MINIREVIEW Alternative splicing: good and bad effects of translationally silent substitutions M. Raponi and D. Baralle Academic Unit of Genetic Medicine, Human Genetics Division, University of Southampton, Southampton General Hospital, UK Introduction Splicing is an important part of a post-transcriptional mechanism where introns are removed and exons are joined together, allowing the resulting mature mRNA to be translated into a specific protein product. This mechanism is supported by the spliceosome machine, which recognizes the well-characterized splicing con- sensus sequences at the exon–intron junctions (donor and acceptor sites) and their proximities (branch points). Other cis-acting elements involved in the deter- mination of the splicing outcome are recognized by trans-acting factors that can either act as splicing silencers or enhancers. Alteration of splicing may occur whenever cis varia- tions alter the recognition of splicing regulatory sequences [1,2]. This could result in altered isoform proportions, activation of a control mechanism such as nonsense-mediated decay, as well as the creation or loss of splicing variants. As this process has a signifi- cant impact on protein abundance and ⁄ or functional- ity, it follows that sequence variants in translationally silent exonic positions that modify splicing are crucial in genetic diagnosis and their role as a possible cause of disease cannot be ignored. Equally important is the role that these silent sequences may have in evolution. For example, many algorithms used to calculate evolu- tionary distances are normalized against the transla- tionally ‘silent’ sequence variants, which until recently were considered evolutionarily neutral. We now know that many so-called neutral substitutions are instead causative, as they produce the skipping of the exon or Keywords minigene; NF1; pre-mRNA; silent; splicing; translation Correspondence D. Baralle, Academic Unit of Genetic Medicine, Human Genetics Division, University of Southampton, Duthie Building (Mailpoint 808), Southampton General Hospital, Tremona Road, Southampton SO16 6YD, UK Fax: +44 2380794346 Tel: +44 2380796162 E-mail: D.Baralle@soton.ac.uk (Received 26 August 2009, revised 4 November 2009, accepted 17 November 2009) doi:10.1111/j.1742-4658.2009.07519.x Nucleotide variations that do not alter the protein-coding sequence have been routinely considered as neutral. In light of the developments we have seen over the last decade or so in the RNA processing and translational field, it would be proper when assessing these variants to ask if this change is neutral, good or bad. This question has been recently partly addressed by genome-wide in silico analysis but significantly fewer cases by laboratory experimental examples. Of particular relevance is the effect these mutations have on the pre-mRNA splicing pattern. In fact, alterations in this process may occur as a consequence of translationally silent mutations leading to the expression of novel splicing isoforms and ⁄ or loss of an existing one. This phenomenon can either generate new substrates for evolution or cause genetic disease when aberrant isoforms altering the essential protein func- tion are produced. In this review we briefly describe the current under- standing in the field and discuss emerging directions in the study of the splicing mechanism by integrating disease-causing splicing mutations and evolutionary changes. Abbreviations K a , ratio of nonsynonymous substitutions; K s , ratio of synonymous substitutions. 836 FEBS Journal 277 (2010) 836–840 ª 2010 The Authors Journal compilation ª 2010 FEBS changes in the alternative splicing (AS) isoforms (that become substrates of both positive and negative natural selection). Two main categories of translationally silent varia- tions can alter splicing: (a) intronic variations – changes outside the coding exonic sequence; (b) synonymous changes – variations that alter the exonic sequence, but not the codon information, for an amino acid. Translationally silent variations that affect splicing and disease Clinical studies identifying aberrant splicing mutations are of great importance for genetic counselling, as a good proportion of unclassified variants are often found to be the cause of inappropriate RNA process- ing (recently reviewed by Baralle et al. [3]). Such vari- ants affecting splicing can be classified as pathogenic mutations or genetic variations causing predisposition to disease. The first category usually has a devastating effect on splicing, with a substantial loss of original protein function or even acquisition of an antagonistic function. An explanation for the second category relies on the fact that a weakly tolerated effect on splicing can be enhanced by additional phenomena such as affected expression of trans-acting factors that regulate splicing [4]. Both intronic and synonymous nucleotide substitu- tions can sufficiently alter splicing and cause genetic dis- ease. A list of silent mutations associated with altered splicing was reported by Cartegni et al. [5]. However, the number of examples reported to date in the litera- ture is not as large as expected and this may be because synonymous variations have previously always been considered neutral and because of an existing bias to search for alterations in the protein functional ⁄ struc- tural properties. Likewise, deep intronic variations that do not affect the canonical splice sites have rarely been taken into consideration or even reported owing to technical difficulties in sequencing entire genes (also until recently an expensive task) and in the underappre- ciated belief that sequence variations far away may affect far away splicing signals. With increased aware- ness we predict that this will change in the future. There are a number of different ways that such variations may affect splicing, including: l disruption of exonic ⁄ intronic splicing enhancer ⁄ silen- cer sequences or creation of exonic ⁄ intronic splicing silencer ⁄ enhancer sequences; l alteration of RNA secondary structure; l creation or disruption of splice sites; In addition to the clinical importance of discovering the aberrant effect of such mutations, they also repre- sent an essential clue and wealthy resource for the study of novel splicing regulatory mechanisms. There is substantial precedence for identifying novel splicing regulatory sequences and splicing factors by molecular analysis of splicing aberrations caused by disease-caus- ing mutations. A good example of this was the mechanistic study of a deep intronic GTAA deletion in the ATM gene that permitted the identification of a novel intronic splicing processing element [6]. Further functional studies have shown how U1 binding to such intronic elements can inactivate the inclusion of aberrant exons [7]. Studies of this kind provide significant insights into the splicing regulation of many genes, but this approach has been poorly undertaken with regards to synonymous changes that affect splicing. Apart from synonymous variations causing disease by creating or affecting the canonical splice sites, most of them still lack experimental approaches directed at identifying the exact mechanism involved. In spite of the well-known lack of reliability of in silico approaches, many of the synonymous changes causing aberrant splicing are thought to alter exonic silencer ⁄ enhancer sequences only on this basis [8,9]. For instance, a silent mutation in exon 7 of the POMGNT1 gene, in a patient with congenital muscular dystrophy, was shown to promote skipping of this exon. Here an extensive in silico analysis predicted the creation of various splic- ing regulatory sequences, including an exon splice silen- cer, as well as a change in secondary structure [8]. Another more characterized example comes from an in silico analysis of two PDHA1 exon 5 silent variants. Each variant determines exon 5 skipping and were pre- dicted to disrupt the splice enhancer SRp55 motif. Using a minigene system, the inefficient exon 5 inclu- sion was corrected by strengthening the intron 5 donor site, suggesting that the putative SRp55 motif compen- sates for the weak donor site [9]. Translationally silent variations that affect splicing and evolution With increasing understanding of the importance of cis regulatory sequences located either in the introns or in the coding sequence for the splicing process, the scien- tific world has become aware that there is a selective constraint for evolution, not only against sequence variations that alter the protein information, but also against variations that are harmful for the splicing pro- cess. As the last category includes even supposedly silent changes, which do not alter the amino acid code, it follows that both intronic and synonymous M. Raponi and D. Baralle Alternative splicing FEBS Journal 277 (2010) 836–840 ª 2010 The Authors Journal compilation ª 2010 FEBS 837 variations are not neutral for evolution. Understanding this has important consequences in the way routine diagnostic testing is approached. In addition, the concept of non-neutrality for synon- ymous variations will force an adjustment of the tradi- tional way of measuring sequence evolution based on the K a ⁄ K s ratio (where K a is the ratio of nonsynony- mous substitutions and K s is the ratio of synonymous substitutions). This method, where the metric is based on the neutrality of K s , has now become relatively inaccurate. Although this approach is still in use, researchers are aware that the K s may not always be neutral, but is potentially affected by at least the splic- ing constraint. As a consequence, a new approach has emerged where the detection of K a ⁄ K s ratio peaks in genes, using a sliding window analysis, is assumed to be an index of selective constraint acting on silent sites. An example of this type of approach is a fascinating conservation analysis comparing BRCA1 orthologues where the aligned coding sequences were used for two independent sliding window analyses (mouse–rat and human–dog) [10]. This analysis showed a strong puri- fying selection at silent sites in a critical region of this gene [10], spanning the 3¢ end of exon 10 and the 5¢ end of exon 11. Purifying selection is the force that drives negative selection to eliminate deleterious muta- tions that would otherwise alter protein function. The possibility that this biased synonymous codon usage reflects the necessity to maintain regulatory sequences associated with splicing regulation was subsequently suggested by the identification of two putative exon splicing enhancers within the critical region [11]. However, this type of approach contains several pit- falls, and it is important to acknowledge the existence of recent bioinformatic studies showing that the synon- ymous substitution rate reduction observed with the sliding window analysis may often be artefactual [12,13]. As a result, there is a strong recommendation that all these studies should be complemented by fur- ther experimental support to demonstrate purifying selection at silent sites in the gene of interest and to demonstrate that such constraint is necessary to main- tain correct splicing of the gene. In fact, although we acknowledge that codon bias is a potential index of splicing constraint, it should not be forgotten that other selective forces may act at silent sites, such as translational accuracy, mRNA binding and mRNA stability (for a review see [14]). In addition, missense variations can also affect splicing. Therefore, a low K a may not only represent negative selection at amino acid substitution, but also splicing constraint. Therefore, the detection of both lower K a and K s in one region is probably an index of splicing constraint rather than the detection of K a ⁄ K s peaks, which may be due to a high K a ratio and not to selec- tive constraint at silent sites. Notwithstanding the controversy surrounding the measurement of purifying selection at silent sites, the fact that synonymous substitutions are under selective constraint because they have to ensure splicing effi- ciency has already been experimentally demonstrated for the CFTR gene [15]. In addition to reporting that  30% of the synonymous substitutions in human CFTR exon 12 significantly reduce its inclusion, this study has also brought new evidence that protein func- tion optimization can be constrained in exchange for the maintenance of proper splicing efficiency. These results were confirmed by an additional evolutionary study that used CFTR exon 12 as a model and showed suboptimal composition at silent sites for splicing effi- ciency in the human exon and proposed a way by which exon loss may represent a substrate for evolu- tion when a combination of synonymous changes induce partial exon skipping [16]. From an evolutionary point of view, however, the most frequently described substrate for natural selec- tion of new splicing variants is exon gain, although, as for exon loss, the precondition that allows a new splic- ing variant to evolve freely is the maintenance of origi- nal protein function. Therefore, nucleotide variations that preserve the coding capacity (such as synonymous or intronic substitutions), but also induce the inclusion of a new exon in only a minor fraction of the mature transcript represent the best candidates in the creation of new splicing substrates for evolution. In this way, the newly generated alternative splicing exon has a bet- ter chance of being tolerated by the cell metabolism and is then free to evolve. Integrating evolutionary and splicing disease-related mechanistic studies – an example The importance of clinical studies is not simply to obtain important knowledge that a mutation has caused a splicing defect, but also to provide a clue for subsequent splicing functional studies, therapeutic approaches and further elucidation of this complex and interesting system. Evolutionary studies represent another important field for the investigation of the ele- ments involved in splicing regulation and the integra- tion of all these approaches will give us the best chance of finally understanding the splicing mechanism itself. An example from our own laboratory is the NF1 splicing mutation c.293–279A>G. This mutation was Alternative splicing M. Raponi and D. Baralle 838 FEBS Journal 277 (2010) 836–840 ª 2010 The Authors Journal compilation ª 2010 FEBS found to activate a pseudoexon and subsequent experi- ments showed a novel mechanism by which the levels of polypyrimidine tract binding proteins limit the damaging pseudoexon inclusion [17]. The discovery of such repression is of great relevance for further gene therapy applications rescuing the patient’s wild-type phenotype. This dependency on trans-acting factor expression levels may also represent an important observation with regards to explaining the variable characteristics of disease, such as why particular organs are affected by a mutation, age of onset, individual susceptibilities, etc. In addition, this discovery also brings insight into the speculation that evolutionary changes may protect against aberrant splicing due to a mutation as well as predispose to disease. Indeed, the same variation is nor- mally present in the canine gene sequence where no splicing alteration is observed. As shown in Fig. 1, we demonstrated the compensatory relevance of some nu- cleotides that differentiate the dog sequence from human. Canine nucleotide substitution in the human minigene for splicing assay harbouring the c.293– 279A>G mutation was enough to mimic the normal pseudoexon exclusion observed in dog and in human normal phenotypes. These data make it clear that com- pensatory changes in dog protect against additional variations that would produce intron exonization. Con- versely, in the human there is a predisposition to muta- tions causing pseudoexon inclusion in NF1 intron 30, which is only partially counteracted by the presence of polypyrimidine tract binding protein binding sites. However, it would be wrong to conclude that evolu- tionary changes happening in human introns should be considered ‘bad’ because of a predisposition to aber- rant splicing, as this may not represent the whole story. In fact, from an evolutionary point of view, the procliv- ity of human intron 30 to exonize may be ‘good’ if looked at as the ability to produce a new substrate for evolution, as previously suggested [18]. Indeed, the loss of the intronic A>G variation from dog versus human, which creates a functional acceptor splice site only in combination with the human cryptic donor site 171 nu- cleotides downstream, has probably allowed the crea- tion of the latter. Overall, this donor site is probably alternatively spliced in humans to produce a minor fraction of transcripts where 67 nucleotides of intron 30 are retained [18]. Such tolerated splicing variants can evolve freely in the pseudointronic sequence and thus acquire a new function. In conclusion, we need to reassess our view of nucle- otide variations that were previously considered neu- tral, particularly with regards to their effect on splicing. A variety of tools are available to us for this purpose and further investigation of these sequence variants will not only further our understanding of the splicing mechanism and improve clinical diagnostic testing, but is also important for understanding gene evolution. References 1 Pagani F & Baralle FE (2004) Genomic variants in exons and introns: identifying the splicing spoilers. Nat Rev Genet 5, 389–396. 2 Cooper TA & Mattox W (1997) The regulation of splice-site selection, and its role in human disease. Am J Hum Genet 61, 259–266. 3 Baralle D, Lucassen A & Buratti E (2009) Missed threads. The impact of pre-mRNA splicing defects on clinical practice. EMBO Rep 10, 810–816. 4 Nissim-Rafinia M & Kerem B (2005) The splicing machinery is a genetic modifier of disease severity. Trends Genet 21, 480–483. 5 Cartegni L, Chew SL & Krainer AR (2002) Listening to silence and understanding nonsense: exonic muta- tions that affect splicing. Nat Rev Genet 3, 285–298. 6 Pagani F, Buratti E, Stuani C, Bendix R, Do ¨ rk T & Baralle FE (2002) A new type of mutation causes a splicing defect in ATM. Nat Genet 30, 426–429. Fig. 1. Human–dog inactivating substitutions completely repress pseudoexon inclusion. The pseudoexon sequence in uppercase is compared with the dog (Canis familiaris) sequence. Asterisks indicate nucleotide matches and dashes indicate sequence gaps. The )279 a>g mutation and the nucleotide substitutions of human versus dog pseudoexon sequence are shown (t1 = G>T; t2 = A>T; g1 = A>G; g2 = T>G). Transfection in Hela cells of hybrid minigenes carrying both single and combinations of substitutions always show pseudoexon exclusion (data not shown). M. Raponi and D. Baralle Alternative splicing FEBS Journal 277 (2010) 836–840 ª 2010 The Authors Journal compilation ª 2010 FEBS 839 7 Lewandowska MA, Stuani C, Parvizpur A, Baralle FE & Pagani F (2005) Functional studies on the ATM intronic splicing processing element. Nucleic Acids Res 33, 4007–4015. 8 Oliveira J, Soares-Silva I, Fokkema I, Gonc¸ alves A, Cabral A, Diogo L, Gala ´ n L, Guimara ˜ es A, Fineza I, den Dunnen JT et al. (2008) Novel synonymous substi- tution in POMGNT1 promotes exon skipping in a patient with congenital muscular dystrophy. J Hum Genet 53, 565–572. 9 Boichard A, Venet L, Naas T, Boutron A, Chevret L, de Baulny HO, De Lonlay P, Legrand A, Nordman P & Brivet M (2008) Two silent substitutions in the PDHA1 gene cause exon 5 skipping by disruption of a putative exonic splicing enhancer. Mol Genet Metab 93, 323–330. 10 Hurst LD & Pa ´ l C (2001) Evidence for purifying selec- tion acting on silent sites in BRCA1. Trends Genet 17(2), 62–65. 11 Orban TI & Olah E (2001) Purifying selection on silent sites – a constraint from splicing regulation? Trends Genet 17, 252–253. 12 Schmid K & Yang Z (2008) The trouble with sliding windows and the selective pressure in BRCA1. PLoS ONE 3, e3746. Erratum PLoS ONE 3(11), e3746. 13 Parmley JL & Hurst LD (2007) How common are intragene windows with KA > KS owing to purifying selection on synonymous mutations? J Mol Evol 64, 646–655. 14 Parmley JL & Hurst LD (2007) How do synonymous mutations affect fitness? Bioessays 29, 515–519. 15 Pagani F, Raponi M & Baralle FE (2005) Synonymous mutations in CFTR exon 12 affect splicing and are not neutral in evolution. Proc Natl Acad Sci USA 102, 6368–6372. 16 Raponi M, Baralle FE & Pagani F (2007) Reduced splicing efficiency induced by synonymous substitutions may generate a substrate for natural selection of new splicing isoforms: the case of CFTR exon 12. Nucleic Acids Res 35, 606–613. 17 Raponi M, Buratti E, Llorian M, Stuani C, Smith CW & Baralle D (2008) Polypyrimidine tract binding protein regulates alternative splicing of an aberrant pseudoexon in NF1. FEBS J 275, 6101–6108. 18 Raponi M, Upadhyaya M & Baralle D (2006) Functional splicing assay shows a pathogenic intronic mutation in neurofibromatosis type 1 (NF1) due to intronic sequence exonization. Hum Mutat 27, 294–295. Alternative splicing M. Raponi and D. Baralle 840 FEBS Journal 277 (2010) 836–840 ª 2010 The Authors Journal compilation ª 2010 FEBS . MINIREVIEW Alternative splicing: good and bad effects of translationally silent substitutions M. Raponi and D. Baralle Academic Unit of Genetic Medicine,. process may occur as a consequence of translationally silent mutations leading to the expression of novel splicing isoforms and ⁄ or loss of an existing one. This