Inhibition of myogenesis in transgenic mice expressing the human DMPK 3′-UTR

Abstract

Myotonic dystrophy (DM1) is a multisystemic disorder caused by a CTG repeat expansion within the 3′-UTR of the DMPK gene. DM1 is characterized by delayed muscle development, muscle weakness and wasting, cardiac conduction abnormalities, cognitive defects and cataracts. Recent studies have demonstrated that the disease mechanism involves a dominant gain-of-function conferred upon mutant transcripts by expanded repeats. However, further attempts to model aspects of DM muscle pathology in cultured myoblasts suggest that 3′-UTR sequences flanking the CTG repeat tract are also required for full expression of the disease phenotype. Here, we report that overexpression of the DMPK 3′-UTR including either wild-type (11) or expanded (91) CTG repeats results in aberrant and delayed muscle development in fetal transgenic mice. In addition, transgenic animals with both expanded and wild-type CTG repeats display muscle atrophy at 3 months of age. Primary myoblast cultures from both 11 and 91 repeat mice display reduced fusion potential, but a greater reduction is observed in the 91 repeat cultures. Taken together, these data indicate that overexpression of the DMPK 3′-UTR interferes with normal muscle development in mice and that this is exacerbated by inclusion of a mutant repeat. This suggests that the delayed muscle development in DM1 involves an interplay between the expanded CTG repeat and adjacent 3′-UTR sequences.

INTRODUCTION

Myotonic dystrophy type I (DM1) is the most prevalent inherited neuromuscular disorder in adults, affecting ∼1/8000. It is a multisystemic disorder clinically divided into an adult and a more severe congenital form. Features of adult DM1 include myotonia and progressive muscular weakness and wasting, cardiac conduction defects, testicular atrophy, cataracts and premature balding (1). Congenital DM infants are hypotonic, have excessive numbers of muscle satellite cells and experience difficulties suckling due to poorly developed facial muscles (1). In addition, they are often mentally retarded.

The causative mutation for DM1 is a CTG trinucleotide repeat expansion in the 3′-untranslated region (UTR) of the myotonic dystrophy protein kinase (DMPK) gene (2–4). This repeat is polymorphic in normal individuals with alleles ranging from five to 37 in length. Repeats exceeding a threshold of ∼50–80 result in disease, with increasing severity and an earlier age-of-onset as the repeats lengthen. Repeats in adult patients range from roughly 100 to 1000 in length while in congenital DM they range from 1000 to 3000 repeats (2). Recently, the mutation responsible for a second form of DM (DM2) was identified as a CCTG tetranucleotide repeat within the first intron of the ZNF9 gene on chromosome 3. Expansions of the CCUG repeat in the most severe DM2 cases can be as large as 44 kb (5).

Although it is not entirely clear how CTG repeat expansions result in DM pathology, recent evidence has shown that an RNA-mediated dominant gain-of-function conferred upon mutant transcripts plays a major role. Overexpression of the DMPK 3′-UTR including an expanded CUG repeat (6) or repeat tracts alone (7) inhibits differentiation of the mouse myoblast cell line C2C12, which has been used to model the impaired muscle development seen in congenital DM1 cases. Mutant DMPK mRNA has been shown to be selectively retained in aberrant foci within the nuclei of patient cells, strongly suggesting that perturbed transcript processing underlies the dominant gain-of-function (8). As a potential mediator of this effect, a protein was characterized which binds to CUG repeat oligonucleotides in vitro and displays increasing affinity as the repeat sequence lengthens (9). This protein was identified as the human homolog (MBNL) of Drosophila muscleblind, which is important for muscle and eye differentiation in flies (10). Furthermore, MBNL and related isoforms were found to co-localize with mutant transcript foci in both DM1 and DM2 patient cell nuclei (11–13). Finally, insertion of 250 CUG repeats into the 3′-UTR of a foreign gene (mouse skeletal muscle actin) was sufficient to elicit specific features of DM muscle pathology in transgenic mice, notably myotonia (14).

Other experiments suggest that sequences within the DMPK 3′-UTR which flank the repeats may be important to the full expression of the disease phenotype. We found that overexpression of a wild-type 3′-UTR inhibits C2C12 differentiation and mapped this activity to the region 5′ of the repeat tract (15). In agreement, in a separate study the inhibition of differentiation by a mutant 3′-UTR also required the same region 5′ of the repeat (16). Also, transgenic mice overexpressing a wild-type DMPK gene displayed certain features of DM1, including type I fibre atrophy, increased numbers of central nuclei and reduced fusion potential of cultured myoblasts (17). To test this hypothesis, we overexpressed the DMPK 3′-UTR with either a wild-type (11) or mutant repeat sequence (91) in transgenic mice and examined the effects on muscle differentiation in vivo. We found that both strains of mice displayed defects in muscle differentiation at an early stage of development, including developmental abnormalities of face, jaw, back and neck musculature. In addition, young adult mice from both strains show type I and/or type II muscle fiber atrophy. Primary myoblasts obtained from young adult mice showed reduced fusion potential compared with cultures derived from control mice, but interestingly, this effect was more pronounced for 91 repeat mice compared with 11 repeat mice. This is the first demonstration of inhibition of myogenesis in vivo by overexpression of a wild-type DMPK 3′-UTR. This phenotype is exacerbated by the presence of a CUG repeat expansion and therefore suggests that cooperativity between the DMPK 3′-UTR sequences and the CUG repeat may be involved in the full expression of DM pathology in skeletal muscle.

RESULTS

Generation and characterization of transgenic mice overexpressing the human DMPK 3′-UTR

In vitro studies have demonstrated that overexpression of a normal DMPK 3′-UTR (15) or of mutant CTG repeats (6,7) inhibits the differentiation of C2C12 myoblasts. As these cells represent a faithful model of in vivo muscle differentiation, we investigated the possibility that overexpression of a normal (11 CTG repeats) or a mutant (91 CTG repeats) DMPK 3′-UTR in the muscle of transgenic mice could reproduce the observations seen in C2C12 myoblasts. Wild-type and mutant DMPK 3′ UTR sequences were fused downstream of the reporter gene GFP to facilitate detection of the gene product (Fig. 1). To direct expression of the transgene in muscle, we utilized a region derived from 5′-UTR of DMPK that was previously shown to have promoter activity in combination with a region from the first intron that had skeletal muscle specific enhancer activity (18); these sequences were fused upstream of GFP. The activities of these elements in the non-muscle tissues in which DMPK is normally expressed have not been investigated. Transgenic mice were generated and founders were identified by Southern analysis using a GFP probe (Fig. 1B).

To confirm that the DMPK regulatory elements employed conferred the appropriate expression patterns in vivo and to assess expression levels between different transgenic lines, northern blot analysis was performed using a human specific DMPK 3′-UTR probe (Fig. 2A). Total RNA was isolated from hind-limb muscle, heart and brain from five lines of transgenic mice containing 11 CUG repeats (CUG11; Fig. 2A and B and Table 1) and three lines with 91 CUG repeats (CUG91; Fig. 2B, Table 1 and data not shown) at 3–6 months of age. Transgene expression in all lines was highest in heart, followed by hind-limb muscle, while low amounts were detected in brain in some lines. This general expression pattern is consistent with that of the human DMPK gene (2,3). Transgene expression was generally higher in the CUG11 lines, with line CUG11–1913 being highest of the two CUG11 lines used in subsequent experiments (CUG11–1913, CUG11–1923: Fig. 2B and Table 1). Using quantitative real-time PCR analysis, we found that, of the three CUG91 founder lines generated, expression was comparable to that of the CUG11 lines in only one line (2043-1: Fig. 2B and Table 1). This was not unexpected as expanded CTG repeats have been shown to significantly decrease the expression of a reporter gene in cis in C2C12 (relative to wild-type repeats); these mutant constructs exerted a negative effect on differentiation despite their lower expression levels (6). Similar effects of repeat expansions on DMPK are also observed in patient myoblasts (19). Despite this difference in expression level, we used two CUG91 lines (2043-1 and 2038-3) in key experiments and noted similar results in both lines (Figs 5,6 and 8 and data not shown). We also analyzed the expression of GFP in tissue extracts by western blot. Tongue, hind-limb muscle, heart and brain (data not shown) were analyzed from CUG11 line 1923 and a wild-type control littermate (Fig. 2C). An extract from C2C12 myoblasts transfected with a GFP expression construct was included as a positive control. GFP was clearly detected in the tongue and the heart and faintly in the hind-limb muscle, in agreement with the RNA analysis.

In situ hybridization analysis of transgene expression

To determine the precise expression pattern of the transgene and identify potentially affected tissues, in situ hybridization was performed on transgenic mouse embryos aged 20 days post-coitum (pc). Antisense probes for GFP and myogenin were used to detect the transgene mRNA and to identify developing muscle, respectively (Fig. 3A–F). GFP and myogenin were detected in the tongue, snout and to a lesser extent in muscles of the neck and throat (Fig. 3A and B). GFP was also expressed in several regions of the brain including the frontal lobe, midbrain, hindbrain and choroid plexus and ependymal cells of the subventricular layer (arrowheads, Fig. 3A and B). In the torso area, expression of both genes was found in intercostal muscles, muscles of the back and the diaphragm (Fig. 3C and D). GFP was also expressed in the heart and was evident in the smooth muscle lining of pulmonary and amniotic blood vessels (arrows and arrowheads, Fig. 3C and D). Finally, in the gut area, only GFP was observed in smooth muscle lining the large intestines whereas both genes were expressed in skeletal muscle of the lower limbs (arrows and arrowheads, Fig. 3E and F). These data are consistent with both in situ hybridization data for the endogenous mouse DMPK gene (20) and the known expression profile in humans. Therefore, in addition to achieving expression in skeletal muscle, the regulatory elements utilized conferred accurate expression of the transgene in heart, smooth muscle and brain. However, no overt pathology was observed in transgenic embryos at this (late) developmental stage.

DMPK 3′-UTR overexpression disrupts early muscle development

We examined transgenic mice for effects of DMPK 3′-UTR overexpression at day 12 pc, a time that coincides with the onset of primary myogenesis (21). Embryos from CUG11 line 1923 were identified as transgenic by in situ hybridization with an antisense GFP probe (Fig. 4A, embryo 3) and serial sections from these and wild-type littermates were stained using a probe for myogenin as a marker for developing muscle (22). Two wild-type and one transgenic embryo are shown (Fig. 4). Myogenin staining was clearly observed in the somites in all embryos (small arrows, Fig. 4B) and in developing muscles of the neck region in wild-type embryos (small arrowhead, embryos 1 and 2, Fig. 4B). In contrast, there was a striking lack of staining in this region of the neck in the CUG11 embryo (small arrowhead, embryo 3, Fig. 4B). GFP-positive tissue can be seen in this region (extending dorsal and ventral from the large arrowhead, embryo 3, Fig. 4A) but expresses little myogenin (large arrow and arrowhead, embryo 3, Fig. 4B). This suggests that pre-muscle tissue is present but has not fused to form myotubes. GFP-positive tissue can also be seen in a more ventral position (large arrow, embryo 3, Fig. 4A), in the area of developing jaw muscles. Several corresponding myogenin-positive structures are present in wild-type embryos (for example, black arrowheads in embryos 1 and 2, Fig. 4B), but only one can be seen in the transgenic embryo (large arrow, embryo 3, Fig. 4B). These data indicate that overexpression of the DMPK 3′-UTR disrupts primary myogenesis in the developing mouse embryo, particularly in future facial and neck areas.

DMPK 3′-UTR overexpression delays secondary muscle development

The above data suggests that overexpression of a wild-type DMPK 3′-UTR can delay normal muscle development in vivo, as previously speculated (15,17). To determine whether overexpression of a mutant DMPK 3′-UTR resulted in any additional pathology, we performed immunohistological analysis by staining CUG11, CUG91 and wild-type animals for myogenin at day 15 pc, a time which marks the boundary between primary and secondary myogenesis. In back, neck and tongue musculature, a lack of fully developed myotube structures was observed in transgenic embryos from CUG11 (line 1913; Fig. 5B, D and F) compared with a wild-type littermate (Fig. 5A, C and E). This phenotype was also observed in the other CUG11 line (1923) as well as both CUG91 lines examined (2038-3 and 2043-1; data not shown). Transgenic embryos had either disorganized, incompletely formed myotubes containing only one or a few myonuclei or many mononucleated myogenic cells, instead of large multinucleated myotubes (arrowheads, Fig. 5D and F). Immunostaining of transgenic CUG91 (2038-3) E15 embryos for total sarcomeric myosin protein revealed, like myogenin staining, poorly developed myotube structures (Fig. 6). Myotubes in the developing back and neck musculature of the transgenic animals were very thin and underdeveloped (Fig. 6B, D), while no myotubes were evident in the jaw musculature of the transgenic embryo (compare Fig. 6E and F). In addition, it was clearly evident that sarcomeric myosin protein expression was considerably lower in the developing muscle of the transgenic embryo compared with the wild-type, indicating reduced myogenic differentiation (Fig. 6A–F). Again, this phenotype was observed in all other lines analyzed (1913, 1923 and 2043-1).

A differentiation defect seen in primary myoblasts from transgenic animals is exacerbated by mutant CTG repeats

In order to quantitate the difference in myotube development, primary myoblast cell cultures were established from one CUG11 line (1923), one CUG91 line (2043-1) and from wild-type mice and differentiated in vitro by growth factor withdrawal. Myotubes were poorly developed in both CUG11 (not shown) and CUG91 cultures compared with wild-type (Fig. 7A and B). Fusion indices were determined (see Materials and Methods) and a clear reduction was seen in both transgenic cultures compared with wild-type (Fig. 7C). Interestingly, CUG91 cultures exhibited an ∼50% lower fusion index than CUG11 cultures. These results are in agreement with both the poor development of muscle seen in vivo in these mice and with a study demonstrating inhibition of myoblast fusion as a result of 3′-UTR expression in C2C12 myoblasts (16).

Muscle fibre atrophy in young transgenic mice

Finally, we closely examined muscle from mice at various times after birth. A histological analysis of muscle at 6 months revealed no overt changes in CUG11 and CUG91 mice, consistent with the overtly normal appearance of day 20 embryos. As type I muscle fiber atrophy is a hallmark feature of DM, we determined the mean fiber area in soleus muscle from younger transgenic and wild-type animals. Cross sections were stained for type I MHC to distinguish type I and type II fibers and the cross sectional areas were plotted (Fig. 8A and B). In 1 to 3-month-old CUG11 mice, significant type I fiber atrophy was observed in line 1913 (P<0.05, Student's t-test) compared with wild-type littermates, whereas line 1923 showed significant type I and II fiber atrophy (Fig. 8B). The CUG91 mice analyzed also displayed significant type I and II fiber atrophy (Fig. 8B). We are uncertain whether the unexpected type II atrophy is reflective of a more pronounced defect than that normally seen in humans, of aberrant expression of the transgene at the fiber level, or of some difference between mouse and human physiology. In older mice (6 months and 1 year of age) muscle hypertrophy, but never atrophy, was sometimes observed. Taken together, our data demonstrate that the effects of DMPK 3′-UTR overexpression are pronounced early in development but diminish with age.

DISCUSSION

Considerable evidence now supports the hypothesis that the dominant nature of the DM mutation is exerted at the RNA level. In particular, expression of 250 CTG repeats placed within the 3′-UTR of an α-actin transgene causes myotonia and other histological features of DM in mouse skeletal muscle (14). However, a number of observations suggest that expression of expanded repeats alone does not fully account for disease. Weakness and wasting of skeletal muscle was not observed in the α-actin/mutant repeat mice, suggesting that the myopathy observed was incomplete. In studies with C2C12 myoblasts, we and others have found that expression of the DMPK 3′-UTR inhibits differentiation (6,15,16). While these studies have disagreed on the importance of a mutant repeat, they have agreed that the region 5′ of the repeat within the 3′-UTR is necessary. Interestingly, transcripts with repeats alone (i.e. removed from their normal 3′-UTR context) aggregate in nuclear foci, yet no longer inhibit differentiation, demonstrating for the first time that the phenomena are separable (16). Finally, it is noteworthy that DM2 does not have a severe congenital form despite the fact that MBNL (and related isoforms) are likewise sequestered by mutant ZNF9 transcripts. While this might simply be due to intrinsic properties of CCUG versus CUG repeat sequences, it is equally possible that the sequence context for expanded repeats plays a role. Together, these studies suggest an interplay between repeat expansions and some normal function of the 3′-UTR, perhaps related to development or differentiation.

We chose to study the effects of overexpression of the human DMPK 3′-UTR using an endogenous promoter/enhancer system (18), in an attempt to most closely mimic the normal timing and expression patterns for DMPK. It was hoped that this approach would avoid possible artefacts due to very high transgene levels and eventually extend the tissues available for study. As noted, transgene expression levels in the heart, skeletal muscle and brain generally paralleled those of the endogenous gene. Transgene expression was also seen in ventricular epithelium and various smooth muscle compartments, also sites of DMPK expression (20,23). Thus, our construct contains the regulatory elements necessary for faithful expression in several tissues and therefore represents an excellent system for the extension of our studies into the heart, brain and smooth muscle.

Skeletal muscle in our mice appeared to be normal at later stages in development but we observed obvious defects at the time at which primary myogenesis occurs (E12–E15) (21). Muscle groups in the neck and face that normally strongly express myogenin were conspicuously unstained, while the somites were normal. At this time, we cannot be certain whether poorly differentiated precursor tissue is present at the correct position or whether muscle groups are ectopically localized, but favor the former explanation. In any case, this observation is particularly intriguing as these muscle groups display the most consistent involvement in DM. At the onset of secondary myogenesis, staining for myogenin and sarcomeric actin again indicated that back, neck and tongue muscle remain poorly developed in both CUG11 and CUG91 transgenic animals, indicating that the delay in myogenesis continues through this stage.

Although the myogenic phenotype observed in CUG11 and CUG91 transgenic embryos was similar, differentiated myoblasts cultured from CUG91 mice had the lowest fusion index compared with CUG11 and wild-type primary cultures. Clearly, the presence of a CUG expansion further impaired the fusion of primary myoblasts, indicating cooperation between the CUG repeat and other 3′-UTR sequences, as previously suggested (16). This exacerbation of the myogenic phenotype occurs despite the typically lower expression of the CUG91 transgene (Table 1).

It should be noted that, although very high levels of GFP have been reported to cause cellular pathology in some cases (24), GFP expression was consistently much higher in the heart of our mice than in other tissues and yet we observed no developmental defects here (data not shown). More importantly, as mentioned, the myoblast fusion deficit was more severe in CUG91 lines despite the fact that they typically expressed less transgene than the CUG11 lines. These results strongly argue against a non-specific toxicity in transgenic mouse skeletal muscle.

While the effects of the transgene were primarily developmental, post-natal defects in muscle composition were also seen. Type I and sometimes type II fiber atrophy was seen in the soleus muscle of 3-month-old animals from both CUG11 and CUG91 transgenic mice, generally consistent with our previous study of DMPK overexpression (17). DMPK 3′-UTR expression in adult muscle may cause mild muscle degeneration or, alternatively, muscle maturation in younger transgenic mice may continue to be delayed as a result of the delay during development. Obviously, we favor the latter explanation as the fiber atrophy does not persist after 6 months. Type II fiber atrophy was not expected because type I fiber atrophy is predominately observed in DM (1). The type II fiber atrophy may be due to differences between mouse and human physiology or may be the result of ectopic transgene overexpression in type II fibers.

It is not clear at this point how the DMPK 3′-UTR inhibits muscle development, but we suspect that it may normally have a negative role during differentiation and that overexpression increases this function to a deleterious state. A role for the 3′-UTR's of structural genes such as tropomyosin during muscle differentiation has been elucidated (25), but in this case these sequences actually promote differentiation by suppressing proliferation. In any case, according to our proposal, mutant repeats may also enhance the negative impact of the DMPK 3′-UTR in a specific fashion, which could only be observed in primary myoblast cultures in this study. One possibility is that sites for RNA-binding proteins exist in the portion of the 3′-UTR upstream of the repeat and that both overexpression and repeat expansion increase factor binding, such that they become depleted.

Congenital myotonic dystrophy presents with a distinct set of symptoms, notably a profound delay in muscle maturation, while other pathological features seen in adult muscle (e.g. sarcoplasmic masses, centronucleation) are not observed. Infants that survive this phase often have improved status and go on to develop symptoms of the adult disease later (1). Therefore, very large repeats have additional effects that exert themselves at an early developmental phase and may not be simply extensions of the adult disease. As mentioned, MBNL sequestration probably also plays a role in DM skeletal muscle pathology. Therefore, factors that bind upstream of the repeat sequence must operate in addition to the role proposed for MBNL. It is possible that these unknown factors interact with MBNL. Further studies that delineate the possible roles of MBNL during human muscle development and in adult muscle will shed further light in this regard. Finally, recent studies have shown that myotonia in DM is a consequence of mis-splicing of the ClC-1 muscle chloride channel pre-mRNA, ultimately resulting in loss of the protein from the muscle membrane (26,27). Myotonia is a symptom of adults and thus the associated splicing defect may not be directly related to the congenital developmental defect. As mentioned, the α-actin/mutant repeat mice did not exhibit muscle weakness and wasting (26). Taken together, these observations raise the possibility that repeat expansion may cause various symptoms of DM through multiple mechanisms.

In conclusion, we demonstrate for the first time in vivo that overexpression of the human DMPK 3′ UTR utilizing DMPK regulatory elements results in myogenic defects. Aspects of the myogenesis phenotype are exacerbated by inclusion of a CUG repeat expansion within the 3′-UTR. Therefore, the molecular basis of the myogenic defect in DM may involve an interplay between the CUG repeat expansion and adjacent 3′-UTR sequences. Elucidation of these molecular interactions will be critical to further our understanding of this disease.

MATERIALS AND METHODS

Transgene construction and generation of transgenic mice

A 1.8 kb EcoRV/SmaI fragment (GenBank L08835-nt 291–2139) with promoter activity from the DMPK 5′-UTR (19) was ligated into pBluescript (SK-). A 2.1 kb fragment from the DMPK first intron (nt 2362–4470) with muscle-specific enhancer activity was cloned into a NotI site downstream of the DMPK promoter. To omit a first intron ATG codon and not exclude the end of the first intron, a 261 bp fragment (nt 4474–4735) was ligated into an SpeI site downstream of the enhancer. A fragment encoding GFP was excised from pEGFP-N1 (Clontech) using BamHI and NotI (blunted with Klenow) and cloned into BamHI/EcoRI (blunt) sites downstream of the DMPK promoter/enhancer. DMPK 3′-UTR fragments were PCR amplified from normal (11 CTG repeats) or patient DNA (91 CTG repeats) cloned with BamHI (blunt) and HindIII (nt 12129–13408) and ligated into EcoRV/HindIII sites downstream of GFP. The final 3′-UTR fragment (nt 13408–13747) was PCR-amplified and cloned into the HindIII site to give the transgene constructs. Both transgenes including CTG repeats were verified by sequencing and restriction digests. Transgenes were excised from pBluescript and 2 ng was microinjected into oocytes obtained from the mating of CBA/C57BL6 F₁ mice. Identified founder transgenic mice were crossed onto a C57BL6 background.

Nucleic acid isolation and analysis

Mouse litters were weaned at 3 weeks of age at which time pups were ear-tagged and tail-clipped. DNA was isolated from mouse tails using the DNeasy DNA isolation kit (Qiagen) according to the manufacturer's instructions. DNA was digested with BamHI and electrophoresed on a 1% agarose gel. Southern blotting was performed as described (2). RNA was isolated from animal tissues using Trizol reagent (Gibco-BRL) according to manufacturer's instructions. Northern blotting was performed as described (18).

Real-time quantitative RT–PCR

GFP mRNA (isolated from skeletal muscle, brain and heart) was measured using real-time quantitative RT–PCR as per the Taqman method (6,28). The GFP forward and reverse primers (forward 5′ CATGGTCCTGCTGGAGTTC, reverse 5′ TTACTTGTACAGCTCGTCCATGC) and fluorescent probe (probe 5′ [6∼FAM]CCGCCGCCGGGATCACTCTC[Tamra∼Q]), were designed to amplify a 68 bp region of GFP. Total RNA was isolated as described above and further purified using Rneasy mini-spin columns combined with DNase I treatment (Qiagen). For each sample, 5 ng of total RNA were used for the analysis. The RNA was reverse-transcribed and PCR amplified using the Taqman EZ RT–PCR kit (Applied Biosystems) on an ABI Prism 7700 Sequence Detector (PE Applied Biosystems). Results were quantified using the cycle threshold (Ct) method and normalized to GAPDH mRNA levels, determined using PE-ABI supplied mouse primers and probe (JOE or VIC-labeled). GFP primers and probes were used at concentrations of 600 and 200 nM, respectively. The RT reaction was carried out at 50°C for 2 min, 60°C for 30 min, and 95°C for 5 min and was followed by 45 cycles of PCR at 94°C for 20 s and 60°C for 1 min.

Protein lysates and western blotting

Protein lysates from tissues were obtained by homogenization on ice with a polytron in 10 mM Tris pH 8.0, 150 mM NaCl, 2 mM MgCl₂, 1 mM PMSF. Samples were spun for 5 min at 500 rpm in a microfuge; supernatants were adjusted to 2% SDS and boiled for 20 min. Lysates from cultured cells were prepared in RIPA buffer (150 mM NaCl, 1% NP40, 0.5% deoxycholate, 0.1% SDS, 50 mM Tris–HCl pH 7.5, 100 µg/ml PMSF, 2 µg/ml aprotinin, 1 µg/ml pepstatin A, 2 µg/ml antipain). Protein concentrations were determined using a Pierce micro BCA kit according to the manufacturer's instructions. A 30–50 µg sample of lysate was mixed with 3× loading buffer (187.5 mM Tris–HCl pH 6.8, 6% SDS, 30% glycerol, 0.03% bromphenol blue) and boiled for 5 min. Samples were electrophoresed on a 10% SDS–polyacrylamide gel and transferred to PVDF membrane by semi–dry electroblotting. The membrane was incubated with a polyclonal GFP antibody (Clontech; 1 : 1000 dilution), washed in PBST and visualized by chemiluminescence (ECL kit, Amersham) according to the manufacturer's instructions.

³³P in situ hybridization

Riboprobes encoding GFP and myogenin were generated by run-off transcription using a MAXIscript kit (Ambion) supplemented with ³³P-rUTP according to the manufacturer's instructions. A 396 bp fragment of the GFP cDNA (nt 126–522) was PCR-amplified and cloned into pCR 2.1 (linearized with BamHI) and the most 3′ 449 bp (EcoRI fragment) of the myogenin cDNA also cloned into pCR2.1 (linearized with BamHI) were used as templates. Embryos were obtained from pregnant C57 Bl6 mice. The day the vaginal plug appeared was designated E0. Pregnant mice were euthanized by CO₂ and embryos removed at various times between E9.5 and E20. Subsequent procedures were performed as described (22).

Immunostaining of cryosectioned embryos and muscle

Embryos were flash frozen in dry-ice cooled isopentane and sectioned on a cryostat. 12 µm cryosections were dipped in H₂O for 1 min and then fixed in 4% PFA in PBS (8 mM Na₂HPO₄, 2.2 mM NaH₂PO₄, 137 mM NaCl, 2.7 mM KCl) for 3 min at room temperature. All antibody incubations and washes were performed in PBS. Slides were washed three times for 3 min with shaking. Specimens were blocked in 5% serum in PBS at room temperature for 15 min. Sections were incubated with primary antibody solution containing 0.3% Triton X-100 overnight at 4°C, washed three times for 3 min with shaking and incubated with secondary antibody solution (1 : 200 dilution) containing 0.3% Triton X-100. Slides were washed three times for 3 min with shaking and mounted in antifade solution (Dako).

Primary myoblast culture and fusion assay

Primary myoblast cultures were isolated essentially as described (29). Myoblasts were grown to a density of 1×10⁴ cells in 24-well collagen coated dishes. Cells were then induced to differentiate in DMEM medium containing 5% FBS for 5 days. Cells were fixed in 4% PFA for 10 min and stained for MHC-I with MF20 hybridoma surpernatant (1 : 5 dilution). Hoechst dye (1 µl of 1 mg/ml) was used to visualize nuclei. Fusion index was calculated by tabulating the percentage of total nuclei located within myotube structures. Ten fields were counted, ∼400 nuclei per field. The values shown represent the average fusion index of the 10 fields. Two separate primary myoblast isolations were performed on one CUG11 and CUG91 mouse and corresponding littermate controls, shown is a representative experiment.

Muscle sectioning and fiber size determination

Mouse hindlimb muscle was hemisectioned such that the soleus was bisected at the midline. Muscles were submersed in liquid nitrogen cooled isopentane for ∼30 s, sectioned and adhered to precleaned Superfrost-Plus slides (VWR). Sections were stained for neuromuscular junctions with α-bungarotoxin–CY3 (1 : 500; Molecular Probes) as described to ensure sections were from the middle third of the soleus muscle. Sections were stained with monoclonal antibody BAD5 (1 : 5 dilution) specific for type I myosin heavy chain (MHC-I) as described above except that the secondary antibody was conjugated to horseradish peroxidase. The peroxidase reaction was visualized with 3, 3′-diaminobenzidine tetrahydrochloride (DAB) substrate. Images of type I (DAB positive) and type II (DAB negative) fibres from transgenic and wild-type mice were traced manually and cross section areas determined using the ScionImage program (Scion corporation). Cross sectional areas were determined for one transgenic and wild-type littermate for two lines of CUG11 (1913, 1923) and two lines of CUG91 (2038-3, 2043-1) mice. For type II fibers, 50 cross section areas were included in the calculations and 100 for type I fibers.

ACKNOWLEDGEMENTS

Thanks to Catherine Neville and Jin-Ying Xuan for sequencing. We are grateful to Martine St Jean and Aegera Therapeutics for technical assistance. C.J.S. was supported by the Aurthur Minden pre-doctoral fellowship from the Muscular Dystrophy Association of Canada (MDAC). This work was supported by grants from the MDA (USA), the Canadian Institutes of Health Research (CIHR) and the Canadian Genetic Diseases Network (CGDN). R.G.K. is a CIHR Senior Investigator and a Howard Hughes Medical Institute (HHMI) International Research Scholar.

Figure 1. DMPK CUG11 and CUG91 transgenes and identification of transgenic mice by Southern blot analysis. (A) The transgene used in the construction of CUG11 and CUG91 transgenic mice is shown. The BamHI sites used in Southern blot analysis are indicated and the location of the probe used for screening Southern blots for transgene positive animals. (B) Southern blots were probed with the GFP cDNA. Five 91 CUG repeat transgenic mice and one 11 CUG repeat transgenic animal are shown, highlighting the mobility difference between the two transgenes.

Figure 2. Expression analysis of mRNA and protein from various tissues of several lines of CUG11 transgenic mice. (A) Northern blot analysis was performed on total RNA from skeletal muscle, heart and brain of five CUG11 transgenic mouse lines (1912, 1913, 1921, 1923 and 1927). The northern blot was probed with the 3′ end of the DMPK 3′ UTR. To control for RNA loading, the blot was probed with the Pgk cDNA. (B) Quantitative real-time PCR expression analysis of CUG11 and CUG91 transgenic mice. Total RNA from heart, muscle and brain of two CUG11 and two CUG91 transgenic mouse lines were analyzed. Tabulated transgene expression levels were normalized to GAPDH expression levels. Shown are relative expression profiles for CUG11 line 1923 and CUG91 line 2043-1. CUG91 line 2043-1 expressed less transgene mRNA than CUG11 line 1923 in heart and muscle but more in brain. Experiment was performed three times in triplicate, twice with one set of primers and once with a second set of primers yielding near identical results. Shown is a representative experiment. (C) Western blot analysis using a polyclonal GFP antibody was performed on tissue lystates from CUG11 transgenic mouse and wild-type littermates. Lanes 1 and 2 are protein lysates from C2C12 cells transfected with a GFP expression vector or mock transfected. Protein lysates from tongue, skeletal muscle and heart from transgenic (lanes 3, 5 and 7) and wild-type littermates (lanes 4, 6 and 8) were analyzed. Expression of CUG11 mRNA and protein is highest in cardiac tissue and is present but at lower levels in tongue and skeletal muscle.

Figure 3. Expression of the CUG11 transgene (GFP) in E20 embryos. In situ hybridization was performed on transgenic (line 1923) E20 embryo sections using antisense probes for myogenin (A, C, E) and GFP (B, D, F). (A, B) Expression of myogenin and the GFP is clearly observed in tongue (T), muscles of the head and neck area (M), cell linings of the ventricles and within regions of the forebrain, midbrain and hindbrain (arrowheads) (B). Myogenin and GFP are present in intercostal muscles (m), muscles of the back and neck (m) and in the diaphragm (d) (C, D). Expression of GFP, but not myogenin was observed in smooth muscle cells lining the blood vessels (bv) and cardiac cells of the heart (h). Expression of both myogenin and GFP is observed in the muscles of the lower limb (m) (E, F). Only GFP is expressed in the smooth muscle cell (smc) lining of the large intestines and amniotic blood vessels (arrows, arrowheads) (F). GFP is expressed in skeletal muscle, heart, smooth muscle and brain at E20. This pattern of expression indicates that regulatory elements required for tissue specific expression of DMPK were included in the transgene regulatory elements.

Figure 4. Expression of the CUG11 mRNA during early mouse embryogenesis. (A) In situ hybridization was performed on embryo (12 days pc) sections using an antisense GFP probe. The presence of GFP transcripts indicates embryo 3 is transgenic. GFP is expressed in skeletal muscle precursor cells of the somite (small arrows) and in muscles of the future neck, face and jaw (small arrowheads). (B) Primary myogenesis was detected in vivo using an antisense probe for myogenin, which is expressed in developing skeletal muscle. GFP is expressed in a similar pattern to myogenin confirming in vivo muscle specificity of the DMPK regulatory elements employed. The white arrowhead in embryo 3 indicates neck musculature that does not express myogenin transcripts (compare small white arrowhead in embryo 2 and 3). Small white arrows mark myogenin expression in somites. The black arrowheads in embryos 1 and 2 indicate expression of myogenin in facial muscles of wild-type embryos while the white arrow in embryo 3 indicates poor myogenin staining of facial muscle in the transgenic embryo. Transgenic embryos display a muscle development phenotype during primary myogenesis, particularly in developing neck and facial muscle.

Figure 5. Inhibition of muscle differentiation in E15 embryos from CUG11 and CUG91 mice. Wild-type and transgenic embryo littermates from two lines of CUG11 (1913, 1923) and CUG91 (2038-3, 2043-1) were sectioned and stained with anti-myogenin antibody F5D. Shown are muscles of the back (A, B), neck (C, D) and tongue (E, F) of CUG11 (1913) mouse and a wild-type control. An identical phenotype was observed in embryos from both (2038-3, 2043-1) CUG91 lines and CUG11 line 1923. Formation of myotubes in vivo is inhibited by expression of the DMPK 3′-UTR with 11 or 91 CUG repeats in developing muscle.

Figure 6. Impairment of muscle differentiation and reduction of total myosin levels in E15 embryos from CUG11 and CUG91 mice. Wild-type and transgenic embryos from two lines of CUG11 (1913, 1923) and CUG91 (2038-3, 2043-1) and wild-type littermates were sectioned and stained with an antibody against total sarcomeric myosin (MF20). Shown are muscles of the back (A, B), neck (C, D) and jaw (E, F) of CUG91 (line 2038-3) mouse and a wild-type control. An identical phenotype was observed with embryos from CUG 91 line 2043-1 and CUG11 lines 1913 and 1923. Myotube formation and expression of total sarcomeric myosin is inhibited in vivo as a result of expression of the DMPK 3′ UTR with 11 or 91 CUG repeats in developing muscle.

Figure 7. Inhibition of muscle differentiation in myoblasts obtained from CUG11 and CUG91 mice. (A, B) Primary myoblasts were obtained from CUG11 and 91 mice and wild-type littermates, induced to differentiate for 5 days then stained for total sarcomeric myosin (MF20) and DAPI. Shown is a typical field of (A) wild-type and (B) CUG91 myotube cultures. (C) Degree of in vitro differentiation (fusion index) was calculated (see Materials and Methods). Myoblasts cultured from both CUG11 and CUG91 transgenic mice exhibited reduced differentiation potential compared with wild-type myoblasts, although the effect was more pronounced in CUG91 myoblasts. Experiment was performed twice from separate myoblast isolations, shown is a representative experiment.

Figure 8. Determination of type I and II fibre sizes from cross-sectioned soleus muscle of CUG11 and 91 transgenic mice. Soleus muscle was cross-sectioned through the middle third of the muscle. This was confirmed by α-bungarotoxin staining to detect neuromuscular junctions (not shown). Mean type I and II fibre areas for two lines of CUG11 (A) and CUG91 (B) transgenic mice are shown. Both CUG11 lines exhibit significant type I atrophy in the transgenic sample but only line 1923 had type II atrophy. Both CUG91 lines analyzed showed significant type I and II atrophy in the transgenic sample. Significance is indicated with asterisks (P<0.05, type 1 fibres, n=100, type II fibres, n=50).

Table 1.

Relative transgene expression in muscle, brain and heart of CUG11 and CUG91 lines assessed by quantitative RT–PCR

Line	Muscle	Brain	Heart
CUG 11–1913	1.85±0.33	13.58±0.25	43.16±4.44
CUG 11–1923	1±0.01	0.09±0.01	19.66±0.70
CUG 91–2038-3	0.02±0.00	0.01±0.00	0.51±0.06
CUG 91–2043-1	0.81±0.09	0.37±0.02	11.66±0.96