|Home | About | Journals | Submit | Contact Us | Français|
Nedd4 (neural precursor cell expressed developmentally down-regulated gene 4) is an E3 ubiquitin ligase highly conserved from yeast to humans. The expression of Nedd4 is developmentally down-regulated in the mammalian nervous system, but the role of Nedd4 in mammalian neural development remains poorly understood. Here we show that a null mutation of Nedd4 in mice leads to perinatal lethality: mutant mice were stillborn and many of them died in utero before birth (between E15.5–E18.5). In Nedd4 mutant embryos, skeletal muscle fiber sizes and motoneuron numbers are significantly reduced. Surviving motoneurons project axons to their target muscles on schedule, but motor nerves defasciculate upon reaching the muscle surface, suggesting that Nedd4 plays a critical role in fine-tuning the interaction between the nerve and the muscle. Electrophysiological analyses of the neuromuscular junction (NMJ) demonstrate an increased spontaneous miniature endplate potential (mEPP) frequency in Nedd4 mutants. However, the mutant neuromuscular synapses are less responsive to membrane depolarization, compared to the wildtypes. Ultrastructural analyses further reveal that the pre-synaptic nerve terminal branches at the NMJs of Nedd4 mutants are increased in number, but decreased in diameter compared to the wildtypes. These ultrastructural changes are consistent with functional alternation of the NMJs in Nedd4 mutants. Unexpectedly, Nedd4 is not expressed in motoneurons, but is highly expressed in skeletal muscles and Schwann cells. Together, these results demonstrate that Nedd4 is involved in regulating the formation and function of the NMJs through non-cell autonomous mechanisms.
Protein ubiquitination and ubiquitin–proteasome mediated degradation have emerged as important mechanisms regulating synapse formation and function (Bingol and Schuman, 2006; Ding et al., 2007; Ehlers, 2003; Speese et al., 2003; Watts et al., 2003). The ubiquitination process is a cascade of reactions catalyzed, sequentially, by the E1 ubiquitin-activating enzyme, the E2 ubiquitin-conjugating enzyme and the E3 ubiquitin ligase (Scheffner et al., 1995), each of which are critically involved in synaptogenesis. For example, bendless, an E2 enzyme in Drosophila, is required for the giant fiber axon to synapse with its motoneuron target (Thomas and Wyman, 1984); UBC-25, an E2 enzyme in C. elegans, is required for the maintenance of neuromuscular function (Schulze et al., 2003). Furthermore, mutations of rpm-1 (regulator of pre-synaptic morphology), a RING (really interesting novel gene) finger-domain E3 ligase in C. elegans (Schaefer et al., 2000; Zhen et al., 2000), or highwire, the homologue of rpm-1 in Drosophila (DiAntonio et al., 2001; Wan et al., 2000; Wu et al., 2005) lead to neuromuscular synaptic overgrowth and transmission defects. Similarly, axonal projection and neuromuscular defects have been reported in the mouse mutant Phr1 (pam, highwire, rpm-1), the murine orthologue of highwire/rpm-1 (Bloom et al., 2007; Burgess et al., 2004; Lewcock et al., 2007), and in zebrafish mutant Esrom, the fish orthologue of highwire/rpm-1 (D’Souza et al., 2005).
Like rpm-1/highwire/Esrom/Phr1, Nedd4 is an E3 ligase, but it is a member of a distinct family of E3s, the HECT (homologous to E6-AP carboxyl terminus) domain E3 family (Ingham et al., 2004; Rotin et al., 2000)). Nedd4 contains one C2 domain, three (or four) WW domains and one HECT domain; such domain architecture is highly conserved from yeast to humans (Ingham et al., 2004; Kumar et al., 1997; Staub and Rotin, 2006). In mammals, Nedd4 is known to down-regulate the epithelial sodium channel (ENaC) (Abriel et al., 1999; Dinudom et al., 1998; Goulet et al., 1998; Harvey et al., 1999; Staub et al., 1996) and other voltage-gated sodium channels including neuronal specific (Fotia et al., 2004) and cardiac specific (van Bemmelen et al., 2004) voltage-gated sodium channels. Studies in Drosophila have shown that DNedd4 (the Drosophila orthologue of Nedd4) regulates cell surface expression of the roundabout (Robo) receptor (Myat et al., 2002) and the transmembrane protein Commissureless (Comm) (Ing et al., 2007), both of which are crucial for target recognition and stabilization of the fly NMJ. It is thought that DNedd4 promotes neuromuscular synaptogenesis in Drosophila by facilitating the endocytosis of Comm (Ing et al., 2007; Wolf et al., 1998).
A question arises what role(s) Nedd4 plays in the formation and function of the mammalian NMJs. In fact, Nedd4 was initially identified as a developmentally down-regulated gene from the mouse brain; Nedd4 mRNA is detected in the neural tube (E8.5), the head (E10–E11) and mouse brain (E13-neonatal stage) and its level regresses as development proceeds (Kumar et al., 1997; Kumar et al., 1992). However, the role of Nedd4 in the development of mammalian nervous system remains largely unknown. To address this issue, we have examined mutant mice deficient in Nedd4. Our analyses show that Nedd4 mutants are perinatal lethal and display profound defects in neuromuscular formation and function. In Nedd4 mutants, the skeletal muscle fiber sizes and motoneuron numbers are significantly reduced. Surviving motoneurons project axons to their muscles on schedule, but motor nerves defasciculate upon reaching the muscle surface, suggesting Nedd4 is not required for muscle target recognition, but is important for proper interaction between the nerve and the muscle. Interestingly, neuromuscular synapses in Nedd4 mutants are established in the central region of the muscle, in a pattern similar to the wildtypes. However, an individual synapse in the Nedd4 mutant is composed of more numerous, yet smaller nerve terminal profiles, compared to the wildtypes. Consistent with these structural defects, electrophysiological analyses revealed functional impairments of neuromuscular synaptic activities in Nedd4 mutant embryos. Interestingly, Nedd4 was not detected in motoneurons, but was highly expressed in skeletal muscles and Schwann cells in the periphery. These results suggest that Nedd4 functions through non-cell autonomous mechanisms to regulate nerve terminal differentiation and function.
Nedd4 mutant mice (Nedd4Gt(IRESBetageo)249Lex) were generated by Lexicon Genetics Incorporated, via retroviral insertion strategy (Raymond and Soriano, 2006). The Nedd4 gene was disrupted between exons 17 and 18, and this strategy also introduced a reporter gene, β-galactosidase (lacZ), into the Nedd4 locus, thus the expression of lacZ could be used to identify cells expressing endogenous Nedd4 (http://www.mmrrc.org/strains/11742/011742.html). We have obtained the heterozygote Nedd4Gt(IRESBetageo)249Lex mice, which are viable, fertile and devoid of gross phenotypic defects (http://www.mmrrc.org/strains/11742/011742.html), from the Mutant Mouse Regional Resource Centers (MMRRC). To generate homozygous mutants (Nedd4Gt(IRESBetageo)249Lex/Gt(IRESBetageo)249Lex, hereafter as Nedd4−/−, or Nedd4 mutant), heterozygotes were time-mated, and the day when a vaginal plug first appeared was designated as embryonic (E) day 0.5. After selected intervals of development, embryos (E10.5–E18.5) were collected by Cesarean section of anesthetized pregnant female mice. Nedd4 mice were genotyped by PCR, using the following primer set: Nedd4 wild type allele-forward GGA GTC TTT GGA TAT TGT AAG AGC, reverse GAG CGT GCG CCT CAC AAG TAT GA; Nedd4 mutant allele-forward AAA TGG CGT TAC TTA AGC TAG CTT GC; reverse GAG CGT GCG CCT CAC AAG TAT GA. All experimental protocols followed NIH Guidelines and are approved by the UT Southwestern Institutional Animal Care and Use Committee.
Immunofluorescence staining was carried out as previously described (Liu et al., 2008). Briefly, muscle samples were fixed in 2% paraformaldehyde (PFA) in 0.1 M phosphate buffer (pH 7.3) at 4 °C overnight, blocked in dilution buffer (500 mM NaCl, 0.01 M phosphate buffer, 3% BSA and 0.01% thimerosal), and then incubated with primary antibodies. The sources of primary antibodies were as follows: S100β(1:500, Swant Swiss Antibodies, Switzerland), neurofilament (NF150) (1:1000, Chemicon, Temecula, CA), synaptophysin (1:100, Dako, Carpinteria, CA), SV2 (1:1000, Developmental Studies Hybridoma Bank, the University of Iowa, Iowa City, IA), HB9 (1:5000, gifts from Dr. Samuel Pfaff, Salk Institute, La Jolla, CA) (Thaler et al., 1999), MuSK (1:500, gift from Dr. Steve Burden, Skirball Institute, NYU Medical Center, NY) and synaptotagmin-2 (1:1000, gift from Dr. Thomas Südhof, Stanford University School of Medicine, Palo Alto, CA) (Pang et al., 2006). Samples were then incubated with fluorescein isothiocyanate-conjugated secondary antibodies and Texas-Red conjugated α-bungarotoxin (α-bgt) (2 nM, Molecular Probes), washed with PBS and mounted in 90% glycerol, 10% Tris Buffer (pH 8.5,10 mM) containing n-propyl gallate (20 mM) to reduce photobleaching (Giloh and Sedat, 1982). Quantification of fluorescence intensity and sizes of AChR clusters was generated from confocal images acquired with identical, sub-saturating gains. The mean gray value (integrated density/total pixels), area, perimeter and Feret’s diameter (the length of the greatest axis) were measured using NIH ImageJ.
Detection of β-galactosidase (LacZ) activity was carried out based upon previously described procedures (MacGregor et al., 1995; Sanes et al., 1986). Briefly, embryonic (E11.5–E18.5) tissue sections (12 μm) or whole diaphragm muscles (E13.5–E18.5) of heterozygote or wildtype (as negative control) mice were fixed in 2% paraformaldehyde (PFA) in 0.1 M phosphate buffer for 1 h, washed three times for 10 min each time in PBS and then incubated in an incubation buffer containing sodium phosphate (150 mM), MgCl2 (2 mM), sodium deoxycholate (0.01%), NP-40 (0.02%), potassium ferricyanide (5 mM), potassium ferrocyanide (5 mM), and 5-bromo-4-chloro-3-indolyl-beta-D-galactopyranoside (1 mg/ml, X-Gal), at 37 °C. Images were acquired on an Olympus BX51 upright microscope with Nomarski optics (for sections) or a Zeiss stereo microscope (SteREO Discovery) (for wholemount diaphragm muscles).
Embryonic tissues were collected at E18.5 and homogenized in Tris buffer containing 50 mM Tris–NaOH, pH 7.4, 150 mM NaCl, 2 mM EDTA, 1 mM PMSF and protease inhibitor cocktail (Roche Applied Science, Indianapolis, USA). Tissue homogenates were separated by SDS-PAGE, transferred to a nitrocellulose membrane, and blocked in 5% milk in Tris-buffered saline. The membrane was incubated with anti-β-galactosidase (Cappel, 1:1000), or anti-Nedd4 (1:10,000, Upstate), or mouse monoclonal anti-alpha-tubulin (1:1000, Sigma-Aldrich) antibodies, followed by peroxidase-conjugated secondary antibodies (1:10,000, Biomol), and visualized with Enhanced Chemiluminescence (ECL, Amersham Biosciences) reagents.
Detection of AChE was based on the methods previously described (Enomoto et al., 1998). Briefly, diaphragm muscles were fixed with 2% PFA, rinsed in PBS and incubated in 0.2 mM ethopropazine, 4 mM acetylthiocholine iodine, 10 mM glycine, 2 mM cupric sulfate, and 65 mM sodium acetate solution at pH 5.5, for 2–4 h at 37 °C. Staining for AChE was developed by incubating the wholemount diaphragm for 2–5 min in sodium sulfide (1.25%, pH 6.0), followed by extensive wash. The diaphragms were then cleared with 50% glycerol in PBS and flat mounted and images were acquired on a Zeiss stereomicroscope.
Wildtype or heterozygote (N=5) and Nedd4 mutant (N=3) embryos (E18.5) were immersion fixed in Bouins solution following decapitation and evisceration. The trunk was trimmed resulting in the thoracic–lumbar spinal column with adjacent vertebra and trunk muscles. The tissue was then processed for paraffin embedding, serially sectioned at 8 μm and stained with thionin (Sun et al., 2003). For motoneuron counts, all motoneurons were counted at 600× in every 10th section through the entire lumbar enlargement using criteria described previously that control for counting the same neuron twice (Clarke and Oppenheim, 1995). The raw cell counts were multiplied by 10 for estimating the total number of lumbar motoneurons. Cell counts were done blinded as to genotype by two independent observers, whose motoneuron counts differed by less than 3%. Counts of DRG neurons were carried out bilaterally in the L4 ganglion (every 5th section) at 400×. The length of the lumbar segments of the spinal cord was estimated by multiplying the total number of sections by the section thickness (8 μm). The volume (μm2) of the lumbar spinal cord was determined from outline drawings using a camera lucida of every 10th cross-section at 100× and this value was multiplied by 10. The volume of the L4 DRG was determined in a similar fashion from every 5th section at 250×. The soma size of lumbar motoneurons and L4 DRG neurons (i.e., cross-section in μm2) was determined from outline drawings of 40–70 cells per embryo at 600× by sampling cells at rostral, middle and caudal lumbar regions and using the same inclusion criteria as for motoneuron cell counts (Clarke and Oppenheim, 1995).
Intracellular recording was performed in acutely isolated phrenic nerve–diaphragm muscle preparation (E18.5) in oxygenated rodent Ringer’s solution (136.8 mM NaCl, 5 mM KCl, 12 mM NaHCO3, 1 mM NaH2PO4, 1 mM MgCl2, 2 mM CaCl2, and 11 mM D-glucose, pH 7.3) as previously described (Liley, 1956). Endplates were located by visual inspection through an Olympus BX51WI upright microscope with a long working distance (3.30 mm) water immersion objective (40×, NA 0.80) and impaled with glass micropipettes (20–40 MΩ). Miniature endplate potentials (mEPPs) were acquired by an intracel-lular amplifier (AxoClamp-2B, Molecular Devices), digitized with Digidata 1332 (Molecular Devices), analyzed with pClamp 9.0 (Molecular Devices) and Mini Analysis Program (Synaptosft, Inc., Decatur, GA).
Samples (phrenic nerve, diaphragm and intercostal muscles) were collected from E18.5 embryos and fixed in 2% paraformaldehyde and 2% glutaraldehyde in 0.1 M sodium cacodylate buffer, pH 7.4. The samples were then rinsed with 0.1 M sodium cacodylate buffer and post-fixed in 2% osmium tetroxide, dehydrated in a graded series of ethanol, infiltrated and polymerized in Epon 812 (Polyscience Inc.). Ultrathin sections were stained with uranyl acetate and lead citrate and electron micrographs were recorded using a JEOL 100CXII operated at 80 KeV.
The data were presented as mean±standard error of the mean (SEM). Statistical differences between the control and mutant groups were determined using Student’s t- or chi-square test, wherever appropriate. A p value of <0.001 was considered to be highly significant, and p value of <0.05 significant.
To investigate the role of Nedd4 during development, we analyzed Nedd4 mutant mice generated from mating of heterozygote (Nedd4+/−) mice. In our initial analyses, we genotyped more than 200 offspring from Nedd4+/− breeding, but we did not obtain any homozygote Nedd4 mutant mice after weaning (21 days after birth). We therefore turned our focus to collecting newborn pups, and we observed that some pups from heterozygous crosses were stillborn and were noticeably smaller than their littermates at postnatal day 0 (P0). We genotyped these mice and confirmed that they were homozygote Nedd4 mutants (Fig. 1A). This led us to further examine mouse embryos at E18.5, a stage immediately prior to birth. At E18.5, Nedd4 mutant embryos showed no spontaneous movement, but displayed weak and uncoordinated movement in response to mild pinches on their tails or legs, suggesting the neural circuits for reflective responses were intact in these mutants. To check the expression of Nedd4 protein in the wildtype and mutant embryos, we collected various embryonic (E18.5) tissues (brain, spinal cord, skin and muscles), and immunoblotted the tissue homogenates with antibodies against Nedd4, or β-galactosidase (β-gal), since a reporter gene lacZ was inserted into the Nedd4 genomic locus (http://www.mmrrc.org/strains/11742/011742.html). As expected as a reporter, β-gal was detected in Nedd4 mutant (Fig. 1B) and heterozygotes (data not shown), but not in the wildtype tissues (Fig. 1B). Consistent with previously reported Nedd4 gene expression pattern in mouse embryos (Kumar et al., 1997; Kumar et al., 1992), Nedd4 protein was detected in the brain, spinal cord, skeletal muscle and skin homogenates from the wildtype embryo (Fig. 1B). However, no Nedd4 protein was detected in Nedd4−/− embryo (Fig. 1B); these results demonstrated the insertional mutation in the Nedd4 gene results in a null phenotype.
To determine the lethal period of Nedd4 null mice, we collected and genotyped 282 embryos (E10.5–E18.5) from heterozygote crosses. We compared the observed ratio of wildtype (+/+), heterozygote (+/−) and homozygote (−/−) embryos to the expected Mendelian ratio of 1:2:1. These results are summarized in Table 1. At E10.5–E13.5, ratios were very similar to the predicted Mendelian ratio. By E15.5–E16.5, fewer Nedd4 homozygote embryos were obtained compared to the number predicted from the Mendelian ratio (chi-square test, p=0.399), and by E18.5, the number of Nedd4 homozygote embryos was significantly reduced (chi-square test, p=0.004). These data indicate that many Nedd4−/− embryos died perinatally in the uterus (between E15.5 and E18.5).
Previous Northern blot and in situ hybridization analyses have demonstrated that Nedd4 is highly expressed in developing mouse brain (Kumar et al., 1997; Kumar et al., 1992). However, the tissue expression pattern of Nedd4 in the nervous system has not been carefully documented. For example, it is not clear where Nedd4 is expressed in the spinal cord or in the peripheral nervous system. To address this issue, we carried out immunohistochemical staining using Nedd4 antibodies. Unfortunately, antibodies against Nedd4, while able to recognize Nedd4 by Western blotting (Fig. 1B), did not give rise to specific labeling in mouse tissues in situ – a common situation for some antibodies that react well with denatured conformation (e.g., after exposure to SDS), but not with native conformation (e.g., in tissue sections) (Willingham, 1999). In order to infer the distribution of endogenous Nedd4, we examined the distribution of β-galactosidase encoded by the lacZ reporter gene inserted into the Nedd4 genomic locus. Since the heterozygote Nedd4 mice were grossly normal, the experiment was conveniently performed in heterozygote mice using X-gal staining. At E11.5, LacZ was readily detected in the skin (Figs. 1C, D, arrow) and within the marginal layer of the embryonic spinal cord (Figs. 1C, D, arrowhead); however, no LacZ was detected in the rest of the spinal cord, except a small region around the central canal (Figs.1C, E, and data not shown). At later stages (E13.5–E18.5), LacZ staining remained highly concentrated to the surface of the embryonic spinal cord (Fig. 1F, arrow); higher magnification of the region revealed that the staining was localized to the embryonic meninges (Fig. 1I). Within the spinal cord, LacZ was detected only in a small group of cells localized around the central canal (cc) (Fig. 1J); the major regions of the spinal cord, including motoneurons, were not labeled by LacZ (Fig. 1E, and Fig. 1H, arrowheads). In contrast, the dorsal root ganglion (DRG) was heavily labeled (Figs. 1F, G); the LacZ-positive cells within the DRG appeared to be predominately small, flattened Schwann cells, or satellite cells (Fig. 1G). Schwann cells in the peripheral nerves such as the phrenic nerves were also LacZ-positive (Fig. S1, D). In addition, the cartilage primordium of vertebrae surrounding the spinal cords was also intensely labeled (Fig. 1F, white arrow). Furthermore, the skeletal muscles were positively labeled by X-gal staining. LacZ in the muscle was detected as early as E13.5 (Supplementary Fig. S1A) and persisted throughout development (E15.5, Supplementary Fig. S1, B, and E18.5, Figs. 1K–L). To determine the expression of Nedd4 in the brain, we carried out LacZ staining on cross-section at various embryonic stages (E11.5, E13.5, E16.5 and E18.5). We consistently observed that LacZ-positive cells were primarily localized to embryonic meninges (data not shown) and the ventricular zones surrounding the ventricles (Fig. S2, B). In addition, LacZ-positive cells were also detected across the retina, especially within the retina ganglion cell layer (Fig. S2, A).
Since Nedd4 is expressed in embryonic skeletal muscles (Figs. 1K, L, Supplementary Fig. S1, A, B), we assessed muscle development in Nedd4 mutant embryos. Under a dissecting microscope, we noticed that diaphragm muscles in Nedd4 mutant embryos were markedly thinner and much more fragile than their wildtype or heterozygote littermate controls. The mutant muscle fibers appeared wavy and disorganized, and this phenotype was consistent among all mutants examined (N=15). Figs. 2A, B illustrates phalloidin-stained diaphragm muscles from the wildtype vs. Nedd4 mutant. In the wildtype, muscle fibers are aligned in parallel arrays (Fig. 2A), whereas in Nedd4 embryos, muscle fibers appear wavy and disorganized (Fig. 2B). Cross-sections of diaphragm muscles revealed that the calibers of individual muscle fibers were significantly reduced in Nedd4 mutants (Fig. 2D); the average area of muscle fibers was 89.4±3.5 μm2 in the Nedd4 mutants (n=146, number of muscle fibers), vs. 135.8±4.7 μm2 (n = 183, number of muscle fibers) in wildtype muscles; the difference in muscle fiber sizes between the mutant and wildtype is highly significant (p<0.001). Reduced muscle fiber sizes were also detected in limb muscles of the Nedd4 mutant embryos (data not shown). There was no evidence of muscle necrosis or regeneration in the Nedd4 mutants, since muscle nuclei were peripherally localized in Nedd4 mutant muscles, similar to the wildtypes (Figs. 2C, D). To further examine the fine structure of muscle fibers, we examined the diaphragm muscle sections under electron microscope. Interestingly, despite reduction in sizes, ultrastructure was well-preserved in Nedd4 mutant muscles; the Z-lines were clearly visible and the sarcomeres were properly assembled (Figs. 2E, F), suggesting muscle ultrastructure developed normally in the absence of Nedd4.
Since Nedd4 mutant mice were stillborn, we asked whether the formation and function of their neuromuscular junctions were affected in the absence of Nedd4. We began our analyses on the diaphragm muscle, since neuromuscular synapses in embryonic diaphragm muscles are among the earliest to form (Allan and Greer, 1997; Bennett and Pettigrew, 1974). During development, the phrenic nerves (p) enter the diaphragmatic muscle near the mid-costal regions and then divide into secondary (s) intramuscular branches that innervate the costal and crural diaphragm (Bennett and Pettigrew, 1974). Consistent with the previous reports (Allan and Greer, 1997; Bennett and Pettigrew, 1974), our results show that, in the wildtype, the intramuscular nerve branches appear as single nerve bundles that travel perpendicular to the long axis of the muscle fibers; numerous tertiary nerves (t) then extended from secondary branches and terminated within the central region of the muscle (Fig. 3E). Such a pattern was established as early as E13.5 (Fig. 3A). In Nedd4−/−embryos, the diaphragm muscle was also innervated at E13.5 (Fig. 3B), indicating that axons were properly guided to their target muscles. However, the secondary intramuscular nerves in the Nedd4−/−muscles were markedly defasciculated throughout development (arrowheads in Figs. 3B, D, F and H), and often developed abnormal loops (arrows in Figs. 3B, D and F). Furthermore, as development progresses to later stages (E15.5–E18.5), it became apparent that the phrenic nerves in the Nedd4−/− embryo (N=8) failed to completely innervate the whole diaphragm, leaving a gap at the ventral region of the diaphragm (demarcated by a bracket in Fig. 3D and Fig. 3H). In contrast, the ventral portion of the diaphragm in the wildtypes was fully innervated by phrenic nerves (N=6) (Figs. 3C, G). Thus, the absence of innervation in the ventral portion of the diaphragm in the Nedd4 mutant can be attributed to the failure of nerve growth, instead of nerve degeneration. Interestingly, despite defasciculation of the secondary nerve branches, the tertiary nerve branches in the Nedd4−/− embryos were mainly confined to the central region of the muscle, in a pattern similar to that shown in the wildtype embryos (Figs. 3E and F). These data demonstrate that while Nedd4 is not required for general target recognition of muscle, it is required for proper nerve fasciculation during development.
Next we asked if Schwann cell development was affected in Nedd4 mutants, since Nedd4 expression was detected in Schwann cells (Fig. 1G), and Schwann cells are involved in nerve fasciculation (Lin et al., 2000; Woldeyesus et al., 1999; Wolpowitz et al., 2000) and neuromuscular synapse formation and function (Herrera et al., 2000; Reddy et al., 2003). Using Schwann marker S-100β, we observed that S-100βpositive Schwann cells were present in spinal nerve roots (E18.5), pre-terminal nerves and nerve terminals (E15.5–E18.5) of both wildtype and Nedd4 mutant embryos (data not shown). Under EM, Schwann cells were also detected in nerve trunks (Figs. 4G, H) and nerve terminals (Figs. 8A, B) of both wildtypes and Nedd4 mutants. Thus, Schwann cells appeared to differentiate and migrate normally, although we cannot rule out possible functional defects of Schwann cells in Need4 mutant embryos.
From our wholemount analyses, we noticed that the mutant phrenic nerve appeared thinner than the wildtype phrenic nerve (Figs. 3E, F), suggesting a possible reduction of axon number and/or axon size in the mutant nerves. We therefore examined cross-sections of the phrenic nerves by electron microscopy and measured the number and size of individual axons. As shown in Fig. 4, the caliber of the phrenic nerves was markedly reduced in Nedd4 mutant (Fig. 4H), compared to the wildtype (Fig. 4G). We counted all axons from three pairs of wildtype and Nedd4 mutant phrenic nerves: there were 53±10.9 axons in the mutant, and 259±7.2 in the wildtype. The number of individual axons was significantly reduced in Nedd4 mutant embryos (p<0.001, Student’s t-test). Axon size, however, remained similar between the two genotypes: the average diameter of individual axons in the wildtype was 1.54±0.03 μm (n=224, number of axons); whereas the average diameter of individual axons in the Nedd4 mutant was 1.39±0.08 μm (n=36, number of axons).
Decrease of axon number in Nedd4 mutant nerves suggested a possible decrease in number of motoneurons. We therefore investigated motoneurons in embryonic spinal cords in Nedd4−/− embryos. First, we carried out immunostaining on E10.5 spinal cord sections using antibodies against HB9, a motoneuron marker (Arber et al., 1999; Li et al., 1999; Thaler et al., 1999). As shown in Fig. 4B, HB9-positive motoneurons were localized to ventral spinal cords in both wildtype (Fig. 4A) and Nedd4 mutant (Fig. 4B) embryos, but HB9-positive motoneurons appeared fewer in the Nedd4 mutant compared to the wildtype. We further quantified motoneuron numbers in a defined segment of the spinal cord (lumbar spinal cord, L4) and confirmed that the number and size of motoneurons were indeed significantly reduced in Nedd4 mutants (Fig. 4F, Table 2), compared to their littermate controls (Fig. 4E, Table 2). In addition, L4 spinal cord length, volume, and L4 DRG volume were also significantly reduced in the Nedd4 mutants compared to the controls (Figs. 4C, D, Table 2). Thus, the reduction of muscle fiber sizes, motoneuron numbers, spinal cord and DRG volumes likely reflects overall reduction of animal size in the Nedd4 mutants (Fig. 1).
We next examined the development of neuromuscular synapses in Nedd4−/− embryos. First, we stained diaphragm muscles with Texas-Red conjugated α-bungarotoxin to label post-synaptic AChRs. As illustrated in Fig. 5, the distribution of AChRs in the wildtype and Nedd4−/− muscles was similar: AChRs were clustered along the central region of the muscle (the endplate band) (Fig. 5A). The mean gray value of fluorescence intensity of individual AChR clusters was similar between the wildtype and Nedd4−/−muscles: wildtype,100± 15% (n=197 clusters, N=4 embryos); mutant: 96±15%, (n=145 clusters, N=3 embryos) (Fig. 5B). The average area, perimeter and Feret’s diameter (the length of the greatest axis) of AChR clusters were slightly increased in Nedd4−/− muscles, when compared to wildtype, but the difference was statistically insignificant (Fig. 5B, P>0.05). Thus, formation of AChR clusters in Nedd4−/− muscles appears normal. Consistent with AChR clustering, the enzyme AChE, a reliable marker for post-synaptic membrane (Rotundo, 2003), was also localized to the central region in Nedd4−/− muscles, a pattern similar to that seen in the wildtype muscles (Fig. 5C).
To further examine the spatial relationship between pre-synaptic nerve terminals and post-synaptic AChR clusters, we double-labeled wholemount diaphragm muscles with post-synaptic marker α-bungarotoxin and a number of pre-synaptic markers. In both wildtype and Nedd4 mutant muscles, nerve terminals were intensely labeled by synaptic vesicle proteins such as synaptotagmin 2 (arrowheads in Fig. 5D), synaptophysin (Supplementary Fig. S3) or SV2 (data not shown), and the nerve terminals formed juxtaposition with post-synaptic AChR clusters (Fig. 5D, Supplementary Fig. S3). These results demonstrate that neuromuscular junctions were established in the absence of Nedd4.
The muscle-specific receptor tyrosine kinase, MuSK, plays critical roles in neuromuscular synaptogenesis (DeChiara et al., 1996; Kim and Burden, 2008), and its surface expression is regulated by a PDZ domain containing RING-finger E3 ligase (PDZRN3) (Lu et al., 2007). To investigate if synaptic localization of MuSK requires Nedd4, we immunolabeled muscle sections with anti-MuSK antibodies. As shown in Supplementary Fig. S4, MuSK was localized to synaptic sites in Nedd4−/− muscles, in a pattern similar to that seen in wildtype muscles. Thus, Nedd4 is not required for MuSK expression or localization.
To examine the function of the NMJs in Nedd4 mutants, we carried out electrophysiological analyses. First, we measured spontaneous miniature endplate potentials (mEPPs) at resting membrane potential. Consistent with previously reported low mEPP frequencies in mammalian embryonic muscles (Dennis et al., 1981; Diamond and Miledi, 1962), the average mEPP frequencies in wildtype muscles was 1.0±0.1 min−1 (n=32 muscle fibers, N=6 embryos). Interestingly, the mEPP frequencies in Nedd4 mutants were significantly increased: the average mEPP frequencies were 2.6±0.3 min−1 (n=49 muscle fibers, N =4 embryos) (p<0.001, Student’s t-test) (Fig. 6B). The average mEPP amplitudes were moderately increased: wildtype, 2.4±0.3 mV (n=20 muscle fibers, N=3 embryos), Nedd4−/−, 3.2±0.2 mV (n=47 muscle fibers, N=4 embryos), but this increase was not statistically significant (p>0.05, Student’s t-test).
To further analyze activity-dependent synaptic transmission in the Nedd4 mutants, we evoked neurotransmitter release by stimulating the nerves electrically. We observed that nerve stimulation invariably resulted in muscle action potentials followed by muscle contraction, with similar amplitudes in both the wildtypes and Nedd4 mutants (Fig. 6D). Occasionally, we were able to record sub-threshold, evoked endplate potentials (EPPs). However, since muscle sizes were significantly reduced in Nedd4 mutants (Fig. 2), the interpretation of various sizes of EPPs between wildtype and Nedd4 mutant muscles was complicated by changes of input resistance of muscle fibers, which are inversely related to the muscle fiber diameter (Bekoff and Betz, 1977; Harris and Ribchester, 1979). We therefore focused our analyses on mEPP frequency changes induced by membrane depolarization, which induces activity- and calcium-dependent neurotransmitter release at the NMJs (Plomp et al., 2000; Protti and Uchitel, 1993; Schoch et al., 2006). Under membrane depolarization (40 mM K+), synaptic activity (measured by changes in mEPP frequency) was drastically increased in both wildtype and Nedd4 mutants. However, the increases in mEPP frequencies induced by membrane depolarization were far greater in the wildtype, compared to the Nedd4 mutant: in the wildtype, membrane depolarization led to over 100-fold increases of mEPP frequency (105.9±9.6 min−1, n=22, compared to spontaneous mEPP frequency 1.0±0.1 min−1, n=32); whereas in the Nedd4 mutant, identical condition led to an increase of mEPP frequency by only 25-fold (68.8±8.9 min−1, n=12, compared to spontaneous mEPP 2.6±0.3 min−1, n=49). Thus, a given membrane depolarization evoked less neurotransmitter release in the Nedd4 mutant NMJs, compared to the wildtypes.
We next asked if the neurotransmitter release evoked by membrane depolarization at the NMJs of the Nedd4 mutants was Ca2+-dependent. We measured the frequencies of mEPPs at various external Ca2+ concentrations (0.5, 1, 2 and 4 mM) in the presence of 40 mM K+. As illustrated in Fig. 6C, despite lower mEPP frequencies in Nedd4 mutant when compared to wildtype, the Ca2+-dependency of mEPP frequency change was similar and the Ca2+-curves were in parallel between wildtype and Nedd4 mutant. These results demonstrate that Nedd4 is not required for Ca2+-dependent neurotransmit-ter release at the NMJs.
Why were the spontaneous mEPP frequencies higher, whereas the evoked mEPP frequencies lower, in Nedd4 mutants, compared to the wildtypes? We considered the following two possibilities. First, individual muscle fibers in Nedd4−/− embryos may contain more synaptic sites than wildtype muscle fibers, thus the increases of mEPP frequencies could result from a summation of synaptic activity in multiple synapses. Alternatively, neuromuscular synapses in Nedd4−/− muscles might contain more nerve terminals (compared to wildtype), thus the increases of mEPP frequencies would be due to releases of neurotransmitter from more nerve terminals at the same synaptic site (Bennett and Pettigrew, 1974).
To test the first possibility, we dissociated embryonic diaphragm muscles (E18.5) into a single muscle fiber and counted the number of AChR clusters in each fiber. As shown in Fig. 7, the majority of muscle fibers in both wildtype (n=562 fibers, N=3 embryos) and Nedd4−/−(n=368 fibers, N=3 embryos) contained only one AChR cluster (1 patch). That is, a single AChR cluster was seen in 97.0±0.4% of wildtype muscle fibers, and 96.7±0.7% of Nedd4−/− muscle fibers; a small but comparable fraction of the muscle fibers had two clusters (2 patches): 3.0±0.4% in wildtype, 3.3±0.7% in Nedd4−/− muscles. Thus, the majority of muscle fibers in both wildtype and Nedd4−/−embryos contain a single synaptic site (single endplate).
To test the second possibility, we carried out ultrastructural analyses of the NMJs. As illustrated in Fig. 8, the overall ultrastructure of the NMJs was preserved in the Nedd4−/−(Fig. 8B), compared to the wildtype (Fig. 8A). In both genotypes, pre-synaptic terminals were filled with synaptic vesicles and capped by terminal Schwann cells, and the basal lamina was clearly visible at the synaptic cleft. Interestingly, multiple nerve terminal profiles were seen at the same synaptic site. Thus we counted the number of nerve terminal profiles at each synaptic site in both wildtype and Nedd4−/− embryos. The average number of nerve terminal profiles per synapse was 2.4±0.4 (n=9) in the wildtype, which is consistent with previously reported numbers (Bennett and Pettigrew, 1974; Purves and Lichtman, 1980). Thus, on average, each embryonic neuromuscular synapse in rodents is innervated by 2–3 nerve terminal profiles. However, compared to the wildtype synapses, there was a 50% increase in the number of nerve terminal profiles at the mutant synapses: the average number of nerve terminal profiles per synapse was 3.7±0.4 (n=13) (Figs. 8D, E). In contrast to the increased nerve terminal profile numbers, the profiles of individual nerve terminal branches were significantly reduced in size in the mutant: the average size of the nerve terminal profiles in Nedd4−/− embryos was 0.36 ±0.05 μm2 (n = 46), compared to 0.66±0.12 μm2 (n=20) in wildtype (Fig. 8C, p<0.005, Student’s t-test). Thus, there were more numerous, yet smaller nerve terminals at the NMJs of Nedd4 mutants, compared to the wildtype.
Within each nerve terminal profile, docked vesicles (the vesicle that attached to the pre-synaptic membrane facing post-synaptic specialization) were present in both wildtypes and Nedd4 mutants. Since vesicle numbers varied from terminal to terminal, we determined the ratio of docked and undocked vesicles. In the wildtypes (n=12 nerve terminals), there were 23 docked, and 401 undocked vesicles (ratio of docked/undocked=0.057); in the Nedd4 mutants (n=33 nerve terminals), there were 42 docked, and 779 undocked vesicles (ratio of docked/undocked=0.054). Thus, there were fewer vesicles per nerve terminal profile in the Nedd4 mutant (24.9±3.2, n=33), compared to the wildtypes (n=35.3±10.2, n=12). However, the ratio of docked and undocked vesicle remained similar between wildtypes and Nedd4 mutants. These ultrastructural changes are consistent with the electrophysiological alternation of the NMJs in the mutants (Fig. 6): the increase in spontaneous mEPP frequency in the Nedd4 mutants could be attributed to the increase of nerve terminal profiles per synapse; whereas the decreased response (compared to the wildtypes) to membrane depolarization could be attributed to the reduction of synaptic vesicle numbers at each nerve terminal profile in the mutants. Thus, the neuromuscular synapses in the Nedd4 mutants appear less mature, compared to the wildtypes.
Here we have investigated the physiological role of an E3 ubiquitin ligase Nedd4 in mammalian neural development by analyzing Nedd4 null mutant mice. Our results reveal that Nedd4-deficiency leads to perinatal lethality and abnormal neuromuscular structure and function. This suggests that Nedd4 plays crucial roles in mammalian embryonic development and is required for survival. One explanation for the observed lethality is a possible respiratory failure resulting from motoneuron loss and/or muscle weakness. Alternatively, the lethality may be attributed to a broad disruption of essential cellular functions, such as cell growth and/or proliferation, as Nedd4 mutant mice are markedly smaller than controls. Consistent with this possibility, a recent report demonstrates that Nedd4 regulates animal growth by positively regulating insulin-like growth factor 1 (IGF-1) and insulin signaling (Cao et al., 2008). It is of interest to note that Rsp5, a yeast orthologue of Nedd4, is essential for yeast viability (Hein et al., 1995), and that Nedd4-2, a homologue of Nedd4, is not required for mouse survival (Shi et al., 2008).
Using lacZ as a reporter gene for Nedd4 expression, we show that Nedd4 is selectively expressed in specific regions of the brain and spinal cord during development, most notably in the embryonic meninges and regions around the ventricles (in the brain) and central canal (in the spinal cord). Nedd4 was also detected in the skeletal muscle and DRG throughout development. In contrast, Kumar et al. (1997) reported that Nedd4 was ubiquitously expressed in the embryonic brain by in situ hybridization. However, individual cells were not readily recognizable in the results reported by Kumar et al., probably due to the methodology (in situ hybridization). Nevertheless, their results (Fig. 4B from Kumar et al., 1997) indicate that some regions of the embryonic brain are more intensely labeled (e.g., the surface of the brain and regions surrounding the ventricles), compared to other regions of the brain (Kumar et al., 1997). In this regard, our results obtained from LacZ staining are consistent with this differential expression pattern revealed by in situ hybridization (Kumar et al., 1997). That is, LacZ-positive cells were primarily localized to embryonic meninges and the ventricular zones surrounding the ventricles in the brain. However, we cannot rule out the possibility that low levels of Nedd4 beyond the limit of detection by LacZ staining may also be present in other regions of the brain.
Our results show that Nedd4 is involved in multiple stages of neural development in mice. During early development, Nedd4 is important for normal cell growth and/or proliferation; in the absence of Nedd4, cell number and sizes are grossly reduced. This phase of Nedd4 function appears to be global. For example, the sizes of muscle fiber, the volume of spinal cord and DRG are all markedly reduced in Nedd4−/− mice. Interestingly, Nedd4 is up-regulated in disuse (unloading or denervation) -induced muscle atrophy in adult rats (Koncarevic et al., 2007), suggesting Nedd4 is involved in activity-dependent muscle atrophy in adult animals. However, the underlying mechanism of disuse-induced atrophy in adult muscles is likely distinct from reduction of muscle fiber sizes displayed in Nedd4 mutant embryos.
In Nedd4−/− embryos, motor axons are properly guided to their target muscles, thus Nedd4 is not required for general target recognition. However, Nedd4 is specifically required for proper nerve fasciculation, as motor nerves become defasciculated upon reaching their muscle surface. These results suggest that Nedd4 plays important roles in mediating nerve–muscle interaction. This phenotype is reminiscent of the muscle innervation defects reported in Drosophila following a reduction of dNedd4 levels in the muscle by RNAi (Ing et al., 2007). These results suggest that Nedd4 plays important, evolutionarily conserved roles in mediating nerve–muscle interaction.
The phenotype developed in Nedd4 mutants appears distinct from several other mutant mice that display developmental defects at the NMJs. For example, mutant mice lacking MyoD, a member of the myogenic regulator factors (Buckingham, 1992), although viable and fertile with no apparent morphological abnormalities in the skeletal muscle (Rudnicki et al., 1992), develop abnormalities in their NMJs. Their endplate bands are expanded and their pre-synaptic nerves branch extensively, compared to the wildtypes (Wang et al., 2003). What remains unclear is whether the neuromuscular synaptic transmission is compromised in MyoD mutants. On the other hand, skeletal muscle fiber sizes are markedly reduced in mutant mice deficient in choline acetyltransferase (ChAT) (Brandon et al., 2003; Misgeld et al., 2002) or the γ-subunit of AChRs (Koenen et al., 2005; Liu et al., 2008; Yampolsky et al., 2008). The endplate bands are also broadened and nerve branching is increased in these mutants. In contrast, the endplate bands are confined within the central region of the muscle in Nedd4 mutants, despite marked reduction of muscle fiber size in Nedd4 mutants. Thus, reduction in muscle fiber sizes does not appear to influence synaptogenesis in Nedd4 mutants, although we could not rule out the possibility that the decreases in muscle fiber sizes may attribute, at least in part, to the reduction of motoneuron number in these mutants.
The most striking pre-synaptic phenotype in Nedd4 mutants is defasciculation of the intramuscular nerves. It remains to be further elucidated why Nedd4 is needed for proper nerve fasciculation. Our results have demonstrated that Nedd4 is expressed in the skeletal muscles and the Schwann cells, but not in motor neurons. Thus, the underlying mechanism of nerve defasciculation in the absence of Nedd4 is likely non-cell autonomous. Further identification of specific target molecules regulated by Nedd4 will shed light on elucidating the underlying mechanisms. One potential candidate could be β-catenin, a molecule critically involved in Wnt signaling. Interestingly, the defects of nerve fasciculation in the Nedd4 mutant are strikingly similar to those reported in β-catenin-deficient muscles (Li et al., 2008). In addition, the reduction of tissue mass in Nedd4 mutants also resembles that observed in β-catenin loss of function mutant mice (Armstrong et al., 2006; Zechner et al., 2003). Furthermore, β-catenin has been identified as a target substrate of RING-finger E3 ligase (Ozz-E3) (Nastasi et al., 2004). It is conceivable that Nedd4 may regulate β-catenin levels in the muscle. For example, Nedd4 may be involved in stabilizing β-catenin in the muscle, which is involved in retrograde regulation of pre-synaptic differentiation at the NMJs in mice (Li et al., 2008). Thus, further examining the involvement of Nedd4 in Wnt/β-catenin signaling may provide important insights.
Despite defasciculation of the nerves and reduction of muscle fiber sizes and motoneuron numbers, functional synaptic connections were ultimately established in the Nedd4 mutants. In fact, the spontaneous mEPP frequency was increased significantly in the mutants. The increase in mEPP frequency could be caused, in principle, by a number of factors – increased endplate number per muscle fiber, increased release sites (active zones) per nerve terminal, or increased nerve terminal profiles per synapse. By manually teasing out single muscle fibers and counting the number of endplates on each muscle fiber, we found no evidence of increased endplates per muscle fiber. However, there are increased nerve profile numbers per synapse in the mutants. Therefore, the increase in spontaneous mEPP frequency in the mutants may likely be a result of more nerve terminals (per synapse) spontaneously releasing a neurotransmitter. Since total motoneuron numbers are reduced in the mutants, the increases in nerve terminal profiles are likely due to increased nerve terminal branching at synaptic sites. Interestingly, each nerve terminal profile in the Nedd4 mutants is smaller and filled with fewer synaptic vesicles. Thus, less vesicles are available for membrane depolarization (by high K+)-induced exocytosis, compared to the wildtypes. This may explain why the increases of mEPP frequency are less dramatic in the Nedd4 mutants (25-fold increase), compared to the wildtypes (100-fold increase).
The developmental defects of neuromuscular junction in Nedd4 mutant mice resemble in some aspects, the defects present in mice deficient for the RING-finger E3 ligase Phr1. For example, in both Nedd4 and Phr1 mutants (Bloom et al., 2007; Burgess et al., 2004), the phrenic nerves are markedly reduced in size and fail to innervate the ventral region of the diaphragm. The width of the endplate bands is also reduced in both Nedd4 and Phr1 mutants (Bloom et al., 2007). However, there are also noticeable differences between Nedd4 and Phr1 mutants. For example, DRG size is reduced in Nedd4, but not in Phr1 (Burgess et al., 2004) mutant, and nerve terminals develop excess sprouting in Phr1 (Bloom et al., 2007; Burgess et al., 2004), but not in Nedd4 mutant. In addition, defasciculation of intramuscular nerves occurs in Nedd4, but not in Phr1 mutant. Thus, while both Phr1 (a RING-finger E3 ligase) and Nedd4 (a HECT-domain E3 ligase) are involved in regulating neuromuscular synapse formation in mice, the underlying mechanisms and substrates are likely distinct. Phr1 is expressed in the motoneuron and regulates neuromuscular development cell-autonomously (Bloom et al., 2007), whereas Nedd4 is expressed in the post-synaptic muscles and functions in a non-cell autonomous manner. The present study, together with a number of recent studies in which molecular manipulations specifically in muscles, but not in motoneurons, e.g. β1 integrins (Schwander et al., 2004), γ-AChR subunit (Koenen et al., 2005; Liu et al., 2008, Yampolsky et al., 2008 #13306) and β-catenin (Li et al., 2008), lead to abnormal development of pre-synaptic nerve terminals, further support an emerging view that post-synaptic muscle cells play critical roles in neuromuscular synaptogenesis.
We would like to thank Laurie Mueller and David Prevette for their excellent technical assistance, and Drs. Natalie Kim and Steven Burden for their valuable suggestions on single muscle fiber preparation. We are indebted to Drs. Thomas Südhof, Jane Johnson, Keith Wharton, Kim Huber, Ege Kavalali and Jonathan Terman for their critical comments on the manuscript. This work was supported by grants (to W. Lin) from NIH/NINDS (NS055028), the Edward Mallinckrodt, Jr. Scholar Program and the Cain Foundation in Medical Research, and NIH/NINDS grant NS53527 (R. W. Oppenheim).
Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.ydbio.2009.03.023.