(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .



Genome-wide analyses reveal the contribution of somatic variants to the immune landscape of multiple cancer types [1]

['Wenjian Bi', 'Department Of Medical Genetics', 'School Of Basic Medical Sciences', 'Peking University', 'Beijing', 'People S Republic Of China', 'Center For Medical Genetics', 'Medicine Innovation Center For Fundamental Research On Major Immunology-Related Diseases', 'Zhiyu Xu', 'Regor Pharmaceuticals Inc.']

Date: 2024-02

It has been well established that cancer cells can evade immune surveillance by mutating themselves. Understanding genetic alterations in cancer cells that contribute to immune regulation could lead to better immunotherapy patient stratification and identification of novel immune-oncology (IO) targets. In this report, we describe our effort of genome-wide association analyses across 22 TCGA cancer types to explore the associations between genetic alterations in cancer cells and 74 immune traits. Results showed that the tumor microenvironment (TME) is shaped by different gene mutations in different cancer types. Out of the key genes that drive multiple immune traits, top hit KEAP1 in lung adenocarcinoma (LUAD) was selected for validation. It was found that KEAP1 mutations can explain more than 10% of the variance for multiple immune traits in LUAD. Using public scRNA-seq data, further analysis confirmed that KEAP1 mutations activate the NRF2 pathway and promote a suppressive TME. The activation of the NRF2 pathway is negatively correlated with lower T cell infiltration and higher T cell exhaustion. Meanwhile, several immune check point genes, such as CD274 (PD-L1), are highly expressed in NRF2-activated cancer cells. By integrating multiple RNA-seq data, a NRF2 gene signature was curated, which predicts anti-PD1 therapy response better than CD274 gene alone in a mixed cohort of different subtypes of non-small cell lung cancer (NSCLC) including LUAD, highlighting the important role of KEAP1-NRF2 axis in shaping the TME in NSCLC. Finally, a list of overexpressed ligands in NRF2 pathway activated cancer cells were identified and could potentially be targeted for TME remodeling in LUAD.

Recent studies have found that some genetic changes help cancer cells to evade the immune surveillance. To systematically understand the impact of cancer cell genetic alterations to immune regulation, we examined 74 immune traits across 22 cancer types. The tumor microenvironment (TME), crucial for cancer development, varies based on gene mutations in different cancers. Notably, the KEAP1 gene in lung adenocarcinoma (LUAD) emerged as a key player, explaining over 10% of immune trait variations. KEAP1 mutations activate the NRF2 pathway, creating a suppressive TME in LUAD with lower T cell infiltration and heightened T cell exhaustion. Additionally, genes such as CD274 (PD-L1), associated with immune checkpoints, exhibit high expression in NRF2-activated cancer cells. By developing a NRF2 gene signature, we found that it more effectively predicts anti-PD1 therapy responses than CD274 alone in non-small cell lung cancer. Lastly, we identified ligands overexpressed in NRF2-activated cancer cells, suggesting potential targets for reshaping the LUAD microenvironment. In essence, understanding these genetic interactions helps improve lung cancer treatment and enhance the efficacy of immunotherapy.

Competing interests: I have read the journal’s policy and the authors of this manuscript have the following competing interests: Zhiyu Xu., F.L., Zhi Xie, H.L., X.Z., W.Z., and X.T. are employees of Regor Pharmaceuticals Inc., Cambridge, Massachusetts, USA.

Funding: This research was supported by National Natural Science Foundation of China (62273010, W. B.). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Copyright: © 2024 Bi et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Comprehensive genomic analyses have identified somatic mutations and other alterations in the KEAP1 or NRF2 genes in various types of cancer, and Nrf2 mutations occur less frequently than Keap1 mutations. [ 19 ] Disruptions in the Keap1-Nrf2 pathway is frequently associated with poor prognosis and chemotherapeutic resistance in NSCLC. [ 20 ] Since the Keap1-Nrf2 pathway plays as primary regulator of key cellular processes that contribute to resistance against chemotherapy drugs, NRF2 has been studied as a potential therapeutic target molecule in NSCLC and some other cancers. [ 21 ] Our findings explore the molecular features and the impact of KEAP1-NRF2 to TME, which will be beneficial for novel treatment approaches in NSCLC in the near future.

To validate our findings and further characterize the underlying mechanism, we collected and analyzed three scRNA-seq datasets of LUAD and confirmed the immunosuppressive role of KEAP1 mutations. [ 13 – 15 ] We found a NRF2 gene signature activated in KEAP1-mutated samples. This signature was significantly correlated to lower T cell infiltration, higher macrophage and monocyte, higher Treg percentage, and higher T cell exhaustion, indicating an inflamed but immune suppressive TME subtype. Interactions of several ligand-receptor pairs between cancer cells and immune cells (CD274:PDCD1, FAM3C:PDCD1, PVR:TIGIT) are predicted to be enhanced in NRF2 signature high samples which could be potential targets to inhibit to remodel the TME. Furthermore, using clinical trial data [ 16 ], we found that non-small cell lung cancer (NSCLC) patients with an activated NRF2 signature are more likely to experience a durable response compared to other patients (p-value: 0.018). By contrast, CD274, a biomarker that has already been adopted for use in lung cancer clinical traits (NCT04294810) [ 17 , 18 ], failed to separate responsive patients from others (p-value: 0.51).

In this study, we analyzed 74 TME traits and 22 cancer types in The Cancer Genome Atlas (TCGA), using genome-wide gene-level association analyses, to identify genes in which somatic variants significantly alter TME traits. Totally, 451 significant gene-trait associations were reported across different cancer types. Among these associations, 14 genes were found to regulate 3 or more TME traits, which suggests their important roles in shaping multiple aspects of the TME. Of these, KEAP1 was identified as the top hit in lung adenocarcinoma (LUAD) and explained a large proportion (>10%) of variance for multiple immune traits, including interferon pathway, MHC class II expression, and the NK cell signature score. Other important gene-trait associations included TP53 mutations associated with B cell receptor (BCR) function in breast cancer (BRCA), PBRM1 mutations associated with neutrophil in kidney renal clear cell carcinoma, IDH1 mutations associated with lower lymph vessel signature in Brain Lower Grade Glioma (LGG) and BRAF mutation associated with NK cell and macrophages in Thyroid Carcinoma (THCA).

Recently, the regulation of TME related to somatic alternations in intrinsic pathways in cancer cell has received increasing attention. [ 3 , 9 ] A large number of studies have shown that previously known oncogenes or tumor suppressors regulate the cancer TME by altering the activities of cancer intrinsic pathways. [ 10 – 12 ] These recurrently mutated genes in cancer cells activate or inhibit various chemokines and cytokines, resulting in different TME subtypes. Characterizing the associations between these cancer cell genetic alterations and TME traits can help to better stratify patients by TME subtypes, which could be crucial for IO therapy development.

In the past, the study of cancer TME was largely restricted by the technology available for TME traits retrieval. Only a small number of TME traits can be derived from expensive and laborious experiments, such as flow cytometry [ 4 ] and immunohistochemistry [ 5 ]. Nowadays, with the advancements in omics technology and bioinformatics tools, various TME traits can be derived from RNA-seq data through de-convolution methods or gene signature enrichment analysis. [ 6 , 7 ] These bulk RNA-seq-derived traits have been shown to be highly consistent with immune traits obtained using cell flow cytometry or scRNA-seq. In a recent study by Sayaman et al., 139 TME traits were collected from multiple studies, mostly from bulk RNA-seq data. [ 8 ] Their results showed that germline variants account for no more than 20% of the variation in TME traits, which leaves a significant proportion of variance unexplained. [ 8 ].

Over the last decades, the discovery of immune checkpoints and their applications in cancer therapy have revolutionized the treatment of various cancer types. [ 1 ] Immune checkpoint inhibition (ICI) therapies have been utilized as single agents or in combination with chemotherapies to treat over 50 types of cancer. Despite of these tremendous success, however, only a limited percentage of patients have achieved long-lasting benefits. [ 2 , 3 ] The ineffectiveness of immune-oncology (IO) therapies could be at least partially attributed to the imprecise selection of patients resulted from limited understanding of tumor microenvironment (TME).

Results

Overview of genetic test to associate somatic variants with TME traits We conducted genome-wide gene-level association analyses to identify genes in which somatic variants alter TME traits significantly. Of the TME traits Sayaman et al. analyzed [8], we selected 74 traits (S1 Table), most of which were derived from bulk RNA-seq data of TCGA tumor samples by scoring different gene signatures using ssGSEA or by deconvoluting bulk RNA-seq using CIBERSORT. [8,22] These immune traits were selected to represent proportion of different immune cell types, activity of immune pathways, and states of different immune cells. To ensure data processing workflow consistency across different cancer types, we used somatic mutations, log2 copy number alterations, and clinical data from TCGA panCancer project. [23] In total, 22 cancer types with > 100 samples were selected for analysis (S2 Table). Somatic mutations in non-coding regions (UTR or intron) except splicing changing variants were excluded from analysis. For each cancer, genes mutated in ≥ 5 tumor samples were fed into association test pipeline. Immune traits were transformed to either quantitative or binary values depending on their distributions, and then passed to linear or logistic regressions (see Methods). In addition to the confounding covariates adjusted by Sayaman et al. (including gender, days to birth, and age at initial pathologic diagnosis), radioactive therapy status and chemotherapy status were also incorporated in the regression model [8]. Both somatic mutations and copy number alterations (CNA) can impact cancer via TME regulation. However, CNA usually spans multiple gene regions and thus it is challenging to distinguish a driver gene from other passengers in the same CNA region without prior knowledge or additional experiments. Hence, we mainly focused on associations between mutations and TME traits, and included CNA as an additional confounder. Although both quantitative log2CNA value and categorized CNA value are available in cbioportal [24], we incorporated log2CNA as a covariate since it has higher correlation with gene expression. The log2CNA values were adjusted by tumor purity prior to analysis (see Methods). In the germline variants association study, Sayaman et al. combined multiple cancer types together for analysis. [8] In this study, we analyzed 22 cancer types separately due to a significant diversity of mutations in cancer cells. A total of 1,628 (74×22) genome-wide analyses were conducted for 22 cancers and 74 TME traits (Fig 1). To avoid a large number of spurious positives, analyses p-values were adjusted using genome control lambda (gc lambda, see Methods). For each genome-wide analysis, we calculated false discovery rate (FDR) adjusted p-values as q-values and selected genes whose q-values < 0.05 as top genes that are significantly associated with TME traits. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 1. Overview of genetic association test to identify immune regulators in different cancer types using TCGA data. Somatic mutations and copy number variations from 22 cancer types are downloaded from cbioportal to associate with 74 immune traits collected from Sayaman et al. (2021) [8]. https://doi.org/10.1371/journal.pgen.1011134.g001

Contribution of somatic variants to TME traits varies across cancer types We identified 451 significant gene-trait associations with q-values < 0.05 across 22 cancer types (S3 Table). For cancer types of brain Lower Grade Glioma (LGG), Thyroid Carcinoma (THCA), and Breast Cancer (BRCA), 99, 81, and 62 significant gene-trait associations were identified, respectively. Fig 2A demonstrated the 14 genes significantly associated with 3 or more TME traits, of which each of the 13 genes except TP53 was identified in only one corresponding cancer type, suggesting the importance of tissue context on TME regulation. Based on the identified gene-trait associations, we calculated the proportion of trait variance that is explained by somatic variants, including both somatic mutations and CNA, for each cancer type (see Methods). Totally, 32 TME traits from 9 cancer types have >20% variance explained by somatic variants, suggesting a non-negligible contribution of somatic variants to TME (Fig 2B and S4 Table). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 2. Different TME characteristics of different cancers are shaped by differentially mutated genes. (a). Top genes that are significantly associated with > = 3 TME traits. Of the 451 significant gene-trait associations, 14 genes were highlighted. The y-axis is the gene name and the x-axis is the number of the traits that are significantly associatedwith the gene. Different colors are for different cancer types. (b). Immune traits with > 20% explained variance by somatic mutations. The y-axis is the name of TME trait and the x-axis is the proportion of the variance explained by somatic mutation. Different colors are for different cancer types. https://doi.org/10.1371/journal.pgen.1011134.g002

TP53 regulates TME traits for multiple cancer types As aforementioned, TP53 is the only gene associated with multiple TME traits (> = 3) in multiple cancer types (Fig 2A). In BRCA, TP53 mutations are correlated to higher tumor-infiltrating lymphocyte (TIL) fraction, higher interferon gamma, higher macrophage, and higher B cell receptor (BCR) richness and diversity (S5 Table), which is consistent with previous results that TP53 mutations may promote immunogenic activity in BRCA. [25] On the contrary, in head and neck squamous cell carcinoma (HNSC), TP53 mutations are associated with inhibitory immune features such as lower lymphocyte infiltration signature score and lower CD8 cell signature score, which was also reported in previous studies. [26] The example of TP53 highlights the importance of tissue context for TME regulation, emphasizing the need of cancer-type-specific TME stratification for targeted immunotherapy.

Kidney Renal Clear cell carcinoma (KIRC) In KIRC, we identified neutrophils signature, for which, 28% trait variance are explained by somatic variants (Fig 2B). Neutrophils were known as the first line of defense against microbial infection. They circulate in the blood and are recruited rapidly to the site of tissue injury. Recent studies showed that neutrophils have pro-tumoral or anti-tumoral functions under different contexts. For KIRC specifically, previous studies suggested that tumor-infiltrating neutrophils act as an independent adverse prognostic feature. [27] Higher tumor-infiltrating neutrophils (TINs) were significantly associated with worse overall survival and higher metastasis rate.[27] Of the genes constituting the trait of neutrophils signature in KIRC, we found PBRM1 contributes mostly to the trait variance (~22% variance explained, S5 Table) and mutations of PBRM1 are associated with higher neutrophils (p value: 1.95×10−9). This perhaps is not surprising as PBRM1 encodes a protein that is involved in the regulation of chromatin remodeling and inflammation-related genes are highly regulated by chromatin remodeling genes in KIRC. [28] PBRM1 is highly mutated (~30%) in KIRC samples in which more than 85% of the genetic alterations lead to loss of function (deletion, truncating mutation, and splice mutation). PBRM1 mutation cause activation of inflammation-related genes, which can trigger neutrophil-dependent lung metastasis in advanced KIRC. [28] Generally speaking, experiment and analyses in both mice and human validate that PBRM1 loss of function defines a nonimmunogenic tumor phenotype associated with checkpoint inhibitor resistance in renal carcinoma. [28,29]

Brain Lower Grade Glioma (LGG) In LGG, we identified lymph vessels, for which, 38% trait variance are explained by somatic variants. Of the genes significantly associated with the lymph vessels trait, IDH1 gene contributed mostly to the trait variance (33% variance explained). IDH1 is highly mutated in LGG tumor samples (77%) in which almost half of the mutations are annotated as truncating mutations. IDH1 mutations are associated with lower lymph vessel signature (p-value: 7.97×10−40). IDH1 is reported to regulate podoplanin (PDPN) expression in glioma by regulating its promoter methylation status. [30] And PDPN is strongly expressed in higher-grade IDH1-wild-type glioma but almost undetectable in IDH1-mutated glioma. Consistent with our analysis that IDH1 loss of function mutations is associated with lower lymph vessel signature, upregulation of PDPN induces lymphangiogenesis and metastasis in tumor. [31]

Thyroid Carcinoma (THCA) For THCA, somatic variants account for 42.8% variance of NK cells, 28.8% variance of TGFB score, 28.6% variance of macrophages, and 22.9% variance of dendritic cells (DC). They are mainly driven by BRAF gene mutations which account for 17.8% - 39.5% of variance for the four traits. In THCA, 56% of patients are carrying the same mutation BRAF V600E mutation which is almost the only mutation found in BRAF. Based on our analysis, BRAF V600E mutation is significantly associated with high macrophages (p-value: 2.42e-15). In addition, patients carrying the BRAF mutation are having lower CD8 T cells % (p-value: 4.00e-7) and higher Treg cell % (p-value: 6.64e-12), which is consistent with another study of human thyroid cancer.

Breast Cancer (BRCA) Besides TP53, another major influencer to TME traits in BRCA is CDH1. Most of CDH1 mutations in TCGA-BRCA cohort are loss of function mutations, of which > 80% are from LumA subtype. Deficiency of CDH1 protein is associated with higher NK cell signature score and lower macrophage cell population. This is in line with its function of encoding a ligand for interacting with killer cell lectin-like receptor G1 (KLRG1) on NK cells and memory T cells to trigger inhibitory signals. [32] It is interesting that CDH1 loss of function (LOF) mutations are highly frequent in LumA breast cancer, suggesting that high expression of LOF CDH1 mutants in LumA cancer indicates worse prognosis. Furthermore, deficiency of CDH1 maybe oncogenic for cancer initiation and over expression of CDH1 mutants could advance cancer progression by inhibiting NK cells and T cells. Several studies have shown synergistic effect by blocking KLRG1 and PD-1 together in mouse models for multiple cancer types. [33, 34] Our analysis supports the notion that CDH1-high BRCA patients can be potentially treated by a combination therapy of KLRG1 inhibitor and PD-1/PD-L1 inhibitor.

Lung Adenocarcinoma (LUAD) In LUAD, KEAP1 stands out as the top hit associated with multiple TME traits, explaining 5% to 15% trait variance (Fig 3A). Boxplot of trait value distribution shows that many immune traits are downregulated in KEAP1 mutated patients, such as TGFβ signaling, different T cell subtypes gene signatures, NK cell gene signature, MHC expression, macrophage gene signature, and interferon pathways (Fig 3B). Per our interest of LUAD, we used scRNA-seq data to further confirm these associations and characterize the molecular mechanisms behind the associations. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 3. KEAP1 is the gene shaping multiple TMEs in LUAD. (a). Immune traits that are associated with KEAP1 mutations in LUAD. The y-axis is -log10(p values) in which the p value correspond to the tests associating somatic mutation in KEAP1 with TME traits. The x-axis is the proportion of trait variance been explained by KEAP1 mutations. (b). Immune traits value distribution across KEAP1 mutated group (red) and KEAP1 wild type group (blue) in LUAD. The x-axis is for different TME traits and the y-axis are for TME traits after transformation. The p-values in the x-axis are calculated using two-sample students t-test. https://doi.org/10.1371/journal.pgen.1011134.g003

[END]
---
[1] Url: https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1011134

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.