Alternative pre-mRNA splicing is a complex event in which cis
elements located in both exonic and intronic sequences play an important role. Compared to exonic cis
elements, the role of intronic cis
elements in the modulation of alternative splicing is poorly understood (54
). Further, there are no algorithms to accurately predict intronic cis
elements. The task of identifying intronic cis
elements is also complicated by the large size of introns. Here we describe a novel intronic cis
element, ISS-N1 (CCAGCAUUAUGAAAG), which is located in the nonconserved portion of the final intron (intron 7) of human SMN
. This element was found to be an important component of a regulatory network that modulates alternative splicing of SMN
exon 7, which is linked to the pathogenesis of SMA.
We began this study as a follow-up to one of the important findings of our in vivo selection: a weak 5′ ss is the dominant cause of exon 7 exclusion in SMN2
). Consistently, we demonstrated that exonic sequences near the 5′ ss of exon 7 form an inhibitory element called 3′-Cluster. Interestingly, 3′-Cluster was found to overlap sequences that are not conserved among mammals, suggesting an evolutionary constraint in which a novel regulatory network evolved just before the duplication of SMN
. Such an arrangement may even have favored SMN
duplication as SMN expression was already under the tight control of inhibitory cis
elements that rendered the 5′ ss of exon 7 poorly accessible. We reasoned that the nonconserved intronic sequences might have played a key role in creating a poorly accessible 5′ ss. Thus, we analyzed a series of deletion mutations within adjacent intron 7. This led to the identification of ISS-N1, an inhibitory cis
element located immediately downstream of the 5′ ss in intron 7 of human SMN
genes (Fig. ).
The observation that substitutions in the nonconserved portions of intron 7 restore exon 7 inclusion in SMN2
reveals the significance of noncoding sequences that acquired an additional function during evolution to downregulate SMN expression (Fig. ). Our results support a hypothesis that human-specific alternative splicing of exons may be linked to the evolutionarily divergent intronic sequences. Interestingly, the sequence between positions 7 and 30 of intron 7 was found to be more conserved (~67%) than expected (~55%) based on the statistical prevalence of conserved residues between the human and mouse intronic sequences downstream of a constitutively spliced exon (54
). This underscores that the type of mutations, not their number, played a significant role in creating novel cis
elements during evolution. It is not known if the mouse sequence that corresponds to ISS-N1 contains an enhancer element that is recognized by a splicing factor. Interestingly, this region in the mouse contains a UUUAA motif, which has been found to be a winning pentamer within low G+C content intronic sequences downstream of the 5′ ss (62
). Winning pentamers supposedly regulate splicing by an as yet unknown mechanism. Since an ISS-N1-deleted mutant promoted exon 7 inclusion in all cell lines, including mouse cells, we conclude that tissue-specific factors are not involved in ISS-N1-mediated regulation of alternative splicing of exon 7 (Fig. ). These results also demonstrate how the relatively conserved pre-mRNA processing machinery uses nonconserved intronic cis
elements to achieve human-specific regulation.
We used an antisense oligonucleotide-based approach to conclusively demonstrate the inhibitory nature of ISS-N1 (Fig. ). In this approach, the ISS-N1 sequence was blocked by antisense oligonucleotide Anti-N1. As expected, Anti-N1 fully restored exon 7 inclusion in the SMN2 minigene. The positive effect of Anti-N1 was very specific to SMN exon 7, as no off-target effect on alternative splicing was observed in a comprehensive test that used eight other minigenes (Fig. ). In addition, we found that the Anti-N1-mediated stimulatory effect is exclusively dependent upon base pairing with the target (Fig. ). Consistently, a mutant oligonucleotide (Anti-I7-25) that restored base pairing with a mutated form of ISS-N1 (SMN2/I7-25) also promoted exon 7 inclusion in SMN2. Since both deletion of ISS-N1 and substitutions within ISS-N1 promoted exon 7 inclusion in SMN2 (Fig. and ), we conclude that antisense oligonucleotide-mediated stimulation of exon 7 inclusion is due solely to blocking of ISS-N1 and not to a nonspecific effect of an RNA-RNA duplex which is formed by annealing of Anti-N1 to ISS-N1. The fact that antisense oligonucleotides that annealed to sequences downstream of ISS-N1 produced no changes in the splicing pattern (Fig. ) also rules out the possibility of any stimulatory effect merely due to a double-stranded RNA.
The observation that deletion of ISS-N1 compensates for a number of abrogated stimulatory cis
elements suggests that ISS-N1 plays a critical role in alternative splicing of exon 7. ISS-N1 is strategically placed in the vicinity of the 5′ ss, which is poorly defined for several reasons, including poor base pairing with U1 snRNA, a component of U1 snRNP. Consistently, we have previously demonstrated that a single nucleotide substitution (A54G) that increased base pairing between the 5′ ss and U1 snRNA from six to eight nucleotides promoted exon 7 inclusion in SMN2
). Interestingly, we found a striking similarity between the stimulatory effects produced by the A54G mutation and the ISS-N1 deletion. In both cases, exon 7 inclusion was restored despite the abrogation of a number of positive cis
elements. These results support a very tight control by which the 5′ ss of SMN
exon 7 is recognized. One of the possible mechanisms by which ISS-N1 exerts its inhibitory impact may involve rendering the 5′ ss inaccessible for the recruitment of U1 snRNP. Consistently, moving ISS-N1 away from the 5′ ss promoted exon 7 inclusion in SMN2
The splicing pattern of SMN2 mutants in which ISS-N1 was moved 5 or 10 nucleotides away from its original location demonstrated the portability of ISS-N1. These mutations also delineated the 5′ and 3′ boundaries of ISS-N1 and confirmed that intronic sequences immediately upstream and downstream of ISS-N1 are not associated with the inhibitory role of ISS-N1. However, the portability of ISS-N1 was not absolute as moving ISS-N1 20 or 35 nucleotides away from its original location restored exon 7 inclusion in SMN2. The limitation of ISS-N1 portability was also revealed by the insertion mutation in which addition of five U residues between the 5′ ss and ISS-N1 substantially increased exon 7 inclusion in SMN2. This underscores the specific role of an inhibitory context in which ISS-N1 must cooperate with other negative elements to produce the net effect of exon 7 exclusion during pre-mRNA splicing of SMN2. Insertion of U residues may have disrupted this context by creating an enhancer element and/or by providing better accessibility to the 5′ ss through a favorable RNA structure. Consistently, we observed that ISS-N1 was less inhibitory in a heterologous context when it was inserted nine nucleotides away from the 5′ ss of exon 6 of the Casp3 minigene. However, the impact of ISS-N1 on Casp3 exon 6 was still greater than that on exon 7 of SMN1, which differs from SMN2 by a critical C6U mutation. This observation highlights how the impact of an intronic cis element is altered by exonic mutations that are located more than 60 nucleotides away (C6U in SMN2 and mutant SMN1/1U in Fig. ).
Use of antisense oligonucleotide Anti-N1 confirmed the presence of ISS-N1 in the context of the endogenous gene, which is transcribed from a different promoter than the minigene. Further, the endogenous transcript is more than 25-fold larger than the transcript derived from the SMN2
minigene used in our study. There are several ways in which transcription and the transcript size could affect alternative splicing (1
). The fact that blocking of ISS-N1 by low concentrations of Anti-N1 restored exon 7 inclusion in mRNA derived from endogenous SMN2
demonstrated the feasibility of an intron-interacting oligonucleotide to fully overcome the inhibitory effect of an exonic mutation responsible for producing a truncated protein in patient cells. Such a strong antisense effect is possible only when the antisense target is easily accessible. The strong stimulatory effect of Anti-N1 may also suggest that blocking of ISS-N1 lessens the dependency on exon 7 inclusion-inducing factor Tra2-β1, which is downregulated in SMA (17
The majority of SMA cases are caused by the deletion of both SMN1
). Fortunately, most patients retain a copy of SMN2
that could be used to produce the full-length SMN protein. For antisense oligonucleotide-based therapy of SMA, an intronic site is likely to present a better target because an intron-annealed oligonucleotide will not interfere with downstream mRNA metabolism, including exon-junction complex formation, transport, and translation (44
). Consistently, we observed increased levels of SMN protein in SMA cells treated with Anti-N1 at a concentration as low as 5 nM. In contrast, a prior study that used a bifunctional oligonucleotide which annealed to exon 7 and required the presence of splicing factor SF2/ASF produced a stimulatory effect on SMN2
exon 7 inclusion at a concentration of 100 nM (53
). Effects on the protein level for other antisense oligonucleotides that promoted exon 7 inclusion in SMN2
minigenes have not been reported (6
Antisense oligonucleotide-based therapy might have utility in a number of human diseases associated with aberrant splicing (10
). A recent study demonstrated antisense oligonucleotide-mediated correction of aberrant splicing resulting in the restoration of expression of dystrophin in body-wide skeletal muscles of the mouse model of Duchenne muscular dystrophy (33
). Mouse models of SMA are available (20
). However, for therapeutic applications of antisense oligonucleotides in neuromuscular diseases such as SMA, additional modifications may be required for oligonucleotides to cross the blood-brain barrier. Recent advancements in nucleic acid technology and delivery strategies have broadened the scope of antisense oligonucleotide usage for neuronal disorders (14
). Currently, SMA has no therapy. Our discovery of ISS-N1 provides a very specific and accessible intronic target to develop an effective antisense oligonucleotide-based therapy for SMA.
In summary, we have discovered a novel splicing-inhibitory cis element, ISS-N1, which is located in the final intron of human SMN genes. Deletion of ISS-N1 fully corrected SMN2 splicing and compensated for the loss of a number of positive splicing elements within SMN1. This is the first report demonstrating the profound impact of an intronic cis element on alternative splicing of human SMN genes. Blocking of ISS-N1 by an antisense oligonucleotide restored exon 7 inclusion in SMN2 and elevated the levels of SMN protein in SMA patient cells. These findings bring new insights into our understanding of human SMN splicing and provide a unique target for correcting the defective gene associated with a devastating disease of infants and children.