Molecular defects of presynaptic, synaptic, or postsynaptic proteins at the motor endplate impair neuromuscular transmission and result in congenital myasthenic syndromes (CMS).¹ Mutations in the acetylcholine receptor (AChR) ε subunit gene (CHRNE; OMIM 100725) cause endplate AChR deficiency and/or kinetic abnormalities of AChR. CHRNE mutations causing endplate AChR deficiency include 13 missense, 27 frameshift, 6 nonsense, 8 splice site, 3 promoter region, and 1 chromosomal microdeletion mutations.¹ Only three of the eight splice site mutations have had their consequences characterised: IVS7-2A→G² and IVS9+1G→T³ result in skipping of an adjacent exon, and IVS9-1G→C causes retention of intron 9.⁴ Four of the five uncharacterised mutations were reported by us. Exclusive expression of CHRNE by subsynaptic nuclei has previously prevented analysis of splicing consequences of these mutations. We recently reported that the cloned entire CHRNE exhibits the same splicing properties as its pre-mRNA in the native state in transfected COS cells.⁴ Therefore we used this method to analyse the splicing consequences of four previously reported and five novel splice site mutations in CHRNE.

MATERIALS AND METHODS

All human studies were in accord with the guidelines of the institutional review board of the Mayo Clinic.

Patients

Patients 1–5 (respectively a 59 year old woman, a 23 year old man, a 2.5 year old girl, a 6 year old boy, and a 44 year old man) have moderate to severe myasthenic symptoms that have been present since birth or infancy, decremental EMG responses, and no AChR antibodies. All respond partially to pyridostigmine. Patient 4 underwent an intercostal muscle biopsy for diagnosis, which showed severe endplate AChR deficiency (6% of normal) and compensatory expression of the fetal γ-AChR at the endplate.

Construction of CHRNE clones for splicing analysis

To examine the consequences of the identified splice site mutations, we used the previously constructed pRBG4-CHRNE plasmid, which carries a cytomegalovirus (CMV) promoter and the entire CHRNE spanning 12 exons and 11 introns.⁴ For extended analysis of g.IVS10-9_c.1167dup16, we also constructed a pRBG4 minigene, spanning nucleotide 880 in exon 9 to nucleotide 1457 in exon 12, where position +1 represents the first nucleotide of the first codon of the mature peptide.

Mutations were engineered using the QuikChange Site Directed Mutagenesis kit (Stratagene). Presence of the expected mutation and absence of unwanted artefacts were confirmed by sequencing the entire insert.

Reverse transcription PCR analysis of cytoplasmic RNA of transfected COS cells

Wild type and mutant CHRNE clones were introduced into COS-7 cells, and cytoplasmic total RNA was isolated as described.⁴ We used two or more sets of PCR primers for each construct to screen for skipping of an exon and retention of an intron.

RESULTS

Each patient carries two mutant CHRNE alleles

We detected a total of seven CHRNE mutations in five patients (table 1). Patients 1, 2, and 3 carry homozygous splice site mutations. Five mutations affect pre-mRNA splicing. Analysis of family members reveals that affected siblings carry two mutant CHRNE alleles, whereas unaffected relatives harbour one or no mutant allele (data not shown), indicating that each mutation is recessive.

Splicing consequences of five novel and four previously reported mutations

Muscle mRNA was available only from patient 4. In this patient, RT-PCR analysis revealed that 1259del23 causes retention of intron 11 (fig 1, table 1). As no muscle specimens were available from the other patients for mRNA analysis, we introduced the four novel (table 1) and the four previously reported (table 2) splice site mutations into the cloned entire CHRNE and analysed the cytoplasmic RNA of the transfected COS cells by RT-PCR. Unlike minigenes, entire CHRNE clones mostly yielded a single splicing product. RT-PCR analysis revealed that IVS4-2A→C, IVS6+1G→T, IVS10+2T→G, and 857G→T cause skipping of an adjacent exon; IVS7+2T→C and 1259del23 result in retention of the mutant intron; IVS9-1G→A causes both exon skipping and intron retention; IVS6-1G→C activates a cryptic 3′ splice site; and g.IVS10-9_c.1167dup16 silences the downstream copy of the 3′ splice site (fig 1, tables 1 and 2).

Why is the downstream copy of duplicated 3′ splice sites silent?

The 16 bp duplication (g.IVS10-9_c.1167dup16) generates two copies of 3′ splice sites, but only the upstream copy is used for splicing. To understand the underlying mechanism, we engineered a series of artificial mutations into a minigene spanning CHRNE exons 9 to 12 (fig 2A and B).

Firstly, we examined a role of the polypyrimidine tract of the upstream and downstream copies (fig 2C). The pyrimidine ratios in the polypyrimidine tract are the same for the two copies (18/24 = 75% for the upstream copy and 30/40 = 75% for the downstream copy), and are not likely to account for selection of the splice site. Substitution of “ac” for the invariant “ag” dinucleotide of the upstream copy activated the downstream copy (Mt-AC in fig 2D), indicating that the increased distance from the branch point to the “ag” dinucleotide does not hinder splicing. Because a stretch of t bases in the polypyrimidine tract is more efficient in splicing than c bases,⁵ we mutated “cctt” to “tctt” (Mt-TC), “cttt” (Mt-CT), and “tttt” (Mt-TT), but none activated the downstream copy (fig 2D).

The role of the branch point sequence was then examined. Displacement of an invariant “a” nucleotide downstream (Mt-Br1 in fig 2E), disruption of the native branch point sequence (Mt-Br2 in fig 2E), or both (Mt-Br3 in fig 2E) had no effect on splicing. Preserved splicing even in the absence of the branch point consensus sequence CURAY in Mt-Br2 indicates that an “a” residue somewhere in intron 10 serves as a branch point, and confirms that the position and context of the branch point sequence is degenerative in mammals. We next swapped the seven residue segments and placed the native branch point sequence 16 residues downstream (Mt-Br4 in fig 2E). Mt-Br4 activated both the upstream and downstream copies of the 3′ splice sites, probably because the shortened polypyrimidine tract rendered the upstream copy of the splice acceptor site less competitive than the downstream copy.

DISCUSSION

Sizes of flanking introns predict exon skipping or intron retention

We analysed splicing consequences of nine mutations in CHRNE. Three other splicing mutations had been previously characterised.^2–4 To summarise, exons 5, 6, 8, 9, and 10 are skipped, and introns 7, 9, and 11 are retained. The three retained introns are all short (intron 7, 82 bp; intron 9, 83 bp; and intron 11, 109 bp), whereas four of five skipped exons flank medium to long introns on either or both sides (exon 5 is flanked by 129 and 306 bp introns; exon 6 by 306 and 334 bp introns; exon 8 by 82 and 1210 bp introns; and exon 9 by 1210 and 83 bp introns). This is in accordance with collation of splicing mutations,⁶ and is consistent with the model of exon recognition in vertebrate splicing.⁷ Exon 10, however, is flanked by 83 and 90 bp introns, and is skipped by two distinct mutations, indicating that the size of the flanking introns is only one of several factors that determine the splicing consequence. Indeed, even when we eliminated 1127 bp in the middle of intron 8 (IVS8+17 to IVS8-59) to reduce its size to 83 bp, both IVS7-2A→G and 857G→T still resulted in skipping of exon 7 (data not shown).

Why does a mutation at the 3′ end of an exon affect pre-mRNA splicing?

Analysis of 1801 human 5′ splice sites¹⁷ revealed that a mismatch at position –1 to U1 snRNA is mostly compensated for by a match at position +6 (fig 3). As CHRNE intron 8 has a mismatch at position +6 (fig 1), 857G→T at position –1 probably prevents U1 snRNA from recognising the 5′ splice site of intron 8, and hence causes skipping of exon 8.

To date, 23 splicing mutations of the last nucleotide of an exon have been reported in human to date (table 3). In 17 mutations, the nucleotide at position +6 does not match to U1 snRNA (underlined nucleotides in table 3). In six other mutations that have a matched T nucleotide at position +6, 12 out of 18 nucleotides at positions +3 to +5 are mismatched, whereas in 836 human 5′ splice sites that carry a matched T nucleotide at position +6, only 766 out of 2508 nucleotides are mismatched¹⁷ (Fisher’s exact test, p = 0.003). These observations also support the idea that a mismatch at position –1 is compensated for by matches at positions +3 to +6, especially at position +6.

Duplication of the 3′ splice site

We constructed and analysed a series of artificial mutants to understand the mechanism by which the duplicated 3′ splice sites resulting from g.IVS10-9_c.1167dup16 silence the downstream copy. Scanning model of recognition of the 3′ splice site indicates that the first "ag" dinucleotide after the branch point is used for splicing catalysis.^8,9 Three exceptions have been reported:^10–12 (a) an "ag" dinucleotide less than 13 nucleotides downstream of the branch point is not recognised, probably due to steric effects of trans acting elements; (b) the first "ag" dinucleotide is hidden in a stable secondary structure; and (c) two "ag" dinucleotides that are <12 nucleotides apart compete for being recognised by the splicesome. As the naturally occurring duplication mutant and all artificial mutants except for Mt-Br4 conform to none of the exceptions, they followed the scanning model that favours the first "ag" after the branch point. On the other hand, displacement of a branch point sequence 16 residues downstream (Mt-Br4 in fig 2) placed the "ag" dinucleotide <13 nucleotides downstream of the branch point, and made the upstream copy less competitive than the downstream copy.

Pathogenic duplication of the 3′ splice site has been reported in two other human genes. Both follow the scanning model of recognition of the 3′ splice site. An 18 nucleotide duplication comprising 16 intronic and 2 exonic residues of HEXB encoding the β subunit of β-hexosaminidase results in an active upstream copy of the 3′ splice sites.¹³ A 69 nucleotide duplication comprising 7 intronic and 62 exonic residues of SLC4A1 encoding anion exchanger member 1 also results in an active upstream copy of the 3′ splice sites.¹⁴

Thus, the scanning model of recognition of the 3′ splice site applies to most physiological and pathological duplications of the 3′ splice sites, though exceptions do occur and await explanation.^10,15,16