SCNM1, a Putative RNA Splicing Factor That Modifies Disease Severity in Mice

See allHide authors and affiliations

Science  15 Aug 2003:
Vol. 301, Issue 5635, pp. 967-969
DOI: 10.1126/science.1086187


The severity of many inherited disorders is influenced by genetic background. We describe a modifier interaction in C57BL/6Jmice that converts a chronic movement disorder into a lethal neurological disease. The primary mutation (medJ) changes a splice donor site of the sodium channel gene Scn8a (Nav1.6). The modifier mutation is characteristic of strain C57BL/6Jand introduces a nonsense codon into sodium channel modifier 1 (SCNM1), a zinc finger protein and a putative splice factor. An internally deleted SCNM1 protein is also predicted as a result of exon skipping associated with disruption of a consensus exonic splicing enhancer. The effect of the modifier mutation is to reduce the abundance of correctly spliced sodium channel transcripts below the threshold for survival. Our finding that genetic variation in a putative RNA splicing factor influences disease susceptibility in mice raises the possibility that a similar mechanism modifies the severity of human inherited disorders.

About 10% of human disease mutations alter pre-mRNA splice sites. Genetic variation in proteins that regulate splicing has been predicted to result in trans-acting modification of disease severity for splice-site mutations (1, 2). With the use of the mouse as a model, we describe here an example of genetic interaction between a putative splicing factor and a splice-site mutation in a neuronal sodium channel gene. Scn8a encodes the sodium channel Nav1.6, which is localized on dendrites and axons throughout the nervous system and concentrated at nodes of Ranvier in myelinated axons (3). Mutations in mouse Scn8a cause inherited movement disorders that range in severity from tremor to ataxia, dystonia, and juvenile lethality (46). The severity of the hypomorphic allele Scn8amedJ is determined by the unlinked modifier gene Scnm1 (7).

Scn8amedJ (medJ) contains a mutation in the splice donor site of intron 3 (Fig. 1A). The abundance of Scn8a mRNA is normal in medJ homozygotes, but there is a mixture of correctly and incorrectly spliced transcripts, with predominance of the incorrect transcript (68). This transcript skips exon 2 and exon 3 and encodes a truncated, nonfunctional protein (Fig. 1A). The severity of the medJ movement disorder is dramatically affected by strain background. This effect was mapped to a single Mendelian locus, sodium channel modifier 1 (Scnm1), which determines the proportion of correctly spliced transcripts and hence disease severity (7). C3H and other common inbred strains carry the resistance allele of Scnm1. Resistant medJ/medJ mice produce 10% correctly spliced transcripts, exhibit a progressive disorder with dystonia and ataxia, and live for >1.5 years (8). C57BL/6J mice carry the recessive susceptibility allele of the modifier. Susceptible medJ/medJ mice produce only 5% of correctly spliced transcripts, become paralyzed, and do not survive beyond 1 month (7, 8). Reduction of correctly spliced transcripts to 5% in medJ/– compound heterozygotes is also lethal (8).

Fig. 1.

Molecular variants of Scn8a and Scnm1 interact to influence the severity of neurological disease. (A) A 4-bp deletion in the donor splice site of exon 3 in the Scn8amedJ allele results in exon skipping (+5 to +8). The major processed transcript of the Scn8amedJ allele skips both exon 2 and exon 3 because of the arrangement of U2 and U12 introns (6). (B) Protein domains and amino acid sequence of SCNM1from mammals and fish. The mouse R187X mutation and the acidic domain are boxed, the nuclear localization signal (NLS) is in bold, and the zinc finger (ZnF) is underlined. H, Homo sapiens; B, Bos taurus; M, Mus musculus; F, Fugu rubripes. (C) The R187X mutation is shared by the closely related C57 and C58 strains. Genealogy suggests that the R187X mutation arose in the common ancestor to these strains about 80 years ago (29). The wild-type allele (+) was detected in 31other inbred strains (fig. S1).

We previously mapped the Scnm1 locus to a 950-kb region on mouse chromosome 3 containing about 34 candidate genes and a recombination hot spot (9). To identify Scnm1, we compared the finished genomic sequence of the nonrecombinant region from strain C57BL/6J (9) with the human orthologous sequence on chromosome 1q21 (10). This sequence comparison identified a premature stop codon, R187X, in the C57BL/6J allele of the predicted gene MGC3180 (11). The wild-type MGC3180 gene in strain C3H contains seven exons and encodes a protein of 229 amino acids (Fig. 1B). The stop codon in C57BL/6J removes the C-terminal 43 residues (Fig. 1B).

Reverse transcription polymerase chain reaction (RT-PCR) of RNA from C57BL/6J mice with primers in exon 5 and exon 7 resulted in amplification of the normal 330-base-pair (bp) product and a unique 135-bp product that lacks exon 6 (Fig. 2A). The transcript lacking exon 6 retains an open reading frame and encodes a 164-residue protein, SCNM1Δ133-196. Analysis of the exon 6 sequence with software that detects exon splice enhancer sites (12) indicated that the C-to-T substitution in codon 187 destroys a predicted exon splice enhancer recognized by the arginine- and serine-rich protein ASF/SF2 (Fig. 2B). C57BL/6J mice are predicted to produce two abnormal proteins, one that is prematurely truncated at residue 186 and a smaller 164-residue protein, SCNM1Δ132-196, that lacks the residues encoded by exon 6 (Fig. 1B).

Fig. 2.

Disruption of a consensus exonic splice enhancer in exon 6 of Scnm1 by the R187X mutation in strain C57BL/6J. (A) Aberrant splicing of Scnm1 in C57BL/6J. The transcript that skips exon 6 is detectable by RT-PCR of adult brain RNA from the susceptible strain, C57BL/6J, but not from the wild-type (resistant) strain, C3H. (B) The SF2/ASF consensus exonic splice enhancer sequence is reprinted from (30) with permission. The nucleotide substitution in the mutant allele reduces the match to the consensus from a significant value of 2.1to an insignificant value of –0.4, consistent with loss of function (12).

We determined the strain distribution of the wild-type and R187X alleles by amplifying exon 6 from genomic DNA of 36 inbred strains and digesting the product with the diagnostic restriction enzyme Hpy99I. The wild-type allele was present in 31 strains (fig. S1), including all five strains known to carry the resistance allele of Scnm1 (9). The R187X allele was found only in the closely related C57 and C58 strains, suggesting that the mutation arose in a common ancestor of these strains about 80 years ago (Fig. 1C). Strain C57BLKS/J carries the DBA/2J haplotype in this chromosome region (fig. S1C).

To investigate whether the mutant allele of MGC3180 accounts for the disease susceptibility of strain C57BL/6J, we tested the ability of the wild-type allele to rescue the lethal phenotype. BAC (bacterial artificial chromosome) clone 26A24, isolated from the resistant strain 129S6, contains the wild-type MGC3180 gene as well as 10 flanking genes (fig. S2). Purified BAC DNA was microinjected into fertilized eggs from strain C57BL/6J. Transgenic founder #405 was crossed to the congenic line C57BL/6J-medJ/+, and the resulting transgenic, medJ/+ offspring were backcrossed to the congenic line. The medJ/medJ, Tg405–positive offspring had a typical resistant phenotype, including dystonic postures, demonstrating rescue by the BAC transgene (Table 1). The abundance of the wild-type MGC3180 transcript in brain RNA from line Tg405 was 50% that of the endogenous R187X transcript (fig. S3).

Table 1.

Rescue of C57BL/6J-medJ/medJ mice by transgenic expression of wild-type SCNM1. Rescue is indicated by the survival of transgenic mice beyond 40 days of age and by their typical dystonic phenotype. B indicates C57BL/6J; H, C3H.

Scn8a genotype Scnm1 genotype Line No. surviving >40 days Lifespan (to present) Full-length Scn8a transcript relative level
medJ/medJ BB C57BL/6J-medJ 0/11 30 ± 4 days 1.0 ± 0.2 (n = 3)
medJ/medJ BB Tg405 (BAC) 6/6 >8 months 1.9 ± 0.1 (n = 3)
medJ/medJ BB Tg580 (cDNA) 6/7 >2 monthsView inline 2.1 ± 0.3 (n = 4)
medJ/medJ BH see (View inline) see (View inline) >8 months 1.9 ± 0.3 (n = 3)
medJ/medJ HH see (View inline) see (View inline) >8 months 1.9 ± 0.2 (n = 3)
  • View inline* Killed at 2 months.

  • To confirm that MGC3180 is responsible for BAC rescue of the susceptible phenotype, we carried out a second rescue experiment with the use of a wild-type cDNA from strain C3H under the control of a ubiquitously expressed chicken β-actin promoter (13). The abundance of the wild-type MGC3180 transcript in brain RNA from transgenic line C57BL/6J-Tg580 expressing the cDNA construct was twice that of the endogenous R187X transcript, demonstrating that excess transcripts are tolerated (fig. S3). Transgenic mice from this line were backcrossed to the C57BL/6J-medJ/+ line as described above. C57BL/6J-medJ/medJ, Tg580 mice exhibited the dystonic phenotype and were rescued from juvenile lethality (Table 1). Thus, the predicted gene MGC3180 is the sodium channel modifier gene Scnm1.

    To examine the splicing of the Scn8amedJ pre-mRNA in the rescued transgenic mice, we determined the relative amounts of full-length and mutant transcripts with the use of a primer extension–chain termination assay (8, 14). The proportion of correctly spliced Scn8a transcripts in both lines of transgenic mice was twofold higher than in nontransgenic medJ/medJ littermates (Table 1). The level of correctly spliced transcript in rescued transgenic mice was comparable to the level previously measured in mice with the wild-type (resistance) allele of Scnm1 (8), demonstrating transgenic rescue of the C57BL/6J splicing defect. The fivefold difference in wild-type SCNM1 expression levels between Tg405 and Tg580 did not change the proportion of correctly spliced Scn8amedJ pre-mRNA, indicating that SCNM1 is not rate-limiting for splicing in this concentration range.

    Orthologs of Scnm1 were identified in mammals, chicken, fish, and the urochordate Ciona with the use of public sequence databases, but no orthologs were detected in flies, worms, or yeast. Analysis of protein domains in the coding sequence revealed one C2H2 zinc finger, a basic nuclear localization signal, and a C-terminal acidic domain (Fig. 1B). The C2H2 zinc finger domain is 100% conserved in the mammalian orthologs of SCNM1 (Fig. 1B, underlined). Phylogenetic comparison of zinc finger sequences places SCNM1 within the U1C subfamily of RNA binding proteins that is commonly found in RNA-processing proteins (fig. S4). The U1C splice factor functions in recognition of splice donor sites (15, 16). Two other subfamily members, SF3A2 and WBP4, are also involved in splicing (17, 18). The homology between the zinc fingers in SCNM1 and established splice factors suggests that SCNM1 may also function in splicing.

    The basic nuclear localization signal of SCNM1 matches the bipartite consensus (K/R)2 X10-12 (K/R)3 (19) (Fig. 1B). To examine subcellular localization, we fused the coding sequences of wild-type SCNM1229 and SCNM1186 downstream of the green fluorescent protein. Both fusion proteins were localized exclusively to the nucleus in transfected cells (fig. S5).

    The R187X mutation in Scnm1 impairs the in vivo splicing of Scn8amedJ in C57BL/6J mice. Because Scnm1 is widely expressed in mouse embryonic and adult tissues (20), the mutation could also result in genome-wide changes in pre-mRNA processing, as observed for deletion of individual splicing factors in yeast (21). Impaired splicing of unlinked genes could be responsible for quantitative trait loci alleles and modifiers in strain C57BL/6J that map to this chromosome region, including vestibular dysfunction and alcohol preference (2224). R187X could also confer general susceptibility to de novo splice-site mutations. Coisogenic C57BL/6J-Tg580 mice expressing wild-type SCNM1 will be useful for evaluating these predicted effects of Scnm1.

    The effect of genetic background on the severity of inherited disorders has long been recognized (25), but only a few modifiers have been molecularly identified to date (26, 27). Trans-acting splice factors have been suggested as one class of modifiers of the severity of human inherited disorders (28). Scnm1 provides an example of this type of disease susceptibility through its trans-effect on the Scn8amedJ transcript. The role of human SCNM1 in modulating splicing defects can now be tested with the use of linkage markers on human chromosome 1q21.

    Supporting Online Material

    Materials and Methods

    Figs. S1to S6

    References and Notes

    References and Notes

    View Abstract

    Navigate This Article