LINC00982-encoded protein PRDM16-DT regulates CHEK2 splicing to suppress colorectal cancer metastasis and chemoresistance

Abstract

Metastasis is one of the key factors of treatment failure in late-stage colorectal cancer (CRC). Metastatic CRC frequently develops resistance to chemotherapeutic agents. This study aimed to identify the novel regulators from "hidden" proteins encoded by long noncoding RNAs (lncRNAs) involved in tumor metastasis and chemoresistance. Methods: CRISPR/Cas9 library functional screening was employed to identify the critical suppressor of cancer metastasis in highly invasive CRC models. Western blotting, immunofluorescence staining, invasion, migration, wound healing, WST-1, colony formation, gain- and loss-of-function experiments, in vivo experimental metastasis models, multiplex immunohistochemical staining, immunohistochemistry, qRT-PCR, and RT-PCR were used to assess the functional and clinical significance of FOXP3, PRDM16-DT, HNRNPA2B1, and L-CHEK2. RNA-sequencing, co-immunoprecipitation, qRT-PCR, RT-PCR, RNA affinity purification, RNA immunoprecipitation, MeRIP-quantitative PCR, fluorescence in situ hybridization, chromatin immunoprecipitation and luciferase reporter assay were performed to gain mechanistic insights into the role of PRDM16-DT in cancer metastasis and chemoresistance. An oxaliplatin-resistant CRC cell line was established by in vivo selection. WST-1, colony formation, invasion, migration, Biacore technology, gain- and loss-of-function experiments and an in vivo experimental metastasis model were used to determine the function and mechanism of cimicifugoside H-1 in CRC. Results: The novel protein PRDM16-DT, encoded by LINC00982, was identified as a cancer metastasis and chemoresistance suppressor. The down-regulated level of PRDM16-DT was positively associated with malignant phenotypes and poor prognosis of CRC patients. Transcriptionally regulated by FOXP3, PRDM16-DT directly interacted with HNRNPA2B1 and competitively decreased HNRNPA2B1 binding to exon 9 of CHEK2, resulting in the formation of long CHEK2 (L-CHEK2), subsequently promoting E-cadherin secretion. PRDM16-DT-induced E-cadherin secretion inhibited fibroblast activation, which in turn suppressed CRC metastasis by decreasing MMP9 secretion. Cimicifugoside H-1, a natural compound, can bind to LEU89, HIS91, and LEU92 of FOXP3 and significantly upregulated PRDM16-DT expression to repress CRC metastasis and reverse oxaliplatin resistance. Conclusions: lncRNA LINC00982 can express a new protein PRDM16-DT to function as a novel regulator in cancer metastasis and drug resistance of CRC. Cimicifugoside H-1 can act on the upstream of the PRDM16-DT signaling pathway to alleviate cancer chemoresistance.

Keywords: Cimicifugoside H-1; Colorectal cancer; Drug resistance.; Fibroblasts; PRDM16-DT.

Figure 1

PRDM16-DT encoded by LINC00982 is identified as a cancer metastasis suppressor and is of clinical significance in CRC. (A) Schematic diagram of CRISPR/Cas9 screening (B) Heatmap of lncRNAs involved in cancer metastasis. (C) Read counts of LINC00982 sgRNAs are significantly increased in sgRNA-transduced cells (KO) as compared to input cells. (D) QRT-PCR assay displays markedly decreased LINC00982 expression in I8 cells compared to their parental cells. n = 3/experiments. (E) Structure of the PRDM16-DT-GFP fusion plasmids. Mut means ATG (start codon) is mutated to ATT. (F-G) PRDM16-DT-GFP fusion protein in ORF-GFPmut-transfected cells using GFP fluorescence (F) and Western blotting (G). n = 3/experiments. (H) Structure of the Flag fusion plasmids. Mut means ATG (start codon) is mutated to ATT. (I-J) Using confocal microscopy (I) and Western blotting (J), PRDM16-DT-Flag fusion protein expression is detected in ORF-Flag-transfected cells. n = 3/experiments. (K) Western blotting showing lower PRDM16-DT expression in CRC cells than NCM460 cells (normal colonic epithelial cells), and a further decrease in highly invasive I8 cells. n = 3/experiments. (L) Higher PRDM16-DT expression level in normal tissues than in CRC tissues (n = 14, P < 0.01). (L) Different staining scores for PRDM16-DT in CRC tissues. Scale bar, 50 μm. (N-O) IHC assay showing higher PRDM16-DT expression in most normal tissues than primary tumors (T1 & 2, T3 & 4) and metastatic tumors (L). Scale bar, 50 μm. (P) Based on PRDM16-DT expression, Kaplan-Meier survival analysis shows a shorter survival in low PRDM16-DT-expressing patients than those with high PRDM16-DT expression (n = 93, P = 0.04). (Q) Representative staining results of PRDM16-DT expression in CRC and normal tissues. Scale bar, 50 μm. (R) Comparison of PRDM16-DT expression in 93 cases of CRC primary tumor tissues and 87 cases of adjacent normal tissues. Bars, SD; **, P < 0.01; ***, P < 0.001.

Figure 2

PRDM16-DT inhibits CRC metastasis. (A) Successful establishment of stable and doxycycline (DOX)-inducible ORF-overexpressing and ORFmut-overexpressing cell lines. Cells were treated with DOX (1 mg/mL) for 48 hours. n = 3/experiments. (B) PRDM16-DT significantly inhibits CRC migration and invasion ability. Cells were treated with DOX (1 mg/mL) for 48 h. n = 3/experiments. Scale bar, 100 μm. (C) Successful knockdown of LINC00982 in CRC cells. n = 3/experiments. (D-E) LINC00982 knockdown cells were transfected with ORF and ORFmut plasmids for 48 h and migration assay revealed that LINC00982 knockdown with shRNA sequence significantly promotes the CRC migration and invasion ability, restores expression with ORF, and significantly abrogates the promotion effect of silencing LINC00982. n = 3/experiments. Scale bar, 100 μm. (F-G) Bioluminescence imaging (n = 6 mice/group) (F), and H&E staining (n = 3 mice/group) (G) show that PRDM16-DT reduces metastatic potential in mice. Mice were fed with either DOX containing or normal diet. Scale bar, 100 μm. Bars, SD; **, P < 0.01; ***, P < 0.001; ns, no significant difference.

Figure 3

PRDM16-DT, transcriptionally regulated by FOXP3, suppresses cancer metastasis by interacting with HNRNPA2B1. (A) Predicted IRF2, YY1, HNF-1, CEBPA, and FOXP3-binding motifs located at the promoter of PRDM16-DT. CRC cells were treated with siRNAs (100 nM) of IRF2, YY1, HNF-1, CEBPA, and FOXP3 for 48 h. qRT-PCR assay shows marked upregulation of PRDM16-DT expression by siFOXP3. n = 3/experiments. (B) ChIP assay to determine FOXP3 binding with -917/-912 region of the PRDM16-DT promoter. n = 3/experiments. (C) Increased transcriptional activity of construct with -917/-912 deleted (ΔPRDM16-DT) in CRC cells. FL: Firefly luciferase; RL: Renilla luciferase. n = 3/experiments. (D) Increased transcriptional activity of the PRDM16-DT reporter construct with the intact region (-917/-912) following FOXP3 knockdown. n = 3/experiments. (E) Upregulation of PRDM16-DT in CRC cells following FOXP3 knockdown. n = 3/experiments. (F) PRDM16-DT immuno-precipitates in CRC cells on a gel. (G) Enrichment of RNA splicing in the proteins interacting with PRDM16-DT. n = 3/experiments. (H) Co-IP assay with anti-HNRNPA2B1 antibodies showing the endogenous interaction of HNRNPA2B1 with PRDM16-DT, even in the presence of RNAase treatment (30 mg/mL RNAase for 30 min). IgG was used as a control. n = 3/experiments. (I) An HA tag was added to the truncated fragments of HNRNPA2B1, and Co-IP assay with anti-HA antibodies showing reduced binding between PRDM16-DT and HNRNPA2B1 upon GRD domain deletion. IgG was used as a control. n = 3/experiments. (J) Co-IP assay to analyze the binding of PRDM16-DT with wild-type HNRNPA2B1 and its AGG mutant in CRC cells. n = 3/experiments. (K-L) HNRNPA2B1 overexpression in CRC cells treated with DOX (1 mg/mL, 48 h) (K) Significant abrogation of the inhibitory effect of PRDM16-DT on CRC cells by HNRNPA2B1 (L). n = 3/experiments. Scale bar, 100 μm. Bars, SD; **, P < 0.01; ***, P < 0.001; ns, no significant difference.

Figure 4

PRDM16-DT promotes the formation of long CHEK2 (L-CHEK2). (A) PRDM16-DT-mediated AS events of mRNAs analyzed by RNA-seq. (B-C) RNA-seq showing the inclusion of CHEK2 exon 9 (E9) promoted by PRDM16-DT. (D-E) MeRIP-qPCR analysis to validate the higher m6A modification of CHEK2 in I8 cells (D) and tumor tissues (E) than parental cells and normal tissues. n = 3/experiments. (F) RT-PCR assay showing the inclusion of exon 9 of CHEK2 promoted by PRDM16-DT. n = 3/experiments. (G) LINC00982 knockdown cells were transfected with the indicated plasmid, and the RT-PCR assay showed that knockdown of LINC00982 suppressed the inclusion of exon 9 of CHEK2, restored expression with ORF but not ORFmut reversed the effect. n = 3/experiments. (H) CRC cells treated with DOX (1 mg/mL) for 48 hours were transfected with HNRNPA2B1 overexpression plasmid. RT-PCR assay showing that HNRNPA2B1 determined the PRDM16-DT-regulated splicing of CHEK2. (I) RNA pulldown assay showing a strong interaction between HNRNPA2B1 and E9 (50-80) labeled with m6A modification. n = 3/experiments. (J) RIP assay confirming the weaker binding of HNRNPA2B1 and E9 with mutation on its m6A-binding site. n = 3/experiments. (K) Cells were transfected with ORF plasmid (0 μg, 1 μg, 2 μg). RNA pulldown assay shows decreased binding of HNRNPA2B1 to CHEK2 in an ORF dose-dependent manner. n = 3/experiments. (L) ORF and ORFmut vectors were transfected into CRC cells. RNA pulldown assay shows markedly reduced binding of HNRNPA2B1 to CHEK2 exon 9 by ORF but not ORFmut. n = 3/experiments. (L) Lack of the AGG mutant of HNRNPA2B1 binding to CHEK2. n = 3/experiments. (N) FISH assay showing HNRNPA2B1 co-localization with CHEK2. n = 3/experiments. Scale bar, 5 μm; 2.5 μm. Bars, SD; **, P < 0.01; ***, P < 0.001.

Figure 5

L-CHEK2, not S-CHEK2, suppresses metastasis of CRC. (A-B) siRNA (100 nM) of CHEK2 with L-CHEK2 or S-CHEK2 vector were transfected in CRC cells (A); increased cell migration and invasion by CHEK2 silencing, which can be repressed to the control level upon re-expression with L-CHEK2 (B). n = 3/experiments. Scale bar, 100 μm. (C-D) CRC cells with stable overexpression of L-CHEK2 or S-CHEK2 were intravenously injected into mice via the tail vein, after 4 weeks, the reduced metastatic potential in mice by L-CHEK2 was analyzed by bioluminescence imaging (n = 6 mice/group) (C), and H&E staining (n = 3 mice/group) (D), Scale bar, 100 μm. (E) Knockdown of L-CHEK2 in CRC cells treated with DOX (1 mg/mL, 48 h) showing efficient restoration of cell metastasis to control levels reduced by ORF overexpression. n = 3/experiments. Scale bar, 100 μm. (F) QRT-PCR assay showing decreased L-CHEK2 mRNA levels in tumor tissues compared to normal tissues. Bars, SD; **, P < 0.01; ***, P < 0.001; ns, no significant difference.

Figure 6

PRDM16-DT promotes E-cadherin secretion to inhibit fibroblast activation in a paracrine manner. (A) Heatmap of gene profiles in L-CHEK2 and S-CHEK2-overexpressing CRC cells. (B) Volcano plot of DEPs regulated by L-CHEK2. (C) GSEA analysis showing EMT enrichment in DEGs of L-CHEK2-overexpressing cells. (D) Expression of EMT markers (Twist, E-cadherin) in L-CHEK2-overexpressing cells compared with S-CHEK2-overexpressing and control cells. n = 3/experiments. (E) Decreased Twist expression and increased E-cadherin expression in ORF-overexpressing CRC cells, with no expression change in ORFmut cells. n = 3/experiments. (F) Reversal of effects of EMT markers in ORF-overexpressing cells by L-CHEK2 silencing. n = 3/experiments. (G-H) Increased E-cadherin secretion in the conditioned medium (CM) from ORF-overexpressing CRC cells compared to control cells by ELISA (G) and Western blotting (H). n = 3/experiments. (I) Decreased expression of α-SMA and FAP in the CM from ORF-overexpressing CRC cells by Western blotting. n = 3/experiments. (J) Schematic diagram for coculture system. (K) Invasive ability of CRC cells treated with CM from different fibroblasts as indicated by the metastasis assay. n = 3/experiments. Scale bar, 100 μm. (L) Decreased MMP9 secretion in fibroblasts cocultured with CM from ORF-overexpressing CRC cells compared to control cells by ELISA. n = 3/experiments. (L) MMP9 overexpression in fibroblasts. n = 3/experiments. (N) Metastasis assay showing MMP9 overexpression in fibroblasts and reversal of the inhibitory effect of fibroblasts cultured with CM from ORF-overexpressing CRC cells on the metastasis ability of CRC cells as indicated. n = 3/experiments. Scale bar, 100 μm. Bars, SD; **, P < 0.01; ***, P < 0.001.

Figure 7

Identification of cimicifugoside H-1 as a novel sensitizer for CRC oxaliplatin treatment. (A) HCT116-Luc-OR cells transfected with PRDM16-DT-expressing plasmid (1 μg) or vector control were treated with oxaliplatin (20 μM) for 48 h. Metastasis assay showing significant and synergistic suppression of HCT116-Luc-OR metastasis by the combined treatment of PRDM16-DT overexpression with oxaliplatin. n = 3/experiments. Scale bar, 100 μm. (B) Low endogenous expression of PRDM16-DT in OR CRC tissue and cells. n = 3/experiments. (C) Identification of anticancer drugs from the library comprising 50 plant natural products (10 μM, 24h) using the invasion assay. Significant inhibitory effect of cimicifugoside H-1, among the 50 compounds tested on the invasion of HCT116 cells. n = 3/experiments. (D) Increased PRDM16-DT expression in CRC cells by cimicifugoside H-1 (10 μM, 20 μM, 24 h) as evidenced by Western blotting. n = 3/experiments. (E) Increased E-cadherin secretion by cimicifugoside H-1 (10 μM, 20 μM, 24 h). n = 3/experiments. (F-G) Migration assay showed that Significant inhibition of invasion and migration of CRC cells (F) and OR CRC cells (G) by cimicifugoside H-1 (10 μM, 20 μM, 24 h). n = 3/experiments. Scale bar, 100 μm. (H) Molecular docking to predict the binding sites of cimicifugoside H-1 with FOXP3 protein. The potential binding sites are labeled. (I) Establishment of FOXP3-deficient CRC cells. n = 3/experiments. (J) Overexpression of wild-type OR mutated FOXP3 in FOXP3-deficient CRC cells followed by metastasis assays to compare their sensitivity toward cimicifugoside H-1 treatment (10 μM, 24 h). n = 3/experiments. Scale bar, 100 μm. (K) Binding of FOXP3 with cimicifugoside H-1 (Kd=5.2×10^-8) as determined by the Biacore assay. n = 3/experiments. FOXP3-deficient CRC and control cells were intravenously injected into mice via the tail vein. After one week, mice were divided into two subgroups for treatment with cimicifugoside H-1 (20 mg/kg) or the vehicle (0.5% CMC-Na). and lung metastasis was monitored. (L) No inhibition of lung metastasis by cimicifugoside H-1 in FOXP3-knockout cells. n = 6 mice/group. (L) H&E staining of the lungs. Mice were divided into three groups and treated by oral administration with vehicle (0.5% CMC-Na), cimicifugoside H-1 (20 mg/kg), or oxaliplatin (20 mg/kg). n = 3 mice/group. Scale bar, 100 μm. (N) HCT116-Luc-OR and control cells were intravenously injected into mice via the tail vein. After one week, mice were divided into three groups for treatment with cimicifugoside H-1 (20 mg/kg), oxaliplation (20 mg/kg) or the vehicle (0.5% CMC-Na), and lung metastasis was monitored. Superior suppressive effect of cimicifugoside H-1 compared to oxaliplatin on tumor metastasis. n = 6 mice/group. (O) H&E staining of the lungs. n = 3 mice/group. Scale bar, 100 μm. Bars, SD; **, P < 0.01; ***, P < 0.001.

Figure 8

Schematic diagram summarizing cancer metastasis and chemoresistance suppression by PRDM16-DT.