Introduction
In all living organisms, as the third layer of epigenetics, more than 150 RNA modifications including 5-methylcytosine (m
5C), N6-methyladenosine (m
6A) and N1-methyladenosine (m
1A) have been identified [
1,
2]. Among these modifications, m
6A RNA methylation, which are widely found in the mRNA, lncRNA as well as miRNA, is recognized as the most prominent and abundant form of internal modifications in eukaryotic cells, of whose abundance account for 0.1–0.4% total adenosine residues [
3‐
5]. Similar to the modification of DNA and protein, m
6A modification is a kind of dynamic reversible process in mammalian cells, which is regulated by methyltransferases, demethylases and binding proteins, also known as “writers”, “erasers” and “readers” [
6]. The formation process of m
6A methylation is catalyzed by methyltransferases consisting of RBM15, ZC3H13, METTL3, METTL14, WTAP and KIAA1429, while the removal process is mediated by demethylases including FTO and ALKBH5. In addition, a group of specific RNA-binding proteins composed of YTHDF1/2/3, YTHDC1/2, HNRNPA2B1, LRPPRC, FMR1 and so on can recognize m
6A motif, thus affecting m
6A functions [
7,
8]. The in-depth understanding of these regulators would help reveal the role and mechanism of m
6A modification in post-transcriptional regulation. It has been reported that the m
6A regulators play a crucial role in a variety of biological functions in vivo [
9‐
11]. Increasing evidence demonstrated that dysregulated expression and genetic changes of m
6A regulators were correlated with the disorders of multiple biological process including dysregulate cell death and proliferation, developmental defects, tumor malignant progression, impaired self-renewal capacity, and immunomodulatory abnormality [
12‐
14].
Immunotherapy represented by immunological checkpoint blockade (ICB, PD-1/L1 and CTLA-4) has demonstrated astounding clinical efficacy in a small percentage of patients with durable responses. Unfortunately, the majority of patients experienced minimal or no clinical benefit, far from a met clinical need [
15]. Traditionally, the tumor progression has been considered as a multi-step process that only involves the genetic and epigenetic variation in tumor cells. However, numerous studies have shown that the microenvironment in which tumor cells depend for growth and survival also play a crucial role in the tumor progression. The tumor part was composed of a complex tumor microenvironment (TME) that not only contained cancer cells but also stromal cells such as resident fibroblasts (cancer associated fibroblast; CAF) and macrophages, and distant recruited cells such as infiltrating immune cells (myeloid cells and lymphocytes), bone marrow-derived cells (BMDCs) such as endothelial progenitor and hematopoietic progenitor cells, secreted factors such as cytokines, chemokines, growth factors, and new blood vessels. Of these, five distinct myeloid populations including tumor-associated macrophages (TAM), tumor-associated neutrophils (TANs), dendritic cells, myeloid-derived suppressor cells (MDSCs) and Tie2-expressing monocytes comprised the tumor-associated myeloid cells (TAMCs) [
16]. Cancers cells elicited multiple biological behavior changes through direct and indirect interactions with other TME components such as inducing proliferation and angiogenesis, inhibiting apoptosis, avoiding hypoxia as well as inducing immune tolerance. As the understanding of the diversity and complexity of tumor microenvironment has deepened, emerging evidence reveals its critical role in the tumor progression, immune escape, and its effect on response to immunotherapy. Predicting the response to ICB based on the characterization of TME cell infiltration is a key procedure on increasing the success of existing ICBs and exploiting novel immunotherapeutic strategies [
17,
18]. Therefore, by comprehensively parsing the TME landscape heterogeneity and complexity, different tumor immune phenotypes are likely to be identified, and the abilities of guiding and predicting immunotherapeutic responsiveness would also improve. Additionally, the promising biomarkers could be revealed, which will prove highly effective in recognizing patients’ response to immunotherapy and will benefit the search for new therapeutic targets [
19,
20].
Recently, several studies have revealed the special correlation between TME infiltrating immune cells and m
6A modification, which can’t be explained via RNA degradation mechanism. Dali et al. reported that binding of YTHDF1 to the transcripts encoding lysosomal proteases modified by m
6A methylation improved the translational efficiency of lysosomal cathepsins in dendritic cells (DCs), while suppression of cathepsins in DC significantly strengthened its ability to cross-present tumor antigens, which in turn enhanced tumor infiltrating CD8+ T cell antitumor response. And YTHDF1 inhibition also improved the therapeutic efficacy of anti-PD-L1 blockade [
21]. The study of Huamin et al. revealed that METTL3-mediated m
6A modification promoted the activation and maturation of DCs. Declining expression of co-stimulatory molecules CD80 and CD40 resulted by METTL3 specific depletion reduced the ability of stimulating T cell activation. And down-regulation of Tirap inhibited the transmission of the TLR4/NF-κB signaling pathway and decreased the secretion of pro-inflammatory cytokines [
22]. In addition, some studies have focused on the intrinsic oncogenic pathways induced by dysregulated expression and genomic variation of m
6A regulators. For example, Qiang et al. found that METTL3 overexpression promote gastric cancer (GC) malignant progression and liver metastasis through angiogenesis and glycolysis pathway [
23].
However, the above studies have necessarily been confined to only one or two m6A regulators and cell types owing to technical limitations, while the antitumor effect is characterized by numerous tumor suppressor factors that interact in a highly coordinated manner. Therefore, comprehensive recognizing the TME cell infiltration characterizations mediated by multiple m6A regulators will contribute to enhancing our understanding of TME immune regulation. In this study, we integrated the genomic information of 1938 gastric cancer samples to comprehensively evaluate the m6A modification patterns, and correlated the m6A modification pattern with the TME cell-infiltrating characteristics. We revealed three distinct m6A modification patterns, and surprisingly found that the TME characteristics under these three patterns were highly consistent with the immune-excluded phenotype, immune-inflamed phenotype and immune-desert phenotype, respectively, suggesting the m6A modification played a nonnegligible role in shaping individual tumor microenvironment characterizations. For that, we established a set of scoring system to quantify the m6A modification pattern in individual patients.
Discussion
Increasing evidence demonstrated that m6A modification took on an indispensable role in inflammation, innate immunity as well as antitumor effect through interaction with various m6A regulators. As most studies focus on single TME cell type or single regulator, the overall TME infiltration characterizations mediated by integrated roles of multiple m6A regulators are not comprehensively recognized. Identifying the role of distinct m6A modification patterns in the TME cell infiltration will contribute to enhancing our understanding of TME antitumor immune response, and guiding more effective immunotherapy strategies.
Here, based on 21 m
6A regulators, we revealed three distinct m
6A methylation modification patterns. These three patterns had significantly distinct TME cell infiltration characterization. Cluster A was characterized by the activation of innate immunity and stroma, corresponding to immune-excluded phenotype; cluster B was characterized by the activation of adaptive immunity, corresponding to immune-inflamed phenotype; cluster C was characterized by the suppression of immunity, corresponding to immune-desert phenotype. The immune-excluded and immune-desert phenotypes could be regarded as non-inflamed tumors. The immune-inflamed phenotype, known as hot tumor, show by a large number of immune cell infiltration in TME [
39,
43,
44]. Although the immune-excluded phenotype also showed the presence of abundant immune cells, the immune cells were retained in the stroma surrounding tumor cell nests rather than penetrate their parenchyma. The stroma could be confined to the tumor envelope or may penetrate the tumor itself, making the immune cells appear to be really inside the tumor [
45‐
47]. The immune-desert phenotypes were associated with immune tolerance and ignorance, and lack of activated and priming T-cell [
48]. Consistent with the above definitions, we found cluster A exhibited a significant stroma activation status, including the highly expressed angiogenesis, EMT and TGF-β pathways, which were considered T-cell suppressive. Combined with the TME cell-infiltrating characteristics in each cluster, it confirmed the reliability of our classification of immune phenotypes for different m
6A modification patterns. Therefore, after fully exploring the TME cell–infiltrating characterization induced by distinct m
6A modification patterns, it was not surprising that cluster A had the activated innate immunity but poorer prognosis.
Further, in this study, the mRNA transcriptome differences between distinct m
6A modification patterns have been proved to be significantly associated with m
6A and immune related biological pathways. These differentially expressed genes were considered as m
6A-related signature genes. Similar to the clustering results of the m
6A modification phenotypes, three genomic subtypes were identified based on m
6A signature genes, which were also significantly correlated with stromal and immune activation. This demonstrated again that the m
6A modification was of great significance in shaping different TME landscapes. Therefore, a comprehensive assessment of the m
6A modification patterns will enhance our understanding of TME cell-infiltrating characterization. Considering the individual heterogeneity of m
6A modification, it was urgently demanded to quantify the m
6A modification patterns of individual tumor. For that, we established a set of scoring system to evaluate the m
6A modification pattern of individual patients with gastric cancer—the m
6A gene signature. The m
6A modification pattern characterized by immune-excluded phenotype exhibited a higher m6Ascore, while the pattern characterized by immune-inflamed phenotype showed a lower m6Ascore. In addition, In IMvigor210 cohort with the determined immune phenotype, these results were well validated [
33]. This suggested m6Ascore was a reliable and robust tool for comprehensive assessment of individual tumor m
6A modification patterns, which could be used to further determine the TME infiltration patterns, that was, tumor immune phenotypes. Integrated analyses also demonstrated that m6Ascore was an independent prognostic biomarker in gastric cancer. Patients with EBV and MSI subtypes, sensitive to checkpoint immunotherapy [
42], was significantly related to lower m6Ascore. Considering the low mutation burden but high immune infiltration in EBVþ tumors [
41], our m6Ascore showed a predictive advantage in precision immunotherapy for gastric cancer.
Our data also revealed a markedly negative correlation between m6Ascore and tumor mutation burden. Consistent with previous studies, EMT and GS molecular subtypes demonstrated the lowest m6Ascore, underlining the core role of stromal activation in resistance to checkpoint immunotherapy [
49,
50]. This indicated that response to checkpoint immunotherapy was not only associated with antigen processing, and improved cytolytic activity, but also related to suppression of angiogenesis, fibroblast activation, TGF beta pathway components and the EMT. Previous studies confirmed that the EMT- and TGFbeta-related pathway activation resulted in decreased trafficking of T-cell into tumors as well as their weakened tumor killing effects [
33,
49]. The above suggested that the activated stromal TME in the activated immune TME could mediate therapeutic resistance to immune-checkpoint blockade, as well as influence the individual precise immunotherapy of gastric cancer. In this work, we showed m
6A methylation modification patterns played a nonnegligible role in shaping different stromal and immune TME landscape, implying m
6A modification could affect the therapeutic efficacy of immune checkpoint blockade. The m
6A gene signature with integrated various biomarkers including mutation load, neoantigen load, PD-L1 expression, stromal and immune TME and MSI status, could be the more effective predictive strategy for immunotherapy. We also confirmed the predictive value of the m6Ascore in two cohort with anti-PD-1 and anti-PD-L1 immunotherapy. A significantly difference on m6Ascores existed between non-responders and responders.
In short, in clinical practice, the m6Ascore could be used to comprehensively evaluate the m6A methylation modification patterns as well as their corresponding TME cell infiltration characterization within individual patient, further to determine the immune phenotypes of tumors and guide the more effective clinical practice. We also demonstrated the m6Ascore could be utilized for assessing patients’ clinicopathological features including stages of tumor inflammation, tumor differentiation levels, clinical stages, histological subtypes, molecular subtypes, genetic variation, MSI status, EBV infection and tumor mutation burden etc. The detailed relationships between m6Ascore and clinicopathological features could be found in our study. Similarly, m6Ascore could act as an independent prognostic biomarker for predicting patients’ survival. We could also predict the efficacy of adjuvant chemotherapy and the patients’ clinical response to anti-PD-1/PD-L1 immunotherapy through m6Ascore. More importantly, this study has yielded several novel insights for cancer immunotherapy that targeting m6A regulators or m6A phenotype-related genes for changing the m6A modification patterns, and further reversing the adverse TME cell infiltration characterization, that was the transformation of “cold tumors” into “hot tumors”, may contribute to exploiting the development of novel drug combination strategies or novel immunotherapeutic agents in the future. Our findings provided novel ideas for improving the patients’ clinical response to immunotherapy, identifying different tumor immune phenotypes and promoting personalized cancer immunotherapy in the future.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.