INHIBITION OF MYOSTATIN WITH EMPHASIS ON FOLLISTATIN AS A THERAPY FOR MUSCLE DISEASE

Abstract

In most cases, pharmacologic strategies to treat genetic muscle disorders and certain acquired disorders, such as sporadic inclusion body myositis, have produced modest clinical benefits. In these conditions, inhibition of the myostatin pathway represents an alternative strategy to improve functional outcomes. Preclinical data that support this approach clearly demonstrate the potential for blocking the myostatin pathway. Follistatin has emerged as a powerful antagonist of myostatin that can increase muscle mass and strength. Follistatin was first isolated from the ovary and is known to suppress follicle-stimulating hormone. This raises concerns for potential adverse effects on the hypothalamic–pituitary–gonadal axis and possible reproductive capabilities. In this review we demonstrate a strategy to bypass off-target effects using an alternatively spliced cDNA of follistatin (FS344) delivered by adeno-associated virus (AAV) to muscle. The transgene product is a peptide of 315 amino acids that is secreted from the muscle and circulates in the serum, thus avoiding cell-surface binding sites. Using this approach our translational studies show increased muscle size and strength in species ranging from mice to monkeys. Adverse effects are avoided, and no organ system pathology or change in reproductive capabilities has been seen. These findings provide the impetus to move toward gene therapy clinical trials with delivery of AAV-FS344 to increase size and function of muscle in patients with neuromuscular disease.

Strategies to increase muscle size and strength through inhibition of the myostatin pathway show promise for clinical application.³⁴ Follistatin is a potent antagonist of myostatin that takes advantage of its ability to hinder access to signaling receptors on skeletal muscle. The muscle-building properties of follistatin are well demonstrated,³⁶ but because it is a peptide with multiple functions, concerns have been raised regarding off-target effects when considering its appropriateness for treatment of muscle disease. The goal of this review is to thoroughly discuss these complex interactions and demonstrate a strategy that takes advantage of known follistatin properties that can be harnessed to promote efficacy to increase muscle mass and muscle strength in the absence of adverse clinical effects.

Emphasis on myostatin inhibition emerges because treating muscle disorders by most pharmacologic approaches has been disappointing. Androgen steroids, popular among athletes, pose long-term risks⁶⁶ including: (1) endocrine (gonadal atrophy and sterility)²⁸; (2) somatic (changes in blood lipid profiles and cardiac hypertrophy)^3,30,37; and (3) neuropsychiatric (anxiety, depression, hostility, paranoia)⁵⁷; and attempts to treat muscle disorders have been disappointing.⁵ Glucocorticosteroids, the only beneficial drug treatment for muscular dystrophy, are virtually entirely targeted toward the Duchenne muscular dystrophy (DMD) population.^46,50 Even in this patient group the mechanism of benefit is poorly understood, and the evidence that muscle mass is increased is meager.⁴⁶ For genetic muscle diseases, gene manipulation strategies are on the horizon, including gene replacement,^12,19,20,62 exon skipping,^1,44 and mutation suppression.^7,23 Despite enthusiasm, experimental studies suggest that these approaches usually fall short of returning function to normal.⁴⁰ Combinational approaches that include partial correction of the underlying defect (i.e., micro-dystrophin) combined with increasing muscle size and strength appear to offer more.² For muscle diseases where correction of the underlying defect might not be an option, increasing muscle size and strength may be opportune for both genetic and acquired muscle diseases where treatment options are limited. Examples include some forms of muscular dystrophy where gene manipulation strategies are not yet applicable (e.g., facioscapulohumeral dystrophy, FSHD), acquired disorders such as sporadic inclusion body myositis, where pharmacologic treatment failures predominate, or cachectic disorders related to cancer or aging that may be ideally suited for a muscle-enhancing approach.

MYOSTATIN PATHWAY

The potential for follistatin as a therapeutic agent for muscle disease cannot be fully understood without knowledge of the myostatin pathway. Myostatin is a member of the transforming growth factor-beta (TGF-β) superfamily of signal peptides. It is expressed specifically in developing and adult skeletal muscle.⁴⁵ During development, myostatin expression limits the size of the muscle in concert with multiple factors that sculpt the limbs in relation to skeletal, vascular, and ectodermal patterns of growth.⁴ Myogenic cells respond to myostatin by downregulating the expression of Pax-3 and Myf-5, important transcriptional regulators of myogenic cell proliferation, and Myo-D, an early marker of muscle differentiation. In their sentinel report in 1997, McPherron et al.⁴⁵ demonstrated the biological effect of targeted disruption of growth and differentiation factor-8 (GDF-8) gene in the mouse. GDF-8 null mice were significantly larger in size than wildtype animals, and there was widespread increase in skeletal muscle mass (Fig. 1). Individual muscles of mutant mice weighed 2−3 times more than those of wildtype animals. The increase in mass was the result of a combination of muscle hypertrophy and hyperplasia. These experiments established the GDF-8 peptide as a major player for inhibiting muscle growth, with the designated name “myostatin.”

FIGURE 1.

Myostatin null animals exhibit increased muscle mass. Adult myostatin null mice demonstrating increased size (right) as compared to wildtype (left) animals. Reprinted with permission from Lee SJ, McPherron AC. Curr Opin Genet Dev 1999:5:604−607.

MYOSTATIN SYNTHESIS

The human myostatin gene (MSTN) maps to chromosome 2q32.2.⁶⁷ The gene contains three exons and three putative transcription start sites that encode a 376-amino acid precursor protein composed of a signal peptide, an N-terminal propeptide domain and a C-terminal domain that gives rise to the active peptide (Fig. 2). Myostatin activation requires stepwise proteolytic cleavages of the precursor protein. Initially, furin family enzymes remove the signal peptide (24-amino acid). A second cleavage event at amino acid sites 240−243 leaves two fragments: an N-terminal propeptide domain of 27,640 Da and C-terminal domain of 12,400 Da destined to become the active myostatin protein.³⁴ Parallel fragments of the myostatin C-terminal are linked through a disulfide bond, referred to as the myostatin C-terminal dimer that remains noncovalently complexed to the N-terminal propeptide.^33,71 This noncovalent complex circulates in the blood and maintains the myostatin C-terminal dimer in a latent, inactive state.^25,33 A third cleavage at amino acid 76 is required for the myostatin C-terminal to become active.⁷⁹ This occurs via a different enzyme group, a metalloproteinase that belongs to the bone morphogenic protein (BMP)-1/tolloid (TLD) family.

FIGURE 2.

Blocking the myostatin pathway. Myostatin (M) activation requires stepwise proteolytic cleavages of the precursor protein. After the signal peptide (SP) is removed, a second cleavage event leaves two fragments: an N-terminal propeptide domain of ≈28 kD and C-terminal domain of 12.5 kD destined to become the active myostatin protein. Parallel fragments of the myostatin C-terminal are linked through a disulfide bond, referred to as the myostatin C-terminal dimer that remains noncovalently complexed to the N-terminal propeptide. This noncovalent complex circulates in the blood maintaining the myostatin C-terminal dimer in a latent, inactive state. A third cleavage at amino acid 76 affects the ability of the propeptide to bind the active C terminal domain. Myostatin can be found in the serum or locally in an inactive state when bound to follistatin (FS), follistatin-related gene (FLRG), and growth and differentiation factor-associated serum protein-1 (GASP-1). These peptides block the activation of the myostatin pathway. If the pathway is not inhibited, the active myostatin dimer binds to the activin receptor type IIB (ActRIIB), which then recruits and activates by transphosphorylation the type I receptor (ALK4 or ALK5). Smad2 and Smad3 are subsequently activated and form aggregates with Smad4 and then are translocated to the nucleus, activating target gene transcription.

Myostatin signaling acts through the activin receptor type IIB (ActRIIB) on skeletal muscle by setting in motion an intracellular cascade of events. First, there is presumed recruitment of a type I co-receptor.³⁴ Activin receptor-like kinases 4 and/or 5 (ALK-4, ALK-5) represent candidate coreceptors that are phosphorylated by ActRIIB.⁵⁹ This in turn leads to phosphorylation of TGF-β specific Smads 2 and 3 that form a complex with Smad 4. The Smad 2/3/4 complex is translocated to the nucleus to regulate expression of targeted genes such as MyoD and myogenic regulatory factors (MRFs) (Fig. 2).^32,33,42

Apart from an essential role in muscle growth, recent evidence indicates that myostatin has a regulatory role in skeletal muscle fibrosis. Li et al.³⁸ revealed that myostatin and the ActRIIB receptor are expressed on muscle fibroblasts, thus inducing their proliferation and the production of extracellular matrix proteins. This proliferation leads to the induction of the canonical Smad signaling pathway in fibroblasts by Smad3 phosphorylation and downstream p38 MAPK and Akt pathways.³⁸ This enhances the therapeutic potential for myostatin inhibition that could lead to muscle enlargement while at the same time decreasing muscle fibrosis. In many muscle disorders, active fibrosis leads to the irreversibility of the condition, be it inherited or acquired.

FOLLISTATIN SYNTHESIS, ISOFORMS, AND PHYSIOLOGIC ROLE

Follistatin, secreted as a glycoprotein, was originally identified in porcine ovarian follicular fluid and received its name because it suppresses synthesis and secretion of follicle-stimulating hormone (FSH) from the pituitary gland.⁵⁶ It is highly conserved, with overall species homology of 83% and 95% in mammals. Two groups isolated and published their results in 1987. One coined the term follistatin,¹⁴ and the other named it FSH-suppressing protein (FSP).⁶¹ With time, follistatin became the popular designation, but the name hardly does justice to a peptide with functions that extend beyond FSH suppression.

The follistatin gene localizes to chromosome 5q11.2. It is composed of a relatively small 6-kb genomic DNA consisting of six exons. There is an alternative splice site that generates two major species, a full-length version that encodes a 344-amino acid preprotein differing by a 27-amino acid sequence from its carboxy-shortened version of the 317-amino acid form missing exon 6 (Fig. 3).^64,65 Prior to activation, follistatin, like myostatin, undergoes further posttranslational modification to lose another 29 amino acids by removal of the signal peptide that results in polypeptides of 315 (FS315), often referred to as the long isoform and 288 (FS288), called the short isoform. There is also evidence to suggest that FS315 can be proteolytically cleaved in vivo at the carboxy-terminal to give an intermediate isoform of 303 amino acids.⁶⁹

FIGURE 3.

The follistatin gene consists of six exons. Alternative splicing generates two isoforms, FS317 and FS344. Alternative splicing occurs at the 3′ end of the gene between exon 5 and exon 6. Splicing out of intron 5 generates a stop codon immediately following the last amino acid of exon 5, and leads to the termination of the coding sequence for FS317. An alternative splice site results in the inclusion of exon 6 and generates FS344. After translation and prior to activation, follistatin undergoes further posttranslational modification by cleavage of the 29 amino acid signal peptide. This results in polypeptides FS315 (long-isoform from FS344) and FS288 (short-isoform from FS317).

At the time follistatin was first isolated, little was known of its mechanism of action. In a major breakthrough, follistatin was found to be an activin-binding protein.⁵² An important function of follistatin is its collaborative role in reproductive physiology with other TGF-β superfamily members, activin and inhibins. These TGF-β family peptides have overlapping autocrine/paracrine functions. All three were initially purified from gonadal fluids and characterized based on their ability to modulate FSH. In addition to gonadal sites of production (ovary/testes), these peptides are all produced by cells in the hypothalamic–pituitary axis (gonadatropes and folliculostellate cells). Follistatin binds activin and attenuates the release of FSH. Activin is secreted by the follicle of the ovary and serves to enhance FSH secretion. Inhibins, which are secreted in two forms (A and B), inhibit the release of FSH at the hypothalamic–pituitary level. In addition, it is well documented that follistatin can abrogate the effects of GnRH in stimulating FSH secretion.^77,78 This is also due in part to the blocking of transcriptional activation of the GnRH receptor gene by activin.¹⁶

This complex interaction of follistatin in relation to pituitary and gonadal function has raised concerns about its potential use as a therapeutic agent in the clinic. However, potential recombinant products can take advantage of differences between the isoforms in their ability to bind heparin sulfate. A well-recognized follistatin heparin-binding site is present at residues 72−86, which is a region rich in basic amino acids.⁷³ In contrast, the carboxy-terminal 27 amino acid sequence of FS-315, composed of 44% acidic amino acids, interferes with the heparin site.⁷⁰ These considerations take on a novel perspective with regard to gene therapy considering potential transgene products. FS-288, the shorter alternatively spliced product has an ≈10-fold higher affinity to activin compared to FS-315.^24,69,70 In addition, FS-288 targets heparin sulfate proteoglycan binding sites at cell surfaces, while FS-315 represents a soluble serum-based or circulating follistatin isoform.⁶³ In developing a gene therapy product for clinical use, we have taken advantage of this property. In our preclinical research studies, adeno-associated (AAV) virus that carries cDNA FS-344 delivers a gene therapy product (FS-315) without interruption in reproductive capabilities in either males or females in species ranging from mice to monkeys. Our results support prior observations that impairment of activin binding is more closely allied with FS-288 and its cell surface-binding properties mediated by heparan sulfate proteoglycans. This strategy greatly enhances the margin of safety for clinical trials because the FS-315 isoform has a limited effect on activin modulation by protecting the pituitary–gonadal axis from unwanted alterations. The same can be said for avoiding off-target effects mediated by cell surface binding of follistatin, including functions related to cellular differentiation, repair, and apoptosis.⁵

The origin of follistatin under normal physiologic conditions is not entirely understood. Clearly, follistatin is produced locally in the pituitary gland and in gonads, ovaries, and testes. Overall, measurements of follistatin during the menstrual cycle show few changes.^15,18,29 However, a notable exception is during pregnancy, when follistatin concentrations rise toward term in parallel with activin.^15,76,79 Follistatin is widely distributed throughout multiple organs, and the majority of follistatin found in the circulation is likely secreted from the walls of blood vessels.

GENETICALLY ALTERED MICE OVER- AND UNDEREXPRESSING FOLLISTATIN

A component of understanding the functional role of follistatin can be gleaned from studies of genetically modified mice. Studies that evaluate the overexpression of the follistatin gene through genetically induced gain-of-function mutations are worth study to examine the potential for off-target effects. However, information derived from such models requires cautious interpretation because of species differences and influences of overexpression during development that are not clinically relevant. Despite caveats, the findings in a transgenic model in which the follistatin gene was introduced under control of a muscle-specific myosin light chain promoter are encouraging.³³ Muscle mass increase was significantly greater than observed in the myostatin null mutant mouse. These results suggest that at least part of the effect of follistatin results from impact on another pathway independent of myostatin inhibition. This hypothesis is reinforced by additional studies in which mice that overexpress transgenic follistatin were crossed with myostatin null animals.³⁶ The resulting phenotype appeared to be additive, with a quadrupling of muscle mass in follistatin⁺/myostatin^−/− mice, outstripping the effects of either myostatin nulls or follistatin-overexpressing animals alone. These findings emphasize that other signaling pathways could be exploited to increase muscle size and strength.³⁶

In another gain-of-function mutant mouse line, the metallothionein (MT)-1 promoter was placed upstream of the follistatin gene.²¹ Observations on gonadal changes in this model are thought-provoking, but they may not be directly related to clinical translational considerations. On a positive note, the MT-follistatin transgenic offspring were viable and developed to adults. In addition, no deleterious effects were seen in any organ system other than gonadal tissue. In males testes size was decreased, with variable Leydig cell hyperplasia, an arrest in spermatogenesis and seminiferous tubular degeneration leading to infertility. Females had thin uteri and small ovaries, and many became infertile. Interpreting these results in the context of what we would anticipate in a clinical trial of gene therapy requires caution. In the patient, our concern would be that high serum levels of follistatin would bind activin and result in reduced serum FSH levels leading to gonadal dysfunction. Instead, the follistatin-overexpressing mice had normal FSH levels. This implies that in the transgenic mouse model, follistatin overexpression exerted its effects through gene expression in the matrix of the end organ, disrupting local regulation. This would not parallel the gene therapy paradigm, where the transgene product, FS344, is nontissue bound, and has no effect on reproductive function.

On the opposite side of the spectrum, a loss-of-function mutant mouse was created by a targeted deletion of the follistatin gene.⁴³ The mutant mice survived until birth but died within hours of delivery. Defects included growth-retardation and shiny, taut skin. There was poor whisker development, hyperkeratosis of skin, abnormal tooth development, defects in the hard palate, and reduced size of intercostal and diaphragm muscles. The central and peripheral nervous systems, however, were intact. At best, this short-lived model has little relevance to our goals of studying follistatin as a potential therapeutic agent, but it does reinforce that follistatin may mediate effects through pathways of the TGF-β family other than myostatin.

TRANSLATIONAL APPROACHES THAT INHIBIT MYOSTATIN IN MUSCULAR DYSTROPHY

Increasing muscle size and strength in animal models of muscle disease in experimental translational studies has important implications for neuromuscular patients. Table 1 summarizes the strategies that have been used to reinforce the potential for this approach. In the mdx mouse, a model for DMD, there was an increase in muscle size and strength using a monoclonal antibody that inhibited myostatin.⁸ In further support of myostatin inhibition in muscular dystrophy, the mdx mouse was crossed with the myostatin knockout, resulting in muscles of larger size associated with improved grip strength.⁷⁴ An impressive outcome in this model was reduced fibrosis in the diaphragm muscle, a potential clinically meaningful result. Another approach to achieve myostatin blockade utilized the propeptide to keep the myostatin C-terminal dimer inactive, blocking access to the ActRIIB receptor. In one experimental paradigm, the propeptide was stabilized by fusion to IgG-Fc and systemically administered to the mdx mouse, which resulted in increased muscle size concomitant with treatment.⁹ In a second strategy, AAV was used to deliver a mutant myostatin propeptide in a mouse model of limb girdle muscular dystrophy (LGMD)2A that harbored a mutation of calpain-3. The mutation prevented cleavage of the propeptide and maintained the myostatin C-terminal domain in an inactive state. The calpain-3 deficient mice had increased muscle mass and improved force generation.⁶

Table 1.

Translational and clinical studies of myostatin inhibition.

Preclinical study	MD model	Conclusions	Reference
Functional improvement of dystrophic muscle by myostatin blockade	mdx	Monoclonal antibody to myostatin resulted in increased muscle weight, size, and absolute strength in mdx mice	8
Loss of myostatin attenuates severity of muscular dystrophy in mdx mice	mdx;mstn −/−	Increased muscle mass was maintained over time in the Mstn −/− mdx mouse resulting in increased grip strength and reduced fibrosis in the diaphragm	74
Myostatin propeptide-mediated amelioration of dystrophic pathophysiology	mdx	Pharmacological blockade with a myostatin propeptide stabilized by fusion to IgG-Fc improved pathophysiology in mdx	9
Muscular atrophy of caveolin-3–deficient mice is rescued by myostatin inhibition	CAV3 Tg	CAV3 normally suppresses the myostatin-mediated signal, thereby preventing muscular atrophy. In CAV3 mutant, atrophy is prevented by overexpressing myostatin propeptide or by IP injection of ActRIIB receptor	53
AAV-mediated delivery of a mutated myostatin propeptide ameliorates calpain 3 but not alpha-sarcoglycan deficiency	capn3 −/−sgca −/−	Delivery of mutated myostatin propeptide by AAV increased force generation, in capn3 −/−deficient mice. sgca −/− mice did not exhibit increased muscle mass or protection from Evans blue dye uptake	6
Myostatin blockade improves function but not histopathology in a murine model of limb-girdle muscular dystrophy 2C	sgcg −/−	Antibody mediated myostatin blockade led to increased fiber size, muscle mass, and absolute force with no improvement in muscle histopathology	10
Age-dependent effect of myostatin blockade on disease severity in a murine model of limb-girdle muscular dystrophy	sgcd −/−	Early loss of myostatin achieved by monoclonal antibody administration or by gene deletion improved muscle mass regeneration in scgd −/−; late loss demonstrated no improvement	54
Elimination of myostatin does not combat muscular dystrophy in dy mice but increases postnatal lethality	dyW/dyW	dyW/dyW mice lacking myostatin showed increased muscle regeneration muscle mass, but demonstrated increased mortality	39
Long-term enhancement of skeletal muscle mass and strength by gene transfer of follistatin	mdx, wild-type	Gene transfer of myostatin inhibitory proteins increased muscle mass and strength >2 years	22
A phase I/II trial of MYO-029 in adult subjects with muscular dystrophy	BMD, FSHD, LGMD patients	Safety was demonstrated in clinical trial assessing neutralizing antibody to myostatin, MYO-029 in adult muscular dystrophy patients. Clinical efficacy was not found but bioactivity was demonstrated by DEXA and histology in a limited number of subjects	75

BMD, Becker muscular dystrophy; capn3^−/−, calpain3 deficient mouse; CAV3 Tg, caveolin-3 transgenic mouse deficient for caveolin-3; DEXA, dual-energy X-ray absorptiometry; dy^W/dy^W, laminin-alpha2 (merosin) deficient mouse; FSHD, facioscapulohumeral dystrophy; IP, intraperitoneal; LGMD, limb girdle muscular dystrophy; mdx^−/−, Duchenne muscular dystrophy mouse model deficient for dystrophin; mstn^−/−, myostatin knockout mouse; sgca ^−/−, alpha-sarcoglycan deficient mouse; sgcd ^−/−, delta-sarcoglycan deficient mouse; sgcg ^−/−, gamma-sarcoglycan deficient mouse.

Other investigations have tempered enthusiasm and cautioned beneficial treatment strategies given the age and stage of dystrophy at treatment. For example, in a model of delta-sarcoglycan deficiency (scgd^−/−) representing LGMD2F, loss of myostatin activity in very young mice using monoclonal antibody administration or gene deletion (mstn^−/−; scgd^−/−), increased muscle size, improved muscle regeneration, and reduced fibrosis.⁵⁴ However, antibody-mediated myostatin inhibition at a late stage in the scgd^−/− mouse failed to prevent progression of muscle pathology despite increasing muscle size. Another disappointing result was observed in the dy^W/dy^W mouse,³⁹ a mouse model for merosin deficient congenital muscular dystrophy (MDC1A).⁴⁷ The dy^W/dy^W mouse exhibits poor muscle regeneration³¹ that might be expected to show a favorable response to myostatin inhibition. However, attempts to increase muscle strength and prolong life-span by eliminating myostatin by crossing the dy^W with the myostatin null mouse were disappointing. The homozygous dy^W/dy^W;mstn^−/− improved muscle regeneration, but dystrophic features including necrosis, inflammation, and fibrosis were unaffected and the pre-weaning mortality was increased. The poor outcome was attributed to a reduction of brown fat in the neonatal period that was made worse by the additive effects of the underlying mutation combined with myostatin inhibition (dy^W/dy^W; mstn^−/−). The gloomy prediction of this study should be balanced by findings that myostatin effects on adipocytes are influenced by age and gene targeting. For example, in adult mice the administration of myostatin has no effect on fat tissue.⁶⁸ When myostatin overexpression is achieved with a muscle-specific promoter (MCK or MCK-3E) there are no effects on adipocytes.⁶⁰

In another model of inherited muscle disease related to caveolins, the findings in translational studies set the stage for a possible favorable outcome for a myostatin inhibition clinical trial. Caveolins are integral membrane proteins and are the principal components of the sarcolemmal invaginations referred to as caveolae. These proteins play important roles in signal transduction and vesicular trafficking.^17,55 Caveolin-3 (CAV3) gene mutations cause a spectrum of muscle disorders, including: autosomal dominant LGMD1C, rippling muscle disease (RMD), sporadic and familial forms of hyperCKemia, and distal myopathy.¹¹ In a transgenic mouse mutant for P104L of CAV3, phosphorylated Smad2 was increased.⁵³ These findings suggest that CAV3 normally suppresses the myostatin-mediated signal, thereby protecting the muscle from atrophy. Blocking myostatin by crossing CAV3 mutants with transgenics that overexpress the myostatin propeptide (CAV-3P104L/MSTNPro) or by intraperitoneal administration of the soluble ActRIIB receptor prevented muscle atrophy with evidence of suppression of phosphorylated-Smad2. These findings potentially bode well for myostatin inhibition in patients with CAV3 mutations.

The translational study cited above that looked at the potential efficacy of the neutralizing monoclonal antibody to myostatin (MYO-029) has been studied in a double-blind randomized clinical trial. One hundred sixteen muscular dystrophy subjects with varied illnesses including LGMD, Becker muscular dystrophy, and facioscapulohumeral dystrophy were included.⁷⁵ Subjects were divided into sequential dose-escalation cohorts (Cohort 1 at 1 mg/kg; Cohort 2 at 3 mg/kg; Cohort 3 at 10 mg/kg; Cohort 4 at 30 mg/kg). MYO-029 showed good safety and tolerability, with the exception of a cutaneous hypersensitivity rash at the 10 and 30 mg/kg doses. There was no improvement in muscle strength or function, but the study was underpowered for efficacy.

Nevertheless, bioactivity of MYO-029 was supported by a trend in a limited number of subjects that demonstrated increased muscle size using dual-energy radiographic absorptiometry and muscle histology with a dose-dependent increase at 3 mg/kg and 10 mg/kg doses. This was the first clinical trial based on the hypothesis that systemic administration of myostatin inhibitors could be an appropriate strategy for patients with a variety of muscle disorders. Most important, there were no significant organ-related adverse events beyond the hypersensitivity skin reactions to the antibody. This study clearly paves the way for more potent myostatin inhibitors to be used for stimulating muscle growth in muscular dystrophy.

FOLLISTATIN EFFECTS ON MUSCLE ENHANCEMENT THROUGH GENE THERAPY

A gene therapy approach to myostatin inhibition represents an important consideration for patients with muscle disease. In contrast to pharmacologic administration of myostatin inhibitors such as neutralizing antibodies,^8,69 or recombinant pharmacologic agents that have been used in translational studies (propeptide,^9,58 soluble ActRIIB receptor,³⁵ or drugs such as trichostatin A⁴⁸), gene therapy offers the potential for a single administration of vector carrying the follistatin gene with persistent expression for many years, even perhaps throughout the lifetime of the individual. AAV is the vector of choice for muscle disease. It has a proven safety record in more than 40 clinical trials in a diverse group of diseases⁵¹ and demonstrated efficacy in Leber's congenital amaurosis.⁴¹ As muscle is the natural host for AAV, the advantage for treating a wide range of muscle disease is obvious. Several clinical trials that incorporate direct muscle injections are under way for alpha-1-anti-trypsin deficiency as well as Duchenne and LGMD, and no adverse effects have been encountered (unpubl. obs., Jerry Mendell, Barry Byrne, Terrence Flotte). While there is controversy over the immunogenicity of AAV,⁴⁹ this is not an insurmountable obstacle, since immune suppression is being studied in viral gene transfer in the ongoing clinical trials.

An advantage of myostatin inhibition is the potential to treat both genetic and acquired diseases. Achieving clinically meaningful outcomes using current means of gene delivery to single muscle targets is also a realistic goal. An example of a potential recipient of this approach is the patient with quadriceps (knee extensor) muscle weakness who is disabled by frequent falls. These patients also suffer knee pain because of joint stress from genu recurvatum (back kneeing). Examples of muscle disorders that are predisposed to weakness of knee extensor muscles include myotonic muscular dystrophy type 1, Becker muscular dystrophy, and sporadic inclusion body myositis.

The therapeutic potential of AAV-mediated follistatin gene therapy has been extensively studied in our laboratories at the Center for Gene Therapy, Nationwide Children's Hospital (Columbus, Ohio), with demonstrable clinical implications.²² The follistatin gene used for these studies is the alternatively spliced cDNA FS-344 (Fig. 3). The final product of this transgene is the FS-315 circulating isoform. Because of its reduced affinity for heparin,^70,69 cell surfaces of the gonadal–pituitary axis are much less likely to be affected, thus increasing the safety profile for this transgene product. This is reflected in our preclinical studies, as will be described.

To determine the optimal gene product for clinical advancement, follistatin was directly compared with other myostatin-inhibitory genes, including growth and differentiation factor-associated serum protein-1(GASP-1)²⁶ and the follistatin-related gene (FLRG).⁷² The quadriceps and tibialis anterior muscles of wildtype mice were injected with AAV sero-type1 that encodes FS-344, FLRG, and GASP-1 under the control of a strong cytomegalovirus (CMV)-based promoter. Increased body mass and muscle enlargement were observed with all three transgenes tested; however, the greatest increase occurred in FS-344-treated animals (Fig. 4a,b). The increased muscle mass was found in the injected hindlimb muscles and at sites remote to the injection, including upper extremity muscles such as the triceps (Fig. 4c). Thus, these inhibitors were secreted into the circulation from the site of muscle injection in the hindlimbs. The enlarged muscle mass was accompanied by functional improvement, demonstrated by an increase in hindlimb grip strength (Fig. 4d). There was no effect on heart size or histological appearance of cardiomyocytes, indicating that myostatin inhibition was selective to skeletal muscle tissue. We found no change in reproductive capacity in mice treated with our AAV1 carrying the FS-344 transgene (AAV1-FS, Table 2). Furthermore, we found no histological/pathological alterations in the gonadal tissue, heart, liver, or kidneys of FS-344-treated mice compared with controls, indicating that this treatment appeared to be safe and well tolerated.

FIGURE 4.

Myostatin inhibitor proteins increase muscle mass and strength in wildtype C57Bl/6 mice. (a) Gross hindlimb muscle mass is increased in all myostatin-inhibitor-protein-treated mice at 725 days of age compared with AAV1-GFP-injected controls. (b) Total body mass is significantly increased in AAV1-FS344-injected (**P < 0.01) and AAV1-GASP-1-injected (*P < 0.05) mice compared with AAV1-GFP controls at 725 days of age (n = 10). (c) The mass of individual hindlimb and forelimb muscles is increased in mice injected with AAV expressing myostatin inhibitory proteins (n = 10). *P < 0.05. (d) Hindlimb grip strength improves >2 years in all treated mice with the greatest differences in AAV1-FS344 treated animals compared with AAV1-GFP controls (n = 10). Error bars represent standard error.

Table 2.

Normal reproduction in C57BL/6 and mdx mice treated with AAV1.FS

Group	Treatment	Mean litter size (SD)
C57BL/6	AAV1.FS treated normal male mated with normal female (n=4)	9.0 (2.582) (n=4)
	AAV1.FS treated normal female mated with normal male (n=4)	9.25 (1.708) (n=4)
	Untreated normal male mated with untreated normal female (n=4)	9.0 (2.160) (n=4)
mdx−/−	AAV1.FS treated mdx−/− male mated with mdx−/− female (n=3)	4.5 (0.707) (n=3)
	AAV1.FS treated mdx−/− female mated with mdx−/− male (n=2)	2.0 (0) (n=2)
	Untreated mdx−/− male mated with untreated mdx−/− female (n=6)	3.83 (1.169) (n=6)

Subsequently, we tested the potential for AAV1-FS-344 to increase muscle mass and strength along with delaying muscle deterioration in the mdx mouse. The quadriceps and tibialis anterior muscles were injected at 3 weeks of age and were followed for 5 months before necropsy. A 15-fold increase in serum follistatin was seen in mice that received high dose (1.5 × 10¹² vg/kg) and a 6-fold increase was detected at low dose (1.5 × 10¹¹ vg/kg). Treated mice displayed a significant increase in muscle size (Fig. 5a), with the greatest individual muscle weight increase in high-dose follistatin-treated animals (Fig. 5b). As shown in previous studies in wildtype mice, effects could be seen at sites remote to the viral injection (e.g., triceps; Fig. 5b). Increased muscle mass translated to a dose-dependent improvement in muscle strength in the hindlimbs and forelimbs of treated animals compared with controls (Fig. 5c). AAV-FS-344-treated mice also demonstrated a significant reduction in serum creatine kinase (Fig. 6a). This is of interest, given that follistatin was protective despite its lack of correction of the underlying dystrophin deficiency. One might speculate that increasing the strength of individual fibers makes them more resistant to damage from the stress of normal activities.

FIGURE 5.

Single injection of AAV1.FS-344 increases muscle mass and strength in young mdx mice. (a) Gross hindlimb muscle mass is increased in AAV1.FS-344-injected mdx animals at 180 days of age compared with AAV1.GFP-injected controls. (b) The mass of individual hindlimb and forelimb muscles is increased at 180 days of age in mice injected at 3 weeks of age with AAV1.FS-344 paired with AAV1.GFP controls (n = 15). *P ≤ 0.05. (c) Grip strength is improved in a dose-dependent manner in young mdx mice injected at 3 weeks of age with AAV1-FS-344 followed for 180 days (n = 15). Red, high-dose AAV1.FS; blue, low-dose AAV1.FS-344; green, AAV1.GFP controls. Error bars represent standard errors.

FIGURE 6.

Effects of AAV1.FS344 on muscle enzymes, strength, and morphology in older mice. (a) Serum creatine kinase levels (units/liter) are decreased at 3 months after injection with AAV1.FS-344 compared with AAV1.GFP-injected controls. (*P < 0.05; n = 10.) Error bars represent standard errors. Aged mdx mice treated with AAV1.FS-344. (b) Hindlimb grip strength is significantly increased (P < 0.05) at 275 days and beyond in aged mdx mice treated with AAV1.FS-344 at 210 days of age (n = 15). Red, high-dose AAV1.FS; green, AAV1.GFP controls. (c) Hematoxylin and eosin (H&E) stain of aged gastrocnemius (pre-treatment 180 days) demonstrates reduced pathology at 560 days.

We next evaluated the potential for AAV1-FS-344 to increase muscle strength in mdx animals treated at an older age. Mice injected in the quadriceps and tibialis anterior muscles at 210 days of age demonstrated increased muscle strength measured by hindlimb grip strength for more than 60 days after administration. After reaching a plateau, increased strength persisted long-term, throughout the 560 days evaluated in this study (Fig. 6b,c). Reduced inflammation and endomysial connective tissue in the diaphragm supported previous findings related to follistatin, broadening its potential for clinical application. In the diaphragm of the AAV1-FS-344-treated mdx mouse, the muscle that most closely recapitulates the dystrophic process in humans,⁷⁴ we found a reduction in endomysial connective tissue (Fig. 6d). In addition, mdx muscle examined at 560 days of age demonstrated fewer focal groups of necrotic muscle fibers and mononuclear cell infiltrates. This antiinflammatory effect of follistatin has been studied in other translational paradigms. In models of experimental (trinitrobenzene sulfonic acid, oral dextran sulfate sodium) and spontaneous colitis (interleukin-10 gene-deficiency), follistatin pretreatment increased survival, decreased plasma levels of inflammatory cytokines, and reduced tissue inflammation.¹³ Follistatin has also been shown to reduce proinflammatory cytokines induced by lipopolysaccharide stimulation.²⁷

Given the positive results in mice, we also extended our follistatin gene therapy studies to nonhuman primates to establish evidence that direct muscle injections are a viable means of increasing strength in a larger model that more closely simulates the clinical setting. All too often, studies in mice fail to translate to the clinic; however, demonstrating efficacy and safety using the same means of viral transfer and the same cassette carrying the gene of interest in a larger animal species provides reasonable assurance of success for patients. Our studies in the cynamologous macaque have been remarkably successful (first presented at the American Society for Gene Therapy Meeting 2008; in preparation). We have studied AAV1-FS-344 under control of a specific (muscle creatine kinase) and a nonspecific (CMV) promoter. We have seen a remarkable increase in muscle size and strength with both promoters. Extensive observations on these monkeys with regard to reproductive organ function and postmortem tissue analysis have not revealed any organ system pathology, including organs of the pituitary–gonadal axis.

Our studies therefore indicate that AAV-mediated follistatin gene therapy has potential for treatment of muscular dystrophy, and it is well positioned to benefit certain acquired diseases. We have also observed that combinational treatment, meaning replacement of the defective gene, in combination with follistatin gene therapy may offer more than either alone. For example, ongoing studies in our laboratory have demonstrated that combining AAV-mediated micro-dystrophin gene therapy with AAV-delivered follistatin increases force generation of the muscle and protection against eccentric contraction to a greater degree than either can achieve alone. The antiinflammatory effects of follistatin that accompany AAV-mediated follistatin gene therapy are also of potential benefit to certain patient groups such as those with sporadic inclusion body myositis. The dual effects of increased muscle strength and reduced mononuclear cell infiltration provide a promising combination for treatment of an otherwise refractory muscle disease.

In conclusion, gene therapy using follistatin to inhibit myostatin holds promise for the treatment of muscle disease. Our findings that demonstrate increased muscle size and strength with reduced fibrosis in the mdx mouse, as well as our success in translating the findings to nonhuman primates, sets the stage for human clinical trials. We have encountered no adverse effects, no effect on reproductive capacity, and no immunogenicity of the follistatin transgene or its product. Therefore, the antiinflammatory effects of follistatin will facilitate efforts in a variety of diseases, particularly those like sporadic inclusion body myositis, where medical management has left many patients disabled and without treatment options.

Abbreviations

AAV

adeno-associated virus

ActRIIB

activin receptor type IIB

ALK-4

activin receptor like kinase 4, 5

BMP

bone morphogenic protein

capn3^−/−

calpain3 deficient mouse

CAV-3 Tg

caveolin-3 transgenic mouse deficient for caveolin-3

DMD

Duchenne muscular dystrophy

dy^W/dy^W

laminin-alpha2 (merosin) deficient mouse

FLRG

follistatin related gene

FS

follistatin

FSH

follicle stimulating hormone

FSHD

facioscapulohumeral dystrophy

FSP

FSH suppressing protein

GASP-1

growth and differentiation factor-associated serum protein-1

GDF-8

growth and differentiation factor 8

LGMD

limb girdle muscular dystrophy

mdx^−/−

Duchenne dystrophy mouse model deficient for dystrophin

MSTN

myostatin

mstn^−/−

myostatin deficient mouse

sgca^−/−

alpha-sarcoglycan deficient mouse

sgcd^−/−

delta-sarcoglycan deficient mouse

TGF-β

transforming growth factor beta

TLD

tolloid