The n6-methyladenosine (m6A) is the most enriched internal epigenetic modification in eukaryotic mRNA.. In terms of molecular mechanisms, m6A is involved in nearly every step of RNA metabolism, including translation, degradation, splicing, export, folding of mRNA , and processing of mRNA and non-coding RNA (ncRNA) . We found alterations in m6A levels are involved in cancer pathogenesis and development by modulating the expression of tumor-associated genes. The effects of lncRNA in tumors is not uniform, with some acting as carcinogens and others as tumor suppressors, whose aberrant expression, mutations and SNPs are closely associated with tumorigenesis and metastasis [25, 26].
There has been increasing evidence that m6A modifications interact with lncRNAs as an important link affecting tumor development. For example, long-stranded non-coding RNA FAM225A drives nasopharyngeal carcinogenesis and metastasis via its role as a ceRNA for sponge miR-590-3p/miR-1275 and upregulation of ITGB3 . Wu et al. found m6A -inducible lncRNAs RP11 induces the propagation of colorectal cancer cells by upregulating Zeb1.Wang et al. believed lncRNA LINRIS protects IGF2BP2 by stabilizing it and promoting aerobic glycolysis in colorectal cancer . In addition, Ni et al. found the long-stranded non-coding RNA GAS5 interacts with YAP and triggers its phosphorylation and degradation to inhibit colorectal cancer progression, which is under negative regulation by the m6A reader YTHDF3. However, the overall characteristics of TME mediated by the interaction of m6A with lncRNAs and their impact on therapy and prognosis still lack a comprehensive understanding. Therefore, identifying the mode of action of different m6A-related lncRNAs in the tumor immune microenvironment can provide deeper understanding of the effect of m6A-lncRNA interactions in anti-tumor immune responses, as well as the tumor prognostic features of COAD mediated by m6A-related lncRNAs, which will help develop more effective and precise immunotherapeutic strategies.
We demonstrated the expression and prognostic value of m6A-associated prognostic lncRNAs in COAD, as well as their association with tumor immune microenvironment and ICI therapy. Subsequently, we built a prognostic module on the basis of m6A-associated prognostic lncRNAs, and screened sensitive drugs targeting high-risk genes.
The 43 m6A-associated prognostic lncRNAs we detected were all prognostic risk factors for COAD and differed significantly in both COAD tumors and normal paracancerous tissues. Among them, SNHG26, AC145285.2, AC026367.1, AC092944.1, GLIDR, AL512306.3, U91328.1, ATP2B1-AS1, PRKAR1B-AS2, SEPTIN7-DT, AC104819.3, NIFK-AS1, AC019205.1, LINC01588, LINC00861 had significantly lower expression in COAD tissues. The rest are highly expressed in COAD.
We also made a comparison of the expression differences of eight common immune checkpoints in COAD tumor tissues and paraneoplastic tissues and found that they were not consistently expressed. This suggests that the expression of these immune checkpoints differs between cancer and normal tissues, and also suggests that there are differences in immune checkpoint expression between individual COAD patients, and that clarifying these differences and using different immune checkpoint treatments may bring different benefits to patients, and continuing to explore the differences in immune checkpoint expression between different subgroups facilitates more effective individualized treatment.
The differences in tumor immune microenvironment (TIME) between subtypes of clusters A, B, and C were significant. clusters A and C with good prognosis both had significantly higher immune scores than cluster B. The emarkable difference in survival rates between clusters A and C may be correlated to the higher immunization scores in clusters A and C. However cluster A did not outperform the other two groups in assessing the IPS score for immunotherapy, which may be due to the complex TME effect.
Immune and stromal cells are affected by tumor cell and TME interactions and can contribute in part to tumor development. TME components and immune system biomarkers are of importance for cancer detection, prognostic assessment, and therapeutic response. The tumor-related mesenchyme provides nutrients, oxygen, enzymes, and stroma-bound growth factors, which promote tumor progression and proliferation[31–33].
Further analysis suggested that the infiltration of 23 immune cell types was generally higher in group A than in groups B and C. However, this also included some immunosuppressive cells such as Treg and MDSC, and some stroma-related expression was also higher in group A than in groups B and C. Recent studies have previously shown that immune cells prefer keeping in the stroma around the tumor cell nests than penetrating the parenchyma of the tumor cells, a phenotype also known as the immune rejection phenotype. In addition, multiple immunosuppressive immune cells are commonly found in TME of colorectal cancer. MDSCs are proven immunosuppressive cells, which are similar to queen bees, can promote the formation of Tregs . Besides, MDSCs often become TAMs and facilitate the differentiation of fibroblasts into CAFs, even shadowing the risk of death in tumor-ridden patients and increasing the risk of checkpoint inhibitor (CPI) resistance [38, 39]. Similar to MDSCs, the role of regulatory T cells (Tregs) in colorectal cancer has not been fully elucidated. Treg participates in effector t cell-mediated inflammation suppression through various mechanisms, including the release of transforming growth factor-β and interleukin-10 . The mean content of Tregs in the blood in colorectal cancer is higher than that of healthy volunteers. The average content of Tregs in tumor is higher than that in adjacent non-tumor. In addition, many studies have shown that Tregs in tumor can inhibit the proliferation of autologous CD4+ and CD8+ T cells , and the frequency of Tregs is negatively correlated with the expression of IFN-γ and IL-2 in tumor tissues . Hence, some studies suggested they are associated with poor prognosis.
Hence, we hypothesized that the higher immune score in cluster A, but not better IPS score and OS than the other two clusters, may be the result of the retention of immune cells in the stroma around the tumor cell nests and immune tolerance induced by immunosuppressive cells that do not fully exert immunocidal effects.
GSVA results suggest the presence of downregulation of cancer suppressor pathways including P53, and apoptosis in cluster B. P53 is an important tumor suppressor encoded by the oncogene TP53 and is involved in many vital biological procedures including cell cycle arrest, senescence, and apoptosis[45, 46]. P53 has also recently been shown to regulate cellular metabolism maintaining intracellular iron homeostasis and drive iron death to inhibit tumor development [47–49], while perturbative deletion or mutation of P53 regulates immune recognition and thus promotes an immunosuppressive environment through mechanisms such as reduced MHC-I presentation and increased suppressive myeloid cells and Treg . This may be an important reason for the lower OS in the cluster B group compared to the remaining two groups.
In addition to P53 pathway, some immune-related pathways are activated in cluster C but exhibit inactivation in cluster B, such as interferon-α responses.
The IFNα pathway has been shown to assist the signaling cascade by boosting the expression of IL21,IFN-γ, IL15[20, 50, 51]and other cytokines in immune cells to drive the maturation of DCs, which differentiated CD4+ T cells into Th1  and increased the activation and cytotoxicity of CD8+ T cells. IFNα has also been shown to mediate these immunomodulatory roles also in the absence of reliance on intermediate cytokine production. This is consistent with previous results that differences in interferon-alpha response may be responsible for differences in cluster A and C immune cell infiltration as well as prolonged OS.
Cluster A, however, exhibits paradoxical antitumor immunity, with both enrichment of antitumor immune cells and enrichment of Treg and MDSCs mediating immunosuppression, with both the highest immune and stromal scores.
Previous studies have shown that IL2 activation of STAT5 induces Foxp3 to sustain regulatory T cell development[53–55], while activation of EMT and transforming growth factor-β-related pathways impedes lymphocyte infiltration into the tumor parenchyma. This may account for Treg enrichment and mesenchymal activation in cluster A. Immune checkpoint blockade (ICB) combination therapy targeting these immunosuppressive mechanisms may improve the outcome of patients in cluster A and further prolong OS[57, 58].
We evaluated the prognostic value of m6A-related prognostic lncRNAs in COAD patients and derived 13 prognostic risk signals from them. The risk signals effectively classified COAD patients into two groups: high-risk and low-risk. In the TCGA training and validation cohort, patients in the high-risk group had significantly shorter OS than those in the low-risk group. Among these predictors, some have been reported, such as silencing of ATP2B1-AS1 that blocks NFKBIA-mediated NF-κB signaling pathway, LINC00861 as a protective factor in ovarian cancer, as a ceRNA for miR-513b-5p in cervical cancer, regulating PTEN/AKT/mTOR signaling pathway to inhibit cervical cancer cell progression, and Closely associated with PD1, PD-L1 and CTLA4 in prostate cancer[60–62]. AC003101.2 was predicted to play a ceRNA role in colorectal cancer, and AC156455.1 was reported to be a prognostic predictor associated with genomic instability in renal clear cell carcinoma. AC005229.4 was also reported to be an autophagy-related prognostic indicator in hepatocellular carcinoma, bladder cancer and endometrial cancer [65–67].
Moreover, linking the risk score and clinical attributes that we create a nomogram and examined it using data from TCGA, which has a great capacity to predict OS in COAD patients beneficially helping us to predict patient survival and clinical decision making.
Transforming growth factor β receptor II (TGFβRII), encoding β-catenin (CTNNB1), epidermal growth factor receptor (EGFR), And bcl-2-associated x protein (bax), which can encode microsatellite, often mutates in dMMR/MSI CRC. It has been shown that patients with dMMR/MSI early COAD have a better prognosis and survival compared to patients with pMMR/MSI mutations[18, 68], in addition, tumors possessing these features typically have a higher density of tumor-infiltrating lymphocytes (TIL) than MSS tumors, possess a stronger anti-tumor immune response[69, 70], and are more sensitive to ICIs sensitive to ICIs, which may be due to a large number of new tumor antigens produced by MSI CRC, which is caused by the mutation of synthetic truncated protein caused by unrepaired frameshift mutation . There is an overlap between MSI-H/dMMR and tumors with high TMB, but a large proportion of CRC with high TMB does not have defects in the MMR pathway, making TMB a more inclusive biomarker and thus greatly increasing the number of potential patients who may be identified as good candidates for ICI therapy. However, TMB assessments are more expensive, the standardization of TMB scoring needs to be further improved, and the applicability of the predicted values to MS/pMMR still needs to be evaluated.
We, therefore, continued to explore the TMB, MSI, and IPS characteristics of COAD patients in the risk score group, and our study found that TMB was higher in the low-risk score group, and risk scores were lower in COAD patients with MSI-H status, and that PD1 or PD1 combined with CTLA4 had a better effect in the IPS prediction in the low-risk score group. This is consistent with some previously reported results, which implies that these 13 lncRNAs could influence the efficacy of immunotherapy and could be used as a surrogate for improving patient stratification and assessing the effectiveness of immunotherapy. In addition, we valuated the top 20 most mutated genes in COAD and found that the mutation frequency was upper in the low-risk group overall, except for tumor suppressor genes such as TP53 and APC, which had a higher mutation frequency in the high-risk group, which was consistent with the previous GSVA analysis.
In addition, the combination of TMB with other biomarkers, such as immune checkpoints, may be more effective in forecasting the therapeutic efficacy of immunotherapy and provide guidance for clinical practice, so we proceeded to compare the differences in immune checkpoint expression between the high-risk scoring group and the low-risk scoring group. We observed that the expression of the immune checkpoint CD274 was higher in the low-risk scoring group than in the high-risk scoring group. The results based on IPS scores as well as TMB and the differential expression of CD274 also suggest that m6A-related lncRNAs may be able to guide individual patient treatment strategies and improve stratification of ICB therapy in COAD patients.
The tumor microenvironment plays an important regulatory role in tumorigenesis, which is regulated by lncRNAs during dynamic changes. This has been demonstrated in previous studies, for example, lncRNA MALAT1 was previously shown to adsorb miR195 to promote the development of diffuse large B-cell lymphoma and immune escape. In addition to this, lncRNA UCA1 overexpression protects PDL1 expression from miRNA inhibition and promotes immune escape in gastric adenocarcinoma cells [74-,75].
However, current tumor knowledge of m6A-associated prognostic lncRNAs in COAD remains limited. In the present study, the risk scores of 13 prognostic predictors constructed based on m6A-related prognostic lncRNAs were inconsistently associated with 23 immune cell infiltration. LINC00861, ATP2B1-AS1 were mainly positively associated with immune cell infiltration, but the other predictors were overall negatively correction with the infiltration of immune cell. This may explain the lack of difference in immune scores between the high-risk and low-risk score groups.
In addition we screened for sensitive drugs that can inhibit high-risk genes in prognostic models of m6A-related lncRNAs, such as quinostatin, mecamylamine, and piperidolate .The ability of quinostatin to inhibit the PI3K-MTOR pathway is probably significant for its anti-tumor role. And mecamylamine inhibits the α7nAChR/NF-ĸB p100/p52 pathway, promotes apoptosis and disrupts the anti-inflammatory effect in macrophages, perhaps also taking the corresponding effect in the treatment of COAD. Piperidolate is an anticholinergic agent, and previous studies have demonstrated that inhibition of cholinergic receptors can inhibit the proliferation of a variety of cancers, suggesting a potential role for piperidolate in oncology treatment.
This study was validated only in the TCGA dataset, which has limitations. Further and more independent COAD cohorts should be utilized to confirm the role of the detected m6A-related prognosis-related lncRNAs for prognostic stratification of COAD. Moreover, the role and mechanism of m6A-related lncRNAs for TME as well as COAD development still need to be proved in vitro and vivo. Our results may contribute to provide some usable clues for further experimental studies.