The extensive molecular genetic heterogeneity seen with inherited eye disease is a major barrier to the development of gene-based therapeutics. The underlying molecular pathology in a considerable proportion of these diseases however are nonsense mutations leading to premature termination codons. A therapeutic intervention targeted at this abnormality would therefore potentially be relevant to a wide range of inherited eye diseases. We have taken advantage of the ability of aminoglycoside drugs to suppress such nonsense mutations and partially restore full-length, functional protein in a zebrafish model of choroideraemia (chmru848; juvenile chorio-retinal degeneration) and in two models of ocular coloboma (noitu29a and gupm189; congenital optic fissure closure defects). In vitro cell-based assays showed significant readthrough with two drugs, gentamicin and paromomycin, which was confirmed by western blot and in vitro prenylation assays. The presence of either aminoglycoside during zebrafish development in vivo showed remarkable prevention of mutant ocular phenotypes in each model and a reduction in multisystemic defects leading to a 1.5–1.7-fold increase in survival. We also identified a significant reduction in abnormal cell death shown by TUNEL assay. To test the hypothesis that optic fissure closure was apoptosis-dependent, the anti-apoptotic agents, curcumin and zVAD-fmk, were tested in gupm189 embryos. Both drugs were found to reduce the size of the coloboma, providing molecular evidence that cell death is required for optic fissure remodelling. These findings draw attention to the value of zebrafish models of eye disease as useful preclinical drug screening tools in studies to identify molecular mechanisms amenable to therapeutic intervention.
The number of people worldwide who suffer from blindness is calculated to be in excess of 35 million people (1). The relative contribution of genetic defects to this cohort is not known exactly; however, it is estimated that genetic defects account for over half of the cases of blindness in children under the age of 16 (2,3). Furthermore, the hereditary retinopathies are the commonest causes of registered visual handicap in developed countries (4). Hereditary retinopathies are also, surprisingly, genetically heterogeneous with literally hundreds of genes associated with retinopathy. However, these molecular genetic abnormalities may be divided into subtypes. Our analysis of the human gene mutation database (http://www.hgmd.cf.ac.uk/ac/index.php) suggested that ∼12% of all mutations causing human genetic disease are single-point mutations that result in a premature termination codon. Within this, we identified a significant number of eye diseases. For example, OMIM (http://www.ncbi.nlm.nih.gov/) lists 46 premature termination codon mutations in 21 different genes causing nine clinically distinct inherited retinal degenerations (e.g. retinitis pigmentosa, Usher syndrome, Bardet–Biedl syndrome), although this is likely to be an underestimate of the true scale, as not all mutations are listed at OMIM. In addition, in a study of 107 patients with CHARGE syndrome (featuring ocular coloboma), 65% of individuals had mutations in the CHD7 gene, and of these, nearly half were nonsense mutations (5). A molecular therapy that targets these premature termination codons could therefore treat a substantial proportion of patients, making the approach both practical and economical.
Aminoglycosides are a class of drugs which have the ability to bind to the eukaryotic 18S ribosomal RNA A-site resulting in a conformational change that interferes with the translational fidelity of proteins (6). If a premature termination codon exists in the mRNA, a near-cognate tRNA anti-codon can be substituted in place of a release factor, leading to the production of full length, functional protein. A number of null mutation in vitro and in vivo studies have been undertaken to test the relevance of aminoglycoside treatment in conditions such as cystic fibrosis (7), Duchenne muscular dystrophy (8), Hurler syndrome (9) and retinitis pigmentosa (10). Positive results have led to a number of clinical trials (11–14). Interestingly, levels of aminoglycoside-driven readthrough ranged only from 1 to 25% in these studies (6). The efficacy of readthrough is thought to be dependent on the type of aminoglycoside used, the sequence of the premature termination codon (UGA > UAG > UAA) and the nucleotide base immediately downstream of the stop codon (C > U > G ≥ A). In disease states where protein expression is almost obsolete, notably recessive disorders, the presence of as little as 1% functional protein seems to modify disease phenotype (15).
Zebrafish models of visual disorders are valuable tools for understanding the pathogenesis of eye disease (16). There is a strong homology between zebrafish and human genes, and accordingly disease models can display very similar phenotypic features. Examples of zebrafish ocular phenotypes similar to those seen in human disease include choroideraemia and ocular coloboma. Choroideraemia is an X-linked recessive condition characterized by a progressive chorioretinal dystrophy, with an incidence of 1 in 50 000 (17). Affected males commonly present during their teenage years with night blindness, and subsequent constriction of visual fields leads to complete loss of vision by middle age (18). The molecular defect in over one-third of these cases are nonsense mutations, occurring in the Rab escort protein 1 (REP1) gene (19,20). The choroideraemia mutant zebrafish (chmru848) has a recessive nonsense mutation in the second exon of rep1 (C→T) converting a glutamine codon (CAA) to a premature stop codon (UAA) at amino acid position 32 (Q32X). The ocular morphological features include microphthalmia, irregular eye pigmentation, cataracts, disrupted retinal layers characterized by areas of RPE hypertrophy with invasion into the photoreceptor layer and small dense pyknotic cells throughout the retina (21,22).
Ocular coloboma is a congenital eye anomaly arising from incomplete fusion of the optic fissure during weeks 5–7 of embryogenesis. It is a significant cause of childhood visual impairment and blindness worldwide, reaching a prevalence of 7.5 per 10 000 births (23,24). Nonsense mutations have been identified in many cases of ocular coloboma, for example, within the paired box 2 (PAX2) gene causing renal-coloboma syndrome (24), and in the CHD7 gene in CHARGE syndrome (5). The no isthmus (noitu29a) zebrafish mutant has a recessive nonsense mutation in exon 7.1 of pax2.1 (C→T) resulting in an arginine codon (CGA) converting to a stop codon (UGA) at amino acid position 139 (R139X) (25). This model exhibits defects in the formation of the optic stalk, leading to an open optic fissure, lack of midbrain–hindbrain boundary, plus optic tectum, cerebellum, inner-ear and pronephric duct abnormalities (26). The second zebrafish model of ocular coloboma is the grumpy (gupm189) mutant, which has a recessive nonsense mutation in lamb1 (27). Phenotypic features include optic fissure closure defects, lens dysplasia, disruption of the retina and a shortened body axis associated with lack of notochord differentiation (28).
Using these zebrafish models, we assessed the ability of two different aminoglycosides to readthrough the null mutations and modify the phenotype. We have undertaken in vitro cell culture experiments to assess the efficacy of gentamicin and paromomycin readthrough and then in vivo histological and biochemical assessments to determine whether these drugs can modify phenotype.
Therapeutic dose for aminoglycoside readthrough in zebrafish models
Aminoglycosides can cause serious nephro- and oto-toxic effects in humans and zebrafish (29,30). To determine the highest dose of drug that could be tolerated, we carried out survival experiments. The survival of wildtype (wt) embryos dosed continuously from 10 h post fertilization (hpf) (prior to eye development) with paromomycin or gentamicin (50 µm–1 mm) was measured at 6 days post fertilization (dpf) (Fig. 1A). Embryos treated with 100 µm gentamicin or paromomycin displayed 100% survival rates at 6 dpf and showed normal retinal morphology (Fig. 1B). Doses of >100 µm for both drugs yielded toxic side effects such as balance defects indicated by abnormal swimming behaviour, absence of the startle response, morphological abnormalities such as shorter body length, bent tails, pericardial oedema and smaller eyes (microphthalmia). Doses of ≥0.5 mm gentamicin and paromomycin were lethal.
On the basis of these results in wt fish, the effect of gentamicin on survival in chm mutants was determined following the administration of 50, 100 or 150 µm gentamicin at 10 hpf compared with untreated embryos (Fig. 1C). There was no difference in survival times for the 50 µm treatment group (P = 0.25) compared with untreated controls, whereas treatment with 100 or 150 µm gentamicin significantly increased survival time (P < 0.0001). The highest survival was seen for the group treated with 100 µm (8.0 days), and toxicity effects were noted at 150 µm. As dose–response relationships for gentamicin and paromomycin were very similar, the 100 µm concentration was established as the optimum therapeutic dose for in vivo testing in other models.
Aminoglycoside-mediated readthrough of chm, noi and gup nonsense mutations
To assess the efficiency of aminoglycoside-mediated readthrough of the nonsense mutations causing choroideraemia and ocular coloboma in the zebrafish models, in vitro luciferase reporter assays were performed. When cloning the gup luciferase construct, we found that the mutation was Q524X (CAG to UAG), which differed from that published by Parsons et al. (27), who reported Q544X. Genotyping was performed on homozygous gup mutants to validate this amendment (data not shown). COS-7 cells were transiently transfected with wt or mutant chm Q32X, noi R139X or gup Q524X constructs in the presence or absence of 100 µm gentamicin or paromomycin.
This concentration of aminoglycosides was found to be the optimal dose based on previous reporter assays (10) and was consistent with the dose–response experiments performed on wt and mutant embryos (Fig. 1A and C). Increased readthrough levels of 2.1 and 4.0% were observed for the mutant chm Q32X construct in the presence of gentamicin and paromomycin, respectively (Fig. 2A). A 3.4 and 5.5% increase in readthrough with gentamicin and paromomycin, respectively, was seen for the noi R139X mutation. Analysis of the gup Q524X construct resulted in increased readthrough levels of 3.2 and 4.3% with gentamicin and paromomycin, respectively. These results show that although both drugs can suppress nonsense mutations, paromomycin had a higher readthrough efficiency than gentamicin with all three constructs. In addition, readthrough of the noi R139X (UGA) mutation exceeded that of the gup Q524X (UAG) mutation, which was better than the chm Q32X (UAA) mutation with both aminoglycosides, and this is consistent with the preferential readthrough of specific stop codons (UGA > UAG > UAA) (31).
Aminoglycoside treatment increases the survival of chm, noi and gup mutants
Mean survival of chm, noi and gup mutant embryos in vivo was determined in the presence or absence of 100 µm gentamicin or paromomycin administered from 10 hpf (Fig. 2B). Untreated chm mutants survived an average of 4.8 days, whereas chm embryos treated with gentamicin lived for 8.3 days, or with paromomycin lived for 8.4 days, representing a 1.7-fold increase in survival. Untreated noi mutants survived an average of 5.8 days, whereas noi embryos treated with gentamicin lived for 9.0 days, or with paromomycin lived for 9.1 days, representing a 1.5-fold increase in survival. Untreated gup mutants survived an average of 5.0 days, whereas gup embryos treated with either gentamicin or paromomycin lived for 8.6 days, representing a 1.7-fold increase in survival. Treatment of chm, noi and gup mutant embryos with either drug significantly increased their survival time (P < 0.0001) compared with untreated mutant controls; however, the difference between survival of gentamicin-treated and paromomycin-treated groups was negligible.
Gentamicin–Texas Red conjugate penetration into the retina
Cellular uptake of aminoglycosides occurs by fluid-phase and receptor-mediated endocytosis, which can be validated using conjugation of the aminoglycoside to a Texas Red (TR) tracer (32). To confirm that the aminoglycoside drugs could penetrate the eye wt and mutant chm, noi and gup larvae were assessed by incubating 10 hpf embryos in 100 µm gentamicin–TR (GTTR) conjugate until either 6 or 9 dpf. Multiple confocal images through the retina were taken and reconstructed to show drug penetration. GTTR was seen throughout the retina of wt (Fig. 3A and F) and all mutant larvae up to 9 dpf (Fig. 3B–D and G–I). There was no apparent reduction in the penetration of the aminoglycoside conjugate as the embryo grew older, thus suggesting continued pharmacological activity. wt embryos exposed to TR conjugate alone showed minimal background fluorescence indicating very little or no penetration into the retina at 6 or 9 dpf (Fig. 3E and J).
rep1 protein detection in chm mutants following aminoglycoside treatment
To determine whether aminoglycoside treatment increased rep1 protein expression, western blot analysis was undertaken on total zebrafish extracts from wt, untreated and aminoglycoside-treated chm mutants at 6 and 9 dpf. (Fig. 4A). Untreated chm mutants at 6 dpf demonstrated levels of 15.3% rep1 protein expression relative to gapdh, and this source is attributed to residual maternal rep1 derived from the persistent yolk sac. Gentamicin- and paromomycin-treated chm larvae at 6 dpf expressed 37.3 and 32.4% rep1 protein, respectively (Fig. 4B). Deduction of background maternal contribution results in a 22 and 17% respective increase in full-length protein secondary to aminoglycoside-mediated readthrough. By 9 dpf, with no maternal protein contribution at this time, gentamicin- and paromomycin-treated chm mutants were found to express 8.1 and 18.7% rep1 protein, respectively (Fig. 4B). This suggests that aminoglycoside readthrough was directly responsible for protein expression 9 days after continual drug exposure. Although there were differences in protein expression levels at 6 and 9 dpf between the drugs, this did not translate into different morphological, histological or survival characteristics of the drug-treated mutant embryos.
In vitro prenylation assay
The cytosolic extracts from whole chm mutant zebrafish have an accumulation of unprenylated Rab proteins owing to the reduced rep1 activity. To determine whether the aminoglycoside readthrough generated increased rep1 functional activity, we used an in vitro prenylation assay (33). In this assay, the cytosolic extracts were a source of unprenylated Rab substrates that were incubated with recombinant REP and RGGT in the presence of [3H]GGpp. Under these conditions, we observed the presence of prenylated [3H]GG-Rabs in chm zebrafish extracts (95%) compared with only 10% in wt extracts (Fig. 5A). Thus, a high level of prenylation in chm mutant extracts reflects a low level of rep1 activity in vivo, whereas a low level of prenylation in wt extracts reflects a high level of rep1 activity in vivo. In fish treated with gentamicin or paromomycin, the amount of prenylated Rabs was reduced to 88 and 82%, respectively (Fig. 5B). This demonstrated that there was a lower level of unprenylated Rabs in chm zebrafish that had been treated with the aminoglycosides and hence a higher level of rep1 activity in vivo, compared with untreated fish.
Effect of aminoglycoside treatment on chm morphology and retinal histology
Mutant chm larvae at 6 dpf displayed systemic defects such as pericardial and abdominal oedema, shorter body length, an uninflated swim bladder, persistent yolk sac and microphthalmia with irregular eye pigmentation due to reduced iridophore pigment cells (Fig. 6B). Coronal sections of the retina showed widespread degeneration with pyknotic cells throughout all layers of the retina, diminishing lamination, areas of RPE hypertrophy which extend into the photoreceptor layer compressing outer segments of these cells and small cataractous lenses (Fig. 6B).
Treatment with 100 µm gentamicin or paromomycin resulted in milder and comparable chm mutant phenotypes. Gross morphology at 6 dpf revealed reduced pericardial and abdominal oedema, increased body length and larger eye diameter compared with untreated mutants, however wt parameters were not fully reached (Fig. 6C and D). Coronal sections of treated chm mutant retina at 6 dpf showed remarkable normalization of lamination including a regular RPE layer, absence of pyknotic cells and normal lens. The main difference between treated chm mutant and wt retina was notably thickened inner nuclear and ganglion cell layers (Fig. 6C and D) with less densely packed nuclei. In addition, the ganglion cell layer was closely apposed to the lens in the treated mutants owing to ganglion cell proliferation. At 9 dpf, treated chm larvae showed no signs of a persistent yolk sac, minimal oedema, the overall eye size and body length remained less than age-controlled wt larvae and the larvae appeared thinner (Fig. 6F and G). Importantly, a functional inflated swim bladder failed to develop in treated mutants even at 9 dpf, and this consequently impaired proper feeding, leading to starvation (21). Retinal sections appeared remarkably similar to wt fish and only the ganglion cell layer remained thicker and closely apposed to the lens (Fig. 6F and G). No signs of chorioretinal degeneration were seen at 9 dpf following aminoglycoside treatment.
Effect of aminoglycoside treatment on noi morphology and retinal histology
Gross morphological defects of mutant noi larvae at 6 dpf include significant pericardial and abdominal oedema, residual yolk sac, absence of a functional swim bladder, shorter and curved body axis, irregular eye pigmentation and microphthalmia (Fig. 7B). Parasagittal sections of the retina showed a persistent open optic fissure in the inferior aspect of the eye, extending through all the layers of the retina. Ingression of RPE into the retina demarcates the colobomatous defect with irregular retinal lamination and a small cataractous lens (Fig. 7B).
Mutant noi larvae treated with 100 µm gentamicin or paromomycin showed similar morphological and retinal histological results. Treated noi at 6 dpf demonstrated reduced pericardial and abdominal oedema, increased body length, a large residual yolk sac and larger eyes compared with untreated mutants (Fig. 7C and D), although they did not reach wt levels. Parasagittal retinal sections of treated noi mutants at 6 dpf showed almost complete fusion of the optic fissure; a small notch was visible at the most inferior aspect of the eye (Fig. 7C and D). At 9 dpf, treated noi larvae showed no signs of a persistent yolk sac; overall eye size and body length resembled age-controlled wt larvae. However, these older mutants displayed mild recurrent pericardial and gut oedema and failed to develop an inflated swim bladder, thus preventing feeding (Fig. 7F and G). By 9 dpf, the optic fissure was fully closed with a smooth continuous circumference and regular retinal lamination (Fig. 7F and G).
Effect of aminoglycoside treatment on gup morphology and retinal histology
Phenotypic features of gup mutants at 6 dpf include a shortened body axis, pericardial oedema, persistent yolk sac, uninflated swim bladder, small irregular shaped eyes and large optic fissure closure defects (Fig. 8B). In addition, abnormalities in lens capsule integrity and fibre cell differentiation lead to the formation of small dysplastic lenses which are not anchored in the anterior chamber of the eye, as with wt embryos, but are situated in the retinal neuroepithelium (34). Parasagittal sections of gup retina reveal the extent of the coloboma with highly irregular lamination, discontinuous RPE and photoreceptor cell layers and the absence of a lens (Fig. 8B).
The gup mutants treated with either drug had comparable results. At 6 dpf, they showed a longer body length, reduced pericardial oedema and larger eyes with complete fusion of the optic fissure, compared with untreated gup mutants (Fig. 8C and D). The persistent yolk sac, body length and the lack of a discernible pupil with irregular iridophore patterning remain the most differentiating features from age-matched wt controls. At 9 dpf, treated gup larvae remain short in length (<2 mm, compared with wt counterparts which reach ∼5 mm in length), but their eyes were larger displaying almost pin-point pupils surrounded by dense iridophore pigment cells and a fused optic fissure. There was no functional swim bladder which precludes feeding, and the presence of a residual yolk sac and mild pericardial oedema was seen (Fig. 8F and G).
Retinal histology of treated gup mutants confirmed the fusion of the optic fissure in parasagittal views. A shallow notch at the inferior border of the eye in the gentamicin-treated larvae was present (Fig. 8C and D); however, this was fully resolved by 9 dpf (Fig. 8F and G). In addition, at 6 and 9 dpf, the retinal sections showed absence of lens material in treated gup larvae, the corresponding void being filled with proliferating ganglion cells and ectopic cellular laminae (Fig. 8C–G). In the GTTR experiment, aminoglycoside entry was restricted to the retina in wt larvae at 6 dpf, and GTTR did not penetrate the lens (Fig. 3A). This suggests that the primary site of drug action is the retina, hence closure of the optic fissure is seen in all treated gup larvae, but rescue of lens material has not been successful. The RPE and photoreceptor cell layer appear normal, but other retinal nuclear layers are thicker than wt controls.
Treated gup mutants were further examined to determine whether the optic fissure closed at the normal time of 48 hpf. wt embryos showed complete fusion of the optic fissure by 48 hpf (Fig. 9A), whereas untreated gup mutants show a large open optic fissure with a small dysplastic lens suspended superiorly from the retinal neuroepithelium and unattached inferiorly (Fig. 9B). Treated gup mutants at 48 hpf show a spherical eye with partial optic fissure closure, but the process is not complete (Fig. 9C and D). Lens material is present, but appears to have no anchorage and is positioned superficially on the surface of the retina.
Effect of aminoglycoside treatment on apoptotic cell death in zebrafish mutants
The histological analysis of mutant chm, noi and gup retina treated with aminoglycosides revealed increased thickness of the inner nuclear and ganglion cell layers. Therefore, TUNEL assays were performed on wholemount and coronal retinal sections of wt, untreated and treated mutant embryos at 6 dpf to assess levels of cell death. In all mutants treated with either 100 µm gentamicin or paromomycin at 10 hpf, apoptotic activity was greatly reduced throughout the larvae and within the retina compared with untreated mutants (Fig. 10), which may have accounted for the thicker nuclear layers. Cell death was particularly marked at the site of the coloboma in gup mutants, but in treated larvae, apoptosis was negligible in the corresponding region traversed by the fused optic fissure (Fig. 10F and G). This indicated that apoptosis might play a significant role in optic fissure morphogenesis.
Anti-apoptotic therapy in gup mutants
Curcumin is an anti-apoptotic agent which can modulate enzyme activity and gene expression (35,36), whereas zVAD-fmk is a specific cell-permeable peptide caspase inhibitor (37). These compounds were used to test whether a reduction in apoptosis could induce closure of the optic fissure in coloboma mutants. Dose–response experiments suggested a dose of 5 µm curcumin had no toxic side effects in wt larvae at 6 dpf (Fig. 1D), and this was consistent with previous studies in zebrafish embryos (38). At 48 hpf, curcumin-treated gup mutants displayed a slightly smaller optic fissure and systemic features similar to untreated embryos (Fig. 9E). At 6 dpf, the size of the optic fissure was greatly reduced (Fig. 11C) and there was no cell death at the position of the optic fissure (Fig. 11D). The reduced size of the coloboma and reduction in TUNEL staining in embryos treated with zVAD-fmk (Fig. 11E and F) were very similar to the curcumin-treated embryos. Embryos treated with the zFA-fmk control inhibitor still displayed a large optic fissure (Fig. 11G) and TUNEL labelling at the site of the persistent optic fissure (Fig. 11H).
In this study, we have demonstrated that aminoglycoside-mediated readthrough can substantially ameliorate zebrafish eye phenotypes due to premature termination codons and also an increased overall survival of these models with otherwise embryonic lethal mutations.
The zebrafish model of choroideraemia is the only animal model available with a nonsense mutation in the rep1 gene, similar to the human condition (55% homology between the human and zebrafish rep1 proteins). Zebrafish lack an orthologue to the human REP2 gene, thus the uncompensated loss of rep1 results in multisystemic defects which are not seen in humans (22). Western blot analysis revealed low levels of rep1 protein in untreated chm mutants, predominantly from residual maternal rep1 derived from the yolk sac. This contribution enables relatively normal development of embryos until around 3–4 dpf, as the maternal rep1 is utilized, retinal degeneration proceeds with worsening pericardial and abdominal oedema, leading to death at around 5 dpf. Aminoglycoside treatment resulted in the rescue of the mutant embryos by significantly reducing systemic defects, thus increasing survival, and dramatically improving the ocular pathology to a near-normal appearance. Western blotting, reporter experiments and in vitro prenylation assays provided evidence that this rescue was due to functional full-length rep1 being produced by aminoglycoside-mediated readthrough. At 9 dpf, protein expression was still apparent although at lower levels, but this did not correspond with a worsening of the ocular phenotype. The reduction in protein expression could be due to less readthrough efficiency secondary to decreased aminoglycoside availability, although the GTTR experiments did show that gentamicin was still present in the retina up to 9 dpf after a single dose.
It is not known which amino acids replace the stop codons during readthrough, although it will be a near-cognate tRNA anti-codon, thus limiting the options for substitution (39). The amino acid substitution may have no effect on the function of the protein or it may have some functional constraint affecting protein activity, thus acting like a REP1 missense mutation (40). The in vitro prenylation assays confirmed that there was additional rep1 activity, so presumably the amino acid substitution did not have a substantial inhibitory effect. The Q32 amino acid is in the conserved SCR1B domain at the N-terminal of rep1 (41), but is not a consensus residue that is found in evolutionarily diverse species (GxxVLHxDxxxYYG, where the boldfaced x is amino acid 32). The R139X mutation of pax2.1 is in the third α-helix of the C-terminal subdomain of the conserved 128 amino acid-paired DNA-binding domain (PSVSSINRIIRTKVQQ, where the boldfaced R is amino acid 139). No mutations have been found in PAX2 in this subdomain, so the effect of an amino acid substitution is not known. However, comparisons across all nine known paired box genes indicate that the arginine at this position is highly conserved (42). The Q524X in lamb1 is in the fifth EGF-like domain consisting of 60 amino acids, which contains eight conserved cysteine residues that are critical for disulphide bridges (43). The glutamine residue is next to one of the cysteines, but whether a missense mutation at this position would have an effect on the formation of the disulphide bridges is not known.
In addition to the readthrough mechanism of aminoglycosides, we also noted reduced apoptotic cell death activity in all three treated zebrafish models, especially in the region of the optic fissure of the eye. In previous studies, TUNEL-positive cells have been observed in the presumptive optic fissure area prior to its formation, in the lips of the embryonic fissure, but are absent after the fissure has closed in the mouse eye (44). In mouse embryos with spontaneous open-fissure abnormalities, the area adjacent to the persistent fissure was highly TUNEL-positive. This suggests a transient apoptotic requirement for the formation and persistence of the optic fissure. We therefore hypothesized that a reduction in cell death would be associated with the closure of the optic fissure. To investigate this, gup coloboma mutant embryos were treated with apoptosis inhibitors curcumin or zVAD-fmk. Our results showed that the size of the coloboma and the level of apoptosis were greatly reduced by both drugs, suggesting that reducing cell death improved optic fissure morphogenesis in this mutant, although the precise mechanism through which this occurred remains to be identified. However, in no case was the optic fissure fused. During normal optic fissure closure, laminin expression (defective in gup mutants) is strong at the site of the fusing optic fissure (45), thus inhibiting apoptosis alone would not be sufficient to overcome the underlying fusion defect. Nevertheless, these observations suggest that a second treatment modality might be possible in ocular coloboma, via caspase inhibition, to limit the severity of the colobomatous defect. Interestingly, zVAD-fmk has been administered to pregnant mice carrying embryos heterozygous for a Pax2 mutation modelling renal–coloboma syndrome, without noticeable side effects (46). The caspase inhibitor rescued the defective nephrogenesis in the kidney; however, the eye phenotype was not evaluated.
The clinical applications of aminoglycoside treatment could be far-reaching. Genetic eye diseases currently remain largely untreatable. Choroideraemia, for example, usually affects adolescent males who present with night blindness and progress to total blindness by middle age. Our in vivo results suggest that early administration of aminoglycosides to individuals who have been genotyped for nonsense mutations could prevent, arrest or slow disease progression. However, it is likely that treatment would have to be continuous in these cases. In terms of treating congenital malformations of the eye, the management is more controversial, as drug administration would have to take place prenatally. The associated oto- and nephrotoxicity are the limiting factors when considering the use of aminoglycosides as readthrough agents (29,30). Local administration could partly overcome this problem, for example using intravitreal injections or implantation of slow release drug systems within the eye. Furthermore, a critical aspect to readthrough efficiency is the level of nonsense transcripts available for drug action. Normally, these transcripts are degraded by the nonsense-mediated mRNA decay (NMD) pathway (47). In a study examining the varying response of cystic fibrosis patients to gentamicin treatment, patients whose cells had high levels of nonsense transcripts had improved readthrough efficiency, and a better response to the drug in vivo (48). Patients whose cells had low levels of transcripts were treated with cycloheximide to inhibit NMD and this improved the readthrough efficiency in these cells. Thus, the ability of pharmacological compounds to be effective readthrough agents may ultimately depend on the NMD state of target cells.
Recently, a new class of small molecule drugs have been developed on the basis of high-throughput screens identifying compounds that promoted nonsense suppression (6). PTC-124 is an achiral compound with 1,2,4-oxadiazole linked to fluorobenzene and benzoic acid rings. It is not an antibiotic and shares no structural similarity to aminoglycosides, consequently lacking the associated toxic side effects. PTC-124 is orally bioavailable and can be used at significantly lower dosages to obtain higher levels of readthrough compared with aminoglycosides (49). Phase-I studies in healthy volunteers have begun to characterize the safety profile of this drug (50). Phase-II clinical trials are currently underway using PTC-124 in cystic fibrosis and Duchenne muscular dystrophy patients.
In conclusion, this study provides proof of concept for using pharmacological agents to readthrough premature termination codons to treat genetic eye disease caused by nonsense mutations. This is the first study that validates a viable approach for treating choroideraemia and ocular coloboma by using in vivo animal models. Future use of safer, more efficient readthrough drug compounds could provide a practical treatment option for a range of patients facing the prospect of blindness secondary to nonsense mutations. The zebrafish model system provides a rapid preclinical screening tool, and the use of several mutant models in this study highlights the potential applicability of aminoglycoside and/or anti-apoptotic treatment across a range of diseases.
MATERIALS AND METHODS
Zebrafish strains (wt AB, chmru848, noitu29a and gupm189) were maintained in aquaria according to standard protocols (51). Embryos were generated by natural pair-wise matings of identified heterozygous carriers and raised at 28.5°C on a 14 h light/10 h dark cycle in a 100 mm2 petri dish containing aquarium water. The developmental stages are given in hpf/dpf, according to morphological criteria (52). For some experiments, 0.2 mm phenylthiourea (PTU, Sigma) was added to the embryos at 10 hpf to prevent pigment formation.
Luciferase reporter assay
For cloning zebrafish wt and mutant luciferase constructs, 50 wt embryos at 24 hpf were homogenized in Trizol (Invitrogen) and total RNA was extracted following the manufacturer’s instructions. The ThermoScript™ RT-PCR System (Invitrogen) was used to generate cDNA from 1 µg total RNA. RT-PCR was carried out using 100 ng cDNA, PfuUltra II Fusion HS DNA polymerase (Stratagene) and gene-specific primers containing NcoI linkers listed in Table 1. PCR products were purified using S-400 microspin columns (Amersham Pharmacia Biotech) and cloned into pCR Blunt-II TOPO vector (Invitrogen). Clones were sequenced to ensure validity and then insert sequences were excised with NcoI and cloned into the NcoI site of pGL3 control vector (Promega). Nonsense mutations (boldfaced) were engineered into the wt constructs by site-directed mutagenesis and confirmed by sequencing (chm mutagenesis primer CTGCTCCAGAGTTGGATAAAGTGTTTTGCATCTGG, noi mutagenesis primer CAGGATAATTTGAACCAAAGTTCAGCAGCC and gup mutagenesis primer CCTGTATCGGGTTAGTGTCAGTGTCGGGAGC). Plasmid DNA for transfection was purified with maxiprep columns (Qiagen) and quantified using the NanoDrop® ND-1000 Spectrophotometer.
|Gene||Primers||Annealing (°C)||Size (bp)|
|chm forward primer||ACCATGGGCCGCCACCATGGCTGCGGAGGATCTCCCGTCT||64||496|
|chm reverse primer||TCCATGGGCGCAGAATCTTCATTCTCTTC|
|noi forward primer||ACCATGGGCCGCCACCATGGATATTCACTGCAAAGCAGAC||71||538|
|noi reverse primer||TCCATGGGGTCATTGGAAGCGCTTGACACAGGTGG|
|gup forward primer||GCCACCATGGGTGGTGCAATCAACAACAATTGCTCTCCT||68||192|
|gup reverse primer||TCCATGGGTCTTGAGGATGATTTCTCGGGACAAC|
|Gene||Primers||Annealing (°C)||Size (bp)|
|chm forward primer||ACCATGGGCCGCCACCATGGCTGCGGAGGATCTCCCGTCT||64||496|
|chm reverse primer||TCCATGGGCGCAGAATCTTCATTCTCTTC|
|noi forward primer||ACCATGGGCCGCCACCATGGATATTCACTGCAAAGCAGAC||71||538|
|noi reverse primer||TCCATGGGGTCATTGGAAGCGCTTGACACAGGTGG|
|gup forward primer||GCCACCATGGGTGGTGCAATCAACAACAATTGCTCTCCT||68||192|
|gup reverse primer||TCCATGGGTCTTGAGGATGATTTCTCGGGACAAC|
COS-7 cells were maintained in Dulbecco’s modified Eagle’s medium, supplemented with 10% fetal calf serum and 10 000 U penicillin/streptomycin at 37°C in a humidified 5% CO2 incubator. Lipofectamine™ (Invitrogen)-mediated transient transfections were carried out as described previously (53) using 1 µg DNA/well in a six-well plate. Three independent transfections were performed per construct with Renilla luciferase (pRL-CMV vector) co-transfected as an internal control reporter of transfection efficiency. After 24 h, cells were dosed with 100 µm gentamicin (Sigma) or 100 µm paromomycin (Sigma) and further incubated for 48 h. The Dual-Luciferase® Reporter Assay System (Promega) was used to measure dual-luciferase activity in cell lysates using a Luminometer 400 (Nichols Institute Diagnostics).
Dechorionated wt embryos at 10 hpf were treated with test compounds (gentamicin, paromomycin from 50 µm to 2 mm) that were directly added to the aquaria water and then raised until 6 dpf. For each treatment, 30 embryos were used and three independent experiments were performed. Outcomes were based on mortality and toxicity effects including behavioural and morphological defects at 6 dpf. Results are presented as mean ± SEM. Embryos surviving to 6 dpf with no signs of toxicity were prepared for histological analysis of the retina.
Gentamicin and paromomycin treatment
chm, noi and gup mutant embryos were dechorionated at 10 hpf and treated with either 100 µm gentamicin, 100 µm paromomycin or kept in control (drug-free) aquaria water. Survival of mutant larvae was recorded in days, and statistical analysis using the Mann–Whitney test compared survival between control (untreated) and experimental (treated) groups. Significance was defined as P < 0.001, n = 50 for each treatment group. Further experiments using 50 and 150 µm gentamicin were tested on chm embryos to confirm that the optimum dose for this study was 100 µm on the basis of initial survival experiments.
Retinal histology and morphological studies
At 6 and 9 dpf, embryos from each treatment group were collected and fixed in 4% paraformaldehyde (PFA) overnight at 4°C. They were washed three times in PBS and dehydrated through a graded ethanol series (50, 70, 90% and three times in 100%). Embryos were transferred into cedar wood oil and then embedded in paraffin wax. Microtome sections of 5 µm were cut, mounted on Superfrost Plus slides and stained with H&E. Retinal histology was imaged using a digital camera mounted on an Olympus 1×71 microscope and wholemount images for morphological analysis were taken with a Leica MZ16F stereomicroscope.
To assess aminoglycoside penetration into the retina, mutant and wt embryos were dosed with GTTR conjugate. A 2 mg/ml stock solution of Texas Red®-X, succinimidyl esters (Invitrogen Molecular Probes) was prepared in DMSO. For conjugation, 2.2 ml of gentamicin (50 mg/ml stock) was added to 0.3 ml of TR stock solution and agitated overnight at 4°C in the dark. The high molar ratio of gentamicin to TR esters (300:1) produces a conjugation ratio of one TR molecule to a single gentamicin molecule. chm, noi, gup and wt embryos were dosed with 100 µm GTTR conjugate from 10 hpf to 6 dpf. Stacked images of the retina were taken using the Leica SP5 inverted confocal microscope.
PTU-treated embryos at 6 dpf from each treatment group were fixed in 4% PFA and embedded in wax as described before. Retinal sections were dewaxed by washing twice in histoclear (National Diagnostics), followed by two washes in 100% ethanol and once in 70% ethanol, before rinsing in deionized H2O. For wholemount TUNEL assays, embryos were dehydrated through a graded methanol series (25, 50, 75% and twice in 100%) and stored in 100% methanol at −20°C. After rehydration, both sections and wholemount embryos were digested with proteinase K (10 µg/ml) for 15 min and 1 h, respectively. Embryos were refixed with 4% PFA for 20 min at room temperature, followed by several washes in PBS. Following the manufacturer’s instructions, the ApopTag® Peroxidase In Situ Apoptosis Detection Kit (Millipore) was used to detect levels of apoptotic cell death.
Dose–response experiments for curcumin (1–50 µm) were performed on wt embryos to determine the highest dose that could be used without toxic effects. Twenty gup mutant embryos were dechorionated at 10 hpf and treated with 5 µm curcumin, 300 µm zVAD-fmk, 300 µm zFA-fmk (negative control for zVAD-fmk) or kept in control (drug-free) aquaria water until 6 dpf. Embryos were fixed in 4% PFA overnight at 4°C, followed by wholemount TUNEL assays to assess levels of apoptotic cell death. Images were acquired for morphological analysis.
Protein was isolated from 20 embryos per treatment condition at 6 and 9 dpf: wt, untreated chm (only 6 dpf as mutants do not live to day 9) and chm mutants treated with 100 µm gentamicin or 100 µm paromomycin. Embryos were snap-frozen in liquid nitrogen and homogenized in lysis buffer (10 mm Tris, pH 7.5, 10 mm NaCl, 1% SDS, 1× Roche Protease Inhibitor Cocktail) by sonication. Protein concentration was measured using the BCA™ Protein Assay Kit (Pierce). For each sample, 30 µg of protein was boiled for 5 min with an equal volume of SDS sample buffer. Proteins were separated on a 10% SDS–polyacrylamide gel and transferred onto an Immun-Blot™ PVDF membrane (BioRad). The membrane was blocked for 2 h in blocking solution [5% dry milk, PBS/0.1% Tween (PBST)], washed three times in PBST and then incubated overnight at 4°C with 1:1000 primary anti-Rep J905 antibody (54). Following three washes in PBST, the membrane was incubated with a secondary anti-rabbit IgG HRP conjugate (dilution 1:10 000, GE Healthcare) for 2 h. The membrane was washed three times in PBST before chemiluminescent detection using the ECL™ Western Blotting Detection Kit (GE Healthcare). Using Restore™ Western Blot Stripping Buffer (Pierce), the membrane was stripped and reprobed with polyclonal anti-gapdh antibody (Abcam) as a loading control for each sample. AIDA Image Analyzer software was used to determine the relative abundance of rep1 protein compared with corresponding levels of control gapdh.
In vitro prenylation assays
Twenty embryos per treatment condition at 6 dpf were used: wt, untreated chm and chm mutants treated with 100 µm gentamicin or 100 µm paromomycin. Embryos were snap-frozen in liquid nitrogen and homogenized in hypotonic buffer (50 mm HEPES, pH 7.5, 10 mm NaCl, 1 mm DTT, 1× Protease Inhibitor Cocktail) by sonication. Nuclei were pelleted by centrifugation at 800g for 10 min, then the postnuclear supernatant was collected and centrifuged at 100 000g for 1 h at 4°C using a TLA45 Beckman rotor to obtain cytosolic protein extracts. In vitro prenylation was performed in reaction buffer containing 50 mm HEPES, pH 7.5, 5 mm MgCl2, 1 mm NP40, 1 mm DTT, 1× Protease Inhibitors (Roche), 1× Phosphatase Inhibitors (Sigma). Each 25 µl reaction contained 2 µm recombinant REP, 100 nm recombinant rat RabGGTase, 50 nm [3H]GGpp supplemented with 1 µm unlabelled GGpp and 100 µg zebrafish cytosolic extract containing a source of unprenylated Rabs. HEK 293 cells are prenylated and act as a negative control, whereas HEK 293 cells treated with the prenylation inhibitor mevastatin contain unprenylated Rabs, hence this was used as a positive control for the assay; 100 µg of each was used. After incubation for 45 min at 37°C, 5 µl of loading buffer was added to each reaction and aliquots were boiled for 5 min at 95°C. Aliquots were separated by 12.5% SDS–polyacrylamide gel, and the proteins fixed by incubation for 1 h in a solution of 50% methanol and 10% acetic acid. Gels were incubated for 30 min in AMPLIFY solution (Amersham) before drying in a 60°C oven and exposure to autoradiography film for 2 weeks. AIDA Image Analyzer software was used to determine the relative abundance of prenylated Rab proteins compared with levels of control HEK 293 cells treated with mevastatin.
This research was supported by St Mary’s Hospital Development Trust (grant ref. PVB/MRW 2005 and L44 2007), The British Retinitis Pigmentosa Society (grant ref. GR550) and the Crystal Family Charitable Trust.
The authors thank Dr Derek Stemple at the Wellcome Trust Sanger Institute for kindly providing us with the gup zebrafish line, Dr Kate Lewis at the University of Cambridge for the noi zebrafish line, Dr Martin Spitaler at the Imperial College Imaging Facility (FILM) for his valuable expertise and Miss Loraine Lawrence, Senior Technician and Histologist, Department of Leukocyte Biology, Imperial College London.
Conflict of Interest statement. None declared.