Many disease-causing mutations affecting donor splice site recognition are reported in the literature. One of the more frequently observed nucleotide changes causing aberrant splicing are due to mutations in the donor splice site which lower the strength of base pairing with U1 snRNA (small nuclear RNA). However, recent data have highlighted the possibility of a recognition mechanism for weak donor splice sites that are at least partially U1-independent. This is important as most of the donor splice site prediction programs currently in use are based on the U1 snRNA 5′-splice site base pairing and would not pick this up. We review these mechanisms and how an up-to-date donor splice site mutation repertoire indicates the heterogeneity of the molecular mechanism. We suggest that, in clinical molecular genetics, it is important to evaluate sequence variants for aberrant splicing even in those cases where the variant is not thought to alter the U1 snRNA interaction.
Introns are removed during pre-mRNA maturation through two transesterification reactions where snRNPs (small nuclear ribonucleopreteins) and additional factors associate to form a large complex called a spliceosome . The spliceosomal snRNPs U1, U2, U4, U5 and U6 are involved in the majority of the pre-mRNA splicing.
In the formation of the pre-spliceosome, the 5′-end of the U1 RNA base pairs with the 5′ junction (donor site or 5′ splice site), whereas other spliceosomal components U4/U5/U6 interact with the donor site after disruption of the U1 snRNA (small nuclear RNA) base pairing.
In particular, the U1 snRNA 5′-end is perfectly complementary to the donor site consensus sequence CAG/GTRAGT, while the U5 snRNA and the U6 snRNA can only bind to positions −1 and −2, and +2, +5 and +6 respectively . Since the donor site does not always conform to the consensus sequence, but can instead have a degenerate pattern feature, it is understandable that many other additional elements such as splicing silencer and enhancer complexes participate in splice site selection.
Functional analysis of mutations affecting the donor splice site consensus sequence is important to elucidate the poorly understood 5′ splice site recognition mechanism. In the last 10 years, several studies based on in vitro splicing, in silico prediction and U1 complementarity experiments have shown the possibility of alternative donor site recognition mechanisms where factors other than the simple U1 snRNA base pairing determine 5′ splice site usage.
Disease-causing mutations affecting the donor splice site
Splicing mutations at the donor site provide a significant contribution to human disease. Functional studies based on these mutations have been useful for a better understanding of the 5′ splice site recognition mechanism.
The effect of nucleotide substitutions affecting the 5′ splice site is not always predictable: in particular, nucleotide variations at the same donor site position can be splicing mutations in one context but not in others , with splice site sequences that deviate from the consensus not necessarily producing significantly lower amounts of spliced mRNA .
A recent study extrapolated from the variety of 5′ splice site prediction methods the models that best predict the localization of cryptic or de novo 5′ splice sites that were activated in vivo . They have shown that cryptic 5′ splice sites activated in human disease by disruption of the consensus sequence of the authentic sites were best predicted by models that consider nucleotide dependencies at the 5′ splice site, rather than models based on weight-matrix. In keeping with this, a second recent study showed how disruption of conserved patterns between nucleotides at different positions within the 5′ splice site is often a frequent cause of genetic disease .
U1-independent splicing was initially suggested a decade ago when several pre-mRNAs were successfully spliced from U1-depleted nuclear extract enriched in SR (serine/arginine-rich) proteins . Further experiments demonstrated that the addition of the SR protein SC35 alone was enough to splice a class of specific RNA substrates and that the U6 snRNA interaction can be rate-limiting for U1-independent splicing . Subsequently, a class of pre-mRNAs that can be spliced in U1-depleted extract without requiring additional amounts of SR proteins was found . Such pre-mRNAs included chimaeric constructs and a natural pre-mRNA from the Drosophila ftz (fushi tarazu) gene. In a recent study, we have found a further example of a naturally occurring pre-mRNA that can be spliced in U1-depleted HeLa nuclear extracts in NF1 (nuclear factor 1) (M. Raponi, E. Buratti, E. Dassie, M. Upadhyaya and D. Baralle, unpublished work). In addition, we have demonstrated the lack of strong direct and indirect interactions with U1 and the 5′ splice site under study. For this particular case, aberrant splicing due to mutations at position +5 at the donor site can not be corrected in U1 complementarity experiments, suggesting the loss of U6 binding at the 5′ splice site as the cause of the defect in a U1-independent/U6-dependent recognition mechanism.
U1 complementarity failure
Aberrant splicing mutations at the 5′ splice site can often be suppressed by co-expression of Ul snRNAs with appropriate compensatory changes, but several exceptions have been reported. For example, it has been shown that a change at position +5 in the intron 14 donor site of the human TCIRG1 gene is not suppressed by a compensatory change in Ul snRNA that instead induces an increase in aberrant splicing .
Other examples where compensatory nucleotide changes between U1 and a 5′ splice site failed to suppress the aberrant cleavage were reported in yeast [11,12], where it was shown that changes in U1 snRNA that compensate for a mutation at position +5 of the donor site can nevertheless stimulate aberrant cleavage outside the paired region. This observation introduced the idea that other factors are involved in the fidelity of the 5′ splice site cleavage. Finally, 8 years later, Hwang and Cohen  showed that U1 snRNA base pairing did not define the 5′ splice site, but instead promoted the nearby 5′ splice site choice by U6 in mammalian cells.
U6 selection of the donor site
Base-pairing interaction of the 5′ splice site with U6 snRNA is known to play a role in 5′ splice site selection [2,13]. Compensatory changes in the U6 snRNA have been shown to suppress mutations at donor site positions +5 and +6, and an extended interaction between the donor site and U6 snRNA has been suggested.
The fact that both U1 and U6 co-operate in the donor site definition, with U1 serving to bring U6 into the vicinity of the site, it raises the question of how U6, on the basis of minimal complementarity, can be involved in the process of U1-independent donor site selection. Studies that support this possibility include those in mammalian extracts, where the U4/U6·U5 snRNP associates with the 5′ splice site independently of U1 snRNP . In addition, using a Saccharomyces cerevisiae in vitro trans-splicing system has shown that, when the 5′ end of U1 snRNA is deleted, the 5′ splice site is still recognized by U4/U6, suggesting that 5′ splice site recognition is not dependent on the 5′ end of U1 , although splicing is inhibited in this system. It is possible that the previously described cis-acting elements distinct from the donor site involved in this unusual recognition mechanism are binding sites for SR proteins or simply determinants of specific secondary structures.
A further possibility arises from the work of Lund and Kjems  who have shown that stabilizing U1 snRNA binding to the 5′ splice site increases the competitive strength of a splice site. From this perspective, we can speculate that the scant number of U1-independent 5′ splice sites is due to the fact that they may not prevail in competition. However, a picture emerges where distinct or simpler interactions may be required to ensure recognition of U1-independent 5′ splice sites, rather than the complicated network required to discriminate between competing and stronger 5′ splice sites.
Different recognition mechanisms are possible where U1 snRNA 5′ splice site base pairing is not absolutely required. However, we have no clear knowledge of which mechanism allows recognition of particular donor splice sites in the absence of U1 interaction.
Other factors that are not related to U1 base pairing in the recognition of the donor sites may be involved and should be characterized. When U1 determination of the site of splicing is not necessary, for example if other competitive donor sites are not present in the vicinity, the recognition of a U1-independent 5′ splice site may simply proceed through a U6-dependent mechanism.
It is clear that a complete mechanistic understanding of 5′ splice site selection is lacking. Clinical genetic analysis can give clues as to what controls U1-independent recognition mechanisms through the study of pathological sequence variants found at the 5′ splice site, in turn these alternative pathways of 5′ splice site recognition need to be properly elucidated, not only for safer molecular diagnosis in patients, but also for the development of techniques and therapeutic tools for the correction of related splicing mutations.
RNA UK 2008: Independent Meeting held at The Burnside Hotel, Bowness on Windermere, Cumbria, U.K., 18–20 January 2008. Organized and Edited by David Elliot (Newcastle, U.K.), Sarah Newbury (Sussex, U.K.) and Alison Tyson-Capper (Newcastle, U.K.).