Identification of LINC01977 as a super-enhancer-associated cancer-testis lncRNA
To characterize the dysregulated lncRNAs regulated by SEs, super-enhancer-associated lncRNA (SE-lncRNA) microarrays were used for five paired tumor and adjacent normal tissues from LUAD patients (Additional file 2: Table S1), which revealed that 515 lncRNAs were differentially expressed, including 189 significantly up-regulated and 326 significantly down-regulated lncRNAs in malignant tissues compared with normal tissues. By overlapping the dysregulated lncRNA from TCGA and our microarray analysis data, we obtained 55 dysregulated lncRNAs (Additional file 1: Figure S1A, B), of which 18 lncRNA transcripts have been validated in NCBI (https://www.ncbi.nlm.nih.gov/). We focused our attention on the transcript LINC01977 due to the following reasons: (1) It was the most significantly differentially expressed lncRNA in the analysis (Fig. 1A); (2) it was localized in 17q25.3 (Fig. 1B), whose genomic instability has been reported to be associated with the malignant progression of NSCLC [20]; and (3) analysis of the GTEx dataset demonstrated that LINC01977 showed high expression in the normal testis (Fig. 1C), referring to cancer-testis genes, which have potential roles in tumorigenesis and progression [21].
Additionally, pan-cancer analysis from the public database showed that LINC01977 showed higher expression in cancer cell lines (CCLE, CGP, and Genentech) (Fig. 1D) and LUAD tumor tissues (ImmLnc dataset) (Fig. 1E). Subsequently, we performed qRT-PCR on 40 paired LUAD tissues and observed that LINC01977 was highly expressed in paired LUAD tumor tissues (Fig. 1F), which was validated in the other 186 unmatched LUAD tumor tissues (Additional file 1: Figure S1C) and the GEO dataset of LUAD (Additional file 1: Figure S1D, E). Next, we performed RNAscope in situ hybridization (ISH) to detect LINC01977 expression in the tissue microarray -containing 98 pairs of well-annotated LUAD patients (Fig. 1G, Additional file 2: Table S2). Our results demonstrated that LINC01977 was highly expressed in LUAD tissues and correlated with poorer overall survival (Fig. 1H).
As LINC01977 was annotated as a SE-associated lncRNA in the aforementioned microarray, we assessed the impact of the SE on the transcriptional regulation of LINC01977. We mined the H3K27ac and H3K4me1 ChIP-seq data, the active markers of the SE, in the canonical LUAD cell lines (A549 and PC-9) and normal lung tissue from the Encyclopedia of DNA Elements (ENCODE) database, and an aberrantly activated SE region located upstream of LINC01977 spanning 48 kb was identified in A549 cells (Fig. 1I). However, in the PC-9 cell line, there were only modest changes in these specific peaks, which may have been due to the tissue and cell line specificity of SE [22]. Moreover, Hi-C data of A549 cells from the public database highlighted that the SE region directly interacted with the promoter region of LINC01977 (Fig. 1J). To clarify whether the SE region we identified in A549 cells could affect the transcription of LINC01977, the SE region of LINC01977 was divided into six constituents (E1–E6), which were designed as dual-luciferase reporter gene plasmids, containing the LINC01977 promoter region (Fig. 1K). Consistent with SE-specific peaks we obtained from ChIP-seq data, majority of SE sub-region (5/6) were generally higher in A549 cells compared to PC-9 cells. The results suggested that SE caused strong transcription-enhancing activity in LINC01977 in A549 cells, while moderate transcription-enhancing activity in PC-9 cells (Fig. 1L). Moreover, treatment with the BRD4 inhibitor JQ1 decreased the LINC01977 mRNA level in A549 cells, while mRNA alteration of the neighboring gene CBX4 was not significant, consistent with the feature of SE-lncRNAs (Additional file 1: Figure S1F). Additionally, the LINC01977 mRNA level was different in A549 cells compared with the other LUAD cell lines and normal lung cell line (Additional file 1: Figure S1G). To exclude the possibility that this difference is due to the driver mutation status in LUAD cell lines, we performed correlation analysis in TCGA-LUAD dataset, which suggesting that no differences were detected in LINC01977 expression by mutation status (Additional file 1: Figure S1H-L). Collectively, these findings indicated that the cancer-testis lncRNA, LINC01977, was driven by the aberrantly activated SE in LUAD.
LINC01977 exerts oncogenic roles in vitro and in vivo
To investigate the function of LINC01977 in the malignant progression of LUAD, we performed loss-of-function experiments in vitro and in vivo. Antisense oligonucleotides (ASO) were selected to inhibit LINC01977 expression. The transfection efficiency in LUAD cell lines were evaluated by qRT-PCR (Additional file 1: Figure S2A, B). We performed EdU assays and demonstrated that LINC01977 promoted proliferation of the LUAD cell line (Fig. 2A, Additional file 1: Figure S2C). Additionally, cell cycle assessment suggested that the absence of LINC01977 blocked the G1/S cell cycle transition (Fig. 2B, Additional file 1: Figure S2D). Next, annexin V staining followed by flow cytometry analysis showed increased apoptosis in LINC01977-deficient LUAD cells compared with the control LUAD cells (Fig. 2C, Additional file 1: Figure S2E). The migration and invasion abilities of LUAD cells were detected by wound healing, Transwell, and Matrigel invasion assays, which suggested that the absence of LINC01977 significantly reduced cell migration and invasion (Fig. 2D, Additional file 1: Figure S2F–H). Furthermore, 3-D tumor sphere formation and 2-D plate clone formation were assessed to evaluate the clonogenicity of LUAD cells, and the results revealed that the absence of LINC01977 reduced the clonogenicity of LUAD cells (Fig. 2D, Additional file 1: Figure S2I, J). These results revealed that LINC01977 deficiency attenuated LUAD cells proliferation and malignant phenotypes in vitro.
Next, we investigated the biological functions of LINC01977 in vivo. Nude mice were injected with A549 cells to establish a subcutaneous tumor bearing model, and LINC01977-ASO or the control was injected intratumorally every 3 days (Fig. 2E). The results indicated that tumors injected with LINC01977-ASO were smaller in size, lighter in weight, and retarded in growth compared with controls; meanwhile, efficient LINC01977 knockdown was confirmed by qRT-PCR (Fig. 2F–J). Additionally, we assessed whether inhibition of LINC01977 affected tumor metastasis in the xenograft metastasis model (Fig. 2K). Cancer cell colonization in secondary organs was monitored by the IVIS (Fig. 2L). Analysis of lung metastatic colonies indicated that LINC01977-ASO suppressed the formation of distant metastases (Fig. 2L, M, Additional file 1: Figure S2K). Taken collectively, these findings indicated that LINC01977 behaved as an oncogene to promote the malignant progression of LUAD.
LINC01977 binding to SMAD3 protein is dependent on the MH2 domain
To determine the exact full length of LINC01977, we performed 5’ and 3’ rapid amplification of complementary DNA ends (RACE) assays. The results revealed that LINC01977 was a 2791-nt lncRNA, which was longer than the sequence we obtained from NCBI and contained stem-loop structure (Fig. 3A-C, Additional file 1: Figure S3). According to the coding potential assessment tool (CPAT), the score of LINC01977 was very low compared with GAPDH, a well-known coding RNA, which was consistent with the characteristics of lncRNAs (Fig. 3D). To explore the subcellular localization of LINC01977, we detected its expression in cytoplasmic and nuclear fractions by qRT-PCR analysis, which indicated that LINC01977 was predominantly localized in the nucleus (Fig. 3E). This was further verified by RNA fluorescence in situ hybridization (FISH) assays (Fig. 3F).
It has been widely reported that lncRNAs exert their biological effects by binding to proteins [23]. To uncover the mechanism underlying the role of LINC01977 in LUAD, we performed RNA pull-down assays and subjected the precipitates to mass spectrometry analysis to identify the potential LINC01977-interacting proteins. The silver staining results showed enrichment of several bands of proteins potentially combined with LINC01977, distributed in the ~ 40–55 kDa and > 180 kDa regions (Fig. 3G–I). According to the unique peptides and cover percentages from the assays, we focused on the SMAD3 protein for the next experiment (Additional file 2: Table S3). CatRAPID was used to predict the potential SMAD3-binding regions in LINC01977 and the potential protein-binding domains of LINC01977 in SMAD3 (Additional file 1: Figure S4A). RIP and RNA pull-down assays confirmed that LINC01977 interacted with SMAD3, and not with SMAD2 or SMAD4 (Fig. 3J–K, Additional file 1: Figure S4B, C), the important component of the R-SMAD complex.
To clarify the nucleotide sequence of LINC01977 that binds to SMAD3, we constructed a series of LINC01977 deletion mutants according to the stem-loop structures (Fig. 3C, Additional file 1: Figure S4D). The results showed that deletion of the RNA fragment ranging from 1799 to 2791 significantly decreased the binding to SMAD3, which revealed that this region was critical for the interaction of LINC01977 with SMAD3 (Fig. 3L). Similarly, to assess the precise functional domains of SMAD3 interacting with LINC01977, we generated N-terminal HA fusion-tagged full-length (FL) or truncated SMAD3 constructs containing the MH1 domain (amino acids 1–130), linker domain (amino acids 130–230), and MH2 domain (amino acids 230–425) (Fig. 3M). RNA pull-down assays revealed that LINC01977 bound to the MH2 domain of SMAD3, which is critical for the translocation, phosphorylation, and transcription factor/co-activator binding of SMAD3 (Fig. 3N) [24]. We also demonstrated that high expression of SMAD3 was associated with poor prognosis in patients with LUAD in both TCGA-LUAD and LUAD GEO datasets (GSE31210 and GSE13213) (Additional file 1: Figure S4E). Taken collectively, these results revealed that LINC01977 bound with the MH2 domain of SMAD3 protein.
LUAD cells adapt to TAM2-induced TGF-β enrichment in the tumor microenvironment via SE-LINC01977-dependent TGF-β/SMAD3 signaling pathway activation
TGF-β signaling is mainly mediated through the canonical SMAD3 signaling pathway. TGF-β-mediated phosphorylation of SMAD3 on Ser423/425 is essential for TGF-β/SMAD3 signal transduction [25]. To assess whether LINC01977 was involved in the TGF-β/SMAD3 signaling pathway, we detected phosphorylated SMAD3 Ser423/425 in a time-dependent TGF-β treated manner. We found that phosphorylated SMAD3 protein expression was elevated within 30 min of TGF-β treatment in LINC01977-overexpressing LUAD cells, suggesting that LINC01977 contributes to the SMAD3-mediated early TGF-β response (Fig. 4A, Additional file 1: Figure S5A, B). To clarify the impact of LINC01977 on p-SMAD3, immunofluorescence assays were performed in A549 cells. The results indicated that p-SMAD3 mainly appeared in the nucleus in TGF-β treated A549 cells and predominantly accumulated in the nucleus following LINC01977 overexpression (Fig. 4B). Next, LUAD cells were harvested for nuclear and cytoplasmic fractions at 24 h after gain- or loss- of LINC01977. Fractionation followed by western blotting analysis demonstrated that LINC01977 promoted p-SMAD3 nuclear accumulation, which was consistent with the immunofluorescence results (Fig. 4C, Additional file 1: Figure S5C, D). Furthermore, LINC01977 deficiency attenuated p-SMAD3 nuclear translocation as well under TGF-β stimulation in LUAD cells (Fig. 4D, Additional file 1: Figure S5E, F).
To analyze the intracellular distribution of p-SMAD3 in A549 cells, we used two-photon confocal microscopy. Unsurprisingly, LINC01977 overexpression significantly promoted p-SMAD3 nuclear accumulation after TGF-β stimulation (Fig. 4E). Moreover, perturbations in the SMAD3 mRNA level were not observed in the presence or absence of LINC01977 in A549 cells (Fig. 4F, G). Next, we assessed the impact of the TGF-β/SMAD3 signaling pathway on the expression of LINC01977. We used the p-SMAD3 inhibitor SIS3 and the TGF-β activation inhibitor SB431542, and the results revealed that LINC01977 expression was significantly decreased during TGF-β inhibition and p-SMAD3 inactivation in LUAD cells (Fig. 4H, Additional file 1: Figure S5G). These results indicated that LINC01977 expression was associated with the activation of the TGF-β/SMAD3 signaling pathway.
A previous study has reported that SMAD3 is a key transcription factor in the malignant progression of tumors. In LUAD, survival analysis revealed that high-expression SMAD3 was associated with poor prognosis (Additional file 1: Figure S4E) [26]. Motif analysis showed three predicted SMAD3-binding sites in the LINC01977 promoter region (Fig. 4I). The results of ChIP-qPCR revealed SMAD3 occupancy in the LINC01977 promoter (Fig. 4J). Furthermore, luciferase reporter assay revealed that LINC01977 expression was promoted in a TGF-β/SMAD3 partially dependent manner (Additional file 1: Figure S5H). Interestingly, SMAD3 occupancy was also observed in the SE region of LINC01977 in LUAD cells, which demonstrated that SMAD3 transcriptionally regulates LINC01977 via promoter binding and epigenetic regulation of SE occupancy (Fig. 4K, Additional file 1: Figure S5I). Indeed, luciferase reporter assays supported this conclusion (Fig. 4L).
Tumor-associated macrophages (TAMs), highly abundant components in the tumor microenvironment, construct pre-metastasis niches [27]. In the malignant progression of LUAD, the abundance of TGF-β in the tumor microenvironment was attributed to the high infiltration of M2-like TAMs (TAM2) [28]. By GSEA, we also found that high M2 infiltration in LUAD was significantly associated with TGF-β signaling pathway activation and enriched in early-stage LUAD (Additional file 1: Figure S5J, K). Accordingly, we used the human monocyte THP-1-derived in vitro model that mimics this phenotype of TAM2 for co-culture with LUAD A549 cells (Fig. 4M), and the results were validated by the TAM2 specific marker CD68/CD206 via both IF and Flow cytometry analysis (Fig. 4N, Additional file 1: Figure 5L). Next, the secretion of TGF-β was confirmed by enzyme linked immunosorbent assay (ELISA) (Additional file 1: Figure S5M). qRT-PCR and functional experiments provided evidence that co-culturing with TAM2 increased LINC01977 expression in LUAD cells and promoted the EMT process, consistent with TGF-β-induced malignant progression. These observations were also validated by LINC01977 and TGF-β dependent experiments (Fig. 4O, Additional file 1: Figure S5N-P). Taken collectively, these results confirmed that TAM2 promoted the LUAD malignant phenotype.
It has been reported that lncRNAs are often enriched in a tissue- or cell-specific manner [29]. Studies on epigenetic diversity and heterogeneity in cancer have demonstrated that SEs are also highly cell-specific [30, 31]. To investigate how LUAD cells adapt to complex tumor microenvironments, we first detected LINC01977 in LUAD cells and tumor microenvironment-associated cells. To this end, we obtained tumor microenvironment-associated cells derived from peripheral blood mononuclear cells (PBMCs), including CD4+ T cells, CD3+ T cells, CD8+ T cells, and DCs, as well as cancer-associated fibroblasts (CAFs). We observed that LINC01977 was highly enriched in LUAD cell lines. Additionally, upon stimulation with TGF-β, increased LINC01977 expression was observed in LUAD cells (A427, H1299, SW1573, and H1975), but the changes were not significant in both tumor microenvironment-associated cells and human normal bronchial epithelium (HBE) cells (Additional file 1: Figure S5Q, R). These observations potentially reflected that high LINC01977 expression was the specific response of LUAD cells to TAM2 infiltration induced by TGF-β enrichment in the tumor microenvironment. As previously shown, LINC01977 was involved in the TGF-β/SMAD3 signaling pathway, and these results were likely due to LUAD cell adaption to TAM2 infiltration. Additionally, this adaptive behavior was validated by functional experiments (Fig. 4O, Additional file 1: Figure S5O). Thus, our finding uncovered that LUAD cells adapt to TAM2-induced TGF-β enrichment in the tumor microenvironment in two ways: (i) by sensing and activating the canonical TGF-β signaling pathway and (ii) by SE-LINC01977 binding and tethering SMAD3 in the nucleus to promote TGF-β/SMAD3 signaling transduction via epigenetic regulation.
Epigenetic transcriptional activation of the SMAD3/CBP/P300 complex is LINC01977-dependent and can potentially serve as a target for combination therapy
CREB-binding protein (CBP) and P300 (CBP/P300), transcriptional co-activators, have been reported to bind to SMAD3 and enhance SMAD-induced transactivation of target genes [32]. In this study, we identified that CBP/P300 was among the listed proteins of the quantitative LC/MS assays in which LINC01977 RNA pulled-down (Fig. 3H, Additional file 2: Table S3). Subsequently, OmiXcore was used to predict the interaction between CBP/P300 and LINC01977 (Fig. 5A). Both RNA pull-down and RIP results validated the interaction (Fig. 5B, C). To assess the impact of LINC01977 on SMAD3/CBP/P300, we performed co-IP assays in LUAD cells (Fig. 5D, Additional file 1: Figure S6A-C). The results revealed that LINC01977 promoted the interaction between p-SMAD3 and CBP/P300 upon TGF-β stimulation. Next, we repeated the co-IP assays, where RNase was used to decrease LINC01977 expression and RNase inhibitor was used to protect LINC01977 from degradation. Unsurprisingly, the absence of LINC01977 significantly abolished the interaction between p-SMAD3 and CBP/P300, while the interaction between CBP and P300 was unchanged in LUAD cells (Fig. 5E, Additional file 1: Figure S6D, E). As a well-known histone acetyltransferase (HAT), CBP/P300 is primarily responsible for acetylating H3K27 and regulating gene transcription [33]. To investigate whether LINC01977 affects histone modification, in situ IF assays were performed to detect the transcriptional activation marker H3K27Ac level, which suggested that high expression of LINC01977 indeed resulted in the acetylation of H3K27 (Fig. 5F). Furthermore, to verify whether the transcription of LINC01977 could benefit from CBP/P300 and induce H3K27Ac, we performed ChIP-qPCR for detection. Interestingly, similar to the SMAD3 occupancy (Fig. 4I–K), co-occupancy of CBP/P300 and H3K27Ac was also detected in the SE (Fig. 5G, Additional file 1: Figure S6F, G) and promoter region of LINC01977 (472–480 and 1466–1477) (Fig. 5H, Additional file 1: Figure S6H, IL). Taken collectively, these observations demonstrated that SE-LINC01977 not only participated in the TGF-β/SMAD3 pathway in LUAD cells, but also leveraged the LINC01977-dependent SMAD3/CBP/P300 epigenetic transcriptional complex.
Recently, a growing number of small molecule inhibitors of key epigenetic regulators and nucleic acid drugs have been tested in cancer clinical trials [34]. Thus, we attempted to explore whether SGC-CBP30, the small molecule inhibitor of CBP/P300, could delay the malignant progression of LUAD. In vitro functional experiments suggested that combination therapy consisting of LINC01977-ASO and SGC-CBP30 significantly decreased LUAD cells migration and invasion (Fig. 5I, Additional file 1: Figure S6J-L). To better understand the effect of combination therapy in LUAD, the metastatic lung colonization model was used to mimic the in vivo environment (Fig. 5J). The results revealed that combination therapy resulted in remarkable suppression of lung metastasis, while SGC-CBP30 or LINC01977-ASO alone also produced inhibitory effects (Fig. 5K). Moreover, combination therapy-treated mice had significantly longer morbidity-free survival (Fig. 5L). Furthermore, consistent with tumor suppression, combination therapy significantly reduced p-SMAD3 and CBP levels in xenograft tumor mouse models compared with single-agent treatment (Fig. 5M). Thus, these results indicated that a significant therapeutic advantage was found by combining SGC-CBP30 with LINC01977-ASO compared with either agent alone.
LINC01977-dependent SMAD3/CBP/P300 complex epigenetically up-regulates ZEB1, the central switch of EMT process, in LUAD cells
To determine the major pathways affected by LINC01977, we performed RNA-seq analysis in A549 cells to explore the disturbed transcriptome affected by LINC01977 and to identify the potential downstream targets. There were 835 up-regulated genes (LINC01977 vs pcDNA3.1) and 866 down-regulated genes (LINC01977-ASO vs scramble) (Fig. 6A, Additional file 2: Table S4). Analysis from Kyoto Encyclopedia of Genes and Genomes (KEGG) showed that highly expressed LINC01977 was associated with cancer, signal transduction, and transport and catabolism (Additional file 1: Figure S7A). Gene Ontology (GO) enrichment analysis also revealed that LINC01977 was associated with protein binding and nuclear processes (Additional file 1: Figure S7B). By screening overlapping differentially expressed genes and qRT-PCR validation, ZEB1 was noted to have a consistent correlation tendency with LINC01977 expression (Fig. 6B, C). Among these genes, ZEB1 was the most attractive candidate based on the following: (i) ZEB1, the core transcriptional inducer of EMT, is a key factor for cell plasticity and cancer metastasis [35], (ii) ZEB1 maintains cell fate and stem cell quiescence [36], and (iii) its elevated expression in poor-prognosis LUAD (Fig. 6D). We therefore proceeded to examine whether ZEB1 was the downstream target gene of SE-LINC01977 in LUAD cells.
A positive correlation between SMAD3 and ZEB1 was observed in the LUAD public dataset (Fig. 6E). Moreover, motif analysis indicated that the SMAD3 specific binding motif, SMAD-binding element (SBE), was found in the ZEB1 promoter region by JASPAR (Fig. 6F). According to the identified binding sites, we constructed wild-type (WT) and mutant (MT1-MT4) luciferase reporter gene plasmids (Fig. 6G). The results revealed that transcription of ZEB1 was promoted by TGF-β, and it was profoundly elevated by LINC01977-SMAD3 overexpressed together, indicating that LINC01977-SMAD3 played an important role in the transcriptional activation of ZEB1 (Fig. 6H). Subsequently, to clarify which SMAD3-binding site plays the major role in ZEB1 regulation, we performed luciferase activity assays, which suggested that a mutation in binding site #4 directly abolished the regulation from TGF-β/LINC01977-SMAD3 to ZEB1 (Fig. 6I, Additional file 1: Figure S7C, D). Furthermore, ChIP-qPCR results also provided evidence that the LINC01977-dependent SMAD3/CBP/P300 complex co-occupied the wild-type binding site #4, along with increased acetylation at H3K27 (Fig. 6J, Additional file 1: Figure S7E, F). With LINC01977 overexpression in A549 cells, these co-occupancies were enhanced together (Fig. 6K). Additionally, to further validate whether loss-of chaperon proteins attenuate ZEB1 expression, we employed CRISPR/Cas9 system to knockout SMAD3, CBP and P300 in LUAD cells. Gene editing efficiency were verified by PCR amplification (Additional file 1: Figure S7G, H). With LINC01977-overexpressed LUAD cells, loss-of chaperon proteins significantly decreased ZEB1 expression, suggesting that LINC01977-mediated transcriptional regulation of ZEB1 was dependent on LINC01977-SMAD3/CBP/P300 complex (Additional file 1: Figure S7I). Furthermore, as an EMT activator, ZEB1 plays a key role in cell plasticity and promotes metastasis of cancers. EMT is also associated with greater metastatic potential [37]. High expression of ZEB1 was associated with the EMT pathway activity by GSCA bioinformatic analysis (Additional file 1: Figure S7J). To confirm the impact of LINC01977 on the EMT process, we performed qRT-PCR of EMT-associated genes and observed that LINC01977 promoted EMT in LUAD cells (Additional file 1: Figure S7K). Subsequently, we detect EMT and cell cycle related proteins expression in gain- or loss- of LINC01977 LUAD cells. The results of western blot assays suggested that LINC01977 promoted EMT and cell cycle progression (Additional file 1: Figure S7L). Additionally, we performed IHC analysis in tumor tissues we obtained from subcutaneous model and xenograft metastasis model for in vivo validation. As expected, LINC01977 deficiency decreased p-SMAD3, ZEB1, Ki-67, and PCNA expression; meanwhile, expression of epithelial marker E-cadherin and apoptosis marker cleaved caspase 3 was elevated, which revealed that LINC01977 promotes EMT and proliferation in vivo (Additional file 1: Figure S7M). Taken together, these results established an axis for the SE-LINC01977-dependent SMAD3/CBP/P300 complex that epigenetically up-regulated ZEB1, thereby promoting the EMT process via ZEB1 in LUAD cells.
LINC01977 is a promising predictor of poor outcome in early-stage LUAD
To investigate the correlation between LINC01977 and TAM2 infiltration, the most abundant stromal cell population in LUAD, we assessed the expression of LINC01977 and TAM2 using serial sections in a cohort of LUAD tissues (n = 186) coupled with qRT-PCR and immunofluorescence staining of CD68+/CD206+, respectively (Fig. 7A). Interestingly, high LINC01977 expression was observed in early-stage LUAD, and TAM2 were significantly enriched at the invasive margin (Fig. 7B, C). Moreover, these results were validated in TCGA-LUAD dataset (Additional file 1: Figure S5K). Correlation analysis demonstrated that LINC01977 was positively correlated with TAM2 infiltration in early-stage LUAD (Fig. 7D). As TAM2-induced TGF-β signaling pathway activation, higher chromatin accessibility in the SE region was observed in LUAD patients with high expression of TGF-β by ATAC-seq analysis from TCGA-LUAD (Fig. 7E). Additionally, the LINC01977 mRNA level in LUAD patients with high SMAD3 protein expression was similarly elevated in TCGA-LUAD dataset (Fig. 7F). Moreover, high-TAM2 infiltration-induced TGF-β signaling pathway activation was associated with poor prognosis in early-stage LUAD, which was also validated in TCGA-LUAD dataset (Fig. 7G). Taken collectively, these results indicated that TAM2 were highly infiltrated at the invasive margin, which was correlated with the high expression of LINC01977.
To explore the correlation between LINC01977 expression and the clinical-pathological date in LUAD, we performed disease-free survival (DFS) analysis in the TCGA-LUAD dataset, which revealed that high expression of LINC01977 was significantly associated with poor prognosis in early-stage LUAD (Stage I) (Fig. 7H), whereas the predictive power decreased with advanced clinical stage (Fig. 7I, J). Collectively, we speculated that LINC01977 functioned as a biomarker of poor prognosis in early-stage LUAD.