Báo cáo khoa học: Low U1 snRNP dependence at the NF1 exon 29 donor splice site doc

Low U1 snRNP dependence at the NF1 exon 29 donor splice site Michela Raponi 1,2 , Emanuele Buratti 3 , Elisa Dassie 1 , Meena Upadhyaya 4 and Diana Baralle 1,2 1 Department of Pathology, University of Cambridge, UK 2 Human Genetics Division, University of Southampton, UK 3 Department of Molecular Pathology, ICGEB, Trieste, Italy 4 Institute of Medical Genetics, Cardiff University, UK The mechanisms involved in the inclusion of an exon in the mature mRNA molecule are complex, with an ever-increasing number of sequence and protein ele- ments being involved in its definition [1,2]. At the most basic level, canonical splice sites (SSs) are present at the 5¢-ends and 3¢-ends of the exons, and other classic splicing signals, the polypyrimidine tract and the branch point, are present upstream of the 3¢SS [3,4]. To increase the overall fidelity of the splicing reaction, additional enhancer and silencer elements are present in the exons [exon splicing enhancers (ESEs); exon splicing silencers (ESS)] and/or introns [intron splicing enhancers (ISEs); intron splicing silencers (ISS)], allow- ing the correct SSs to be distinguished from the many cryptic SSs that have very similar signal sequences [5–7]. In addition, it is now clear that many other factors, such as RNA secondary structure, tran- scription rates, genomic context, and external stimuli, can profoundly affect the working of the splicing machinery [8–11]. Not surprisingly, therefore, many disease-causing splicing mutations described in the lit- erature produce changes in these regulatory sequences and processes [12,13]. It is the study of these new cases that often provides novel insights into the basic mecha- nisms of SS recognition by the spliceosome [14–16]. Neurofibromatosis type 1 (NF1) is a dominantly inherited multisystem disorder with complete pene- trance by age 5 years. Mutations in the NF1 gene have been found to span the entire coding sequence. In particular, a high incidence of splicing aberrations Keywords donor; NF1; splicing; U1 snRNP Correspondence D. Baralle, Human Genetics Division, University of Southampton, Duthie Building (Mailpoint 808), Southampton General Hospital, Tremona Road, Southampton SO16 6YD, UK Fax: +44 0 238 079 4346 Tel: +44 0 238 079 6162 E-mail: D.Baralle@soton.ac.uk (Received 25 September 2008, revised 26 January 2009, accepted 30 January 2009) doi:10.1111/j.1742-4658.2009.06941.x Many disease-causing splicing mutations described in the literature produce changes in splice sites (SS) or in exon-regulatory sequences. The delineation of these splice aberrations can provide important insights into novel regula- tion mechanisms. In this study, we evaluated the effect of patient variations in neurofibromatosis type 1 (NF1) exon 29 and its 5¢SS surrounding area on its splicing process. Only two of all nonsense, missense, synonymous and intronic variations analyzed in this study clearly altered exon 29 inclu- sion/exclusion levels. In particular, the intronic mutation +5g>a had the strongest effect, resulting in total exon exclusion. This finding prompted us to evaluate the exon 29 5¢SS in relation to its ability to bind U1 snRNP. This was performed by direct analysis of the ability of U1 to bind to wild- type and mutant donor sites, by engineering an in vitro splicing system to directly evaluate the functional importance of U1 snRNA base pairing with the exon 29 donor site, and by coexpression of mutant U1 snRNP mole- cules to try to rescue exon 29 inclusion in vivo. The results revealed a low dependency on the presence of U1 snRNP, and suggest that exon 29 donor site definition may depend on alternative mechanisms of 5¢SS recognition. Abbreviations ATM, ataxia telangiectasia mutated gene; EDB, extra type III homology B or extra domain B; EMSA, electromobility shift analysis; ESE, exon splicing enhancer; ESS, exon splicing silencer; ISE, intron splicing enhancer; ISS, intron splicing silencer; NF1, neurofibromatosis type 1; SNP, single-nucleotide polymorphism; SS, splice site. 2060 FEBS Journal 276 (2009) 2060–2073 ª 2009 The Authors Journal compilation ª 2009 FEBS has been reported in this gene, making it an ideal model with which to study how this process is involved in disease. Examples have included muta- tions interrupting U1 snRNA binding [17,18] and ESEs [19,20]. Given the high number of NF1 patients found to carry single-nucleotide substitutions in NF1 exon 29 and its 5¢SS, we developed a functional assay to distin- guish which genomic variants cause aberrant splicing. Only two of the substitutions analyzed in this study caused exon skipping in the range of 30–100% of the total processed mRNA. Interestingly, the intronic mutation +5g>a, at the 5¢ SS of exon 29, had the strongest effect, resulting in total exon exclusion. This finding prompted us to evaluate carefully the impor- tance of 5¢SS definition in NF1 exon 29. In general, the concept of 5¢SS definition itself has proven to be surprisingly sensitive to sequence context, and several papers have analyzed this issue in detail in the recent past [21–27]. Indeed, although the importance of U1 snRNA interaction with the 5¢SS is considered to be one of the fundamental steps in SS definition [28,29], the binding of this molecule through a straight- forward RNA–RNA interaction has sometimes been found to be dispensable for proper splicing [30,31], and other accompanying factors may play an impor- tant role therein [21,32,33]. From a clinical point of view, this complexity is well reflected in the fact that identical genomic variations around 5¢SSs can be innocuous in one exon or cause disease-related skip- ping in another [7]. For this reason, the susceptibility of the NF1 exon 29 5¢SS to aberrant splicing was fur- ther analyzed at the molecular level in order to delin- eate the regulatory elements involved. Results NF1 exon/intron 29 mutations and evolutionarily conserved splicing regulatory elements The sequence variations identified in this study in NF1 exon/intron 29 of patients with clinically defined NF1 are shown in Fig. 1. Initially, in silico analysis of splice regulatory sites was used to analyze the impact Fig. 1. NF1 exon/intron 29 mutations and evolutionary conservation. NF1 exon 29 (upper-case) and partial intron 29 (lower-case) sequences, showing the location of the nucleotide changes. The amino acid changes are in parentheses. Nucleotide positions that are not conserved between humans and other placental mammals are underlined. Intron genomic sequence alignments are shown at the bottom. The 5¢SS conserved sequences are boxed. M. Raponi et al. Defective interaction between U1 snRNP and 5¢SS FEBS Journal 276 (2009) 2060–2073 ª 2009 The Authors Journal compilation ª 2009 FEBS 2061 of each exonic variation on splicing elements. The results where a significant variation was predicted either to create or interrupt a putative splice regula- tory site are reported in Table 1. In addition, evolu- tionary conservation analysis was used to improve the specificity of this ESE prediction tool, as suggested recently by Pettigrew et al. [34]. In this analysis, the potential evolutionary conservation of ESEs was examined using the ucsc genome browser in all avail- able placental mammals (chimp, mouse, rat, horse, armadillo, cat, cow, and dog). This revealed that none of the predicted ESE sequences interrupted by our nine exonic variations were evolutionarily conserved (compare Fig. 1 and Table 1). However, several exon- ic variations were predicted to create new silencer sites (Table 1). Comparative genomic analysis among placental mammalian species was also performed to evaluate the conservation of the first 23 nucleotides of the intron 29 sequence (Fig. 1). Whereas the downstream sequence (nucleotides +6 to +23) presents high variability among the different species, the 5¢SS consensus sequence is completely conserved, suggesting high susceptibility to mutations of this region. Table 1. Splicing Sequence Finder analysis for the nine exonic variations. New site refers to putative regulatory sequences created by the variation. Broken site refers to putative regulatory sequences disrupted by the variation. Proteins predicted to bind the putative splicing regu- latory sequences and the matrix sources are given. Variation Matrix Protein Sequence Effect c.5224C>T (Q1742X) ESE finder matrices SRp55 TAAGTA New site RESCUE ESE examers CAAGTA Site broken hnRNP motifs hnRNPA1 TAAGTA New site Silencer motifs from Sironi et al. CTGTCTAA New site c.5234C>G (S1745X) ESE finder matrices SRp40 CTTCAGC Site broken SF2/ASF CAGCAGA Site broken RESCUE ESE examers AACTTC Site broken ACTTCA Site broken CTTCAG Site broken PESE octamers from Zhang & Chasing ACTTCAGC Site broken c.5242C>T (R1748X) RESCUE ESE examers AGTGAA New site PESE octamers from Zhang & Chasing CGAACA Site broken GCGAACAA New site c.5264C>G (S1755X) ESE finder matrices SRp40 GGGCAATG New site SRp55 TGAGTC New site RESCUE ESE examers CAATCA Site broken AATCAG Site broken Silencer motifs from Sironi et al. AATGAGTC New site c.5290G>T (A1764S) ESE finder matrices SRp55 TGCTTC Site broken Silencer motifs from Sironi et al. TTCTTCGG New site c.5388T>A (C1796X) ESE finder matrices SRp40 TGAGAAG New site RESCUE ESE examers TGAGAA New site PESE octamers from Zhang & Chasing TGTGAAGC New site ESE motifs Tra2 GAGAAG New site c.5425C>A (R1809S) ESE finder matrices SF2/ASF CGGACCA New site SF2/ASF CAGCTGG New site PESE octamers from Zhang & Chasing CGCTGGG Site broken CGGACCCG Site broken GGACCCGC New site ACCCGCTG New site CCCGCTGG New site GACCAG New site ESE motifs Tra2 GAGAAG New site c.5426G>T (R1809L) ESE finder matrices SF2/ASF CGCTGGG Site broken SRp40 GGACCCTC New site PESE octamers from Zhang & Chasing CGGACCCG Site broken c.5427C>T (R1809R) ESE finder matrices SF2/ASF GACCCGT New site SRp40 CCGCTGG Site broken SF2/ASF CGCTGGG Site broken Defective interaction between U1 snRNP and 5¢SS M. Raponi et al. 2062 FEBS Journal 276 (2009) 2060–2073 ª 2009 The Authors Journal compilation ª 2009 FEBS Splicing analysis of NF1 exon\intron 29 variations by the minigene approach On the basis of these considerations, we conducted in vivo splicing assays using a genomic portion of NF1 exon 29 and part of the flanking introns 28 and 29. A schematic diagram of the hybrid minigene (pTB NF1– 29) used in transient transfection splicing assays is shown in Fig. 2A. The transfection results for the pTB NF1–29 hybrid minigenes carrying all exonic and intronic nucleotide substitutions described in Fig. 1 are shown in Fig. 2B. Transfection of the normal wild-type pTB NF1–29 minigene in HeLa cells followed by RT-PCR amplifi- cation generated mostly transcripts of 580 bp contain- ing exon 29, and a smaller proportion of transcripts of 239 bp lacking this exon. This leaky splicing of NF1 exon 29 in the minigene context was analogous to that already reported in leukocytes from normal individuals [35]. To further confirm the reliability of our system, transfection of the minigene carrying the +19t>a sub- stitution, which is considered to be a common single- nucleotide polymorphism (SNP) [36], was undertaken, and resulted in the same splicing outcome as for the wild-type minigene. Moreover, transfection of the minigene carrying the R1748X change gave the same result as trasnfection of the wild-type minigene and as that obtained when analyzing the RNA sample from leukocytes available from the patient (data not shown). With regard to the variations found in patients, only a minority of the substitutions observed in patients resulted in low to moderate modifications of splicing levels: R1809S and S1755X led to the highest amount of exon 29 inclusion, whereas the Q1742X minigene resulted in approximately 30% exon skipping with respect to wild-type levels (Fig. 2B). Interestingly, the strongest effect was observed for the minigene carrying the intron 29 substitution +5g>a, which resulted in 100% exon skipping. It should also be noted that simi- lar results were obtained when the minigenes were transfected into COS cells (data not shown). We then proceeded to characterize in more detail the impor- tance of the NF1 exon 29 donor site in regulating recognition of this exon. Defective NF1 intron 29 5¢SS recognition First, natural human 5¢SS sequences with similar degrees of mismatching with the mutated donor site were selected from the most up-to-date Homo Sapiens Splice Sites Dataset, released by Sahashi et al. [37]. Interestingly, only 3.62% of the 1492 natural 5¢SSs carrying +4g+5a (and thus containing the same in- tronic mismatches with U1 snRNA as the mutated donor site) present a cytosine as the mutated donor site in position –2, whereas an adenosine is over-repre- sented in this position (Fig. 3A). The adenosine frequency is 89.14%, as opposed to 63.5% of the over- all 189 249 human 5¢SSs [37]. In addition, the 542 functional 5¢SSs with both +5a and )2c rarely have a guanosine in position +4, as for the mutated donor site, whereas 64.31% of those sites have an adenosine in this position (Fig. 3A). As the significant differences Fig. 2. Hybrid minigene transient transfection assay. (A) Schematic representation of the hybrid minigene (pTB NF1–29) used in transient transfection splicing assays. Minigene exons and introns are indicated as boxes and lines, respectively. The exon 29 sequence (white box) was tested for splicing efficiency, using specific primers (arrows). Dotted lines show the two NF1 exon 29 alternative splicing possibilities. (B) RT-PCR products from transfection experiments. The minigenes were transfected into HeLa cells, and RT-PCR analysis was performed on a 2% agarose gel. RNA splicing variants corresponding to exon 29 inclusion (+) and exclusion ()) are shown. The mean levels of exon 29 inclusion, together with standard deviations (SDs) from three different experiments, are reported below. M. Raponi et al. Defective interaction between U1 snRNP and 5¢SS FEBS Journal 276 (2009) 2060–2073 ª 2009 The Authors Journal compilation ª 2009 FEBS 2063 between the group of functional 5¢SSs and the mutated SS described here were in positions )2 and +4, we decided to study the splicing effect of artificial muta- tions in the NF1 exon 29 5¢SS to determine whether or not a cooperative effect between those positions and the +5 position exists. Therefore, )2c was replaced by a purine both in the wild-type and in the +5g>a pTB NF1–29 minigene. The )2c>a substitution did not have any effect on splicing, either in the wild-type or the +5g>a pTB NF1–29 minigenes (Fig. 3B). Nevertheless, according to several 5¢SS prediction programs (Fig. 3C), the )2c>a substitution should have been quite sufficient to restore the U1 complementarity of the mutated +5g>a exon 29 donor site, thus allowing 5¢SS usage, as is the case in the functional 5¢SS donor sequences described in Fig. 3A. Moreover, transfection of the minigene carrying the artificial )2c>g substitu- tion showed the same effect as the natural mutation +5g>a, resulting in 100% exon 29 exclusion, even though the available 5¢SS prediction programs were not able to predict the dramatic effect of this substitu- tion (Fig. 3B,C). These results suggested that other factors may play a role in determining exon 29 recog- nition besides U1 snRNA complementarity. Finally, the only substitution in the mutant +5g>a 5¢SS that was predicted to raise the SS score to 0.98 according to the nnsplice program is represented by +4g>a (Fig. 3C). We therefore replaced the +4g with adeno- sine, both in the wild-type and in the +5g>a minigenes. This time, in keeping with predictions, although the +4g>a substitution alone had no effect in improving wild-type exon 29 inclusion levels, the double mutation +4g>a\+5g>a partially restored Fig. 3. NF1 intron 29 5¢SS recognition. (A) Nucleotide frequency (%) at human natural (true) 5¢SSs with similar degrees of mismatching with U1 snRNA as the +5g>a mutated donor site. The 5¢SS consensus sequence and the intron 29 +5g>a 5¢SS sequences are shown. Nucleo- tide positions in the 5¢SSs are numbered. (B) Point mutations were introduced into the IVS29 5¢SS, and the resulting minigene variants were transfected into HeLa cells. The splicing pattern of these mutants was analyzed by RT-PCR on a 2% agarose gel, and the mean levels of exon 29 inclusion, together with standard deviations from three different experiments, are shown below. (+) and ()) indicate exon 29 inclu- sion and exclusion, respectively. The additional band below the (+) band is due to heterodimer formation. (C) Table showing the 5¢SS score calculated by the maximum entropy method (ME), the Neural Network website (NN), the Shapiro and Senapathy matrix (S&S), and hydrogen bond base pair formation for the wild-type and mutant IVS29 donor sites. Defective interaction between U1 snRNP and 5¢SS M. Raponi et al. 2064 FEBS Journal 276 (2009) 2060–2073 ª 2009 The Authors Journal compilation ª 2009 FEBS exon 29 inclusion levels (Fig. 3B). In addition, the +4g>a and +7t>a substitutions together rescued exon inclusion when inserted in the )2c>g minigene, a result that is generally in keeping with program predictions (Fig. 3C). These results are in support of U1 snRNP dependency when the 5¢SS sequence is sufficiently altered. However, reproduction of similar complemen- tarity using mutant U1 snRNPs did not result in exon inclusion in the original IVS29 context (see below). Interaction between U1 snRNP and IVS29 donor sites We then decided to investigate the ability of all these donor sites to bind U1 snRNA/U1 snRNP directly, under splicing conditions. In these assays, we used a sequence from the ATM gene as a positive control that was previously shown to bind U1 snRNP with high affinity [15]. Synthetic oligonucleotides of all these sequences (Fig. 4A) were end-labeled with a- 32 P and UV-cross- linked with a HeLa nuclear extract that was treated with RNAse H in the presence of a specific antisense oligonucleotide in order to inactivate the endogenous U1 snRNP molecules (Fig. 4B, minus lanes). A mock- inactivated sample obtained using a random oligonu- cleotide was also used as a control (Fig. 4B, plus lanes). Figure 4B shows that, under splicing condi- tions, all IVS29 donor sites bound U1 snRNA with substantially reduced efficiency as compared to the ATM wild-type sequence (compare the intensities between lane 2 and lanes 4, 6, 8, and 10 in Fig. 4B). Although this experiment ruled out a strong and direct interaction between U1 snRNA and the IVS29 donor site sequences, there also remained the possibility that U1 snRNP might still be recruited to the IVS29 donor site by other indirect interactions. Moreover, the much stronger signal for the ATM SS than for the NF1 SS could have simply reflected the higher number of com- plementary nucleotides in this oligonucleotide with respect to the IVS29 oligonucleotides. In order to address these questions, we then per- formed an electromobility shift analysis (EMSA) of the RNAÆprotein complexes assembled on the ATM oligo- nucleotide and each of the IVS29 oligonucleotides, using nuclear extract depleted or mock-depleted of U1 snRNA. Figure 4C shows, as expected, that in the Fig. 4. UV crosslinking and EMSA analysis. (A) Sequences of the RNA oligonucleotides used in these experiments: ATM wild-type (WT), IVS29 WT, IVS29 +5A, IVS29 +4A+5A, and IVS29 ) 2A+5A. (B) The reactivity of each 5¢-end-labeled single-stranded oligonucleo- tide with U1 snRNA following RNAÆRNA UV crosslinking analysis is shown. For each RNA oligonucleotide, this reaction was performed using U1 snRNP-depleted HeLa nuclear extract ()) and mock- depleted extract (+). The RNAÆRNA crosslinked complexes were separated by EMSA, and the region where the U1 snRNAÆRNA complexes migrate is indicated by an open box. (C) EMSA analysis of proteinÆRNA complexes that are formed on each of these oligo- nucleotides in the presence of U1 snRNP-depleted HeLa nuclear extract () ) and mock-depleted extract (+). The region involved in the U1 snRNPÆRNA interaction is indicated by an open box. M. Raponi et al. Defective interaction between U1 snRNP and 5¢SS FEBS Journal 276 (2009) 2060–2073 ª 2009 The Authors Journal compilation ª 2009 FEBS 2065 case of the wild-type ATM oligonucleotide, only one major complex was formed in the presence of the mock-U1-depleted nuclear extract and that formation of this complex was completely abolished after RNa- se H treatment (compare lanes 1 and 2 in Fig. 4C). In the case of the IVS29 oligonucleotides, a series of complexes were also assembled on each oligonucleo- tide, most of them occurring outside the region defined by the ATM–U1 snRNA complex (see open box). Moreover, no major effect on any of these complexes could be observed following U1 snRNP removal from the nuclear extract (Fig. 4C; compare lanes 3 and 4, 5 and 6, 7 and 8, and 9 and 10). In vitro analysis of the functional relationship between U1 snRNP and the NF1 IVS29 donor site In order to functionally test the dependency of the NF1 exon 29 5¢SS and its ability to base-pair with U1 snRNA, we next set up an in vitro splicing system based on the PY7 tropomyosin plasmid (see Fig. 5A for a schematic diagram). Initially, a preliminary in vitro splicing analysis was performed to confirm that, even in these very different conditions, the +5A and +4A+5A mutations had a similar effect to that observed for the IVS29 mutants in the pTB context. The results of this analysis confirmed that these in vitro splicing substrates were spliced as observed in the minigene splicing system (Fig. 5B). An in vitro splicing reaction was then performed in the presence of 50 ng of an antisense oligonucleotide (U1AS) aimed at blocking U1 snRNA interactions. In this respect, it should be noted that the simple incuba- tion of HeLa nuclear extract with an antisense oligo- nucleotide against the 5¢-end U1 snRNA has previously been reported to activate the endogenous RNase H activity, with consequent inactivation of this factor [38]. In our case, this was confirmed by incuba- tion of approximately 150 lg of commercial HeLa nuclear extract with increasing concentrations of U1AS oligonucleotide, followed by phenol/chloroform extraction of total RNAs and reverse primer extension analysis using a labeled oligonucleotide localized in U1 loop 2 [39]. The results of this analysis showed that, after 30 min, the quantity of intact U1 snRNA was reduced by more than 90% in the presence of 12.5 ng of U1AS oligonucleotide (Fig. 5C, lane 2), and that it became undetectable after addition of 25 ng of oligonucleotide (Fig. 5C, lane 3). The U1-depleted mix was then used to test whether the absence of U1 snRNP could affect the in vitro recognition of the IVS29 donor site. As shown in Fig. 5D, addition of Fig. 5. In vitro splicing of NF1 5¢SS in the absence of U1 snRNP. (A) Schematic diagram of the hybrid NF1 exon 29 construct used in the in vitro splicing assays. The NF1 exon/intron sequence present in these construct is shown in full. The open box represents tropomyosin exon 3, and the black line the remaining 94 nucleotide tropomyosin intron. The position of the T7 promoter used to transcribe this RNA is also indicated. (B) In vitro splicing efficiency after 2 h of each RNA template carrying the IVS29 wild-type (WT) donor sequences and the IVS29 +5A and IVS29 +4A+5A mutants. A schematic diagram of the different RNA species is given on the right. (C) Reverse primer exten- sion analysis of the effects of endogenous RNase H activity on U1 snRNP in the presence of an antisense oligonucleotide (U1AS) targeted against the 5¢-end of U1 snRNA. Lane 1: no oligomer. Lanes 2–5: increasing oligomer concentrations. The positions of wild-type uncleaved U1 snRNA, U1 (+1), and of the cleaved RNase H products, U1 (+7/+11), are indicated. (D) RT-PCR analysis of an in vitro splicing assay using the IVS29 wild-type construct in the presence of 50 ng of U1AS oligonucleotide (+U1AS) or of a mock oligonucleotide (+mock). Defective interaction between U1 snRNP and 5¢SS M. Raponi et al. 2066 FEBS Journal 276 (2009) 2060–2073 ª 2009 The Authors Journal compilation ª 2009 FEBS the U1AS oligonucleotide had no effect on the splicing efficiency of the exon 29 donor site with respect to the addition of a mock oligonucleotide (Fig. 5D; compare lanes 1 and 2). It should be noted that the ability of U1AS addition to inhibit splicing efficiency was previ- ously confirmed using other two intron/three exon in vitro substrates (data not shown). Rescue of U1 complementarity in the +4 and +5 positions of IVS29 using mutant U1 snRNA sequences One potential criticism of these in vitro binding and splicing experiments is that the RNA substrates used in these assays lacked regions that are necessary to sta- bilize U1 snRNP binding to the donor site. To investi- gate this possibility, we determined whether rescue of exon inclusion from the +5g>a mutant could be obtained following in vivo complementation experi- ments with mutant U1 snRNAs designed to function- ally complement different mutated positions of this 5¢SS. In this respect, it should be noted that in other NF1 settings, such as an NF1 exon 3 donor site carry- ing a +5g>c disease-causing mutation, this strategy was particularly efficient in rescuing donor site muta- tions [17]. Therefore, three U1 snRNA mutants were engineered to achieve this in the IVS29 donor site con- text. The first was designed to complement specifically the +5a position (U1+5A), the second was designed to bind specifically at both the +4g and +5a (U1+4G+5A) positions, and the third was designed to bind with full complementarity to the +5g>a intron 29 5¢SS (U1–2C+4G+5A+7U). Complementation experiments were thus attempted by cotransfecting the +5g>a IVS29 splice site with U1+5A, U1+4G+5A and U1–2C+4G+5A+7U mutant U1 snRNPs and the +4g>a+5g>a IVS29 mutant with just the U1+5A mutant U1 snRNP (see Fig. 6A for a schematic diagram of the predicted interactions between these mutant donor sites and the wild-type and engineered U1 snRNA molecules). The results in Fig. 6B show that U1+4G was incapable of A B C Fig. 6. U1 complementarity experiments. (A) Base-pairing between IVS29 +5g>a, IVS29 +4g>a+5g>a and IVS29 )2c>g donor sites and either U1+5A, U1+4G+5A, U1–2C+4G+5A+7U, or U1+4G. Ver- tical lines indicate Watson–Crick base pairs, and circles indicate wobble base pairs. (B) These U1 variants were cotransfected into HeLa cells together with pTB NF1–29 minigenes carrying different 5¢SS changes, and the splicing pattern was analyzed by RT-PCR on a 2% agarose gel with respect to inclusion (+) and exclusion ())of this exon. The mean levels of exon 29 inclusion, together with standard deviations from three different experiments, are shown below. (C) Amplification of each mutant U1 cDNA, to check for comparable and correct expression in the transfected cells. ()) indi- cates cotransfection of the minigene with the U1empty vector. Minus RT controls confirmed that plasmid DNA was not being amplified (data not shown). M. Raponi et al. Defective interaction between U1 snRNP and 5¢SS FEBS Journal 276 (2009) 2060–2073 ª 2009 The Authors Journal compilation ª 2009 FEBS 2067 rescuing the IVS29 )2c>g minigene, and that both U1+5A and U1+4G+5A were incapable of rescuing the IVS29 +5g>a minigene, whereas U1– 2C+4G+5A+7U had a very weak effect. In addition, U1+5A had only a weak positive effect with regard to promoting exon 29 inclusion in the IVS29 +4g>a+5g>a minigene (Fig. 6B). These results indi- cate a situation where, even in vivo, the expression of U1 mutants carrying more extensive U1 snRNA com- plementarity for the IVS29 donor site mutants than U1–IVS29 donor site wild-type complementarity are very inefficient at promoting donor site recognition. As a control, correct and comparable expression levels of the U1+5A, U1+4G+5A, U1)2C+4G+5A+7U and U1+4G constructs were confirmed by amplifying their cDNA with specific oligonucleotides following their transfection into HeLa cells (Fig. 6C). Discussion In this study, we found that only a minority of the mutations found in NF1 patients in NF1 exon 29 and in its 5¢SS led to significantly altered levels of exon skipping. This was a rather surprising observation, as in silico analysis of the sequences disrupted by these mutations was predicted to alter several putative splic- ing regulatory sequences. However, other studies have shown that analysis of polymorphic variations and their putative effects on splicing can show a discrep- ancy between splicing defect, protein binding, and ESE predictions [40–43]. The only substitution that can be considered with certainty to be an aberrant splicing mutation is an IVS29 +5g>a intronic substitution, which led to 100% exon skipping. In this respect, it should be noted that a missense mutation, c.5546g>a (R1849Q), in NF1 exon 29 affecting the 5¢SS has also been described previously [36,44]. One of the most surprising findings of this analysis concerns the different splicing outcomes, strikingly divergent from in silico predictions, observed for the nucleotide substitutions in position )2 (Fig. 3). For example, the )2c>g substitution, which generally has no effect on exon skipping unless a guanosine is in the )1 position and a weak 5¢ consensus is present, in our case had a dramatic effect on splicing when inserted in the wild-type pTB NF1–29 minigene. This may well be because the )2 position is also affecting some other context-specific element. This speculation is reinforced by the results obtained when adenosine was placed in the )2c position in the +5g>a pTB NF1–29 minigene; splicing remained completely abolished, instead of hav- ing a compensatory effect, as expected from the overall in silico predictions. Indeed, our mutational data sug- gest that each position in the 5¢SS of exon 29 is critical for the splicing outcome. This result, and the low num- ber of natural 5¢SSs with guanosine in position +4 and adenosine in position +5, strongly suggests the existence of a mutual relationship between these two positions, as previously suggested from a human– mouse comparative genomic analysis by Carmel et al. [45], and that positions +4 and +5 are less tolerant to base combinations that differ from the consensus sequence. This situation could, however, be similar to a recent example in NF1 involving the 5¢SS of exon 3. In this case, an IVS3 +5g>c mutation caused aberrant splic- ing and disease, whereas the same sequence in IVS1 and IVS7 of the NF1 gene did not affect splicing. Fur- ther mechanistic studies of this region showed that hnRNP H binding at the 5¢SS inversely correlated with U1 snRNP binding to this site, and thus with the path- ological effect of this intron 3 mutation [14]. It should be noted, however, that the IVS29 donor site presents a distinct difference with respect to this situation. In fact, the IVS3 +5g>c mutation could be rescued suc- cessfully by a U1 snRNP molecule carrying a compen- satory mutation in position +5 [17], whereas, in our case, cotransfection with a modified U1 designed to complement specifically the +5g>a position was incapable of rescuing the IVS29 +5g>a minigene (Fig. 6B). It is interesting to note that a recent analysis of mutations occurring within the TCIRG1 gene also showed lack of U1 rescue of a +5g>a mutation occurring in the donor site of exon 14, but successful recovery for a +4a>t mutation in for exon 2 [46]. In addition, an early analysis in yeast showed the impor- tance of the +5G residue for the correct identification of the position of the 5¢SS cleavage, suggesting that this is not related to base pairing with U1 but to the sequence composition itself [47]. These results suggest that, in a growing number of cases, U1 snRNP com- plementarity may not necessarily represent the most important step in the initial 5¢SS definition, although one potential criticism of this view is the observation that past studies have reported a rather mixed response of mutant U1 snRNP carrying compensatory substitutions for different mutations in the same 5¢SS [48]. In fact, this lack of suppression by mutant U1 snRNPs may simply be due to the fact that, for some unknown reason, not all mutant U1 snRNPs are equally effective at compensating for particular nucleo- tide substitutions. However, it may also suggest that other factors besides the interaction with U1 snRNP Defective interaction between U1 snRNP and 5¢SS M. Raponi et al. 2068 FEBS Journal 276 (2009) 2060–2073 ª 2009 The Authors Journal compilation ª 2009 FEBS may be more important in the definition of the IVS29 donor site. For example, it is well known that U6 interacts with the fifth nucleotide of the intron [49], and it is well established that base pairing between U6 and the donor site can affect SS recognition [50,51]. An alternative explanation may involve the effect of other cis-acting elements, as has been hypothesized to be the case for other U1-independent model systems. [52]. As a side note, it is important to note that for the recognition of the other end of the exon, represented by the 3¢SS sequence, it has long been known that the absence of some key splicing factors, such as U2AF, can be overcome by an excess of SC-35 protein [53]. Moreover, it has been recently reported that introns with unconventional polypyrimidine tracts are unaf- fected by U2AF inactivation, suggesting that, in some of these cases, there may be additional mechanisms of 3¢SS identification other than that traditionally described [54]. Taken together, these results suggest that both 5¢SS and 3¢SS alternative splicing assembly pathways may coexist within the eukaryotic cell, and further work will be required to clarify this issue. From a clinical point of view, the practical conse- quence of our work resides in the conclusion that most substitutions that occur within or near the NF1 exon 29 donor site may have profoundly adverse con- sequences on its inclusion levels, even in cases where they are not predicted to disrupt U1 snRNA inter- actions. Experimental procedures GenBank access The human NF1 exon\intron 29 sequence can be derived from the genomic NF1 sequence (GenBank accession number: AY796305). Hybrid minigene splicing assay To generate the hybrid minigene constructs, human geno- mic DNA was amplified from normal and mutated exon 29 to generate fragments that contain the exon along with the intronic flanking sequence, using the following oligonucleo- tides: NF29-F, 5¢-ttcattcatatgaccatttgaatatacaatggt-3¢; and NF29-R, 5¢-aagtaacatatgatggagaaaggacatatat-3¢. Both oli- gonucleotides carry an Nde1 site in their 5¢-ends, and they were used to clone the product into a modified version of the a-globin-fibronectin extra type III homology B or extra domain B (EDB) minigene, in which the alternatively spliced EDB exon has been removed to generate a site for the insertion of the genomic sequence under study [55]. For analysis of the splicing pattern in this hybrid minigene expression system, 1 lg of each minigene plasmid was transfected into HeLa and COS cells with Invitrogen lipofectamine reagent. RNA extraction and RT-PCR analysis were performed as previously described [56], using primers complementary to sequences in the flanking fibronectin exonic sequence. Mutant U1 snRNA cotransfection experiments The parental U1 snRNA clone was pG3U1 (WT-U1), a derivative of pHU1 [57]. We created the variants U1+4G, U1+5A, U1+4G+5A and U1–2C+4G+5A+7U by site- directed mutagenesis using pGEM Hind R (5¢-aagctatttagg tgacactatagaa-3¢) as a reverse primer, and U1+4G_Bgl2F (5¢-ccaagatctcatacctacctggcag-3¢), U1+5A_Bgl2F (5¢-ccaag atctcatatttacctggcag-3¢), U1+4G+5A_Bgl2F (5¢-ccaagatct catatctacctggcag-3¢), and U1-2C+4G+5A+7U_Bgl2F (5¢- ccaagatctcaaatctaccgggcaggggaga-3¢), respectively, as the forward primers. Primers carry the appropriate restriction site in order to replace the sequence between the BglII and HindIII sites with mutant clones. We transfected HeLa cells with the lipofectamine reagent, with 1 lg of each minigene plasmid, and with 0.8 lg of the U1 snRNA coding plas- mids. The expression of the transfected U1 variant minig- enes was tested by amplifying the cDNA with specific oligonucleotides, using pGEM Q (5¢-atcgaaattaatacgactca -3¢) as a forward primer and U1_QR (5¢-ctgggaaaac caccttcgt-3¢) as the reverse primer. RNase H digestion to inactivate U1 snRNA Approximately 300 lg of commercial HeLa nuclear extract (CilBiotech, Mons, Belgium, approximate concentration 15 lgÆlL )1 ) were digested with 5 units of RNase H (USB) at 37 °C for 30 min, according to the manufacturer’s instructions, in a 60 lL final reaction volume. In order to inactivate U1 snRNP, a small oligonucleotide (5¢-ccagg taagtat-3¢, U1AS oligonucleotide) was added to the reaction mixture at a final concentration of 5 ngÆlL )1 , and a mock- depleted extract was prepared by adding a random oligo- nucleotide. In order to assess the endogenous RNase H inactivation of U1 snRNP, different concentrations of U1AS oligonucleotide were added directly to 5 lL of HeLa nuclear extract and incubated for 30 min at 37 °C. Total RNA was then isolated by phenol/chloform extraction and ethanol precipitation. The uncleaved and cleaved forms of U1 snRNA were visualized by reverse primer extension, using a protocol described by Mccullough & Berget [39]. The primer used for the reverse transcription reaction corresponds to the U1 loop 2 region of U1 snRNA (5¢-cgg agtgcaatg-3¢). Primer extension products were run on a 6% polyacrylamide sequencing gel before being dried and exposed overnight with X-Omat autoradiographic film (Kodak). M. Raponi et al. Defective interaction between U1 snRNP and 5¢SS FEBS Journal 276 (2009) 2060–2073 ª 2009 The Authors Journal compilation ª 2009 FEBS 2069 [...]... associated with A–>G transition at +4 of the IVS33 splice donor site of the neurofibromatosis type 1 (NF1) gene Hum Mol Genet 3, 663–665 Colapietro P, Gervasini C, Natacci F, Rossi L, Riva P & Larizza L (2003) NF1 exon 7 skipping and sequence alterations in exonic splice enhancers (ESEs) in a neurofibromatosis 1 patient Hum Genet 113, 551–554 Zatkova A, Messiaen L, Vandenbroucke I, Wieser R, Fonatsch... assays The NF1 IVS29-based in vitro splicing system was prepared by amplifying part of this exon/ intron sequence to join it to the tropomyosin-based sequences of the PY7 plasmid [65] The sense primer used to achieve this contained a T7 promoter at the 5¢-end, in order to allow capped RNA synthesis from the amplified products and either the wild-type 2070 exon 29 donor sequence or those carrying the +4A... binding at the 5¢ splice site correlates with the pathological effect of two intronic mutations in the NF-1 and TSHbeta genes Nucleic Acids Res 32, 4224–4236 Pagani F, Buratti E, Stuani C, Bendix R, Dork T & Baralle FE (2002) A new type of mutation causes a splicing defect in ATM Nat Genet 30, 426– 429 Akker SA, Misra S, Aslam S, Morgan EL, Smith PJ, Khoo B & Chew SL (2007) Pre-spliceosomal binding of U1 snRNP. .. mutations The in vitro splicing assay conditions used in these experiments have been extensively described elsewhere [66] U1 snRNP inactivation was achieved by adding 50 ng of U1AS antisense oligonucleotide (5¢ccaggtaagtat-3¢) to the 20 lL splicing mixture Following RNA extraction, the spliced/unspliced mRNA products were amplified by RT-PCR, using a sense primer that contained the exon sequence of either... Samples were incubated at room temperature for 15 min, and were then loaded onto a 4% native acrylamide gel in 0.5· TBE that was run at a constant 150 V for 2 h at 4 °C The gel was then dried and exposed to Kodak XAR autoradiographic film In order to UV-crosslink U1 snRNA to the RNA oligonucleotides, the reaction mixture used for the band-shift analysis was subjected to irradiation at 254 nm for 15 min,... JA (1994) SR proteins can compensate for the loss of U1 snRNP functions in vitro Genes Dev 8, 2704–2717 31 Crispino JD, Blencowe BJ & Sharp PA (1994) Complementation by SR proteins of pre-mRNA splicing reactions depleted of U1 snRNP Science 265, 1866–1869 32 Du H & Rosbash M (2002) The U1 snRNP protein U1C recognizes the 5¢ splice site in the absence of base pairing Nature 419, 86–90 33 Puig O, Bragado-Nilsson... (1991) Influences of separation and adjacent sequences on the use of alternative 5¢ splice sites J Mol Biol 217, 265–281 28 Seraphin B & Rosbash M (1989) Identification of functional U1 snRNA–pre-mRNA complexes committed to spliceosome assembly and splicing Cell 59, 349–358 29 Eperon IC, Ireland DC, Smith RA, Mayeda A & Krainer AR (1993) Pathways for selection of 5¢ splice sites by U1 snRNPs and SF2/ASF EMBO... either NF1 exon 29 or the b-globin sequence and a common antisense primer that was localized on the tropomyosin exon 3 sequence For this reason, no splicing intermediates could be detected RT-PCR was performed in conditions that gave a linear relationship between the input RNA and the PCR products over an extended range of PCR conditions (from four-fold less to four-fold more than the amounts used for the. .. Italy) The mixture was then digested using proteinase K (Sigma, Milano, Italy) at a final concentration of 2 lgÆlL)1 for 30 min at 37 °C before being loading onto the gel Site- directed mutagenesis Mutations were introduced by the two-step PCR method [58], using primers carrying each substitution, and the flanking primers used were NF29-F and NF29-R (sequences available upon request) Bioinformatics predictions... understanding nonsense: exonic mutations that affect splicing Nat Rev Genet 3, 285 298 6 Pagani F & Baralle FE (2004) Genomic variants in exons and introns: identifying the splicing spoilers Nat Rev Genet 5, 389–396 7 Kralovicova J & Vorechovsky I (2007) Global control of aberrant splice- site activation by auxiliary splicing FEBS Journal 276 (2009) 2060–2073 ª 2009 The Authors Journal compilation ª 2009 FEBS . NF29-F, 5¢-ttcattcatatgaccatttgaatatacaatggt-3¢; and NF29-R, 5¢-aagtaacatatgatggagaaaggacatatat-3¢. Both oli- gonucleotides carry an Nde1 site in their 5¢-ends,. between U1 snRNA and the IVS29 donor site sequences, there also remained the possibility that U1 snRNP might still be recruited to the IVS29 donor site by other