Duchenne's muscular dystrophy: animal models used to investigate pathogenesis and develop therapeutic strategies

Abstract

Duchenne's muscular dystrophy (DMD) is a lethal childhood disease caused by mutations of the dystrophin gene, the protein product of which, dystrophin, has a vital role in maintaining muscle structure and function. Homologues of DMD have been identified in several animals including dogs, cats, mice, fish and invertebrates. The most notable of these are the extensively studied mdx mouse, a genetic and biochemical model of the human disease, and the muscular dystrophic Golden Retriever dog, which is the nearest pathological counterpart of DMD. These models have been used to explore potential therapeutic approaches along a number of avenues including gene replacement and cell transplantation strategies. High-throughput screening of pharmacological and genetic therapies could potentially be carried out in recently available smaller models such as zebrafish and Caenorhabditis elegans. It is possible that a successful treatment will eventually be identified through the integration of studies in multiple species differentially suited to addressing particular questions.

Duchenne's muscular dystrophy (DMD) is a lethal X-linked myopathy characterized by the near absence of dystrophin protein in skeletal muscles. The dystrophin–glycoprotein complex (DGC) connects the actin cytoskeleton of myofibres to the extracellular matrix and is therefore integral to the contractile structure of muscle (Watkins et al. 1997; Michele & Campbell 2003). Mutations which cause disruption of the component proteins of the DGC lead to a host of myopathies, of which DMD is both the most severe and the most common, affecting one in 3500 live male births. The dystrophin gene is highly conserved; homologues have been identified not only in vertebrates (mammals, birds and fish) but also in the popular invertebrate laboratory models Caenorhabditis elegans and Drosophila melanogaster.

In humans, the functional significance of dystrophin in skeletal muscle is demonstrated by the acute pathology that results from its absence. Muscle histology in DMD patients is almost normal before the onset of clinical symptoms at 3–5 years of age. The preliminary stage of the disease is characterized by the presence of focal groups of necrotic myofibres, muscle hypertrophy and abnormally high levels of muscle creatine kinase. In the second (pathological) phase, repeated cycles of degeneration exhaust the regenerative capacity of muscle-specific stem cells (satellite cells), and fibrotic mechanisms cause the progressive replacement of contractile muscle tissue with collagenous connective tissue (Rafael et al. 1997). This process leads to joint contractures, loss of ambulation by 10–12 years and death in the twenties from respiratory or cardiac failure (Wells & Wells 2002).

Therapeutic strategies

Several different therapeutic strategies have been pursued. The most perfect solution would be to place a normal copy of the dystrophin gene into muscle cells, and hence restore sufficient protein expression to improve structure and function. At 3.0 mB the dystrophin gene is vast (Hoffman et al. 1987), and as successful therapy would require massive and sustained gene transfer, this is a daunting task (Thioudellet et al. 2002). Nevertheless, the availability of high-efficiency viral infection (Roberts et al. 2002; Scott et al. 2002; Cerletti et al. 2003) and nonviral transfection (Thioudellet et al. 2002; Gollins et al. 2003; Lu et al. 2003) methods and the development of functional mini-dystrophin genes still allow it as a possibility (Thioudellet et al. 2002; Wells & Wells 2002). An alternative to replacing the faulty gene is to modulate its expression, e.g. by employing antisense oligonucleotides that alter RNA stability or splicing (Rando 2002; Lu et al. 2003), thereby resulting in the production of a functional protein.

Transplantation of normal donor (or genetically corrected host) muscle precursor cells (myoblast transfer) has also been explored as a method for restoring dystrophin protein to dystrophic muscle. Transplanted muscle precursors can differentiate to form dystrophin-expressing muscle and persist as part of the myotome indefinitely (Law et al. 1988; Partridge et al. 1989; Morgan et al. 1990). This technique avoids the problems of manipulating the large and unwieldy dystrophin gene, but is still constrained by the difficulties associated with treating large volumes of muscle with long-lasting effect. An alternative approach is to upregulate the expression of an endogenous protein that effects some functional replacement (Krag et al. 2001).

Currently, the only treatment to prove clinically efficacious is dosage with the steroid drug prednisone/prednisolone, which results in a modest increase in strength and delays, but does not halt, the progress of the disease (Backman & Henriksson 1995; Dubowitz et al. 2002).

Animal models of DMD

The most commonly used laboratory model of DMD is the C57Bl/10ScSn mdx (mdx) mouse (Bulfield et al. 1984). Whilst the mdx is a genetic and biochemical homologue of the disease, it has a somewhat milder phenotype. Muscle pathology is comparatively moderate and mechanical function is less seriously compromised, resulting in an almost normal lifespan. This is of evidence that dystrophin protein is less critical to muscle function in mice than it is in humans.

Whilst this apparent species-specific response to dystrophin deficiency may be informative regarding the function of the protein (Cullen & Jaros 1988; Partridge 1997), it also implies that the findings of animal studies should not be extrapolated to the human disease without caution. Historically, therapies that looked promising in animal models (such as myoblast transfer) have repeatedly yielded disappointing results in clinical trials (Partridge 1996; Miller et al. 1997). However, the understanding of comparative dystrophic pathology, has advanced in recent years, the results of animal experiments can now be analysed in a more informed, and consequently informative, way. Other than the ubiquitous mdx mouse, several experimental models have been established, ranging in evolutionary terms from the nematode worm, C. elegans, to the muscular dystrophic Golden Retriever dog.

This review discusses the relative merits and caveats of the various animal models used in DMD research and suggests how data compiled from studies using diverse species could potentially both advance knowledge of dystrophic pathogenesis and hone therapeutic strategies.

Murine models

Dy/dy and mdx mice

The dystrophin gene (mutations of which cause DMD) was not identified until 1987 (Hoffman et al. 1987). Before this time, several unauthenticated models of murine dystrophy were used in DMD research, selected for their pathological similarities to the human disease. A significant amount of work, including early studies of therapeutic myoblast transfer carried out by Peter Law and his associates, was carried out using the C57Bl/6 J-dy2J (dy/dy) mouse. However, it was later discovered that the genetic basis of dy/dy murine dystrophy is a mutation in the gene coding for laminin M-chain (Sunada et al. 1994; Xu et al. 1994) rather than dystrophin, disqualifying it as a model of DMD. At the same time, other groups were working with another murine model of muscular dystrophy, the mdx mouse (Coulton et al. 1988; Morgan et al. 1989). Mdx mice were derived from a naturally occurring mutant that arose within a C57Bl/10 colony, initially identified from its abnormally high plasma levels of creatine kinase (Bulfield et al. 1984). Whilst concerns were expressed that the mild mdx phenotype did not adequately model the human disease (Dangain & Vrbova 1984; Tanabe et al. 1986), the later discovery of the dystrophin gene fully authenticated the mdx as a genetic model of DMD (Hoffman et al. 1987).

Pathogenesis in the mdx mouse

Muscle pathology is most pronounced in the mdx between 2 and 8 weeks of age, a period characterized by the presence of necrotic foci, newly regenerated centrally nucleated myofibres and high plasma concentrations of creatine kinase. Cyclical degeneration and regeneration peaks between weeks 3 and 4; this window is thought to model DMD most closely (Partridge 1997). Mild myopathy with its associated fibrosis persists for the remainder of the animal's life, but does not become acute until senility. Detailed analysis of mdx muscle pathology has shown that whilst some muscles (such as the masseter) are spared, others (such as the gastrocnemius and in particular, the diaphragm) are severely affected in older animals (Muller et al. 2001). Less obviously pathological muscles such as the tibialis anterior are also more susceptible to contraction-induced injury than in the wildtype (Dellorusso et al. 2001); consequently moderate exercise can accelerate the course of the disease (De Luca et al. 2003). It is possible that the relative mildness of the mdx phenotype is, in part, an artefact of the animal house environment, which does not require or encourage much active movement and may therefore spare muscle.

Gene therapy studies in the mdx mouse

The amenability of mice to transgenic technology has permitted gene therapy studies to be carried out from two different angles. The therapeutic efficacy of a particular construct can be tested throughout the entire myotome by the creation of transgenic mice on an mdx background. Such experiments have been performed using a variety of constructs comprising both full and mini-versions of human and mouse dystrophin and have been used to investigate the molecular size, and quantity, of dystrophin necessary to prevent dystrophy (Wells & Wells 2002). Mdx mice have also been used to explore somatic gene transfer techniques; this approach is required to treat DMD patients. A number of both viral and nonviral systems have been successfully utilized as a proof of principle (Wells & Wells 2002) and oligonucleotide therapies have also been successfully employed (Rando 2002; Bertoni et al. 2003); however, it is debatable whether these results are likely to be reproducible in human patients without further development in larger models that are more closely homologous to DMD.

The mdx/utrn–/– (dystrophin and utrophin deficient) mouse

It is thought that mdx pathology is moderated by the partial functional replacement of dystrophin by the homologous protein utrophin (Rybakova et al. 2002). Mice which are null for both dystrophin and utrophin (mdx/utrn–/–) have a much more severe phenotype, comparable to human DMD (Deconinck et al. 1997). In contrast to the mdx, mdx/utrn–/– mice exhibit growth retardation, weight loss, spinal curvature and joint contractures, early diaphragmatic pathology and premature death (Deconinck et al. 1997). Mdx dystrophic pathology can be ameliorated by high levels of utrophin protein if expression is initiated early (Squire et al. 2002); likewise, when utrophin is delivered to the muscle of neonatal mdx/utrn–/– mice, the development of necrotic foci is substantially reduced (Wakefield et al. 2000). This supports the view that dystrophin and utrophin are functionally interchangeable (Rybakova et al. 2002).

These experiments demonstrate how the comparison of pathogenesis in physiologically different species and strains can reveal novel therapeutic targets; the amenability of the mouse to transgenic technology makes it particularly suitable for this type of study. Whilst dystrophin is a foreign protein in DMD patients and could potentially provoke an immune reaction, utrophin is endogenous and therefore nonimmunogenic (Wells & Wells 2002; Cerletti et al. 2003), a possible clinical advantage.

Immunodeficient mouse strains

Transgenic technology has been used to create several different immunodeficient strains of mouse that have proved useful in the development of cell transplantation therapies. Mdx mice generated on a nu/nu background have been shown to support murine donor-derived muscle formation and thus demonstrate myoblast transfer therapy as a proof of principle (Partridge et al. 1989). It is also possible to generate human donor-derived muscle in mice that are double knockouts for recombinase activating gene 2 and the common cytokine receptor (RAG2–/–/γc–/–) (Cooper et al. 2001). Results in these models, however, are only optimal after pretreatment of the host implantation site by either irradiation (Morgan et al. 1993) or cryodamage (Morgan et al. 1987). Ultimately, these treatments combined with the innate immunodeficiency result in a model that is further removed from the human patient.

The disparity between DMD boys and these models (including other obvious differences, such as size) probably contributed to the failure of clinical trials of myoblast transfer therapy carried out in the 1990s (Skuk & Tremblay 2001), which were perhaps initiated before potential problems such as immune rejection had been adequately addressed.

Another caveat associated with the use of immunodeficient strains is that certain aspects of the dystrophic phenotype may be altered. The absence of T cells in mdx mice bred on an nu/nu background was found to reduce fibre loss and collagen deposition compared to immunocompetent mdx (Morrison et al. 2000). Whilst this is an interesting and significant observation, it further decreases the integrity of the mdx nu/nu mouse as an accurate model of DMD.

Limitations of murine models

Mice cannot be used to address certain issues associated with DMD, such as the problems of performing gene or cell therapy on large volumes of muscle, and their differing pathological expression of the disease prevents them from being a precise substitute for DMD boys in experimental trials (Partridge 1997). However, provided that their physiological differences are properly acknowledged, the genetic tractability and convenient size of mice make them invaluable tools in DMD research.

Canine models

Dogs as models of muscular dystrophy

Spontaneous mutations of the dystrophin gene resulting in X-linked muscular dystrophy have been identified in several breeds of the domestic dog: the Golden Retriever (Cooper et al. 1988), the Rottweiler (B. Cooper in Partridge 1997) and the German short-haired pointer (Schatzberg et al. 1999); a beagle model has also been produced (Shimatsu et al. 2003). Of these, the Golden Retriever muscular dystrophic (GRMD) dog has been the most extensively studied and best characterized. Dogs are not ideal laboratory animals; colonies are expensive to maintain, the species is not genetically manipulable and is highly sentient and emotive. The GRMD dog, however, exhibits muscle changes which model human pathogenesis far more faithfully than the mdx mouse does (Cooper et al. 1988). Adult dogs also have a body mass that is comparable to DMD patients (Howell et al. 1997). For these reasons (and in the absence of other large models of DMD), the GRMD model has found favour in recent years.

Pathogenesis in the GRMD dog

GRMD pathogenesis manifests in utero with the development of lingual muscle lesions (Valentine et al. 1988; Nguyen et al. 2002). Extensive necrosis of the muscles of the limbs, trunk and neck can be identified from birth onwards (Nguyen et al. 2002). This is most pronounced from days 2–30, with the number of necrotic fibres declining as the animal reaches maturity (Cozzi et al. 2001).

During this period and up to 60 days (postnatal muscle maturation occurs at 60 days in normal dogs; this is delayed in comparison with other animals, including humans (Cozzi et al. 2001).), transforming growth factor-β1 levels are elevated and are thought to be involved in the initiation of fibrosis (Passerini et al. 2002). As with both the human disease and the mdx, creatine kinase levels are also extremely high, reaching a plateau at 100X normal by 6–8 weeks of age (Valentine et al. 1988). Expression of utrophin is reported to be abnormal (Wilson et al. 1994). By 6 months, severe fibrosis and joint contractures (e.g. in the tarsus) develop (Valentine et al. 1988; Kornegay et al. 1994); these features of the human disease are much less pronounced in the mdx model. As with DMD patients, young GRMD dogs frequently die from cardiac or respiratory failure, although some survive to reach several years of age (Nguyen et al. 2002).

Evaluation of experimental therapies in the GRMD model

The extensive homology between GRMD and DMD pathogenesis seems to qualify the dog as the most appropriate substitute for human patients in clinical trials. The small numbers of therapies investigated in both dogs and humans, such as myoblast transfer, have produced similarly negative results (Skuk & Tremblay 2001), whilst giving apparently encouraging results in mice. Other therapeutic approaches, such as gene transfer, are currently being evaluated in dogs with a view to establishing workable human protocols.

Preliminary gene therapy experiments established the theoretical possibility of introducing bacterial plasmid or adenoviral vectors into canine muscle (Howell et al. 1997), although the expression of the foreign genetic material was shown to occur at low frequency. High-level expression of a functional dystrophin mini-gene in canine muscle was later achieved using replication-deficient adenoviral vectors, an effect which was prolonged by cyclosporin immunosuppression (Howell et al. 1998). Prolonged maintenance of functional dystrophin in GRMD muscle has additionally been achieved through oligonucleotide therapy (Bartlett et al. 2000). It has also been demonstrated that, as with mdx mice, the expression of exogenous utrophin (also under cyclosporin immunosuppression) ameliorates the dystrophic phenotype of GRMD dogs by reducing fibrosis and increasing the expression of endogenous DGC proteins (Cerletti et al. 2003).

Limitations of canine models

For practical reasons, the GRMD is never likely to supersede the mdx in high-throughput studies. However, dogs can be used to evaluate the feasibility of applying mouse-developed technologies to subjects that more closely resemble DMD patients both in terms of size and in the pathological expression of the disease. The relatively positive results of GRMD gene transfer experiments have recently led to the initiation of similar clinical trials in human patients, although the results of these trials are not yet published (Thioudellet et al. 2002).

Feline models

Dystrophin deficiency in cats results in hypertrophic feline muscular dystrophy (HFMD). Similarly to mdx pathology, the skeletal muscle of the HFMD cat undergoes repeated cycles of degeneration and regeneration but does not develop the debilitating fibrosis that is characteristic of both DMD and GRMD. As its name suggests, one of the most obvious features of HFMD muscle is marked hypertrophy, which can ultimately lead to lethal oesophageal obstruction and kidney failure (Kohn et al. 1993; Gaschen & Burgunder 2001). Affected animals also suffer from cardiomyopathy (Gaschen et al. 1999). Whilst the HFMD cat is interesting in the sense that it provides another example of a species-specific response to dystrophin deficiency, it is not widely used as a model because it combines the practical disadvantages of being a large, sentient and emotive animal, with the scientific objection of having limited pathological similarity to DMD.

Primates

There are currently no primate models of dystrophin deficiency. Whilst nondystrophic primates have been used to optimize the conditions for treatments such as myoblast transfer (Skuk et al. 2000; Skuk et al. 2002c), they are not used in DMD research to any significant extent.

Non-mammalian models

Mammalian vs. non-mammalian models

Dystrophic pathology can only be accurately reproduced in mammalian models, which cannot easily be used for high-throughput studies and whose complexity and individual variation reduces the reproducibility of experiments. Non-mammalian models of disease such as the zebrafish (Chambers et al. 2001; Rubinstein 2003) and the nematode C. elegans (Baumeister & Ge 2002) oppose mammalian models in that whilst they have different musculature and pathology from human patients, they can be maintained in large numbers, are readily genetically manipulable and, particularly in the case of C. elegans, exemplify physiological simplicity. They are therefore used to perform elegant and reproducible experiments. Fortuitously, both species express a dystrophin homologue which has led to their use in both DMD-related gene analysis and drug discovery studies.

Zebrafish

Zebrafish are attractive as models of myopathy because they have a high skeletal muscle content and express orthologues of most human DGC proteins with similar membrane localization (Chambers et al. 2001; Guyon et al. 2003). Whilst gene-targeting technologies are not currently available, equivalent experiments can be carried out using oligonucleotide analogues (morpholinos) which disrupt the translation of specific mRNA transcripts. This technique has been used to create dystrophin-deficient zebrafish which have an unstable DGC, bent morphology and lower activity (Guyon et al. 2003), an interesting homologue of mammalian dystrophy.

C. elegans

Dystrophins have been identified in several invertebrate species, including C. elegans (Bessou et al. 1998; Chamberlain & Benian 2000), Drosophila (Neuman et al. 2001) and the sea urchin (Neuman et al. 2001). C. elegans moves by contraction of longitudinal striated muscles which express a dystrophin homologue called dys-1 (Bessou et al. 1998) that interacts with other DGC proteins (Gieseler et al. 1999; Chamberlain & Benian 2000). Mutations of dys-1 result in hyperactivity and hypercontraction and increased sensitivity to the neurotransmitter acetylcholine and its inhibitor (Bessou et al. 1998). The outstanding advantages of nematodes over other animal models are that they are by far the most readily genetically manipulable (since they undergo parthenogenesis the production of clones is also easy) and can be grown en masse in microtitre plates, making them ideal for high-throughput genetic and pharmacological studies (Baumeister & Ge 2002).

Conclusions

Since the discovery of the dystrophin gene in 1987 (Hoffman et al. 1987), much progress has been made in understanding the genetics and pathogenesis of DMD, and several genetic homologues have been identified. Sadly, this has not, so far, resulted in a cure for the disease, which has a 100% mortality rate. This failure could be attributed to the fact that the majority of research has been carried out in small animal models (largely the mdx mouse), which do not adequately model the pathology of the disease.

There is a dearth of large animal models of DMD, with the GRMD dog being the only real player. Whilst the GRMD dog is undoubtedly the most closely homologous model of the human disease, its scope is restricted because studies are necessarily expensive, small scaled and slow and currently exclude transgenic approaches. Inter-animal variation also makes standardization difficult. However, in the relatively recently available zebrafish and C. elegans models of muscular dystrophy, large-scale screening of potential pharmacological and gene therapies is entirely realistic. Whilst the evolutionary distance of these animals from humans precludes experimental successes from being directly transposed to the DMD patient, results can be used to identify probable approaches to be tested and refined in mammalian homologues.

The rapid rise of transgenic technology may eventually lead to the development of new large models of DMD myopathy. The pig is perhaps the most obvious candidate for such an approach because, whilst it is similar to humans in terms of size and physiology, it has practical advantages over the dog in that it can be intensively produced and is genetically manipulable (although the technology is in its infancy). Studies using such large animals could be carried out in parallel with larger scale screening using smaller models that give greater reproducibility.

As yet, there is no definitive model. However, the variety of DMD homologues available is differentially suited to addressing particular questions and therefore gives great scope for research.