Linc00942 promotes TMZ resistance and STAT3 signaling activity in GBM cells.
To determine the differential lncRNA expression profiles between TMZ-resistant GBM cells and parental cells, we analyzed lncRNA microarrays (27) and found that Linc00942 was highly expressed in a variety of TMZ-resistant GBM cells (Fig. 1A) and had a potential biological function in promoting TMZ resistance in glioma according to siRNA screening assays (Supplementary Fig. 1). Subsequently, we identified the sequence of Linc00942 by RACE assay, which showed that Linc00942 consisted of two exons with a full length of 2230 nt, transcribed on chromosome 12 from site 1,500,521 to site 1,504,440 (5'-3') (Supplementary Fig. 2A-D). ORF analysis (Supplementary Fig. 2E-F) and coding ability analysis (Supplementary Fig. 2G-H) revealed that Linc00942 lacked coding ability. Linc00942 was enriched in both the nucleus and cytoplasm, as detected by FISH and qRT‒PCR (Supplementary Fig. 2I-K). To further explore the effect of Linc00942 on TMZ resistance in GBM, we knocked down Linc00942 in TMZ-resistant GBM cells by utilizing shRNA (Supplementary Fig. 3A) and confirmed the effect of Linc00942 on TMZ sensitivity both in vitro and in vivo. In in vitro experiments, Linc00942 knockdown resulted in a lower IC50 of TMZ and a lower EdU-positive rate (Fig. 1B-C, Supplementary Fig. 3B-D), suggesting that knockdown of Linc00942 led to significantly reduced cell viability and DNA replication under TMZ treatment. Glioma stemness is one of the main factors influencing drug resistance, and TMZ-resistant glioma cells have a greater capacity for self-renewal under TMZ treatment (28, 29). In this study, we performed a tumor sphere formation assay and found that knockdown of Linc00942 significantly suppressed self-renewal in TMZ-resistant glioma cells (Supplementary Fig. 3E-F). In vivo, we established orthotopic mouse models using TMZ-resistant LN229R GBM cells with or without Linc00942 knockdown and paired parental LN229 GBM cells (Fig. 1D). The bioluminescence imaging results showed that Linc00942 knockdown attenuated the difference in the therapeutic effect of TMZ between the LN229R group and the parental group (Fig. 1E). Mice bearing LN229R cells exhibited significantly poorer survival than mice bearing parental LN229 cells. In addition, knockdown of Linc00942 prolonged the survival time of mice bearing LN229R cells (Fig. 1F). These results indicated that Linc00942 contributes to TMZ resistance in GBM cells and that knockdown of Linc00942 partly restored TMZ sensitivity in TMZ-resistant GBM cells. Next, we established a Linc00942-overexpressing GBM model (Supplementary Fig. 4A) to perform CCK-8, EdU, and tumor sphere formation assays. Compared to scramble GBM cells, the overexpression of Linc00942 in GBM cells significantly improved cell viability, DNA replication, and self-renewal during TMZ treatment (Fig. 1G-H, Supplementary Fig. 4B-E). To further investigate how Linc00492 promotes TMZ resistance in GBM cells, we performed RNA-seq in Linc00492-overexpressing LN229 cells and scramble LN229 cells to identify transcriptomic changes. The volcano plot showed that 716 genes were significantly upregulated in Linc00492-overexpressing LN229 cells (Supplementary Fig. 5A). We also confirmed differentially expressed mRNAs in LN229R cells by a gene ranking plot according to the microarray data published in our previous study (27) (Supplementary Fig. 5B). A Venn diagram showed that 278 genes were coupregulated in both Linc00492-overexpressing LN229 cells and TMZ-resistant LN229R GBM cells (Supplementary Fig. 5C). Transcription factor enrichment analysis by UCSC showed that these coupregulated genes were significantly enriched in STAT targets (Supplementary Fig. 5D). ssGSEA and GSEA suggested that STAT3 signaling and the JAK/STAT pathway, but not STAT5 signaling, were activated in both LINC00492-overexpressing LN229 cells and TMZ-resistant LN229R GBM cells (Fig. 1I). Collectively, these data demonstrate that Linc00492 significantly increased TMZ resistance in GBM cells and was related to activated STAT3 signaling.
TPI1 and PKM2 are direct binding partners and downstream factors regulated by Linc00942.
lncRNAs often exert their diverse biological functions by interacting with RNA-binding proteins (5, 30). To further elucidate the potential mechanism of Linc00942, we next used ChIRP followed by mass spectrometry (ChIRP-MS) to identify potential Linc00942-binding factors in GBM cells (Fig. 2A, Supplementary Fig. 6A-D). A total of 19 specific biotin-labeled Linc00942 probes were used to capture Linc00942 from total cellular extracts of TMZ-resistant GBM cells LN229R (Supplementary Table 7), while negative probes were used as a control. Twelve proteins were significantly enriched with Linc00942 probes vs. negative probes (Supplementary Fig. 6E). Among these proteins, we found that TPI1 and PKM1/2, two glycolytic enzymes, were the top 2 Linc00942 binding factors (Fig. 2B). We further confirmed that TPI1 and PKM2, but not PKM1, significantly interacted with Linc00942 by ChIRP-WB (Fig. 2C, Supplementary Fig. 6F) and RIP assays (Fig. 2D, Supplementary Fig. 6G-H) in both TMZ-resistant GBM cells and Linc00942-overexpressing GBM cells. Recent studies have shown that phosphorylation and nuclear translocation of TPI1 and PKM2 result in significant changes in biological processes, leading to therapy resistance and oncogene expression in many cancers (31, 32). Therefore, we detected the expression levels of TPI1 and PKM2 by WB and found that there was no significant difference in the total expression levels of TPI1 and PKM2. However, the phosphorylation and nuclear translocation of TPI1 and PKM2 were significantly increased in TMZ-resistant GBM cells and Linc00942-overexpressing GBM cells (Fig. 2E, Supplementary Fig. 7A). We also confirmed that there was significant nuclear translocation of TPI1 and PKM2 in TMZ-resistant GBM cells and Linc00942-overexpressing GBM cells by IF assay and colocalization analysis (Fig. 2F-G). In addition, a variety of metabolic enzymes and substrates have been reported as regulators of histone modification. Both TPI1 and PKM2 can promote histone acetylation through nuclear translocation, especially nuclear translocation of TPI1, which leads to significant histone acetylation in several types of cancer cells by modulating HDAC activity (7, 10). Therefore, we performed HDAC activity assays and WB to detect whether changes in histone acetylation occurred in TMZ-resistant GBM cells and Linc00492-overexpressing GBM cells. In the end, we found that the activity of HDAC3 was significantly decreased and the acetylation of H3K4 was increased (Fig. 2H, Supplementary Fig. 7B-C). Taken together, these results revealed that Linc00942 could bind with TPI1 and PKM2 and regulate histone acetylation via the activity of HDAC3 in GBM cells.
Interactions of TPI1 and PKM2 with Linc00942 are essential for Linc00942-mediated TMZ resistance.
Both RNA molecules and proteins fold into complicated three-dimensional (3D) structures to perform a broad range of cellular functions (33, 34). To further investigate the mechanism by which Linc00942 binds to TPI1 and PKM2, we performed molecular docking analysis to identify the specific binding site. Alpha fold was used to predict the 3D structures of TPI1 and PKM2 (Supplementary Fig. 8A-B). The 3dRNA server was used to establish a 3D model of Linc000942 based on the minimal folding free energy (MFE) structure, and the best model was selected for further analysis (Supplementary Fig. 8C-D). Molecular docking analysis was performed by AutoDock Vina, docking site analysis and docking scores were evaluated by HDOCK, and the model of molecular docking was visualized by PyMol (Fig. 3A-B). The results of molecular docking analysis showed that Linc00942 interacted with TPI1 through a kissing loop at 1000~1600 nt of Linc00942, and the first-rank binding site of PKM2 was located at a pseudoknot structure, 1800~2100 nt of Linc00942 (Supplementary Fig. 8E-F). Several studies have shown that lncRNAs interact with and modulate corresponding proteins through specific substructures (35, 36). MFE and 3D structure analysis revealed that Linc00942 was composed of three main substructures. We named these three substructures p1 (1-219/1786-2230 nt), p2 (220-885 nt), and p3 (886-1785 nt). To verify the binding sites of TPI1 and PKM2, in combination with the results of molecular docking analysis, we constructed mutant Linc00942 ΔP1 (del1-219/1786-2230 nt) and ΔP3 (del886-1785 nt) plasmids according to the predicted binding site of Linc00942 and TPI1/PKM2 (Fig. 3C). A biotin-labeled RNA pull-down assay showed that TPI1 interacted with ΔP1 (del1-219/1786-2230 nt), and PKM2 mainly bound to ΔP3 (del886-1785 nt) (Fig. 3D). GBM cells transfected with ΔP1 and ΔP3 were designated MUT-1 and MUT-2 GBM cells, respectively (Fig. 3E, Supplementary Fig. 9A-B). We further performed RIP assays in MUT-1 and MUT-2 GBM cells, and the results showed that TPI1 mainly bound to mutant Linc00942 in MUT-1 and PKM2 mainly bound to mutant Linc00942 in MUT-2 (Fig. 3F-G, Supplementary Fig. 9C-D). We then investigated the role of the interaction of Linc00942 with TPI1 and PKM2 in TMZ resistance. The IC50 assay showed that both mutations of Linc00942 partly attenuated TMZ resistance induced by Linc00942 overexpression in GBM cells. However, both MUT-1 and MUT-2 GBM cells exhibited higher TMZ resistance than scramble GBM cells (Fig. 3H, Supplementary Fig. 9E). EdU assays indicated that Linc00942 mutations led to a significantly decreased proportion of EdU-positive cells compared with that among Linc00942-overexpressing GBM cells during TMZ treatment (Fig. 3J-I). In addition, the mutations of Linc00942 weakened the ability of Linc00942 to promote tumor sphere formation (Supplementary Fig. 9F). These results indicate that Linc00942-induced TMZ resistance depends on interactions with both TPI1 and PKM2.
Linc00942 promotes the acetylation of H3K4 and activation of STAT3 through the interactions of TPI1 and PKM2
Based on the above results, we found that TMZ resistance induced by Linc00942 was dependent on the interaction of TPI1 and PKM2. However, it is unclear whether TPI1 and PKM2 are regulated by direct interaction. Therefore, we next determined the expression levels of TPI1 and PKM2 in MUT-1 and MUT-2 GBM cells. We found that the phosphorylation level and nuclear expression level of TPI1, but not PKM2, were significantly increased in MUT-1 GBM cells compared to scramble GBM cells. In addition, phosphorylation and nuclear translocation of PKM2 were significantly increased in MUT-2 GBM cells but not in MUT-1 GBM cells (Fig. 4A, Supplementary Fig. 10A). The quaternary structure of proteins is closely related to phosphorylation, so phosphorylation can lead to significant changes in biological function. Both TPI1 and PKM2 can undergo complicated polymerization associated with crucial biological processes (17, 37). Therefore, we further analyzed the polymerization of TPI1 and PKM2 by native PAGE. The results showed that the expression of the TPI1 dimer was significantly increased in Linc00942-overexpressing and MUT-1 GBM cells. In addition, the expression of the dimer but not the tetramer of PKM2 was significantly increased in Linc00942-overexpressing and MUT-2 GBM cells (Fig. 4B, Supplementary Fig. 10B). Furthermore, the TPI1 dimer and PKM2 dimer were highly expressed in TMZ-resistant GBM cells (Fig. 4C, Supplementary Fig. 10C). IF assays and colocalization analysis revealed that the nuclear translocation of TPI1 was increased in MUT-1 GBM cells and that the nuclear translocation of PKM2 was increased in MUT-2 GBM cells (Fig. 4D). These results indicated that the mutations of the TPI1/PKM2 binding sites in Linc00942 regulated the complicated polymerization processes of TPI1 and PKM2. We next determined HDAC3 activity and H3K4ac expression in MUT-1 and MUT-2 GBM cells. Compared to the scramble cells, MUT-1 GBM cells showed a significant reduction in HDAC3 activity and higher H3K4ac expression, and the results were similar for Linc00942-overexpressing GBM cells (Fig. 4E, Supplementary Fig. 10D). Recent studies have shown that PKM2 is a direct regulator of STAT3 that and nuclear translocation of PKM2 significantly promotes the phosphorylation of STAT3 (16, 17). Therefore, we performed immunoprecipitation followed by WB to determine whether Linc00942 promotes the interaction of PKM2 and STAT3. The results showed that the interaction of PKM2 and STAT3 was significantly enhanced in Linc00942-overexpressing GBM cells and MUT-2 GBM cells compared with scramble cells. Moreover, a similar increase in the interaction between PKM2 and STAT3 was found in TMZ-resistant GBM cells compared with parental cells. In addition, we confirmed a markedly activated STAT3/P300 axis in Linc00942-overexpressing GBM cells, TMZ-resistant GBM cells, and MUT-2 GBM cells (but not in MUT-1 cells), suggesting that the interaction between Linc00942 and PKM2 played a key role in activating the STAT3 pathway (Fig. 4F, Supplementary Fig. 10E). Taken together, our results demonstrate that the posttranslational modification of TPI1 and PKM2 regulated by interactions with Linc00492 contributes to the acetylation of H3K4 and the activation of the STAT3 signaling pathway in GBM cells.
Linc00942 promotes the transcriptional activation of SOX9, regulating TMZ resistance in GBM cells
Acetylation of H3K4 in specific promoter regions could be regulated by both HDAC3 and p300, which could contribute to marked transcriptional changes (38, 39). To determine how Linc00942-induced transcriptional alterations contribute to TMZ resistance in GBM cells, we analyzed the expression of STAT target genes via a heatmap (Fig. 5A). The results showed that SOX9 was significantly upregulated in Linc00942-overexpressing LN229 cells and TMZ-resistant LN229R GBM cells. Recent studies have reported that SOX9 is a key stemness regulator in GBM, and high expression of SOX9 results in drug resistance in various cancers, including TMZ resistance in GBM cells (24-26, 40). We further detected the expression of SOX9 by q-RT‒PCR. The results showed that the expression level of SOX9 was significantly increased in Linc00942-overexpressing GBM cells and TMZ-resistant GBM cells compared with scramble/parental GBM cells. Moreover, knockdown of Linc00942 inhibited SOX9 expression in TMZ-resistant GBM cells. However, only a slight increase in SOX9 was observed in MUT-1 and MUT-2 GBM cells (Fig. 5B, Supplementary Fig. 11A-C). WB and tumor sphere IF assays confirmed the qRT‒PCR results. In addition, p-STAT3 expression was significantly increased in Linc00942-overexpressing GBM cells, MUT-2 GBM cells, and TMZ-resistant GBM cells compared with scramble/parental GBM cells, which is consistent with our previous results (Fig. 5C-F, Supplementary Fig. 11D-E). To further elucidate the mechanisms of transcription of SOX9, we downloaded the ChIP-Seq data of p300 (GSM935545), STAT3 (GSM935591), HDAC3 (GSE127356), and H3K4ac (GSM521894) from the GEO database (Supplementary Fig. 12A). We found that p300, STAT3, HDAC3, and H3K4ac were significantly enriched in the SOX9 promoter region (Supplementary Fig. 12B). We then performed a ChIP‒qPCR assay in LN229 and HG7 cells. The results showed that HDAC3 enrichment in the SOX9 promoter region was significantly decreased in Linc00942-overexpressing and MUT-1 GBM cells, and the interaction of STAT3 with the SOX9 promoter region was significantly increased in Linc00942-overexpressing and MUT-2 GBM cells compared with scramble cells. However, the level of STAT3 enrichment in the SOX9 promoter region was much lower in MUT-2 GBM cells than in Linc00942-overexpressing cells. Dramatically increased p300 and H3K4ac levels in the SOX9 promoter region were detected only in Linc00942-overexpressing GBM cells (Supplementary Fig. 12C). Changes in the levels of p300, STAT3, HDAC3, and H3K4ac were also confirmed in TMZ-resistant GBM cells, and these effects could be reversed by knockdown of Linc00942 (Supplementary Fig. 12D-E). Next, we knocked down SOX9 in Linc00942-overexpressing GBM cells and TMZ-resistant GBM cells by shRNA (Supplementary Fig. 13A). The IC50 assay showed that the reduction in SOX9 significantly inhibited the viability of Linc00942-overexpressing GBM cells and TMZ-resistant GBM cells during TMZ treatment (Fig. 5G, Supplementary Fig. 13B). IF assays showed that knockdown of SOX9 resulted in a significantly decreased EdU-positive rate in Linc00942-overexpressing GBM cells and TMZ-resistant GBM cells (Fig. 5H, Supplementary Fig. 13C-D). Sphere formation assays showed that knockdown of SOX9 significantly suppressed the self-renewal of Linc00942-overexpressing GBM cells and TMZ-resistant GBM cells (Supplementary Fig. 13E). These results showed that Linc00492 promoted TMZ resistance in GBM in a manner dependent on the marked transcriptional activation of SOX9 through both HDAC3 activity suppression and STAT3 activation.
In vivo experiments validated that Linc00942 promotes TMZ resistance by interacting with TPI1 and PKM2
To further verify the above results that Linc00942 regulates TMZ resistance via interaction with TPI1 and PKM2, we established orthotopic mouse models bearing GBM xenografts by transplantation of Linc00942-overexpressing, MUT-1, MUT-2, and scramble LN229 cells (Fig. 6A). Five days after GBM cell implantation, mice were treated intraperitoneally with TMZ or DMSO (60 mg kg-1 day-1 per mouse) every 7 days. Simultaneously, the mice underwent bioluminescence tomography every 7 days starting on day 7. The bioluminescence image showed that the treatment efficacy of TMZ was significantly inhibited in mice bearing GBM derived from Linc00942-overexpressing cells. Moreover, mutations in Linc00942 (MUT-1 and MUT-2) in GBM cells partly inhibited TMZ resistance in vivo (Fig. 6B). Survival analysis and bioluminescence analysis indicated that TMZ treatment significantly prolonged the survival time of mice bearing MUT-1, MUT-2 and scramble LN229 cells compared with that of those bearing Linc00942-overexpressing GBM cells (Fig. 6C). The IHC results showed that the nuclear translocation of TPI1 and expression of H3K4ac were significantly increased in mice bearing Linc00942-overexpressing and MUT-1 LN229 cells. Nuclear translocation of PKM2 and expression of p-STAT3 were significantly increased in mice bearing Linc00942-overexpressing and MUT-2 LN229 cells. A marked increase in the expression of SOX9 was observed only in mice bearing Linc00942-overexpressing LN229 cells (Fig. 6D-E). In addition, we performed an IHC assay in mice bearing GBM cells derived from LN229 cells and TMZ-resistant LN229R cells with or without Linc00942 knockdown. The results showed that SOX9, p-STAT3, and H3K4ac were highly expressed in mice bearing TMZ-resistant LN229R cells. Furthermore, TPI1 and PKM2 translocation was significantly promoted in TMZ-resistant LN229R mice. Knockdown of Linc00942 attenuated these effects (Supplementary Fig. 14A-B). Furthermore, we established orthotopic models with Linc00942-overexpressing LN229 cells and TMZ-resistant LN229R GBM cells, with or without SOX9 knockdown. Survival analysis and bioluminescence analysis indicated that knockdown of SOX9 significantly restored TMZ sensitivity in the in vivo experiment (Supplementary Fig. 14C-D). Overall, our results revealed that the TMZ resistance induced by Linc00942 is related to a marked increase SOX9 expression, which is regulated by the direct interaction of TPI1 and PKM2 with Linc00942.