Long noncoding RNAs (lncRNA) are involved in tumorigenesis and drug resistance. However, the roles and underlying mechanisms of lncRNAs in colorectal cancer are still unknown. In this work, through transcriptomic profiling analysis of 21 paired tumor and normal samples, we identified a novel colorectal cancer–related lncRNA, MNX1-AS1. MNX1-AS1 expression was significantly upregulated in colorectal cancer and associated with poor prognosis. In vitro and in vivo gain- and loss-of-function experiments showed that MNX1-AS1 promotes the proliferation of colorectal cancer cells. MNX1-AS1 bound to and activated Y-box-binding protein 1 (YB1), a multifunctional RNA/DNA-binding protein, and prevented its ubiquitination and degradation. A marked overlap between genes that are differentially expressed in MNX1-AS1 knockdown cells and transcriptional targets of YB1 was observed. YB1 knockdown mimicked the loss of viability phenotype observed upon depletion of MNX1-AS1. In addition, MYC bound the promoter of the MNX1-AS1 locus and activated its transcription. In vivo experiments showed that ASO inhibited MNX1-AS1, which suppressed the proliferation of colorectal cancer cells in both cell-based and patient-derived xenograft models. Collectively, these findings suggest that the MYC–MNX1-AS1–YB1 axis might serve as a potential biomarker and therapeutic target in colorectal cancer.
This study highlights the discovery of a novel colorectal cancer biomarker and therapeutic target, MNX1-AS1, a long noncoding RNA that drives proliferation via a MYC/MNX1-AS1/YB1 signaling pathway.
Colorectal cancer is a heterogeneous and molecularly complicated disease (1, 2). Patients with colorectal cancer generally receive stratified treatment options; however, the treatment options are limited and lead to diverse clinical outcomes (1–3). In the past few years, diagnostic tumor biology has improved and advanced the effectiveness of treatments (1, 2, 4). Recently, molecular stratification has been implemented in clinical practice to identify subgroups who may benefit from novel treatments, especially among patients with metastatic colorectal cancer who lack treatment options (1, 2, 4). Patients suffering from metastatic colorectal cancer are typically offered chemotherapy (fluoropyrimidines plus either oxaliplatin or irinotecan) sometimes in combination with antibodies targeting VEGF (bevacizumab or regorafenib) and, if they lack mutations in RAS, EGFR (cetuximab or panitumumab), which leads to improved overall survival (OS; refs. 1, 2, 4).
Long noncoding RNAs (lncRNA) play a pivotal role in mediating crosstalk between various cellular components, including proteins, RNAs, and DNAs, that are involved in signal transduction (5–7). Numerous studies have described the importance of lncRNAs in tumorigenesis, such as initiation and progression (8, 9). For example, lncRNA EPIC1 interacts with MYC and promotes cancer progression (9). The identification of specific lncRNAs involved in cancer progression has led to innovative clinical applications, such as diagnostic biomarkers, prognostic indicators, stratification standards, drug sensitizers, and therapeutic targets (10–12). In colorectal cancer, lncRNAs are also associated with tumor progression and drug resistance and could be promising therapeutic targets (13–16). For instance, lncRNA N-BLR leads to colorectal cancer invasion and migration (13). SNHG5 promotes colorectal cancer cell survival (14). MIR100HG mediates colorectal cancer cetuximab resistance, and CCAL promotes colorectal cancer progression and induces multidrug resistance (MDR; refs. 15, 16). Novel therapeutic strategies that target lncRNAs, including antisense oligonucleotides (ASO) that downregulate the transcription of lncRNAs via RNase H-dependent degradation, LNAs that can lock nucleic acids (17), and small molecule inhibitors that interfere with lncRNA–protein interactions (18) have also been developed. Targeting lncRNAs using an siRNA-based strategy, including ASOs or LNAs, has been successful in several preclinical models (7, 19, 20). Clinical trials using LNAs targeting the oncoprotein Bcl-2 (21), HIF1α (NCT00466583), and AR (22) have shown promising preliminary results, prompting us to search for novel lncRNAs associated with colorectal cancer progression that could serve as therapeutic targets (23).
We attempted to identify novel lncRNAs with significantly increased expression and positive correlations with clinical outcomes in metastatic colorectal cancer; patients with metastatic colorectal cancer miss the opportunity for radical tumor resection and depend only on chemotherapy and molecular targeted therapy (24). We integrated RNA-sequencing (RNA-seq) and bioinformatics analyses and identified a novel lncRNA, MNX1-AS1, that may contribute to the oncogenesis and progression of colorectal cancer. In this study, we found that MNX1-AS1 promotes colorectal cancer proliferation by accelerating the cell cycle in vitro and in vivo, and this effect partially depends on inhibiting the ubiquitination-mediated degradation of YB1. We also demonstrated that MYC binds to the MNX1-AS1 promoter and promotes its transcription, indicating that MNX1-AS1 is a transcriptional target of the oncoprotein MYC. Our study revealed that MNX1-AS1 may be a promising biomarker and therapeutic target for colorectal cancer treatment.
Materials and Methods
Human tissue specimens
Clinical samples were collected from Sun Yat-sen University Cancer Center (SYSUCC; Guangzhou, China). All the patients were histologically diagnosed with colorectal cancer. Written informed consent was obtained from all the patients, and the patients received regular follow-up. All the clinicopathologic information is provided in Supplementary Tables S1 and S4. The study was approved by the Medical Ethics Committee of SYSUCC.
In vivo xenograft study (in situ tumor proliferation study)
A stable MNX1-AS1-knockdown cell line that also stably expressed luciferase was constructed, and, with the corresponding negative control cell lines, was injected subcutaneously into the dorsal flanks of 4-week-old female BALB/c nu/nu mice. When the maximum diameters of the tumors were approximately 1 cm, the tumors were removed and cut into 1 mm × 1 mm tissue blocks. The tissue blocks were read for use. The nude mice were anesthetized with isoflurane for the operation. A small incision was made at the midline of the abdomen through the skin, abdominal muscles, and momentum majus to locate the cecum. The spare tumor tissue block was sewn on the serous surface of the cecum with rich blood supply. Primary fluorescence imaging was performed immediately after the operation to determine the tumor reference size. After the operation, the tumor growth was observed by fluorescence imaging every 2 weeks, and the mice were sacrificed at the appropriate time. The volumes and weights of the tumors were recorded in situ.
MNX1-AS1 RNA copy number analysis
Total RNA was isolated from 1 × 106 cells using a RNeasy Mini Kit (Qiagen, #74104). Full-length MNX1-AS1 RNA was transcribed in vitro with the MEGAscript T7 Transcription Kit (Invitrogen, AM1334), T7 RNA polymerase (Roche, #10881775001), and RNase-free DNase I (Promega, #M198A). cDNA was synthesized using 1 μg of total RNA or full-length MNX1-AS1 RNA. Serial 10-fold dilutions (105–1011 molecules per test) of cDNA from in vitro–transcribed MNX1-AS1 RNA were used as reference molecules to calculate the standard curve. Real-time PCR was performed as described previously (25).
RNA pull-down assay
Biotin-labeled full-length and truncated fragments of MNX1-AS1 RNA were transcribed in vitro with a Pierce RNA 3′ End Desthiobiotinylation Kit (Thermo Fisher Scientific, #20163) and T4 RNA ligase (Invitrogen, MEGAscipt, #AM1330) using PCR products as a template. Cells were collected by trypsinization, washed twice with sterilized PBS, and prepared using Pierce IP lysis buffer (Thermo Fisher Scientific). RNA pull-down assays were performed with a Pierce Magnetic RNA–Protein Pull-Down Kit (Thermo Fisher Scientific, #20164). According to the manufacturer's instructions, biotinylated RNA was captured with streptavidin magnetic beads and then incubated with the cell lysates or purified protein (20 μg) at 4°C for 6 hours before washing and elution of the RBP complex. The eluted proteins were subjected to MS analysis or Western blotting. The in vitro binding assay of biotin-labeled MNX1-AS1 RNA was performed as described previously (25).
RNA immunoprecipitation (RIP) assays were performed with a Magna RNA-Binding Protein Immunoprecipitation Kit (Millipore), according to the manufacturer's instructions. The cells were subsequently washed twice with cold PBS, and the pellets were resuspended in RIP lysis buffer on ice for 15 minutes before use. Negative control IgG, human anti-YB1 antibody (1:20, Abcam), and anti-FLAG tag antibody (1:20, Sigma) were used in this study. After proteinase K digestion, the immunoprecipitated RNAs were extracted, purified, and subjected to qPCR. The RNA levels were normalized to the input (10%).
MS2-tagged RNA affinity purification
Colorectal cancer cells RKO and SW480 were cotransfected with the pcDNA3.1-MS2 or pcDNA3.1-MS2-MNX1-AS1 and MCP-3FLAG plasmids (OBiOTechnology). After 48 hours, the live cells were irradiated with 254 nm UV light at 400 mJ/cm2. Then, the cells were lysed for 10 minutes on ice and centrifuged at 13,000 × g for 10 minutes. The FLAG-tagged proteins were immunoprecipitated with anti-FLAG M2 affinity gel (Sigma-Aldrich). After three washes with low-salt wash buffer, the agarose gels were boiled in loading buffer, and the proteins were detected by Western blotting analysis.
In vivo therapeutic study
To generate patient-derived xenograft (PDX) models, fresh colorectal cancer tumor samples from patients were immediately subcutaneously inoculated into both flanks of nude mice. When the successfully established PDXs (P1) reached ∼500 mm3, the tumors were transplanted into other mice (P2). Eventually, mice bearing P3 grafts were used to examine the therapeutic effects of the ASO MNX1-AS1 inhibitor (5 nmol per injection; RiboBio) and/or oxaliplatin every 3 days. RKO colorectal cancer cells (2 × 106) were subcutaneously injected into the dorsal flanks of 4-week-old female BALB/c nu/nu mice (three mice per group) to generate cell-based models (3 mice/group each) along with PDX-based models (4 mice/group each). Tumor-bearing mice were randomly assigned to four groups: (i) the control group, which received 200 μL of PBS every 3 days; (ii) the oxaliplatin group, which received 10 nmol/kg oxaliplatin every 3 days by intraperitoneal injection; (iii) the ASO MNX1-AS1 inhibitor group, which received 5 nmol ASO per injection every 3 days by intratumoral injection; and (iv) the combination group. The animals were treated for 1 month. All the tissues from the cell-based xenografts or PDXs were subjected to further pathologic analysis. The animal studies were conducted with the approval of the Sun Yat-sen University Institutional Animal Care and Use Committee.
All the data are presented as the mean ± SD. Student paired or unpaired t tests and chi-square tests were used for the comparison of significant differences between groups with GraphPad Prism software. Correlations between the levels of MNX1-AS1 and the expression of target genes downstream of YB1 were analyzed with the Pearson correlation analysis. Survival analyses were performed using the Kaplan–Meier method and assessed using the log-rank test with SPSS software. The levels of significance were set to *, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001.
The study was approved by the Institutional Review Board of SYSUCC. Subjects provided written informed consent. The animal study was approved by the Institutional Animal Care and Use Committee of Sun Yat-sen University.
RNA-seq and bioinformatics analyses identified specific lncRNAs in colorectal cancer
Because lncRNAs play an important role in many types of cancer, including colorectal cancer and esophageal squamous cell carcinoma (ESCC; refs. 25, 26), we aimed to identify tumor-promoting lncRNAs in colorectal cancer. Twenty-one pairs of tumor tissues and adjacent normal tissues from patients with stage IV colorectal cancer were subjected to transcriptome/RNA-seq analyses (Supplementary Table S1). As a result, 2,746 lncRNA genes were identified, and of these genes, 713 were upregulated and 588 were downregulated in the cancer tissues (Fig. 1A). On the basis of the P values and the base mean of these genes, we chose the top 30 candidates for further confirmation (Supplementary Table S2). We validated the RNA-seq results by performing RT-PCR on the same cohort of colorectal cancer and normal colorectal tissues that were used for RNA-seq. Of the 30 lncRNAs, 8 candidates were confirmed to be more highly expressed in colorectal cancer tissues than in normal colorectal tissues (Supplementary Table S3). Next, we used an independent cohort of 96 colorectal cancer tissues and paired normal colorectal tissues for further validation, and seven lncRNAs were still significantly upregulated in the cancer tissues (Supplementary Fig. S1A). Furthermore, we used The Cancer Genome Atlas (TCGA) database to determine whether the expression of these lncRNAs was associated with the prognosis of patients with colorectal cancer. Among these seven lncRNAs, high MNX1-AS1 expression was significantly associated with poor outcome in patients with colorectal cancer, suggesting that its oncogenic potential is worth further study (Fig. 1B).
MNX1-AS1 is upregulated in colorectal cancer and correlated with poor prognosis
MNX1-AS1 is an antisense lncRNA located on chr7: q36.3 (Fig. 1C; Supplementary Fig. S1B). Analysis using the TCGA RNA-seq dataset showed that lncRNA ENSG00000243479 (MNX1-AS1) was upregulated in multiple types of cancer, including colorectal cancer (Fig. 1D). Because MNX1-AS1 is an antisense lncRNA, we wondered whether the depletion of MNX1-AS1 affects the transcription of its neighboring gene MNX1. We analyzed the expression of MNX1 after knocking down MNX1-AS1 by shRNA transfection, and MNX1 showed little change after MNX1-AS1 was silenced (Supplementary Fig. S1C).
To explore the predictive values of MNX1-AS1 in colorectal cancer, we compared the predictive value of the MNX1-AS1, KRAS, BRAF, and ERBB2 mRNA expression levels for prognosis using the TCGA-COAD dataset, and the results showed that the prognostic efficacy of MNX1-AS1 was similar to that of BRAF (Fig. 1E). Consistent with the TCGA dataset analysis (Fig. 1D), we found that MNX1-AS1 was overexpressed in colorectal cancer tissues compared with normal colorectal tissues from a cohort of 150 patients with complete follow-up clinical information (Fig. 1F). In this cohort, the patients with high MNX1-AS1 levels showed significantly worse OS (Fig. 1G; SYSUCC, n = 150; the clinicopathologic information is provided in Supplementary Table S4). The correlation between MNX1-AS1 expression and OS remained significant after adjusting for other prognostic factors, including age and clinical stage (Supplementary Table S5).
MNX1-AS1 functions as a potential oncogenic lncRNA by promoting cell-cycle progression
To evaluate the role of MNX1-AS1 in colorectal cancer, we analyzed the MNX1-AS1 expression levels in a panel of human colorectal cancer cells and normal colorectal epithelial cells (CCD112, CCD841, and NCM460) using RT-PCR. Consistent with the upregulation of MNX1-AS1 in colorectal cancer tissues, the MNX1-AS1 levels were significantly higher in tumor cells than in normal cells (Supplementary Fig. S1D). The copy number of MNX1-AS1 was also analyzed in colorectal cancer cells, and MNX1-AS1 was expressed at higher levels in colorectal cancer cells than in normal colorectal cells CCD112 and CCD481 (Fig. 1H; Supplementary Fig. S1E). Moreover, cell fractionation PCR and subcellular RNA analyses revealed that MNX1-AS1 was located both in the cytoplasm and the nucleus, and this result was further verified by RNA fluorescence in situ hybridization (FISH) assays (Fig. 1I and J; Supplementary Fig. S1F).
We selected two shRNA sequences that target MNX1-AS1 and used a lentiviral system to stably knockdown MNX1-AS1 expression (Supplementary Fig. S2A). In addition, stable MNX1-AS1–overexpressing RKO and SW480 cell lines were established (Supplementary Fig. S2B). MNX1-AS1 knockdown resulted in a decrease in the proliferation of HCT116 and RKO colorectal cancer cells (Fig. 2A). Colony formation assays further demonstrated that MNX1-AS1 knockdown significantly inhibited the anchorage-independent growth of cancer cells (Fig. 2B). Furthermore, RT-PCR showed significantly reduced mRNA transcription of a panel of stemness-associated genes (NANOG, OCT-4, BMI-1, NOTCH-1, and SMO) and cancer stem cell-associated surface antigens (CD24, CD44, and CD133; Supplementary Fig. S2C). Moreover, cell-cycle analysis revealed that silencing MNX1-AS1 resulted in G1–S arrest in HCT116 and RKO cells (Fig. 2C). Annexin-V/PI double staining in both HCT116 and RKO cells demonstrated that knockdown of MNX1-AS1 induced cell apoptosis (Supplementary Fig. S2D). Overexpression of MNX1-AS1 significantly accelerated cell proliferation (Supplementary Fig. S2E), anchorage-independent growth (Supplementary Fig. S2F), and G1–S transition (Supplementary Fig. S2G).
To examine whether cell proliferation was also accelerated by MNX1-AS1 in vivo, we subcutaneously injected MNX1-AS1 (sh#1 and sh#2)-knockdown or shCtrl-transduced HCT116 and RKO cells into nude mouse dorsal flanks and implanted tumor tissue blocks on the serous surface of the cecum (Fig. 2D). Compared with the shCtrl-transduced cells, the sh-MNX1-AS1–transduced HCT116 and RKO cells generated smaller and lighter orthotopic tumors in the cecum (Fig. 2E–G). Pathologic analysis with hemotoxin and eosin (H&E) staining showed more necrosis and fibrosis in the shMNX1-AS1 groups (Fig. 2H). Furthermore, we performed IHC staining for Ki67 to determine the tumor proliferation index; the results revealed that knockdown of MNX1-AS1 could suppress the growth of colorectal cancer (Supplementary Fig. S2H and S2I). Collectively, these results suggest the oncogenic activity of MNX1-AS1 and that MNX1-AS1 might serve as a potential therapeutic target in colorectal cancer treatment.
MNX1-AS1 is directly associated with YB1
Accumulating data have suggested that key signaling mediators, such as receptors, protein kinases, and transcription factors, directly associate with lncRNAs and are directly regulated by the lncRNAs to which they bind (5). To study the potential molecular mechanism by which MNX1-AS1 promoted the proliferation of colorectal cancer cells, we performed RNA pull-down assays and subsequent mass spectrometry (MS) analysis to explore the proteins that potentially bind to MNX1-AS1 (Fig. 3A). By comparing the proteins pulled down with MNX1-AS1 in RKO cells with those in control cells, we identified 34 candidate proteins (Supplementary Table S6). On the basis of the protein score, which incorporates protein coverage and the exponentially modified protein abundance index (emPAI, see Materials and Methods for details), YB1 was the most enriched protein that could be precipitated by in vitro–transcribed biotinylated MNX1-AS1 (Supplementary Table S6).
We then performed RIP and MS2-tagged RNA affinity purification (MTRAP) to confirm the direct interaction between endogenous MNX1-AS1 and the YB1 protein. In RKO, HCT116, and SW480 cells, compared with the negative control, YB1 could significantly enrich MNX1-AS1 in vitro (Fig. 3B), as shown by RIP, and coexpression of the MS2-MNX1-AS1 and MCP-3FLAG plasmids also led to significant enrichment of YB1 in vivo, demonstrating that YB1 specifically bound to MNX1-AS1 (Fig. 3C). To map the MNX1-AS1 function responsible for YB1 binding, we conducted in vitro RNA pull-down assays using a series of truncated MNX1-AS1 fragments according to their predicted secondary structures calculated by RNA fold software (Supplementary Fig. S3A). This analysis revealed that nucleotides 391–580 and 581–992 of MNX1-AS1 (MNX1-AS1 391–580 nt and 581–992 nt) were both sufficient to enrich YB1 (Fig. 3D). In RKO and SW480 cells, 581–992 nt was the YB1-interacting region, whereas in HCT116 cells, both 391–580 nt and 581–992 nt were the YB1-interacting regions, although the interaction predominantly occurred within the 391–580 nt region (Fig. 3D). The 1–390 nt of MNX1-AS1 could not bind to YB1 in RKO, SW480, and HCT116 cells (Fig. 3D). Deletion of the 1–580 nt region abolished MNX1-AS1's interaction with the YB1 protein in RKO and SW480 cells, and consistent with the fact that HCT116 cells contained two MNX1-AS1–interacting regions (391–580 nt and 581–992 nt), deletion of the 1–580 nt region did not completely abolish MNX1-AS1's interaction with YB1 in HCT116 cells (Fig. 3E). These data suggested that the MNX1-AS1 391–580 nt and 581–992 nt regions were both necessary for MNX1-AS1 to bind to the YB1 protein in colorectal cancer cells.
Interestingly, the YB1 protein was obviously downregulated with MNX1-AS1 knockdown in RKO and SW480 cells (Fig. 3F). YB1 was upregulated in colorectal cancer tissues compared with normal colorectal tissues (Supplementary Fig. S3B), and the expression of YB1 and MNX1-AS1 was positively correlated (Supplementary Fig. S3C).
To further explore the function of YB1 in colorectal cancer cells, we knocked down YB1 using siRNAs (Supplementary Fig. S3D), which resulted in decreased cell proliferation, anchorage-independent growth, and cell-cycle arrest in SW480 and RKO cells (Supplementary Fig. S3E–S3G). To further study the roles of the MNX1-AS1–YB1 regulatory axis in colorectal cancer, we knocked down YB1 in RKO, HCT116, and SW480 cells that stably overexpressed MNX1-AS1 and observed that the MNX1-AS1-induced cell proliferation was attenuated by YB1 knockdown (Fig. 3G and H; Supplementary Fig. S3H). These results suggested that MNX1-AS1 and YB1 are closely related and that their interaction plays an important role in colorectal cancer development.
MNX1-AS1 stabilizes YB1 by preventing YB1 ubiquitination
LncRNAs play a role in regulating YB1 expression (27). For example, lncRNA MIR22GH was shown to prevent the proteasomal degradation of YB1 in lung cancer cells, thus contributing to YB1 overexpression (28). Posttranslational modifications, including ubiquitination, may also play an important role in YB1 expression (29). First, we treated MNX1-AS1-knockdown cells with the proteasome inhibitor MG132 and observed increased levels of endogenous YB1 protein compared with those observed in the control-treated cells (Fig. 4A). In contrast, compared with the control, treatment of MNX1-AS1-knockdown cells with the protein synthesis inhibitor cycloheximide resulted in a notably shorter half-life of YB1 (Fig. 4B). These results suggested that MNX1-AS1 may play a critical role in the ubiquitin–proteasome-mediated degradation of the YB1 protein. Indeed, the ubiquitination of YB1 was dramatically increased in cells in which MNX1-AS1 was knocked down compared with control cells (Fig. 4C; Supplementary Fig. S4A). Collectively, these results indicate that the interaction of MNX1-AS1 with YB1 prevents the ubiquitination and degradation of YB1.
Fourteen ubiquitination sites of YB1 were reported from the CPLM database (30)—10 of which were mainly located in the N-terminus (Fig. 4D). According to this prediction, FL (1–520 nt), N-terminus (N, 1–245 nt), and C-terminus (C, 246–520 nt) constructs were transfected into colorectal cancer cells. As expected, the IP assays conducted with an anti-FLAG antibody in MNX1-AS1-knockdown cells expressing FLAG-tagged N-terminus or C-terminus YB1 yielded positive ubiquitin staining, and the FLAG-tagged N-terminus pulled down significantly more ubiquitin molecules (Fig. 4E; Supplementary Fig. S4B). RIP confirmed that MNX1-AS1 precipitated with the N-terminus of YB1 (Fig. 4F; Supplementary Fig. S4C). These data suggest that MNX1-AS1 is negatively related to YB1 ubiquitination.
We further identified the lysine residues in YB1 predominantly subjected to ubiquitination, which may be affected by MNX1-AS1. We focused on 10 ubiquitinated lysine (K) residues in the N-terminus of YB1, including K26, K52, K53, K58, K64, K81, K92, K93, K118, and K137 (Fig. 4D). We mutated each of these 10 lysine (K) residues to arginine (R) and performed IP and RNA pull-down assays. Among these 10 mutants, YB1 K52R showed no increase in ubiquitination in response to MNX1-AS1 knockdown in RKO cells (Fig. 4G), and an RIP assay confirmed that MNX1-AS1 failed to precipitate with the mutant YB1 K52R (Fig. 4H). Unlike RKO, we demonstrated that MNX1-AS1 can't bind to the YB1 K53R and K93R mutants in HCT116 cells or to the YB1 K58R, K81R, and K92R mutants in SW480 cells (Supplementary Fig. S4D and S4E). Taken together, these results suggest that MNX1-AS1 might stabilize YB1 by preventing the ubiquitination of multiple residues through direct binding in colorectal cancer cells.
MNX1-AS1–mediated regulation of YB1 partially depends on downstream targets of MYC related to cell cycle
Although not extensively explored, it has been suggested that lncRNAs are essential mediators of intracellular signaling pathways (5). The lncRNA MAYA, via its interaction with the scaffold protein LLGL2 and the methyltransferase NSUN6, forms an RNA–protein complex for the methylation of MST1, which is a master regulator of the Hippo–YAP pathway (31). Another lncRNA, AK023948, positively regulates the AKT pathway in breast cancer cells through DHX9 and p85 (32). Therefore, we aimed to explore the signaling pathways downstream of MNX1-AS1 by analyzing RNA-seq data from RKO cells transduced with shRNAs targeting MNX1-AS1. To exclude the possibility of off-target effects on gene expression associated with single shRNAs, we only focused on genes regulated in the same direction in all three independent transfection experiments. The knockdown of MNX1-AS1 in RKO cells resulted in the dysregulation of 1,168 genes (upregulation of 1,036 genes and downregulation of 132 genes; Fig. 5A). Gene set enrichment analysis (GSEA) showed that cell-cycle-related biological processes, such as “MYC targets” and “G2M checkpoint”, were significantly enriched in MNX1-AS1–associated genes (Fig. 5B). Among these processes, the MYC pathway/targets were the prominent gene sets enriched in the MNX1-AS1–regulated genes.
With the observation that MNX1-AS1 directly interacts with YB1, we next investigated the impact of altered MNX1-AS1 levels on YB1 downstream signaling. Our results demonstrated that MNX1-AS1 had an effect on colorectal cancer cell-cycle progression, so we mainly focused on the downstream targets of YB1 that are involved in cell-cycle regulation. We explored the correlations between the levels of MNX1-AS1 and the mRNA levels of genes downstream of YB1 in a cohort of 72 clinical colorectal cancer tissue samples, and we observed positive correlations between MNX1-AS1 and CCNA2, CCND1, CDC25A, CDC45, CDC20, MYC, BAX, XRCC5, AKT, MSH2, and EIF4G in the colorectal cancer tissues (Supplementary Fig. S5A). Thus, some downstream targets of YB1, including CCND1, CDC20, MAD2L1, and CCNA2, overlapped with differentially expressed genes in MNX1-AS1-knockdown cells. We then confirmed that these genes were significantly downregulated by MNX1-AS1 knockdown (Fig. 5C and D; Supplementary Fig. S5B). To further determine the role of the MNX1-AS1-YB1 regulatory axis in colorectal cancer, we knocked down YB1 in RKO and SW480 cells that stably overexpressed MNX1-AS1 and observed that the MNX1-AS1–mediated regulation of the expression of YB1 targets was attenuated by YB1 knockdown (Fig. 5E). These observations were consistent with our observation that MNX1-AS1 knockdown resulted in cell-cycle arrest in colorectal cancer cells and that the oncogenic role of MNX1-AS1 partially depended on YB1 and the YB1 downstream targets CCNA1, CCND1, CDC20, and MAD2L1. To confirm the correlation between the expression of MNX1-AS1 and the MNX1-AS1–regulated genes CCNA1, CCND1, CDC20, and MAD2L1, colorectal cancer tissues from a 100-patient cohort were analyzed, and the results showed that MNX1-AS1 had a significant correlation with the downstream targets of MYC and YB1 (Fig. 5F and G).
There is evidence that YB1 and lncRNAs can affect MYC translation (33). In addition, our results demonstrated that MNX1-AS1 can affect the downstream targets of both MYC and YB1; therefore, MNX1-AS1 or YB1 may be associated with the MYC protein. We next investigated the relationship between MNX1-AS1 and MYC or YB1. RIP showed that MNX1-AS1 failed to immunoprecipitate MYC (Supplementary Fig. S5C). The total mRNA and protein levels of MYC remained unchanged after the knockdown or overexpression of MNX1-AS1 or the knockdown of YB1 in colorectal cancer cells (Supplementary Fig. S5D and S5E). Furthermore, in colorectal cancer tissues, there was no correlation between MYC and YB1 at the RNA level (Supplementary Fig. S5F). Taken together, these results suggested that the oncogenic role of MNX1-AS1 might be associated with the downstream targets of YB1 and MYC without direct binding to MYC.
MYC promotes transcription of MNX1-AS1
To explore the underlying mechanisms of upregulated MNX1-AS1 in colorectal cancer tissues, we screened transcription factors that may regulate the expression of MNX1-AS1. We searched for transcription factors associated with MNX1-AS1 expression in TCGA database and found 18 positively correlated and 28 negatively correlated transcription factors (correlation coefficient, Pearson R2 > 0.3; Supplementary Fig. S6A). Next, we knocked down the top 5 positively correlated transcription factors by siRNA transfection. The results showed that silencing MYC but not the other four proteins, including MNX1, TFAP4, ETV4, and E2F1, significantly downregulated MNX1-AS1 expression (Fig. 6A; Supplementary Fig. S6B). In addition to the significant correlation between MNX1-AS1 and MYC in TCGA database (Fig. 6B), this correlation, at both the RNA and protein levels, was also confirmed in colorectal cancer tissues of a 71-patient cohort from SYSUCC (Fig. 6C and D).
MYC binds to DNA and functions as a transcription factor by forming a heterodimer with another transcription factor, MAX (34). We analyzed the potential transcription factors that bind to the promoter of MNX1-AS1 using the JASPAR database. We found a high possibility that MYC:MAX may bind to the promoter region of MNX1-AS1 (Fig. 6E). Furthermore, we performed MYC chromatin immunoprecipitation (ChIP) assays in RKO and SW480 cells, and these ChIP assays enriched fragments of the MNX1-AS1 promoter (Fig. 6F). Moreover, luciferase reporter assays confirmed the transcriptional activity of MYC in the MNX1-AS1 promoter regions (Supplementary Fig. S6C). To further identify the region by which MYC regulates MNX1-AS1 transcription, we expressed the full-length promoter of MNX1-AS1, the truncated promoter of MNX1-AS1 excluding the MYC-binding sites, and the promoter of MNX1-AS1 containing mutated MYC-binding sites in 293T, RKO, and SW480 cells. MYC-binding site truncation or mutation in the MNX1-AS1 promoter abolished its active transcription (Fig. 6G), indicating that MYC directly regulates the transcription of MNX1-AS1 through these binding sites.
MNX1-AS1–YB1 axis is a potential therapeutic target for treating colorectal cancer
Targeting lncRNAs has shown promising therapeutic potential (18). ASOs can be used to knockdown lncRNAs, and these molecules have been shown to have effects in mouse models of Angelman syndrome, a neurogenetic disease characterized by severe mental and developmental disorders (35). Oxaliplatin is the prominent drug in the standard treatment of colorectal cancer. To determine whether the combination of ASOs targeting MNX1-AS1 combined with oxaliplatin might be more therapeutically effective in vivo, xenograft assays were carried out in both cell-based and PDX models (Supplementary Fig. S7A). Compared with the control group, the ASO MNX1-AS1 group showed significantly inhibited tumor growth (Fig. 7A and B; Supplementary Fig. S7B and S7C), with smaller and lighter orthotropic tumors (Fig. 7C; Supplementary Fig. S7D). Tumor growth was further suppressed by treatment with the combination of ASO MNX1-AS1 plus oxaliplatin compared with treatment with singe agents (Fig. 7A–C; Supplementary Fig. S7B–S7D). Pathologically, H&E staining and microscopic examination showed more necrosis and fibrosis in the ASO MNX1-AS1–treated tumors than in the control tumors. In the combination treatment group (ASO MNX1-AS1 + oxaliplatin), the tumors exhibited the highest levels of necrosis and fibrosis, both of which showed strong therapeutic effects according to NCCN guidelines (Fig. 7D; Supplementary Fig. S7E). IHC staining of Ki67 in the tumor sections further revealed that cancer cell proliferation was significantly decreased in the combination-treated group compared with that in either single agent–treated group (Fig. 7D; Supplementary Fig. S7E).
For patients with metastatic colorectal cancer, although the resection of certain metastases might lead to curative outcomes (4), approximately 25% of patients with metastatic colorectal cancer have a poor prognosis (36). Standard care for these patients is chemotherapy (fluoropyrimidines plus either oxaliplatin or irinotecan) as the first-line therapy and biological drugs targeting VEGF (bevacizumab or regorafenib) and EGFR (cetuximab or panitumumab for RAS wild-type patients) as the second-line therapy (4, 37, 38). However, these treatments are only effective in the short-term and eventually result in drug resistance and a lack of therapeutic options. Moreover, prolonged use of standard chemotherapy alone or in combination with biological drugs escalates drug toxicity in patients with metastatic colorectal cancer (24). Therefore, there is an urgent need to identify novel drugs with greater efficacy and lower toxicity. With improved silencing techniques, targeting lncRNAs has become increasingly promising as novel therapeutic targets (23, 39); and some ASO-based therapies have been tested in clinical trials (17). In this study, we aimed to identify lncRNAs that are aberrantly expressed in stage IV colorectal cancer and have prognostic value and to test their therapeutic effects in colorectal cancer as single agents or in combination with standard chemotherapy. Therefore, we analyzed the transcriptomes of 21 pairs of stage IV colorectal cancer tumor tissues and normal colorectal tissues and identified seven novel lncRNAs that were significantly upregulated in colorectal cancer; of these lncRNAs, the expression of 3, namely, MNX1-AS1, LINRIS (linc00920; ref. 40), and FAM83H-AS1, was correlated with prognosis in TCGA dataset. Furthermore, using the TCGA-COAD dataset, we confirmed that MNX1-AS1 may serve as an effective prognostic indicator comparable to well-known prognostic indicators, such as KRAS, BRAF, and erbB2, which also suggested a role for MNX1-AS1 in colorectal cancer progression.
LncRNAs play pivotal roles in mediating tumorigenesis and progression by directly binding to crucial proteins, which are involved in signal transduction in cancer cells (5). For example, lncGata6 promotes intestinal tumorigenesis and maintains the stemness of intestinal stem cells (8). In contrast, lncRNA APC1 can inhibit colorectal cancer pathogenesis by directly binding to Rab5b mRNA and reducing its stability (26). Antisense lncRNA DSCAM-AS1 mediates breast cancer progression and tamoxifen resistance (11). Here, we mechanistically validated the novel antisense lncRNA MNX1-AS1 as a potential oncogene in colorectal cancer. Antisense lncRNAs may encode peptides to influence tumor growth (41). MNX1-AS1 showed little ability to encode peptides, as calculated by the CPAT algorithm. Next, we found that MNX1-AS1 significantly affected colorectal cancer cell proliferation by directly binding to and stabilizing YB1, a DNA/RNA-binding protein that has already been reported to be regulated by lncRNA HOXC-AS3 (42). YB1 plays important roles in apoptosis, cell proliferation, differentiation, and stress responses (43). LncRNAs play a role in regulating YB1 expression (27). Our study showed that MNX1-AS1 mainly interacted with YB1 through its 391–992 nt region in RKO and SW480 cells, whereas in HCT116 cells, MNX1-AS1 interacted with both the 391–580 nt and 581–992 nt regions. More than one segment of MNX1-AS1 was involved in its interaction with YB1 in different colorectal cancer cells, which may be due to the genetic and epigenetic features of the colorectal cancer cell lines (44). LncRNA MIR22GH was shown to prevent the proteasomal degradation of YB1 in lung cancer cells, which contributed to YB1 overexpression (28). Posttranslational modifications, including ubiquitination, may also play an important role in YB1 expression (29). We demonstrated that MNX1-AS1 might stabilize YB1 by preventing the ubiquitination of multiple residues in colorectal cancer cells. Our results showed that MNX1-AS1 prevented the ubiquitination and degradation of YB1, leading to enhanced proliferation of colorectal cancer cells.
Next, we explored the downstream targets regulated by MNX1-AS1 and YB1. Many lncRNAs can affect more than one hallmark of cancer biology; for example, colorectal cancer–associated lncRNA (CCAL) not only promotes colorectal cancer progression but also induces MDR by activating Wnt signaling and suppressing AP-2a (16). LINC00152 can be activated by the transcription coactivator YAP1 and regulates Fascin actin-bundling protein 1 (FSCN1) expression, thus promoting the malignant dissemination and metastasis of colorectal cancer (45). We used RNA-seq analysis and found a marked overlap in genes that are differentially expressed by MNX1-AS1 knockdown and genes transcriptionally regulated by YB1, suggesting that MNX1-AS1 binds to YB1 to promote tumorigenesis in colorectal cancer. The knockdown of YB1 mimicked the reduced proliferation of colorectal cancer cells observed upon MNX1-AS1 silencing.
There is emerging evidence that MYC plays an important role in lncRNA transcriptional activation. Cohesion regulator noncoding RNA (CONCR) is required for cell cycle and is transcriptionally activated by MYC in multiple cancer types (46). MYC forms a heterodimer with a partner protein, MAX, via a basic-helix-loop-helix-leucine zipper domain. The MYC–MAX complex binds directly to the DNA sequence (CACA/GTG), which is part of the general E-box (CACGTG) DNA recognition sequence, and functions as a transcriptional activator or repressor (34, 47, 48). Our results indicated that MYC directly regulated the expression of MNX1-AS1 through the binding of the MYC:MAX complex to the MNX1-AS1 promoter regions. MNX1-AS1 and MYC protein and mRNA expression was positively correlated in primary colorectal cancer tumors, suggesting that dysregulation of the MYC–MNX1-AS1–YB1 axis leads to colorectal cancer proliferation.
With technological development, the possibility of targeting lncRNAs as treatments is increasingly feasible (49). Our results showed that lncRNA MNX1-AS1 promoted the proliferation of colorectal cancer cells by stabilizing YB1, suggesting that MNX1-AS1 might be a potential therapeutic target (Supplementary Fig. S8). With the development of RNA silencing technologies, emerging evidence has revealed roles of lncRNAs in drug resistance (49) and effective therapeutics (17, 23). Recent studies identified lncRNA CCAT1, which enhances the sensitivity of colorectal cancer to JQ1, a BET inhibitor (50). MIR100HG has been shown to correlate with cetuximab resistance in colorectal and head and neck squamous cell cancer (15). Overexpression of the antisense lncRNAs EGFR-AS1 and DSCAM-AS1 could induce resistance to TKIs or tamoxifen, respectively (10, 11). The combined application of drugs with different mechanisms can enhance the therapeutic effects and prevent resistance to single drugs. These previous reports suggest that lncRNAs, including antisense lncRNAs, can be potential therapeutic targets in colorectal cancer, and targeting lncRNAs combined with conventional chemotherapies may enhance chemosensitivity and improve treatment responses (51). In this study, in both cell-based xenografts and PDX models, ASO MNX1-AS1 significantly suppressed the growth of colorectal cancer. Treatment with the combination of ASO MNX1-AS1 and oxaliplatin significantly suppressed tumor growth compared with treatment with either agent alone. These data indicate that MNX1-AS1/YB1 could serve as a potential therapeutic target in metastatic colorectal cancer.
No disclosures were reported.
Q.-N. Wu: Conceptualization, data curation, formal analysis, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. X.-J. Luo: Data curation, formal analysis, validation, investigation, visualization, methodology, writing–original draft. J. Liu: Formal analysis, validation, investigation, visualization, methodology, writing–original draft. Y.-X. Lu: Data curation, formal analysis, investigation, methodology. Y. Wang: Data curation, investigation, visualization, methodology. J. Qi: Investigation, writing–review and editing. Z.-X. Liu: Software, formal analysis, investigation. Q.-T. Huang: Resources, investigation. Z.-K. Liu: Formal analysis, validation. J.-B. Lu: Resources, validation, methodology. Y. Jin: Resources, formal analysis, methodology. H.-Y. Pu: Resources, writing–original draft, discussion and analysis. P.-S. Hu: Validation, methodology. J.-B. Zheng: Validation, investigation, writing–review and editing. Z.-L. Zeng: Resources, funding acquisition, validation, investigation, methodology. H.-Q. Ju: Resources, funding acquisition. D. Xie: Conceptualization, resources, methodology, writing–review and editing. Q. Zhao: Conceptualization, resources, data curation, software, formal analysis, supervision, funding acquisition, investigation, methodology, writing–original draft, writing–review and editing. R. Xu: Conceptualization, resources, supervision, funding acquisition, investigation, project administration, writing–review and editing.
This research was supported by the National Natural Science Foundation of China (81930065 and 81802438), Science and Technology Program of Guangdong (2019B020227002), Science and Technology Program of Guangzhou (201904020046, 201803040019, and 201704020228), Fundamental Research Funds for the Central Universities (SYSU; 19ykpy184 and 19ykpy173), and CAMS Innovation Fund for Medical Sciences (CIFMS; 2019-I2M-5-036).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.