In this study, a series of complex bioinformatics analyses were carried out, which identified three lncRNAs associated with HCC prognosis, including AC015908.3, AC068987.4 and AL365203.2. The immune-related risk model constructed based on these three lncRNAs was able to distinguish low-risk from high-risk samples with a relatively high prognosis accuracy. For the training set, HCC patients in low-risk group had longer OS than those in high-risk group, and similar results were also obtained in the testing set and the entire set. Moreover, when patients were stratified based on age (≤60/>60), gender (female/male), clinical stage (I and II/III and IV), grade (G1 and G2/G3 and G4) and T stage (T1 and T2/T3 and T4), the as-constructed prognosis model also displayed the potentials to predict the differential prognosis between high and low risk groups. Further PCA confirmed that our prognosis signature had grouping capacity. Therefore, the identified immune-lncRNA signature might be implicated in HCC initiation and progression, which rendered them with the potentials as the valuable clinical biomarkers. In addition, the gene expression profiles of high risk HCC cases selected by our signature were enriched in various immune-related gene sets, and subsequent analysis further confirmed the increased infiltration of B cells, CD4+T cells, CD8+T cells, macrophages, neutrophils, and dendritic cells in TME of high risk HCC patients, especially for the latter three immune cell types.
To the best of our knowledge, no existing study has reported the direct correlation between AC015908.3, AC068987.4 or AL365203.2 expression and HCC prognosis, and the functions of these three lncRNAs in HCC remain unclear so far. Moreover, the detailed studies with regard to these three novel lncRNAs are lacking at the moment. AC068987.4 and AL365203.2 were found to be the negative prognostic genes, while AC015908.3 was the positive prognostic gene. Accordingly, AC068987.4 and AL365203.2 expression in high-risk groups among the three sets was remarkably higher than that in low-risk groups. Compared with normal tissues, the expression levels of the above-mentioned two adverse prognostic factors in TCGA HCC samples were also remarkably up-regulated. There are few reports on the immune-related functions of these three genes, and the identified lncRNAs may regulate immune function either directly or indirectly. Among the above-mentioned immune-related gene sets, GSE39110 displayed an immunization regimen, which delivered the T cell receptor (TCR) signals through a defined antigenic peptide, transmitted the inflammatory signals through lipopolysaccharide (LPS), and propagated the growth and differentiation signals through IL-2R that initially favored Ag-specific CD8+ T cells, for the sake of rapid and substantial development of TCR transgenic OVA-specific OT-I CD8+T cells into T effector-memory cells [28]. Compared with low-risk samples, the high-risk samples were enriched with genes that were down-regulated in CD8+T cells at 3 days after immunization, which might suggest that high-risk patients had poor ability to transform Ag-specific CD8+ T cells into T effector-memory cells. Of interest, previous studies have demonstrated that, the tumor‐infiltrating and interleukin‐33–producing effector‐memory CD8+ T cells in the resected HCC prolong patient survival[29], thus reflecting a protective effect. In another gene set, GSE14308, which contained genes that were up-regulated in Th2 cells compared with in Th17 cells, was also closely relate to the high-risk groups. The markedly increased levels of anti-inflammatory cytokines, such as IL-4, IL-5, IL-8, and IL-10, which were secreted by Th2 cells, were also reported to be unique to liver tissues in metastatic HCC patients[30], and such results might reflect a more malignant state in high risk patients. On the other hand, genes that were up-regulated in CD25+ T cells compared with CD25- T cells were also suggested to be associated with the high-risk groups. For instance, Lee et al illustrated that the increased CD4+ CD25+ T cells, which appeared to suppress the dendritic cells-activated immune response, was correlated with HCC tumor size in TME[31].
Moreover, the relationships of the signature with immune cell infiltration were deliberated to reflect the immune microenvironment status of HCC. Interestingly, our analysis indicated that the immune-related signature constructed in this study was positively correlated with the infiltration of six immune cell types, especially for macrophages, neutrophils and dendritic cells (DCs), suggesting that the higher infiltration levels of these immune cells might be observed in high-risk patients. Consistently, recent studies have expounded that the high levels of tumor-infiltrating macrophages and neutrophils predict dismal prognosis for primary HCC patients. Typically, the intramural neutrophil infiltration is elaborated to be promoted by the CXCL5 and CXCR2-CXCL1 axis; besides, it displays marked correlation with shorter OS and HCC recurrence, and serves as an independent prognostic factor[32-34]. Moreover, the increased infiltration of tumor-associated macrophages dominated by the M2 macrophages, which may be ascribed to the deletion of the Hippo signaling, has been reported to produce the Wnt/β-catenin signaling and trigger the elevated intramural FoxP3+ Treg population, while this in turn accelerates HCC progression [35-37]. In addition, accumulating evidence reveals a role of DCs as an adverse prognostic factor for HCC. For instance, Zhou et al. set forth that, the intratumoral infiltration by plasmacytoid DCs was a novel indicator of the poor prognosis for HCC patients, which might be achieved through inducing the immune tolerogenic and inflammatory TME comprising regulatory T and IL-17-producing cells[38]. Such results have underscored the importance of tumor-associated DC cells in predicting the prognosis for HCC patients. Taken together, the as-constructed model not only predicted the prognosis for HCC patients, but also suggested the different immune conditions of patients in high- and low-risk groups. The immune-related signature was correlated with survival, which might serve as the biomarkers to assess the feasibility of various immunotherapies. Nevertheless, to investigate the detailed relationship between immune-related lncRNA expression and the actual immune phenotype, future studies should make efforts to development more novel immune-related signatures in RNA sequencing datasets and further validate them experimentally. The three lncRNAs identified in this study provided clues for discovering the particular immunotherapies for HCC patients.
It is known that lncRNAs participate in various biological processes, such as transcription, translation, cell differentiation, chromatin modification, as well as regulation of gene expression, cell cycle and nuclear-cytoplasmic trafficking[39, 40]. To further explore the broader functions of these three lncRNAs, co-expression and functional enrichment analyses were performed subsequently. After co-expression analysis on those three ncRNAs, mRNAs with the Pearson correlation of >0.4 were selected for further enrichment analysis. The results demonstrated that these three lncRNA-related protein-coding genes were mainly involved in cadherin binding, helicase activity, and ATPase activity, which thus indirectly revealed the potential functions of the signature-lncRNAs from another perspective. Numerous works have elaborated that, the reduced expression of E-cadherin as well as catenin complex, such as α-, β-, γ-catenin, occurs frequently in HCC, which contributes to tumor progression and development [41, 42]. On the other hand, several genes that are involved in ATPase activity or helicase activity are also illustrated to have a bearing on HCC progression. RuvB‐like 2 (RUVBL2), an ATPase and putative DNA helicase known to interact with β‐catenin and cellular v‐myc myelocytomatosis viral oncogene homolog (c‐myc), is confirmed to be over-expressed in most HCCs, which is associated with the enhanced tumorigenicity [43]. Further, Grigoletto et al. reported that the ATPase activity of Reptin, which was linked to the Walker A and B domains, was required for its effects on tumor cell growth and viability in HCC[44]. Based on the KEGG pathway enrichment analysis, the spliceosome pathway was revealed to be the most enriched one, and most of its genes were demonstrated to be up-regulated in HCC[45]. The spliceosome is a ribonucleoprotein complex to control the (alternative) splicing, which participates in cell cycle, invasion, metastasis and angiogenesis [46]. Previous studies indicate that, targeting spliceosome reduces cancer cell proliferation [14], and triggers the mTOR blockade and autophagy[47]. Our founding suggested that, the signature-lncRNAs might also promote HCC cell metastasis and proliferation through the spliceosome pathway. However, further study is warranted to characterize the tumor-specific role of lncRNAs. Moreover, mRNAs showing the highest Pearson correlation coefficient with signature-lncRNAs were also examined. Specifically, TMEM220, which showed the highest positive correlation coefficient with AC015908.3 according to our co-expression analysis (Cor: 0.662, P < 0.001), was proved to be remarkably down-regulated in TCGA HCC tissues compared with normal liver tissues, and it was deemed as the favorable prognostic factor. ARID3A, a nuclear matrix-associated transcription factor that stimulates the expression of immunoglobulin heavy chain (IgH) expression as well as the progression of the Cyclin E1/E2F-dependent cell cycle, displays marked correlation with AC068987.4 (Cor: 0.636, P < 0.001), as mentioned in a transcriptional regulatory network in the context of HCC[48, 49]. In our study, ARID3A expression in HCC samples was markedly up-regulated, but its effect on prognosis should be further investigated. Additionally, SLC38A1, a crucial glutamine transporter, plays an essential role in nutrient uptake, energy production, chemical metabolism, detoxification, and neurotransmitter cycling[50]. As the mRNA that had the highest positive correlation with AL365203.2 in our study (Cor: 0.65, P < 0.001), SLC38A1 was demonstrated to be activated in HCC tissues[51], which was linked with the worse OS for HCC patients. Taken together, the potential functions mentioned in this study have enriched the scope of action of the novel signature-lncRNAs and assisted in understanding these three genes at a deeper level.
Nonetheless, some limitations should be noted in this study, which should be taken into consideration when interpreting our results. Firstly, the transcriptomic analysis only reflected some aspects of immune status, rather than the global map. Secondly, the immunocyte-specific gene sets applied in this study were limited to 6 major immune cell types, as a result, differences in the more specialized immune cell subtypes (like the differently polarized macrophages or myeloid-derived suppressor cells) might not be recognized in this study, even though they were known to be mechanistically linked to HCC progression and stage[16, 52]. Finally, our results were not validated via another independent cohort, which was also a limitation of this study, and the reliability of our molecular results was still challenged by the lack of experiments in vitro or in vivo.