Modulation of survival motor neuron pre-mRNA splicing by inhibition of alternative 3' splice site pairing.

Spinal muscular atrophy is caused by the loss of functional survival motor neuron ( SMN1 ) alleles. A transla-tionally silent nucleotide transition in the duplicated copy of the gene ( SMN2 ) leads to exon 7 skipping and expression of a nonfunctional gene product. It has been suggested that differential SMN2 splicing is caused by the disruption of an exonic splicing enhancer. Here we show that the single nucleotide difference reduces the intrinsic strength of the 3 (cid:1) splice site of exon 7 2-fold, whereas the strength of the 5 (cid:1) splice site of the exon 7 is not affected. Thus, a decrease in splice site strength is magnified in the context of competing exons. These data suggest that lower levels of exon 7 definition not only reduce intron 6 removal but, more importantly, increase the efficiency of the competing exon 7 skipping pathway. Antisense oligonucleotides were tested to modulate exon 7 inclusion, which contains the authentic translation stop codon. Oligonucleotides directed toward the 3 (cid:1) splice site of exon 8 were shown to alter SMN2 splicing in favor of exon 7 inclusion. These results suggest that antisense oligonucleotides could be used as a therapeutic strategy to counteract the progression of SMA.

Spinal muscular atrophy is caused by the loss of functional survival motor neuron (SMN1) alleles. A translationally silent nucleotide transition in the duplicated copy of the gene (SMN2) leads to exon 7 skipping and expression of a nonfunctional gene product. It has been suggested that differential SMN2 splicing is caused by the disruption of an exonic splicing enhancer. Here we show that the single nucleotide difference reduces the intrinsic strength of the 3 splice site of exon 7 2-fold, whereas the strength of the 5 splice site of the exon 7 is not affected. Thus, a decrease in splice site strength is magnified in the context of competing exons. These data suggest that lower levels of exon 7 definition not only reduce intron 6 removal but, more importantly, increase the efficiency of the competing exon 7 skipping pathway. Antisense oligonucleotides were tested to modulate exon 7 inclusion, which contains the authentic translation stop codon. Oligonucleotides directed toward the 3 splice site of exon 8 were shown to alter SMN2 splicing in favor of exon 7 inclusion. These results suggest that antisense oligonucleotides could be used as a therapeutic strategy to counteract the progression of SMA.
Proximal spinal muscular atrophy (SMA) 1 is a common human genetic disease that is the leading cause of hereditary infant mortality (1)(2)(3). It is characterized by the progressive degeneration of the anterior horn stem cells of the spinal cord with consequent paralysis of the trunk and limbs. Three clinical groups of the disease (I-III) have been described based on the decreasing severity of the symptoms (4). SMA has been linked to deletions or mutations of the survival of motor neuron (SMN) gene, which has been mapped as an inverted repeat to chromosome 5 at 5q13 (4). Homozygous absence of the telomeric copy (SMN1) correlates with development of SMA (5,6). By contrast, alterations within the centromeric SMN gene (SMN2) do not produce any known phenotype (4). It has been demonstrated that SMN levels produced from the SMN2 locus modify the severity of the disease in a dose-dependent manner (7,8). However, SMN2 alone cannot provide protection from SMA (9 -11).
The majority of the SMN transcript consists of 9 exons encoding for a 294-amino acid (38 kDa) gene product. The ubiquitously expressed protein is detected at especially high levels in neuronal cells (7,8). Until recently, it was unclear why only mutations in the telomeric SMN1 mutations result in SMA. A genomic sequence comparison of the two genes revealed that SMN1 and SMN2 encode for the identical protein (12). However, three alternatively spliced transcripts generated with different efficiencies have been described for each locus (4). SMN1 primarily produces the full-length form of SMN, whereas differential splicing of the SMN2 pre-mRNA predominantly produces an isoform lacking exon 7 (SMN⌬7). Comparison of SMN transcripts revealed a direct relationship between SMA and exon 7 skipping. Furthermore, it was demonstrated that SMN⌬7 is unable to efficiently self-associate, a property that correlates with disease severity (13). Thus, SMN⌬7 is not functionally equivalent to full-length SMN, and the failure of SMN2 to fully compensate for mutations in SMN1 is because of a differential pre-mRNA-splicing event.
The only critical nucleotide change between SMN1 and SMN2 affecting the inclusion of exon 7 has been pinpointed to a single C to T base difference located six nucleotides inside exon 7 (12,14) (Fig. 1). This transition may disrupt a putative exon-splicing enhancer (ESE) (14), a cis-acting element that generally serves as binding sites for members of the serinearginine (SR)-rich family of splicing factors (15). An assembled splicing enhancer complex in turn promotes splice site and exon recognition by assisting in the recruitment of the splicing machinery to the adjacent intron through protein-protein interactions (16,17). Previous work demonstrated that the recognition of weak splice site signals could be aided significantly by the presence of ESEs. By contrast, ESEs have little or no detectable effect on the activation of strong splice sites that conform to the consensus sequence (15)(16)(17)(18). Most ESEs studied in mechanistic detail activate 3Ј splice sites that contain suboptimal polypyrimidine tracts (py), a stretch of nucleotides immediately upstream of the exon/intron junction. Because py tracts and the 3Ј splice site serve as binding sites for the U2 small nuclear ribonucleoprotein auxiliary factor (U2AF) (19 -23), a model has emerged suggesting that enhancer complexes assembled on ESEs recruit U2AF to a weak 3Ј splice site (24 -26). In the case of SMN2, it was suggested that the C to T nucleotide difference in exon 7 might disrupt the integrity of a putative ESE to a degree that renders spliceosomal assembly and exon 7 recognition insignificant (12,14). As a consequence, exon 7 is skipped in SMN2, and SMN⌬7 is the product of the alternative processing event. A recent report has tested this hypothesis and demonstrated the presence of an ESE within exon 7 (27). However, the critical C to T transition observed between SMN1 and SMN2 was not contained within this re-gion. In fact, within the nucleotide sequences around the C to T transition, no detectable ESE activities were observed. Thus, it is still unclear why the C to T transition leads to exon 7 skipping in SMN2.
Because SMN2 cannot compensate for mutations within SMN1, SMN1 is the SMA-determining gene. More importantly, however, because all individuals with SMA have retained their SMN2 allele, therapy directed toward increasing SMN2 exon 7 inclusion could provide a promising tool to lower the clinical severity of SMA. Recent studies have demonstrated that overexpression of hTra2, a splicing factor that was shown to interact with exon 7 of SMN, was able to tilt the balance of exon 7 inclusion in the SMN2 splicing pathway such that the majority of SMN2 mRNAs were full-length (28).
To gain insights into the mechanism that leads to SMN2 exon 7 skipping, we have evaluated and compared the relative strength of exon 7 splice sites. Here we provide evidence that the C to T transition within exon 7 of SMN2 decreases the intrinsic strength of the upstream 3Ј splice site (3Јss) by 2-fold. By contrast, the 5Ј splice site (5Јss) of exon 7 is not affected by the nucleotide transition. We conclude that the competition between the 3Јss of exons 7 and 8 to pair with the 5Јss of exon 6 amplifies the relatively minor differences in exon 7 splice site strength. These observations suggested that modulation of the 3Јss competition could provide the means to facilitate the inclusion of exon 7, which contains the authentic translation stop codon into the SMN2-splicing pathway. We find that the application of antisense oligonucleotides that reduce the recognition of the 3Ј splice site of exon 8 increases the incorporation of exon 7 into the fully processed transcript.

MATERIALS AND METHODS
Plasmids-SMN1 and SMN2 mini-gene constructs containing exon 6, intron 6, exon 7, intron 7, and exon 8 in a pCI mammalian expression vector background were a gift from Elliot Androphy. SMN⌬8 constructs were generated by deleting a NotI and XcmI fragment from the SMN mini-gene. SMN⌬6 was generated through the deletion of a PstI and NheI fragment of the SMN mini-gene. hetSMN constructs were generated through XhoI and BclI digestion of the SMN mini-gene. The deleted region contained exon 6 and most of intron 6. However, in contrast to the SMN⌬6 constructs, the hetSMN constructs retain a pseudointron inherent to the pCI backbone. The pseudointron is located upstream of the SMN sequences and contains a 5Ј splice site derived from exon 1 of ␤-globin and a 3Ј splice site derived from IgM.
Transfections and Reverse Transcription-PCR Analysis-Human cervical carcinoma (C33a) cells were grown in minimal essential media (BioWhittaker) supplemented with 2 mM glutamine and 10% fetal bovine serum. Transfections were performed using the LipofectAMINE 2000 transfection kit (Life Technologies, Inc.). 1 g of DNA was used to transfect ϳ3 ϫ 10 5 cells plated 24 h before transfection. At 24 h post-transfection, cells were harvested, and total RNA was isolated using TRIzol reagent (Life Technologies, Inc.). Total RNA was reversetranscribed with Moloney murine leukemia virus reverse transcriptase for 1 h at 37°C using poly-dT priming. Radioactive PCR analysis was then performed with different primer sets to identify spliced products. PCR was done for 20 cycles in the presence of 0.2 mM dNTP supplemented with trace amounts of [␣-32 P]dATP. Each cycle consisted of a 45-s denaturation step at 94°C, a 45-s annealing step at 56°C, and a 90-s extension step at 72°C. For analysis of transfections with SMN⌬8, primers used selectively amplified both unspliced and spliced products. Primers used to amplify unspliced products were pCI-for (5Ј-GCTAAC-GCAGTCAGTGCTTC-3Ј) and pCI-rev (5Ј-GTATCTTATCATGTCTGC-TCG-3Ј). To amplify spliced products, primers used were Tin6-for (5Ј-TAATACGACTCACTATAGGCCATATAAAGCTATCTATATATAGC-3Ј) and pCI-rev. For analysis of transfections with SMN⌬6 and het- SMN, pCI-for and pCI-rev primers were used. As a normalization control, the hypoxanthine phosphoribosyltransferase gene (HPRT) was simultaneously amplified by using the primers HPRT-Fw 5Ј-AAGGA-GATGGGAGGCCAT and HPTR-Rev 5Ј-GTTGAGAGATCATCTCCAC-CAAT. Reaction products were boiled and fractionated on a 6% polyacrylamide gel. The resulting bands were visualized and quantitated using a PhosphorImager. The splicing efficiency was calculated after normalization to the internal control HPTR by computing the fraction of spliced product counts over the sum of spliced and unspliced counts. Transfection experiments were repeated at least three times.
Antisense Oligonucleotide Application-Stable cell lines expressing either the wild type SMN1 or SMN2 mini-gene were generated by G418 selection. Briefly, C33a cells were co-transfected with the plasmid of interest and pSV neo , a neo selection plasmid (a gift from Dr. Marian Waterman), using the LipofectAMINE 2000 reagent (Life Technologies, Inc.). After 12 h, G418 was added to the medium to a final concentration of 2 mg/ml. Selected clones were sub-cultured and amplified. All antisense oligonucleotides containing 2Ј-O-methyl and phosphorothioate backbone modifications were obtained from Dharmacon Research, Inc., Boulder, CO. Antisense oligonucleotides were added to stable cell lines in exponential growth phase using LipofectAMINE 2000. The final concentration of antisense oligonucleotides in the culture medium was 100 nM unless stated otherwise. Cells were harvested, and total RNA was isolated using TRIzol reagent (Life Technologies, Inc.) 24 h after antisense oligonucleotide addition unless stated otherwise. As a loading control, the hypoxanthine phosphoribosyltransferase gene (HPRT) was simultaneously amplified by using the primers HPRT-Fw 5Ј-AAGGAG ATGGGAGGCCAT and HPTR-Rev 5Ј-GTTGAGAGATCATCTCCA CCAAT. The sequence of the antisense oligonucleotides directed toward the branch point (oligo bp ) is 5Ј-CUUCAUAUGUCAGAGUGUACAG-3Ј, toward the polypyrimidine tract (oligo py ) is CCUGCAAAUGAGAAAU-UAG, and toward the splice site junction (oligo jnct ) is CUAGUAUUUC-CUGCAAAUGAG.

RESULTS
It has been suggested that differential processing of the SMN1 and SMN2 pre-mRNAs is caused by the disruption of an ESE located within exon 7 of the transcript (12,14). The single nucleotide difference within exon 7 of SMN1 and SMN2 is located six nucleotides downstream of the 3Ј splice site junction. Whereas the C to T transition does not change the coding sequence of the transcript, the recognition of exon 7 by components of the splicing machinery is severely altered. Less than 20% of SMN2 pre-mRNAs are spliced to include exon 7, whereas more than 90% of SMN1 pre-mRNAs are processed to include exon 7. Generally, exon skipping is a consequence of the inability of the splicing machinery to recognize and define the exon (29). Exon definition in turn predominantly relies on the strength of 3Ј and 5Ј splice site signals flanking the exon. Cis-acting RNA elements like ESEs or exonic splicing silencer elements facilitate this process by recruiting general splicing factors to the adjacent exon/ intron junctions, thus augmenting the intrinsic strength of 3Ј or 5Ј splice site (16,30). Because the C to T transition in SMN2 is proposed to disrupt a putative ESE element, it is expected that the efficiency at which the splicing machinery recognizes the 5Ј or 3Ј splice sites will be reduced. To test this hypothesis, we generated a set of SMN mini-genes that was assayed in cell transfection splicing assays.
The C to T Transition in SMN2 Decreases the Utilization of the 3Ј Splice Site of Exon 7-To test the effects the C to T transition has on the strength of the 3Јss or 5Јss of exon 7, we designed two sets of identical mini-genes for SMN1 and SMN2 ( Fig. 2A). SMN⌬8 lacks the majority of intron 7 and all of exon 8 and, thus, tests the activity of the 3Јss of exon 7. SMN⌬6 lacks exon 6 and the vast majority of intron 6 and, thus, tests the activity of the 5Јss of exon 7. Compared with the wild type substrate, these pre-mRNAs do not contain competing exons. Therefore, splice site activation leads to a single intron removal event. Processed as well as unprocessed RNA were amplified from total RNA preparations by PCR using intron-specific or exon-specific primer sets (see "Materials and Methods"). In the native context of competing splice sites, processing of the SMN1 and SMN2 wild type mini-genes lead to the characteristically strong difference in exon 7 inclusion (Fig. 2, B and C). However, as demonstrated in Fig. 2B, the difference of intron 6 removal was observed to be less pronounced for the SMN1⌬8 and SMN2⌬8 mini-genes (lanes 5 and 6). Before calculating processing efficiencies, band intensities were normalized to the intensity of an internal control, the amplification product of the hypoxanthine phosphoribosyltransferase gene (HPRT) (see "Materials and Methods"). Within a data set of four independent transfection experiments, a reproducible difference in splice site strength was determined to be approximately 2-fold (compare splicing efficiencies of 20 Ϯ 3% for SMN1⌬8 with 9 Ϯ 2% for SMN2⌬8) (Fig. 2B, Table I). Thus, the C to T transition in SMN2 decreases the intrinsic strength of the 3Јss strength of exon 7 by 2-fold.
To investigate if the C to T transition in SMN2 leads to a decrease in the strength of the 5Јss of exon 7, we compared the SMN⌬6 substrates. As demonstrated in Fig. 2C, the steady state levels of intron 7 removal were observed to be surprisingly low (0.5 Ϯ 0.3% for SMN1⌬6 and 0.8 Ϯ 0.4% for SMN2⌬6, Table I). Because we did not detect a statistically significant difference between the efficiencies of SMN1⌬6 and SMN2⌬6 splicing, we conclude that the C to T transition in SMN2 does not affect the 5Јss strength of exon 7 within the context of the single intron pre-mRNA.
The experiments described above indicate that the 3Јss of exons 7 and 8 compete to pair with the 5Јss of exon 6. To evaluate the importance of exon 6 nucleotides in this competition we exchanged exon 6 of SMN with exon 1 derived from ␤-globin to generate a heterologous substrate, hetSMN ( Fig.  2A). As summarized in Table I, quantitative analysis of the transfection assays shows that the ratio of exon 7 inclusion observed for the hetSMN pre-mRNAs are very similar to the levels observed with the wild type SMN mini-genes (Table I).
We conclude that the decision to include or exclude exon 7 into the splicing pathway is independent of the nature of the upstream exon and its 5Ј splice site. Taken together, these results demonstrate that the nucleotide difference in SMN2 decreases the utilization of the 3Јss but not of the 5Јss of exon 7. Interestingly, the data also suggest that the relatively minor differences in the strength of the SMN1 and SMN2 exon 7 splice sites are drastically amplified in the native cis-competition context.
Antisense Oligonucleotide Application to Modulate the Competition of Exon 7 and Exon 8 Increase the Ratio of Exon 7 Inclusion in SMN2-Because a greater number of SMN2 gene copies in SMA patients correlates with a milder phenotype of SMA, it has been postulated that the low amount of full-length SMN generated from SMN2 modulates disease severity (7,8). Alteration of the SMN2-splicing pattern to generate higher fractions of full-length SMN gene product may therefore be a promising therapeutic approach to treat SMA. Our mutational analysis suggested that the splicing machinery recognizes and processes exon 7 efficiently in the absence of a splice site choice. Therefore we reasoned that an antisense oligonucleotide directed toward the 3Ј splice site of SMN2 exon 8 could potentially tilt the balance in favor of exon 7 inclusion. As illustrated in Fig. 3A, this antisense oligonucleotide strategy may also lead to intron 7 retention. However, intron 7 retention does not change the coding region for SMN because the translation stop codon is located within the last nucleotides of exon 7. Stable cell lines expressing the SMN1 and SMN2 wild type mini-genes were created by G418 selection. The cells were then incubated in the presence of various antisense oligonucleotides designed to hybridize to components of the 3Јss of exon 8, the proposed branch point sequence (oligo bp ), the polypyrimidine tract (oli-go py ), or the splice site junction (oligo jnct ). As illustrated in Fig.  3B, treatment of the cell lines with oligo jnct , but not with oligo bp (or oligo py , data not shown), led to a reproducible increase in the generation of exon 7 containing mRNA. As expected, we also detected a significant fraction of stable SMN RNA that retained intron 7. Quantitative analysis indicates that the antisense oligonucleotide application leads to a 5-fold increase in the ratio of exon 7 inclusion over exon 7 skipping. The absolute levels of exon 7 inclusion increased from 20% in the absence of antisense oligonucleotides up to 50% in the presence of the junction antisense oligonucleotide. Thus, treatment of the stable cell lines expressing the SMN2 mini-gene with the junction antisense oligonucleotide modulates splice site pairing to include exon 7. As illustrated in Fig. 4A, the application of the antisense oligonucleotide is concentration-dependent, with an apparent K d of 20 Ϯ 10 nM. Maximal splice site modulation is reached at ϳ100 nM oligo jnct . In comparison, incubation of the bp or py antisense oligonucleotides at concentrations of up to 500 nM did not lead to any measurable increase in the fraction of exon 7 inclusion (data not shown). Fig. 4B shows a time course of the antisense oligonucleotide treatment. Modulation of SMN2 splicing is observed within the first 7 h of oligo jnct transfection and is maintained throughout the length of the experiment (72 h). Maximal efficiency was observed after 24 h of treatment, and the steady decrease in antisense activity thereafter is most likely because of the degradation of the 2Ј O-methyl phosphorothioate oligonucleotide. Although oligo jnct manipulated the ratio of exon 7 inclusion in SMN2 pre-mRNA splicing, its effect on SMN1 processing was manifested predominantly by the generation of intron 7 retaining SMN1 RNAs (data not shown). We conclude that an antisense oligonucleotide treatment directed toward the intron 7/exon 8 junction efficiently and specifically modulates the splicing pattern of SMN2.

DISCUSSION
SMA is caused by gross mutations in the telomeric copy of the duplicated SMN genes. The centromeric copy of SMN, SMN2, cannot compensate for the loss of SMN1 because the majority of SMN2 pre-mRNAs undergo differential RNA splicing. A single C to T nucleotide difference that lies within exon 7 of the SMN transcript leads to preferential exon 7 exclusion in SMN2, ultimately leading to production of a truncated and unstable peptide. Thus, the C to T transition drastically affects the efficiency and kinetics at which the splicing machinery recognizes and defines exon 7. To gain insights into the mechanisms by which the C to T transition alters exon 7 definition, we have compared the strength of the SMN1 and SMN2 exon 7 splice sites. Our data demonstrate that in the absence of competing splice sites, i.e. in the context of a single intron removal event, a 2-fold difference was observed between the 3Ј splice sites of SMN1 and SMN2. In contrast, we did not detect a difference between the 5Ј splice sites. Thus, the C to T transition does not appear to drastically decrease the intrinsic strength of the exon 7 splice sites. The 2-fold variance clearly cannot account for the greater difference observed in the native context of competing exons. Although 90% of SMN1 mRNAs include exon 7, only 20% of exon 7-containing mRNAs are observed for SMN2. Assuming similar transcriptional activities, the difference in the mole number of full-length SMN mRNAs produced from each SMN locus is therefore approximately 5-fold. To evaluate these differences, it is useful to express the percentage of exon 7 inclusion by the fraction of the apparent rate constant of exon 7 inclusion (k inc ) over the apparent rate constant of exon exclusion (k ex ). Thus, the fraction of exon 7 inclusion equals k inc /k ex . Because the ratio of exon 7 inclusion is 90% for SMN1 pre-mRNA processing, the rate of exon inclusion is 9 times faster than the rate of exon skipping. In case of SMN2, k ex is 5 times faster than k inc , leading to a 20% inclusion efficiency. In the context of competing exons, k inc /k ex therefore changes from 9 in SMN1 to 0.2 in SMN2, a 45-fold difference. Removal of either intron 6 or intron 7 most likely commits the splicing pattern to exon 7 inclusion. Thus, changes in the rate of exon inclusion depend on the efficiency of either intron 6 or intron 7 splicing. Because our comparison of intron 6 and intron 7 removal indicates that neither process is altered by more than 2-fold, it is tempting to argue that the SMN2 characteristic exon 7 skipping phenomena arises mainly by increasing the efficiency of the competing exon exclusion reaction. Based on the above comparisons, a 2-fold decrease in the efficiency of intron 6 or intron 7 removal is expected to be  hrs accompanied by a 20-fold increase in the rate of k ex . These considerations suggest that a more efficient recognition of exonic sequences by the action of exonic splicing enhancers, for example, affect not only the rate of intron removal but also the competing rate of exon skipping. Therefore, exon definition may incorporate a mechanism that also influences rates of splice site juxtapositioning. Alternative splicing of the CD45 pre-mRNA has recently been attributed to the combined actions of exonic enhancer and silencer elements (31). It was shown that a single nucleotide polymorphism like SMN, located within exonic sequences, changed the ratio of k inc /k ex by ϳ30-fold. Using single intron substrates it was further demonstrated that the efficiency of only one of the two intron removal events was altered by approximately 4-fold. By analogy to the above considerations, these results are in agreement with the notion that an increase in the level of exon definition divergently alters the rates of exon inclusion and exon exclusion. A more efficient recognition and assembly of the exonic region by the splicing machinery leads to a less efficient splice site pairing of the upstream splice site with the competing downstream splice site. As demonstrated recently, recruitment of spliceosomal factors to nonsplicing competent regions caused inhibition of intron removal. 2 In this case, it was argued that the non-productive splicing complex efficiently competes for splice site pairing across the intron, thereby reducing the probability of a productive juxtapositioning event with a downstream splice site. We would like to suggest that the C to T transition in SMN2 leads to an overall decrease in the ability of exon 7 to be recognized as exonic. As the affinity of splicing factors to interact with exon 7 decreases, so does the activity of flanking intron removal. In addition, a less defined exon 7 region reduces its ability to prevent exon 7 skipping, and consequently, the competing splice site pairing event between the 5Ј splice site of exon 6 and the 3Ј splice site of exon 8 becomes the preferred splicing pathway. However, we cannot exclude the possibility that the design of our single intron substrates deleted cis-acting elements required for the full manifestation of the splicing effect caused by the C to T transition.
Our experiments indicate that the C to T transition in SMN2 does not drastically alter the kinetics of intron 6 or intron 7 removal. If unprocessed and processed SMN⌬8 and SMN⌬6 transcripts are assumed to have similar half-lives, a comparison of their steady state levels strongly suggests that intron 6 removal occurs much faster than intron 7 removal. These conclusions imply that intron 6 removal precedes processing of the terminal intron. Consistent with this interpretation, we reproducibly detect low steady state levels of the SMN1 intron 7 retaining splicing intermediate (Fig. 3B).
Because all individuals with SMA have retained their SMN2 allele, therapy directed toward increasing SMN2 exon 7 inclusion could provide a very promising tool to lower the clinical severity of SMA. Our comparison of SMN1⌬8 and SMN2⌬8 processing demonstrated a low but measurable difference in the efficiency of intron 6 removal. These data suggested that the 3Јss of exon 7 of SMN1 and SMN2 have similar affinities for components of the splicing machinery and that the C to T transition in SMN2 does not abolish the recognition of the 3Јss of SMN2 exon 7. These observations prompted us to test an antisense oligonucleotide-based approach to modulate the frequency of exon 7 inclusion in SMN2. By blocking the 3Ј splice site of exon 8 with an antisense probe, we hoped to manipulate the apparent competition between the 3Ј splice sites of exon 7 and exon 8. In designing the antisense oligonucleotide, we took advantage of our observation that the 3Ј splice site of SMN2 exon 7 is quite competent to pair with the 5Ј splice site of SMN2 exon 6 and the fact that the SMN translation termination codon resides within exon 7. Of the three tested oligonucleotides, only one, oligo jnct , showed any significant effect in modulating SMN2 exon 7 inclusion. As expected from the experimental design, oligo jnct caused an accumulation of RNA species that retained intron 7. In addition, oligo jnct marginally increased the fraction of fully spliced product. The failure of oligo py and oligo bp to induce any splicing alterations may either be caused by accessibility complications or by their inability to reduce spliceosomal assembly, even when associated with their target sequence. The concentration dependence of oligo jnct illustrates the efficiency of an interaction between oligo jnct and the target sequence. The concentration of oligo jnct required for half-maximal modulation (20 Ϯ 10 nM) favorably compares with other reports utilizing antisense oligonucleotide approaches to alter pre-mRNA-splicing patterns (32)(33)(34)(35). As expected from half-life considerations, the oligo jnct splicing effect diminishes over time.
It is unclear at the present time what minimal levels of SMN expression are required to protect against SMA or to prevent the progression of the neuro-degenerate disease. The overall efficiency of the oligo jnct application is illustrated in Fig. 3B. At the conditions tested, the presence of the oligo jnct leads to a 2.5-fold increase in the concentrations of RNAs containing exon 7. This numerical increase appears modest in magnitude. However, because it has been demonstrated that increased levels of SMN2 copy numbers correlate with decreasing severity of SMA, a 2.5-fold increase in the levels of exon 7 containing RNA may be very significant in raising SMN levels to concentrations that allow some level of protection against disease progression. Although the application of oligo jnct demonstrates the feasibility of the antisense approach, it also serves as a stepping stone for the search of related oligonucleotides with higher efficacy.
A recent study demonstrated that the overexpression of hTra2, a human splicing factor that has been shown to interact with an exon 7 ESE common to both SMN1 and SMN2 pre-mRNAs, is able to efficiently alter the fraction of exon 7 inclusion in SMN2 (28). The strategy used in this approach was to increase the probability of an interaction between hTra2 and the pre-mRNA. Activating exon 7 definition in SMN2 therefore committed the splicing machinery to include exon 7 more frequently into the splicing pattern. Although the results of the exon 7 activation studies were very encouraging, the overexpression of a splicing factor with general functions in pre-mRNA splicing could lead to undesired alterations of other important pre-mRNAs. Here we have tested a splicing modulation approach that by design incorporates a high degree of specificity. The mechanism of oligo jnct action relies on its specific interactions with the 3Ј splice site of exon 8. Association of oligo jnct with its target renders it inaccessible for recognition by the splicing machinery. Oligo jnct therefore competes with the splicing machinery for interactions with the 3Јss of exon 8. Although antisense oligonucleotide applications are highly specific in nature, their efficacy depends largely on target accessibility, as demonstrated by the failure of oligo bp and oligo py to affect SMN2 pre-mRNA splicing. These oligonucleotides in essence served as controls to demonstrate that the processing modulation caused by oligo jnct is a consequence of a specific and productive interaction with the target RNA and not by a nonspecific antisense oligonucleotide effect.
Although the antisense approach appears very promising in principle, the ultimate test of the application will be to evaluate if the presence of oligo jnct or a related oligonucleotide leads to an increase in the levels of full-length SMN protein. Although oligo jnct directed intron 7 retention will generate a mRNA that, when translated, will produce functional SMN protein, it is unknown if SMN intron 7 retention alters the stability of the RNA or its ability to cross the nuclear membrane. Further experiments will address the outcome of these important issues.