A-44G transition in SMN2 intron 6 protects patients with spinal muscular atrophy

Abstract

Spinal muscular atrophy (SMA) is a neuromuscular disease caused by reduced expression of survival of motor neuron (SMN), a protein expressed in humans by two paralogous genes, SMN1 and SMN2. These genes are nearly identical, except for 10 single-nucleotide differences and a 5-nucleotide insertion in SMN2. SMA is subdivided into four main types, with type I being the most severe. SMN2 copy number is a key positive modifier of the disease, but it is not always inversely correlated with clinical severity. We previously reported the c.859G > C variant in SMN2 exon 7 as a positive modifier in several patients. We have now identified A-44G as an additional positive disease modifier, present in a group of patients carrying 3 SMN2 copies but displaying milder clinical phenotypes than other patients with the same SMN2 copy number. One of the three SMN2 copies appears to have been converted from SMN1, but except for the C6T transition, no other changes were detected. Analyzed with minigenes, SMN1C6T displayed a ∼20% increase in exon 7 inclusion, compared to SMN2. Through systematic mutagenesis, we found that the improvement in exon 7 splicing is mainly attributable to the A-44G transition in intron 6. Using RNA-affinity chromatography and mass spectrometry, we further uncovered binding of the RNA-binding protein HuR to the -44 region, where it acts as a splicing repressor. The A-44G change markedly decreases the binding affinity of HuR, resulting in a moderate increase in exon 7 inclusion.

Figure 1.

Diagrams of a partially converted SMN2 allele and the new SMN1/2 minigenes, and splicing analysis of the minigenes and SMN1 C6T mutant. (A) Diagram of part (from exon 6 to exon 8) of an allele of SMN2, showing incomplete conversion from SMN1. The natural sequence differences between SMN1 and SMN2 are listed below the diagram. This allele was identified in 5 out of 53 three-SMN2-copy patients with mild clinical phenotypes. (B) Diagram of the new SMN1/2 minigenes cloned in pCI-neo. The new minigenes comprise all the sequences that naturally differ between the two genes except the G236A change in exon 8. The genomic DNA is shown for comparison, and the intron and exon sizes are indicated. (C) Exon 7 splicing analysis of the SMN1/SMN2 minigenes, and their mutants SMN1 C6T and SMN2 c.859G>C. Each plasmid was transfected into HEK293 cells, and splicing was analyzed as described in Materials and Methods. As shown in the gel and histogram, the C6T mutation in exon 7 in the SMN1 minigene setting decreased exon 7 inclusion by 47% (from 97 to 50%), but this is still better than SMN2 (50% versus 29%) and similar to the c.859G>C mutation in the SMN2 minigene setting (50% versus 53%). Inc%: percentage of exon 7 inclusion. **P<0.001 (n = 4).

Figure 2.

The G-44A transition accounts for the splicing deterioration in SMN2 compared to SMN1 C6T. (A) Screening of double mutations in the SMN1 C6T setting. The second mutation was either G-980A, Ins5nt (5-nt AGGCA insertion at -908), A-849C, G-549A, T-478C, T-255C, G-44A, A100G, or A215G, and was introduced in the parental plasmid, pCI-SMN1 C6T. SMN1, SMN2, SMN1 C6T, and SMN2 T6C were used as controls. Each plasmid was transfected into HEK293 cells, and the effect of each mutation was analyzed. Two mutations resulted in significant splicing suppression: G-44A in intron 6 reduced exon 7 inclusion by ∼20%, and A100G in intron 7 reduced it by 9%. **P < 0.001 versus SMN1 C6T (n = 4). (B) Mutations analyzed in the SMN1 C6T/G-44A setting. Each mutation, including G-980A, Ins5nt, A-849C, G-549A, T-478C, T-255C, A100G, and A215G was introduced in the parental double-mutant plasmid pCI-SMN1 C6T/G-44A, and analyzed in HEK293 cells. The effect of A100G was weaker than in the double-mutant context (Panel A), but there was a significant reduction in exon 7 inclusion (∼3%). On the other hand, some mutations, such as G-980A and G-549A, which had no effect in the double-mutant context, significantly improved splicing in the triple-mutant context (4–5% increase in exon 7 inclusion). *P < 0.01 versus parental SMN1 C6T/G-44A (n = 5). (C) SMN1 C6T and SMN1 C6T/G-44A (+G-44A) mutants were analyzed in five different cell types, including neural cells: HEK293, HeLa, COS1, NSC34, and SH-SY5Y. The splicing patterns in all cell types are consistent with a 15–20% decrease in exon 7 inclusion caused by the G-44A substitution. **P < 0.001 versus SMN1 C6T (n = 4).

Figure 3.

RNA secondary structure is not involved in splicing repression caused by the G-44A substitution. (A) An RNA secondary structure was predicted by UNAFold at a region from position -41 to -60 in SMN1 intron 6. The arrow indicates position -44. (B) List of mutations in the SMN1 C6T setting for structure analysis, corresponding sequences, predicted minimum free energy (ΔG, Kcal/mol) of the secondary structure, and exon 7 inclusion of each mutant (Inc%). The underlined sequence is the loop, and mutations are shaded. Minimum free energy values were calculated by UNAFold. Some mutations strengthen the putative stem-loop structure, whereas others destabilize it. Exon 7 inclusion was analyzed as in Panel C; mean values are shown (n = 4). (C) All mutant plasmids in the SMN1 C6T setting were analyzed in HEK293 cells. SMN1 and SMN2 wild types and SMN1 C6T were used as controls. % exon inclusion is shown below this representative gel. (D) The percentage of exon 7 inclusion of each tested mutation, as well as SMN1 C6T, is presented as a scatter plot against the calculated minimum free energy (listed in Panel B). No correlation was observed by SPSS Pearson analysis (R²=0.11).(E) Deletion mutants Del1, Del2, Del3 and Del4 were created in the SMN1 C6T setting, and their splicing was analyzed in HEK293 cells. SMN1 WT, SMN1 C6T and SMN1 C6T/G-44A were used as controls. The histogram shows the quantitation of splicing from four experiments similar to the one shown on the left. Four mutants displayed exon 7 inclusion levels of 40, 73, 83 and 60%, respectively, all of which are significantly better than SMN1 C6T/G-44A. **P < 0.001 versus SMN1 C6T (n = 4).

Figure 4.

Analysis of proteins bound to the intronic splicing silencer around position -44 by RNA-affinity chromatography. (A) Two 16-nt RNA oligonucleotides with WT (corresponding to SMN1) and G-44A mutant (corresponding to SMN2) sequences were used for RNA-affinity chromatography. (B) Agarose beads covalently linked to the RNAs shown in (A) were incubated with HeLa cell nuclear extract (NE) under splicing conditions, and the beads were washed four times with Buffer D containing 100 mM KCl. Captured proteins were eluted with SDS sample buffer, separated by SDS-PAGE and stained with Coomassie Blue. A reaction without RNA (No RNA) was used as a control. Protein bands around 36-37 kDa were prominent on the gel in both WT and G-44A samples, with the bands from the G-44A sample appearing stronger. Bands were excised from the gel and subjected to mass spectrometry. (C) Mass spectrometry analysis revealed multiple proteins bound to the two RNA species. Four proteins with sizes between 35 kDa and 38 kDa had PSMs over 4. Among them, HuR had the largest increase in PSMs from the WT to G-44A samples (2.5 fold). The mass spectrum of a peptide derived from HuR, identified in the pulldown, is shown below. (D) Western blot analysis of the eluted proteins with anti-hnRNP A1, anti-hnRNP A2 and anti-HuR antibodies. 7.5% input of NE was loaded on the right lane. Band intensities were measured using Image J software. Normalized band intensity was displayed in the histogram. HnRNP A1, hnRNP A2 and HuR had a fold increase of 1.04, 1.15, and 1.37, respectively, in binding to the G-44A mutant RNA, compared to the WT RNA. **P < 0.001 (G-44A versus WT RNA, n = 3)

Figure 5.

MS2 tethering assay revealed HuR as a splicing repressor when bound to the -44 site. (A) Diagram of the SMN1 minigene with an MS2 hairpin (MS2 SMN1). The sequence from −47 to − 36 in intron 6 (TATGTCTATATA) was deleted and replaced with an MS2-binding sequence (5’ -ACATGAGGATCACCCATGT -3’). The pre-mRNA transcribed from the MS2 SMN1 minigene forms a stem-loop structure that binds to MS2 CP. (B) The effects of various expressed proteins on exon 7 splicing were analyzed in HEK293 cells. The MS2 SMN1 minigene (1 μg) and various expression plasmids (100 ng each for transfection with one expression plasmid or 50 ng each for co-transfection with 2 expression plasmids) were co-transfected into cells. T7-Empty: pCGT7 empty vector was transfected as a blank control. T7-CP: T7-tagged coat protein; T7-CP-A1: hnRNP A1 fused to T7-CP; T7-CP-A2: hnRNP A2 fused to T7-CP; T7-CP-HuR: HuR fused to T7-CP. T7-A1: T7-tagged hnRNP A1; T7-A2: T7-tagged hnRNP A2. Western blot (WB) analysis with an anti-T7 antibody showed proper expression of all proteins. Cy5-labeled RT-PCR analysis of exon 7 splicing was quantitated and shown in the histogram. **P<0.001 versus T7-CP (n = 3). *Nonspecific band.

Figure 6.

RRM1 and RRM2 of HuR are sufficient for splicing repression. (A) Diagram of the primary structure of HuR and its deletion mutants. HuR has three RNA recognition motifs (RRM1, RRM2 and RRM3) and a hinge region between RRMs 2 and 3. The residue numbers at domain boundaries are indicated. The T7 tag is denoted by a hexagon. (B) The effects of the indicated HuR mutants on exon 7 splicing were analyzed in HEK293 cells. The MS2 SMN1 minigene (1 μg) and each protein expression plasmid (100 ng) were co-transfected into cells. Western blot (WB) analysis using anti-T7 antibody was performed to examine protein expression. Splicing analysis with Cy5-labeled RT-PCR was performed, and quantitation of exon 7 inclusion is shown in the histogram. **P < 0.001, ^#P>0.05 versus T7-CP-HuR (n = 3). *Nonspecific band.

Figure 7.

Effects of HuR overexpression and knockdown. (A) A negative correlation between the A/T content and exon 7 splicing was observed in SMN1 C6T, SMN1 C6T/G-44A and SMN1 C6T/G-44A/C-42A (see Fig. 3). The number of A or T residues within a 15-nt segment is indicated for each construct. (B) Effect of HuR overexpression on exon 7 splicing, analyzed with SMN1 WT, SMN1 G-44A, SMN1 G-44A/C-42C and SMN1 Del4 (see Fig. 3). In the SMN1 setting, the -44 specific and A/T content-dependent effects of HuR could be distinguished with these WT and mutant minigenes. Each minigene plasmid (1 μg) and pCGT7-HuR (1 μg) were co-transfected into HEK293 cells, and exon 7 splicing was analyzed with Cy5-labeled RT PCR; the histogram on the right shows the quantitation of exon 7 inclusion. Western blot (WB) analysis using an anti-T7 antibody confirmed expression of T7-HuR. **P<0.001, T7-HuR versus T7-Empty (n = 3). (C) Effect of HuR knockdown on exon 7 splicing, analyzed with SMN1 C6T and SMN1 C6T/G-44A. Each siRNA (100 nM) and one minigene plasmid (1 μg) were co-transfected into HEK293 cells. HuR was detected with a monoclonal anti-HuR antibody, with β-tubulin as a loading control. Two siRNAs (siRNA-1 and siRNA-2) robustly reduced HuR expression (to 52% and 48%, respectively) compared to a control siRNA (siNC). After HuR knockdown, exon 7 splicing in SMN1 C6T/G-44A showed a greater increase than in SMN1 C6T (18–20% versus 7–10%). **P < 0.001 compared to siNC (n = 3).