Splicing arrays reveal novel RBM10 targets, including SMN2 pre-mRNA

Abstract

Background: RBM10 is an RNA binding protein involved in message stabilization and alternative splicing regulation. The objective of the research described herein was to identify novel targets of RBM10-regulated splicing. To accomplish this, we downregulated RBM10 in human cell lines, using small interfering RNAs, then monitored alternative splicing, using a reverse transcription-PCR screening platform.

Results: RBM10 knockdown (KD) provoked alterations in splicing events in 10-20% of the pre-mRNAs, most of which had not been previously identified as RBM10 targets. Hierarchical clustering of the genes affected by RBM10 KD revealed good conservation of alternative exon inclusion or exclusion across cell lines. Pathway annotation showed RAS signaling to be most affected by RBM10 KD. Of particular interest was the finding that splicing of SMN pre-mRNA, encoding the survival of motor neuron (SMN) protein, was influenced by RBM10 KD. Inhibition of RBM10 resulted in preferential expression of the full-length, exon 7 retaining, SMN transcript in four cancer cell lines and one normal skin fibroblast cell line. SMN protein is expressed from two genes, SMN1 and SMN2, but the SMN1 gene is homozygously disrupted in people with spinal muscular atrophy; as a consequence, all of the SMN that is expressed in people with this disease is from the SMN2 gene. Expression analyses using primary fibroblasts from control, carrier and spinal muscle atrophy donors demonstrated that RBM10 KD resulted in preferential expression of the full-length, exon 7 retaining, SMN2 transcript. At the protein level, upregulation of the full-length SMN2 was also observed. Re-expression of RBM10, in a stable RBM10 KD cancer cell line, correlated with a reversion of the KD effect, demonstrating specificity.

Conclusion: Our work has not only expanded the number of pre-mRNA targets for RBM10, but identified RBM10 as a novel regulator of SMN2 alternative inclusion.

Keywords: Alternative splicing; Cancer; RBM10; SMN2; Spinal muscular atrophy; Splicing array.

Fig. 1

Electropherogram results for KITLG and FN1b ASEs from Array-96 and for SMN2 and SIAHBP1 ASEs from Array-191. The electropherograms on the left are from the control transfections using scrambled siRNA, while the electropherograms on the right are from the RBM10 KD transfections using the siRNA indicated. Each row represents the results from the cell line indicated. The “expected” amplicon size is shown below the electropherogram, while the “found” amplicon size is shown above

Fig. 2

Heatmap and unsupervised hierarchical clustering of genes changed in Array-96 and cell lines used (built using R package ‘pheatmap’). Each row represents the gene that had a change in at least one cell line, and each column represents the cell line used in the experiment. Preferential exon exclusion shown in blue; preferential exon inclusion shown in red. The Euclidean distance metric was used for clustering, while the linkage method used was “average” linkage

Fig. 3

Heatmap and unsupervised hierarchical clustering of genes changed in Array-191 and cell lines used (built using R package ‘pheatmap’). Each row represents the gene that had a change in at least one cell line, and each column represents the cell line used in the experiment. Preferential exon exclusion shown in blue; preferential exon inclusion shown in red. The Euclidean distance metric was used for clustering, while the linkage was done using the “ward.D” algorithm

Fig. 4

Changes associated with RBM10 KD in Array-191. a Venn diagram showing the relative distribution of changes amongst the five cell lines. Modified from http://faculty.ucr.edu/~tgirke/Documents/R. b Summary of changes, by cell line, that occurred in three or more cell lines

Fig. 5

SMN RNA expression in human fibroblasts following transient RBM10 KD. a RBM10 and SMN RNA expression in parental fibroblasts: (i) representative agarose gel following RT-PCR, and (ii) graphed densitometric data from n = 1, representing the clearest data from the indicated transfections T2 and T4. T2 and T4 refer to two different transfections, numbered 2 and 4, respectively. b RBM10 and SMN RNA expression in mock transfected (mock) and RBM10 siRNA transfected (10KD) fibroblasts, 48 h following transfection. c RBM10 and SMN RNA expression in mock transfected and RBM10 siRNA transfected fibroblasts, 72 h following transfection. T1–T5 designate data from five different transfections. The RBM10 and SMN results presented for any particular time point and fibroblast type were from the same transfection

Fig. 6

SMN protein expression in human fibroblasts following transient RBM10 KD. Expression was quantified using the Odyssey CLx Imaging System. a Three transfections into wt, carrier and SMA Type 1 fibroblasts, with analysis of RBM10 and SMN expression at 96 h post-transfection. The infrared picture for SMN expression in the SMA Type 1 fibroblasts was enhanced to enable visualization of the bands. b The ratio of SMN to GAPDH expression for four transfections (n = 4) was graphed. Results were expressed as means and standard errors. Significance was measured using an unpaired Student’s t test, where *p = 0.046 and **p = 0.01. NT non-transfected, mock mock-transfected, Tr transfected

Fig. 7

SMN RNA and protein expression in a stable RBM10 KD MCF-7 subline. M300.5 was a stable MCF-7 subline expressing a scrambled shRNA, and M29/30.2 was a stable MCF-7 subline expressing two RBM10 exon 6-specific shRNAs. a, b RNA expression. a Lane 1 ladder, lane 2 end-point PCR of M300.5 control using GAPDH primers, lanes 3 and 4 a duplicate load of an end-point PCR of M300.5 using the SMN primers; lane 5 blank, lane 6 end-point PCR of M29/30.2 RBM10 KD using GAPDH primers, lanes 7 and 8 a duplicate load of an end-point PCR of M29/30.2 using the SMN primers. b RNA expression following transient transfection, as previously described [26], of a pcDNA3-based plasmid containing RBM10v1. c Protein expression. (i) One representative Western blot of SMN protein expression. (ii) % SMN^FL, compared to SMN^Δ7, from two biological replicates. Results were expressed as means and standard errors