A Muscleblind Knockout Model for Myotonic Dystrophy

See allHide authors and affiliations

Science  12 Dec 2003:
Vol. 302, Issue 5652, pp. 1978-1980
DOI: 10.1126/science.1088583


The neuromuscular disease myotonic dystrophy (DM) is caused by microsatellite repeat expansions at two different genomic loci. Mutant DM transcripts are retained in the nucleus together with the muscleblind (Mbnl) proteins, and these abnormal RNAs somehow interfere with pre-mRNA splicing regulation. Here, we show that disruption of the mouse Mbnl1 gene leads to muscle, eye, and RNAsplicing abnormalities that are characteristic of DM disease. Our results support the hypothesis that manifestations of DM can result from sequestration of specific RNAbinding proteins by a repetitive element expansion in a mutant RNA.

Aberrant expansion of microsatellites is associated with a number of neurological diseases (14). Myotonic dystrophy (dystrophia myotonica, DM) is often characterized by myotonia, or delayed muscle relaxation due to repetitive action potentials in myofibers, and muscle degeneration (3). Manifestations of DM may also include heart block, ocular cataracts, hypogonadism, and nervous system dysfunction. DM type 1 (DM1) is caused by a (CTG)n expansion (n = 50 to >3000) in the 3′-untranslated region (3′UTR) of the DMPK gene, whereas DM type 2 (DM2) is caused by a (CCTG)n expansion (n = 75 to ∼11,000) in the first intron of ZNF9 (4, 5). How does the expansion of noncoding CTG or CCTG repeats result in dominantly inherited neuromuscular disease?

Both DM1 and DM2 mutant transcripts accumulate in foci within muscle nuclei (510). An indication that these transcripts are pathogenic comes from studies on HSALR mice that express a large CTG repeat in the 3′UTR of a human skeletal actin transgene (9). These transgenic mice develop myonuclear RNA foci, myotonia, and degenerative muscle changes similar to those seen in human DM. The myotonia in HSALR mice is caused by loss of skeletal muscle chloride (ClC-1) channels due to aberrant pre-mRNA splicing (11, 12). Similar ClC-1 splicing defects exist in DM1 and DM2. However, the connection between accumulation of mutant DM transcripts in the nucleus and altered splice site selection has not been established (1115).

Proteins in the MBNL (muscleblind-like) family bind to expanded CUG repeats in vitro and colocalize with mutant DM and HSALR transcripts in vivo (810, 16, 17). Human muscleblind genes MBNL1, MBNL2, and MBNL3 are homologous to the Drosophila gene muscleblind, which is essential for muscle and eye differentiation (8, 18, 19). MBNL1, the major MBNL gene expressed in human skeletal muscle, encodes multiple protein isoforms, including some that bind to expanded CUG repeats (41 to 42 kD) and others that fail to bind (31-kD isoform generated by exon 3 skipping) (fig. S1) (8, 17). Expression of CUG and CCUG expansion RNAs induces MBNL recruitment into nuclear RNA foci, but there is no evidence that this relocalization results in muscleblind depletion and functional impairment (810, 20).

To test the hypothesis that sequestration of MBNL proteins contributes to DM pathogenesis, we generated mice with a targeted deletion of Mbnl1 exon 3 (E3) (Fig. 1A) (21). We postulated that this targeting strategy would approximate the situation in DM by eliminating synthesis of CUG-binding isoforms (8). Genomic blot analysis demonstrated successful deletion of Mbnl1 E3 in Mbnl1ΔE3E3 mice (Fig. 1B). Loss of E3 expression was confirmed by reverse transcription polymerase chain reaction (RT-PCR); primers in exons 3 and 6 amplified a cDNA product from either Mbnl1+/+ or Mbnl1+/ΔE3 mice that was absent in Mbnl1ΔE3E3 mice (Fig. 1C). As expected, Mbnl1 expression was not fully eliminated in Mbnl1ΔE3E3 mice; RT-PCR products were apparent with primers in constitutively spliced exons 10 and 12, or within exon 13. To confirm elimination of the Mbnl1 41- to 42-kD proteins in Mbnl1ΔE3E3 mice, we used monoclonal antibody 3A4, which recognizes Mbnl1 proteins containing exon 5 (20) (fig. S1). The 41- to 42-kD isoforms in Mbnl1+/+ and Mbnl1+/ΔE3 mice were missing in Mbnl1ΔE3E3 (Fig. 1D). Previous studies have suggested that elevated levels of another RNA-binding protein, CUGBP1, are responsible for DM-associated RNA splicing changes (1315). However, Mbnl1ΔE3E3 mice did not show increased Cugbp1 expression (Fig. 1D).

Fig. 1.

Characterization of Mbnl1ΔE3E3 mice. (A) Targeted disruption of Mbnl1. Illustration includes C57BL/6J Mbnl1 exon organization (open boxes, UTRs; black boxes, open reading frame) together with the 129S1/SvImJ insert (black rectangle), the 129 genomic region with Eco RV (E) (E site in C57BL/6J shown by black box with white E), Xba I (X), and Bam HI (B) sites, the targeting construct with a thymidine kinase marker (TK), floxed (black triangles, loxP sites), neomycin cassette (stippled box with white N), the 129 region (thick black line), and locations of hybridization probes I and II. (B) Genomic DNA analysis of Mbnl1 mice with the use of probe I. The 11-kb Eco RV fragment is derived from C57BL/6J; the mutant is 6.5 kb. (C) Loss of Mbnl1 E3 expression in Mbnl1ΔE3E3. RT-PCR with primers in Mbnl1 exons 3 and 6 (top panel), exons 10 and 12 (middle panel), and exon 13 (lower panel). Predicted lengths (in nucleotides) are from cDNAs. (D) Immunoblot analysis (total spleen protein) showing absence of Mbnl1 41- to 42-kD proteins in Mbnl1ΔE3E3.

Mbnl1ΔE3E3 mice display overt myotonia beginning around 6 weeks of age. Delayed muscle relaxation was most noticeable after a period of rest and showed improvement during activity (movie S1). A similar “warm up” phenomenon is characteristic of myotonia in human DM. Electromyographic recordings confirmed myotonic discharges in all Mbnl1ΔE3E3 mice tested (n = 10) (Fig. 2A). Because myotonia in DM1 and DM2 muscle is associated with aberrant ClC-1 splicing (11, 12), we used RT-PCR assays to investigate the effect of loss of Mbnl1 E3 on ClC-1 (encoded by Clcn1) expression (Fig. 2B). Remarkably, Mbnl1ΔE3E3 mice showed abnormal inclusion of Clcn1 cryptic exons 7a and 8a in a pattern similar to that seen in HSALR mice. Also, some full-length ClC-1 cDNA clones from Mbnl1ΔE3E3 mice showed abnormal inclusion of intron 2 (21), as has been observed in DM and HSALR muscle (11, 12). Notably, these abnormal splice isoforms have premature termination codons and do not encode functional chloride channels (11). By contrast, splicing of the Scn4a sodium channel, the only other ion channel previously associated with myotonia, was normal in Mbnl1ΔE3E3 muscle (21).

Fig. 2.

Myotonia and cataracts. (A) Electromyography (EMG) of Mbnl1 vastus muscle. Arrow (top panel) indicates normal EMG electrode insertional activity in wild-type muscle, whereas insertion triggers myotonic discharges in Mbnl1ΔE3E3 muscle (bottom panel). (B) ClC-1 splicing in DM mouse models. Functional chloride channels are produced when Clcn1 exons 6, 7, and 8 are spliced directly together, whereas isoforms that include cryptic exons 7a or 8a encode truncated nonfunctional proteins (11). Clcn1 exons 6 to 8 are illustrated (open boxes) with primer positions indicated (horizontal arrows). Inclusion of exons 7a and 8a occurs at low levels in wild-type (FVB wt, Mbnl1+/+) and Mbnl1+/ΔE3 muscle but at increased levels in Mbnl1ΔE3E3 and HSALR muscle. (C and D) Loss of ClC-1 protein in Mbnl1ΔE3E3 vastus muscle. Representative images of sections from 11-week-old mice showing reduced ClC-1 immunostaining in Mbnl1ΔE3E3 mice (D) relative to wild-type mice (C). Scale bar, 20 μm. (E and F) Equivalent dystrophin (Dys) levels in Mbnl1+/+ (E) and Mbnl1ΔE3E3 (F) muscle. (G and H) Abnormal muscle histology. Hematoxylin and eosin (H&E)–stained vastus from wild-type (G) and Mbnl1ΔE3E3 (H) mice, showing split myofibers (black arrowhead) and centralized myonuclei (white arrowhead). Scale bar, 30 μm. (I to L) Cataract development. Dilated eyes of 18-week-old mice showing a clear wild-type lens (I) but dust-like opacities (white arrowhead) in Mbnl1ΔE3E3 mice (K). Center bright spot is the lamp reflection. H&E-stained anterior sections (J, L) highlight increased fragmentation (black arrowhead) and opacities (white arrowhead) in Mbnl1ΔE3E3 lens (L) compared to wild-type lens (J).

These results suggested that changes in splice site selection result in the loss of functional ClC-1 from myofiber membranes. Immunofluorescence analysis confirmed a major reduction of ClC-1 protein in Mbnl1ΔE3E3 muscle relative to the muscle of wild-type sibs (Fig. 2, C and D), whereas the membrane-associated proteins dystrophin (Fig. 2, E and F) and α-sarcoglycan (fig. S2) were unaffected. Because abnormalities of ClC-1 splicing in Mbnl1ΔE3E3 muscle are more pronounced than in HSALR muscle, and considering that HSALR mice have a >80% reduction of chloride conductance (11), it is likely that myotonia in Mbnl1ΔE3E3 mice is due to improper ClC-1 pre-mRNA splicing.

Histological analysis of Mbnl1ΔE3E3 mice up to 11 weeks of age did not show major degeneration of muscle fibers. Pathological features in Mbnl1ΔE3E3 muscle included an increase in nuclei with an abnormal (central) position and splitting of myofibers (Fig. 2, G and H). Histologic abnormalities were not observed in Mbnl1+/+ or Mbnl1+/ΔE3 muscle. Besides muscle abnormalities, distinctive ocular cataracts that progress from subcapsular “dust-like” opacities to mature cataracts are a prominent DM-associated feature (3, 4). Similar cataracts were observed in all Mbnl1ΔE3E3 eyes examined (n = 24; 3 to 8 months old) but not in wild-type siblings (Fig. 2, I to L).

Abnormal regulation of alternative splicing has been observed in DM1 muscle for cardiac troponin T (TNNT2), insulin receptor (INSR), and ClC-1 (1315). Of these, analysis of INSR is uninformative because human patterns of INSR alternative splicing are not conserved in mice (22). However, Mbnl1ΔE3E3 adult heart shows abnormal retention of the Tnnt2 “fetal” exon 5 (Fig. 3A), as was observed for DM1 (13). To determine whether alternative splicing of other genes is disrupted in Mbnl1ΔE3E3, we assessed fast skeletal muscle troponin T (Tnnt3). Primers in Tnnt3 exons 2 and 11 produced a single major RT-PCR product in adult Mbnl1+/+ and Mbnl1+/ΔE3 mice that was undetectable in Mbnl1ΔE3E3 mice (Fig. 3B). Instead, a cluster of larger cDNAs, all containing a “fetal” (F) exon (23), was prominent. In contrast, mutually exclusive splicing of Tnnt3 exons 16 and 17 was unaffected in Mbnl1ΔE3E3 mice; this finding shows that altered Mbnl1 expression has specific effects on splice site selection even within the same pre-mRNA (Fig. 3C). Subsequently, we found similar alterations of TNNT3 splicing in adult DM1 muscle (Fig. 3D).

Fig. 3.

Pre-mRNA splicing. (A) Adult retention of Tnnt2 exon 5 in Mbnl1ΔE3E3 heart. RT-PCR products with (+) and without (–) exon 5 (black box) are indicated (brackets). Size markers are pBR322 Msp I fragments. (B) Tnnt3 fetal (F) exon inclusion in adult Mbnl1ΔE3E3. The Tnnt3 protein contains variable N-terminal (alternative splicing of exons 4 to 8 and F) and C-terminal regions (exons 16 and 17) (23). RT-PCR (11-week-old mice) of Tnnt3 exons 2 to 11 (left panel) is shown with alternatively spliced exons 4 to 8 and the fetal (F) exon (black boxes). The F exon contains a Bsr BI site (arrowhead) resulting in comigrating smaller fragments in Mbnl1ΔE3E3 (right panel). (C) RT-PCR of Tnnt3 exons 15 to 18 after Msc I digestion. (D) Retention of TNNT3 fetal (F) exon in adult DM1 skeletal muscle (left panel). Right panel shows cDNAs containing the F exon (bracket) cleaved with Bbs I (arrowhead).

These studies confirm a key prediction of the MBNL protein sequestration hypothesis for DM pathogenesis (8). Loss of specific Mbnl1 isoforms that associate with expanded (CUG)n and (CCUG)n RNAs is sufficient to cause myotonia, cataracts, and RNA splicing defects that are similar to those seen in DM. Although muscleblind-like proteins may influence gene expression at multiple levels, our results raise the possibility that these proteins play a direct role in splice site selection. Indeed, recent studies confirm that MBNL proteins bind to distinct RNA sequence elements and influence exon use during splicing (24).

Young Mbnl1ΔE3E3 mice do not develop the severe neonatal muscle weakness associated with congenital DM1, and we do not yet know whether cardiac conduction problems develop in this model. Thus, some aspects of the DM phenotype may not result from loss of MBNL1 function alone. Additional muscleblind proteins (MBNL2 and MBNL3) are also recruited to nuclear RNA foci (17), so their sequestration may be required to fully replicate the multisystemic DM phenotype. Alternatively, other effects of the repeat expansion mutation on gene expression may be required for the full range of disease manifestations (3, 4, 25, 26).

Supporting Online Material

Materials and Methods


Figs. S1 and S2

Movie S1

References and Notes

View Abstract

Stay Connected to Science

Editor's Blog

Navigate This Article