CRISPRa screening identifies Linc01056 as a candidate regulator of sorafenib sensitivity in HCC

To identify the critical lncRNA regulator of sorafenib sensitivity in HCC, we applied a global screening approach involving a CRISPR/Cas9 lncRNA SAM pooled library containing 96,458 sgRNAs that targeted the TSSs of 10,504 unique lncRNAs [18]. The human HCC cell line MHCC97L, which exhibited a moderate response to sorafenib (Fig. S1a), was used to establish an in vitro model for CRISPRa screening of potential lncRNA regulators. MHCC97L cells with stable expression of the Cas9 protein were transduced with lentiviral sgRNAs and were then treated with vehicle or 5 µM sorafenib for 7 days (Fig. 1a). A 7-day treatment with sorafenib, which provided strong pressure for the selection of positive and negative lncRNA regulators of sorafenib sensitivity in HCC cells, significantly suppressed cell proliferation and induced cell death in MHCC97L cells (Fig. 1b & S1b). MHCC97L cells treated with vehicle or sorafenib were then subjected to next-generation sequencing to identify lncRNAs that were negatively and positively associated with sorafenib sensitivity. Model-based analysis of genome-wide CRISPR/Cas9 activation library was used to identify hits from our CRISPRa screening based on a previous study [27]. Quality control assessment suggested that the sgRNAs resulted in high-purity and clean reads (Fig. S1c). Sorafenib-treated group has a slightly higher average normalised read count, suggesting the successful screening of the lncRNA (Fig. 1c). Using a cut-off of |log2FC|≥1, we identified 67 lncRNAs that were downregulated and 79 lncRNAs that were upregulated in the surviving cells after sorafenib treatment (Fig. 1d). The lncRNA-specific sgRNA were confirmed that can effectively activate the corresponding targets when transfected into MHCC97L cells using qPCR (Fig. S1d). From the result of sequencing, Linc01056 was one of the most downregulated lncRNAs in the surviving MHCC97L cells after 7 days of 5µM sorafenib treatment (Fig. 1e). Linc01056 is a lncRNA located at chromosome 20, 63,038,011–63,053,863, with an exon-spiced length of 1,234 nt (Table S1), while lacking protein coding potentials (Fig. S1e). It does not overlap with any known protein-coding genes (Fig. S1f). According to the published dataset GSE30611, Linc01056 mainly presents high copy numbers in breast cancer, gastric cancer, and HCC, while presents low copy numbers in lung cancer, ovarian cancer, and leukaemia [28]. To examine whether Linc01056 expression is suppressed in sorafenib-resistant HCC, we established in vivo-generated sorafenib-resistant HCC tumours according to our previous study [29]. Significant downregulation of Linc01056 expression was observed in sorafenib-resistant HCC tumours compared to their sorafenib-sensitive counterparts (Fig. 1f). In addition, we challenged MHCC97L cells with 10 µM sorafenib for 24 h and 7 days. Intriguingly, we observed an impulsive stimulated expression of Linc01056 in 24 h sorafenib treatment, but significantly suppressed expression after 7-day exposure (Fig. 1g). The sorafenib resistance was greatly enhanced with the reduction of Linc01056 expression after 7-day exposure (Fig. 1h), indicating that the reduced level of Linc01056 in response to sorafenib treatment may be associated with sorafenib resistance. Our previous study illustrated the change in transcription factor profile in sorafenib-treated MHCC cells [29], where ETS Proto-Oncogene 1 (ETS1) was found to be a responder of sorafenib. To check if ETS1 is also responsible for the transcription of Linc01056, we predicted the binding motif of ETS1 in JASPER and located that AGGAAG from -895 to -900 before the promoter of Linc01056 is a key binding motif (Fig. S1g). Chromatin-immunoprecipitation (ChIP) assay proved the binding of ETS1 to the promoter region of Linc01056 with increased level after sorafenib treatment (Fig. S1h) Induction of Linc01056 expression was suppressed with siETS1 (Fig. S1i), and the sorafenib sensitivity was also reduced (Fig. S1j). The results suggested that ETS1 is a transcriptional regulator of Linc01056. Collectively, we suggest that expression of Linc01056 has a role in response to sorafenib of HCC cells.

Fig. 1

CRISPRa screens identified Linc01056 as a regulator of sorafenib response in HCC cells. (a) Flowchart of CRISPRa screening on MHCC97L cells. (b) 7-day treatment of sorafenib significantly suppressed the proliferation ability of MHCC97L cells. (c) Violin plot of the normalized read count of the sequencing result. The average count of the sorafenib-treated group was slightly higher. (d) Volcano plot of the changes of expression of lncRNAs upon 7-day exposure of sorafenib in HCC cells. (e) Linc01056 was one of most downregulated lncRNA upon acquisition of sorafenib resistance in HCCs. The acquired sorafenib resistance MHCC97L cells were obtained by prolonged seven-day 5 µM sorafenib treatment. (f) In sorafenib-resistant HCC tumour cells, the expression of Linc01056 was potently suppressed. (g) HCC cells were exposed to 1-, 7- and 12-day sorafenib at the dose of 10µM, it was observed that 1-day treatment of sorafenib induced Linc01056 expression, while long-term treatment of sorafenib suppressed Linc01056 expression. (h) Cell viability of MHCC97L were measured against sorafenib treatment for wild-type or 1-day or 7-day treated cells. *p < 0.05, **p < 0.01, ***p < 0.001

Linc01056 is essential for the sorafenib sensitivity of HCC in vitro and in vivo.

To identify the functional role of Linc01056 in the sorafenib sensitivity of HCC cells, we first generated MHCC97L and PLC/PRF/5 cells with stable knockdown of Linc01056. Significant suppression of Linc01056 was observed in HCC cells stably expressing the shRNA plasmids targeting Linc01056 (Fig. 2a). With a better knockdown effect, sh-1056-1 was chosen for the remaining study. Cell viability assays revealed that knockdown of Linc01056 in HCC cells resulted in a significantly attenuated response to sorafenib treatment (Fig. 2b). To confirm there is no off-target effect, we rescued expression of Linc01056 in resistant MHCC97L cells established by 7-day sorafenib intervention by a Linc01056-expressing plasmid. Rescue of Linc01056 potentiated the resistant HCC cells to sorafenib treatment (Fig. 2c). While knockdown of Linc01056 showed a minimal effect on cell growth in the absence of sorafenib, cells with lower Linc01056 expression exhibited a stronger colony formation ability upon long-term sorafenib treatment (Fig. 2d). Linc01056 knockdown significantly reduced apoptosis in HCC cells upon sorafenib treatment compared to that in vector control cells (Fig. 2e). Moreover, treatment of HCC cells with a nontoxic concentration of sorafenib potently suppressed the in vitro motility as well as invasion through the extracellular matrix, and these abilities were significantly restored upon knockout of Linc01056 expression (Fig. S2a & S2b). To investigate if the level of Linc01056 is related to the sorafenib resistance, we performed a CRISPR-KO targeting Linc01056 also on the MHCC97L and PLC/PRF/5 cell line (Fig. S2c). The CRISPR-KO cells showed a higher IC50 value to sorafenib than Linc01056-knockdown cells (Fig. 2b, Fig.S2d), and the results of the apoptosis assay was consistent (Fig. S2e).

Fig. 2

Linc01056 was required for sorafenib response in HCC. (a) Knockdown of Linc01056 in MHCC97L and PLC/PRF/5 cells by shRNA interference. (b) Knockdown of Linc01056 increased cell viability in sorafenib-treated HCC cells. (c) Rescue of Linc01056 potentiated the resistant HCC cells to sorafenib treatment. (d) Knockdown of Linc01056 improved colonic formation of HCC cells in the presence of sorafenib. (e) Knockdown of Linc01056 reduced sorafenib-induced apoptosis in HCC cells. Stable knockdown of Linc01056 (f) accelerated in vivo tumour growth and (g) end-point tumour size, (h) reduced expression of cell apoptosis marker cleaved caspase-3 and (i) promoted the cell proliferation marker Ki67. *p < 0.05, **p < 0.01, ***p < 0.001

To examine the role of Linc01056 in HCC in vivo, we established an orthotopic HCC model in mice via implantation of luciferase reporter-expressing MHCC97L cells with or without Linc01056 knockdown. We observed that Linc01056 knockdown significantly reduced the in vivo tumour response to sorafenib treatment without a change in body weight (Fig. S2f), as indicated by the rate of tumour growth (Fig. 2f). At the end of the study, the livers were harvested, and Linc01056 knockdown was found to result in a larger tumour size and higher tumour weight in sorafenib-treated mice bearing HCC tumours (Fig. 2g). Increased expression of Ki67 and a decreased level of cleaved caspase-3 were observed in tumour tissues from sorafenib-treated mice implanted with Linc01056-knockdown MHCC97L cells compared to those in mice implanted with the corresponding vector control cells (Fig. 2h and 2i). Furthermore, Linc01056 knockdown greatly increased the probability of lung metastasis of MHCC97L cells in mice exposed to sorafenib (Fig. S2g). Taken together, our findings suggest that Linc01056 expression is required for the response of HCC cells to sorafenib treatment both in vitro and in vivo.

Loss of Linc01056 mediates the metabolic switch towards FAO in sorafenib-treated HCC

To further explore the possible mechanisms underlying the reduced sorafenib response in HCC cells with Linc01056 knockdown, we performed proteomics analysis to compare the protein expression profile between sorafenib-treated MHCC97L cells transduced with the vector control plasmid or the Linc01056 shRNA plasmid. Differential changes in protein expression were observed (Fig. 3a). We then shortlisted the upregulated and downregulated proteins upon Linc01056 knockdown (Fig. 3b and 3c) and performed Gene Ontology (GO) analysis to determine the possibly enriched biological processes. We found that the proteins upregulated by Linc01056 knockdown were enriched primarily in FAO-related terms, while the downregulated proteins were enriched in glycolysis-related terms, indicating that Linc01056 knockdown may activate a metabolic switch from glycolysis towards FAO upon sorafenib pressure (Fig. 3d). Consistent enriched pathways were obtained from the proteomics analysis of sorafenib-treated PLC/PRC/5 cells (Fig. S3a). Gene set enrichment analysis (GSEA) confirmed that MHCC97L cells with Linc01056 knockdown showed higher enrichment of FAO activity (Fig. 3e). The expression of genes related to FAO was significantly increased but that of glycolysis-associated genes was markedly reduced in sorafenib-treated HCC cells with Linc01056 knockdown (Fig. S3b & S3c).

Fig. 3

Linc01056 knockdown induced metabolic shift towards fatty acid oxidation. (a). Proteomic analysis on sorafenib-treated MHCC97L cells with or without Linc01056 knockdown. Pathway enrichment on differential gene expression showed that (b) increased genes enriched in pathways related to fatty acid oxidation, while c) reduced genes enriched in pathways related to glycolysis/gluconeogenesis. (d) Changes in expression of FAO-related proteins upon Linc01056 knockdown. (e) GSEA analysis showed enrichment of genes related to FAO. (f) Linc01056 knockdown maintained cellular ATP level upon sorafenib treatment in HCC cells. Knockdown of Linc01056 (g) increased the OCR and (h) decreased the ECAR in HCC cells in the presence of sorafenib. Linc01056 knockdown (i) increased mitochondrial ROS, and (j) maintained the basal respiratory and (k) maximal respiratory capacity in sorafenib-treated HCC cells. (l) Linc01056 increased fatty acid storage in sorafenib-treated HCC cells.(m) Linc01056 knockdown increased content of C16 intermediates of fatty acid in sorafenib-treated HCC cells. *p < 0.05, **p < 0.01, ***p < 0.001

The acquisition of sorafenib resistance in HCC cells requires a high level of intracellular energy to maintain cell growth and survival under sorafenib pressure [30]. We found that Linc01056 knockdown in MHCC97L and PLC/PRF/5 HCC cells resulted in higher levels of intracellular ATP, indicating that the metabolic switch from glycolysis towards FAO may confer an advantage on energy production in HCC cells (Fig. 3f). We then profiled the metabolic characteristics of sorafenib-treated HCC cells with Linc01056 knockdown using Seahorse XF assays. Knockdown of Linc01056 in HCC cells significantly increased the oxygen consumption rate (OCR) and decreased the extracellular acidification rate (ECAR), confirming that HCC cells with Linc01056 knockdown preferentially utilize oxidative phosphorylation (OXPHOS) instead of glycolysis to generate ATP (Fig. 3g and h). In addition, the mitochondrial reactive oxygen species (mtROS) level was markedly elevated upon Linc01056 knockdown in HCC cells (Fig. 3i). Knockdown of Linc01056 significantly increased the basal and maximal respiratory capacities of sorafenib-treated HCC cells (Fig. 3j and k). Knockdown of Linc01056 in HCC cells resulted in reduced glucose uptake and suppressed extracellular lactic acid production, as well as cellular LDH activity (Fig. S3d–f), but significantly increased the consumption of intracellular free fatty acids (Fig. 3l). Metabolic profiling of fatty acids derived from sorafenib-treated MHCC97L cells with Linc01056 knockdown suggested an increased content of C16 intermediates compared with that in the vector control counterpart cells (Fig. 3m). Collectively, these observations suggested that Linc01056 knockdown resulted in a metabolic switch from glycolysis towards FAO in sorafenib-treated HCC cells that increased the efficiency of energy production.

Linc01056 loss-induced FAO is required for the acquisition of sorafenib resistance in HCC

Both glycolysis and FAO have been reported to be hyperactive in HCC cells [31, 32]. To explore whether the Linc01056 knockdown-induced metabolic reprogramming from glycolysis towards FAO is due to direct inhibition of glycolysis by Linc01056, we cotreated HCC cells with the glycolytic inhibitor 2-DG in the presence of sorafenib. Inhibition of glycolysis by 2-DG sensitized HCC cells to sorafenib (Fig. S4a). Consistent with this finding, apoptosis induction by sorafenib treatment in HCC cells was increased in the presence of 2-DG regardless of Linc01056 knockdown surprisingly (Fig. S4b). These results suggested that alone, inhibition of glycolysis in sorafenib-treated HCC cells was not sufficient to induce sorafenib resistance. Hence, we hypothesized that Linc01056 knockdown confers sorafenib resistance through direct activation of FAO in sorafenib-treated HCC cells. Increased fatty acid uptake was observed in HCC cells with Linc01056 knockdown (Fig. 4a), as was enhanced expression of genes related to fatty acid uptake (Fig. S4c). Interestingly, we did not observe significant changes in the expression of genes associated with de novo lipogenesis (Fig. S4d). These observations suggested that the Linc01056 knockdown-associated metabolic switch was directly related to the activation of fatty acid β-oxidation.

Fig. 4

FAO inhibition sensitised Linc01056-knockdowned HCC cells upon sorafenib treatment. (a) Knockdown of Linc01056 resulted in an increase in fatty acid uptake in HCC cells, which was further augmented under sorafenib treatment. FAO suppression by etomoxir (b) increased cytotoxicity of sorafenib, (c) suppressed colonic capacity, and (d) induced apoptosis in HCC cells with Linc01056 knockdown. FAO suppression by etomoxir (e & f) reduced in vivo tumour growth and (g & h) end-point tumour size, (i) increased expression of cell apoptosis marker cleaved caspase-3 and reduced the cell proliferation marker Ki67 in sorafenib-treated in vivo HCC tumours without Linc01056 knockdown. *p < 0.05, **p < 0.01, ***p < 0.001

To test whether FAO activation induced by Linc01056-KD contributes to sorafenib resistance in HCC cells, we applied etomoxir, a CPT1 inhibitor that blocks fatty acid transport and utilization in mitochondria, to suppress FAO. Etomoxir showed a minimal effect on sorafenib sensitivity in HCC cells transduced with vector but significantly improved the HCC cell response to sorafenib in HCC cells with Linc01056 knockdown (Fig. 4b). Similarly, etomoxir strongly reduced the colony formation capacity of HCC cells with Linc01056 knockdown under sorafenib treatment (Fig. 4c). Treatment of etomoxir re-sensitized the Linc01056 knockdown HCC cells by increasing the sorafenib-induced apoptosis (Fig. 4d). Moreover, etomoxir further increased the inhibitory effect of sorafenib on the in vitro motility and invasion of HCC cells with Linc01056 knockdown (Fig. S5a & S5b). In vivo, etomoxir was applied to investigate the role of FAO activation in Linc01056 knockdown-induced sorafenib resistance in the orthotopic HCC mouse model. Etomoxir treatment improved the sorafenib response of orthotopic HCC tumours by suppressing their growth (Fig. 4e, f) and resulted in smaller tumour sizes (Fig. 4g, h). The improvement in the sorafenib response induced by etomoxir in HCC tumours with Linc01056 knockdown was further supported by the reduced expression of Ki67 and increased level of cleaved caspase-3 in the tumour tissues (Fig. 4i). Furthermore, etomoxir further enhanced the inhibitory effect of sorafenib on the lung metastasis of MHCC97L cells with Linc01056 knockdown (Fig. S5c). These observations confirmed that FAO activation played an essential role in mediating sorafenib resistance in HCC cells upon Linc01056 knockdown.

Linc01056 loss activates PPARα-mediated transcription of FAO-associated genes

Given that Linc01056 functions in regulating sorafenib sensitivity by maintaining intracellular energy metabolism homoeostasis, we hypothesized that Linc01056 critically regulates PPARα, the cellular sensor that suppresses glycolysis, while inducing FAO activation [33]. Indeed, knockdown of Linc01056 in HCC cells resulted in transcriptional activation of PPARα-specific target genes, including EHHADH, ACAA1 and ACOX1 (Fig. S6a) and resulted in nuclear localization of PPARα (Fig. S6b), confirming the transcriptional activation of PPARα. The ChIP results suggested that PPARα bound to the transcriptional activation binding site in the promoter regions of FAO-related genes (Fig. 5a) upon Linc01056 knockdown, a finding that confirmed the regulatory role of Linc01056 in PPARα transcriptional activity. LncRNAs regulate the activity of transcription factors via multiple mechanisms [34]. Interestingly, we did not observe obvious changes in the mRNA and protein expression of PPARα (Fig. S6c & Fig. 5b), suggesting that Linc01056 regulates PPARα transcriptional activity through posttranslational mechanisms. In situ hybridization revealed that Linc01056 localized to the cytoplasm but not the nucleus in MHCC97L cells regardless of sorafenib treatment (Fig. 5c). In addition, the RIP assay results showed that cytoplasmic Linc01056 could specifically bind to PPARα but not PPARɤ or FOXO1 in MHCC97L cells (Fig. 5d). Moreover, the immunoprecipitation assay results confirmed that PPARα bound to Linc01056 but not another lncRNA, MALAT1 (Fig. 5e). These observations indicated that Linc01056 specifically bound to PPARα to prevent its nuclear localization and transcriptional activity.

Fig. 5

Linc01056 interfered PPARα transcription activity-associated FAO activation. (a) Linc01056 knockdown induced binding of PPARα on to the promoter region of FAO-related genes. (b) Linc01056 knockdown did not change the protein expression of PPARα in HCC cells. (c) Linc01056 located in the cytoplasm of MHCC97L cells with or without sorafenib treatment. (d) Linc01056 specifically bound to PPARα but not PPARγ or FOXO1. (e) PPARα specifically bound to Linc01056 but not other lncRNA like MALAT1. Suppression of PPARα activity by GW6471 (f) reversed the increase of OCR and (g) decrease of ECAR in Linc01056-knockdown HCC cells upon sorafenib treatment. GW6471 inhibited (h) production of mitochondrial ROS, (i) intracellular ATP, (j) cell viability, (k) colonic capacity, and (l) apoptosis of Linc01056-knockdown HCC cells upon sorafenib treatment. *p < 0.05, **p < 0.01, ***p < 0.001

To explore the role of PPARα in mediating FAO activation upon Linc01056 knockdown, we treated HCC cells with the PPARα-specific inhibitor GW6471. Treatment with GW6471 significantly reversed the Linc01056 knockdown-induced increase in the OCR in sorafenib-treated HCC cells, while the increase of ECAR upon PPARα inhibition was subtle (Fig. 5f and g). Moreover, GW6471 potently decreased the basal and maximal respiratory capacities of sorafenib-treated HCC cells with Linc01056 knockdown (Fig. S6d & S6e). Thus, the induction of mtROS production upon Linc01056 knockdown was strongly inhibited by GW6471 (Fig. 5h), and intracellular ATP production was partially reduced upon GW6471 treatment (Fig. 5i). These observations confirmed the role of PPARα in mediating FAO activation upon Linc01056 knockdown in sorafenib-treated HCC cells. Furthermore, inhibition of PPARα by GW6471 in HCC cells with Linc01056 knockdown restored sorafenib sensitivity, as measured by a cell viability assay (Fig. 5j), and suppressed colony formation (Fig. 5k). Sorafenib-induced apoptosis was further increased in HCC cells with Linc01056 knockdown by GW6471 treatment (Fig. 5l). In addition, GW6471 treatment further increased the inhibitory effect of sorafenib on the in vitro motility and invasion of HCC cells with Linc01056 knockdown (Fig. S6f & S6g). Collectively, these observations suggested that activation of PPARα-associated gene transcription mediates FAO activation in sorafenib-treated HCC cells with Linc01056 knockdown.

Clinicopathological significance of Linc01056 in HCC

To determine the clinicopathological significance of Linc01056 in HCC, we examined the expression of Linc01056 and Linc01056-related signalling molecules that we identified in this study using a combination of in situ hybridization and multiplex immunofluorescence. The tissue array containing tumour sections from 90 patients was analysed (Fig. S7a; representative image in Fig. 6a, patient information in Supplementary Table S3). We found that the cytoplasmic expression of Linc01056 was significantly downregulated in HCC tumour tissue compared to non-tumour adjacent tissue (Fig. 6b), consistent with the data reported from two other HCC patient cohorts, GSE62232 and GSE76297 (Fig. S7b & S7c). The expression of Linc01056 in HCC tissues and overall survival or progression-free survival in the patients are negatively correlated (Fig. 6c and 6d). Also, the level of Linc01056 is significantly lower in patients with recurrence (Fig. S7d). However, Linc01056 expression was not associated with HCC stage (Fig. S7e) or tumour size (Fig. S7f). To identify the clinical correlation between Linc01056 expression and the expression of the signalling molecules identified in this study, we performed staining for PPARα and the fatty acid transporter CD36 and CPT1, as representatives of FAO activity, in HCC tumours and quantified their expression levels. We found significant negative correlations between Linc01056 expression and the expression of CPT1 (Fig. 6e). The expression of Linc01056 was positively correlated with cytoplasmic localisation of PPARα (Fig. 6f). Notably, PPARα was expressed significantly higher in the sorafenib non-responder group compared to the responder group, as observed in the published patient cohort GSE109211 (Fig. S7g) Consistently, expression of PPARα in HCC was positively correlated with FAO-related CPT1 and CD36 expression from our immunostaining (Fig. S7h & S7i). Phospho-Erk was previously reported as a predictive marker of the sorafenib response in HCC patients, and we observed that Linc01056 expression was positively correlated with the phospho-Erk level in HCC tissues [35], indicating the clinical association of Linc01056 expression with the sorafenib response in HCC patients (Fig. 6g). Collectively, our findings indicate the clinicopathological significance of Linc01056 expression in HCC.

Fig. 6

Clinicopathological significance of Linc01056 in HCC. (a) representative image of multiplex staining on tissue microarray of human HCC samples. (b) Expression of Linc01056 was significantly downregulated in HCC compared to adjacent liver. High expression of Linc01056 predicted the good prognosis of (c) overall survival and (d) progression-free survival of HCC patients. (e) Expression of Linc01056 was negatively correlated with FAO marker CPT1 in human HCC samples. (f) Expression of Linc01056 was positively correlated with the cytoplasmic localization of PPARα. (g) Expression of Linc01056 was positively correlated with the sorafenib sensitivity marker p-Erk. *p < 0.05, ***p < 0.001