Identification of a Cis-acting Element for the Regulation ofSMN Exon 7 Splicing*

Spinal muscular atrophy results from the loss of functional survival motor neuron (SMN1) alleles. Two nearly identical copies of SMN exist and differ only by a single non-polymorphic C to T transition in exon 7. This transition leads to alteration of exon 7 splicing; that is, SMN1produces a full-length transcript, whereas SMN2 expresses a low level of full-length transcript and predominantly an isoform lacking exon 7. The truncated transcript of SMN encodes a less stable protein with reduced self-oligomerization activity that fails to compensate for the loss of SMN1. In this paper, we identified a cis-acting element (element 1), which is composed of 45 bp in intron 6 responsible for the regulation of SMN exon 7 splicing. Mutations in element 1 or treatment with antisense oligonucleotides directed toward element 1 caused an increase in exon 7 inclusion. An ∼33-kDa protein was demonstrated to associate with a pre-mRNA sequence containing both element 1 and the C to T transition in SMN exon 7 but not with the sequence containing mutated element 1, suggesting that the binding of the ∼33-kDa protein plays crucial roles in the skipping ofSMN exon 7 containing the C to T transition.

Spinal muscular atrophy results from the loss of functional survival motor neuron (SMN1) alleles. Two nearly identical copies of SMN exist and differ only by a single non-polymorphic C to T transition in exon 7. This transition leads to alteration of exon 7 splicing; that is, SMN1 produces a full-length transcript, whereas SMN2 expresses a low level of full-length transcript and predominantly an isoform lacking exon 7. The truncated transcript of SMN encodes a less stable protein with reduced self-oligomerization activity that fails to compensate for the loss of SMN1. In this paper, we identified a cis-acting element (element 1), which is composed of 45 bp in intron 6 responsible for the regulation of SMN exon 7 splicing. Mutations in element 1 or treatment with antisense oligonucleotides directed toward element 1 caused an increase in exon 7 inclusion. An ϳ33-kDa protein was demonstrated to associate with a pre-mRNA sequence containing both element 1 and the C to T transition in SMN exon 7 but not with the sequence containing mutated element 1, suggesting that the binding of the ϳ33-kDa protein plays crucial roles in the skipping of SMN exon 7 containing the C to T transition.
Spinal muscular atrophy (SMA) 1 is a common autosomal recessive disorder with progressive paralysis caused by the degeneration of motor neurons in the spinal cord (1). The survival of the motor neurons (SMN) gene has been identified as the disease gene of SMA and is present on chromosome 5 at 5q13 (2,3). Humans contain two nearly identical copies of the SMN gene, SMN1 and SMN2. These genes encode an identical protein, a 294-amino acid RNA-binding protein. Only homozygous deletions or mutations of SMN1 result in the SMA phenotype, and the levels of SMN expression driven by SMN2 in motor neurons inversely correlate with the severity of the disease (4 -15).
SMN1 mRNA expresses a full-length transcript, whereas SMN2 produces a low level of full-length transcript and predominantly an isoform lacking exon 7 (SMN⌬7) (2,16,17). The SMN⌬7 is less stable (18), and it was reported that SMN⌬7 cannot oligomerize or self-associate as efficiently as the protein produced from the full-length SMN transcript (2,19,20). Therefore, a deficiency in the full-length SMN protein correlates with the disease. The critical difference between SMN1 and SMN2 is a silent nucleotide transition in SMN exon 7. SMN1 contains a C located six nucleotides inside exon 7, whereas SMN2 contains a T at this position. This transition is considered to inhibit one of the splicing regulatory elements within exon 7, which are called exonic splicing enhancers (ESE) (21). A recent report demonstrated the presence of an ESE within exon 7 and that human Tra2-␤1, a member of the serine-arginine-related proteins of splicing factors, binds to the elements and stimulates an ESE (22). However, the critical C to T transition is not contained within the element. Furthermore, the transition does not change the binding activity of Tra2-␤1 to the ESE. Thus, it is still unclear why the C to T transition leads to a lack of exon 7 in SMN2.
In this paper, we tried to elucidate the molecular mechanisms of SMN2 exon 7 skipping by determining the critical cis-acting elements on the SMN pre-mRNA responsible for the aberrant splicing of SMN exon 7, which contains the C to T transition. Moreover, we tried to develop strategies to enhance the inclusion of exon 7 by using antisense oligonucleotides that can inhibit the function of the cis-acting element, which we identified in this study.

EXPERIMENTAL PROCEDURES
Cell Cultures-COS-7, HEK293T, and SK-N-SH cells were used for in vivo splicing assays. COS-7 and HEK293T cells were grown in 10% fetal bovine serum/Dulbecco's modified Eagle's medium, and SK-N-SH cells were cultured in ␣-minimum essential medium with 10% fetal bovine serum. Prior to transfection, cells were plated at a density of 60 -80% confluency on 3.5-cm dishes.
In Vivo Splicing-Constructs of SMN1 and SMN2 mini-genes containing exon 6 -exon 8 in a pCI mammalian expression vector were gifts from Drs. Elliot Androphy (Tufts University) and Christian Lorson (Arizona State University) (21). Mini-genes containing SMN1 exon 6 -exon 8 and the C to T transition in exon 7 cloned into the pCI vector were mutated in element 1 by site-directed mutagenesis. The constructs (1.0 g) were transfected into cells using LipofectAMINE reagent or LipofectAMINE ACE Reagent (Invitrogen) according to the manufacturer's protocol. Transfected cells were lysed in buffer RLT (Qiagen), and total cellular RNA was purified using the RNeasy Mini Kit (Qiagen). First-strand cDNA was synthesized in a 20-l reaction volume using a random primer (TaKaRa) and Moloney murine leukemia virus reverse transcriptase (Invitrogen). PCR amplification analysis of the plasmid-derived cDNAs was performed using the primer set, pCI forward (5Ј-GCT AAC GCA GTC AGT GCT TC-3Ј) and pCI reverse (5Ј-GTA TCT TAT CAT GTC TGC TCG-3Ј). PCR was performed in a total volume of 50 l that contained 1.2 g of first-strand cDNA, 0.4 M each primer, 0.2 M dNTPs supplemented with trace amounts of [␣-32 P]dCTP, 5 units of rTaq DNA polymerase, and 10ϫ PCR buffer (TaKaRa). Amplification conditions were as follows: an initial denaturation step (94°C/2 min), 30 cycles (94°C/30 s; 56°C/1.5 min; 72°C/1 min), and a final extension step (72°C/10 min). Reaction products were resolved by electrophoresis through a 5% acrylamide gel. PCR products were cloned into the pGEM-T vector (Promega) and sequenced. Quan-tification of the density of each band was carried out using the Densitography Program (ATTO). The ratios of inclusion of exon 7 were quantified and expressed as percentages of inclusion relative to the total intensities.
Exon Trapping Systems-Various deletion mutants of SMN1 containing intron 6, exon 7, and intron 7 were generated by PCR using the following primer sets: OF (5Ј-AAG CTT GAC TAT CAA CTT AAT TTC TGA TC-3Ј) and 600Rev (5Ј-GGA TCC CTG CTG TGT CTG CCT ACT AGT G-3Ј) for DM1 ( Fig. 2A); O-IF (5Ј-AAG CTT GTA AAA TGT CTT GTG AAA C-3Ј) and 600Rev for DM2; IF (5Ј-AAG CTT GCT ATC TAT ATA TAG CTA TCT ATG-3Ј) and 600Rev for DM3; 600Fwd (5Ј-AAG CTT GGC ATG AGC CAC TGC AAG AAA AC-3Ј) and OR (5Ј-GGA TCC GAG AAT TCT AGT AGG GAT GTA G-3Ј) for DM4; 600Fwd and OR-IR (5Ј-GGA TCC GTT TTA CAT TAA CCT TTC AAC T-3Ј) for DM5; and 600Fwd and IR (5Ј-GGA TCC CAC AAA CCA TAA AGT TTT AC-3Ј) for DM6. These PCR products were digested with NotI and BamHI and inserted into the exon trapping vector pSPL3 (Invitrogen) (23). Mutations in DM1-T were generated by PCR using the template as a mutant SMN mini-gene with mutagenized oligonucleotides (5Ј-AGG AAA AAA AGA AGG ATA TAA AGC TAT CTA TAT ATA G-3Ј and 5Ј-TTT TGT TTC ACA AGA CAT TTT ACT TAT TTT ATT CAA C-3Ј), and these PCR products were also digested with NotI and BamHI and cloned into the vector. All constructs were sequenced before use in experiments. Total cellular RNA was isolated from transfected COS-7 cells 24 h after the transfection using the RNeasy Mini Kit (Qiagen). Aliquots of 3.0 g of RNA were reverse-transcribed using the SA2 primer (5Ј-ATC TCA GTG GTA TTT GTG AGC-3Ј) and Moloney murine leukemia virus reverse transcriptase (Invitrogen). Splicing products were detected by PCR using a pSPL3 vector-specific primer set, SD6 (5Ј-TCT GAG TCA CCT GGA CAA CC-3Ј) and SA2. PCR was performed in a total volume of 40 l. Amplification was conducted as follows: 1 min at 94°C, 1 min at 60°C, and 1 min at 72°C for 30 cycles followed by 72°C for 5 min. The PCR products were electrophoresed in a 5% acrylamide gel.
Introduction of Antisense Oligonucleotides-All antisense oligonucleotides containing 2Ј-O-methyl modifications were synthesized by Japan Bio Service Inc. The sequences of the antisense oligonucleotides were 5Ј-GAU AGC UAU AUA UAG AU-3Ј (oligo-con), 5Ј-GAU AGC UAU AUA UAG AU-3Ј (oligo-pyr), and 5Ј-UGG AUG UUA AAA AGU A-3Ј (oligo-element1). Antisense oligonucleotides and various exon trapping vectors or mini-genes cloned into pCI vectors were co-transfected into COS-7 or HEK293T cells by lipofection. The final concentrations of antisense oligonucleotides in the culture medium ranged from 25 to 200 nM.
RNA-Protein Binding Assay-For the collection of nuclear extracts, SK-N-SH cells were homogenized in 50 volumes of 10 mM HEPES (pH 7.9) containing 10 mM KCl, 1 mM EDTA, 1 mM EGTA, 5 mM dithiothreitol and 1 mM (p-amidinophenyl) methane sulfonyl fluoride at 4°C. Buffers and any other solutions used in this study were sterilized before each use by filtration through a Steritop (Milipore Corporation) with a pore size of 220 nm. Following the addition of 10% Nonidet P-40 to a final concentration of 0.6%, homogenates were centrifuged at 15,000 rpm for 5 min. Pellets were resuspended in 10 volumes of 20 mM Tris-HCl (pH 7.5) containing 400 mM KCl, 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol, and 1 mM phenylmethylsulfonyl fluoride, followed by centrifugation at 15,000 rpm for 5 min. The supernatants thus obtained were stored at Ϫ80°C as nuclear extracts for pre-mRNA binding assays.
Pre-mRNA binding assays were performed as described previously (24). Briefly, sense strand RNAs were transcribed and uniformly labeled with [␣-32 P]GTPs in vitro by T7 RNA polymerase (TaKaRa) from templates of 214-bp sequences containing element 1 and partial sequences of wild-type (SMN1-W) or mutant SMN1 exon 7 (C to T transition, SMN1-T). Furthermore, as a control, a RNA probe was synthesized from the template, which was mutated in element 1 of mutant SMN1 (C to T transition, mSMN1-T). The RNAs were incubated on a heat block at 25°C with SK-N-SH cell nuclear extracts. The RNAprotein complexes were UV-irradiated (300,000 J/cm 2 ) at room temperature for 5 min, digested with 10 g of RNase A (Roche Molecular Biochemicals) at 37°C for 30 min and resolved by 10% SDS/PAGE.

Alternative Splicing of SMN Exon 7 in Various Cells-To
examine the splicing patterns of SMN exon 7, we used SMN1 and SMN2 mini-gene constructs including the genomic exon 6 -exon 8 in a pCI mammalian expression vector (21) (see "Experimental Procedures"). Furthermore, we also used expression vectors containing single nucleotide conversions that were C to T or T to C substitutions in exon 7 of the SMN1 or SMN2 mini-genes, respectively. These constructs were transiently transfected into human neuroblastoma SK-N-SH cells. After collection of total RNA from the cells, RT-PCR was performed using plasmid-specific primer sets (pCI forward and pCI reverse primers) to detect RNA processing of SMN in each construct-transfected cell. The wild-type SMN1 mini-gene expressed full-length SMN transcripts with no detectable SMN⌬7, whereas the wild-type SMN2 mini-gene produced lower levels of full-length SMN and abundant SMN⌬7 (Fig.  1A). Substitution of C to T located six nucleotides inside exon 7 of SMN1 led to the exclusion of exon 7, and the splicing patterns of exon 7 were similar to those of wild-type SMN2. In contrast, the mutant SMN2 (substitution of T to C in SMN2) mini-gene produced full-length SMN transcripts. These findings were consistent with previous results (21,25) and indicated that the C to T transition in SMN exon 7 had disrupted the regulation of SMN splicing.
Next we examined SMN exon 7 splicing in other kinds of cells such as COS-7 and HEK293 cells. The splicing patterns of SMN exon 7 were similar to those observed in the case of SK-N-SH cells (Fig. 1, B and C), suggesting that SMN exon 7 splicing could be regulated by common mechanisms using the same splicing machineries in both neuronal and non-neuronal cells.
Identification of Cis-acting Elements Responsible for the Skipping of SMN Exon 7-To determine the cis-acting ele- ments responsible for the skipping of SMN exon 7, we constructed various deletion mutants of both wild-type SMN1 and mutant SMN1 (C to T transition in exon 7) mini-genes that contained exon 7 with flanking introns 6 and 7, and these were cloned into the exon-trapping vector pSPL3 ( Fig. 2A). When the various deletion mutants of wild-type SMN1 (DM1ϳ6-C) were transfected into COS-7 cells, all constructs expressed the fulllength type of mRNA including SMN exon 7 (ϩexon 7), but never expressed an isoform lacking exon 7 (Ϫexon 7) (Fig. 2B). In contrast, the longest construct of mutant SMN1 (DM1-T), which contained a C to T transition in the SMN1 exon 7 with 156 bp of flanking intron 6 and 284 bp of flanking intron 7, expressed mRNAs with both inclusion and exclusion of exon 7 (40% inclusion) (Fig. 2C). The ratios of exon inclusion using mini-genes containing the C to T transition in exon 7 cloned into exon trap vectors were higher than those using the pCI vector. The differences indicate that there may be some cisacting elements that regulate the splicing of SMN exon 7 in regions that are different from the mini-gene sequences (about 600 bp) cloned into the exon trap vector. However, because the mini-genes containing the C to T transition in SMN1 exon 7 cloned into the exon trap vector (pSPL3) showed a significant increase in the exclusion of exon 7, these constructs are thought to be useful for determining the cis-elements responsible for the exclusion of exon 7 in the 600-bp mini-gene. A deletion mutant, DM3-T, which contained a deletion of 89 bp at the 5Ј-flanking regions of intron 6 from DM1-T, predominantly produced exon 7-included mRNA (82% inclusion), whereas the splicing pattern of deletion mutant DM2-T, which contained a deletion of 44 bp from DM1-T, was similar to that of DM1-T. Thus, the 45-bp region from Ϫ112 to Ϫ68 bp of flanking intron 6 may be a significant element for the exclusion of SMN exon 7 containing the C to T transition. We called this element "ele-ment 1". For the deletion mutants in flanking intron 7, deletion mutant DM6-T, which contained a deletion of 226 bp at 3Јflanking regions from DM1-T, increased exclusion of SMN exon 7 (25% inclusion) although deletion mutant DM5-T produced equivalent amounts of inclusion and exclusion of exon 7 similar to DM1-T and DM4-T (Fig. 2C). Therefore, the 66-bp region from ϩ59 to ϩ124 of flanking intron 7 (element 2) may be a critical element for the inclusion of SMN exon 7 containing the C to T transition. The nucleotide sequences of both elements 1 and 2 in the SMN1 and SMN2 genes are presented in Fig. 2D. The alignment showed that element 1 possessed the unique pyrimidine-rich sequences, but element 2 did not contain these unique sequences. These sequences showed complete matching between SMN1 and SMN2 except for one base substitution in both elements 1 and 2, suggesting that the two elements may be important for the regulation of the splicing of SMN1 exon 7 containing the C to T transition and wild-type SMN2.
We performed further analysis focused on the candidate cis-acting element 1 because it may be possible to develop experimental manipulation to increase the full-length SMN mRNA containing exon 7 if the function of element 1 can be inhibited. To confirm that disruption of element 1 leads to the inclusion of SMN1 exon 7 containing the C to T transition, the pyrimidine-rich sequences in element 1 of DM1 were mutated by substitution of pyrimidines with purines as shown in Fig.  3A. Mutation in DM1-T (mDM1-T) caused a significant increase of exon 7 inclusion compared with DM1-T, and the effects on the increase in exon 7 inclusion were similar to those of deletion mutant DM3-T (Fig. 3B).
Furthermore, to confirm that the cis-acting element is significant for the negative regulation of SMN exon 7 splicing, mini-genes containing SMN1 exon 6 to exon 8 and the C to T transition in exon 7 cloned into the pCI vector were mutated at  7) mini-genes that contain exon 7 with flanking introns 6 and 7 were cloned into pSPL3. BP, branching point; PPT, poly-pyrimidine tract. Element 1 (Ϫ112 to Ϫ68) and Element 2 (ϩ59 to ϩ124) are crucial elements for the exclusion and inclusion of SMN1 exon 7 containing the C to T transition, respectively. B and C, RT-PCR products from in vivo splicing assays. ϩexon 7 and Ϫexon 7 transcripts yield 317-bp and 263-bp fragments, respectively. Each PCR product was excised from the gel, and the DNA sequence was determined. In each case, the splicing occurred at the expected sites. All deletion mutants of wild-type SMN1 (DM1ϳ6-C) express ϩexon 7 transcripts (B), whereas each deletion mutant of SMN1 containing the C to T transition (DM1ϳ6-T) produces both ϩ and Ϫexon 7 (C). The scores shown in B and C are percentages of exon 7 inclusion relative to the total transcripts. Values represent the means of four analyses. Note that the splicing patterns are changed in the cases where DM3-T and DM6-T were transfected compared with the full-length constructs, DM1-T or DM4-T, respectively. D, comparisons of the nucleotide sequences of element 1 and 2 between SMN1 and SMN2. *, differences in the nucleotides between SMN1 and SMN2 in these elements. Element 1 contains pyrimidine-rich sequences (underlined). element 1 (substitution of pyrimidines with purines in a similar manner to the mutations of element 1 in the exon trapping vectors) and then examined to see whether this mutation increased the inclusion of exon 7 of SMN1 containing the C to T transition in exon 7. The inclusion of exon 7 of the construct increased about 3.4-fold compared with that of the original construct (Fig. 3C). Taken together with the exon trapping data, element 1 is demonstrated to be an important sequence for the exclusion of SMN exon 7 containing the C to T transition.

Effects of Treatment with Antisense Oligonucleotides Directed toward Element 1 on the Splicing of SMN Exon 7-Hav-
ing demonstrated that element 1 in intron 6 of the SMN gene plays a crucial role in the regulation of exon 7 splicing, it was of interest to examine whether treatment of cells expressing DM1-T with antisense oligonucleotides was sufficient to lead to an increase in the inclusion of SMN exon 7 containing the C to T transition. We synthesized antisense oligonucleotides directed toward element 1 (oligo-element1, Fig. 4A). We further synthesized antisense oligonucleotides that were directed toward the poly-pyrimidine tract on intron 6 of SMN1 (oligo-pyr) and another sequence in the intron 6 (oligo-con) as controls. These antisense oligonucleotides were co-transfected with the DM1-T vector into COS-7 cells, and total RNA was extracted 24 h after the transfection. RT-PCR was performed to examine the effects of these antisense oligonucleotides on the in vivo splicing of SMN exon 7. Treatment with oligo-element1 led to increased inclusion of SMN exon 7, and the increase was dependent on the amount of antisense oligonucleotide (Fig. 4B). In contrast, treatment with the other oligonucleotides did not lead to increase inclusion of SMN exon 7, but treatment with oligo-pyr, alternatively, showed a slight decrease in the inclusion (Fig. 4, C and D).
We also examined the effects of the antisense oligonucleotides on the mini-gene of SMN1 containing exon 6 -exon 8 and the C to T transition in the exon 7 cloned into the pCI expres-sion vector. RT-PCR analysis showed that treatment with oligo-element1 caused an ϳ2.5-fold increase in the inclusion of SMN exon 7 compared with non-treatment-antisense oligonucleotides (Fig. 5A). The maximal effects on the inclusion of exon 7 were reached at or less than 25 nM oligo-element1. In contrast, treatment with oligo-pyr (Fig. 5C) or oligo-con (data not shown) at concentrations of up to 200 nM did not lead to any changes in exon 7 inclusion. Next, we examined whether oligo-element1 led to the inclusion of exon 7 of wild-type SMN2. As expected, treatment with the antisense oligonucleotides led to a 2.7-fold increase in the inclusion of SMN2 exon 7, similar to the case of SMN1 containing the C to T transition (Fig. 5B). Alternatively, both oligo-pyr and oligo-con did not change the inclusion of SMN2 exon 7 (data not shown). These findings indicate that antisense oligonucleotides directed toward element 1 are an effective tool for increasing exon 7 inclusion in mutant SMN1 containing the C to T transition in exon 7 and wild-type SMN2. Moreover, taken together with the exon trapping experiments, the cis-acting element plays important roles in the regulation of the splicing of SMN exon 7 containing the C to T transition.
A Trans-acting Protein Specifically Binds to the Pre-mRNA Sequences Containing Element 1-To detect proteins associated with element 1, 32 P-labeled RNA and SK-N-SH nuclear extracts were mixed and UV cross-linked. In this experiment, as shown in Fig. 6A, we used synthetic RNAs that were transcribed from templates of 214 bp sequences containing partial sequences of intron 6 with element 1 and exon 7 of wild-type (SMN1-C) or mutant SMN1 (C to T transition, SMN1-T). Furthermore, as a control, we synthesized RNA probes from the template, which was mutated in element 1 of SMN1-T (mSMN1-T) as described under "Experimental Procedures." After digestion of these complexes with RNase A, proteins cross-linked to RNA were resolved by SDS-PAGE followed by autoradiography. Several kinds of bands were detected to bind to each probe in the same pattern, but an ϳ33-kDa protein specifically bound to the SMN1-T probe but not to the SMN1-C probe (Fig. 6B). The binding of the ϳ33-kDa protein could not be detected in binding assays using the mSMN1-T probe, sug-gesting that the ϳ33 kDa protein specifically bound to SMN1 pre-mRNA containing the C to T transition, and that the element 1 was essential for this binding. We performed RNA binding experiments using shorter RNA probes containing element 1 to determine whether the ϳ33 kDa protein bound directly to element 1, but we could not obtain the binding activities of these shorter probes to nuclear proteins including the ϳ33 kDa protein. This suggests that the ϳ33 kDa protein

FIG. 4. Treatment with antisense oligonucleotides directed toward element 1 increases the expression of transcripts containing SMN exon 7.
A, the regions where the three types of antisense oligonucleotides (oligo-element1, oligo-con, oligo-pyr) hybridize. B, RT-PCR of in vivo splicing using the exon trapping system. Treatment with oligo-element1 leads to an increase in the ratio of exon 7 inclusion relative to the total transcripts. The effects observed were dependent on increasing concentrations of oligo-ele-ment1 (lower panel). Treatment with oligo-con did not affect the splicing of SMN exon 7 (C) with oligo-pyr leads to decrease the inclusion (D). The percentages of exon 7 inclusion are shown below each lane and represent the means of four experiments.
FIG. 5. The effects of treatment with antisense oligonucleotides on the splicing of SMN exon 7 using SMN mini-genes cloned into pCI vectors. Both the SMN1 mini-gene (exon 6 -8) containing the C to T transition (A) and the wild-type SMN2 mini-gene (B) produced an increase in the amount of full-length transcripts when the concentration of oligo-element1 was equal to, or more than, 25 nM. The percentages of exon 7 inclusion are shown below each lane and represent the means of four experiments. C, treatment with each concentration of oligo-pyr did not affect the splicing of SMN1 exon 7 containing the C to T transition FIG. 6. An ϳ33-kDa protein specifically binds to the pre-mRNA sequence of SMN1 containing a C to T transition in the exon 7. A, the construction of each probe. Sense strand RNAs were transcribed and uniformly labeled with [␣-32 P] GTPs in vitro by T7 RNA polymerase (TaKaRa) from templates of 214-bp sequences containing element 1 and partial sequences of wild-type (SMN1-C) or mutant SMN1 exon 7 (C to T transition, SMN1-T). An RNA probe was also synthesized from the template that was mutated in element 1 of mutant SMN1 (C to T transition, mSMN1-T). B, nuclear extracts from SK-N-SH cells were incubated with RNA probes, UV-irradiated, and digested with RNase A as described under "Experimental Procedures." An ϳ33-kDa protein (*) specifically binds to the probe SMN1-T but not to the other probes. The arrow shows bands of ϳ40-kDa proteins that bind to the mSMN1-T probe. may bind to the probes by recognizing the higher order structure of the pre-mRNA containing the C to T transition in the SMN exon 7. Therefore, it is still unclear whether or not the ϳ33-kDa protein binds directly to element 1.
We also observed that complexes with molecular masses about 40 kDa specifically bound to the mSMN-T probe (Fig.  6B). However, these complexes did not bind to either the SMN1-C or -T probes. The significance of this binding has yet to be elucidated. DISCUSSION SMN1 mRNA expresses a full-length transcript, whereas SMN2 produces a low level of the full-length transcript and predominantly an isoform lacking exon 7 (2,16,17). The critical difference between SMN1 and SMN2 is a silent nucleotide transition in SMN exon 7; that is, SMN1 contains a C located six nucleotides inside exon 7, whereas SMN2 contains a T at this position. This transition leads to alteration of the recognition of exon 7 by components of the splicing machinery (18,21). Cis-acting RNA elements like ESEs facilitate the splicing by recruiting general splicing factors to the adjacent exon/intron junctions (26 -28). A recent study demonstrated that Tra2␤-1, a member of the serine-arginine-related proteins of splicing factors, interacted with an exon 7 ESE common to both SMN1 and SMN2 pre-mRNAs (22). However, the critical C to T transition is not contained within the ESE where Tra2␤-1 binds directly. Furthermore, the transition does not change the binding activity of Tra2␤-1 to the ESE. Thus, it was still unclear why the C to T transition leads to the lack of exon 7 in SMN2. Therefore, in the present study, we examined the critical cis-acting elements on the SMN pre-mRNA responsible for the skipping of SMN exon 7, which contains the C to T transition.
Deletion analysis of SMN1 pre-mRNA sequences showed that the regions from Ϫ112 to Ϫ68 bp of flanking intron 6 (element 1) and from ϩ59 to ϩ124 of flanking intron 7 (element 2) are significant elements for the exclusion of SMN1 exon 7 containing the C to T transition and for the inclusion of SMN1 exon 7, respectively. However, deletion of these elements from wild-type SMN1 pre-mRNA did not affect the splicing of SMN exon 7. Therefore, combined conditions of deletion of these elements and the C to T transition in exon 7 are necessary for the alteration of SMN exon 7 splicing patterns. The elements that we identified in the present study have not been demonstrated to be critical for the splicing of SMN exon 7. Therefore, it is unknown why the deletion of these elements leads to changes in the processing for exon 7. However, a possible mechanism is that alteration of the splicing patterns of SMN exon 7 may result from the binding of specific regulatory factors to these elements caused by changes in the higher order structure of the pre-mRNA containing the C to T transition in exon 7, and the binding of regulatory factors may sterically hinder the recognition or usage of 3Ј-or 5Ј-splice sites by splicing machineries.
Previously, heterogeneous nuclear ribonucleoprotein (hnRNP) A1 was demonstrated to inhibit splicing of HIV-tat via a hnRNP A1-responsive intron-splicing silencer (29). In this paper, hnRNP binding sites in the tat intron were shown to match the hnRNP A1 consensus binding site, UAGGG(U/A) (30). The element 1 identified in the present study is considered to act for an intron splicing silencer because of the increase in SMN splicing by deletion or mutation of the element. However, element 1 does not contain the hnRNP A1 consensus sequence, although it contains the unique pyrimidine-rich sequence. Therefore, element 1 could not be regulated by hnRNP A1, but a novel molecule(s) may associate with this element to inhibit the splicing of SMN1 containing the C to T transition or wildtype SMN2. Indeed, we found that an ϳ33-kDa nuclear protein could interact with RNA probes containing element 1 and the C to T transition in SMN exon 7, although it is unclear whether or not the ϳ33-kDa protein binds directly to element 1. The protein did not associate with the probe containing the wildtype SMN exon 7, suggesting that the ϳ33-kDa protein does not basically act as a regulator of the splicing of wild-type SMN exon 7, but rather it plays crucial roles in the skipping of exon 7 only in the case of SMN exon 7 containing the C to T transition. However, it remains unclear why the binding of the ϳ33-kDa protein to SMN pre-mRNA sequence containing element 1 inhibits the splicing and leads to the resultant exclusion of SMN exon 7. Identification and characterization of the ϳ33-kDa protein are needed to elucidate the mechanisms.
Only homozygous deletions or mutations of SMN1 result in the SMA phenotype (6,9). SMA patients retain SMN2 alleles. Therefore, increasing the inclusion of SMN2 exon 7 could provide a new therapy to lower the clinical severity of SMA. We have demonstrated that treatment of cultured cells with an antisense oligonucleotide directed toward element 1, which we identified as the cis-acting element for the exclusion of SMN exon 7 containing the C to T transition, increased the inclusion of wild-type SMN2 exon 7 and SMN1 exon 7 containing the transition. The efficiencies of antisense oligonucleotides on the increase in inclusion of SMN exon 7 are dependent on the interaction between the oligonucleotides and the target sequence. The resultant binding of antisense oligonucleotides to the target sequence, such as element 1 in the pre-mRNA of SMN, may inhibit the direct interaction of splicing silencers with the element and lead to the inclusion of SMN exon 7. Under the present conditions, treatment with an antisense oligonucleotide caused a 2.5fold increase in the inclusion of SMN exon 7, but did not induce a shift to complete inclusion of exon 7 due to the technical limits of antisense methods. These effects were equivalent to a previously reported treatment with an antisense oligonucleotide designed to hybridize to the 3Ј splice junction site of exon 8 (25). In this report, the antisense oligonucleotide also increased SMN2 exon 7 inclusion 2.5-fold. Therefore, the method of antisense treatment may be limited in its effects on the increase in inclusion of SMN exon 7. It has been demonstrated that increased levels of SMN2 copy numbers correlate with decreasing severity of the SMA phenotype. Therefore, even if treatment with antisense oligonucleotides leads only to a 2.5-fold increase in fulllength SMN transcripts, it may still lower the clinical severity of SMA.
In the present study, we found a pyrimidine-rich sequence in element 1 and a 66-bp sequence in element 2 that regulate the splicing of SMN exon 7 containing the C to T transition. Further experiments that clarify the functions of these elements, including the identification of their RNA-binding proteins, will address the mechanisms of splicing regulation of SMN exon 7. Moreover, based on these observations, experimental manipulation to modify the function of the cis-acting elements or the trans-acting factors might allow the development of therapeutic strategies for SMA.