAD patients progressed to first or second-generation EGFR-TKI and we were able to stablish paired PDC after progressing to tyrosine kinase inhibitors (TKI) in two of them. We identified an acquired FGFR3-TACC3 fusion in one paired PDC after gefitinib progression (T790M-negative), that led to overexpression of FGFR3 concurrent with an enrichment of squamous cell lineage transcripts (e.g. TP63, SOX2) and MDM2 amplification. Among EGFR wild type (wt) patients, two RET rearrangements, CCDC6-RET and KIF5B-RET, and one EML4-ALK fusion -also detected in the primary tumour- were identified in PDC models. In addition, in two of the samples we discovered novel gene fusions that will be described in detail, involving proteins that are not kinases, and thus, their potential role in cancer is still unknown. In this cohort enriched with never-smoking LuAD patients presenting pleural effusions at diagnosis, the presence of known driver fusions during the disease's course detected by RNA-Seq was 36% (4/11), including a FGFR3-TACC3 fusion as an acquired resistance mechanism to EGFR-TKI. Further study is ongoing in our PDC models to test the functional role of these fusions in order to facilitate precision medicine.