GNE Myopathy is a rare recessively inherited neuromuscular disorder caused by mutations in the GNE gene, which codes for the key enzyme in the metabolic pathway of sialic acid synthesis. The process by which GNE mutations lead to myopathy is not well understood. By in situ hybridization and gne promoter-driven fluorescent transgenic fish generation, we have characterized the spatiotemporal expression pattern of the zebrafish gne gene and have shown that it is highly conserved compared with the human ortholog. We also show the deposition of maternal gne mRNA and maternal GNE protein at the earliest embryonic stage, emphasizing the critical role of gne in embryonic development. Injection of morpholino (MO)-modified antisense oligonucleotides specifically designed to knockdown gne, into one-cell embryos lead to a variety of phenotypic severity. Characterization of the gne knockdown morphants showed a significantly reduced locomotor activity as well as distorted muscle integrity, including a reduction in the number of muscle myofibers, even in mild or intermediate phenotype morphants. These findings were further confirmed by electron microscopy studies, where large gaps between sarcolemmas were visualized, although normal sarcomeric structures were maintained. These results demonstrate a critical novel role for gne in embryonic development and particularly in myofiber development, muscle integrity and activity.
GNE Myopathy (formerly known as hereditary inclusion body myopathy—HIBM) is a rare recessively inherited neuromuscular disorder characterized by adult-onset, slowly progressive distal and proximal muscle weakness, and a typical muscle pathology including cytoplasmic rimmed vacuoles and filamentous inclusions composed of tubular filaments (1). This disorder has been recognized worldwide, with a prominent cluster in the Jewish Persian community, with a prevalence of 1:1500 (2,3), and in other Jewish Middle Eastern populations. A second cluster has been described in Japan (4,5), where it was termed distal myopathy with rimmed vacuoles (DMRV). We first identified GNE as the disease-linked gene in the large Persian Jewish cluster, where all individuals share a single homozygous missense mutation (M712T) (6). Today, more than 70 different mutations in this same gene have been identified worldwide in non-Jewish patients (7,8), including Japanese (9,10), mostly with compound heterozygosity genotype (11). The product of the GNE gene is a 722 amino acid protein, UDP-N-acetylglucosamine 2 epimerase/N-acetylmannosamime kinase (GNE), a bifunctional enzyme with an epimerase and a kinase domain (12,13). The mutations related to GNE Myopathy are dispersed throughout the entire gene, both in the epimerase and in the kinase coding sequences and very often one mutation in each domain are combined in the compound heterozygous patients.
GNE is the key enzyme in the metabolic pathway leading to the synthesis of sialic acid (14), the most abundant terminal monosaccharide on glycoconjugates in eukaryotic cells (15). GNE is essential for embryonic development: inactivation of the gene by knockout in mice results in embryonal lethality at E8.5 (16). Marked GNE deficiency has not been observed in the patients; in fact, GNE protein is expressed at equal levels in patients and normal control subjects (17). Furthermore, no mislocalization of GNE in patients' skeletal muscle could be documented (17,18): the GNE protein is located at the Golgi compartment in a variety of human cells, including muscle. Although this bifunctional enzyme is well known as the limiting factor in the biosynthesis of sialic acid, no clear events have been recognized to account for in GNE Myopathy pathology. The role of sialylation in GNE Myopathy pathogenesis is controversial because of the broad range of GNE enzymatic activity and of membrane-bound sialic acid levels in healthy individuals and GNE Myopathy patients (19–22). Interestingly, an overall reduction in membrane-bound sialic acid by 25% was observed in various organs of heterozygous GNE knockout mice, which were perfectly healthy and did not develop myopathy, even after 2 years (23). In contrast, a transgenic mouse (Tg hGNED176V GNE(−/−)) overexpressing the human GNE epimerase D176V mutation—which is one of the most prevalent mutations among Japanese patients—on a GNE knockout background also showed hyposialylation in most organs, including muscle, (24) similar to the heterozygous GNE knockout mice (23), but in addition those mice present a GNE Myopathy-like muscle pathology. These controversial data illustrate that the process by which GNE mutations lead to myopathy is certainly not well understood.
Since the GNEM712Tin vivo mouse model and in vitro cell culture model are less informative than expected (24–27), it is critical to develop novel model systems to comprehensively examine the functions of the GNE protein in muscle, which could eventually explain the pathophysiological downstream defects caused by GNE mutations. To date, nothing is known of the role of GNE during development, except for brain (28). Zebrafish is recognized as a very potent vertebrate model for the study of development and of various human conditions, in particular neuromuscular disease. Although the main disadvantage is the relatively long evolutionary distance between zebrafish and humans, a very encouraging observation is the fact that all known proteins associated with skeletal muscle diseases in human are expressed in zebrafish and involved in the same physiological pathways and mechanisms. Moreover, zebrafish embryos have been shown to be a useful model to study gene function involved in muscle development, especially in cases in which mouse knockout studies result in early embryonic lethality (29). Thus, this animal model can be used for elucidating disease mechanism and to evaluate therapeutical strategies (30–33).
In the present studies we have shown that the GNE gene and protein are well conserved between zebrafish and the human ortholog, and have established a zebrafish system to elucidate fundamental biological functions of GNE, specifically in muscle. The most frequent GNE mutated site, M712T, is also conserved in zebrafish gne gene. Utilizing knockdown and transgenic technologies in zebrafish, combined with the ease of monitoring development in this species, we show that gne is generally important for embryonic development and particularly essential for myofiber development, muscle integrity and proper locomotor activity.
Zebrafish gne is highly homologous to the human ortholog
Database search in the zebrafish genome identified one putative copy of the gne gene located on chromosome 1 (NC_007112.5 range: 11591629–11630715). A comparison with the human ortholog (NG_008246.1) revealed a conserved genomic structure of 12 exons of similar sizes. The predicted zebrafish gne gene encodes a protein of 725 amino acids (NP_957177.1), containing identical functional domains as in the human protein (NP_005467.1) and showing 81% identity and 90% similarity (Fig. 1A). Further assessment of the two regions in the protein encoding the epimerase and kinase domains revealed 84 and 77% sequence identity, respectively. The putative leucine-rich nuclear export signal (NES) located in amino acids 121–140 (18) is also highly conserved (90% identity). Interestingly, the region containing the provisional allosteric binding site for the feedback inhibitor CMP-Neu5Ac (34) shows even higher sequence identity (92%). Within this region, the two identified mutated amino acids in Sialuria patients, R-263 and R-266 (35), are conserved in the zebrafish gne protein.
Gne is differentially expressed during embryonic development
Spatiotemporal expression of gne was determined by whole-mount in situ hybridization (ISH) at several developmental stages (Fig. 1C–O). A strong zebrafish gne mRNA expression was first observed in one-cell stage embryos, suggesting a maternal expression, which slowly declined with time (Fig. 1C–H). Ubiquitous gne expression was observed until the first day of development. At 1 day post-fertilization (dpf) a strong gne expression was observed in the eyes and central nervous system, including forebrain, midbrain, hindbrain and spinal cord and a weaker expression in muscle (Fig. 1I and J). Brain and retina expression were highest from 1 to 3 dpf (Fig. 1I–L) and then declined and persisted until 7 dpf. Muscle expression was observed in the first 3 days, appearing higher in inter-somitic boundary in the first day (Fig. 1J), and declined in the following days. At 4 dpf gne was also highly expressed in the gastrointestinal tract and possibly in the liver (Fig. 1M), and diffused by 7 dpf (Fig. 1M–O).
Characterization of Tg(-1.9gne:EGFP) fish
In order to visualize gne expression pattern in vivo we isolated its putative promoter, and first non-coding exon (Ensembl ENSDARE00000745152). This fragment (1946 bp) was sub-cloned in a Tol2 vector (Fig. 2A) and microinjected to further generate a stable Tg(-1.9gne:EGFP) line. The gne:EGFP transgenic fish exhibited a maternal expression pattern of EGFP (shown from 1 to 16 cells; Fig. 2B). At later stages a ubiquitous expression was observed (Fig. 2B and D). A comparison of EGFP expression in the F2 transgenic fish with gne mRNA expression in zebrafish embryos at 1 dpf shows they share a strong expression in the brain, eye and gastrointestinal tract and a weaker expression in muscle (Fig. 2C). At later stages of development EGFP expression was detected apparently in the heart, pronephric ducts, liver and pineal gland (although no organ specific markers were stained in order to confirm these locations) (Fig. 2D–F). EGFP expression was observed throughout larval development, in juveniles and persisted in adult fish, where it appeared strong in eyes and pineal gland and lower in skeletal muscles (Fig. 2G–I).
Support for the aforementioned maternal expression was observed in unfertilized eggs which showed EGFP expression when extracted from mature F2Tg(-1.9gne:EGFP) females (Fig. 2J). To asses maternal origin of expression in early developmental stages two crosses were made; initially, Tg(-1.9gne:EGFP) females were crossed with WT males. The resulting offsprings had a strong EGFP expression, starting at one-cell stage embryos (Fig. 2K), confirming a maternal protein deposition. Eggs derived from the reciprocal cross between Tg(-1.9gne:EGFP) males and WT females started showing EGFP expression only ∼8 h post-fertilization, marking the beginning of zygotic expression (Fig. 2L).
Gne knockdown induces developmental abnormalities
To gain insight into the role of gne in zebrafish embryonic development, we knocked down its expression using two different antisense MOs (Fig. 1B). One MO was designed to target the two putative translation initiation sites, which are separated by 18 bp (ATG-MO), and to result in an inhibition of protein translation. The second MO was designed to alter correct splicing by targeting exon 3–intron 3 splice junction (E3I3-MO), potentially leading to a frame shift and premature termination of translation (Fig. 1B). Injection of either ATG-MO or E3I3-MO into one/two-cell stage embryos caused high mortality rate and lead to a variety of dose-dependent morphological phenotypes (Fig. 3A and B). Mortality and phenotypic deformations were evident before 24 hpf and became more prominent by 3 dpf. Morphants displayed a variety of phenotype severity, and were classified at 3 dpf according to their morphological characteristics into three groups (Fig. 3A); embryos in the normal/mild group either showed slightly reduced body length, smaller eyes or no apparent malformations. Embryos in the intermediate group had a curved tail, small eyes and pericardial edema, and embryos in the severe group had a very short undeveloped trunk and showed all mentioned developmental defects.
Injection of a standard control-MO (1 mm) did not induce these phenotypes (Fig. 3B). Embryos injected with ATG-MO (1 mm, n = 165) showed significantly more severe developmental deformations and more lethality than the ones injected with a similar amount of E3I3-MO (P < 0.001, χ2-test, Fig. 3B), indicating importance of the maternal gne mRNA. To confirm that the observed phenotypes are specific to gne deficiency, we conducted rescue experiments. Co-injection of E3I3-MO (1 mm) with full length capped gne mRNA (40 ng/µl, n = 143) resulted in a significant decrease in mortality and reduced phenotype severity compared with injection of only E3I3-MO (1 mm, n = 146) (P = 0.018, χ2-test, Fig. 3B). These results indicate that zebrafish gne mRNA can partially rescue the developmental defects caused by gne knockdown, and that the observed phenotype is specific to gne deficiency. In order to verify that injection of gne E3I3-MO indeed caused mis-splicing, reverse transcriptase-polymerase chain reaction (RT-PCR) analysis of exons 2–4 was conducted on 3 dpf embryos. As shown in Figure 3C, ∼60–80% of the gne mRNA transcripts were mis-spliced. Sequencing of the RT-PCR products indicated that blocking exon 3 splice donor site activated a 3′ dormant splice site in intron 3, which consequently worked with exon 4 acceptor site (Fig. 1B), leading to an 84 bp insertion. Sequence analysis showed that this insertion had caused a frame-shift and introduced three consecutive premature stop codons that should consequently result in a non-functional truncated protein. As expected ATG-MO did not alter splicing (Fig. 3C).
The efficiency of gne knockdown with ATG-MO was monitored by measuring gne epimerase activity. This enzymatic activity (epimerization of UDP-GlcNAc to ManNAc) is unique to the gne protein and can therefore represent gne protein activity in vivo. By quantifying ManNAc production following an addition of a known amount of UDP-GlcNAc (Morgan–Elson method) (36) we found a significant decrease (−35%) in epimerase activity in 2 dpf ATG-MO morphants with a severe phenotype, compared with uninjected control embryos (t-test, P < 0.001, Fig. 3D). These results indicate that gne ATG-MO knockdown lead to a specific decrease in gne protein levels and therefore in vivo protein activity. Epimerase activity was also analyzed in E3I3-MO (1 mm) morphants, and was found to be decreased, however, the decrease was moderate (−3.9%) and not statistically significant.
Gne knockdown results in impaired locomotor activity
To determine whether gne knockdown affects larval locomotor activity we used a computerized video tracking system to monitor larvae motility in a 48 well plate arena. For this purpose, 7 dpf larvae, initially injected with 1 mmgne E3I3-MO or 1 mm control-MO, were monitored. Only morphants exhibiting a normal/mild phenotype were selected for this experiment. We found that these gne morphants swim at a significantly lower average velocity compared with the control-MO injected larvae (Kruskal–Wallis test, P = 0.043, Fig. 4A). To further validate the specificity of gne MO effect on larval swimming activity we performed a rescue experiment. Co-injection of capped gne mRNA along with E3I3-MO completely rescued the impaired activity (Kruskal–Wallis test, P = 0.023, Fig. 4A), indicating early gne deficiency specifically reduces larval locomotor activity.
Gne knockdown impairs muscular development and structure
The observed decrease in swimming activity due to gne deficiency suggested an aberrant muscle development, therefore lead us to determine whether muscle structure was affected by early gne MO knockdown. The normal/mild 7 dpf larvae that were analyzed in the activity monitoring experiment were sacrificed and their muscle fibers were immunostained with an antibody directed against slow-twitch muscle myosin heavy chain (MyHC, F59) (Fig. 4B). We observed a disorganized muscular architecture in gne morphants. Muscle fibers appeared wavy and narrower, separated by large gaps, in contrast to the straight, aligned fibers in control-MO injected embryos. In addition, the number of slow myofibers per somite in these 7 dpf normal/mild gne morphants was significantly reduced compared with control-MO injected embryos [one-way analysis of variance (ANOVA), P = 0.03, Fig. 4D]. The observed muscle phenotype was rescued by co-injecting capped gne mRNA along with gne E3I3-MO (Fig. 4B and D).
This muscle phenotype was already evident at 1 dpf, throughout the entire trunk musculature, and seemed more intense in the severely impacted morphants, which also lost the classic ‘V’ shaped somites structure, and formed either straight or ‘U’ shaped somites (Fig. 4C). To rule out the possibility that the observed muscle phenotype is a MO off-targeting effect mediated through p53 activation (37), p53-MO (0.33 mm) was co-injected with gne ATG-MO (0.4 mm). The simultaneous gne, p53 knockdown did not prevent the muscle defects described above (Fig. 4C), further confirming that the observed phenotypic alterations are specific to gne knockdown.
To gain further insight into fast and slow muscle fibers and sarcomere integrity of gne ATG-MO and E3I3-MO morphants, they were analyzed at 1–3 dpf by immunostaining with antibodies directed against α-actin (thin filaments in slow and fast muscles; Fig. 5A), α-actinin (Z-lines in slow and fast muscles; Fig. 5B) and myomesin (M-lines in fast muscles; Fig. 5C). These stainings revealed that fast myofibers in gne knockdown embryos were also narrower and not as straight and organized as normal control. However, the analyzed proteins seemed to maintain normal localization in muscle fibers, showing typical M-line and Z-line organization (Fig. 5). These findings indicate that Knockdown of gne affects slow and fast muscle development and morphology, but is probably not involved in thick and thin filament arrangement and sarcomere assembly.
Next, the ultrastructure of skeletal muscles in control, gne ATG-MO and E3I3-MO 2 dpf morphants was examined using transmission electron microscopy. Inspection of longitudinal sections of fast-twitch fibers confirmed that their sarcomere architecture remained as aligned and intact as in the control embryos (Fig. 6A–D). M-lines and Z-lines organization, t-tubule and vertical myosepta structures appeared intact. However, the gaps between adjacent myofibers were substantially larger in gne morphants when compared with the closely packed sarcolemmal boundaries between myofibers in the control muscles (Fig. 6). These gaps frequently contained membranous material (Fig. 6B, D and F). A cross section through these fibers emphasizes the substantial gaps separating them and the smaller diameter of morphants myofibers (Fig. 6E and F).
Altogether, our data indicate that gne is essential for myofiber development and muscle integrity, but dispensable for sarcomere organization in fast and slow muscles of zebrafish embryos.
Although the mutated gene associated with GNE Myopathy was discovered more than 10 years ago (6), the mechanism leading from the mutations to the muscle defect remains unexplained. In order to clarify whether GNE is involved in muscle development and function, we used the zebrafish system, an established model for studying neuromuscular disorders (33). This is the first study to describe the zebrafish homologue of the human GNE gene, and to demonstrate a function for gne in zebrafish muscle development and activity.
The remarkable high level of conservation between zebrafish and human GNE proteins implies a conserved biological function and possibly a conserved feedback mechanism.
Our results provide the first documentation of zebrafish gne maternal transcript and protein deposition. In addition, the less severe phenotype resulting with the splicing MO knockdown, which does not interact with the mature maternal gne mRNA, further supports the importance of this maternal transcript for embryonic development. These observations indicate a role for gne at the very first stages of zebrafish development. The early and ubiquitous gne expression shown here is in agreement with gne expression during mouse embryogenesis (38). GNE mRNA and protein were documented to be highly expressed at early stages of mammalian development, including mouse embryonic stem cells and fetal human myoblasts (18,39). These findings are consistent with the occurrence of sialylation very early during mammalian embryonic development, and its importance in morphogenesis regulation. Notably, the strongest gne expression was observed in the developing nervous system in the first 4 days of development, consistent with the well documented role of polysialic acid chains (PSA) on neural cell adhesion molecules (N-CAM) in mammalian neurogenesis and neural plasticity (40). Increased synthesis of PSA-N-CAM was also observed in the embryonic development of the zebrafish brain, reaching a maximum at 27–40 hpf, and was shown to act in guiding outgrowing axons (41). The low gne expression that was observed in skeletal muscle matches the low GNE expression documented in mouse and human muscle (42).
The expression pattern documented from the in situ data was further confirmed by the Tg(-1.9gne:EGFP) zebrafish. Importantly, gne MO knockdown of Tg(-1.9gne:EGFP) zebrafish embryos did not cause an increase in EGFP expression to compensate for the reduction in gne mRNA levels (data not shown). These data are in accordance with the human data provided by Eisenberg et al. (43) who analyzed the genomic expression patterns of muscle specimens from GNE Myopathy patients compared with healthy matched control individuals. GNE transcript levels were found to be similar in GNE Myopathy patients and control, indicating a lack of positive feedback control mechanism which could compensate for a potential reduction in GNE activity.
In the present study, we demonstrate that gne is essential for normal embryonic development, and that gne deficiency results in structurally and functionally malformed skeletal muscles. Both gne MOs used in this study caused highly increased mortality and lead to similar morphological phenotypes, indicating specificity of their action. This was further demonstrated by co-injection of gne mRNA which resulted in decreased mortality and reduced phenotype severity. The high level of mortality is to be expected since GNE is essential for early development, as indicated from gne knockout mice which results in lethality of the embryo at developmental day E8.5 (16), and by the absence of human patients carrying two null mutations.
Since GNE is the key enzyme in the biosynthesis of sialic acid, we sought to examine whether gne knockdown causes a decrease in sialic acid in gne morphants. However, sialic acid quantification by the periodate–resorcinol method (44) was not reproducible or sensitive enough (data not shown). Previous studies indicate that zebrafish fertilized oocytes contain a large panel of oligosialylated and polysialylated glycoproteins and glycolipids (45,46) which can interfere with sialic acid quantification. They have also established that the majority of this oligosialylation is synthesized in the mother's ovary, prior to embryogenesis. Therefore, the fact that the developing embryo has a source of sialic acid which cannot be affected by gne MO knockdown may imply that the observed knockdown phenotype is not exclusively a result of a lack of sialic acid, but from a different, so far unidentified function of gne in early embryonic development. Following the rationale of the presence of intrinsic sialic acid source in the embryo, we did not attempt rescue experiments by providing exogenous sialic acid or its precursors.
Due to a lack of reliable antibody against gne, we were not able to analyze whether gne ATG-MO knockdown completely diminished protein levels in morphant embryos. However, since UDP-GlcNAc 2-epimerase activity is unique to the gne protein we were able to estimate gne ATG-MO knockdown by quantifying epimerase activity in gne morphants. Unexpectedly, we found a relatively low decrease (35%) in epimerase activity in 2 dpf ATG-MO morphants which presented the most severe phenotype. Since heterozygous gne knockout mice show no pathological condition (with only one allele of gne and an overall measured reduction of 25% in membrane-bound sialic acids) (23), a 35% decrease in epimerase activity is not expected to lead to such a drastic developmental disturbance. These observations further support the assumption made previously, that the documented knockdown phenotype may result from a defect in a yet unknown function of gne during early embryonic development.
Our data demonstrate a function for gne in zebrafish embryonic muscle development, as indicated by gne depletion using two MO oligonucleotides. We established that gne knockdown results in disorganized muscles, smaller fibers, increased gaps between adjacent myofibers, diminished number of slow myofibers and lead to a reduction in locomotor activity. The specificity of these structural and functional phenotypes was confirmed by using two independent gne MOs, by rescue experiments with gne mRNA and by the failure of p53-MO knockdown to rescue the observed phenotypes. Notably, gne depletion analysis was only presented for the most mildly affected morphants, which appeared normally developed, and even they presented a significantly reduced number of myofibers and impaired locomotion.
The fact that this muscle phenotype was observed very early in development in gne morphants, together with the documented maternal and early zygotic expression of gne, indicate a role for gne in early muscle formation. Since MO knockdown is only active in the first few days following injection, the locomotion and muscle phenotypes observed on day 7 post-fertilization (already apparent in 24 hpf morphants) indicate an irreversible damage most likely inflicted earlier in muscle development. This assumption is further supported by the ability of gne mRNA to rescue the structural and functional phenotypes, although it is only available to the cell for a short period of time after injection.
In the myofibrils that did form, the ultrastructure of the sarcomeres appeared intact.
Indeed, from the electron microscopy (EM) studies it is clear that the spacing between two adjacent sarcolemmas is larger than normal, and it appears that the bi-layer lipid plasma membrane was not intact, in some regions. In addition, a large amount of membranous material was found in the gap region. Widened myosepta with gap in skeletal muscles and fiber detachment between adjacent myofibers has been shown in dystroglycan-deficient zebrafish (47). However, in those studies it was found that dystroglycan is not required for muscle formation during early embryonic development of zebrafish. Pathologically, no significant myoseptum damage was seen until 7 dpf and the observed myofiber detachments were caused by extensive tearing of extracellular matrix at the myosepta. Also, mutation in laminin-α2 resulting in a severe muscular dystrophy in early stage fish embryos (48) was described to exhibit damaged myosepta and detachment of myofibers. However, the myofibers that remained attached to the myosepta showed disorganized Z and M lines. Further stainings suggested that sarcolemmal damage observed in these lama2 mutant fish was secondary to extracellular matrix defects, which lead to muscle necrosis during advanced stages of disease progression. The mechanism for the observed subcellular defects in gne knocked down zebrafish is not clear. It could be due to defective sarcolemma formation, defective basement membrane or fiber detachment from the basement membrane.
During zebrafish myogenesis two separate muscle populations form: slow-twitch muscle which is superficially located and medial fast-twitch muscle. Slow muscle fibers are derived from the adaxial mesoderm and their differentiation is induced by Hedgehog signaling (49,50) after which they migrate through the somite to form a superficially positioned single layer. The observed reduction in the number of slow fibers might be due to a failure in muscle differentiation or migration. In order to gain insight into the mechanism leading to this muscle pathology, it will be important to study GNE deficiency effect on the expression of myogenic regulatory factors (e.g. MyoD, Myf5, myogenin), cell migration and differentiation in early muscle development. It also remains to be determined whether knockdown of gne could disrupt basement membrane formation, leading to fiber detachment and increased gaps between myofibers. In order to conciliate these findings, pointing to a crucial role of gne in early muscle development, with GNE Myopathy in humans, where muscle function is completely normal at early stages up to early or even more advanced adulthood, it must be taken into consideration that in contrast to the present studies where gne has been knocked down in zebrafish, GNE Myopathy patients present a normal level of GNE protein. This could result in much slower clinical and pathological damage therefore less obvious at early age. Alternatively, gne could have an additional role in more developed muscle tissue, which cannot be unraveled in the early stages of zebrafish development. Altogether, the data presented here demonstrate that zebrafish gne structure and function are highly conserved compared with the human gene, and establishes a novel important role for GNE in muscle embryonic development. These findings provide a rationale for generating stable transgenic zebrafish lines to further examine the role of GNE mutations in GNE Myopathy pathogenesis.
MATERIALS AND METHODS
Fish and embryo maintenance
Zebrafish were maintained according to standard laboratory conditions (51) in a ZebTEC Zebrafish housing system (Tecniplast, S.P.A., Italy). Fish were maintained at 28°C on a 14 h light:10 h dark cycle, and were fed twice a day. Embryos were obtained by natural spawning and kept in 100 mm Petri dishes in egg water containing methylene blue (0.3 ppm). Embryos and larvae were kept in a 28.5°C light-controlled incubator. Egg water was supplemented with 0.2 mm phenylthiourea to prevent pigmentation in embryos subjected to whole-mount ISH and immunostaining experiments.
All procedures involving animals were approved by the Volcani Center Animal Care Committee and conducted in accordance with the council for experiments on animal subjects, Ministry of Health, Israel.
Cloning of zebrafish gne coding sequence
To prepare a probe for whole-mount ISH experiments, the coding sequence of the zebrafish gne mRNA (NM_200883.1) was amplified using primers GNE CDS 2208 F—incorporating a KpnI restriction site (5′ ggtaccatgcagcgagcgcaggagaagat 3′) and GNE CDS 2208 R—incorporating an EcoRI restriction site (5′ gaattcgtaagttctgcgggtggtgtaatccag 3′). cDNA derived from zebrafish PAC2 fibroblast cells (52) served as a polymerase chain reaction (PCR) template. The PCR product was cloned into a pGEM–T Easy vector (Promega) and sequenced. This vector was later used as a template to transcribe a digoxigenin-labeled antisense RNA probe for the ISH analysis (see below).
Zebrafish gne mRNA expression was detected by whole-mount ISH as described (51,53). Embryos and larvae were fixed overnight in 4% paraformaldehyde at 4°C and stored in 100% methanol. The cloned vector described above was linearized with AatII and served as a template for SP6 RNA polymerase, to produce a digoxygenin-labeled 533 base antisense riboprobe according to the manufacturer's instructions (DIG RNA labeling kit; Roche Applied Sciences). Probe was pre-hybridized and hybridized at 65°C (1 ng/µl), 10–15 embryos were used in each sample. To analyze the ISH signal, embryos were placed in 70% glycerol, observed and photographed using a SZX16 stereomicroscope (Olympus, Japan).
Establishment of a stable transgenic line expressing EGFP under the control of the zebrafish gne promoter
To isolate the zebrafish gne promoter, a 1946 bp fragment containing 1896 bp of genomic 5′ flanking region and 50 bp 5′ UTR of the gne gene was amplified from zebrafish genomic DNA using the primers gnePROM 1946 F, incorporating a SalI restriction site (5′ atagtcgacagtcgctgattaaagttctcctg 3ʹ) and gnePROM 1946 R, containing a BamHI restriction site (5′ ataggatccgtccgacgcgattatctgtagttc 3′). The PCR product was cloned in a pCRII-TOPO TA vector (Invitrogen) and sequenced. The insert was double-digested with SalI and BamHI, and sub-cloned into a BamHI/SalI-digested pT2-AL200R150G (51), upstream of the enhanced green fluorescent protein (EGFP) reporter gene to create the pT2-gne:EGFP construct (Fig. 2A). This construct was later used for the preparation of the Tg(-1.9gne:EGFP) line.
To generate a Tg(gne:EGFP) stable transgenic fish, the Tol2 system was used (54). Plasmids were kindly provided by Koichi Kawakami. Capped transposase RNA was synthesized in vitro using mMESSAGE mMACHINE SP6 Kit (Ambion Inc.). Approximately 2 nl of Tol2 transposase RNA and the pT2-gne:EGFP construct were co-injected (25 ng/µl each) into fertilized eggs at one-cell stage, using a micromanipulator and a PV830 Pneumatic Pico Pump (World Precision Instruments, Sarasota, FL). The injected embryos were raised to adulthood, and the pattern of EGFP expression was monitored throughout their development using an epifluorescent stereomicroscope. The injected fish (F0) were crossed and EGFP expressing fish (F1) were raised to adulthood. Transgenic EGFP-positive lines (F1) were out-crossed with wild-type fish. Genome-integrated transgene was PCR verified and sequenced in the F2 progeny.
MO design and injection
Gene knockdown experiments were performed using MO-modified antisense oligonucleotides (MO; Gene Tools, Philomath, OR). MO sequences were as follows: Gene Tools standard control-MO (5′ ctcttacctcagttacaatttata 3′), p53-MO (5′ gcgccattgctttgcaagaattg 3′), gne ATG-MO (5′ atgcagcgagcgcaggagaagatgg 3′) which was designed to target the two consecutive translation initiation putative codons and gne E3I3-MO (5′ ggatcggtatgtagatgaatggaac 3′) which was designed to interfere with splicing by targeting the exon 3–intron 3 boundary (Fig. 1B). MO oligos were diluted to a final concentration of 0.3–1 mm into 0.1% phenol red to allow for easier visualization during injection. One cell-stage embryos were injected 2 nl and incubated as described above. Injected embryos were monitored under a stereomicroscope and sorted into three groups: normally/mildly altered development, intermediately altered development, and severely altered development. Efficiency of gne E3I3-MO, directed against the splice site, was evaluated by RT-PCR; total RNA was extracted from 20 uninjected, gne E3I3-MO and control MO-injected embryos, with the TRI-Reagent kit (Invitrogen) and was used for cDNA synthesis with the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems) according to the manufacturer's instructions. cDNA fragments were then PCR-amplified using primers directed to zebrafish gne exon 2 (zfGNE_Ex2F 5′ gaagccaaagaaagaaagactccgc 3′) and exons 4–5 boundary (zfGNE_Ex4-5R 5′ catctccttactgcctgcgtcaat 3′) yielding a 778 bp amplicon. PCR products were separated by gel electrophoresis, cut out from the gel, purified and sequenced.
Whole-mount immunostaining was carried out as previously described (55). Briefly, zebrafish embryos were fixed in 4% paraformaldehyde (in PBS) for 1 h at room temperature. The fixed embryos were washed for 15 min 3× times in PBST and digested in 1 mg/ml collagenase. Embryos were washed, permeabilized with cold acetone, washed again and blocked with 10% goat serum. Immunostaining was performed with the following primary antibodies: anti-MyHC for slow muscles (F59, DSHB, USA), anti-α-actinin (clone EA-53, #A7811, Sigma, USA), anti-myomesin (mMaC myomesin B4, DSHB), anti-α-actin (Ac1-20.4.2, Progen, Germany). Secondary antibodies were FITC or TRITC conjugates (Sigma).
Larvae expressing fluorescent reporters and whole-mount ISH-stained larvae were observed and photographed using an epifluorescence SZX16 stereomicroscope equipped with a DP72 digital camera (Olympus, Japan).
Confocal imaging of fixed embryos and larvae subjected to immunostaining was performed using either a LSM780 upright confocal microscope (Zeiss, Germany) or an Olympus BX41 fluorescent microscope (Olympus, Japan).
Epimerase activity assay
Epimerase enzyme activity was determined in wild-type and gne morphant zebrafish embryos by a modified method of Hinderlich et al. (19). Briefly, 30 embryos were pooled at 2 dpf, frozen in liquid nitrogen then thoroughly homogenized and lysed with a 26G syringe in 150 μl lysis buffer (10 mm Na2HPO4, pH 7.5, 0.1 mm EDTA, 0.1 mm DTT and 1 mm PMSF). Lysate was centrifuged at 12 000g for 20 min at 4°C and the supernatant was concentrated using a SpeedVac concentrator (Savant, USA) to a final volume of 53 μl. Three microliters were used for protein determination by the Bradford method (56), using bovine serum albumin as a standard. The assay used to determine GNE epimerase activity contained 45 mm Na2HPO4 (pH 7.5), 10 mm MgCl2, 1 mm UDP-GlcNAc and 50 μl of protein extract in a final volume of 120 μl. The reaction was performed at 37°C for 30 min and stopped by heating at 100°C for 1 min. ManNAc produced in this reaction was detected by the Morgan–Elson method (36). In brief, 75 μl of sample was mixed with 15 μl of 0.8 M H2BO3, pH 9.1, and heated at 100°C for 3 min. Then, 400 μl of DMAB solution (1% (w/v) 4-dimethylamino benzaldehyde in acetic acid/1.25% 10 N HCl) was added and incubated at 37°C for 20 min. The absorbance was read at 578 nm and divided by the protein concentration to correct for sample content. Activity was calculated relatively to WT activity. Significant differences in relative activity between the two groups were determined by a t-test (two-sample assuming unequal variances).
Locomotor activity experiments
To determine whether gne knockdown impairs larval movement abilities, locomotor activity levels were tracked under abrupt light to dark transitions. Embryos were microinjected with either Control-MO, gne E3I3-MO or co-injected with gne E3I3-MO and in vitro transcribed gne mRNA. On day 7 post-fertilization, larvae were placed in a 48-well plate containing 1 ml of embryo water at 28°C in the observation chamber of the DanioVision Tracking System (Noldus Information Technology). The larvae were allowed to acclimatize for 30 min before starting video recording. During the experiment, larvae (n = 48) were subjected to 10 min white light on (light) and 5 min light off (darkness). Live video tracking was conducted using the Ethovision 8.0 software (Noldus Information Technology). Activity was measured as the mean velocity of larvae in each treatment group during the experiment (70 min). Significant differences in activity between the three treatment groups were determined by the non-parametric Kruskal–Wallis test since the data did not meet the assumptions of an ANOVA.
For EM analysis, the specimen was fixed in 2% paraformaldehyde, 2.5% glutaraldehyde, 2 mm CaCl2, in 0.1 M cacodylate buffer (pH 7.2) for an hour and stored at 4°C before processing. Specimen was then washed and postfixed in 1% (w/v) osmium tetroxide, 0.1 M cacodylate buffer for 60 min, followed by en bloc staining with 2% (w/v) uranyl acetate, and dehydration using 30, 50, 70, 90 and 100% ethanol in series. After dehydration, specimen was infiltrated and embedded in spurs resin (Electron Microscopy Sciences, PA) following manufacturer's recommendation. Ultrathin sections were cut on a Leica UC6 ultramicrotome (Leica Microsystems, Inc., Bannockburn, IL), post-stained with uranyl acetate and lead citrate, and examined in a FEI Tecnai T12 transmission electron microscope operated at 80 kV. Digital images were acquired by using an AMT bottom mount CCD camera and AMT600 software.
This work was supported by the Israel Science Foundation (235/11) and Patients Associations to S.M.-R., and by the Maryland Stem Cell Research Fund (2011-MSCRFE-0232 to S.D.).
The authors would like to thank the members of the Gothilf lab (Zohar Ben-Moshe, Shahar Alon, Adi Tovin, Sima Smadja) for their invaluable help and advice, Dr Sharona Even-Ram, Dr Lior Appelbaum and Idan Elbaz for their assistance in confocal imaging and Dr Gitai Yahel for statistical analysis. We also would like to acknowledge the technical support of the Core Imaging Facility of the University of Maryland Baltimore for the EM analysis.
Conflict of Interest statement. The authors declare they have no conflict of interests