Introduction

Myotonic dystrophy type I (DM1) is a multi-system disorder occurring with an incidence of 1:8000 worldwide. Three broad forms of DM1 are described, a classic form which has an onset between 10 and 60 years, presenting with myotonia, muscle weakness, smooth and cardiac muscle involvement, CNS dysfunction, somnolence, endocrine disorders and cataracts, a congenital form which is symptomatic at birth and which progresses to manifest many of the symptoms of the classic form of the disease and a minimal form that has its onset after 50 years manifesting with cataracts, myotonia and mild muscle weakness1,2. Heart disease is a well-established feature of the classic and congenital forms of the disease, with cardiac sudden death a prominent cause of mortality1,2. Conduction disorders predominate in DM1. First degree heart block or PR prolongation is commonly observed in DM1 and with increasing time or enhanced disease severity, second and third degree heart block, QRS, QTc widening and bundle blocks increase in frequency1,2,3,4,5. Poor R wave progression is also a feature noted in DM16. Left ventricular hypertrophy, dilation and systole dysfunction occur less frequently and can manifest in the absence conduction disorders1,2,7,8,9,10. Histopathological changes include multi focal myofibrillar loss, fatty infiltration and fibrosis1,2,6,10. A combination of these features is widely believed to contribute to sudden cardiac death9,10,11. Of the electrocardiograph measurements the sum of the QRS duration and PR interval has been shown to be a strong predictor of mortality in DM112.

The genetic mutation in DM1 is the expansion of a CTG repeat sequence located in the 3′ untranslated region of DMPK and immediately 5′ of SIX5 on chromosome 19q13.313,14. In general, larger CTG expansions are associated with a higher incidence and progression of conduction disease and a trend towards an increase in serious cardiac events15. Several lines of evidence demonstrate that expression of expanded CUG repeat sequences plays an important role in the development of key aspects of DM1 pathology16,17,18,19. Toxicity associated with expanded CUG repeats, stems in part from its ability to aberrantly sequester and disable the muscleblind-like (MBNL) family of RNA splice regulators20,21,22. In humans, MBNL1 is highly expressed in skeletal muscle and the heart and shows lower expression in the brain, lung, liver, kidney and pancreas21. In mice, although Mbnl1 expression is more consistent across tissues, the highest expression of Mbnl1 is observed in the heart23. Previous experiments have shown that Mbnl1 loss in mice results in skeletal muscle myotonia and histopathology characteristic of DM1, in conjunction with cataracts and behavioral alterations that are reminiscent of DM1 patients24,25. Mbnl1 is known to play an important role in regulating the transition of its target RNAs from the embryonic splice program to that of the adult26,27. Significantly, previous studies have implicated the persistence of embryonic Clcn1 splice isoforms in adult muscle with the development of myotonia both in Mbnl1 deficient mice and in the HSALR DM1 mouse model, where expanded CUG repeat RNA is expressed in skeletal muscle24,28. As morpholino antisense oligonucleotide targeting prevents aberrant Clcn1 splicing and reverses myotonia in the HSALR mice, a causal relationship is established between abnormal Clcn1 splicing and myotonia29. Taken together these experiments highlight the potential of physiologically relevant Mbnl1 target RNAs that are aberrantly spliced to initiate DM1 pathology.

To test the role of Mbnl1 depletion in the development of DM1 cardiac pathology, we deleted Mbnl1 exon 2 (Mbnl1ΔE2/ΔE2), which encodes the ATG codon, in 129 sv mice. This mutation results in the absence of the Mbnl1 protein. Mbnl1ΔE2/ΔE2 mice show a shortened life span in conjunction with a variety of conduction defects, cardiac hypertrophy, fibrosis and multi-focal myocardial fiber death and calcification. Mbnl1 loss results in the enhanced expression of embryonic splice isoforms in an RNA network which regulates sodium and calcium currents, intra and inter cellular transport, cell survival, sarcomere and cytoskeleton organization and function and encoding structural components of the sarcomere. These results therefore support an important role for Mbnl1 depletion in the development of DM1 cardiac disease and suggest a role for altered splicing in initiating cardiac pathology.

Results

Development of Mbnl1ΔE2/ΔE2 mice

To test the role of MBNL1 deficiency in the development of DM1 cardiac disease we developed Mbnl1loxE2lox mice in which Mbnl1 exon 2 was flanked by lox sites (Fig. 1a i–iii). Southern blot analysis of targeted 129 sv ES cells is shown in Fig. 1b. Chimeric animals derived from targeted 129 sv ES cells were bred to 129 sv wild type animals to derive Mbnl1loxE2lox mice (Fig. 1a iii). Lox mediated deletion of Mbnl1 exon 2 was achieved by crossing Mbnl1loxE2lox mice with 129 sv transgenic mice expressing the Cre recombinase under the control of the protamine 1 promoter30. As the protamine 1 promoter drives expression of the Cre recombinase only in the male germ line, a cross between male Mbnl1loxE2lox/Cre mice and 129 sv wild-type females resulted in Mbnl1+/ΔE2 mice, which were subsequently used to obtain Mbnl1ΔE2/ΔE2 animals using standard breeding schemes (Fig. 1a iv). Deletion of Mbnl1 exon 2 was established by RT-PCR analyses (Fig. 1c). Loss of Mbnl1 protein in Mbnl1ΔE2/ΔE2 mice is shown by western blot analysis (Fig. 1d). Depletion of Mbnl1 resulted in ~2.5 fold increase in the steady-state levels of Mbnl2, a splice regulator, which is homologous to Mbnl1 (Fig. 1d). Analysis of genotype ratios of the progeny of male and female 129 sv Mbnl1+/ΔE2 mice did not reveal a homozygous mutant lethal phenotype.

Figure 1
figure 1

Mbnl1ΔE2/ΔE2 mice demonstrate a reduced life-span.

(a). The wild type (Mbnl1+/+) allele (i), targeting construct (ii), Mbnl1loxE2lox allele (iii) and Mbnl1ΔE2 allele (iv) are shown. (b). Southern blot analysis of wild-type and targeted 129 sv ES cell DNA digested with EcoRV and analyzed with a probe indicted in Panel a. The image shown is cropped. The full-length gel is shown in Supplementary figure S1a. (c). RT-PCR analysis of Mbnl1+/+ and Mbnl1ΔE2/ΔE2 heart RNA using primers located in Mbnl1 exon 2. Gapdh was amplified in parallel as an internal control. The images shown are cropped and full-length gels are shown in Supplementary figures S1b and c. (d). Western blot analysis of Mbnl1+/+ and Mbnl1ΔE2/ΔE2 heart protein lysates using anti-Mbnl1, anti-Mbnl2 and anti-α-tubulin antibodies. The images shown are cropped and the full-length gels are shown in supplementary figures S1d–f. (e). Kaplan-Meyer survival curves are shown. A total of 51 Mbnl1ΔE2/ΔE2 mice (22 males, 29 females) and a corresponding number of male and female wild-type 129 sv mice were used in this study. Sacrificed mice are indicated by vertical bars. There is a statistically significant difference in the survival curves between Mbnl1ΔE2/ΔE2 and Mbnl1+/+ mice (χ2(1) = 85.7, p < 0.00001). The survival curves between Mbnl1ΔE2/ΔE2 males and Mbnl1ΔE2/ΔE2 females is not statistically significant (χ2(1) = 2.8, p < 0.10). (f, g). The recorded causes of death for the male (f) and female (g) Mbnl1ΔE2/ΔE2 mice are indicated.

Full size image

Mbnl1ΔE2/ΔE2 mice show a short life span

Kaplan-Meyer curves were developed to examine differences in survival between wild-type (Mbnl1+/+) and Mbnl1ΔE2/ΔE2 animals. For this analysis 102 gender matched mice [Mbnl1+/+:51 and Mbnl1ΔE2/ΔE2: 51; for each genotype female n = 29 and male n = 22] were followed for up to 14.9 months (Fig. 1e–g). There was a statistically significant difference in the survival curves between Mbnl1ΔE2/ΔE2 and Mbnl1+/+ mice (χ2(1) = 85.7, p < 0.00001). Median survival time was 6.5 months for Mbnl1ΔE2/ΔE2 males (95% CI 6.48–6.59 months) and 4.8 months for Mbnl1ΔE2/ΔE2 females (95% CI 4.05–5.62 months). The survival curves between Mbnl1ΔE2/ΔE2 males and Mbnl1ΔE2/ΔE2 females was not statistically significant (χ2(1) = 2.8, p < 0.10). Death, where no end point symptoms or illness are overt, was observed in ~67% and ~86% of male and female Mbnl1ΔE2/ΔE2 mice respectively. Rectal prolapse (male Mbnl1ΔE2/ΔE2: ~17% and female Mbnl1ΔE2/ΔE2: ~10%) or morbidity from an unknown etiology, where a hunched posture, weight loss and lethargy were observed prior to death (male Mbnl1ΔE2/ΔE2: ~17% and female Mbnl1ΔE2/ΔE2: ~5%) were the other prominent causes of mortality (Fig. 1f & g). No significant difference was observed between the body weights of male and female cohorts of Mbnl1+/+ and Mbnl1ΔE2/ΔE2 mice at 3–4 and 5–7 months of age (Supplementary Fig. S2).

Mbnl1ΔE2/ΔE2mice show QRS, QTc widening, bundle block and STc shortening at 2 and 4 months of age

As cardiac conduction defects are a prominent feature in DM1, surface electrocardiograms were recorded from male and female Mbnl1+/+ and Mbnl1ΔE2/ΔE2 mice at ~2 and 4 months of age under light isoflurane anesthesia. Male and female Mbnl1ΔE2/ΔE2 mice showed a ~28% and ~40% increase in the QRS duration, respectively at 2 months (male: p < 0.0001; female: p < 0.0001) and a ~20% and ~19% increase in QRS duration at 4 months, respectively (male: p = 0.017; female: p < 0.0001) (Fig. 2a & b, c ii & d ii). Interestingly, ~30% of Mbnl1ΔE2/ΔE2 mice studied at both ages showed bundle block, where the QRS complex shows an extra deflection, reflecting the different speeds with which depolarization occurs in the two ventricles (Fig. 2c iii & d iii). Bundle blocks were not observed in Mbnl1+/+ animals. The QTc interval showed ~13% and ~14% increase in duration at 2 months of age in male and female Mbnl1ΔE2/ΔE2 mice respectively (male: p = 0.001; female: p < 0.0001). A trend toward QTc widening was observed at 4 months of age in male and female Mbnl1ΔE2/ΔE2 mice (Fig. 2a & b). The STc interval was ~13% and ~22% shorter in male and female Mbnl1ΔE2/ΔE2 mice at 2 months of age (male: p = 0.01; female: p = 0.0002) and ~14% and ~23% shorter in male and female Mbnl1ΔE2/ΔE2 mice at 4 months (male: p = 0.031; female: p < 0.0001) (Fig. 2a & b). Male and female Mbnl1ΔE2/ΔE2 mice showed increased heart rates and shorter RR intervals (Fig. 2a & b). Data are mean and standard deviation. Recorded values are shown in Supplementary Fig. S3 & S4; Supplementary Table S1 & S2.

Figure 2
figure 2

Mbnl1ΔE2/ΔE2 mice show QRS, QTc widening, bundle block and STc shortening at 2 and 4 months of age.

(a & b). Relative electrocardiogram interval values at 2 and 3–4 months of age for male (2 months: n = 3; 3–4 months: n = 3) and female (2 months: n = 3; 3–4 months: n = 5) Mbnl1+/+ and male (2 months: n = 3; 3–4 months: n = 3) and female (2 months: n = 3; 3–4 months: n = 5) Mbnl1ΔE2/ΔE2 mice are shown. Mice underwent light isoflurane anesthesia at an inducing dosage of 3% isoflurane in oxygen, which was reduced to 1–1.5% isoflurane in oxygen when measurements were recorded. As mouse ECG waveform shapes are different from human, QRS duration values in this study include the Tri (transient re-entry current) wave, as it is part of the ventricular depolarization phase. 80–125 beats were analyzed for each mouse. p-values were calculated using the Student's t-test with significance set at p ≤ 0.05. p values ≤0.05 are indicated in red. (c, d). Shown are representative electrocardiogram traces for male (c) and female (d) Mbnl1+/+ and Mbnl1ΔE2/ΔE2 mice at 3–4 months of age. QRS expansion (cii & dii) and bundle block (ciii & diii) are observed Mbnl1ΔE2/ΔE2 mice. Bundle block was observed in ~30% of Mbnl1ΔE2/ΔE2 mice in both age groups. Recorded values for all intervals are shown in Supplementary Fig. S3 & S4 and in Supplementary Tables S1 & S2.

Full size image

Elongated QTc intervals, diminished R wave amplitudes and sinus node dysfunction occur in Mbnl1ΔE2/ΔE2 mice at 6 months of age

As Mbnl1ΔE2/ΔE2 mice showed elevated heart rates, we reexamined the EKGs in male and female mice at 6 months of age subsequent to anesthesia with ketamine/xylazine, which reduces the heart rate and can therefore potentially uncover additional conduction defects. As predicted the heart rate was lower in both Mbnl1+/+ and Mbnl1ΔE2/ΔE2 mice with ketamine/xylazine administration when compared to that observed with isoflurane (Fig. 3a; Supplementary Table S3). As a prominent S wave was not observed with ketamine/xylazine, we measured QTc intervals. With ketamine/xylazine treatment, the QTc interval showed a prolongation of ~30% and ~66% in Mbnl1ΔE2/ΔE2 male (p = 0.006) and female (p = 0.039) mice respectively (Fig. 3a–d). Data are mean and standard deviation. Recorded values are shown in Supplementary Fig. S5; Supplementary Table S3.

Figure 3
figure 3

Mbnl1ΔE2/ΔE2 mice anesthetized with ketamine/xylazine show prolonged QTc intervals, diminished R wave amplitudes and sinus node dysfunction at 6 months of age.

(a). Relative electrocardiogram interval values for male (n = 4) and female (n = 6) Mbnl1+/+ and male (n = 3) and female (n = 5) Mbnl1ΔE2/ΔE2 mice at 6 months of age are shown. Mice underwent anesthesia via an intraperitoneal injection of 80 mg/kg ketamine and 8 mg/kg xylazine in saline. QRS durations include the Tri (transient re-entry current) wave, as it is part of the ventricular depolarization phase in mice. As S waves, following the triwaves, were not prominent in Mbnl1+/+ and Mbnl1ΔE2/ΔE2 mice after ketamine/xylazine treatment QRS intervals were not measured. Beats analyzed for each mouse were male (n = 74) and female (n = 77) Mbnl1+/+ and male (n = 31) and female (n = 54) Mbnl1ΔE2/ΔE2 mice. p-values were calculated using the Student's t-test with significance set at p ≤ 0.05. p-values ≤0.05 are indicated in red. (b–d). Representative electrocardiogram traces showing diminished R wave amplitudes in male (b) and female (c) Mbnl1ΔE2/ΔE2 mice and expanded Tri-waves in female (d) Mbnl1ΔE2/ΔE2 mice. (e): % Mbnl1+/+ and Mbnl1ΔE2/ΔE2 mice showing diminished R wave amplitudes. (f, g). Sporadic expansion of RR intervals to ≥200% of the average RR interval was observed in ~30% of the Mbnl1ΔE2/ΔE2 mice analyzed [male (n = 4) and female (n = 6) Mbnl1+/+ mice and male (n = 3) and female (n = 5) Mbnl1ΔE2/ΔE2 mice]. Representative electrocardiogram traces from female Mbnl1+/+ (f) and Mbnl1ΔE2/ΔE2 (g) mice are shown. Recorded values for all intervals are shown in Supplementary Fig. S5 and in Supplementary Tables S3.

Full size image

In these experiments we observed additional QRS waveform abnormalities including prolongation of the Tri wave (Transient reentry current) and sharply diminished R wave amplitudes (Fig. 3b–d). The prolonged Tri waves are consistent with QRS widening observed with isoflurane. Significantly, R wave amplitudes were strikingly smaller in ~90% of the Mbnl1ΔE2/ΔE2 mice examined, with the R waves in most cases not reaching the isoelectric point (Fig. 3b, c & e). As poor R wave progression can reflect either lead misplacement, bundle blocks, prior myocardial infarcts, fibrosis or hypertrophy31 we repeated these experiments on a second EKG machine using a different type of acquisition system and leads and with an alternate investigator. As similar results were obtained in both cases, poor R wave amplitudes could reflect bundle blocks, myocardial fiber death, fibrosis or hypertrophy in Mbnl1ΔE2/ΔE2 mice.

In ~25% of the Mbnl1ΔE2/ΔE2 mice examined we observed elongated RR intervals (~1% to 2.5% of beats over a 20 minute interval), which reflected ≥200% increase in the time required for a new P wave to form. SA node firing was therefore inconsistent and sporadically required extended times to fire in Mbnl1ΔE2/ΔE2 mice. Such events were not observed in Mbnl1+/+ mice. A 5 minute interval EKG for a representative Mbnl1+/+ mouse and an Mbnl1ΔE2/ΔE2 mouse with elongated RR intervals is shown in Fig. 3f & g.

Mbnl1ΔE2/ΔE2 mice develop cardiac hypertrophy at 6 months of age

Left ventricular function and chamber dimensions were measured by ultrasound echocardiography at ~2 and 6 month of age in male and female Mbnl1+/+ and Mbnl1ΔE2/ΔE2 mice (Fig. 4; Supplementary Figure S6 and Supplementary Table S4 & S5). No abnormalities were noted at 2 months of age in male Mbnl1ΔE2/ΔE2 mice (Mbnl1+/+ n = 3; Mbnl1ΔE2/ΔE2 n = 3) (Fig. 4a; Supplementary Table S4). Echocardiography could not be performed successfully on female mice at 2 months of age due to their small size. At 6 months of age, posterior wall thickness (PWT) was ~24% greater (p = 0.007), ventricular septal thickness (VST) ~20% greater (p = 0.001) and left ventricular mass ~34% greater (p = 0.054) in male Mbnl1ΔE2/ΔE2 mice when compared to male Mbnl1+/+ mice (Mbnl1+/+ n = 6; Mbnl1ΔE2/ΔE2 n = 5) (Fig. 4b & c; Supplementary Table S5). Left ventricular function was not diminished in male Mbnl1ΔE2/ΔE2 mice. In females significant alterations in the left ventricular function and chamber dimensions were not observed at 6 months of age (Mbnl1+/+ n = 3; Mbnl1ΔE2/ΔE2 n = 5). However 2/5 female Mbnl1ΔE2/ΔE2 mice tested at 6 months of age showed ~200–300% increase in left ventricle mass and ~30–60% decrease in left ventricle percent fractional shortening (Supplementary Table S5).

Figure 4
figure 4

Mbnl1ΔE2/ΔE2 mice show ventricular hypertrophy at 6 months of age.

(a,b). Relative values for ultrasound echocardiographic measurements at 2 and 6–7 months of age for male Mbnl1+/+ (2 months: n = 3; 6–7 months: n = 6) and Mbnl1ΔE2/ΔE2 mice (2 months: n = 3; 6–7 months: n = 5) are shown. p-values were calculated using Student's t-test with significance set at p ≤ 0.05. Values where p ≤ 0.05 are shown in red. (c). 2-D guided M-Mode images of representative 20 week male Mbnl1+/+ and Mbnl1ΔE2/ΔE2 hearts. Ventricular septal thickness and posterior wall thickness are enlarged in the male Mbnl1ΔE2/ΔE2 hearts. The depth from top to bottom is 7 mm. Time is 100 ms/div. Abbreviations: EDD: End-diastolic dimension (green). ESD: End-systolic dimension (blue). VST: Left ventricular septal thickness (red); PWT: Posterior wall thickness (yellow). Recorded values are shown in Supplementary Fig. S6 and in Supplementary Tables S4 & S5).

Full size image

Multi-focal calcification of myocardial fibers and interstitial fibrosis occur with Mbnl1 depletion at 6 months of age

As poor R wave progression can reflect fibrosis or prior myocardial cell death, a histological analysis of Mbnl1+/+ (n = 3) and Mbnl1ΔE2/ΔE2 (n = 3) female hearts at 6 months of age was carried out using H&E and E trichrome stains. Mbnl1ΔE2/ΔE2 heart sections showed multi-focal calcified myocardial fibers and interstitial fibrosis prominently in the interventricular septum and ventricles (Fig. 5). Such structural alterations can act as electrical insulators blocking or altering the course of the action potential depolarization wavefront and thus contributing to the conduction defects observed in Mbnl1ΔE2/ΔE2 hearts at 6 months of age. Histological alterations were not observed at 2 months of age in Mbnl1ΔE2/ΔE2 hearts (data not shown). Thus taken together these analyses demonstrate that Mbnl1 depletion results in progressive cardiac structural and functional abnormalities (Fig. 6).

Figure 5
figure 5

Multi-focal calcification and fibrosis in Mbnl1ΔE2/ΔE2 hearts at 6 months of age.

(a–f). Histological findings in 6 month female Mbnl1+/+ (a, c, e) hearts (n = 3) and female Mbnl1ΔE2/ΔE2 (b, d, f) hearts (n = 3) are shown. In Mbnl1+/+ mice fibrosis and calcifications were not observed ((a): H&E stain ×12.5, (c): H&E stain ×200 and (e): trichrome stain ×400). In the Mbnl1ΔE2/ΔE2 mice there are multiple foci of calcified myocardial fibers that appear dark blue in the H&E stained sections (arrows) ((b): H&E stain ×40, (d): H&E ×200). The inset shows individual blue dots that likely represent calcification of individual mitochondria (x400). In panel (f), a focus of calcification with interstitial fibrosis (light blue lines in this trichrome stained section, ×400) is shown. Both the calcification and the fibrosis were multifocal, involving less than 1% of the myocardial tissue and were seen in all the Mbnl1ΔE2/ΔE2 animals.

Full size image
Figure 6
figure 6

Cardiac pathology observed over 2–6 months of age in Mbnl1ΔE2/ΔE2 mice.

The heart image shown is obtained from ClipArtist.net (http://www.clipartlord.com/free-human-heart-clip-art/).

Full size image

Increased embryonic isoform expression in Mbnl1ΔE2/ΔE2 hearts

As Mbnl1 is known to regulate RNA splice transitions during development, we curated a list of splice events in RNAs reported to be regulated by Mbnl1 in skeletal muscle and which are expressed at low levels in the heart26,32 from RNA-seq data obtained from a 129 sv/BL6 Mbnl1 knockout mouse developed by Kanadia and colleagues24, which was back crossed to wild-type 129 sv mice for more than 4 generations27 and splice events that are developmentally regulated in the heart33 as a reference (Fig. 7 & 8; Supplementary Fig. S7). As Scn5a exon 6a inclusion is developmentally regulated33 and because recent studies have implicated Scn5a exon 6a missplicing in the development of DM1 conduction defects and sudden death34 we studied this splice event and observed enhanced Scn5a exon 6a inclusion in adult Mbnl1ΔE2/ΔE2 hearts in a manner reminiscent of DM1 and similar to that observed in E18 Mbnl1+/+ hearts (Fig. 7b). Three independent RNAs, Junctin, Junctate and Asph are derived from a single genomic locus by alternative splicing35. Asph Exon 4a inclusion is diminished in E18 Mbnl1+/+ hearts and in adult Mbnl1ΔE2/ΔE2 hearts when compared with adult Mbnl1+/+ hearts (Fig. 7b). Asph Exon 4a is present in one of four alternatively spliced Asph isoforms, one of two Junctate isoforms and in both alternatively spliced forms of the Junctin RNA in the mouse heart35. To confirm that the two alternatively spliced isoforms of junctin (Asph Exon1a–5a and Asph Exon 1–5a) show constitutive inclusion of exon 4a in 129 sv mice, we examined the two known Junctin isoforms by RT-PCR. A single band was observed when primers were located in Asph exon 1 and exon 5a and when primers were located in Asph exon 1a and exon 5a (Supplementary Fig. S8), demonstrating that exon 4a is included in both Junctin RNA isoforms. As Asph exon 4a inclusion is diminished in adult Mbnl1ΔE2/ΔE2 hearts, this splice error is predicted to decrease Junctin levels. Steady-state levels of Junctin (Exon 1a–5a) and Junctin (Exon 1–5a) mRNA levels measured by qPCR were ~70% and ~80% reduced respectively, in Mbnl1ΔE2/ΔE2 hearts when compared to controls (Fig. 8a–c; Supplementary Fig. S8). Diminished Atp2a1 exon 22 inclusion, enhanced inclusion Atp11a exon 28a, diminished inclusion of Cacna1s exon 29 and enhanced Ryr2 exon 5 only inclusion (Fig. 7b) are observed in Mbnl1ΔE2/ΔE2 hearts and E18 Mbnl1+/+ hearts when compared with adult Mbnl1+/+ hearts. Thus RNAs regulating sodium and calcium currents show increased expression of embryonic splice isoforms in Mbnl1ΔE2/ΔE2 hearts. With Mbnl1 depletion, three RNAs involved in inter and intra cellular transport Clta, Stx2 and Tjp1/Zo1 are aberrantly spliced in a manner reminiscent of E18 Mbnl1+/+ hearts (Fig. 7c). Atp2a2 exon 18 and Kcnip2 exon 4 splicing was not altered in Mbnl1ΔE2/ΔE2 hearts (Fig. 8; Supplementary Fig. S7).

Figure 7
figure 7

Mbnl1 depletion results in the persistence of embryonic splice isoforms in adult hearts.

(a). A cardiac RNA network that is misspliced in Mbnl1ΔE2/ΔE2 hearts is shown. RNAs are categorized by their biological function and Mesh terms. (b–g). Splicing analysis for the indicated RNAs was performed by RT-PCR with E18 Mbnl1+/+ and 4 months old male Mbnl1+/+ (n = 3) and male Mbnl1ΔE2/ΔE2 (n = 3) hearts. Band intensities were quantified by densitometry. Data are standard error of mean (SEM). p-values were calculated using Student's t-test with significance set at p ≤ 0.05. Primer locations, exon numbers and expected band sizes are indicated. The alternatively spliced exons are shown as green boxes. Exon numbers for all genes except Junctate/Asph are annotated based on Refseq from the UCSC genome browser (NCBI37/mm9); Junctate/Asph is annotated based on the study carried out by Dinchuck et al (Dinchuk et al., 2000). Scn5a exon 6 and 6A were distinguished by digestion with SacI. The SacI site within Scn5 exon 6A is indicated in red. Identified novel exons are indicated by asterisks and the sequences of these exons are shown in Supplementary figure S9. The images shown are cropped. The full-length gels are shown in Supplementary figures S10–S15.

Full size image
Figure 8
figure 8

Tabulation of splice errors in Mbnl1ΔE2/ΔE2 hearts.

(a). Splice events examined in this study are tabulated. PSI: percent spliced in; NAS: Alternative splicing not detected by RT-PCR analysis. (b, c). Steady-state Junctin (E1a–5a) and Junctin (E1–5a) mRNA levels were analyzed by qPCR in E18 Mbnl1+/+ hearts, 4 month male Mbnl1+/+ (n = 3) and 4 month male Mbnl1ΔE2/ΔE2 (n = 3) hearts. Exon numbers and primer positions are shown. Data are standard error of mean (SEM). p-values were calculated using Student's t-test with significance set at p ≤ 0.05. Values where p ≤ 0.05 are shown in red. (d). Potential relationships between splice errors and the cardiac pathology observed are indicated. The heart image shown is obtained from ClipArtist.net (http://www.clipartlord.com/free-human-heart-clip-art/).

Full size image

We examined splice site choice in Titin (Ttn) and Myomesin I (Myom1) RNAs, which encode PEVK or PEVK-like domains that contribute to protein elasticity and can therefore regulate the compliance of the myocardium36,37,38,39,40. In addition, we examined splice events in Capn3, Sirt2, Csda, Clk2 and Madd, which are reported to regulate cell viability41,42,43,44,45,46. We did not observe missplicing of Ttn exon 9, Ttn exon 311, or Ttn exons 114–119, 121–124 or 171–183 in Mbnl1ΔE2/ΔE2 hearts (Supplementary Fig. S7). Mbnl1 loss resulted in enhanced inclusion of Myom1-exon 18 (Fig. 7d), diminished Capn3 exon 16 and 17 inclusion, enhanced Sirt2 exon 2 inclusion and enhanced Csda exon 6 inclusion (Fig. 7e). Mbnl1 depletion did not alter Madd exon 5 or Clk2 exon 4 splicing (Fig. 8; Supplementary Fig. S7). A novel splicing event detected in MyomI is shown in Supplementary Fig. S9.

We tested splice events in additional RNAs encoding sarcomere proteins (Tnnt2, Ldb3/Zasp), proteins regulating sarcomere and cytoskeleton assembly and function (Ablim1, Nrap, Mapt, Trim55/Murf2, Pdlim3, Pdlim5, Sorbs1, Sorbs2, Spag9, Fhod1 and Arhgef7). From this group of RNAs, Tnnt2, Ldb3/Zasp, Trim55, Mapt, Pdlim3, Pdlim5, Sorbs1, Sorbs2, Fhod1 and Spag9 RNAs are aberrantly spliced in Mbnl1ΔE2/ΔE2 hearts, in a manner similar to that observed in E18 Mbnl1+/+ hearts (Fig. 7 d & f). Arhgef7 exon 16, Nrap exon 12 and Ablim1 exon 18 splicing was not altered in Mbnl1ΔE2/ΔE2 hearts (Fig. 8; Supplementary Fig. S7). Novel splice events detected in Pdlim3/Alp are shown in Supplementary Fig. S9.

Lastly, we examined splice events in RNAs known to play a role in heart development (Tbx5)47, myogenesis (Fbxo40)48 and splicing (Mbnl2)22. We observe enhanced inclusion of Mbnl2 exon 6 (Fig. 7g), but splicing of Tbx5 exon 6 and Fbxo40 exon 2 was not altered in Mbnl1ΔE2/ΔE2 hearts (Fig. 8; Supplementary Fig. S7). The RNA network where embryonic isoforms persist in adult Mbnl1ΔE2/ΔE2 heart is shown in Fig. 7a.

Discussion

Mbnl1 depletion in 129 sv mice results in QRS and QTc widening, STc shortening, bundle blocks, diminished R wave amplitudes and SA node dysfunction in conjunction with cardiac hypertrophy, fibrosis, multi-focal myofibrillar death, calcification and sudden death. Concurrently Mbnl1 depletion results in the enhanced expression of embryonic splice isoforms of RNAs regulating sodium and calcium currents (Scn5a, Asph, Junctin, Junctate, Atp2a1, Atp11a, Cacna1s, Ryr2), intra and inter cellular transport (Clta, Stx2 and Tjp1/Zo-1), compliance of the myocardium (Myom1), cell survival (Capn3, Sirt2, Csda) cytoskeleton and sarcomere assembly and function (Trim55, Mapt, Pdlim3, Pdlim5, Sorbs1, Sorbs2, Fhod1 and Spag9) and encoding sarcomere structural proteins (Tnnt2, Ldb3). As several of these features are observed in DM11,2,3,4,5,6,7,8,9,10,11,12,32,34,49,73, this study supports a key role for Mbnl1 depletion in the initiation of DM1 cardiac pathology.

We examined the cardiac pathology that manifests over a period of two to six months in Mbnl1ΔE2/ΔE2 mice and observe that QRS widening manifests at two months, is sustained at 4 months of age and that other distinctive QRS waveform abnormalities manifest at 6 months of age. Bundle block is observed in the 2–4 month period. Interestingly, QRS and QTc widening were more prominent at 2 months when compared to 4 months in Mbnl1ΔE2/ΔE2 mice. This may reflect early mortality of the more severely affected animals (Fig.1e). Interestingly, as with DM1, where functional and structural involvement is associated with age and the male gender10,15, significant cardiac hypertrophy manifested at 6 months of age in male Mbnl1ΔE2/ΔE2 mice, with females showing a trend towards hypertrophy. In contrast, QTc widening in DM1 is associated with being female and older5. Reminiscent of this feature, QTc widening was more prominent at 6 months of age in female Mbnl1ΔE2/ΔE2 mice when compared to males. Cardiac histology in Mbnl1ΔE2/ΔE2 mice showed similarities with DM1 with the manifestation of multi-focal calcification and fibrosis located primarily in the septum and ventricles at 6 months of age10. Such histological alterations could underlie the conduction defects observed at 6 months. The concurrent manifestation of conduction defects, ventricular hypertrophy and histological abnormalities may therefore serve to increase the predisposition to sudden cardiac death with Mbnl1 depletion. In this regard it is of interest to note that QRS, QTc widening and echocardiographic abnormalities have been implicated as risk factors for sudden death in DM1 patients5,9,10,11,12.

Our current and previous analysis of DM1 mouse models shows that DM1 cardiac pathology can result from DMPK, SIX5 and MBNL deficits50,51,52,53. CTG expansion is known to result in an ~20–75% decrease in DMPK54,55 and ~50% decrease in SIX5 levels56,57, as a consequence of DMPK RNA retention in the nucleus and epigenetic silencing of the SIX5 allele linked to the expanded CTG tract. We have shown that first, second and third degree AV block manifest in mice as a function of time and diminishing Dmpk dosage ranging from 50–100% loss50,51,52. In other experiments we have shown that a Six5+/− mice show QRS widening53. As almost a complete sequestration of MBNL1 in CUG foci can occur in DM1 patient hearts with advanced disease49, we examined the consequence of complete Mbnl1 loss in the mouse heart and observe that Mbnl1 loss plays a key role in initiating the electro and echocardiographic abnormalities detailed above. In a recent study, Lee and colleagues have shown that combinatorial loss of Mbnl1 and Mbnl2 in 129 sv/BL6 mice (Mbnl1−/−/Mbnl2+/−) show both PR expansion or first degree heart block in conjunction with QRS expansion58. Taken together these results demonstrate that losses in DMPK, SIX5, MBNL1 and MBNL2 can recreate a great number of the prominent cardiac features observed in DM1. As systole dysfunction and diminished ejection fraction is not observed in these mouse models it is unclear if additional losses in Mbnl3 will be required to fully recapitulate both the echo and electrocardiographic abnormalities described in DM1. Significantly, CUG repeat expression also results in elevated steady state levels of the RNA binding protein, Cug-bp118 and transgenic mice that acutely express Cug-bp1 levels at levels that are ~4–8 fold higher than endogenous levels show PR and QRS prolongation, dilated cardiomyopathy, systolic dysfunction with reduced ejection fractions and widespread necrosis with death occurring 7 days post induction of Cug-bp1 expression59. As the acute induction of high levels of CUG-BP1 is not observed in DM1 patients, it is important to examine the nature of the pathology that manifests with the constitutive expression of Cug-bp1 in mice, at levels that are similar to that observed in DM1 patient hearts. Such experiments will allow better comparisons to be made between the pathology resulting from CUG-BP1 over-expression and reduced levels of DMPK, SIX5 and the MBNL proteins. As Cug-bp1 over-expression alters splice site choice in a manner that partially overlaps with that observed with Mbnl1 loss60, pathophysiology shared by both mouse models has been hypothesized to result from similar RNA defects, while unique features and the disease trajectory of the cardiac pathology observed in Mbnl1ΔE2/ΔE2 hearts are predicted to result from RNA defects resulting exclusively from Mbnl1 loss.

Lee et al.58 have demonstrated that Mbnl1 exon 2 deleted mice developed by Kanadia and colleagues24 on a mixed 129 sv/BL6 background do not show cardiac pathology (annotated as Mbnl1−/− by the authors). These animals showed modest to severe splice errors in 8 RNAs. The lack of a cardiac phenotype was attributed to a compensatory increase in Mbnl2 steady-state levels in 129 sv/BL6 Mbnl1−/− hearts. Reduction in Mbnl2 levels in conjunction with Mbnl1 loss in 129 sv/BL6 Mbnl1−/−/Mbnl2+/− mice resulted in a decrease in life span and the development of cardiac pathology comparable to that described for the 129 sv Mbnl1ΔE2/ΔE2 mice in this study. It should be noted, that both in this study and that of Lee and colleagues Mbl1 exon 2, which encodes the Mbnl1 translational start site, was deleted. Furthermore, both 129 sv/BL6 Mbnl1−/− and 129 sv Mbnl1ΔE2/ΔE2 hearts show an increase in steady-state Mbnl2 levels (Fig. 1d). However, when Tnnt2 exon 5 inclusion, a splice error studied in all three mouse strains is examined interesting differences come to light. In 129 sv/BL6 Mbnl1−/− hearts, aberrant Tnnt2 exon 5 inclusion is ~20%. In contrast both 129 sv/BL6 Mbnl1−/−/Mbnl2+/− hearts and 129 sv Mbnl1ΔE2/ΔE2 hearts show ~70–85% inclusion of Tnnt2 exon 5 (Fig. 7d). In contrast Arhgef7 was misspliced in 129 sv/BL6 Mbnl1−/− hearts but not in 129 sv Mbnl1ΔE2/ΔE2 hearts (Supplementary Fig. S7 and Ref. 58). These results demonstrate that mouse strain differences can alter the outcome of both Mbnl1 mediated splicing and cardiac pathology. This observation is significant, as DM1 is recognized to be a disorder characterized by significant phenotypic variability associated with CTG tract expansions of similar lengths61. Our results support the hypothesis that modifier genes regulating Mbnl1 function can strongly influence disease trajectory in DM1 patients. Thus comparison of these two mouse models provides a unique opportunity to identify modifiers that regulate the severity of DM1 pathology. Identification of such modifier genes will allow better predictions to be made for disease trajectories in individual patients and provide mechanistic insights into Mbnl1 regulation in vivo.

Splice events that we have tested in Mbnl1ΔE2/ΔE2 hearts are shown in Fig. 7 & 8. As several lines of evidence point to alterations in myocyte force generation, force transmission via the cytoskeleton and calcium homeostasis as critical signals that result in cardiac remodeling and disease62, interesting correlations exist between the RNAs that show aberrant splicing and the Mbnl1ΔE2/ΔE2 heart phenotype. A subset of splice events that are noteworthy with respect to the pathology observed in Mbnl1ΔE2/ΔE2 hearts are discussed. A central pathology resulting from Mbnl1 loss is conduction defects, which can reflect both functional and structural cardiac abnormalities. With regard to functional deficits, Scn5a and Junctin/Junctate missplicing is of interest. Adult Mbnl1ΔE2/ΔE2 hearts show ~2.5 fold increase in the inclusion of Scn5a exon 6a (~12%) (Fig. 7b). Inclusion of exon 6a in Scn5a results in a sodium current loss of function phenotype, which manifests with a depolarized shift in steady-state activation, slower kinetics of activation and inactivation, slower recovery from inactivation, an increase in the time for the sodium currents to peak and reduced channel availability63. This splice error has been recently reported in DM1 patients and is predicted to contribute to sudden death in this patient population34. As the QRS interval duration is influenced by the time for sodium currents to peak64, enhanced inclusion of Scn5a exon 6a may underlie QRS widening in Mbnl1ΔE2/ΔE2 mice. Significantly, QRS widening is known to greatly increase the chance of sudden death in the general human population with cardiomyopathy and in DM1 patients11,12,65. Additionally, Scn5a loss of function mutations are known to cause sick sinus syndrome66 where reduction in the flow of sodium ions alters the ability of the SA node to develop and spread electrical signals. Therefore, this splice error may be of significance both with respect to SA node dysfunction and QRS widening in Mbnl1ΔE2/ΔE2 mice. Mbnl1ΔE2/ΔE2 hearts show altered splicing of Junctate and reduction in Junctin mRNA levels. As aberrant levels of junctin and junctate are known to alter calcium homeostasis and result in arrhythmia and sudden death67,68,69,70, it is possible that these splice errors can further enhance the chance of sudden death in Mbnl1ΔE2/ΔE2 mice. Lastly, it is of interest to note that Ryr2 mutations have been reported to cause polymorphic ventricular tachycardia and that Cacna1s missplicing results in aberrant gating of Ca(V)1.1 calcium channel71,72. Thus alterations in sodium and calcium currents may underlie the conduction defects and sudden death observed in Mbnl1ΔE2/ΔE2 mice

Mbnl1 depletion leads to multi-focal myofibrillar death, calcification and fibrosis prominently in the septum and ventricles (Fig. 5). These structural alterations can act as electrical insulators to result in conduction blocks. Enhanced myocyte death or fibrosis in the His-Purkinje system or ventricle can lead to diminished R wave amplitudes and QTc widening. With regard to the development of fibrosis, it should be noted that whole body inactivation of the splice regulator, RBM20 in mice results in altered titin splicing with the persistence of large, elastic embryonic N2BA titin isoforms in adult mouse hearts37. This change increases myocardial compliance and the reduced recoil of flaccid titin filaments is predicted to play a causal role in an adaptive increase in collagen biosynthesis and the development of fibrosis, arrhythmias and sudden death in RBM20 knockout mice36,37. RNAs encoding the large embryonic N2BA titin isoforms are not observed in adult Mbnl1ΔE2/ΔE2 hearts. In contrast, both in DM1 patient hearts and in Mbnl1ΔE2/ΔE2 hearts, enhanced expression of embryonic myomesin I, an isoform that is more elastic and thus more compliant than the adult isoform of myomesin I is observed38,39,40,73. As myomesin is adapted to bear mechanical stress and M-line proteins must be stretched and coiled during muscle contraction and relaxation, increased elasticity of embryonic Myomesin I is predicted to trigger an adaptive increase in fibrosis in a manner similar to that predicted for titin missplicing in RBM20 deficient mice36,37,38,39,40. We observe that Mbnl1 depletion results in diminished inclusion of Capn3 exon 16 & 17. Capn3 is a calcium dependent protease that binds to titin and cleaves several cytoskeletal proteins. Absence of Capn3 mediated substrate cleavage has been shown to result in muscle fiber death41. Analysis of Capn3 splice variants has shown that Capn3 exon 16 is required for fodrin cleavage42. Thus the diminished inclusion of Capn3 exon16, is of interest with respect to the cardiac myofiber death observed in Mbnl1ΔE2/ΔE2 hearts.

We observe a very high rate of Tnnt2 exons 4 and 5 inclusion in Mbnl1ΔE2/ΔE2 mice. This inclusion pattern is more marked than that observed in E18 Mbnl1+/+ hearts (Fig. 7d). The importance of this isoform switching in the development of cardiac hypertrophy, is underscored by an intron 3 polymorphism that alters exon 4 splicing and predisposes to left ventricle hypertrophy74. Mutations in Zasp/Cypher in humans are associated with cardiac hypertrophy75 and re-expression of embryonic RNA isoforms of MyomI, Asph, Mbnl2 have been reported in rat models of cardiac hypertrophy76. Lastly, we observe a cascade of splice errors reminiscent of embryonic splice patterns in RNAs regulating cytoskeleton and sarcomere assembly and function including Trim55, Mapt, Pdlim3, Pdlim5, Sorbs1, Sorbs2, Fhod1 and Spag9 in Mbnl1ΔE2/ΔE2 hearts. We speculate that the aberrant isoform stoichiometry of these proteins could enhance myocyte fragility, death or maladaptive hypertrophy in Mbnl1ΔE2/ΔE2 hearts. Elucidation of the causal role of these splice errors in DM1 cardiac pathology in future studies should provide important mechanistic insights into DM1 cardiac disease.

Human heart failure is associated with high morbidity and is an end-stage of various forms of heart disease77. Several characteristic changes are observed in end-stage failing hearts, with the best documented alterations at the molecular and organ level being a transcriptional and post-transcriptional splicing reprogramming to resemble that of the embryonic heart, various degrees of hypertrophy, mitochondrial dysfunction, alteration of sarcomere and cytoskeleton architecture, aberrant calcium handling, increase in myocyte cell death by apoptosis or necrosis and enhanced extracellular matrix formation78. Although the switchback to an embryonic splice program is well established in heart failure its potential role in either accelerating healing or in triggering further dysfunction and death is unclear. For this reason there has been a great interest in identifying and characterizing splice regulators that control embryonic to adult splice transitions. Although RBM20, ASF/SF2 and SC35 were initially regarded as potential candidates that regulate the transition of embryonic to adult splice programs, it is now clear that the MBNL1 and CUG-BP1, which plays an opposing role to MBNL1, play a central role in this phenomenon. Results from this study suggest that re-expression of embryonic isoforms in heart failure may be maladaptive.

It is significant to note that the persistence of the embryonic splice program observed in Mbnl1ΔE2/ΔE2 hearts is reminiscent of that observed in human heart failure. Indeed several of the classic markers for human heart failure including re-expression of embryonic RNA splice isoforms of Myom I, Tnnt2 and low Junctin levels characterize the Mbnl1ΔE2/ΔE2 hearts33,70,79,80. Interestingly, global analysis of splice changes in human heart failure demonstrates a significant overrepresentation of MBNL motifs in sequences flanking differentially spliced exons79. It should be noted that MBNL1 levels increase and CUG-BP1 levels decrease during heart development60 and molecular events that change these trajectories can reinstate embryonic splice programs in the adult heart. As alterations in MBNL1 and CUG-BP1 levels or function can result in splice errors, cardiac dysfunction and death, it will be particularly important to understand the molecular events that regulate the activity of this pair of proteins in heart disease. Such studies are predicted to uncover novel therapeutic avenues that diminish the rates of progression to heart failure and death in both DM1 patients and in the general population.

Methods

Ethics Statement

All experiments were performed in accordance with the institutional guidelines of both the University of Southern California (USC) and the University of California, Los Angeles (UCLA). The USC protocol was approved by the Institutional Animal Care and Use Committee at the University of Southern California, Los Angeles (Protocol number: 10347). The UCLA protocol (99-028) was approved by the UCLA Office of Animal Research Oversight.

Statistical analysis

Kaplan-Meyer curves were developed to examine differences in survival between Mbnl1+/+ and Mbnl1ΔE2/ΔE2 mice, testing the difference in survival using the Mantel-Cox (log rank) test and adjusting for Mbnl1ΔE2/ΔE2 mice that were sacrificed. Further modeling examined the potential of a gender difference within the Mbnl1ΔE2/ΔE2 mice. Analyses were performed in SPSS (v.21); a = 0.05.

Surface ECG Recording

Electrocardiograms were obtained for at least 15 minutes from each mouse either under light isoflurane anesthesia or subsequent to ketamine/xylazine administration (ketamine 80 mg/kg and xylazine 8 mg/kg) by respectively inserting two Pt needle electrodes (Grass Technologies, West Warwick, RI) or two 29 guage needle electrodes (ADInstruments, Colorado Springs, CO) under the skin in the lead II configuration. Isoflurane induced ECG data were amplified (Grass Technologies) and then digitized for analysis with HEM V4.2 software (Notocord Systems, Croissy sur Seine, France). ECG data recorded through ketamine/xylazine administration were amplified using a Bio Amp (ADInstruments) and digitized using Labchart7 software (ADInstruments). Since mouse ECG waveform shapes are different from human, QRS duration values in this study include the Tri (transient re-entry current) wave, as it is part of the ventricular depolarization phase. T waves tend to be negative going in mice. To correct for the variability in the heart rate observed between mouse genotypes, corrected QTc and STc intervals were calculated using Bazett's Formula: QTc = QT/√RR and STc = ST/√RR.

Ultrasound Echocardiography

Left ventricular (LV) size, mass, wall thickness, ventricular and valve function and blood flow were assessed using methods previously described81,82. Measures of chamber dimensions [end-diastolic dimension (EDD); end-systolic function (ESD); ventricular septal thickness (VST); posterior wall thickness (PWT)], heart rate, ventricular function [left ventricular % fractional shortening (LV%FS); velocity of circumferential fiber shortening (Vcf); and left ventricular ejection fraction (LVEF)] and the early (E) and atrial (A) diastolic filling (E/A ratio) were obtained from mice lightly anesthetized with isoflurane (1.0–1.5%) to permit physiological levels of function.

Morphometry & Histology

All morphometry and histology experiments were carried out primarily as described by Jordan and colleagues82.

RNA analysis

Heart tissue was pulverized under liquid nitrogen. Total RNA was prepared using Trizol (Invitrogen, USA) according to the manufacturer's protocol. RNA from E18 heart tissues was obtained from Zyagen Inc. cDNA was synthesized from 5 μg of total RNA using the cDNA synthesis kit (Thermo Scientific., USA). cDNAs were used for splicing assays using primers and PCR conditions described in Supplementary Table S6. The relative band intensities were measured by densitometry analysis using Gene Tool (Syngene Inc., USA). To identify novel splice isoforms, bands were excised and DNA was extracted using gel extraction kits (Qiagen, USA). Extracted DNAs were cloned into pGEM-T Easy Vector (Promega, USA) according to the manufacturer's protocol. The DNA inserts were sequenced using T7 and SP6 promoter sequencing primers (Integrated DNA Technologies Inc, USA). To measure the level of exon 6a inclusion in Scn5a, amplified PCR products were visualized on 2% agarose gel followed by digestion with Sac I (NEB, USA) for 2 h at 37°C. The relative band intensities were measured by densitometry analysis using Gene Tool.

Western blot analysis

Mouse heart tissues were pulverized in liquid nitrogen and whole cell lysates were prepared by homogenization in modified RIPA buffer (Upstate Inc., USA). Protein extracts were subjected to Western blot analysis with anti-Mbnl1 described by Holt and colleagues83. Anti-Mbnl2 and anti-Tubulin antibodies were obtained from Santa Cruz Biotechnology Inc. USA.