|Home | About | Journals | Submit | Contact Us | Français|
Duchenne muscular dystrophy (DMD) is the most common, lethal, X-linked genetic disease, affecting 1 in 3500 newborn males. It is caused by mutations in the DMD gene. Owing to the large size of the gene, the mutation rate in both germline and somatic cells is very high. Nearly 13–15% of DMD cases are caused by nonsense mutations leading to premature termination codons in the reading frame that results in truncated dystrophin protein. Currently there is no cure for DMD. The only available treatment is the use of glucocorticoids that have modest beneficial effects accompanied by significant side effects. Different therapeutic strategies have been developed ranging from gene therapy to exon skipping and nonsense mutation suppression to produce the full-length protein. These strategies have shown promise in the mdx mouse model of muscular dystrophy where they have been reported to ameliorate the dystrophic phenotype and correct the physiological defects in the membrane. Each of these molecular approaches are being investigated in clinical trials. Here we review nonsense mutation suppression by aminoglycosides as a therapeutic strategy to treat DMD with special emphasis on gentamicin-induced readthrough of disease-causing premature termination codons.
Duchenne muscular dystrophy (DMD) is the most common, lethal, X-linked muscle disorder affecting 1 in 3500 live male births. It is caused by mutations in the DMD gene, the largest gene described to date, consisting of 79 exons, spread over approximately 2.4 million base pairs, encoding a 14-kb mRNA and the 427-kDa dystrophin protein [Koenig et al. 1988, 1987; Hoffman et al. 1987]. Dystrophin localizes to the cytoplasmic surface of the sarcolemma, providing an important link between the intracellular cytoskeleton and the extracellular matrix via dystrophin glycoprotein complex [Ervasti et al. 1990; Yoshida and Ozawa, 1990; Campbell and Kahl, 1989].
The large size of the gene predisposes to an increased incidence of both germline and somatic cell mutations. Estimates of the mutation rate range from 4.3×10−5 to 10.5×10−5 per generation [Gardner-Medwin, 1970; Stevenson, 1958; Walton et al. 1955; Stephens and Tyler, 1951]. Approximately one third of cases represent new mutations of the DMD gene with the remaining inherited on the X chromosome from a carrier mother [Worton, 1992]. Gene deletions of one or more exons represent about 60–65% of mutations. The full spectrum of mutations includes duplications, small insertions and deletions, splice-site and point mutations [Dent et al. 2005]. With regard to the potential for aminoglycoside readthrough, it is estimated that 13–15% are nonsense mutations [Flanigan et al. 2003; Mendell et al. 2001] that create a premature termination codon (PTC) in the reading frame.
Although the DMD gene was identified more than 20 years ago, there is currently no cure for the disease. Glucocorticoids are the only beneficial treatment for the disease, usually administered orally as prednisone [Mendell et al. 1989] or deflazacort [Biggar et al. 2006]. While steroid-induced side effects can be serious, this class of drugs can prolong ambulation and reduce the incidence of severe scoliosis, providing a better quality of life for DMD boys [Bushby et al. 2010; King et al. 2007]. Both gene replacement and pharmacological therapies such as mutation suppression and exon skipping have shown promise in mdx mice, an animal model of DMD [Harper et al. 2002; Barton-Davis et al. 1999; Wilton et al. 1999]. These approaches generate a functional protein which ameliorates the dystrophic phenotype and corrects the physiological defects in the membrane. In this review, we focus on the mutation suppression strategy by aminoglycoside antibiotics with emphasis on gentamicin-induced readthrough of disease-causing PTCs in DMD.
Aminoglycosides are a class of antibacterials used mainly to treat Gram-negative bacterial infections. As far back as the 1960s, certain aminoglycosides such as streptomycin were shown to reduce the fidelity of translation by inhibition of ribosomal proof reading. The increased frequency of erroneous insertions at nonsense codons permitted continued translation to the 3’ end of the gene [Anderson et al. 1965; Davies et al. 1964]. Later reports extended observations to mammalian cells confirming that mutation suppression and stop codon readthrough was not unique to prokaryotes [Martin et al. 1989; Burke and Mogg, 1985]. Genetic and biochemical studies demonstrate that these antibiotics bind to the decoding site of ribosomal RNA of both prokaryotes and eukaryotes and bring about a conformational change that permits codon–anticodon pairing during translation. Aminoglycosides reduce discrimination between cognate and near-cognate tRNA, permitting an amino acid to be inserted at the stop codon. The net effect is continuation of translation through to the natural stop codon [Fan-Minogue and Bedwell, 2008; Fourmy et al. 1996] (Figure 1).
The concept of a readthrough strategy for producing a full-length protein from disease-causing premature stop mutations in mammalian cells was first investigated more than a decade ago in cystic fibrosis (CF) [Howard et al. 1996]. The authors reported the expression of full-length CF transmembrane conductance regulator (CFTR) protein and restoration of its cyclic AMP-activated chloride channel activity in HeLa cells harboring premature stop codon mutations following treatment with two different aminoglycosides G-418 (also known as Geneticin®) and gentamicin. These proof-of-principle studies were provocative and demonstrated the potential for therapeutic application of the readthrough mechanism. Although CF is not the intended target for this review, the findings using aminoglycosides have provided a template for other genetic diseases to follow. In CF, in vivo studies of the transgenic mouse model carrying the G542X-hCFTR mutation provided encouragement for the field. Full-length CFTR protein was generated following treatment with gentamicin, tobramycin or amikacin [Du et al. 2006, 2002]. Functional improvement of the newly translated CFTR protein was assessed by measuring cAMP-stimulated Cl− conductance. In a separate study, co-administration of gentamicin along with poly-L-aspartic acid (PAA) increased the level of readthrough by 20–40% and prolonged gene expression for extended periods of time, even after the readthrough agents were withdrawn. It is also known that PAA reduces gentamicin toxicity, having potential implications for translational clinical studies [Du et al. 2009b].
The chemical composition of aminoglycosides plays a potentially important role in translation. Comparative studies of specific antibiotics tested under well-defined conditions (tissue and specific mutation) can be of value prior to clinical translation. Geneticin is superior to gentamicin and tobramycin in restoring full-length CFTR protein in HeLa cells and mutant bronchial epithelial cell lines [Bedwell et al. 1997; Howard et al. 1996]. In transgenic mice carrying the human CFTR-G542X stop mutation, amikacin proved more effective in mutation suppression than gentamicin [Du et al. 2006]. In mdx mice, negamycin, a dipeptide antibiotic, restores dystrophin expression as reported by immunofluorescence and immunoblotting in skeletal and cardiac muscles as well as in cultured mdx myotubes [Arakawa et al. 2003]. It was further reported that negamycin better maintained body weight and demonstrated less ototoxicity compared with gentamicin.
Another potential consideration in carrying mutation suppression to the clinic is the observation that the source of the antibiotic could have an influence on efficacy related to differential affinity for the A-site of the ribosome. As shown in Figure 2, three variants of gentamicin (C1, C1a, and C2) show only slight differences in chemical structure, but the consequences may be much greater and some investigators point to these differences as important predictors of outcome [Dunant et al. 2003]. In commercial production, a particular mixture of these three isoforms could be more or less effective in the laboratory or in the clinic. For clinical trials, it is important to use gentamicin from the same source; production using the same manufacturing conditions provides greater consistency in the interpretation of results [Dunant et al. 2003; Yoshizawa et al. 1998; Loveless et al. 1984].
The sequence context of PTC is also reported to affect the efficacy of mutation suppression. Myotubes of DMD patients carrying TGA nonsense mutations were reported to preferentially express dystrophin following gentamicin treatment in contrast to those with TAG or TAA stop codons [Kimura et al. 2005]. Politano and colleagues reported a similar preference for TGA stop codons in four gentamicin-treated DMD patients [Politano et al. 2003]. We could not validate this finding in our long-term gentamicin-treated DMD subjects [Malik et al. 2010]. In addition, the flanking sequences especially the nucleotide immediately downstream of the stop codon (+4 position) are potentially key influences in determining the readthrough potential. Prioritization of efficacy is predicted according to the following nucleotide sequence: C>U>A>G at +4 position [Howard et al. 2000; Manuvakhova et al. 2000]. However, this has not always been validated in either dual reporter gene assays [Bidou et al. 2004] or clinical trials [Malik et al. 2010].
Nonsense mediated decay (NMD) is another potential influential factor affecting mutation suppression in response to gentamicin. In studies using epithelial cell lines harboring the W1282X CFTR mutation, siRNA downregulation of NMD increased the level of nonsense transcripts and enhanced CFTR physiologic function in response to gentamicin [Linde et al. 2007]. The results suggest that the efficiency of NMD could have an important role in governing the response to treatments aiming to promote readthrough of PTCs. We have also found evidence favoring this phenomenon in our clinical studies of readthrough in DMD [Malik et al. 2010]. A principle to consider in predicting potential efficacy of mutation suppression is the so-called ‘rule’ for the termination codon position: only those termination codons located more than 50–55 nucleotides upstream of the 3’-most exon–exon junction (measured after splicing) mediate a reduction in mRNA abundance through NMD [Nagy and Maquat, 1998].
Mutation suppression of stop codons by gentamicin specifically in relation to the DMD gene was first reported by Barton-Davis and coworkers in 1999 [Barton-Davis et al. 1999]. The mdx mouse, harboring a PTC in exon 23 of the DMD gene, provided an appropriate animal model to address the translational potential for gentamicin-induced readthrough. A 14-day course of systemically administered gentamicin increased expression of full-length dystrophin correctly localized at the sarcolemmal membrane and was associated with increased force generation and protection against contraction-induced injury. The findings were initially challenged based on differences in chemical structure related to the source of gentamicin [Dunant et al. 2003], but later corroborated in an 8–12-week study that demonstrated efficacy of gentamicin treatment in exercised challenged mdx mice. In these studies, gentamicin-treated mice showed improved histology, increased dystrophin expression in all fibers, and a concomitant recovery of aquaporin-4 known to be linked to the dystrophin glycoprotein complex through α-syntrophin [De Luca et al. 2008]. Plasma creatine kinase (CK) levels were also reduced by 70%. Additional experiments demonstrated that gentamicin increased dystrophin expression in endothelial cells and normalized eNOS expression in arteries of the mdx mouse [Loufrani et al. 2004] potentially relevant to issues of the compromised blood flow to muscle in DMD patients [Sander et al. 2000; Mendell et al. 1971].
Molecular studies of mutation suppression in the mdx mouse served as a stimulus to attempt a similar approach using gentamicin in DMD patients. The initial clinical trial reported in 2001 involved two DMD and two Becker muscular dystrophy (BMD) patients carrying PTC in the dystrophin gene [Wagner et al. 2001]. Each subject received 7.5mg/kg intravenous gentamicin for 2 weeks. They found no full-length dystrophin posttreatment by either immunofluorescence or immunoblotting in any patient although serum CK values dropped significantly. No clinical improvement was found. Findings in this study may differ from later reports based on the duration of treatment or the source of gentamicin. However, the drop in serum CK does suggest a biological effect consistent with our later studies of an extended treatment regimen for 6 months [Malik et al. 2010].
A subsequent trial reported more favorable results using an alternative gentamicin regimen [Politano et al. 2003]. Again the sample size was small including only four subjects. Gentamicin was given over two cycles, each for 6 days with a 7-week hiatus between dosing. Muscle biopsy results revealed dystrophin expression by immunohistochemistry and immunoblotting in three out of four patients with UGA stop codons. Their findings favored an effect related to a greater readthrough permissiveness of the UGA stop codon but the small sample size and the under representation of other PTCs detracts from this conclusion. Serum CK decreased from day 2 to day 6 of the treatment but came back to the initial values subsequently. No clinical meaningful outcomes in terms of walking time or negotiating steps were obtained in any of the patients.
A third clinical trial was conducted by our group at the Nationwide Children’s Hospital and consisted of a two-phase protocol [Malik et al. 2010]. The first phase was intended to establish biopotency in a short-term, 14-day dosing study, simulating what had been done in mdx mice and ensuring adequate translation to the clinic prior to an extended 6-month trial. Ten DMD boys with proven stop codon mutations were recruited for daily gentamicin administration using a standard dosing regimen (7.5mg/kg) for treatment of Gram-negative infections. Serum CK served as the primary outcome measure. A biological effect was established with a reduction of CK by 50% at day 14 (Figure 3). During the trial boys remained ambulatory, excluding a false-positive effect related to decreased activity. Specificity for stop codon readthrough by gentamicin was validated by failure to find a change in serum CK in eight DMD boys with frameshift mutations, treated for the same duration. The dose of gentamicin was well tolerated without any side effects.
Having confirmed a biologically active dosing regimen, the next goal was to determine the feasibility of long-term (6 months) administration of gentamicin. The aims for the long-term trial were: (1) safe gentamicin administration for 6 months; (2) dystrophin expression resulting from readthrough of DMD stop codon mutations; (3) clinically meaningful outcomes in muscle strength and function. The dosing regimen for gentamicin was dictated based on the known prolonged half-life of dystrophin, with estimates of as long as 6–8 weeks [Ghahramani Seno et al. 2008; Ahmad et al. 2000]. Given that the effects of gentamicin on readthrough would accompany each dosing, a reasonable expectation would be a concomitant period of full-length dystrophin translation with every dose. Weekly and twice-weekly dosing regimens would steadily increase dystrophin that would reach a steady state based on the half-life of the protein. The unknown factor was the percentage expression level that could be achieved using a cautious dosing regimen based on established tolerance of the standard 14-consecutive-day dosing schedule. Thus, if gentamicin was given weekly, only after 98 days would the cumulative dose be equivalent to the 14-day course; correspondingly, a twice-weekly regimen would not exceed the 14-day regimen until day 49. This schedule permitted careful monitoring for adverse effects. Enrollment was dependent on normal hearing and kidney function, and negative testing for the known A1555G mitochondrial DNA mutation in the 12S rRNA gene predisposing to aminoglycoside-induced hearing loss [Tang et al. 2002]. We were also aware of the challenges of monitoring renal function in the DMD population because of reduced muscle mass resulting in decreased serum creatinine and creatinine clearance levels. We therefore took advantage of cystatin C for monitoring renal function in DMD patients, validating its use in a separate study [Viollet et al. 2009].
Sixteen subjects with documented stop codon mutations received weekly (n=12) or twice-weekly (n=4) gentamicin. No patient demonstrated a decline in renal or hearing function except one patient given a miscalculated dose (125% of recommended) for the first four administrations resulting in a transient high-frequency hearing loss that returned to normal within 3 months. Owing to the overall small sample size, data analysis included patients taking weekly and bi-weekly gentamicin. Twelve subjects had pretreatment and posttreatment muscle biopsies since three subjects declined to go for posttreatment biopsy and one dropped out of the study (Table 1). Dystrophin expression was assessed by both immunofluorescence and immunoblotting techniques. In nine patients, pretreatment dystrophin levels ranged from 0.8% to 4.5%; six of these showed increased levels following 6-months of gentamicin treatment. In three subjects dystrophin protein increased to a potentially therapeutic range with independent concordance by immunofluorescence (Bioquant image analysis®) and Western blot analyses. The most significant readthrough was seen in three patients increasing dystrophin levels to 13.0%, 15.4%, and 15.4% of wild-type levels (Figure 4). The increases seen in this clinical trial were in the range observed in the mdx mouse gentamicin study (10–20%) resulting in protection against eccentric contraction and increased force generation [Barton-Davis et al. 1999]. Considering the entire group the increase in dystrophin over baseline was significant (6 of 12 patients, p=0.027). The sample size was too small to determine if there was an additive effect of twice-weekly gentamicin versus weekly. There did not appear to be a preferential stop codon predicting readthrough (Table 1) or any relation to the fourth base surrounding the stop codon (Table 1).
Of particular interest the presence of dystrophin in pretreatment biopsies predicted an increase in dystrophin following gentamicin treatment (9/12 subjects; p<0.001) (Table 1). It is worth drawing attention to the finding that the majority of DMD patients with stop codons expressed small amounts of dystrophin prior to gentamicin treatment (9/12 patients; Table 1). These findings simulate the experience reported in CF [Linde et al. 2007] and also demonstrate that pretreatment dystrophin is predictive of aminoglycoside-induced readthrough.
We assume (but have not validated directly) that the observations in the DMD gene parallel findings in CF, where downregulation of NMD governs the level of nonsense transcripts available for readthrough. This argues for a model in which mRNA transcripts escaping NMD serve as a template for stop codon readthrough induced by gentamicin. Thus, it is possible that pretreatment dystrophin expression on muscle biopsies can potentially serve as a marker for those most likely to respond to stop codon readthrough treatment. The caveat, however, is that pretreatment dystrophin can also result from spontaneous exon skipping. For example, in another gentamicin-treated case (Patient 5, Table 1), pretreatment dystrophin levels of slightly more than 4% were easily detected, yet posttreatment levels were unchanged. In this case, it is more likely that alternative splicing resulted in small amounts of in-frame dystrophin that essentially ignored the stop codon.
In addition to dystrophin expression in muscle tissue, serum CK was reduced after 6 months of gentamicin treatment (9851±2109 U/l to 5316±850 U/l, p=0.04), substantiating the validity of the 14-day study. Additional clinical outcome measures were variable and not particularly robust. The average muscle score (AMS), which expected to decline by 0.2 units over 6 months in a corticosteroid-naïve population [Mendell et al. 1989], showed little or no change (pretreatment AMS 5.37±0.39 compared with posttreatment AMS 5.35±0.38). Considering that 12 of 16 cases were not taking corticosteroids, this finding favored a modest clinical effect that was further supported by a slight increase in forced vital capacity (although significance was not reached, p=0.06). Nevertheless, the functional outcomes that could be a surrogate for improved quality of life, failed to improve including time to walk 30 feet, or climb four standard steps, and stand from supine position on the floor. This raises the very important and unresolved question of how much dystrophin is needed to improve patients and whether this can ever be achieved safely with gentamicin or other aminoglycosides. Upregulation to at least 30% of wild-type dystrophin has been estimated as a target for clinical trials [Neri et al. 2007].
A very important question we addressed in a limited number of patients in this trial was the emergence of novel immunogenic epitopes related to dystrophin expression following mutation suppression. The overall issue has relevance to other treatment modalities expressing novel dystrophin epitopes including gene therapy and exon skipping [Mendell et al. 2010]. In the gentamicin clinical trial, one (Subject 11) of three patients tested (Subjects 10–12) exhibited a T-cell response against a dystrophin peptide pool derived from exons 59–70, downstream of his stop codon in exon 58. Peripheral blood mononuclear cells (PBMCs) collected at day 180 (the last day of gentamicin treatment) revealed a brisk IFN-γ response (Figure 5A) and this was corroborated by mononuclear cells isolated directly from muscle. Cellular immunity to dystrophin was mapped to a highly specific peptide fragment spanning amino acids 3325–3344 in exon 69 (Figure 5B). The dystrophin-specific T cells expressed CD4 and thus belonged to the helper subset. Unfortunately the ELISpot assay was performed only posttreatment, thus clouding the timing of appearance of this novel immunogenic peptide. In two other subjects tested (patients 10 and 12), no T-cell immunity was observed to dystrophin. Both of these patients showed dystrophin expression in their pretreatment and posttreatment muscle biopsies (Table 1). It seems clear that future readthrough studies should include pretreatment and posttreatment evaluations of T-cell responses.
In summary, the data presented in this review support the therapeutic concept of mutation suppression by gentamicin in the DMD gene, even though clear clinical efficacy was not achieved in this study. We predict that higher levels of dystrophin will be needed to enhance functional outcomes. Recently, Nudelman and colleagues demonstrated three different novel aminoglycosides, NB54, NB74, and NB84 with superior readthrough efficacy and lower cell-toxicity and ototoxicity compared with gentamicin that could be considered for future clinical trials [Nudelman et al. 2010, 2009]. The design of these new aminoglycosides separates structural elements in the formulation that induce readthrough from those that affect toxicity. Despite these results, the inherent risk of aminoglycoside toxicity and the need for intravenous infusion, favors alternative approaches if possible, using safe, orally available pharmacological agents with potential readthrough capacity [Hirawat et al. 2007; Hamed, 2006].
Nonaminoglycoside readthrough compounds have received considerable attention over the past several years. PTC124 (Ataluren®) is effective in restoring full-length protein in animal models of CF and DMD and proved safe in phase I clinical trials in healthy patients [Kerem et al. 2008; Hamed, 2006]. Initial reports of the efficacy of this drug in a phase II DMD/BMD (ClinicalTrials.gov) have not been encouraging but additional studies are planned to define the most effective dosing regimen. Additional high-throughput screening of compounds may identify other nonaminoglycoside readthrough products that hold promise for DMD patients with stop codon mutations [Du et al. 2009a].
This work was supported by MDA and NIH, NINDS Grant 7R01 NS043186, and Jesse’s Journey.
The authors declare that there are no conflicts of interest.