1. Introduction
TCF12 (also called HEB or HTF4) is a class-I member of the helix–loop–helix (HLH) protein family and functions as a transcription factor that directly binds to the E-box sites (CANNTG) of target gene promoters [ 1 , 2 ]. There are 15,373 genes identified as TCF12 targets based on the ENCODE Transcription Factor Targets Dataset using DNA sequencing analyses following anti-TCF12 chromatin immunoprecipitation (ChIP) [ 3 , 4 ]. The fact that TCF12 has such a vast number of target genes suggests that it can be a highly influential transcription factor. Indeed, TCF12 regulates a wide range of cellular processes, e.g., the mesenchymal transition of epithelial/endothelial cells [ 5 , 6 , 7 ], fibroblast activation and extracellular matrix remodeling [ 8 ], the drug uptake of cancer cells [ 9 ], and muscle development [ 10 ]. In order for TCF12 to correctly regulate the expression of the proper downstream genes in the responsible cells under the right conditions, it likely needs to cooperate closely with other regulatory proteins and RNAs to ensure that it plays an accurate role in the transcriptional machinery complex. For example, TCF12 heterodimerizes with TWIST1, a class-II member of the HLH family, which is required for the proper formation of the junction structure of the frontal and parietal bones during development [ 11 ]. The heterodimer formed by TCF12 and MyoD, another class-II member of the HLH family, is involved in myogenesis [ 10 ]. When TCF12 forms a complex with Id-1, a class-V member of the HLH family, TCF12 is unable to bind to the promoters of target genes for transcriptional regulation [ 9 , 12 ]. Additionally, TCF12 can function as a transcriptional repressor by associating with Bmi1 and EZH2, the components of polycomb-repressive complex 1 (PRC1) and PRC2, respectively [ 5 ]. Apart from the aforementioned regulatory proteins, the long non-coding RNA (lncRNA) MIAT can also form a complex with TCF12, promoting its binding to the NFAT5 gene promoter to activate gene expression [ 13 ]. TCF12 overexpression has been detected in many malignancies such as colorectal cancer [ 5 ], breast cancer [ 8 ], gallbladder cancer [ 14 ], hepatocellular carcinoma [ 15 ], melanoma [ 13 ], and pancreatic cancer [ 16 ].
While ~80% of the human genome can be transcribed into RNA, only ~1.5% of whole genomic DNA is responsible for the synthesis of protein-coding transcripts [ 17 , 18 ]. Most of the human genome expresses non-coding RNA, which is involved in diverse biological processes, including the regulation of mRNA expression. Non-coding RNA with a length >200 nucleotides is generally classified as lncRNA, distinguishing it from small-sized non-coding RNA, including microRNA (miRNA), Piwi-interacting RNA (piRNA), and small nucleolar RNA (snoRNA), among others. The first identified and widely studied lncRNA is Metastasis-Associated Lung Adenocarcinoma Transcript 1 (MALAT1), which is also named Nuclear-Enriched Abundant Transcript 2 (NEAT2) [ 19 ]. MALAT1 has a length of 8664 nucleotides and exerts its versatile functions by interacting with many diverse factors. For example, it is involved in the alternative processing of pre-mRNA by recruiting and modulating splicing-regulatory proteins into nuclear speckles [ 20 , 21 , 22 ]. MALAT1 participates in a transcription-repressive complex by associating with PRC2 components EZH2, SUZ12, and EED to drive the trimethylation of histone H3 at lysine-27 and the repression of target gene promoters [ 23 , 24 ], but it acts as a molecular scaffold promoting the interactions among unmethylated Pc2, E2F, and other transcription factors to trans-activate downstream gene expression [ 25 ]. It can also play a role of guiding RNA to facilitate Sp1 binding to the target gene promoter [ 26 ]. Additionally, MALAT1 can function as a competing endogenous RNA to sponge miRNA and in turn up-regulate the miRNA targets [ 27 , 28 , 29 ]. MALAT1 was discovered because its high expression was associated with tumor metastasis and patients’ poorer survival in early-stage non-small-cell lung cancer [ 19 ]. Its abnormally elevated expression has been subsequently observed in various cancers such as hepatocellular carcinoma [ 30 ], breast cancer [ 31 ], colorectal cancer [ 32 ], prostate cancer [ 33 ], gastric adenocarcinoma [ 34 ], multiple myeloma [ 35 ], and esophageal cancer [ 36 ]. MALAT1 overexpression exhibits tumor-promoting effects through the promotion of tumor cell growth, proliferation, migration, invasion, metastasis, and survival [ 18 ].
Despite the reported clinical relevance of TCF12 and MALAT1 in CRC malignancy [ 5 , 32 ], their individual overexpression in the patient cohort from The Cancer Genome Atlas—Colon Cancer (TCGA-COAD) dataset does not show a significant association with poorer prognosis. Recognizing the multifaceted roles of TCF12 and MALAT1 through interactions with various factors, we employed RNA immunoprecipitation and sequencing to investigate the interactions between TCF12 and lncRNAs. Our findings revealed that MALAT1 is physically associated with TCF12 in CRC cells, and their combined expression was identified as a metastasis-independent prognostic factor for reduced overall survival in CRC patients. We term this relationship between TCF12 and MALAT1 the TCF12–MALAT1 alliance. To uncover downstream factors/events of the alliance contributing to this poorer prognosis, we performed signal pathway enrichment analysis on their shared 2312 target genes, identifying pathways such as Aurora B, ATM, PLK1, and non-canonical WNT as significantly involved. Moreover, through the TCGA-COAD Genome Analyzer (GA) dataset, we identified two distinct CRC patient subgroups: the “Pan-CMS-2 pattern”, associated with better survival, and the “TMBC pattern”, linked to poorer relapse-free and overall survival regardless of metastasis. Our findings provide support for developing personalized medicine to potentially reduce the risk of early mortality in CRC patients.
2. Materials and Methods
2.1. Cell Culture
Human CRC cell lines SW620 and SW480 were cultivated with RPMI-1640 medium supplemented with 10% of fetal bovine serum, 100 units/mL of penicillin, 100 μg/mL of streptomycin, and 2 mM of L-glutamine, which was conducted at 37 °C in a humidified incubator filled with 95% air and 5% CO 2 .
2.2. TCGA-COAD Dataset
We investigated the clinical implications of higher TCF12 mRNA expression, higher MALAT1 expression, and the TCF12–MALAT1 alliance in CRC using RNA sequencing and patients’ clinical data of the TCGA-COAD dataset acquired from the TCGA Hub in the UCSC Xena website [ 37 ]. The RNA sequencing data, transformed into log2 (x + 1) RSEM normalized counts, were downloaded and used for subsequent analyses without any preprocessing. The clinical information, including patient’s gender, age, and pathologic stage, was downloaded from the Phenotypes file, and relapse-free survival (RFS) and overall survival (OS) were determined based on the curated survival data available through Xena. The tumor purity of the TCGA-COAD dataset was estimated using the ESTIMATE (Estimation of STromal and Immune cells in MAlignant Tumor tissues using Expression data) algorithm [ 38 ]. We used ssGSEA (single-sample Gene Set Enrichment Analysis) to calculate the stromal and immune scores based on 141 immune-related genes and 141 stromal-related genes and then predicted tumor purity using the combination of the two scores. Additionally, the patients of the TCGA-COAD dataset were classified into 4 subtypes based on the criteria of the Consensus Molecular Subtype (CMS) classification [ 39 ], including CMS-1 (i.e., microsatellite-unstable subtype), CMS-2 (i.e., canonical WNT subtype), CMS-3 (i.e., metabolism-dysregulated subtype), and CMS-4 (i.e., mesenchymal subtype).
2.3. RNA Immunoprecipitation (RNA IP)
RNA IP was performed to identify TCF12-associated lncRNAs in CRC cells [ 40 ]. SW620 cells were harvested using trypsinization and washed three times with ice-cold PBS. The cell suspension (5 × 10 7 cells in 10 mL of PBS) was added with 162 μL of 18.5% formaldehyde for a cross-linking reaction at room temperature. After 15 min, the reaction was stopped by the addition of 1 mL of 2.5 M glycine and incubation for another 5 min. The cells were collected by centrifugation, washed twice with ice-cold PBS, and further mixed with 1 mL of lysis buffer (consisting of 50 mM of Tris-HCl, pH 7.4, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 1 mM of EDTA, and 150 mM of NaCl) plus 1 mM of dithiothreitol, cocktails of protease inhibitors (Sigma-Aldrich, St. Louis, MO, USA), and 40 U/mL of RNasin (Invitrogen, Carlsbad, CA, USA) on ice for 10 min. The mixture was further homogenized by an ultrasonic homogenizer and then centrifuged at 12,000 rpm for 15 min at 4 °C. One-third of the supernatant was saved as the INPUT fraction. The remaining supernatant was divided into two aliquots for an overnight incubation at 4 °C with 5 μg of control IgG and anti-TCF12 antibody (Cat. No. sc-28364, Santa Cruz Biotechnology, Santa Cruz, CA, USA), respectively. Subsequently, the two aliquots were further incubated for 30 min at room temperature with 50 μL of protein G-conjugated magnetic beads. The beads were collected using a magnet and washed three times with lysis buffer. One half of the beads of each sample were resuspended in 100 μL of Laemmli buffer and heated at 95 °C for 15 min prior to 10% SDS-polyacrylamide gel electrophoresis, followed by conventional immunoblot analysis with anti-TCF12 antibody (1:500 dilution, Cat. No. sc-28364, Santa Cruz Biotechnology). Another half of the beads of each sample were resuspended in 100 μL of reversion buffer (consisting of 50 mM Tris-Cl, pH 7.4, 150 mM NaCl, 5 mM EDTA, 10 mM dithiothreitol, 1% SDS, and 10 μg of proteinase K) and incubated at 42 °C for 30 min and 65 °C for 4 h before RNA isolation by TRIsure–chloroform extraction and RNeasy Mini column chromatography (Qiagen, Germantown, MD, USA). The TCF12-associated MALAT1 was detected by PCR. The primer sequences were as follows: forward, 5′-TCA-TAC-CTA-ACC-AGG-CAT-AAC-A-3′; reverse, 5′-AAG-TGC-TCA-CAA-GGC-AAA-TC-3′. The PCR condition was 95 °C (30 s), 56 °C (40 s), and 72 °C (40 s) for 30 cycles. The size of the PCR product was 206 bp.
2.4. RNA Library Sequencing
RNA samples, isolated from the aforementioned control IgG or anti-TCF12 immunoprecipitates and the INPUT control, were subjected to next-generation sequencing to identify TCF12-associated lncRNAs in CRC cells. First, RNA libraries were constructed according to the manufacturer’s protocol of the NEBNext ® Small RNA Library Prep Kit (New England Biolabs, Ipswich, MA, USA). The 5′ and 3′ adaptors were ligated to total RNA prior to subsequent reverse transcription and PCR amplification. The enriched cDNA fragments were size-fractionated by a 6% polyacrylamide gel electrophoresis, and the 100~200-nucleotide RNA fragments (140~155 nucleotides in length with both adapters) were purified for library construction. Small RNA libraries were subsequently sequenced using the Illumina MiSeq 150-cycle single-read platform (Illumina, San Diego, CA, USA). The generated sequence reads went through a quality control process to remove low-quality reads, followed by 3′ adaptor trimming to generate clean reads [ 41 , 42 ]. Only clean reads with read counts ≥ 2 were included for further analyses. They were aligned to UCSC human hg19 or the Ensembl database by the Bowtie 2 tool [ 43 ], and the significant peaks were called using the cisGenome tool [ 44 ]. The fold changes in the peaks were calculated by dividing the average fragments per kilobase per million (FPKM) across the RNA IP samples to the FPKM of the INPUT control. Transcripts were identified as TCF12-associated RNAs when their fold changes were >2-fold.
2.5. Gene Set Enrichment Analysis
The downstream target genes of TCF12 were obtained from the ENCODE dataset, and the target genes of MALAT1, predicted by the LncRNA2Target database, were searched from the UALCAN website of the University of Alabama at Birmingham [ 45 ]. We applied common target genes of TCF12 and MALAT1 to Model-based Gene Set Analysis (MGSA) for signal pathway enrichment analysis [ 46 ].
2.6. Statistical Analyses
The correlation between the expression levels of the two genes was analyzed using Pearson’s correlation analysis. Differences in the ages and gene expression levels between the two patient subsets were evaluated by an independent-samples t -test. Pearson’s Chi-square analysis was performed to examine the association of gene expression status with various clinicopathological characteristics. The OS and RFS rates of different subsets of CRC patients were calculated using the Kaplan–Meier method and compared by the log-rank test. Prognostic associations of the test variables were evaluated using the univariate and multivariate Cox regression analyses. The interaction term of TCF12 mRNA and MALAT1 expression levels, referred to as “TCF12 × MALAT1”, was also subjected to Cox regression analyses. The evaluation of TCF12 × MALAT1 involves more than simply multiplying their expression levels; it also requires adjusting with their individual effects. As such, “TCF12 × MALAT1” was represented as three covariates in a multivariate Cox regression analysis: TCF12 mRNA expression level, MALAT1 expression level, and the multiplication of TCF12 mRNA and MALAT1 expression levels. These statistical analyses were performed using SPSS 11.0 software (SPSS Inc., Chicago, IL, USA).
3. Results
3.1. The TCF12–MALAT1 Alliance Is Associated with Poor Prognosis in CRC Patients
TCF12 may exert its functions through interacting with other transcription-regulatory proteins and lncRNAs. Therefore, we wondered which lncRNAs could be associated with TCF12 to have some clinical impact on CRC patients. SW620 cells were used for anti-TCF12 RNA IP and next-generation sequencing experiments. Their lysate was prepared for reacting with anti-TCF12 antibody, and then the immunoprecipitates were divided into two aliquots for further extraction of the cellular proteins and RNAs, respectively. The protein fraction was analyzed by an immunoblot assay to confirm that TCF12 was specifically immunoprecipitated (the left panel in Figure 1 A). The RNA fraction was for RNA sequencing and further analyses through the NCBI and Ensembl databases. Several transcripts were suggested to be TCF12-associated lncRNAs, and MALAT1 is one of the lncRNAs of top rank (the right panel in Figure 1 A). To validate whether TCF12 was indeed associated with MALAT1, the lysate of SW620 cells was subjected to anti-TCF12 immunoprecipitation. RNA was extracted from the immunoprecipitates for further RT-PCR analysis of MALAT1 existence. The result showed that MALAT1 was indeed co-immunoprecipitated with TCF12 ( Figure 1 B). This association between TCF12 and MALAT1 can also be confirmed from another CRC cell line SW480 with ectopic TCF12 overexpression ( Supplemental Figure S1 ). Furthermore, we intended to investigate the clinical implication of the TCF12 and MALAT1 combination in CRC patients. The TCGA-COAD dataset includes 456 CRC patients’ data obtained from the GA or HiSeq platform. Concerning the batch effects that probably existed between the GA and HiSeq platforms, we estimated the tumor purities of the employed tumor tissues using the ESTIMATE algorithm. As shown in Figure 1 C, the data from the GA platform exhibited a smaller variation than those from the HiSeq platform. The TCGA-COAD (GA) data were thus used for further analyses. Pearson’s correlation analysis firstly revealed a significant correlation between the levels of TCF12 mRNA and MALAT1 in the human CRC specimens of the TCGA-COAD (GA) dataset (r = 0.377, p < 0.001). A relatively higher average level of MALAT1 expression (12.31 ± 1.34 vs. 11.49 ± 1.43, p < 0.001) was detected in the CRC tissues exhibiting higher (i.e., >median vs. ≤median) TCF12 mRNA expression levels ( Table 1 ), suggesting that TCF12 not only formed a complex with MALAT1 but that also its gene expression was coordinated with that of MALAT1 in CRC patients. We call this relationship between TCF12 and MALAT1 the TCF12–MALAT1 alliance. This alliance not only indicates the physical association between TCF12 and MALAT1 but also includes the interaction between the expression levels of TCF12 mRNA and MALAT1, referred to as “TCF12 × MALAT1”. To comprehensively evaluate the clinical implication of the TCF12–MALAT1 alliance in CRC prognosis, we conducted univariate Cox regression analyses on variates including gender, age, metastasis, TCF12 mRNA expression level, MALAT1 expression level, and TCF12 × MALAT1. As shown in Figure 1 D, TCF12 mRNA expression and MALAT1 expression under an alliance rather than them alone were significantly associated with the poorer OS of CRC patients. Under the TCF12–MALAT1 alliance, the coefficients for TCF12 mRNA expression and MALAT1 expression were 3.696 and 3.354, respectively, with corresponding p -values of 0.029 and 0.023. These associations remained statistically significant even after being adjusted for tumor metastasis and patients’ age, suggesting that TCF12 could collaborate with MALAT1 to exacerbate CRC patients’ OS independently of patients’ age and tumor metastasis ( Figure 1 E). Additionally, we also analyzed the association between the TCF12–MALAT1 alliance and poorer RFS outcome, finding a similar trend as with poorer OS outcome. However, the results did not reach statistical significance ( Supplemental Figure S2 ).
3.2. The TCF12–MALAT1 Alliance Is Associated with β-Catenin-Independent Cyclin D1 Gene Expression
To investigate why the TCF12–MALAT1 alliance is significantly correlated with the poorer prognosis of CRC patients, we identified 2312 overlapping genes from the 15,373 downstream target genes of TCF12 and the 2650 downstream target genes of MALAT1 ( Figure 2 A). Through MGSA signal pathway enrichment analysis, it was suggested that these overlapping genes are involved in pathways such as Aurora B, ATM, PLK1, and non-canonical WNT pathways ( Figure 2 B). Our present study further addressed the relationships between the TCF12–MALAT1 alliance and WNT downstream genes CTNNB1 , CCND1 , and MYC that encode β-catenin, cyclin D1, and c-Myc, respectively. WNT2B is the only WNT ligand revealed on the common target gene list of TCF12 and MALAT1, which can induce both canonical and non-canonical WNT pathways. The expression of cyclin D1 and c-myc genes can be induced through the canonical WNT–β-catenin pathway or mediated by the TCF12–MALAT1 alliance without the involvement of β-catenin. In the TCGA-COAD (GA) dataset, which includes 193 patients, the TCF12 mRNA levels showed significant positive correlations with the MALAT1, β-catenin mRNA, and cyclin D1 mRNA levels. However, the MALAT1 levels were only positively correlated with the TCF12 mRNA and cyclin D1 mRNA levels. The β-catenin mRNA levels, while positively correlated with the TCF12 mRNA levels, showed no significant correlation with the levels of MALAT1, cyclin D1 mRNA, and c-Myc mRNA ( Figure 2 C). Considering tumor heterogeneity occurring among patients, we tried to classify the 193 CRC patients of the TCGA-COAD (GA) dataset based on the criteria of the CMS classification. All of the 184 patients could be successfully assigned to CMS-1 ( n = 32, 17.4%), CMS-2 ( n = 70, 38.0%), CMS-3 ( n = 29, 15.8%), and CMS-4 ( n = 53, 28.8%), respectively ( Figure 2 C). The significant positive Pearson correlations among the TCF12 mRNA, MALAT1, β-catenin mRNA, and cyclin D1 mRNA levels were observed in the patients of CMS-2. For CMS-1 patients, the TCF12 mRNA levels were positively correlated with the levels of MALAT1 and β-catenin mRNA but not with the levels of cyclin D1 mRNA. Their MALAT1 levels were significantly correlated with the TCF12 mRNA levels but not with the β-catenin, cyclin D1, or c-Myc mRNA levels. In CMS-3 patients, TCF12 mRNA expression was correlated with cyclin D1 mRNA but not with MALAT1 or β-catenin mRNA expression. Their MALAT1 levels were not correlated with the TCF12, β-catenin, or cyclin D1 mRNA levels. The tumors in CMS-4 patients showed that the TCF12 mRNA levels were not related to the β-catenin, cyclin D1, or c-Myc mRNA levels, but their MALAT1 levels were significantly correlated with the TCF12 and cyclin D1 mRNA levels. These data together pointed out the existence of complicated relationships among the expression levels of TCF12 , MALAT1 , β-catenin , cyclin D1 , and c-myc genes in CRC. Nevertheless, CMS-2 patients had a relatively better prognosis among all CRC patients ( Figure 2 D). The CRC specimens with higher TCF12 and MALAT1 gene expressions tended to be CMS-2-subtype tumors only when they also exhibited higher β-catenin gene expression or lower cyclin D1 gene expression ( Figure 2 E). We, furthermore, investigated the cyclin D1 and β-catenin mRNA levels in CRC tissues that had higher expression levels of both TCF12 mRNA and MALAT1 (designated as the TCF12 hi MALAT1 hi expression status in Table 2 ). Compared to CRC tissues without simultaneous higher expressions of TCF12 mRNA and MALAT1, the TCF12 hi MALAT1 hi CRC tissues were detected to have a statistically significantly higher expression of cyclin D1 mRNA (12.27 ± 0.68 vs. 11.93 ± 0.60, p = 0.001). However, the difference between their β-catenin mRNA expression levels did not reach statistical significance (13.69 ± 0.45 vs. 13.56 ± 0.56, p = 0.108). These data suggest that β-catenin-independent higher cyclin D1 gene expression can be a tumor-promoting factor responsible for the significant association of TCF12 hi MALAT1 hi status with poor prognosis in CRC patients.
3.3. The TCF12–MALAT1 Alliance Exacerbates CRC Through Low β-Catenin but High Cyclin D1 Gene Expression
Furthermore, we investigated the impact of the β-catenin and cyclin D1 gene expression levels on the overall survival of CRC patients with TCF12 hi MALAT1 hi status. First, among the 193 patients in the TCGA-COAD (GA) dataset, 61 patients exhibited a TCF12 hi MALAT1 hi expression status. When these patients were further divided into two groups based on whether their β-catenin mRNA levels were greater than the median, we found that patients with higher β-catenin mRNA expression had a better prognosis than those with a lower expression ( p = 0.004, Figure 3 A). Among the 193 patients, only 28 had the TCF12 hi MALAT1 hi β-catenin lo expression status and showed a statistically significant poorer prognosis ( p = 0.017, Figure 3 B), but the Chi-square analysis showed no difference in tumor recurrence and metastasis ( p = 0.728 and 0.918, respectively). When the 61 patients with the TCF12 hi MALAT1 hi expression status were divided into two groups based on whether their cyclin D1 mRNA levels were greater than the median, we found that patients with higher cyclin D1 mRNA expression had poorer prognosis compared to those with lower expression ( p = 0.041, Figure 3 C). Among all 193 patients, 37 had the TCF12 hi MALAT1 hi cyclin D1 hi expression status, which was significantly associated with patients’ tumor recurrence ( p = 0.018) but not metastasis ( p = 0.515). Their poorer prognosis did not reach statistical significance ( p = 0.147, Figure 3 D). Next, we evaluated the combined effects of β-catenin and cyclin D1 gene expression levels on the OS of CRC patients with the TCF12 hi MALAT1 hi status. As shown in Figure 4 A, in the TCGA-COAD (GA) dataset, there were 33 patients whose tumor tissues simultaneously expressed higher levels of TCF12 mRNA, MALAT1, and β-catenin mRNA. These patients showed no significant difference in OS regardless of whether their cyclin D1 mRNA expression levels were greater than the median ( p = 0.288). In the TCGA-COAD (GA) dataset, 28 patients exhibited higher TCF12 and MALAT1 gene expression levels but lower β-catenin gene expression (i.e., the TCF12 hi MALAT1 hi β-catenin lo expression status). Among these patients, those with higher cyclin D1 gene expression levels had a noticeably poorer OS, although the difference did not reach statistical significance due to the small sample size ( p = 0.054, Figure 4 B). We further defined the patients with a TCF12 hi MALAT1 hi β-catenin lo cyclin D1 hi expression status as having the TMBC pattern. Among the 193 patients, 16 patients exhibited the TMBC pattern, and they had significantly worse OS compared to those without the TMBC pattern ( p < 0.001, Figure 4 C). Since tumor metastasis is a key factor leading to mortality in CRC patients, we validated the association between TMBC pattern and poorer OS in non-metastatic and metastatic patients, respectively. The results revealed that the patients with the TMBC pattern exhibited worse OS outcomes regardless of their metastasis status ( Supplemental Figure S3 ). Therefore, we further performed univariate Cox regression analyses to assess the impact of both the TMBC pattern and metastasis on OS. The results showed that both the TMBC pattern and metastasis were statistically significant risk factors for poorer OS in CRC patients ( Figure 4 D). Furthermore, a multivariate Cox regression analysis demonstrated that the TMBC pattern and metastasis were both statistically significant independent risk factors for poorer OS ( Figure 4 D). The hazard ratio (HR) of the TMBC pattern was even higher than that of metastasis (4.476 vs. 3.517). The nomogram shows that when the TMBC pattern had a prognostic score of 100, tumor metastasis exhibited a relative score of 84 ( Figure 4 E), suggesting that the TMBC pattern is a metastasis-independent and even stronger prognostic factor for poorer OS in CRC patients. Patients with the TMBC pattern could be predicted to have a one-year survival rate of 60%, but their three-year survival rate dropped to only 30%. These TMBC patients exhibited a worse prognosis regardless of metastasis. We wondered if tumor recurrence could be a primary risk factor contributing to the mortality of this group of patients. As shown in Figure 4 F, these TMBC patients indeed had a significantly shorter RFS, suggesting that early tumor recurrence could be a mortality risk for these patients.
4. Discussion
Despite advancements in medical treatments that have significantly improved the survival time and quality of life for CRC patients, tumor metastasis remains a critical factor that greatly worsens prognosis. The occurrence of metastasis often marks a pivotal turning point in CRC progression, typically associated with poorer survival outcomes. Currently, the five-year survival rate is approximately 90% for the patients with localized CRC (where cancer has not spread out of the colon or rectum). However, once CRC metastasizes to distant organs, patients’ survival rates drop sharply down to around 14%. Analysis of the available data of 447 patients included in the TCGA-COAD dataset also shows that over 70% of non-metastatic CRC patients survive beyond 5 years, but for those patients with CRC metastasis, their five-year overall survival rate reduces dramatically to less than 30%. The occurrence of metastasis signals a warning for disease progression and the patient’s survival, drawing the attention of most CRC researchers worldwide. However, for those CRC patients who do not develop metastasis but still have poor survival outcomes, the causes and mechanisms of mortality are rarely studied. In our study, we discovered that 8.3% of CRC patients, regardless of whether they experience tumor metastasis, have poor prognosis due to their tumor tissues expressing high TCF12 mRNA, high MALAT1, low β-catenin mRNA, and high cyclin D1 mRNA.
TCF12 is a transcription factor that regulates the RNA expression levels of its target genes. MALAT1 also regulates RNA expression through its roles as a molecular scaffold, guide RNA, and miRNA sponge [ 17 , 18 ]. Our present study has uncovered a TCF12–MALAT1 alliance, as MALAT1 was found to physically associate with TCF12 in CRC cells, and its gene expression was coordinated with that of TCF12 . Importantly, the expressions of TCF12 mRNA and MALAT1 under this alliance, rather than their individual expressions, were significantly associated with the poorer OS in CRC patients. Although the overexpression of MALAT1 [ 32 ] or TCF12 [ 5 ] has been independently linked to poorer prognosis in CRC patients, investigations conducted across different cancer types or patient cohorts may have contradictory results. The inconsistencies may partly arise from the involvement of TCF12 and MALAT1 in extensive biological processes mediated by interactions with various regulatory factors. As aforementioned, TCF12 is an E-box-binding transcription factor, with ChIP and DNA sequencing analyses identifying 15,373 candidate downstream target genes [ 3 ], indicating its broad regulatory influence. Due to its HLH protein structure, TCF12 can interact with numerous transcription factors, collectively shaping its role in regulating target gene expressions. TCF12 can also interact with lncRNAs that modulate its activity. For example, the lncRNA MIAT binds to TCF12 and facilitates its binding to the promoter region of NFAT5 gene, thereby activating NFAT5 expression and promoting melanoma cell proliferation, migration, and invasion [ 13 ]. In this study, we observed that TCF12 also binds to lncRNA MALAT1, one of the earliest identified and widely studied lncRNAs. According to the LncRNA2Target database, MALAT1 has approximately 2650 predicted target genes. Furthermore, considering MALAT1′s function as a molecular scaffold or guiding RNA to interact with DNA and transcription regulatory proteins [ 23 , 24 , 25 , 26 ], the total number of its target genes could far exceed this prediction. Notably, the TCF12–MALAT1 alliance was validated experimentally. When we ectopically overexpressed TCF12 in SW480 cells, MALAT1 expression increased correspondingly, and MALAT1 was found to bind to TCF12 ( Supplemental Figure S1 ). SW480, a cell line established from a primary CRC tumor, exhibits low TCF12 expression, whereas SW620, a paired cell line established from lymph node metastasis of the same patient, expresses high levels of TCF12 [ 5 ]. These experimental results are consistent with our description of the TCF12–MALAT1 alliance. Additionally, MALAT1 contains RNA sequences predicted to form triplex structures with several gene promoter regions, suggesting that MALAT1 may function as a guide RNA or molecular scaffold, facilitating the recruitment of TCF12 to promoter regions of target genes and its interactions with other transcriptional regulatory proteins. Consequently, the TCF12–MALAT1 alliance may produce complex and distinct outcomes that are entirely different from those resulting from the individual activities of TCF12 and MALAT1.
We further explored which downstream factors/events of the TCF12–MALAT1 alliance were responsible for the significant association with patients’ poorer OS outcomes. We performed MGSA signal pathway enrichment analysis on 2312 common target genes of TCF12 and MALAT1 and found that these genes are involved in pathways such as Aurora B, ATM, PLK1, and non-canonical WNT pathways. Given that the WNT–β-catenin pathway is a marker for identifying CMS-2 subtype CRC patients who generally have a relatively better prognosis among all CRC patients [ 39 ], the induction of non-canonical WNT pathways (i.e., β-catenin-independent WNT pathways) by the TCF12–MALAT1 alliance may be related to the poorer prognosis of patients. We analyzed the data from 193 patients in the TCGA-COAD (GA) dataset and found that the TCF12 mRNA level was significantly positively correlated with the levels of MALAT1, β-catenin mRNA, and cyclin D1 mRNA, but not with that of c-Myc mRNA. This lack of relationship between the TCF12 and c-Myc mRNA levels was expected, as both belong to the HLH protein family, which involves complex member-to-member interactions [ 1 , 2 ]. However, an interesting and unexpected finding was that the level of MALAT1 was only correlated with the level of cyclin D1 mRNA, not with β-catenin or c-Myc, and the level of β-catenin mRNA was not correlated with the downstream targets of cyclin D1 or c-Myc mRNA expressions. These results suggest that the high expression of cyclin D1 mRNA associated with the TCF12–MALAT1 alliance may occur through a β-catenin-independent pathway. Additionally, an analysis of patients’ OS rates indicated that patients with high expressions of TCF12 mRNA, MALAT1, and β-catenin mRNA had good prognoses, while those with high expressions of TCF12 mRNA, MALAT1, and cyclin D1 mRNA had poor prognoses. Finally, we found that patients with high expressions of TCF12 mRNA, MALAT1, and cyclin D1 mRNA but low expression of β-catenin mRNA—whom we refer to as having the TMBC pattern—had shorter RFS and OS outcomes. Tumor recurrence, rather than metastasis, may be the primary risk factor contributing to the mortality of this group of patients.
The WNT signaling pathways, essentially involved in numerous biological and pathological processes, are commonly classified into the canonical WNT–β-catenin pathway and the non-canonical β-catenin-independent WNT pathways [ 47 ]. The canonical WNT pathway primarily involves the binding of WNT ligands with Frizzled (FZD) receptors and LRP5/6 co-receptors, which prevents β-catenin from degradation by the APC–Axin–GSK-3β complex. The stabilized β-catenin then translocates into the nucleus and partners with other transcription factors (e.g., TCF/LEF) to turn on downstream gene expressions for cell proliferation, survival, and stem cell maintenance. The non-canonical WNT pathways exhibit signaling cascades without the participation of β-catenin and as yet include two branches: the WNT/planar cell polarity (PCP) pathway and the WNT/Ca 2+ pathway [ 47 ]. In the WNT/PCP pathway, WNT ligands bind to FZD receptors along with ROR1/2 or RYK co-receptors, activating downstream molecules RhoA and Rac1, which further regulate cell polarity, migration, and tissue morphogenesis. The WNT/Ca 2+ pathway involves the binding of WNT to FZD, leading to phospholipase C activation and subsequent Ca 2+ release. Increased Ca 2+ levels activate Ca 2+ -dependent proteins, such as phosphatase calcineurin, calmodulin-dependent protein kinase II, and protein kinase C, which play roles in cell adhesion, migration, and gene regulation. In our present study, the TCF12–MALAT1 alliance is significantly associated with a β-catenin-independent induction of cyclin D1 gene expression, which acts as a risk factor for CRC patient mortality, associated more with tumor recurrence than with metastasis. It needs to be clarified whether there are more interactions between the TCF12–MALAT1 alliance and the WNT pathways. With further research, the findings may potentially represent a new branch of the non-canonical WNT pathways.
Cancer is a complicated disease with diverse causes. Even tumors originating from the same organ can exhibit heterogeneity across different patients due to variations in genetic, epigenetic, and tissue microenvironmental factors. This underlines the need for the development of personalized or precision medicine. In our study, we identified two distinct subgroups within CRC patients. One group, comprising 17.1% of patients, showed high levels of TCF12 mRNA, MALAT1, and β-catenin mRNA in tumors, similar to the CMS-2 subtype, and displayed better overall survival; we termed this the “Pan-CMS-2 pattern” ( Figure 5 ). Another group, encompassing 8.3% of patients, exhibited a TMBC pattern with high expressions of TCF12 mRNA, MALAT1, and cyclin D1 mRNA, but low β-catenin mRNA, and had shorter RFS and OS outcomes regardless of whether metastasis occurs. Tumor recurrence seemed to be a major mortality risk for these CRC patients with the TMBC pattern. Cyclin D1 is a key cell cycle regulatory protein that can form complexes with CDK4 and CDK6, thereby activating their kinase activities and promoting G1 to S phase progression in the cell cycle [ 48 ]. Several drugs targeting cyclin D1-associated CDK4/6 have been developed and used in clinics. Examples include palbociclib (Ibrance; Pfizer Inc., New York, NY, USA), abemaciclib (LY2835219; Eli Lilly & Co., Indianapolis, IN, USA), and ribociclib (LEE011; Novartis International AG, Basel, Switzerland), which are highly specific oral inhibitors of CDK4/6 [ 49 ]. Our study suggests that administering CDK4/6 inhibitors to CRC patients with the TMBC pattern may potentially reduce the risk of early mortality.
There are two limitations of this study. First, the findings were derived from only one dataset, and additional CRC patient datasets are needed to validate the clinical implications of the TMBC pattern. Second, this study lacks investigations into the exact mechanism underlying the simultaneous suppression of β-catenin gene expression and upregulation of cyclin D1 gene expression under the TCF12–MALAT1 alliance. Understanding this mechanism could pave the way for developing preventive strategies. Nonetheless, these limitations do not diminish the potential of the TMBC pattern to serve as a biomarker for evaluating the efficacy of precision therapies using cyclin D1-associated CDK4/6 inhibitors.
5. Conclusions
Metastasis is a critical event that significantly worsens the prognosis of CRC patients and has garnered the attention of CRC researchers worldwide. However, for CRC patients who do not develop metastasis but still experience poor survival outcomes, the underlying causes and mechanisms of mortality remain poorly understood. In this study, we identified a subset (8.3%) of CRC patients who, irrespective of tumor metastasis, have a poorer prognosis due to their tumor tissues exhibiting high TCF12 mRNA, high MALAT1, low β-catenin mRNA, and high cyclin D1 mRNA levels, referred to as the TMBC pattern. For these patients, early tumor recurrence, rather than metastasis, appears to be the primary risk factor for mortality. Cyclin D1 is an important cell cycle regulatory protein responsible for activating CDK4 and CDK6 kinase activities and promoting cancer cell proliferation. A precision medicine administering CDK4/6 inhibitors may potentially reduce the risk of early mortality in CRC patients with the TMBC pattern.
Acknowledgments
We acknowledge the technical assistance from the Bioinformatics Core of National Health Research Institutes, which is funded by the National Science and Technology Council (NSTC113-2740-B-400-005-) and the administrative assistance provided by Kyubi Lin.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.
Supplementary Materials
The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/cells13242035/s1 , Supplementary Figure S1: Validation of MALAT1 as a TCF12-associated lncRNA, Supplementary Figure S2: Cox regression analyses of the association of TCF12–MALAT1 alliance with the RFS of CRC patients, and Supplementary Figure S3: Kaplan–Meier OS curves of non-metastatic or metastatic patients of the TCGA-COAD (GA) dataset, who were further divided into subgroups based on the patients with or without TMBC Pattern.
Author Contributions
Conceptualization, C.-M.W., C.-H.C., K.-W.T., S.-H.C. and T.-S.H.; methodology, C.-M.W., C.-H.C., K.-W.T., M.-C.T., F.-Y.T. and S.-S.J.; software, C.-H.C., K.-W.T., S.-S.J. and T.-S.H.; validation, W.-S.C., H.-D.W. and T.-S.H.; formal analysis, C.-H.C., K.-W.T., S.-S.J. and T.-S.H.; investigation, C.-M.W., C.-H.C., K.-W.T., M.-C.T. and F.-Y.T.; data curation, C.-M.W., C.-H.C., K.-W.T. and T.-S.H.; writing—original draft preparation, C.-M.W. and T.-S.H.; writing—review and editing, T.-S.H.; visualization, C.-M.W., C.-H.C., K.-W.T., M.-C.T., F.-Y.T. and T.-S.H.; supervision, H.-D.W. and T.-S.H.; project administration, T.-S.H.; funding acquisition, T.-S.H. All authors have read and agreed to the published version of the manuscript.
Institutional Review Board Statement
Not applicable. All data of CRC patients analyzed in this study were acquired from the publicized dataset TCGA-COAD of the TCGA Hub in the UCSC Xena website.
Informed Consent Statement
Not applicable.
Data Availability Statement
The RNA sequencing data of anti-TCF12 immunoprecipitates are deposited with an accession No. PRJNA1190159 in the Sequence Read Archive (SRA) database, National Center for Biotechnology Information (NCBI), USA. Other data generated or analyzed during this study are included in this published article and the associated online supplemental file .
Conflicts of Interest
The authors declare no conflicts of interest.
References
1. Zhang Y. Babin J. Feldhaus A.L. Singh H. Sharp P.A. Bina M. HTF4, a new human helix-loop-helix protein Nucleic Acids Res. 1991 19 4555 10.1093/nar/19.16.4555 1886779 PMC328652
2. Hu J.S. Olson E.N. Kingston R.E. HEB, a helix-loop-helix protein related to E2A and ITF2 that can modulate the DNA binding ability of myogenic regulatory factors Mol. Cell Biol. 1992 12 1031 1042 1312219 10.1128/mcb.12.3.1031-1042.1992 PMC369535
3. The ENCODE Project Consortium The ENCODE (ENCyclopedia Of DNA Elements) Project Science 2004 306 636 640 10.1126/science.1105136 15499007
4. The ENCODE Project Consortium A user’s guide to the encyclopedia of DNA elements (ENCODE) PLoS Biol. 2011 9 e1001046 21526222 10.1371/journal.pbio.1001046 PMC3079585
5. Lee C.C. Chen W.S. Chen C.C. Chen L.L. Lin Y.S. Fan C.S. Huang T.S. TCF12 functions as a transcriptional repressor of E-cadherin and its overexpression is correlated with the metastasis of colorectal cancer J. Biol. Chem. 2012 287 2798 2809 10.1074/jbc.M111.258947 22130667 PMC3268437
6. Chen W.S. Chen C.C. Chen L.L. Lee C.C. Huang T.S. Secreted heat shock protein 90α (HSP90α) induces nuclear factor-κB-mediated TCF12 protein expression to down-regulate E-cadherin and to enhance colorectal cancer cell migration and invasion J. Biol. Chem. 2013 288 9001 9010 10.1074/jbc.M112.437897 23386606 PMC3610972
7. Fan C.S. Chen W.S. Chen L.L. Chen C.C. Hsu Y.T. Chua K.V. Wang H.-D. Huang T.-S. Osteopontin–integrin engagement induces HIF-1α–TCF12-mediated endothelial-mesenchymal transition to exacerbate colorectal cancer Oncotarget 2018 9 4998 5015 10.18632/oncotarget.23578 29435158 PMC5797029
8. Tang X. Hou Y. Yang G. Wang X. Tang S. Du Y.E. Yang L. Yu T. Zhang H. Zhou M. Stromal miR-200s contribute to breast cancer cell invasion through CAF activation and ECM remodeling Cell Death Differ. 2016 23 132 145 10.1038/cdd.2015.78 26068592 PMC4815985
9. Phadte P. Bishnu A. Dey P. Mehrotra M. Singh P. Chakrabarty S. Majumdar R. Rekhi B. Patra M. De A. Autophagy-mediated ID1 turnover dictates chemo-resistant fate in ovarian cancer stem cells J. Exp. Clin. Cancer Res. 2024 43 222 10.1186/s13046-024-03147-z 39123206 PMC11316295
10. Wang S. Liao Y. Zhang H. Jiang Y. Peng Z. Ren R. Li X. Wang H. Tcf12 is required to sustain myogenic genes synergism with MyoD by remodelling the chromatin landscape Commun. Biol. 2022 5 1201 10.1038/s42003-022-04176-0 36352000 PMC9646716
11. Sharma V.P. Fenwick A.L. Brockop M.S. McGowan S.J. Goos J.A.C.A. Hoogeboom J.M. Brady A.F. Jeelani N.O. Lynch S.A. Mulliken J.B. Mutations in TCF12 , encoding a basic helix-loop-helix partner of TWIST1, are a frequent cause of coronal craniosynostosis Nat. Genet. 2013 45 304 307 10.1038/ng.2531 23354436 PMC3647333
12. Li Y. Yang J. Luo J.H. Dedhar S. Liu Y. Tubular epithelial cell dedifferentiation is driven by the helix-loop-helix transcriptional inhibitor Id1 J. Am. Soc. Nephrol. 2007 18 449 460 10.1681/ASN.2006030236 17202424
13. Lu F. Song Y. Cui S. Zhao H. Chen Y. Du H. LncRNA MIAT promotes the proliferation, migration, and invasion of melanoma cells through recruiting TCF12 and activating NFAT5 Am. J. Transl. Res. 2021 13 12588 12600 34956475 PMC8661192
14. He J. Shen S. Lu W. Zhou Y. Hou Y. Zhang Y. Jiang Y. Liu H. Shao Y. HDAC1 promoted migration and invasion binding with TCF12 by promoting EMT progress in gallbladder cancer Oncotarget 2016 7 32754 32764 10.18632/oncotarget.8740 27092878 PMC5078048
15. Yang J. Zhang L. Jiang Z. Ge C. Zhao F. Jiang J. Tian H. Chen T. Xie H. Cui Y. TCF12 promotes the tumorigenesis and metastasis of hepatocellular carcinoma via upregulation of CXCR4 expression Theranostics 2019 9 5810 5827 10.7150/thno.34973 31534521 PMC6735379
16. Zhou W. Deng X. Wang B. Yuan Y. Ma J. Meng X. HTF4 modulates the transcription of GID2 to promote the malignant biological behavior of pancreatic cancer Pancreatology 2024 24 1073 1083 10.1016/j.pan.2024.08.008 39216997
17. Yoshimoto R. Mayeda A. Yoshida M. Nakagawa S. MALAT1 long non-coding RNA in cancer Biochim. Biophys. Acta 2016 1859 192 199 10.1016/j.bbagrm.2015.09.012 26434412
18. Amodio N. Raimondi L. Juli G. Stamato M.A. Caracciolo D. Tagliaferri P. Tassone P. MALAT1: A druggable long non-coding RNA for targeted anti-cancer approaches J. Hematol. Oncol. 2018 11 63 10.1186/s13045-018-0606-4 29739426 PMC5941496
19. Ji P. Diederichs S. Wang W. Boing S. Metzger R. Schneider P.M. Tidow N. Brandt B. Buerger H. Bulk E. MALAT-1, a novel noncoding RNA, and thymosin β4 predict metastasis and survival in early-stage non-small cell lung cancer Oncogene 2003 22 8031 8041 10.1038/sj.onc.1206928 12970751
20. Anko M.L. Neugebauer K.M. Long noncoding RNAs add another layer to pre-mRNA splicing regulation Mol. Cell 2010 39 833 834 10.1016/j.molcel.2010.09.003 20864030
21. Tripathi V. Ellis J.D. Shen Z. Song D.Y. Pan Q. Watt A.T. Freier S.M. Bennett C.F. Sharma A. Bubulya P.A. The nuclear-retained noncoding RNA MALAT1 regulates alternative splicing by modulating SR splicing factor phosphorylation Mol. Cell 2010 39 925 938 10.1016/j.molcel.2010.08.011 20797886 PMC4158944
22. Lin R. Roychowdhury-Saha M. Black C. Watt A.T. Marcusson E.G. Freier S.M. Edgington T.S. Control of RNA processing by a large non-coding RNA over-expressed in carcinomas FEBS Lett. 2011 585 671 676 10.1016/j.febslet.2011.01.030 21266177 PMC3065235
23. Fan Y. Shen B. Tan M. Mu X. Qin Y. Zhang F. Liu Y. TGF-β-induced upregulation of malat1 promotes bladder cancer metastasis by associating with suz12 Clin. Cancer Res. 2014 20 1531 1541 10.1158/1078-0432.CCR-13-1455 24449823
24. Wang W. Zhu Y. Li S. Chen X. Jiang G. Shen Z. Qiao Y. Wang L. Zheng P. Zhang Y. Long noncoding RNA MALAT1 promotes malignant development of esophageal squamous cell carcinoma by targeting β-catenin via Ezh2 Oncotarget 2016 7 25668 25682 10.18632/oncotarget.8257 27015363 PMC5041935
25. Yang L. Lin C. Liu W. Zhang J. Ohgi K.A. Grinstein J.D. Dorrestein P.C. Rosenfeld M.G. ncRNA- and Pc2 methylation-dependent gene relocation between nuclear structures mediates gene activation programs Cell 2011 147 773 788 10.1016/j.cell.2011.08.054 22078878 PMC3297197
26. Li B. Chen P. Qu J. Shi L. Zhuang W. Fu J. Li J. Zhang X. Sun Y. Zhuang W. Activation of LTBP3 gene by a long noncoding RNA (lncRNA) MALAT1 transcript in mesenchymal stem cells from multiple myeloma J. Biol. Chem. 2014 289 29365 29375 10.1074/jbc.M114.572693 25187517 PMC4200285
27. Hirata H. Hinoda Y. Shahryari V. Deng G. Nakajima K. Tabatabai Z.L. Ishii N. Dahiya R. Long noncoding RNA MALAT1 promotes aggressive renal cell carcinoma through Ezh2 and interacts with miR-205 Cancer Res. 2015 75 1322 1331 10.1158/0008-5472.CAN-14-2931 25600645 PMC5884967
28. Luan W. Li L. Shi Y. Bu X. Xia Y. Wang J. Djangmah H.S. Liu X. You Y. Xu B. Long non-coding RNA MALAT1 acts as a competing endogenous RNA to promote malignant melanoma growth and metastasis by sponging miR-22 Oncotarget 2016 7 63901 63912 10.18632/oncotarget.11564 27564100 PMC5325412
29. Chang S.M. Hu W.W. Long non-coding RNA MALAT1 promotes oral squamous cell carcinoma development via microRNA-125b/STAT3 axis J. Cell. Physiol. 2018 233 3384 3396 10.1002/jcp.26185 28926115
30. Luo J.H. Ren B. Keryanov S. Tseng G.C. Rao U.N. Monga S.P. Strom S. Demetris A.J. Nalesnik M. Yu Y.P. Transcriptomic and genomic analysis of human hepatocellular carcinomas and hepatoblastomas Hepatology 2006 44 1012 1024 10.1002/hep.21328 17006932 PMC1769554
31. Guffanti A. Iacono M. Pelucchi P. Kim N. Solda G. Croft L.J. Taft R.J. Rizzi E. Askarian-Amiri M. Bonnal R.J. A transcriptional sketch of a primary human breast cancer by 454 deep sequencing BMC Genom. 2009 10 163 10.1186/1471-2164-10-163 PMC2678161 19379481
32. Xu C. Yang M. Tian J. Wang X. Li Z. MALAT-1: A long non-coding RNA and its important 3′ end functional motif in colorectal cancer metastasis Int. J. Oncol. 2011 39 169 175 21503572 10.3892/ijo.2011.1007
33. Ren S. Liu Y. Xu W. Sun Y. Lu J. Wang F. Wei M. Shen J. Hou J. Gao X. Long noncoding RNA MALAT-1 is a new potential therapeutic target for castration resistant prostate cancer J. Urol. 2013 190 2278 2287 10.1016/j.juro.2013.07.001 23845456
34. Okugawa Y. Toiyama Y. Hur K. Toden S. Saigusa S. Tanaka K. Inoue Y. Mohri Y. Kusunoki M. Boland C.R. Metastasis-associated long non-coding RNA drives gastric cancer development and promotes peritoneal metastasis Carcinogenesis 2014 35 2731 2739 10.1093/carcin/bgu200 25280565 PMC4247518
35. Cho S.F. Chang Y.C. Chang C.S. Lin S.F. Liu Y.C. Hsiao H.H. Chang J.G. Liu T.C. MALAT1 long non-coding RNA is overexpressed in multiple myeloma and may serve as a marker to predict disease progression BMC Cancer 2014 14 809 10.1186/1471-2407-14-809 25369863 PMC4233101
36. Hu L. Wu Y. Tan D. Meng H. Wang K. Bai Y. Yang K. Up-regulation of long non-coding RNA MALAT1 contributes to proliferation and metastasis in esophageal squamous cell carcinoma J. Exp. Clin. Cancer Res. 2015 34 7 10.1186/s13046-015-0123-z 25613496 PMC4322446
37. Goldman M.J. Craft B. Hastie M. Repečka K. McDade F. Kamath A. Banerjee A. Luo Y. Rogers D. Brooks A.N. Visualizing and interpreting cancer genomics data via the Xena platform Nat. Biotechnol. 2020 38 675 678 10.1038/s41587-020-0546-8 32444850 PMC7386072
38. Yoshihara K. Shahmoradgoli M. Martínez E. Vegesna R. Kim H. Torres-Garcia W. Treviño V. Shen H. Laird P.W. Levine D.A. Inferring tumour purity and stromal and immune cell admixture from expression data Nat. Commun. 2013 4 2612 10.1038/ncomms3612 24113773 PMC3826632
39. Guinney J. Dienstmann R. Wang X. de Reyniès A. Schlicker A. Soneson C. Marisa L. Roepman P. Nyamundanda G. Angelino P. The consensus molecular subtypes of colorectal cancer Nat. Med. 2015 21 1350 1356 10.1038/nm.3967 26457759 PMC4636487
40. Merry C.R. Forrest M.E. Sabers J.N. Beard L. Gao X.H. Hatzoglou M. Jackson M.W. Wang Z. Markowitz S.D. Khalil A.M. DNMT1-associated long non-coding RNAs regulate global gene expression and DNA methylation in colon cancer Hum. Mol. Genet. 2015 24 6240 6253 10.1093/hmg/ddv343 26307088 PMC4599679
41. Cloonan N. Wani S. Xu Q. Gu J. Lea K. Heater S. Barbacioru C. Steptoe A.L. Martin H.C. Nourbakhsh E. MicroRNAs and their isomiRs function cooperatively to target common biological pathways Genome Biol. 2011 12 R126 10.1186/gb-2011-12-12-r126 22208850 PMC3334621
42. John-Aryankalayil M. Palayoor S.T. Makinde A.Y. Cerna D. Simone C.B. II Falduto M.T. Magnuson S.R. Coleman C.N. Fractionated radiation alters oncomir and tumor suppressor miRNAs in human prostate cancer cells Radiat. Res. 2012 178 105 117 10.1667/RR2703.1 22827214 PMC4156966
43. Langmead B. Salzberg S. Fast gapped-read alignment with Bowtie 2 Nat. Methods 2012 9 357 359 10.1038/nmeth.1923 22388286 PMC3322381
44. Ji H. Jiang H. Ma W. Johnson D.S. Myers R.M. Wong W.H. An integrated software system for analyzing ChIP-chip and ChIP-seq data Nat. Biotechnol. 2008 26 1293 1300 10.1038/nbt.1505 18978777 PMC2596672
45. Chandrashekar D.S. Karthikeyan S.K. Korla P.K. Patel H. Shovon A.R. Athar M. Netto G.J. Qin Z.S. Kumar S. Manne U. UALCAN: An update to the intergrated cancer data analysis platform Neoplasia 2022 25 18 27 10.1016/j.neo.2022.01.001 35078134 PMC8788199
46. Bauer S. Robinson P.N. Gagneur J. Model-based gene set analysis for Bioconductor Bioinformatics 2011 27 1882 1883 10.1093/bioinformatics/btr296 21561920 PMC3117381
47. Zhu Y. Li X. Advances of Wnt signalling pathway in colorectal cancer Cells 2023 12 447 10.3390/cells12030447 36766788 PMC9913588
48. Fu M. Wang C. Li Z. Sakamaki T. Pestell R.G. Minireview: Cyclin D1: Normal and abnormal functions Endocrinology 2004 145 5439 5447 10.1210/en.2004-0959 15331580
49. VanArsdale T. Boshoff C. Arndt K.T. Abraham R.T. Molecular pathways: Targeting the cyclin D-CDK4/6 axis for cancer treatment Clin. Cancer Res. 2015 21 2905 2910 10.1158/1078-0432.CCR-14-0816 25941111