Comprehensive analysis of RNA-binding protein SRSF2-dependent alternative splicing signature in malignant proliferation of colorectal carcinoma

Abstract

Aberrant expression of serine/arginine-rich splicing factor 2 (SRSF2) can lead to tumorigenesis, but its molecular mechanism in colorectal cancer is currently unknown. Herein, we found SRSF2 to be highly expressed in human colorectal cancer (CRC) samples compared with normal tissues. Both in vitro and in vivo, SRSF2 significantly accelerated the proliferation of colon cancer cells. Using RNA-seq, we screened and identified 33 alternative splicing events regulated by SRSF2. Knockdown of SLMAP-L or CETN3-S splice isoform could suppress the growth of colon cancer cells, predicting their role in malignant proliferation of colon cancer cells. Mechanistically, the in vivo crosslinking immunoprecipitation assay demonstrated the direct binding of the RNA recognition motif of SRSF2 protein to SLMAP and CETN3 pre-mRNAs. SRSF2 activated the inclusion of SLMAP alternative exon 24 by binding to constitutive exon 25, while SRSF2 facilitated the exclusion of CETN3 alternative exon 5 by binding to neighboring exon 6. Knockdown of SRSF2, its splicing targets SLMAP-L, or CETN3-S caused colon cancer cells to arrest in G1 phase of the cell cycle. Rescue of SLMAP-L or CETN3-S splice isoform in SRSF2 knockdown colon cancer cells could effectively reverse the inhibition of cell proliferation by SRSF2 knockdown through mediating cell cycle progression. Importantly, the percentage of SLMAP exon 24 inclusion increased and CETN3 exon 5 inclusion decreased in CRC samples compared to paired normal samples. Collectively, our findings identify that SRSF2 dysregulates colorectal carcinoma proliferation at the molecular level of splicing regulation and reveal potential splicing targets in CRC patients.

Keywords: SRSF2; alternative splicing; colorectal cancer; proliferation.

Figure 1

SRSF2 is upregulated in human colorectal cancer samples.A, high expression of SRSF2 mRNA levels in CRC samples were revealed by qRT-PCR analysis. qRT-PCR was performed in 51 paired CRC RNA samples. SRSF2 mRNA expression levels were normalized to β-actin. The bar value is the log ratio of SRSF2 mRNA levels between CRC (T) and paired normal tissues (N). p < 0.0001. B, box plot compared SRSF2 mRNA levels in normal colorectal tissues (789 samples) and colorectal tumors (620 samples) in the published TCGA datasets and GTEx datasets. ∗∗∗p < 0.001. C, IHC staining of SRSF2 protein in representative CRC tumor tissues and adjacent nontumor tissues. Scale bar, 100 μm. CRC, colorectal cancer; qRT-PCR, quantitative reverse transcriptase PCR; SRSF2, serine/arginine-rich splicing factor 2; TCGA, The Cancer Genome Atlas.

Figure 2

Knockdown of SRSF2 inhibits the proliferation of colon cancer cells in vitro and in vivo.A, colon cancer cell lines RKO or HT29 cells were stably knockdown using lentiviruses transfected with SRSF2 shRNA (sh-SRSF2#1, sh-SRSF2#2) or negative control shRNA (sh-Luci). SRSF2 knockdown efficiency was assessed by Western blot. B and C, growth curve of RKO (B) or HT29 (C) cells after stably transfected with indicated shRNAs. ∗∗p < 0.01, ∗∗∗p < 0.001. D and E, clonogenic survival assay was performed, and crystal violet staining with representative cells were shown (D). The relative number of focal adhesions described in (D) was quantified as mean ± SD in the bar graph (E). ∗∗p < 0.01, ∗∗∗p < 0.001. F, the representative images of EdU staining assay for cells described in (A) were shown (Left). The results were presented as mean ± SD (Right). ∗∗p < 0.01. G and H, RKO cells stably expressing sh-SRSF2 and sh-Luci were injected into nude mice. Four weeks after injection, tumors excised from the mice were shown (G). Tumor volumes were measured every week, and time course of xenograft tumors growth was shown (H). ∗∗∗p < 0.001. I, mice weights. Results are shown as mean ± SD of mice weights (each with five mice). ns: no significance. J, weight of tumors excised from the mice. Results are shown as mean ± SD of tumor weights (each with initial ten injections). ∗p < 0.05. EdU, 5-ethynyl-2′-deoxyuridine; SRSF2, serine/arginine-rich splicing factor 2.

Figure 3

Alternative splicing profiles affected by SRSF2 in RKO cells.A, RNA-seq strategy was performed using RKO cells transiently transfected with siSRSF2 or siNC independently. B, quantification of SRSF2-affected differential alternative splicing events, as revealed by analysis of RNA-seq data. Among these events, there are 108 cassette, 6 A5SS, 4 A3SS, 4 MEX, 7 IR, 10 Cassette Multi, 14 AltStart, and 20 AltEnd alternative splicing events. C, gene ontology analysis of SRSF2-targeted 173 splicing events. AS, alternative splicing; SRSF2, serine/arginine-rich splicing factor 2.

Figure 4

Validation of SRSF2-regulated differential alternative splicing events in RKO cells.A, heatmap of the significant splicing profiling data by RNA-seq among the RKO/siNC (WT) and RKO/siSRSF2 (KD) groups (FDR < 0.05, Left), and the heatmap of the validated AS events were shown (Right). The data were sorted by the mean value of the indicated WT and KD groups analyzed. Green indicated inclusion, and red indicated exclusion. B and C, representative exon out events regulated by SRSF2 with RT-PCR results, RNA-seq reads coverage, and quantification of the RNA products measured as PSI (percent of splicing in). Note that alternative exons for SRSF2-mediated exclusion were marked in red (B). Representative exon in events affected by SRSF2 were shown and the alternative exons for inclusion were marked in green (C). D, correlation between ΔPSI from RNA-seq AS analysis and 33 RT-PCR validated splicing events. AS, alternative splicing; SRSF2, serine/arginine-rich splicing factor 2.

Figure 5

Knockdown of SLMAP-L splice isoform inhibited coloncancercell growth in vitro.A, schematic diagram of SLMAP splice variants including or lacking alternative exon 24 (SLMAP-L and SLMAP-S). B, RKO cells were stably knockdown using lentiviruses transfected with isoform-specific shRNAs, which targeted against the SLMAP-L or SLMAP-S variants independently. SLMAP knockdown efficiency using sh-SLMAP-L or sh-SLMAP-S compared with sh-Luci in colon cancer RKO cells was assessed by RT-PCR analysis. The quantification of PSI was shown under the RT-PCR results. C, crystal violet staining with representative cells described in (B) in high and low density were shown after clonogenic survival assay performed (Left). The quantification of focal adhesions was shown as mean ± SD in the bar graph (Right). ns: no significance, ∗p < 0.05. D, the representative images of EdU staining assay for cells described in (B) were shown (Left). The results were presented as mean ± SD (Right). ns: no significance, ∗∗p < 0.01. E, growth curve of RKO cells after stably transfected with indicated shRNas. ns: no significance, ∗∗p < 0.01. EdU, 5-ethynyl-2′-deoxyuridine.

Figure 6

SLMAP exon 24 is subject to the regulation of multiple splicing regulators including SRSF2 and binds with the RRM domain of SRSF2.A, diagrams for detection of SLMAP variants including or lacking alternative exon 24 (E24+ or E24−). The product sizes for the two variants of SLMAP are shown. B and C, the indicated siRNAs were transiently transfected into RKO cells, and the RNAs were extracted for RT-PCR analysis of SLMAP exon 24 inclusion/skipping (B). Each PSI value quantification from RT-PCR results of SLMAP splice variants was shown as mean ± SD in the bar graph (C). ns: no significance, ∗∗p < 0.01, ∗∗∗p < 0.001. D, the knockdown efficiency of indicated SR or hnRNP proteins using siRNAs compared with si-NC in RKO cells were assessed by qRT-PCR analysis. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. E, schematic diagram of RRM and RS domains in SRSF2, and constructions of two SRSF2 mutants: ΔRRM (deleting RRM domain), ΔRS (deleting RS domain). Both the mutants and SRSF2-WT plasmids were HA tagged. F, the indicated plasmids were transiently transfected into RKO cells, and RT-PCR was performed to analyze the inclusion/skipping of SLMAP exon 24. Each PSI value quantification from RT-PCR results was shown as mean ± SD in the bar graph. ns: no significance, ∗∗p < 0.01, ∗∗∗p < 0.001. G, Western blot of the endogenous SRSF1, SRSF2, hnRNPM with anti-SRSF1, anti-SRSF2, and anti-hnRNPM antibodies independently and the exogenous SRSF1, SRSF2, and hnRNPM with anti-HA antibodies. H, upper, diagram for the specific primers designed according to exon-intron (E-I) boundary sequences to detect exons 23 to 25 and introns 23 and 24. Bottom, RT-PCR analysis following CLIP assay indicated the direct binding between indicated proteins and endogenous SLMAP RNA fragments. CLIP, crosslinking immunoprecipitation; RRM, RNA recognition motif; RS, serine/arginine-rich domain; SRSF2, serine/arginine-rich splicing factor 2.

Figure 7

Knockdown of SRSF2 or SLMAP-L induces G1 arrest in cell cycle progression.A, cell cycle was analyzed in SRSF2 knockdown RKO cells (sh-SRSF2) and the control RKO cells (sh-Luci). The quantification of the representative DNA content was shown as mean ± SD in the bar graph. ns: no significance, ∗∗p < 0.01, ∗∗∗p < 0.001. B, cell cycle was analyzed in SLMAP-L/SLMAP-S knockdown RKO cells (sh-SLMAP-L/sh-SLMAP-S) and the control RKO cells (sh-Luci). The quantification of representative DNA content was shown as mean ± SD in the bar graph. ns: no significance, ∗∗p < 0.01. C, Western blot of RKO cells knocking down SLMAP-L, SLMAP-S, SRSF2, or sh-Luci using anti-α-Tubulin, anti-Cyclin D1 and anti-CDK6 antibodies, independently. D and E, quantification of the Western blot described in (C). α-Tubulin is used to normalize the results, the relative CCND1 (Cyclin D1), or CDK6 expression of control cells was set as 100%. The data represent three independent experiments, each value was shown as mean ± SD in the bar graph. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. SRSF2, serine/arginine-rich splicing factor 2.

Figure 8

Restoration of SLMAP-L reverses the inhibiting effects of SRSF2 knockdown on RKO cell proliferation.A, RT-PCR detection was performed to analyze the inclusion/skipping of SLMAP exon 24 in SRSF2 knockdown RKO cells stably transfected with SLMAP-L or SLMAP-S. Each PSI value quantification from RT-PCR results was shown under the PCR image. B and C, clonogenic survival assay for cells described in (A) was performed, and crystal violet staining with representative cells were shown (B). The relative number of focal adhesions described in (B) was quantified as mean ± SD in the bar graph (C). ns: no significance, ∗p < 0.05, ∗∗∗p < 0.001. D, growth curve of RKO cells described in (A) was shown. ns: no significance, ∗p < 0.05, ∗∗∗p < 0.001. E and F, the representative images of EdU staining assay for cells described in (A) were shown (E). The results were presented as mean ± SD (F). ns: no significance, ∗p < 0.05, ∗∗p < 0.01. G, cell cycle was analyzed in cells described in (A). The quantification of representative DNA content was shown as mean ± SD in the bar Graph. ns: no significance, ∗∗p < 0.01, ∗∗∗p < 0.001. H, Western blot of cells described in (A) using anti-α-tubulin, anti-Cyclin D1, and anti-CDK6 antibodies independently. I and J, quantification of the Western blot was shown described in (H). α-Tubulin is used to normalize the results, the relative CCND1 (I) or CDK6 (J) expression of control cells was set as 100%. The data represent three independent experiments, each value was shown as mean ± SD in the bar Graph. ns: no significance, ∗p < 0.05, ∗∗p < 0.01. EdU, 5-ethynyl-2′-deoxyuridine; SRSF2, serine/arginine-rich splicing factor 2.

Figure 9

The inclusion of SLMAP exon 24 increase in CRC tumor samples.A and B, representative RT-PCR results for splicing patterns of SLMAP transcripts are shown between colorectal tumors (T) and paired normal tissues (N), and the PSI quantification of each splice variant in SLMAP was shown under the RT-PCR results. Note that alternative exon 24 in SLMAP was marked in green (A). The PSI value quantification from RT-PCR results of SLMAP splice variants were presented as mean ± SD (B). ∗∗p < 0.01. C, schematic diagram of the proposed mechanism by which SRSF2 promotes malignant proliferation of CRC via cell cycle progression through regulating alternative splicing of SLMAP and CETN3 pre-mRNA. CRC, colorectal cancer; SRSF2, serine/arginine-rich splicing factor 2.