Seemingly neutral polymorphic variants may confer immunity to splicing-inactivating mutations: a synonymous SNP in exon 5 of MCAD protects from deleterious mutations in a flanking exonic splicing enhancer

Abstract

The idea that point mutations in exons may affect splicing is intriguing and adds an additional layer of complexity when evaluating their possible effects. Even in the best-studied examples, the molecular mechanisms are not fully understood. Here, we use patient cells, model minigenes, and in vitro assays to show that a missense mutation in exon 5 of the medium-chain acyl-CoA dehydrogenase (MCAD) gene primarily causes exon skipping by inactivating a crucial exonic splicing enhancer (ESE), thus leading to loss of a functional protein and to MCAD deficiency. This ESE functions by antagonizing a juxtaposed exonic splicing silencer (ESS) and is necessary to define a suboptimal 3' splice site. Remarkably, a synonymous polymorphic variation in MCAD exon 5 inactivates the ESS, and, although this has no effect on splicing by itself, it makes splicing immune to deleterious mutations in the ESE. Furthermore, the region of MCAD exon 5 that harbors these elements is nearly identical to the exon 7 region of the survival of motor neuron (SMN) genes that contains the deleterious silent mutation in SMN2, indicating a very similar and finely tuned interplay between regulatory elements in these two genes. Our findings illustrate a mechanism for dramatic context-dependent effects of single-nucleotide polymorphisms on gene-expression regulation and show that it is essential that potential deleterious effects of mutations on splicing be evaluated in the context of the relevant haplotype.

Figure 1.

Analysis of MCAD cDNA from patient fibroblasts. A, We used a LightCycler-based assay for the c.985A→G mutation, to analyze two samples of cDNA from cultured fibroblasts from patient 1 and from four other individuals heterozygous for the c.985A→G mutation. In patient 1, the height of the peak resulting from the allele without the c.985A→G mutation (i.e., the allele with the c.362C→T mutation) was reduced compared with the peak from the allele with the c.985A→G mutation. When four other individuals heterozygous for the c.985A→G mutation were analyzed, the two peaks were of comparable size. This indicates that MCAD mRNA from the allele with the c.362C→T mutation is present in lower amounts than MCAD mRNA from the other MCAD allele. B, Quantitative PCR analysis of the relative amounts of MCAD cDNA relative to β-actin cDNA. Analysis of cDNA prepared from fibroblasts from patient (Pat) 2, who is homozygous for the c.362C→T mutation, showed that MCAD mRNA from the allele with the c.362C→T mutation results in only ∼20% of that in controls (Cont). Patient E, described elsewhere, is homozygous for a 2-bp deletion in exon 11, corresponding to cDNA positions 955–956, that leads to severely decreased amounts of MCAD mRNA.^, The mean of the four control fibroblasts (controls A–D) was assigned the value 100%, and the error bars indicate the range (n=3). C, Amplification of cDNA from patients 1 and 2 and from three controls with primers located in exons 4 and 6 demonstrated that mRNA from alleles with the c.362C→T mutation display a high level of exon 5 skipping. When the bands were excised and sequenced separately, results showed that the normal-sized band from patient 1 had a wild-type sequence at position c.362, whereas the shorter band from the two patients was missing exon 5.

Figure 2.

Minigene analysis. A, Schematic representation showing the MCAD minigene. The variants in which the weak intron 4 3′ splice site was optimized by substitution to pyrimidines at positions −15(c), −10(c), −9(t), and −7(t) were named “IVS.” Part of the sequence at the intron 4–exon 5 splice site of the IVS and wild-type (Wt) versions, as well as the corresponding consensus sequence (Cons), is displayed. Analysis by the Splice Site Prediction by Neural Network program (Fruitfly) gives a score of only 0.12 for the suboptimal intron 4 3’ splice site, whereas the optimized sequence gives a score of 1.0. Also, the positions of the Del3, Del9, and c.362C→T mutations in the MCAD sequence (positions c.351–c.371) are shown. Nearly identical results were observed when we transfected other types of cells (i.e., HEK-293, COS-7, N2A, and NT2) (data not shown). B, Schematic representation of MCAD exon 5 introduced into the VLCAD minigene. Wild-type (VL362C) and c.362C→T mutant (VL362T) exon 5 with 80 bp of intron 4 and 145 bp of intron 5 replaced VLCAD exon 12 and part of the flanking intronic sequences. Gels show the results from amplification of cDNA with minigene-specific primers.

Figure 3.

Characterization of the MCAD c.362 ESE. A, An alignment of all the variant sequences, which were inserted around position c.362 in the MCAD minigene, as well as the relevant part of one of the winner sequences (SF2/ASF-22 [B5]) identified by functional systematic evolution of ligands by exponential enrichment (SELEX). Mutant positions are underlined. A pictogram for the SF2/ASF score matrix is shown above the relevant part of the sequence. Scores from ESEfinder analysis for SF2/ASF, SC35, and SRp40 high-score motifs are shown to the right. B, A c.363T→G suppressor mutation, which reestablished the SF2/ASF high-score motif when present with the c.362C→T mutation, introduced into the wild-type and c.362C→T MCAD minigenes. C, We substituted positions c.355-c.364 of the MCAD sequence in the MCAD minigene with positions 4–12 of exon 7 from the human SMN1 and SMN2 genes, including the SF2/ASF heptamer motif and the inactive mutant form. Nucleotides corresponding to positions c.355–c.366 of the MCAD sequence were substituted with positions 1–12 from the human BRCA1 gene exon 18, thereby introducing the SF2/ASF–dependent ESE. The heptamer with the inactivating BRCA1-NL mutation was also inserted. Nucleotides corresponding to positions c.357–c.368 of MCAD were substituted with nucleotides 20–32 of HBB exon 2, which has been reported to contain a functional SR protein–dependent ESE. The SF2/ASF high-score heptamer and the three flanking nucleotides from the HBB exon 2 ESE exactly match the winner sequence (SF2/ASF-22 [B5]). The SMN1, SMN2, BRCA1-WT, and BRCA1-NL heptamers were also inserted in the MCAD minigene with an optimized intron 4 3′ splice site (IVS). D, Schematic representation showing the BRCA1 exon 18 minigene. We made four BRCA1 exon 18 minigenes with different heptamers at exon 18 positions 4–10: one with BRCA1-WT, one with the inactivating G→T mutation (BRCA1-NL), one construct in which the BRCA1 heptamer was replaced by the wild-type MCAD heptamer (BRCA1-MCAD 362C), and one in which the BRCA1 heptamer was replaced by the c.362C→T mutant MCAD heptamer (BRCA1-MCAD 362T). An alignment of the four different inserted sequences is shown with the calculated scores for SF2/ASF, SC35, and SRp40 from the ESEfinder program. The mutations are underlined. E, The wild-type and c.362C→T mutant MCAD minigenes, as well as the versions with the 3-bp (positions c.361–363) and 9-bp (positions c.358-366) deletions, cotransfected with vectors overexpressing SF2/ASF or SRp40 or with an expression vector without an insert. All the minigenes were transfected into Chang cells, and cDNA was prepared, was amplified by minigene-specific primers, and was analyzed by electrophoresis in 2% agarose gels.

Figure 4.

The c.351A→C synonymous polymorphism inactivates an ESS. Various combinations of mutations were introduced in the MCAD minigene around positions c.351 and c.362. An alignment of the sequences from positions −5 in intron 6 to +16 in exon 7 of the SMN2 gene with positions c.348–c.368 from all the MCAD minigene constructs is shown, with the mutant positions underlined. The exinct element is boxed. The uppercase letters represent the exonic sequence, and the lowercase letters represent the intronic sequences of the SMN2 gene. The functional ESS sequence is shaded in gray, and the inactivated ESS is shaded in black. All minigenes were transfected transiently into Chang cells, and cDNA was prepared, was amplified with minigene-specific primers, and was analyzed by electrophoresis in 2% agarose gels.

Figure 5.

Analysis of the CAGGGG/CAGGGT splicing silencer and its interplay with the MCAD and SMN1 ESEs. Different combinations of the functional (351A) or inactive (351C) MCAD ESS, a prototypical hnRNP A1–binding ESS from HIV-1 tat exon 3 (ESS3A) or an inactive version (ESS3C), and a wild-type (C5A) or mutant (C5G) sequence from exon 10 of the HC5 gene (see the “Discussion” section) were tested in combination with the MCAD wild-type (362C) or mutant (362T) ESE in the pSXN reporter minigene. Also, the wild-type splicing silencer from SMN1/2 (351A) or an inactive version (351C) was tested in combination with either the functional SMN1 ESE or the corresponding nonfunctional SMN2 heptameric sequence. For easy comparison, an alignment of all the inserted sequences is shown. The functional splicing-silencer elements are marked as gray boxes with black letters, and inactivated splicing silencers are marked as black boxes with white letters. All minigenes were transfected transiently into Chang cells, and cDNA was prepared, was amplified with minigene-specific primers, and was analyzed by electrophoresis in 4% agarose gels.

Figure 6.

Analysis for high-score motifs for hnRNP A1 binding and identification of potential splicing-inhibitory hexamers with the use of FAS-ESS. A position-weight scoring matrix for hnRNP A1 binding (top) based on the original SELEX winner sequences identified by Burd and Dreyfuss was recently reported. We used the version with background composition correction for the winner pool, to calculate scores for the known and potential ESS sequences listed. High-score motifs for hnRNP A1 binding are written in black letters, and the corresponding inactive sequences are written in black boxes with white letters. Mutational changes are underlined. Calculated high scores are listed in blue, whereas scores <2.7 are listed in red. Potential splicing-inhibitory motifs identified by analysis of the sequences with the FAS-ESS program are listed to the right.

Figure 7.

Binding of hnRNP A1 to the MCAD c.351A ESS. A, Affinity chromatography with MCAD exon 5 RNA. Four different in vitro–transcribed MCAD exon 5 RNAs used for affinity chromatography with HeLa nuclear extract (NE). The proteins bound were analyzed by SDS-PAGE and western blotting with the use of an hRNP A1 antibody or an SF2/ASF antibody. The in vitro–transcribed MCAD exon 5 RNA harbored either the functional ESS (351A) or the inactivated ESS (351C), in combination with either wild-type (362C) or ESE (362T) sequence. B, RNA oligonucleotide-affinity chromatography. The sequences of the six different RNA oligonucleotides that were used for affinity chromatography in HeLa NE are shown. The bound proteins were analyzed by western blotting with the use of hnRNP A1 and SF2/ASF antibodies. C, RNA interference. For RNA interference, either double-stranded RNA oligonucleotides directed toward hnRNP A1 and hnRNP A2 (A1+A2) or a negative control were transfected into HEK-293 cells. The effect on splicing from four different MCAD minigenes was monitored by amplification of cDNA with minigene-specific primers. Transfections were done in triplicate, and the gel shows the results from one representative experiment. hnRNP A1 down-regulation was monitored by western blotting with the use of hnRNP A1 and hnRNP A2/B1 antibodies (not shown).

Figure 8.

Models for splicing regulation of MCAD exon 5 (A–D), SMN1/2 exon 7 (E and F), and HC5 exon 10 (G and H). A–D, MCAD exon 5. The weak intron 4 3′ splice site is inefficiently recognized by U2AF, and its recognition is probably dependent on assembly of stimulatory complexes on an ESE in exon 5. One possibility is that the purine-rich motif (ESE2?), which is nearly identical to the hTra2-β1–binding ESE2 in SMN exon 7, assists in recognition of the weak intron 4 3′ splice site in this way. Binding of hnRNP A1 to the ESS (351A) inhibits recognition of the weak 3′ splice site, probably by antagonizing the stimulatory effect of splicing-enhancer elements (like ESE2?) and/or more directly by functioning as a primary hnRNP A1-binding site, which stimulates binding of further hnRNP A1 molecules in a way that may directly mask the weak 3′ splice site and/or antagonize the function of crucial ESEs (like ESE2?). The primary function of ESE1, which comprises position c.362, is to antagonize binding of hnRNP A1 to the ESS. The c.362C→T mutation decreases the strength of ESE1, so that hnRNP A1 binding to the flanking ESS is favored. ESE1 is necessary for correct splicing only when the ESS (351A) is functional. In alleles with c.351C, the ESS is not functional, and ESE1 is therefore not required to inhibit hnRNP A1 binding. The ESE1 may function by binding SF2/ASF (as depicted in the figure) or another splicing factor with similar recognition motif and function. E and F, SMN1 and SMN2 exon 7. Recognition of exon 7 is dependent on ESE1 and ESE2 for recognition of the weak intron 6 3′ splice site. Binding of hTra2β1 to ESE2 recruits SRp30c and hnRNPG, and this complex assists, perhaps via interactions of the RS domains of hTra2β1/SRp30c and U2AF, in recognition of the 3′ splice site. An hnRNP A1–binding motif (CAGGGT), which can function as a splicing silencer, overlaps with the 3′ splice site, and it is likely that U2AF and hnRNP A1 compete for binding at this site. In SMN1, binding of SF2/ASF to ESE1 antagonizes binding of hnRNP A1 to the splicing silencer, thereby allowing U2AF binding and exon 7 inclusion. In addition to counteracting hnRNP A1 binding to the splicing silencer, it is also possible that the RS domain of bound SF2/ASF participates in U2AF recruitment. In SMN2 the +6C→T substitution in ESE1 allows binding of hnRNP A1 to the splicing silencer. U2AF binding is thus antagonized, and exon 7 is skipped. G and H, HC5 exon 10. An A→G mutation at the penultimate position of exon 10 has been shown to cause exon 10 skipping. This A→G mutation generates an hnRNP A1–binding site (CAGGGT) identical to the splicing silencer in SMN1/2. Binding of hnRNP A1 at the 5’ splice site would block access by U1 snRNP and result in skipping of the exon. Exonic sequences are represented by uppercase letters (boxed) and intronic sequences by lowercase letters. Blue letters represent the SF2/ASF (blue) recognition sequence (ESE1), green letters the hTra2β1 (green) binding motif (ESE2), and red letters the hnRNP A1 (red) recognition sequence (ESS). U2AF dimer (yellow) and U1snRNP (purple) are shown.