LncRNAs are involved in multiple physiological and pathological processes, the underlying mechanisms remain largely unknown [28–30]. In this study, data showed that ELFN1-AS1 was significantly up-regulated in CRC tissues and promoted immune escape of CRC cells from NK cells. GDF15 is secreted by CRC cells and was one of the key mediators of NK cell activity. We also determined that ELFN1-AS1 regulated the expression of GDF15 through the SND1-GCN5/GDF15 axis. These results indicated that ELFN1-AS1 plays an oncogenic role in CRC progression.
Escape from immune surveillance is a pivotal feature of tumors with distant metastasis [31–33] and activation of immune surveillance is an important strategy for tumor targeted therapy. NK cells play important roles in immune surveillance and can directly kill tumor cells via perforin and granzyme release [34, 35]. There is increasing evidence that the numbers of infiltrating NK cells in tumor tissues are positively correlated with tumor patient survival [36–39]. Consistent with previous research, from TCGA data analysis we found that activated NK cell levels are decreased and resting NK cell levels are increased in CRC tissues, suggesting an altered proportion of NK cell subsets. In addition, a high proportion of resting NK cells also significantly correlated with the poor survival rate of CRC patients. These data indicated that the alteration of NK cell immunity induced by CRC tumor microenvironment may be a major mechanism for tumors to escape immune killing.
Recent studies reported that lncRNAs could regulate immune surveillance [40, 41]. Here we used gene microarrays and identified a specific lncRNA ELFN1-AS1 that was up-regulated in both CRC tissues and cells. ELFN1-AS1 has been linked to the development of multiple tumors such as esophageal [18] and ovarian [17] cancers. In colorectal cancer, ELFN1-AS1 expression was increased and promoted the proliferation and metastasis of tumor cells [21]. MYC-regulated ELFN1-AS may function in cell proliferation and the cell cycle by regulating MYC target genes [42]. Tumor immunity studies also demonstrated that some lncRNAs induce immune cell dysfunction within the tumor microenvironment [43]. Notably, in this study, we found that ELFN1-AS1 tended to be negatively associated with the surface marker CD56 on NK cells in COAD and READ, which suggested the upregulated ELFN1-AS1 may contribute to NK cell suppression. Both in vivo and in vitro, the NK cell cytotoxicity was impaired after co-culture with high level ELFN1-AS1-expressing CRC cells, implying ELFN1-AS1 could promote the immune escape of CRC cells from NK cells. Moreover, we found that the NKG2D and GZMB receptors on NK cells were significantly downregulated and the JNK signaling in NK cells was inhibited after co-culture with high level ELFN1-AS1-expressing CRC cells. JNK signaling is involved in the development and differentiation of immune cells [44]. NKG2D in NK cells can be activated by JNK signaling [45] and elevated NKG2D in turn induces activation of JNK kinase [46] and gradually activates JNK signaling pathways [47]. Inhibition of JNK MAP kinase also blocks granzyme B movement to the immune synapse [48] and the JNK pathway controls expression of CCL5 that is co-released with granzymes in NK cells [49]. Collectively, our results and previous reports suggested that ELFN1-AS1 in CRC cells might directly affect JNK signaling in NK cells to suppress the surface expression of NKG2D and GZMB resulting in a marked deficiency in tumor cytotoxicity.
Previous studies have demonstrated that tumor cells secrete NK cell inhibitory factors such as TGF-β1 and cytokines [50, 51]. In our study, we verified that the level of GDF15 (a secretory ligand of the TGF-β superfamily) was regulated by ELFN1-AS1 in CRC cells. In cervical cancer cells, GDF15 directly promoted cell proliferation and significantly increased cell cycle progression [52]. It is associated with human NK cell dysfunction that leads to the immune escape of cancers [53, 54], as well as TGF-β [55]. In addition, GDF15 is a MYC target and a positive feedback of GDF15/MYC/GDF15 was also verified [56]. Combined with the role of ELFN1-AS1 in MYC-regulated cell phenotypes, we considered that GDF15 was the secretory protein induced by ELFN1-AS1 from CRC cells. Our data also demonstrated that anti-GDF15 antibody could reverse the inhibition of NK cells induced by high ELFN1-AS1 expressed CRC cells via restoring the activity of NKG2D and GZMB in NK cells. This suggested that GDF15 production was an important mechanism used by ELFN1-AS1 to modulate CRC tumor cells to avoid NK cell cytotoxicity.
We also explored biological regulation networks between ELFN1-AS1 and GDF15 in CRC. Additional reports also indicated that EZH2 could impact GDF15 expression via H3K27me3, suggesting that histone modifications are involved in GDF15 regulation [57]. Histone modifications play a pivotal role in gene expression: H3K9ac has been correlated to active enhancers, H3K18ac is generally associated with active gene expression, and H3K27me3 was negatively correlated with transcript levels [58]. Our data indicated that ELFN1-AS1 regulated GDF15 primarily via the H3K9ac modification and not H3K14ac or H3K27me3. Histone acetylation promotes transcription by relaxing chromatin [59], H3K9ac is regulated by the GCN5-SND1 complex and contributes to cancer development [60]. In this process, GCN5 is recruited to the promoter regions to increase chromatin accessibility and acetylates H3 on the chromatin around double-strand breaks (DSB) [61]. Meanwhile, SND1 interacts with GCN5 and plays a role as a recruiter and coactivator [60, 62]. Our data verified that silencing of ELFN1-AS1 attenuated the enrichment of GCN5 on chromatin in CRC cells, leading a decrease of H3K9ac enrichment. SND1 is primarily located in the cytoplasm and is translocated into the nucleus following phosphorylation to form the GCN5-SND1 complex. Here, we identified an interaction between ELFN1-AS1 and SND1. The human SND1 protein contains 4 repeated staphylococcal nuclease-like domains (SN1 to 4) and the downstream TSN domain (Tudor plus SN5 fragments) were identified in the human SND1 protein. Consistent with this, our data demonstrated that the SN2 domain mediated the binding between SND1 and ELFN1-AS1, which linked to GDF15 expression. Moreover, as expected, SND1 silencing in CRC cells directly downregulated GDF15 secretion and co-culture with the SND1 silenced CRC cells restored the cytotoxicity of NK cells against CRC cells. Silencing of GCN5 had similar effects, indicating that ELFN1-AS1 may mediate the production of GDF15 though the GCN5-SND1 complex. GEPIA2 results also demonstrated that the expression of GCN5, SND1 and GDF15 was respectively correlated with ELFN1-AS1, (Fig. S4A-C) and this finding corroborates with our direct experimental observations.
We also analyzed the expression and regulation network of ELFN1-AS1 using RNA microarrays. HPA RNA-seq normal tissues analysis from lncbook indicated that ELFN1-AS1 was highly expressed in brain, rectum and stomach tissues (Fig. S4D). In contrast, in lncexpdb, elevated expression of ELFN1-AS1 was found in stomach, rectum and colon normal tissues (Fig. S4E). ELFN1-AS1 levels in ENCODE primary cell lines indicated a maximum transcripts per million (TPM) in kidney epithelial cells (Fig. S4F) indicating a different expression profile of ELFN1-AS1 in multiple tissues and cells. Methylation analysis also indicated that the methylation levels of promoter (Fig. S5A) and body (Fig. S5B) regions of ELFN1-AS1 are both aberrant in CRC and READ compared with normal tissues. Expression of ELFN1-AS1 is also associated with sample type (Fig. S5C-D). These suggested that aberrant methylation might be responsible for the high expression of ELFN1-AS1 in CRC tissues. Co-expression and KEGG pathway analysis revealed that ELFN1-AS1 was involved in the metabolic pathways and pathways in cancer (Fig. S6A-D), suggesting a major biological function for ELFN1-AS1 in the metabolism of CRC. All of these describe an important role for both the direct action of ELFN1-AS1 on cancer cells and an indirect action on normal cells.