Modifying muscular dystrophy through transforming growth factor-β



E. McNally, University of Chicago, 5841 S. Maryland MC6088, Chicago, IL 60637, USA

Fax: +1 773 702 2681

Tel: +1 773 702 2672



Muscular dystrophy arises from ongoing muscle degeneration and insufficient regeneration. This imbalance leads to loss of muscle, with replacement by scar or fibrotic tissue, resulting in muscle weakness and, eventually, loss of muscle function. Human muscular dystrophy is characterized by a wide range of disease severity, even when the same genetic mutation is present. This variability implies that other factors, both genetic and environmental, modify the disease outcome. There has been an ongoing effort to define the genetic and molecular bases that influence muscular dystrophy onset and progression. Modifier genes for muscle disease have been identified through both candidate gene approaches and genome-wide surveys. Multiple lines of experimental evidence have now converged on the transforming growth factor-β (TGF-β) pathway as a modifier for muscular dystrophy. TGF-β signaling is upregulated in dystrophic muscle as a result of a destabilized plasma membrane and/or an altered extracellular matrix. Given the important biological role of the TGF-β pathway, and its role beyond muscle homeostasis, we review modifier genes that alter the TGF-β pathway and approaches to modulate TGF-β activity to ameliorate muscle disease.


chronic obstructive pulmonary disease


dystrophin glycoprotein complex


Duchenne muscular dystrophy


extracellular matrix


epidermal growth factor


latency-associated peptide


limb girdle muscular dystrophy


large latent complex


latent transforming growth factor-β-binding protein


muscular dystrophy


regulatory Smad protein


single-nucleotide polymorphism


transforming growth factor-β


Muscular dystrophy (MD) is genetically diverse, and arises from mutations in many different single genes, leading to progressive loss of muscle mass, weakness, and eventually loss of muscle function. As in most Mendelian disorders, the primary genetic defect has the most significant effect in determining the age of onset of muscle degeneration and symptoms, the muscle groups most targeted, and, most importantly, the pace at which degeneration and loss of function occur. Some aspects of MD differ with the specific primary gene mutation, but recently there has been an increasing focus on understanding the genetic and molecular basis of the disease variability observed with the identical genetic mutation. Evidence for genetic modifiers of MD derives from both human and animal models, and this review will focus on data from the clinical arena and from models of MD, with a focus on Duchenne MD (DMD) and Becker MD, and the mechanisms by which the transforming growth factor-β (TGF-β) pathway has emerged as a critical pathway that determines outcome in MD. The identification of genetic modifier pathways not only informs the prognosis of individual patients, but also uncovers approaches that can be used to treat disorders for which there is presently no cure.

Skeletal muscle fibers are single, elongated, multinucleate cells that are arranged in parallel. Within a given muscle, myofibers tend to have a uniform fiber diameter, and appear polygonal in cross-section (Fig. 1). Skeletal muscle fibers are organized into fascicles, which are separated by epimysial connective tissue. Myofiber nuclei are generally peripherally located, below the sarcolemmal membrane. Skeletal muscle is dynamic and capable of regeneration after injury. In contrast to normal muscle, regenerating muscle is characterized by internally positioned nuclei and muscle fibers of varying diameter. Regeneration is central to repairing damage and restoring muscle function. However, regeneration has its limits, and may not be sufficient to fully restore the normal muscle architecture. In dystrophic muscle, regeneration is often outpaced by degeneration. Reflecting this, dystrophic muscle appears disorganized; in cross-section, skeletal muscle fiber includes both small and large fibers, and the myofiber nuclei are misplaced and centrally located within a myofiber. Disruption of the dystrophin complex by dystrophin or sarcoglycan gene mutations produces a leaky sarcolemma with markedly increased intracellular calcium. In advanced DMD muscle, fibroblasts and adipocytes replace the myofibers, and the inflammatory response is exacerbated, leading to fibrosis, muscle weakness, and muscle wasting (Fig. 1). The inflammatory response is seen in young DMD muscle, and is therefore an early feature of disease [1]. The mdx mouse model of DMD similarly has a chronic and persistent inflammatory response [2]. In addition to cytokines, there is marked upregulation of mRNA of macrophage products responsible for extracellular matrix synthesis (ECM) and turnover in the mdx limb [2]. The TGF-β pathway was seen in these gene expression profiles as a major proinflammatory and profibrotic cytokine that is important for regulating DMD.

Figure 1.

Fibrotic tissue and fatty and inflammatory infiltrate replace myofibers in advanced dystrophic muscle. Hematoxylin and eosin staining of human skeletal muscle reveals increased fatty and inflammatory infiltrate, and replacement of skeletal muscle fibers by fibrotic tissue (arrows), as DMD pathogenesis progresses from a mild to an advanced state.

Modifiers in multiple forms of human MD

Dystrophin gene (DMD) mutations underlie DMD, and the degree to which dystrophin remains expressed accounts for the milder allelic Becker MD [3, 4]. Even in the presence of premature stop codons, the primary mutation can be bypassed in some muscle fibers through a process known as exon skipping, resulting in a milder phenotype [5]. However, exon skipping and the production of revertant fibers does not entirely explain disease severity, because some patients with no detectable dystrophin production may express a milder phenotype than that predicted from their DMD mutation [6]. Many dystrophin mutations predictably lead to an absence of dystrophin production, but, for at least 16% of DMD mutations, it is possible that small amounts of dystrophin can be produced [3]. The wide range of DMD mutations associated with MD has made it challenging to detect genetic modifiers, as it is easier to identify modifiers with a more homogeneous primary genetic defect.

Limb girdle muscular dystrophy (LGMD) type 2C is caused by mutations in the dystrophin-associated protein γ-sarcoglycan encoded by the SGCG gene. Although a number of different SGCG mutations are responsible for LGMD 2C, there are at least two more frequent SGCG mutations documented in many LGMD 2C patients. The first mutation, Δ521-T, disrupts the reading frame of SGCG, resulting in the absence of protein production; this mutation has been documented in many LGMD 2C patients [7-9]. A second SGCG mutation is a G→A point mutation in codon 283 that replaces a conserved cysteine with a tyrosine in the epidermal growth factor (EGF)-like motif found at the extreme C-terminus in the extracellular domain of γ-sarcoglycan [10]. With these single mutations, individuals with the same mutation vary considerably in their age of onset of muscle weakness, age of ambulatory loss, the degree to which there is associated cardiomyopathy, and rate at which they lose respiratory muscle function [8, 9].

Other forms of MD are also associated with a range of phenotypic expression that cannot be explained by the primary mutation. For example, LGMD 2B and Miyoshi myopathy both arise from mutations in dysferlin, a membrane-associated protein implicated in muscle membrane repair [11, 12]. LGMD 2B has a variable age of onset, but often results in significant loss of muscle mass and reduced function, including loss of ambulation. In contrast, Miyoshi myopathy causes muscle wasting only of the gastrocnemius muscle in the calf, and typically does not cause loss of ambulation. The same dysferlin mutation has been associated with either LGMD 2B or Miyoshi myopathy [13, 14]. Similarly, mutations in the gene encoding lamin A/C (LMNA), a protein of the nuclear membrane, produce a range of muscle dysfunctions and associated cardiomyopathies, as well as electrical conduction system defects affecting the heart [15]. Through a genome-wide analysis of a large French family with a single LMNA mutation in 19 affected individuals, a locus on chromosome 2 was mapped that acted as a modifier [16]. Within this chromosome 2 locus is the DES gene encoding the intermediate filament protein desmin. DES mutations also cause cardiomyopathy and MD [17]. Cumulatively, these human genetic observations emphasize the high degree of variability in muscle disease severity seen in patients carrying an identical gene mutation, and suggest the presence of modifier genes underscoring the disease outcome.

Experimental models for MD also support the presence of modifiers

Both naturally occurring mutations and engineered mutations in MD animal models have been used to demonstrate that genetic backgrounds alter the pace and tempo of disease. There is evidence for genetic modifiers of MD in multiple species, including dogs, mice, fish, and flies. For example, golden retriever muscular dystrophy dogs show great clinical variability in the severity of disease. Affected dogs have a frameshift point mutation in a splice site in intron 6 of DMD, which results in complete absence of dystrophin [18]. Despite carrying the same primary DMD mutation, dystrophic dogs show a wide range of clinical manifestations; some dogs are severely affected and die perinatally, whereas some are less severely affected and live longer [18]. Furthermore, golden retriever muscular dystrophy dogs in a colony established from a single female carrier showed a varying degree of exercise capability, including significant variation in mean step length, maximum jumping height, and the time required to change position [18].

Mouse models of MD have been helpful in identifying genetic modifiers of disease. It has been shown that the most commonly used model of DMD, the mdx mouse, shows an enhanced phenotype when bred into the DBA/2J background [19]. The hindlimb muscles of DBA/2-mdx mice showed lower muscle weight, fewer myofibers and higher levels of fat and fibrotic tissue than C57Bl/10-mdx mice [19], indicating that the DBA/2J genome contributed to the severity of disease as compared with the mild phenotype of the mdx mice. However, mice are not the only dystrophic models that have been useful in identifying strain-induced differences in phenotype. Genetic manipulations in Drosophila melanogaster have also helped to identify modifier genes that interact with the dystrophin glycoprotein complex (DGC) and alter the fly wing phenotype [20]. A genetic screen was performed on three different mutant fly lines, with examination of phenotypic differences in the fly wing cross-vein phenotype, a visible phenotype that is easy to identify. Through this genetic screen, 37 genes were identified as modifiers of wing vein phenotype [20]. The 37 modifier genes clustered into six functional groups, including genes important for muscle development, neuronal/cell migration, motor function, and cytoskeletal components, and genes involved in the TGF-β and EGF receptor signaling pathway [20]. These findings provide a broader understanding of the DGC and interacting proteins.

Zebrafish express many of the same DGC proteins as humans, and the MD phenotype in zebrafish has been used for genetic interaction studies and for chemical screening [21]. The sapje fish mutant harbors a DMD mutation in a splice site at the end of exon 63 [21]. This mutation results in destabilization of the sarcoglycan complex, a common molecular finding in DMD patients, and results in disrupted muscle birefringence, a pattern that is highly visible and useful for chemical and genetic screens.

Mapping genetic modifiers with a genome-wide approach

Because the evidence from humans strongly supported the presence of a genetic basis for LGMD 2C, we took advantage of a mouse model for this disorder, the Sgcg mouse [22]. The Sgcg mouse model was generated by deleting exon 2 of Sgcg, which encodes the cytoplasmic and transmembrane domains, resulting in the absence of any detectable γ-sarcoglycan protein. This Sgcg mouse recapitulates the phenotype seen in LGMD 2C patients, with progressive muscle degeneration and a shortened lifespan. Like mdx mice, Sgcg mice are less severely impaired than their human equivalents, as mice remain ambulatory, unlike humans with these forms of MD. The Sgcg null mutation was bred through 10 generations into four different genetic backgrounds, where it was shown that three of these backgrounds, 129/SVEMS+/J JAX, C57BL/6J JAX, and CD41 VAF+(CD1), conferred a milder phenotype, and one background, DBA2J JAX, enhanced the phenotype of MD [23]. To determine MD severity, two different pathogenic traits were measured. The amount of Evans blue dye uptake by dystrophic muscle was determined as a measure of membrane leakiness, and fibrosis was quantified by assessing the amount of hydroxyproline present in the muscle.

With an unbiased, genome-wide analysis of single-nucleotide polymorphisms (SNPs) that differed between the mild and severe mouse strains carrying the Sgcg mutation, latent TGF-β-binding protein (LTBP)4 (encoded by Ltbp4) was identified as a modifier of muscle disease [24]. Importantly, the Ltbp4 genotype was linked to both muscle membrane leakiness and fibrosis, two features that were previously not thought to be directly related. The idea that the same genetic polymorphism regulates both muscle membrane stability and the amount of fibrosis in muscle indicates that these two features are more intimately related than previously appreciated. The specific mutation responsible for modifying MD is an insertion/deletion polymorphism that affects a domain that alters TGF-β release and therefore TGF-β activity. These results were replicated in a genome-wide analysis in a larger, independent cohort of mice [25]. This study confirmed that the region on chromosome 7 containing Ltbp4 correlated with an enhanced disease phenotype in all of the limb-based skeletal muscles studied. Additional genomic intervals were identified that influenced the severity of disease as it affects heart and trunk muscles [25]. These data point to different genetic bases for disease progression in limb-based skeletal muscle and trunk-based muscles.

Most recently, it was shown that LTBP4 polymorphisms translate to human dystrophinopathy. Human LTBP4 has two major haplotypes that differ by four amino acids, and these residues are hypothesized to modify TGF-β affinity [4]. Human LTBP4 and mouse Ltbp4 polymorphisms are different, but have the same net effect of regulating TGF-β activity. Those polymorphisms that increase TGF-β signaling are associated with greater disease intensity, whereas those that reduce TGF-β signaling ameliorate disease. The mouse Ltbp4 polymorphisms fall in the proline-rich hinge region of LTBP4, whereas the human LTBP4 SNPs are distributed at several points along LTBP4 but not in the hinge region (Fig. 2). In humans, LTBP4 SNPs have also been linked to chronic obstructive pulmonary disease (COPD) [26], colorectal cancer [27, 28], and abdominal aortic aneurysm [27]. The association of LTBP4 SNPs with multiple human diseases reflects the tissue distribution of LTBP4 expression. This constellation of LTBP4 target tissues – muscle, lung, and colon – is also reflected in mice engineered with a hypomorphic allele of Ltbp4 (Ltbp4S−/−) [29]. The Ltbp4S−/− mouse model was generated through a gene trap within the 5′-end of Ltbp4 that markedly reduced LTBP4 expression [29]. These mice develop cardiomyopathy, pulmonary dysfunction and fibrosis, and colon cancer, and these traits reflect the high level of LTBP4 expression in heart, lung, and colon [30].

Figure 2.

LTBP4 isoforms. LTBP4 differs at the 5′-end. Common LTBP4 proteins produced from alternative promoters are shown. The proline-rich hinge region is shown in red. Arrows indicate common SNPs that modify disease outcome in patients with COPD and DMD.

LTBPs and other members of the fibrillin superfamily of proteins are essential ECM proteins

LTBPs belong to the fibrillin superfamily of proteins. There are three fibrillins (fibrillin 1, fibrillin 2, and fibrillin 3) and four LTBPs (LTBP1, LTBP2, LTBP3, and LTBP4) in this superfamily. Features common to all family members include a large number of EGF repeats, many of which contain calcium-binding sequences [31] (Fig. 3). A distinguishing feature between the LTBPs and the fibrillins is the ability of LTBPs, specifically LTBP1, LTBP3, and LTBP4, to directly bind TGF-β. Fibrillins do not bind TGF-β directly, but instead bind LTBPs, and, in doing, so play a critical role in the signaling capacity of the ECM. Like LTBPs, the fibrillins are secreted to the ECM and contribute to the assembly and integrity of the ECM [31, 32]. The Ltbp4S/ genotype in mice also contributes to ECM formation, because elastin and its associated proteins cannot be integrated into the microfibril fibers [31]. These findings further confirm the intimate association between LTBPs, fibrillins, and the integrity of the ECM.

Figure 3.

Fibrillin superfamily of proteins. LTBP1–LTBP4 and FBN1–FBN3 together form the fibrillin superfamily of proteins. LTBPs and the fibrillins share a similar protein structure, and play an integral role in the stability of the ECM.

Autosomal dominant FBN1 mutations cause Marfan syndrome, an inherited connective tissue disorder [33]. Marfan syndrome is characterized by cardiovascular, musculoskeletal and ocular defects that are associated with aberrant TGF-β activity [34]. Fbn1 mutant mice with a Marfan-associated ‘knock-in’ substitution (Fbn1C1039G/+) have elevated TGF-β signaling and elevated pSmad2/3 activity in the lung, skeletal muscle, and heart [34]. The angiotensin II receptor antagonist losartan was shown to reduce TGF-β activity and protect against aortic root aneurysm progression, aortic wall thickening and mitral valve degeneration in Fbn1C1039G/+ mice [35]. Furthermore, losartan administration protects Marfan patients against aortic root dilation [36], and TGF-β receptor antagonist and losartan administration both restore muscle architecture in Fbn1C1039G/+ mice [37].

LTBPs show tissue-enriched expression patterns [38]. LTBP1 is predominantly expressed in the heart, placenta, lung, spleen, kidney, and stomach [38, 39]. LTBP2 is expressed in the lung, skeletal muscle, placenta, and liver [38]. LTBP3 and LTBP4 are both expressed in the heart, skeletal muscle, small intestine, and ovaries [30, 38, 40]. The expression pattern among the LTBPs suggests distinct functional roles in different tissues, and expression overlap in some tissues may provide redundancy to ensure biological function. Exogenously added LTBP1 was shown to stimulate aortic smooth muscle cell migration and thickening of arteries in diabetic rats [41], and elevated LTBP1 mRNA and protein levels have been found in atherosclerotic plaques [42]. In humans, LTBP2 mutations segregate with primary congenital glaucoma [43, 44], and have been shown to disrupt the ECM [45]. Interestingly Ltbp2 null mice die very early in development, consistent with a role in embryo implantation [46]. Ltbp3 null mice have reduced alveolar formation and reduced TGF-β signaling [47], and LTBP3 function is required for proliferation and osteogenic differentiation of human stem cells [48].

In humans, rare loss-of-function recessive alleles of LTBP4 result in a syndrome of impaired pulmonary function and gastrointestinal, musculoskeletal and dermal development [49]. These features are reminiscent of the features described in the Ltbp4S/ mice [29]. These findings support the idea that the developmental requirements for LTBP4 function are probably related to its importance for the activity of multiple TGF-β family members. In particular, the level of bone morphogenetic protein 4 is increased in the absence of LTBP4, consistent with a role of LTBP4 in regulating not only TGF-β activity, but also that of other TGF-β superfamily members [50]. The findings with loss-of-function alleles are in contrast to the common LTBP4 polymorphisms, which are associated with expansion of abdominal aortic aneurysm and impaired exercise tolerance in patients with COPD [26, 27]. We found that the two common LTBP4 alleles are associated with differential TGF-β signaling [4]. Given the chronic injury and inflammation that underlie DMD [2], it is fair to view DMD as a ‘hyper-TGF-β’ state. Reduced TGF-β signaling in both humans and mice with LTBP4 modifications is associated with decreased pathogenesis, especially fibrosis [4, 24]. The observation that a variant of Ltbp4 segregates with worse membrane damage and fibrosis in mice with LGMD [24] is consistent with this model, as a reduction in TGF-β signaling was associated with the improved phenotype. Similarly, a specific allele of human LTBP4 was associated with prolonged ambulation in DMD patients [4]. These data are highly consistent with the idea that TGF-β levels are elevated in DMD, and that partial reduction of this hyper-TGF-β signaling improves outcomes by stabilizing the plasma membrane of muscle and reducing fibrosis (Fig. 4).

Figure 4.

TGF-β secretion and bioavailability are tightly regulated. The secretion and activity of TGF-β are tightly regulated. In a basal state, TGF-β is bound to the LTBPs and kept inactive in the ECM. Upon muscle injury, TGF-β is activated and elicits downstream SMAD signaling, to repair the injury and restore muscle function. Dystrophic muscle is characterized by elevated TGF-β activity, which exacerbates the inflammatory response and aggravates the fibrotic response.

Association of LTBPs with TGF-β

LTBPs bind to and mediate the secretion of inactive TGF-β into the ECM. TGF-β is synthesized as a latent dimerized complex that is unable to engage its membrane-bound TGF-β receptor. The N-terminal prodomain, known as TGF-β latency-associated peptide (LAP), binds to the TGF-β dimer via noncovalent interactions, and keeps this small latent complex inactive. The small latent complex is covalently linked to LTBPs, forming the large latent complex (LLC) [51]. The association of LAP-TGF-β with an LTBP is essential for the secretion and activation of TGF-β. LTBPs not only mediate the secretion of TGF-β into the ECM, but also ensure its recruitment to the ECM microfibrils. The LLC associates with the ECM fibers, and keeps TGF-β inactive until its activity is needed [31, 51]. Furthermore, the association of the LLC with the ECM may provide cells with a readily available TGF-β pool that is positioned to respond to injury [38]. LTBPs are synthesized in molecular excess relative to TGF-β, further suggesting that most secretion of cellular TGF-β occurs through the LLC [52-55].

LTBPs associate with the ECM via an N-terminal ECM-binding domain. The ECM-binding domain contains transglutaminase substrate motifs, and transglutaminase is required for the covalent association of LTBPs with the ECM [56]. The N-termini of human LTBP1 and LTBP4 are alternatively spliced, conferring differential affinity for the ECM [30, 57]. There are two N-terminal isoforms of LTBP1 and of LTBP2 [39, 57, 58], and for LTBP1 the two isoforms have been shown to utilize different promoters [57, 59]. In the case of LTBP4, there are four isoforms that differ at the N-terminus (Fig. 3). The C-termini of the LTBPs have also been implicated in ECM binding, but this is less well understood [60].

The LTBPs have a highly repetitive structure, composed primarily of two distinct cysteine-rich motifs. The first motif is characterized by six cysteines, similarly to the EGF-like domains, and the second consists of eight intramolecularly bound cysteines, commonly referred to as the 8-Cys domain or TB domain, for TB domain [31]. Each LTBP contains four TB domains, separated by multiple EGF-like repeats. The first TB domain is commonly referred to as the ‘hybrid domain’, because it has similarities with both the TB domain and the EGF-like repeats. The second and third TB domains are separated by the proline-rich hinge region and multiple EGF-like repeats. The proline-rich hinge region is divergent among the four LTBPs [61]. LTBPs are cleaved at the hinge region by serine proteases in order to release TGF-β from the ECM [55, 60, 62, 63].

In order for TGF-β to be activated, cleavage of LTBPs must occur. Direct TGF-β activation has been demonstrated in vitro by proteolysis, enzymatic deglycosylation, and acid treatment [38]. The hinge domains of LTBP1 and LTBP3 are alternatively spliced, further affecting susceptibility to cleavage [42, 64, 65]. A splice variant of LTBP1 lacking 53 amino acids in the hinge region, including a heparin-binding site, shows diminished proteolytic cleavage [64]. The third TGF-β-binding domain directly binds to the LAP domain of TGF-β via disulfide bonds, and this mechanism has been described for TGF-β binding in LTBP1, LTBP3, and LTBP4 [58, 66]. How proteolytic cleavage in one region of LTBP alters TGF-β binding at a more distal region of the protein is not clear, and requires further study.

TGF-β signaling and implications for MD

TGF-β superfamily members transduce their signal from the membrane to the nucleus through distinct combinations of transmembrane type I and type II serine/threonine receptors and their downstream effectors, the SMAD proteins [67]. Ligand binding induces the type I and type II receptors to associate, which leads to a unidirectional phosphorylation event whereby the type II receptor phosphorylates the type I receptor, thereby activating its kinase domain [68]. Phosphorylated type I TGF-β receptor recruits regulatory SMAD proteins (R-SMADs) via interactions with SMAD anchor for receptor activation, which results in R-SMAD phosphorylation. Phosphorylated R-SMADs form a complex with common SMAD, a cytosolic protein containing a nuclear localization signal, and migrate to the nucleus to initiate gene transcription [68]. TGF-β also activates other signaling cascades, including the extracellular signal-related kinase, c-Jun N-terminal kinase, TGF-β-activated kinase 1, c-Abl and mitogen-activated protein kinase pathways. TGF-β family members are multifunctional polypeptide growth factors involved in the regulation of many important biological processes, such as growth, differentiation, immune response, secretion, and maintenance of the ECM components and these effects are paramount during injury response, and especially during fibrosis [69, 70]. TGF-β is rapidly induced upon cutaneous injury, and is consistently present in wound fluid throughout the repair process [71]. TGF-β release attracts neutrophils, macrophages, and fibroblasts, which, in turn, release more TGF-β. Expression levels of TGF-β and TGF-β receptors are elevated in fibroblasts of human postburn hypertrophic scars, in keloids that result from an excessive wound healing response, and in keloid-derived fibroblasts [72, 73]. TGF-β induces excess matrix synthesis when injected subcutaneously into mice [74]. Moreover, wound treatment with TGF-β promotes wound closure and scarring in vivo [74]. Treatment of incisional wounds with antibodies against TGF-β or antisense oligonucleotides suppresses ECM synthesis and scarring [75, 76]. Consistent with these observations, TGF-β activity has been shown to aggravate muscle disease states.

Enhanced TGF-β expression has been described in human DMD [77, 78]. Increased levels of TGF-β protein and mRNA are associated with increased canonical (SMAD) and noncanonical (non-SMAD) signaling in both human muscle and mouse models [78]. Increased TGF-β signaling is best described in DMD, but has also been described in other forms of MD [79]. How TGF-β signaling mediates the adverse effects on muscle pathology has not been shown. It is known that the c-Jun N-terminal kinase and extracellular signal-related kinase pathways participate in MD [80, 81], but the relationship to this signaling as a downstream consequence of TGF-β activation has not been determined. Systemic administration of neutralizing antibody against TGF-β or the angiotensin II type 1 receptor blocker losartan was found to normalize muscle architecture, repair and function in the mdx model, suggesting a direct role of excessive TGF-β signaling in muscle disease [37]. Inhibition of TGF-β activity with the same neutralizing antibody against TGF-β (1D11), losartan or a combination of both improved respiratory function in mdx mice [82]. TGF-β inhibition resulted in improved functional respiratory parameters, including normalized enhanced pause (Penh) values, increased peak respiratory flow, and decreased inspiration time and breathing frequency. In addition, administration of both 1D11 and losartan improved grip strength. Treatment with 1D11 proved to be effective in improving grip strength as early as 2 months of age, and treatment with losartan proved to be effective at 9 months of age [82]. Serum creatine kinase levels and hydroxyproline levels significantly decreased following 1D11 treatment, and diaphragm muscle fiber density increased, suggesting improved muscle function [82].

Genetic manipulation of the periostin gene, Postn, also ameliorates the MD phenotype and restores muscle function in mice lacking the δ-sarcoglycan gene (Sgcd /) [83]. Periostin is normally expressed in low amounts in adult tissue, and is upregulated by TGF-β; however, its expression is significantly increased in disease and during fibrogenesis [84]. Circulating levels of periostin are elevated in Sgcd / mice, and immunohistochemical analysis reveals accumulation of periostin in the ECM [83]. Mice lacking both δ-sarcoglycan and periostin (Sgcd /Postn/) show improved histopathology across multiple muscle groups, with no significant change in central nucleation, suggesting that loss of periostin does not interfere with myofiber regeneration [83]. In addition, loss of periostin results in reduced serum creatine kinase levels in Sgcd /Postn/ mice as compared with Sgcd / mice, and improved exercise performance [83]. Ablation of TGF-β signaling with a blocking mAb against TGF-β worsened muscle function in Sgcd /Postn/ mice as compared with mice receiving vehicle treatment, pointing to the conundrum associated with TGF-β activity observed in MD models. TGF-β is tightly regulated and, depending on the amount of active TGF-β that is biologically active, both beneficial and adverse effects have been reported.

Osteopontin is an ECM protein that regulates TGF-β. Deletion of the SPP gene encoding osteopontin has little phenotypic effect in muscle. When the SPP null allele was crossed into the mdx mouse model of DMD, there was a marked reduction in fibrosis and improvement in muscle strength [85]. In addition, deletion of SPP reduced neutrophil but not macrophage invasion into dystrophic muscle and reduced TGF-β mRNA levels [85]. These data identify SPP as a regulator of the inflammatory response, a contributory factor in promoting disease progression in dystrophic muscle. A polymorphism within the SPP promoter was associated with prolonged ambulation in a cohort of DMD patients [86]. The G allele of rs28357094 altered gene promoter function, and was associated with reduced osteopontin mRNA levels in HeLa cells, and, paradoxically, with reduced levels of CD4-positive and CD68-positive cells in DMD [87]. TGF-β has been shown to activate the promoter of SPP1 [88], and a polymorphism in the TGFBR2 promoter correlated with osteopontin mRNA levels, further confirming an interplay between osteopontin and TGF-β [87].

Genetic manipulation of TGF-β and SMAD signaling helps to restore normal heart and muscle function in γ/δ-sarcoglycan null flies (Sgcd[840]) [89]. Partial reduction of SMAD signaling with haploinsufficient alleles that reduced SMAD activity in Sgcd[840] flies improved negative geotaxis, i.e. the ability of the flies to walk upwards against gravity, an ability that is lost in both Sgcd mutant flies and those lacking dystrophin [89-91]. Furthermore, optical coherence tomography showed that reducing TGF-β and SMAD signaling in Sgcd[840] flies restored heart function to wild-type levels. Interestingly, genetic manipulation of various downstream targets of TGF-β signaling revealed that TGF-β signaling involving bone morphogenetic protein has a direct role in improving skeletal muscle function but not heart tube function [89]. These findings further emphasize the intricate nature of TGF-β signaling and the crosstalk between various signaling mechanisms in MD.

Future therapeutic directions

Human patients and animal models with MD confirm the integral role of TGF-β and SMAD signaling in the progression and severity of muscle disease. Mechanisms to reduce this signaling have focused on pharmacological approaches through angiotensin receptor blockade. Interestingly, in the Marfan model, it has been suggested that noncanonical TGF-β signaling may be most beneficial [92, 93]. These data are in contrast to what has been shown with the Drosophila model of MD, where canonical TGF-β signaling was shown to be important for the progression of heart and muscle disease [89]. Whether these differences reflect the underlying differences in the invertebrate system or the differences between vascular tissues and striated muscle is yet unclear. Further studies are needed to determine whether a reduction in TGF-β signaling is required, in order to more fully dissect which intracellular pathways are most beneficial for treating MD.


This work was supported by NIH NS071848, HL61322, AR052646, NS072027, and the Doris Duke Charitable Foundation.