Introduction

Proximal spinal muscular atrophy (SMA) is a common neuromuscular disorder caused by loss of α-motor neurons in the spinal cord due to homozygous deletion or mutation of the survival motor neuron 1 gene (SMN1). SMA is the most frequent genetic cause of infantile death with an incidence of ~1 in 6000~10000 live births and a carrier frequency of 1 in 50.1, 2, 3 The SMN1 gene is located on chromosome 5q134 and produces full-length (FL) SMN mRNA. An almost identical gene, named SMN2, expresses abundant levels of transcript lacking exon 7 due to a C-to-T nucleotide substitution at position 6 within exon 7; the transcript encodes an SMNΔ7 protein, which disrupts SMN oligomerization, thus enhancing degradation of the SMN monomer.5 Childhood SMA is divided into types (I–III) based on the age of onset and achieved maximum motor abilities. SMA severity is inversely correlated with SMN2 gene copy number.6, 7, 8, 9

Alternative pre-mRNA splicing is an important mechanism of gene regulation. Accurate exon identification requires classical splicing signals and various cis-elements such as exonic splicing enhancers (ESEs) and silencers (ESSs) (reviewed in ref. 10, 11, 12). Multiple exonic cis-elements within the SMN gene and its regulators mediate splicing of SMN exon 7, which includes several negative and positive cis-elements13 such as Exinct, SF2/ASF-ESE, hnRNP A1-ESS, Tra2β1-ESE, and the 3'Cluster.

Variants that alter exonic splicing regulatory elements usually affect the functions in human genetic disease.14 In this paper, we identified a rare variant (c.863G>T, r.835_*3del, p.Gly279Glufs*5) in exon 7 of the SMN1 gene in three patients with different phenotypes. This variant lies adjacent to the Tra2β1 enhancer element. We demonstrated the c.863G>T (r.835_*3del, p.Gly279Glufs*5) variant interferes with the correct splicing of exon 7 due to disruption of Tra2β1. Our results suggest the variant abolishes SMN1 exon 7 inclusion in vivo and in an ex vivo minigene splicing assay, thus affecting the function of SMN.

Materials and methods

Patients and controls

Patient 1 (SMA I, ID: sm08120) and patient 2 (SMA II, ID:sm08117) were reported in a previous study.15 Patient 3 (ID:sm14002) was a male newborn infant diagnosed with SMA I. In the late stages of pregnancy, the fetal movement decreased significantly. When the little boy was born, he showed faint cry, profound hypotonia, symmetrical flaccid paralysis; and poor suckling and swallowing abilities after birth. The infant died at age 20 days due to respiratory failure caused by severe pneumonia. He was the second affected patient of this family, the first affected patient was his elder sister with similar severe phenotype. She was born normally, showed abnormal symptom at 2 months old with hypotonia and symmetrical flaccid paralysis, and died at 3 months old with servere pneumonia and respiratory failure. Informed consent was obtained from all three families to participate in this study but RNA would not be obtained from patient 3 because he died prior to genetic diagnosis. In addition, only patient 1 agreed to the establishment of an ex vivo skin fibroblast cell line. The skin fibroblast cell lines from two healthy children (N1 and N2) with two copies of SMN1 gene and two copies of SMN2 gene were also established as the controls. This study was approved by the Capital Institute of Pediatric Ethics Committee.

Sequencing and copy number determination

The SMN gene from exon 7 to exon 8 was PCR-amplified from genomic DNA with primers R111 and 541C1120.4 The PCR products were subcloned and 8–10 SMN1 clones were sequenced on an ABI 3730 automatic sequencer (Applied Biosystems, Foster, CA, USA) as described.16 SMN1, SMN2, NAIP, P44, and H4F5 gene copy number was determined by multiplex ligation-dependent probe amplification (MLPA) with a SALSA MLPA kit (P021-A1; MRC-Holland, Amsterdam, The Netherlands). The sequencing reads were aligned with the SMN1 genomic DNA reference sequence (NG_008691.1) and mRNA reference sequences (NM_000344.3). And the variant was submitted to the Leiden Muscular Dystrophy SMN1 Mutation Database (www.LOVD.nl/SMN1 (database ID SMN1_00057)).

Cell culture

Patient 1 fibroblast primary cell culture was established by the tissue explant method. Fibroblast and HEK293 cells were grown in Dulbecco’s modified Eagle’s medium (Gibco, New York, NY, USA) with 10~15% fetal bovine serum (Gibco, Newcastle, NSW, Australia), 100 U/ml penicillin and 100 μg/ml streptomycin at 37 °C in a humidified 5% CO2 atmosphere.

Transcription of SMN1

Total RNA was isolated from the peripheral blood and/or skin fibroblast cells of patients and controls. First-strand cDNA synthesis was performed with 0.5 μg total RNA, random primers, and M-MLV Reverse Transcriptase (Invitrogen, Carlsbad, CA, USA) according to manufacturer's protocols. Specific PCR primers (SMN57517 and 541C11204) were used to amplify SMN exons 1–8 (including SMN1 and SMN2) with LA Taq polymerase (TAKARA, Kyoto, Japan). The SMN transcript-products were subcloned into pGEM-T Easy (Promega, Madison, WI, USA); SMN1 subclones were screened by restriction digestion (DraI and DdeI) and sequenced.

Real-time PCR was used to quantify SMN transcripts as described by Tiziano et al18 Primers SMN_mgb-F and SMN_mgb-R were used to amplify the SMN1 and SMN2 genes; the specific probes for SMN1 and SMN2 are listed in Table 1. GAPDH served as the internal control (primers GAPDH_abs-F and GAPDH_abs-R; probe GAPDH-MGB).18 Each 20-μl reaction contained 2 × GoldStar TaqMan (KANGWEI, Pekin, China), 20 ng cDNA, 0.4 μl each primer (10 pmol/μl), and 4 pmol of the SMN1, SMN2, or GAPDH probe; reactions were performed on a 7500 Real-Time PCR System (Applied Biosystems). Samples were assayed in duplicate and repeated at least three times; data analysis was performed in SDS version 1.4 (Applied Biosystems).

Table 1 Oligonucleotide sequences
Full size table

Plasmid construction

SMN minigenes (pEasy-M2-SMN1, pEasy-M2-SMN2, and pEasy-M2-SMN1E7/SMN2E8) were constructed as follows. A 5.8-kb segment of the human SMN1, hybrid SMN1E7/SMN2E8, and SMN2 (exon 6, intron 6, 54 nt of exon 7, intron 7, and exon 8) were amplified with primers hSMNE6F and hSMNE8R (Table 1) from human genomic DNA template. The amplified products were cloned into expression vector pEasy-M2 (TRANS, Pekin, China) and sequenced to ensure that no base substitutions had been acquired during cloning. The resulting plasmids were pEasy-M2-SMN1, pEasy-M2-SMN1E7/SMN2E8, and pEasy-M2-SMN2). Short minigenes were constructed by amplifying the three wild-type plasmids with overlapping primers (SMNI6F and SMNI6R). Plasmids were constructed by PCR-based site-directed mutagenesis with fly-pfu DNA polymerase (TRANS) with the three short mini-plasmids as templates. We constructed one splicing-positive plasmid c.853-1G>A and three mutant plasmids (863A, 863C, 863T) with the primers described in Table 1. These short minigenes consisted of exon 6, part of intron 6, 54 nt of exon 7, intron 7, and exon 8.

Transfection and ex vivo splicing analysis

For transfection, 1 × 105 HEK293 cells were seeded in each well of a six-well plate in DMEM with 10% FBS. On the second day, 2 μg of each minigene plasmid was mixed with 10 μl TransLipid Transfection Reagent (TRANS) for 20 min and transfected in 1 ml serum-/antibiotic-free medium. After 4–6 h, the medium was replaced with 2 ml antibiotic-free DMEM with 10% FBS. Cells were harvested after 24 h and total RNA was extracted. Reverse transcription was performed in a total volume of 40 μl from 5 μg total RNA using random primers and M-MLV Reverse Transcriptase (Invitrogen). To ensure amplification of plasmid-derived transcripts, we amplified the cDNA from the transfected minigenes with plasmid-specific forward primer pEasy-M2F (Table 1) and the FAM-labeled SMN-specific reverse primer SMNE8R (Table 1) under the following cycle conditions: 94 °C for 5 min; 28 cycles of 90 °C for 1 min, 60 °C for 1 min, 72 °C for 1 min; 72 °C for 8 min. Amplified products were detected by 2% agarose gel electrophoresis and ethidium bromide staining. To calculate the splicing efficiency, a semiquantitative analysis was performed with a plasmid-specific primer and a FAM-labeled SMN-specific primer; the PCR products were run on the ABI 3730 automatic sequencing system (Applied Biosystems) and the raw data were analyzed with Gene marker version 1.75. Splicing efficiency was calculated by dividing the FL transcripts by the sum of the FL and truncated transcripts. These assays were repeated at least three times.

RNA pull-down

RNA-binding assays were performed with an RNA–protein binding assay kit (Thermo Scientific, Rockford, IL, USA). Biotin-labeled RNA SMNr.863 g (5′-biotin-AAAGAAGGAAGGUGCUCACAUUC) and SMN r.863 u (5′-biotin-AAAGAAGGAAUGUGCUCACAUUC) were synthesized by Invitrogen. The same biotin-labeled level of RNAs were added to an equivalent volume of beads and incubated for 60 min at 4 °C with rotation. After washing, HeLa nuclear extract was added to the RNA-bead mixture and incubated for 2 h at 4 °C with rotation. Bound complexes were washed three times with buffer and eluted in SDS-PAGE loading buffer. Bound proteins were analyzed by SDS-PAGE and immunoblotting with anti-Tra2β1, anti-hnRNP A1, anti-SF2/ASF antibodies (Abcam, Cambridge, UK). The experiments were repeated at least three times.

Results

The rare SMN1.863G>T (r.835_*3del, p.Gly279Glufs*5) variant was associated with lower levels of FL SMN1 transcript in the peripheral blood of two patients and skin fibroblasts of patient 1

Clinical characteristics are described in Table 2. The sequences of SMN1 exon 7-8 genomic DNA from all three patients revealed a base substitution of G to T at position c.863 (position +29 in SMN1 exon 7). All three patients carried only one SMN1 copy. Patients 1 and 2 carried three copies of SMN2, indicating that one SMN1 gene was converted to SMN2. The remaining modifier genes (NAIP, P44, and H4F5) were present in the normal two copies. Patient 3 carried two SMN2 genes, zero copies of NAIP, and one copies of P44 and H4F5 (Table 2).

Table 2 SMA patient characteristics associated with c.863G>T(r.835_*3del, p.Gly279Glufs*5)
Full size table

We used real-time PCR to measure the effect of the SMN1c.863G>T (r.835_*3del, p.Gly279Glufs*5) variant on levels of FL SMN1 (FL-SMN1) transcript in the blood of patients 1 and 2. Our results showed that the amount of FL-SMN1 was significantly reduced, consistent with prior report.16 Sequencing of the SMN1 clones also showed that the SMN1 transcript was missing exon 7. In comparison with normal controls with the same SMN2 copy numbers, the amount of FL-SMN1 transcript was very low and SMN protein levels were significantly reduced (only 22% of control levels) in the skin fibroblasts of patient 1 (Figure 1). Our results also showed no significant difference in the levels of SMN transcripts (FL-SMN1, FL-SMN2, and SMN▵7) between the peripheral blood and skin fibroblasts of patient 1 (Figure 1). Thus, the c.863G>T (r.835_*3del, p.Gly279Glufs*5) variant was associated with reduced inclusion of exon 7 and lower expression of FL-SMN1 mRNA and SMN protein.

Figure 1
figure 1

SMN transcript and SMN protein levels in patient 1. (a) The levels of FL-SMN1, FL-SMN2, and SMN▵7 transcript did not differ between blood cells and skin fibroblasts; FL-SMN1 transcript levels were near the limit of detection. (b) Semiquantitative detection of SMN protein in skin fibroblasts of patient 1. SMN protein expression was very low, only 22% of control levels (two normal individuals carrying two copies of SMN1 gene and two SMN2).

Full size image

Minigene construction and analysis

Two types of minigenes were constructed (Figure 2a). The SMN-Long construct contains exon 6, intron 6, exon 7, intron 7, and exon 8; SMN-Short was missing ~4.2 kb of intron 6. For each plasmid, we constructed three minigenes: SMN1, SMN2, and a hybrid SMN1E7/SMN2E8 gene. To compare the splicing effects of these minigenes, we measured the percentage of FL SMN transcript by RT-PCR after transfection of HEK293 cells. Over 90% of the long and short SMN1 and SMN1E7/SMN2E8 transcripts retained exon 7; in contrast, ~70% of the long and short SMN2 transcripts excluded exon 7 (Figure 2b), although there was no significant splicing difference between long and short plasmid types (P>0.05). Thus, shortening the otherwise long intron 6 had no effect on splicing and most of the intronic 6 sequence is not essential for exon 7 inclusion. In follow-up studies, mutant minigenes were therefore constructed from the short plasmids.

Figure 2
figure 2

Deletion of 4.2 kb of intron 6 has no significant effect on SMN exon 7 inclusion/exclusion. (a) Schematic diagram of the SMN (SMN1, SMN2, and SMN1E7/SMN2E8) minigenes. Interconversion of SMN1 and SMN2 often occurs to form two hybrid genes (SMN2E7/SMN1E8 and SMN1E7/SMN2E8) in normal individuals. The SMN2E7/SMN1E8 hybrid gene leads to exon 7 skipping as in the SMN2 gene, where the exon 7 +6 site is a T. The SMN1E7/SMN2E8 hybrid gene retains exon 7 and does not produce the SMA phenotype. We constructed c.863 variants in the SMN1E7/SMN2E8 hybrid minigene to determine whether its splicing was consistent with that of variation in SMN1. The 4.2 kb central sequence of intron 6 was deleted from the SMN-Short minigene and the five variant nucleotides in SMN1 and SMN2 are showed in the SMN-Short minigenes. (b) Splicing effects were analyzed by qualitative and semi-quantitative RT-PCR; no significant difference was found between the long- and short-type plasmids (P>0.05).

Full size image

The c.863G>T (r.835_*3del, p.Gly279Glufs*5) variant inhibits inclusion of exon 7 in SMN1, SMN2, and hybrid SMN1E7/SMN2E8 genes

To determine whether the c.863G>T (r.835_*3del, p.Gly279Glufs*5) variant is the sole cause of exon 7 exclusion, we performed ex vivo splicing assays using minigenes that mimic the natural context of the SMN1, SMN2, and hybrid SMN1E7/SMN2E8 genes (Figure 3). To evaluate the splicing effect of the mutant minigene, we constructed a positive control splicing minigene that includes the classical splicing site c.835-1G>A in the SMN1-Short plasmid; this construct yielded 100% exclusion of exon 7. The c.863G>T (r.835_*3del, p.Gly279Glufs*5) SMN1 minigene variant reduced SMN1 exon 7 inclusion from 91.7% to 14.8% (t=23.45, P=0.000, Figure 3a). These data are consistent with the results observed in patient blood and skin fibroblasts. Similar results were observed for the SMN2 and hybrid SMN1E7/SMN2E8 minigenes. In the SMN2 gene, the c.863G>T (r.835_*3del, p.Gly279Glufs*5) variant disrupted SMN2 exon 7 inclusion from 33.3% to 0.0% (t=7.56, P=0.002, Figure 3b). In the hybrid SMN1E7/SMN2E8 gene, this variation caused a significant reduction of exon 7 inclusion from 91.1% to 15.5% (t=23.45, P=0.000, Figure 3c).

Figure 3
figure 3

Ex vivo splicing assay of SMN1, SMN2, and hybrid SMN1E7/SMN2E8 minigenes. The mutant minigenes were constructed by PCR-based site-directed mutagenesis and transfected into the HEK293 cells. The picture on the left shows the products of RT-PCR on ethidium bromide-stained agarose gel. The quantitative results are shown on the right. Error bars indicate the SD; P-values were calculated by Student’s t-test (**P<0.01, *P<0.05). The c.863T minigenes showed increased exon 7 skipping; the c.863A minigenes showed a slight increase of exon 7 exclusion; the c.863C minigenes showed increased inclusion of exon 7. (a) SMN1 minigenes; (b) SMN2 minigenes; (c) hybrid SMN1E7/SMN2E8 minigenes.

Full size image

The splicing effect of various c.863 substitutions

We constructed two additional substitutions at c.863 in SMN1, SMN2, and the hybrid SMN1E7/SMN2E8 genes and observed their varying influence on exon 7 splicing. In the SMN1 and SMN1E7/SMN2E8 genes (Figure 3), the c.863G>A variant reduced exon 7 inclusion from 91.7% to 81.3% (t=3.3, P=0.016) and from 91.1% to 80.8% (t=3.02, P=0.057), respectively. The results for the SMN2 gene were similar, as the variant decreased exon 7 inclusion from 33.3% to 19.3% (t=3.06, P=0.038). The c.863G>C variant promoted exon 7 inclusion in all three genes: in SMN1 and SMN1E7/SMN2E8, exon 7 inclusion increased from 91% to 95% (P>0.05); in SMN2, the increase was from 33.3% to 54.9% (t=−4.48, P=0.011).

The c.863G>T (r.835_*3del, p.Gly279Glufs*5) variation disrupts Tra2β1 binding

Variants at c.863, particularly c.863G>T (r.835_*3del, p.Gly279Glufs*5), may influence splicing of SMN exon 7. We then identified the splicing factors that bind the central region of exon 7, which includes the c.863G>T (r.835_*3del, p.Gly279Glufs*5) site (position +29). Pull-down assays were performed with synthetic RNAs corresponding to positions +19 to 41 of exon 7 (Figure 4a). Biotin-labeled RNAs (including wild-type r.863 g and mutant-type r.863 u) were incubated with HeLa nuclear extract and analyzed by western blotting (Figure 4b) for the key regulators (hnRNP A1, Tra2β1, and SF2/ASF19, 20, 21) of SMN exon 7 splicing. The c.863G>T (r.835_*3del, p.Gly279Glufs*5) variant disrupted binding of Tra2β1, the ESE-binding factor that recognizes the center of exon 7 (+19–+27). The splicing enhancer protein SF2/ASF did not bind to wild-type or mutant RNAs, whereas hnRNP A1 interacted with both, suggesting the presence of a binding element for this regulator within exon 7. The c.863G>T (r.835_*3del, p.Gly279Glufs*5) variant did not disrupt hnRNP A1-binding, however, indicating that +29 variant site is not the key nucleotide in the hnRNP A1-binding element. Our results suggest the c.863G>T (r.835_*3del, p.Gly279Glufs*5) variant promotes exon 7 skipping by disrupting the interaction of splicing enhancer Tra2β1.

Figure 4
figure 4

RNA pull-down assay for splicing factors bound to the central region of SMN exon 7. (a) Schematic of the synthetic RNA sequences and the pertinent positions in exon 7 of SMN; the underlined base indicates the r.863 g>u(c.863G>T) variant. (b) Western blot analysis. Synthetic RNAs were labeled with biotin and immobilized on agarose beads; proteins were detected with antibodies specific for hnRNP A1, Tra2β1, and SF2/ASF.

Full size image

Discussion

Here, we identified a rare variant (c.863G>T, r.835_*3del, p.Gly279Glufs*5) in center exon 7 of the SMN1 gene in three patients and this variant abolished SMN1 exon 7 inclusion both in vivo and ex vivo assays. Our results demonstrated that this c.863G>T (r.835_*3del, p.Gly279Glufs*5) variant does not affect SMN protein function via amino-acid substitution; instead, it disrupts the pre-mRNA splicing of SMN1 by interfering with Tra2β1 binding. Sterne et al22 used bioinformatics and biochemical methods to show that ~25% of missense, nonsense, and silent variants result in aberrant splicing of pre-mRNA. These changes were traditional considered in the context of amino-acid substitutions rather than pre-mRNA splicing, although we now know such variants are common causes of human genetic disease. Studies of an SMA phenotype modifier, the SMN2 gene, demonstrated that a single base substitution in exon 7+6 causes abnormal splicing of SMN2 pre-mRNA, resulting in skipping of this exon and reduced expression of the functional SMN protein.5 Our results present clear evidence that the c.863G>T (r.835_*3del, p.Gly279Glufs*5) variant in the central portion of SMN1 exon 7 interferes with correct splicing of this exon, thereby causing the absence of SMN protein and resulting in a severe SMA phenotype. Low-level FL-SMN1 transcript inclusion of exon 7 was confirmed by in vivo studies and sequencing in peripheral blood cells and skin fibroblasts from affected patients. SMN protein levels were significantly reduced in skin fibroblasts from patient 1. Our ex vivo splicing assay confirmed that the c.863G>T (r.835_*3del, p.Gly279Glufs*5) variant promoted exon 7 skipping in SMN pre-mRNA (SMN1, SMN2, and hybrid SMN1E7/SMN2E8), retaining only a small amount of FL transcript. We thus confirmed that the c.863G>T (r.835_*3del, p.Gly279Glufs*5) variant, which increases exon 7 skipping, causes incorrect splicing of SMN1 and is not a missense change in the traditional sense. The change effect of c.863G>T (r.835_*3del, p.Gly279Glufs*5) variant in SMN1 is similar to that in SMN2(c.840C>T). Thus, the amount of functional SMN protein is sharply reduced in vivo owing to rapid degradation of the truncated SMN protein.23 Consequently, motor neurons in the spinal cord anterior horn undergo degeneration and necrosis, resulting in a severe SMA phenotype. This has enhanced our understanding of coding region variants that lead to changes in pre-mRNA splicing.

Over the past decade, extensive studies have been conducted on exon 7 splicing of SMN1 and SMN2 and several splicing-related cis-elements and their corresponding splicing factors have been identified.10, 13, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33 There are overlapping enhancer and silencer sequences within exon 7. For example, Cartegni and Krainer20 identified an SF2/ASF-binding ESE at position +6 of SMN1 exon 7. Kashima et al27 found that a C>T substitution at position +6 of SMN2 exon 7 destroys the SF2/ASF-binding ESE and produces an hnRNP A1-binding ESS. Several studies have shown that a Tra2β1-binding ESE (AAAGAAGGA) is present in the central portion of exon 7 at positions +19–+27.19, 25, 34 A study of the c.859G>C variant (at position +25 of exon 7) showed that the central portion of exon 7 (positions +16–+39) also bind the splicing-inhibitor hnRNP A1 and the splicing activator SF2/ASF.35 Our RNA pull-down assays showed that SF2/ASF does not bind exon 7 at positions +19–+41, consistent with predictions generated in ESEfinder 2.0. The findings of Vezain et al 35 led us to suggest that the important nucleotides for SF2/ASF binding may lie within positions +16–+19 of exon 7, not +19–+41. In addition, the central portion of exon 7 is bound by the hnRNP A1 inhibitor but does not require c.863G. The c.863G>T (r.835_*3del, p.Gly279Glufs*5) variant interferes with binding of the splicing activator Tra2β1. A sharp decline in binding of Tra2β1 to the c.863G>T (r.835_*3del, p.Gly279Glufs*5) site suggests c.863G might be involved in the Tra2β1-ESE sequence, although this finding remains to be verified. Our findings and those of Vezain et al35 suggest the central region of exon 7 binds the Tra2β1 activator and the hnRNP A1 inhibitor. Two recent studies have also shown binding of two splicing factors (hnRNP M and PSF) at sites that partially overlap with the Tra2β1-ESE sequence.32, 33 The evidence suggests sequence overlap between ESE and ESS elements, and between binding sequences for various splicing activators. These components mutually constrain each other to maintain homeostasis, loss of which leads to aberrant splicing. Variants located in or near regulated regions must be considered in the context of splicing effects.

We also showed that exon 7 inclusion in ex vivo splicing assays can be improved or disrupted by variants at c.863 site: the highest level of FL transcript was produced by a G to C substitution, followed by the wild-type base G and A. The base T at this position led to a significant reduction in the level of FL transcript. The splicing effects of these four different bases at c.863 in three minigene constructs (SMN1, SMN1E7/SMN2E8, and SMN2) were similar. Multi-species sequence comparisons suggest c.863G is highly conserved; only an A occurs at this site in mice. In vitro splicing data showed that the c.863G>T variant led to exon 7 skipping in most SMN transcripts. Our patients exhibited severe phenotypes of SMA type I or II and significantly reduced levels of FL-SMN1 transcript. We conclude that the c.863G>T (r.835_*3del, p.Gly279Glufs*5) variant affects function. It is interesting to note that a G to C substitution at c.863 resulted in an increase of FL-SMN transcripts. Our findings and those of Vezain et al34 indicated that the splicing-inhibitor hnRNP A1 can bind the central region of exon 7. The optimal binding sequence for hnRNP A1 is TAGGGA/T. There are two similar sequences (four bases matched the optimal hnRNP A1-binding sequence) in the central region of SMN exon 7: GAAGGA (at positions +22–+27) and GAAGGT (at positions +26–+31). The c.859G>C variant (at position +25 of exon 7) in SMN2, which was first reported in SMA by Feldkötter et al,6 leads to increased inclusion of SMN exon 7 by creating a new binding site for SF2/ASF36 or interfering with hnRNP A1 bindings35 However, increased expression of FL SMN2 rescues SMA phenotypic severity. We found that c.859G>C variant(+25G to C) reduced the hnRNP A1 sequence match to three bases (GAACGA). Variant of +29 G to C alters the sequence to GAACGT, similar to the G>C substitution at position +25. This also led to increased exon 7 inclusion, so we suggest a substitution of C at +29 might disrupt binding of splicing-inhibitor hnRNP A1.

Here, we report three patient carrying the c.863G>T (r.835_*3del, p.Gly279Glufs*5) variant, two with type I and one with type II SMA. Patients 1 and 3 were type I, but Patient 3 had a more severe phenotype. Patient 3 suffered disease onset at late-stage of pregnancy and had a flaccid body at birth, with death occurring at age 20 days due to lung infection and respiratory failure. MLPA results suggested that different modifier gene backgrounds likely cause phenotypic differences between patients. Patient 3 had only two copies of SMN2, zero copies of NAIP, and one copy each of P44 and H4F5. In Patient 3, therefore, full deletion of SMN1 occurred on one chromosome in the presence of a larger deletion that included the NAIP, P44, and H4F5 genes; the other chromosome carried the c.863G>T (r.835_*3del, p.Gly279Glufs*5) variant in SMN1 and a deletion of NAIP. Patient 1 carried three copies of SMN2 and normal complements of NAIP P44, and H4F, suggesting conversion of one copy of SMN1 to SMN2. Previous studies showed that expansion of the SMN1 deletion to NAIP, P44, and H4F5 is always associated with a severe phenotype, including early age of onset and early death.37, 38, 39 Conversion of SMN1 to SMN240, 41 results in increased SMN2 copy number and is inversely correlated with phenotype severity. Our recent report,9 revealed that the median age onset is 1 month in SMA patients with SMN1-SMN2-NAIP genotype of 0-2-0, whereas in patients with 0-3-2 genotype were 8 months. The median survival time of 0-2-0 patients was only 6 months, whereas the lifespan of patients with 0-3-2 is almost not affected. Based on the c.863G>T functional effect, the genotype of patient 3 is similar to 0-2-0, whereas other two patients are similar to 0-3-2. In addition, the plastin 3 (PLS3) gene at position Xq23 is protective in female SMA patients.42,43,44,45 Our recent research45 revealed that the PLS3 gene may have age- and gender-specific role (female protective) in the clinical severity in our SMA patients. So the male patient 3 exhibited a more serious phenotype than the female patients, a phenotypic difference that may partly be attributable to PLS3. Patient 2 had a moderate phenotype with a later age of onset and better motor function than patient 1, but both patients had same SMN1 genotype and SMN2-NAIP-P44-H4F5 copy number, and no significant difference in FL-SMN transcript levels. We suggest other factors such as profilin IIa46, 47 and zinc finger protein48 may also influence SMA phenotype.

In summary, our study clearly demonstrates that the c.863G>T (r.835_*3del, p.Gly279Glufs*5) variant can lead to exon 7 skipping in a majority of SMN1 transcripts. Patients with this variant showed more severe SMA phenotype. Our results also indicate that the c.863G>T (r.835_*3del, p.Gly279Glufs*5) variant results in incorrect splicing of exon 7 by interfering with the binding of splicing activator Tra2β1, but does not promote binding of the splicing-inhibitor hnRNP A1. Our results suggest overlap of the ESE and ESS signals in the central portion of exon 7. Therefore, variants in this region require comprehensive consideration and integrated analysis.