Myotonic Dystrophy Type 2 Caused by a CCTG Expansion in Intron 1 of ZNF9

See allHide authors and affiliations

Science  03 Aug 2001:
Vol. 293, Issue 5531, pp. 864-867
DOI: 10.1126/science.1062125


Myotonic dystrophy (DM), the most common form of muscular dystrophy in adults, can be caused by a mutation on either chromosome 19q13 (DM1) or 3q21 (DM2/PROMM). DM1 is caused by a CTG expansion in the 3′ untranslated region of the dystrophia myotonica–protein kinase gene (DMPK). Several mechanisms have been invoked to explain how this mutation, which does not alter the protein-coding portion of a gene, causes the specific constellation of clinical features characteristic of DM. We now report that DM2 is caused by a CCTG expansion (mean ∼5000 repeats) located in intron 1 of the zinc finger protein 9 (ZNF9) gene. Parallels between these mutations indicate that microsatellite expansions in RNA can be pathogenic and cause the multisystemic features of DM1 and DM2.

DM is a dominantly inherited, multisystemic disease with a consistent constellation of seemingly unrelated and rare clinical features including myotonia, muscular dystrophy, cardiac conduction defects, posterior iridescent cataracts, and endocrine disorders (1). DM was first described nearly 100 years ago (2), but the existence of more than one genetic cause was only recognized after genetic testing became available for myotonic dystrophy type 1 (DM1) (3,4).

DM1 is caused by an expanded CTG repeat on chromosome 19 that is both in the 3′ untranslated region of the dystrophia myotonica–protein kinase (DMPK) gene, and in the promoter region of the immediately adjacent homeodomain gene SIX5(5, 6). How the CTG expansion in a noncoding region of a gene causes the complex DM phenotype remains unclear (5,6). Suggested mechanisms include: (i) haploinsufficiency of the dystrophia myotonica–protein kinase (DMPK) protein (7); (ii) altered expression of neighboring genes, includingSIX5 (8–12); and (iii) pathogenic effects of the CUG expansion in RNA which accumulates as nuclear foci (13, 14) and disrupts cellular function (15–18). Several mouse models have developed different aspects of DM1: a model expressing mRNA with CUG repeats manifests myotonia and the myopathic features of DM1 (19); aDMPK knockout has cardiac abnormalities (20); andSIX5 knockouts have cataracts (21, 22). Taken together, these data have been interpreted to suggest that each theory may contribute to DM1 pathogenesis and that DM1 may be a regional gene disorder (5, 6).

Defining a second human mutation that causes the multisystemic effects of DM, and identifying what is common to these diseases at the molecular level, provides an independent means of determining the pathogenic pathway of DM. Toward this goal, we mapped the myotonic dystrophy type 2 [DM2/proximal myotonic myopathy (PROMM)] locus to chromosome 3q21 (23, 24) and have used positional cloning to identify the DM2 mutation.

The DM2 region was narrowed to a 2-cM interval (25) by analyzing 10 recombinant chromosomes (25). Sequence data (26, 27) from this region, which is partially covered by 14 bacterial artificial chromosomes (BACs), was used to develop 80 short tandem repeat (STR) markers. Linkage disequilibrium analysis (27) was performed on 64 parent-offspring trios in which affected individuals had the clinical features of DM but not the DM1 mutation. Transmission disequilibrium testing (28) and analysis of conserved ancestral haplotypes narrowed the DM2 locus to a region of ∼320 kilobases (kb) (Fig. 1A).

Figure 1

Expanded CL3N58 allele found in DM2 patients. (A) DM2 critical region. Black represents the minimal DM2 critical region, white represents DM2 excluded regions, and gray represents regions in which recombination has occurred. Markers defining recombination events and establishing linkage disequilibrium are shown, along with previously published markers. The relative significance of the P-values are indicated by plusses above the marker names, with “++” ≤ 0.01, “+++” ≤ 0.001 “++++” ≤ 0.0001, and “++++++” ≤ 0.000001. Three BACs (orientation unknown) within the region of linkage disequilibrium are shown. Not drawn to scale. (B) Pedigrees of three different DM2-linked families, each represented by a nuclear family. (C) PCR analysis of CL3N58 marker. The genotype of each individual is shown, with each allele given in base pairs. Unamplified alleles are represented by “–”. (D) Southern-blot analysis of expansion mutations. Individuals with an expanded CCTG track are represented by “EXP,” and individuals with two normal alleles are represented by “N.” The blot was also hybridized with an SCA8 loading control, showing that all but the first lane was evenly loaded. (E) High-resolution sizing of expansions. Lane 3 contains DNA from a control sample. The number of CCTGs of each individual's expanded allele is shown, with “N” representing a normal length CCTG tract.

One of the markers in linkage disequilibrium with DM2,CL3N58 (P ≤ 0.000001), showed an aberrant segregation pattern. All affected individuals appeared to be homozygous by polymerase chain reaction (PCR) (29), and affected children appeared not to inherit an allele from their affected parent (Fig. 1, B and C). Southern (DNA) analysis was performed (30) to investigate the possibility that the aberrant segregation pattern was caused by a repeat expansion or other rearrangement. In addition to the expected normal allele, we detected a variably sized expanded allele, too large to amplify by PCR, that was found only in affected individuals (Fig. 1, B and D). Modified electrophoresis conditions (27, 30) enabled us to resolve a range of expansions between 10 and 48 kb (Fig. 1E).

To determine if this expansion was involved in the DM2 disease process, we performed PCR and Southern analysis on: (i) 51 affected individuals in six families whose disease was consistent with linkage to the DM2 locus; (ii) one affected individual from each of 20 additional families with ancestrally conserved DM2 haplotypes; and (iii) a panel of control genomic samples representing 1360 chromosomes. PCR showed that all 51 affected individuals in the six DM2 families appeared to be homozygous, but each individual had an expanded allele on subsequent Southern analysis (31). The maximum lod scores at Θ = 0.00 between the disease locus and the CL3N58 expansion for the six families were: MN1 = 6.9, MN6 = 1.5, MN10 = 8.2, MN12 = 2.8, F134 = 10.4, and F047 = 1.8. Expanded alleles detected by Southern analysis were also found in affected representatives of all 20 additional families with ancestrally conserved DM2 haplotypes. PCR and Southern analysis identified no control samples with an expansion.

Sequence of the CL3N58 marker contains the complex repeat motif (TG)n(TCTG)n(CCTG)n. In our control group, the size of the (TG)n(TCTG)n(CCTG)nrepeat tract ranged from 104 to 176 base pairs (bp) (Heterozygosity = 0.89) (Fig. 2A). All eight normal alleles that were sequenced (29) had CCTG repeat tracts that were interrupted by both GCTG and TCTG motifs or by one or two TCTG motifs (Fig. 2B). The repeat tract in the largest normal allele (combined TG/TCTG/CCTG repeats of 176 bp) was sequenced and shown to contain 26 CCTG repeats with two interruptions. Smaller expansions from three DM2 patients were sequenced (29), demonstrating that the CCTG portion of the repeat tract is expanded. In contrast to alleles from the control samples, the CCTG repeat tracts on expanded alleles were uninterrupted. Expansion sizes for very large alleles were estimated by Southern analysis assuming that, consistent with the sequenced expansions, lengthening of the CCTG repeat tract accounts for the increase in molecular weight. The range of expanded allele sizes is extremely broad, from 75 to ∼11,000 CCTG repeats with a mean of ∼5000 (Fig. 2C). Shorter expansions were found in individuals with multiple allele sizes in blood, so that the smallest pathogenic size is uncertain.

Figure 2

Analysis of DM2-affected and normal alleles. (A) Distribution of CL3N58 alleles among controls (n = 1360). Alleles represent the total base-pair size of the combined TG, TCTG, and CCTG repeat tracts. (B) Schematic diagram of DM2 expansion region, showing sequence configurations of normal and expanded repeat tracts. (C) Distribution of expanded alleles among 51 affected members of six DM2 families. All expanded allele sizes were included for individuals with multiple bands and, in contrast to (B), are given in CCTG repeat units.

In approximately 25% of the affected individuals, we observed two to four bands in DNA isolated from blood, representing expanded alleles of various sizes (Fig. 3A and Web table 1) (27). Some bands were discrete in size, some appeared as unresolved compression bands at the top of the gel, and others showed a broad variation of molecular weight. Additional examples of somatic instability included: (i) a pair of genetically confirmed (P ≤ 0.001) monozygotic twins (31 years old) (27) had dramatically different expanded alleles (13 and 24 kb) (Fig. 3B); (ii) the expansion size in lymphocyte DNA from an affected individual increased in size by approximately 2 kb during the 3-year interval between blood donations (Fig. 3C); and (iii) the age of affected individuals at the time they donated a blood sample directly correlated (r = 0.41,r 2 = 0.17, P = 0.008) with the size of the expansion (Fig. 3D). Expansion sizes in the blood of affected children are usually shorter than in their parents: the time-dependent somatic variation of repeat size complicates the interpretation of this difference (Web table 1) (27). No significant correlation between age of onset and expansion size was observed.

Figure 3

Instability of the DM2 expansion. (A) Somatic heterogeneity in blood. Southern blots of Bso BI–digested genomic DNA from blood revealed multiple expanded alleles in some affected individuals, some discrete in size (lanes 1 and 2) and others broad (lane 3). (B) Southern blots of Eco RI–digested genomic DNA from blood of monozygotic twins (lanes 4 and 5). (C) Expanded alleles increase in length over time. Southern blot of Eco RI–digested genomic DNA samples from blood taken from a single patient at 28 (lane 6) and 31 (lane 7) years of age, respectively. (D) Correlation between the size of the expanded allele in individuals with a single allele and age at the time blood sample was taken.

The DM2 expansion (CL3N58) is located in a region of the genome for which the available sequence was not completely ordered. To determine the location of the DM2 expansion, we sequenced portions of the BAC RP11-814L21 to assemble unfinished sequence contigs (27). Our sequencing data and sequence from the Human Genome Project (26) indicate that the expansion is located in intron 1 of the zinc finger protein 9 (ZNF9) gene (Fig. 4A) (32), also referred to as the cellular nucleic acid-binding protein gene. ZNF9 contains seven zinc finger domains (33) and is thought to be an RNA-binding protein (34,35). Although the originally reported genomic sequence for ZNF9 (32, 36) did not contain theCL3N58 marker, we have generated additional sequence, used sequence from Celera (32, 37) (x2HTBKUAD8C), and performed Southern (27, 31) and reverse transcription PCR (RT-PCR) analysis (31) to confirm the location of the expansion. ZNF9 transcripts are broadly expressed and are most abundant in heart and skeletal muscle (Web fig. 1) (27), two tissues prominently affected in DM2.

Figure 4

RNA in situ hybridization of the expansion. (A) Genomic organization of the ZNF9 gene. The position of the DM2 expansion in intron 1 is shown. The gene spans 11.3 kb of genomic sequence with an open reading frame of 1.5 kb. (B) In situ hybridization of CAGG probe to DM2 muscle. (C) In situ hybridization of CAGG probe to normal muscle. (D) In situ hybridization of CAG probe to DM1 muscle. Bar is 5 μM.

In situ hybridization has been used to detect nuclear foci containing the CUG expansion in DM1 cells (13). Because DM2 is also caused by an expansion motif, we performed fluorescent in situ hybridization (38) to determine if similar repeat-containing nuclear foci are found in DM2. Fluorescently labeled antisense oligonucleotide probes to the CCUG repeat were hybridized to control, DM2, and DM1 muscle biopsy tissue. The DM2 muscle biopsy was from an affected member of the 3q-linked MN1 family (LOD = 6.9), who had a CCTG expansion detected by Southern analysis. Similarly, DM1 tissue was taken from a genetically confirmed DM1 patient. Numerous intense CCUG-containing nuclear foci were observed in DM2 (Fig. 4B), but not in control, muscle (Fig. 4C). In DM2 muscle, one to five foci were seen per nucleus, with no foci detected in the cytoplasm. In general, more foci were seen per nucleus in DM2 than were seen using antisense probes to the CUG expansions in DM1 muscle (Fig. 4, B and D). The sense CCUG probes showed no nuclear foci, indicating that the probe hybridized to RNA, not DNA (31). Our results show that the CCTG expansion is expressed, but we do not yet know if the RNA foci contain the entire unprocessed ZNF9 transcript. The antisense CCUG probe showed no nuclear foci in DM1 muscle (31). Although the antisense probe to the CUG repeat also hybridized to foci in DM2 muscle (31), we believe this signal was caused by nonspecific cross-hybridization to the extremely large CCUG repeat tract (11,000 repeats).

Our results demonstrate that DM2 is caused by an untranslated CCTG expansion. DM2 shows remarkable clinical similarity to DM1, although the disease course of DM2 is usually more benign. Clinical and molecular parallels between these diseases indicate that the CUG and CCUG expansions expressed at the RNA level can themselves be pathogenic and cause the multisystemic features of DM1 and DM2. The normal function of ZNF9 as an RNA-binding protein (35,36) appears unrelated to any of the proteins encoded in the DM1 region of chromosome 19. Similarly, the genes in the DM2 region (KIAA1160, Rab 11B, glycoprotein IX,FLJ11631, and FLJ12057) bear no obvious relationship to the genes at the DM1 locus, indicating that the clinical features common to DM1 and DM2 are not related to disruptions in the regulation of genes in the vicinity of these expansions.

Given the similarity of the DM1 and DM2 repeat motifs and the fact that the expansions accumulate as RNA foci, RNA-binding proteins that bind to the DM1 CUG expansion may also bind to the DM2 CCUG expansion, causing similar global disruptions in RNA splicing and cellular metabolism (15, 17, 18). One of these proteins has been shown to have a preferential affinity for UG dinucleotides (39), which are found in both DM1 and DM2 expansions. If these same RNA-binding proteins are involved in DM2 pathogenesis, then one could speculate that the longer CCUG repeat tracts cause the milder DM2 phenotype because the affinity of these proteins for the CCUG repeat tract is not as strong. Alternatively, a different set of RNA-binding proteins may bind to the CCUG expansion.

DM2 is the fourth example of a dominant disease that is caused by a microsatellite expansion located in a transcribed but untranslated portion of its respective gene. On the molecular level, the CCTG DM2 expansion has parallels to the untranslated CTG expansions involved in both DM1 (5, 6) and SCA8 (40) as well as the ATTCT expansion in SCA10 (41). The DM2 tetranucleotide and the SCA10 pentanucleotide expansions are generally longer than the expansions associated with the triplet repeat diseases, with the largest DM2 and SCA10 repeats estimated to be ≥11,000 and 4500 repeats (41), respectively.

Repeat instability in DM2 is complicated by the compound repeat motif (TG)n(TCTG)n(CCTG)nand the time-dependent somatic instability of the expansion. Although similar somatic instability is seen in DM1 and FMR1 (42–45), the size differences for DM2 can be much larger, up to 9000 repeats in the blood of one affected individual. Clinical anticipation has been reported in DM2/PROMM families (46). Although we did not observe a correlation between age of onset and expansion size, the somatic instability of the repeat (Fig. 3 and Web table 1) (27) complicates this and other clinical correlations between repeat length and disease.

The clinical similarities between DM1 and DM2 have helped to clarify the extensive role that RNA containing a CUG or CCUG expansion plays in DM pathogenesis. Yet, although DM1 and DM2 phenotypes are strikingly similar, they are not identical. DM2 does not show a congenital form or the severe central nervous system involvement seen in DM1. Defining the downstream differences between the CUG and CCUG expansions will be important for understanding the clinical distinctions between DM1 and DM2. Also of interest will be the identification of the mutation that causes a third type of myotonic dystrophy that has not yet been mapped to a genetic locus (46).

  • * To whom correspondence should be addressed. E-mail: johnday{at} (J.W.D.); ranum001{at}


View Abstract

Navigate This Article