INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Although much is known about the molecular aspects of neuromuscular junction (NMJ) assembly (Hall and Sanes 1993; Sanes and Lichtman 1999), there is less information about how functional transmission is established in vivo. Embryonic junctional potentials in a variety of preparations are small, slow, and variable (Grinnell 1995), and how they are transformed into large, fast and uniform mature events is unclear. The speculation ranges from changes in presynaptic release of acetylcholine (ACh) to the postsynaptic density of ACh receptors (AChRs) and acetylcholinesterase (AChE) but the exact mechanism remains unknown.

The zebrafish is an interesting preparation for the physiological study of NMJ development due to the accessibility of the embryo (Buss and Drapeau 2000; Liu and Westerfield 1988; Nguyen et al. 1999; Westerfield et al. 1986). The miniature end-plate currents (mEPCs) in day-old zebrafish embryos, particularly in superficial muscle cells, have small and variable amplitudes and slow time courses (Nguyen et al. 1999). Two days later in newly hatched larvae, mEPCs are observed more frequently and are much larger and faster. Indeed these mEPCs are among the fastest NMJs (Macdonald and Balnave 1984), consistent with the zebrafish being one of the fastest swimming teleosts of its size (Plaut 2000).

The developmental increase in frequency of zebrafish mEPCs is thought to be due to an increased density of innervation as the muscle cells become poly-innervated (Myers 1985) and remain so in the adult (Westerfield et al. 1986). The increase in mEPC amplitude may be due to an increased density of AChRs (Liu and Westerfield 1992). It could also reflect a slowing of ACh clearance or an increase in the content of the transmitter quantum. The origin of the acceleration in the time course of the mEPCs remains unresolved. It is due partly to a gradual loss of the electrical coupling between muscle fibers that filters extraneous events but also reflects the appearance of faster synaptic events (Buss and Drapeau 2000). How these faster events develop remains unknown.

AChR channels with similar brief openings were observed during stationary exposure to submicromolar ACh in both embryonic and larval muscle cells, and blocking AChE with eserine slowed the mEPC time course in larvae, but not embryos, by twofold (Nguyen et al. 1999). These observations suggested that a change in AChR or AChE properties is insufficient to account entirely for the order of magnitude acceleration in mEPC time course. When larval AChRs are activated under more physiological conditions with brief (1 ms) application of saturating (1-10 mM) ACh, they are blocked initially by positively charged ACh (K_0.5 = 0.5 mM) entering the open channels (Legendre et al. 2000). During synaptic transmission at mature larval NMJs, this yields a rebound current on recovery from open channel block when ACh is removed from the cleft. The occurrence of open channel block can thus serve to detect the presence of a high concentration of released ACh. In the present studies, we examined the properties and distributions of AChRs and AChE and the ultrastructure of the NMJ in zebrafish embryos and larvae to determine their relative contributions to the maturation of fast neuromuscular transmission.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Zebrafish (Danio rerio) were obtained from breeding colonies (Westerfield 1995) and were anesthetized with tricaine before starting the experimental procedures.

Electrophysiology

Isolated patch recordings were obtained as described by Legendre (1998), and the extracellular recording solution (Legendre and Korn 1994) was modified to contain less Ca (0.6 mM) and more Mg (10 mM) to suppress contractions. Outside-out patches (Hamill et al. 1981) were obtained from near the middle of the muscle fibers as the small size of the NMJs prevented us from directly locating them. Fast-flow applications were performed by rapidly moving (in <0.1 ms) a twin-barreled application pipette across the patch pipette, which was displaced with a piezoelectric translator (Model No. P245.30, Physic Instruments). One barrel contained extracellular solution and the other contained in addition 10 mM ACh. Current was recorded with an Axopatch-1D amplifier (Axon Instruments), filtered at 10 kHz (3 dB), and stored using a digital tape recorder. Data were acquired with pClamp 6.0 software (Axon Instruments) by digitizing at 50 kHz and were analyzed off-line with Axograph 3.5 software (Axon Instruments).

Extracellular focal mEPC recordings were obtained as described in Xenopus by Kullberg et al. (1977, 1980) by placing an extracellular electrode at the myoseptal junctions of superficial muscle fibers. Spontaneously occurring synaptic activity was recorded using an Axoclamp 2A (Axon Instruments) in bridge mode. Recordings were performed and analyzed as described in the preceding text.

The mEPCs were recorded in the whole cell mode as described previously (Nguyen et al. 1999) in the presence of 250 nM tetrodotoxin using pipettes with a series resistance of 4-5 M that was compensated by 70-80%. The recordings were made using an Axopatch 1-D (Axon Instruments), were filtered at 2 kHz, and were analyzed as described in the preceding text.

Histology

Zebrafish were anesthetized, skinned and fixed for 1 h at room temperature in phosphate-buffered saline (PBS) containing 2% paraformaldehyde. Small clusters of fixed muscle fibers were isolated from the adult trunk musculature. The fixed preparations were rinsed with PBS and labeled with -bungarotoxin (-BTX) conjugated to Oregon Green and fasciculin (Fas2) conjugated to rhodamine or Oregon Green as described by Peng et al. (1999). The tissues were incubated for 10 min in PBS containing 10% horse serum (PBS/HS) to reduce background, then incubated for 30-60 min in the same buffer containing 1 µg/ml fluorescent -BTX and Fas2. After washing for 30 min in several changes of PBS/HS, the muscle fibers were mounted in 90% buffered glycerol (pH 8.5) containing 1 mg/ml p-phenylenediamine to reduce photobleaching.

Histochemical staining for AChE was performed according to Vacca (1995). Briefly, whole embryos or larvae were fixed overnight in maleate buffer containing 4% paraformaldehyde and treated for 2 h with Triton X-114 prior to incubation with the buffer-substrate mixture containing acetylthiocholine. AChE was not detected in control preparations incubated without substrate or with substrate and 0.5 mM eserine to inhibit esterase activity (not shown).

Biochemical assay of AChE activity

AChE was extracted from whole larvae or isolated adult muscle by homogenization in 10 volumes (wt:vol) borate extraction buffer [20 mM Na borate, pH 9.0, 1.0 M NaCl, 5 mM EDTA, 1% Triton X-100, and a protease inhibitor cocktail (as described in Rotundo 1984)] and centrifuged for 20 min at 14,000g. Aliquots of the supernatant were analyzed by velocity sedimentation on 5-20% sucrose gradients as previously described (Rotundo 1984). The AChE activity in each fraction was determined using a radiometric assay (Johnson and Russell 1975) and the resulting relative counts per minute (CPM) plotted. Markers consisting of Escherichia coli -galactosidase (16 s) and alkaline phosphatase (6.1 s) were routinely included in the samples and assayed using standard procedures.

Electron microscopy

Dechorionated embryos and larvae were anesthetized and fixed overnight at 4°C in PBS containing 3% glutaraldehyde, washed with PBS, and postfixed for 1 h at room temperature in 2% OsO₄ in PBS. The samples were then washed and dehydrated in an alcohol series and embedded in Epon. Thin cross sections were counter-stained with lead citrate and viewed using a JEOL CX-100 transmission electron microscope.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Fast AChR kinetics and open channel block

We compared the properties of embryonic and larval AChRs using fast-flow perfusion of membrane patches isolated from muscle cells. As shown in Fig. 1A, application of 10 mM ACh for 1 ms to an outside-out patch from a 1-day-old embryo activated a steady outward current at +50 mV (average of 8 trials shown). On removal of the ACh, the current decayed monoexponentially (decay time constant  = 0.39 ms). At 50 mV, the same patch showed initially a smaller steady inward current in the presence of ACh (Fig. 1A) due to open channel block (Legendre et al. 2000). On removal of the ACh there was a large, delayed rebound current that decayed more slowly ( = 0.51 ms) during recovery from the block. When the same protocol was used with a patch from a 3-day-old larva, a similar result was obtained (Fig. 1B). The AChR currents observed at 50 mV at both stages had similar activation time courses (20-80% rise time, RT = 0.05-0.06 ms; Fig. 1C), but the embryonic currents decayed twice as slowly [ = 0.55 ± 0.24 (SD) ms, n = 7] as the larval currents ( = 0.27 ± 0.06 ms, n = 10; P < 0.01; Fig. 1D). However, this was only a minor difference in time course when compared with the much slower mEPCs (RT ~ 2 ms,  = 5-10 ms) observed in superficial embryonic muscle cells (Nguyen et al. 1999) (Fig. 1E). Open channel block (K_0.5 = 0.5 mM) was not observed at submillimolar concentrations of ACh, resulting in voltage-independent time courses (Legendre et al. 2000; Nguyen et al. 1999).

View larger version (23K): [in this window] [in a new window]   Fig. 1. Fast kinetics of acetylcholine (ACh) receptors (AChRs). Responses of outside-out patches from embryonic (A) and larval (B) muscle to 1 ms application of 10 mM ACh, shown by the line drawings at top. Top traces: averages of 8 responses from the same patches that were recorded at either +50 mV (top) or 50 mV (bottom). The decay phases are fit with monoexponential functions with the indicated values of . Histograms (bars indicate ± SE) of (C) the 20-80% rise times (RT) and (D) the values of  recorded in patches from embryos (day 1, , n = 7) and in patches (, n = 10), and extracellular recordings (, n = 3) from larvae (day 3). E: scaled averages of miniature end-plate currents (mEPCs) recorded in an embryonic (27 h) muscle fiber at 30 mV (13 mEPCs) and 110 mV (27 mEPCs). F: average of 113 focal mEPCs from a 3-day-old larva fitted with a monoexponential function with the indicated value of .

As the isolated patches were likely from extrasynaptic regions, we next examined whether synaptic AChRs had similar properties. The mEPCs recorded in whole cells (Legendre et al. 2000; Nguyen et al. 1999) were found to be somewhat slower than the single AChR currents (Legendre et al. 2000). This could be due to a difference between synaptic and extrasynaptic AChRs or perhaps to a technical limitation with whole cell recording due to the large size of the muscle fibers. To overcome this technical limitation, extracellular voltage recordings were made of the junctional currents underlying spontaneous events (focal mEPCs) in larval muscle fibers. The 20% largest events were selected for averaging (Fig. 1F; mean = 90 ± 2 µV) as they were presumably the most accurate recordings of events occurring closest to the electrode. The focal mEPCs had similar kinetics (RT = 0.07 ± 0.01 ms,  = 0.220 ± 0.003 ms) as the larval AChRs recorded in isolated membrane patches, indicating the presence of similar synaptic and extrasynaptic AChRs, as reported for other NMJs (Edmonds et al. 1995; Schuetze and Role 1987).

One possible explanation for the slow activation of embryonic mEPCs is that a low (submillimolar) concentration of ACh is released at immature synapses. If this was so, then open channel block should be less effective. Because embryonic muscles contracted at strongly depolarized potentials, we compared the time course of whole cell mEPCs in fibers held at 30 mV, where the block should be lower, and at 110 mV, where the block should be maximal (Legendre et al. 2000). As shown in Fig. 1E, for the averages of the largest 20% of the recorded mEPCs, which permitted reliable detection and comparison, similar rise times were observed at either potential: 0.60 ± 0.08 ms at 30 mV and 0.72 ± 0.08 ms at 110 mV (n = 6; P = 0.36 by signed-rank test). However, the decay time course of embryonic mEPCs was 26 ± 6% slower at the more negative potential ( = 4.4 ± 0.6 ms at 30 mV and  = 5.7 ± 0.5 ms at 110 mV; n = 6; P = 0.03). [These values for RT and  are somewhat smaller than those reported by Nguyen et al. (1999) due to selection of the largest events.] The larger value of  at the more negative potential indicates that open channel block also occurred at embryonic NMJs, reflecting the release of a high (millimolar) concentration of ACh that should not limit the time course of the mEPCs to the extent observed.

Slower increase in density of AChE vs. AChR

As the twofold change in AChR kinetics was insufficient to account for the order of magnitude acceleration in time course of the junctional currents, we examined the distribution and properties of AChRs and AChE. AChRs cluster a few hours after motoneurons contact muscle cells (Liu and Westerfield 1992). In embryos, AChE is present at myoseptal junctions (Hanneman and Westerfield 1989), and in adult zebrafish, it is localized both at the myoseptal junctions and at the multiple, poly-innervating NMJs running orthogonally across deeper fibers (Mos et al. 1983). We examined the pattern of AChE distribution in comparison with AChRs to determine whether it is synaptically localized by the time of maturation of junctional transmission in larvae.

The distribution of these molecules was compared simultaneously by labeling them with the AChE-selective toxin Fas2 and AChR-selective -BTX (Peng et al. 1999). Clustered AChE was undetectable with Fas2 at either embryonic or larval NMJs (the latter is illustrated in Fig. 2A). Because the presence of AChE was detected pharmacologically in larvae as a twofold reduction in mEPC time course following block by eserine (Nguyen et al. 1999), we also carried out a standard histochemical stain. We were thus able to detect AChE at myoseptal junctions in embryos, as reported previously (Hanneman and Westerfield 1989), and at myoseptal junctions and end plate regions along deep muscle fibers in larvae (Fig. 2B).

View larger version (77K): [in this window] [in a new window]   Fig. 2. Distribution of AChRs and AChE. A: staining of 4-day larval muscle AChE with Oregon Green-Fas 2. B: standard histochemical stain for AChE in a 3-day larva. Note the V-shaped myotomal junctions. C: staining of AChE in adult muscle fibers and at mysoseptal junctions (D: horizontal bands at top) with Oregon Green-Fas 2. E: staining of AChRs in 4-day larval muscle with Oregon Green--bungarotoxin (BTX; same preparation as in A). F: staining of AChRs in adult muscle fibers with Oregon Green-BTX. The dark spots in A, B, and E are pigmented skin cells. The scale bar is 46 µm in A and in C-F and 50 µm in B.

As a control to be sure that Fas2 could effectively stain zebrafish AChE and that a lack of staining in embryos and larvae was thus not an artifact due to a lack of Fas2 binding, we also examined the staining pattern in adult muscle fibers. Strong fluorescent staining was detected as string-like clusters of AChE (Fig. 2C) and at myoseptal junctions (Fig. 2D) in the adult fibers. Thus Fas2 could label zebrafish AChE but it appeared to be less well localized in larval muscle.

In contrast to the weak AChE staining in larvae (Fig. 2A), small and broadly distributed AChR clusters were detected along the fibers and at the myoseptal junctions (Fig. 2E; same preparation as in A). In isolated adult muscle fibers, long continuous strings of AChRs were observed to run across the fibers (Fig. 2F), similar to the pattern of Fas2 staining of AChE (Fig. 2C; same preparation as in F). Thus AChE appeared to be less well localized than AChRs in larval muscle even though the synaptic currents in larvae appeared to be mature.

The weaker AChE staining in larvae could be due to lower expression of synaptic AChE or less dense clustering compared with adult NMJs. To distinguish between these alternatives, we extracted AChE from whole larvae and from adult muscle and analyzed the molecular forms by velocity sedimentation. In all vertebrate species (Massoulié et al. 1993), including adult zebrafish (Bertrand et al. 1998), AChE is present as both globular forms and as predominantly synaptically-localized, collagen-tailed forms associated with the synaptic basal lamina. As shown in Fig. 3 (top) an extract of whole larvae showed two major peaks of activity. The heavier peak (to the left) corresponds to the collagen-tailed form of AChE, whereas the second set of peaks corresponds to the globular monomeric, dimeric and tetrameric forms. A similar but better resolved pattern was observed with adult muscle extracts (Fig. 3, bottom). However, the use of entire larvae (including all tissues) and isolated adult muscle prevented a more quantitative comparison. Nonetheless, comparable proportions of globular and collagen-tailed AChE were present, indicating that synaptic AChE was less densely clustered at larval NMJs.

View larger version (14K): [in this window] [in a new window]   Fig. 3. Molecular forms of AChE in larvae and adult muscle. Equal volumes of extracts of entire larvae (top) and adult muscle (bottom) were centrifuged on 5-20% sucrose gradients. Fractions were collected from the bottom of the tube and assayed radiometrically for AChE activity. The activity is plotted for each fraction collected. The internal sedimentation coefficient markers are -galactosidase (left, 16S) and alkaline phosphatase (right, 6.1S).

Ultrastructural maturation of NMJs

While changes in AChR kinetics (preceding text) and AChE activity (Nguyen et al. 1999) could each account for a twofold acceleration in mEPC decay time course during maturation of the NMJ, a large fraction of the order of magnitude changes in synaptic current during maturation remained unaccounted for. We used transmission electron microscopy to examine whether a structural feature of the synapse could contribute to its functional maturation. Mature larval NMJs were readily identifiable, as reported previously (Waterman 1969; Westerfield et al. 1990) and as found for adult zebrafish (Mos et al. 1983). The NMJs were observed at the point where several muscle fibers surrounded the nerve terminals, where a limited region of the terminal showed a synaptic specialization (Fig. 4, bottom). On average the terminals (n = 22) were 1.2 ± 0.5 µm in diameter (mean of large and small axes). These synaptic regions contained small clear vesicles that clustered at the active zones, which on average (n = 31) were 1.8 ± 0.8 µm in length. We counted 156 vesicles within one vesicle diameter along a total of 43 µm of presynaptic membrane at the active zones, for an average of 3.7 vesicles/µm. The NMJs had a presynaptic thickening, a postsynaptic density and a wide cleft (on average 0.12 ± 0.03 µm) with a dense extracellular matrix.

View larger version (115K): [in this window] [in a new window]   Fig. 4. Ultrastructure of embryonic and larval NMJs. Transmission electron micrograph of a 30-h embryonic NMJ (top) and a 3-day larval NMJ (bottom). The scale bar at the bottom right-hand corner of the top micrograph is 1 µm for both micrographs. Notice the large, poorly differentiated nerve ending in the embryo and the smaller terminal in the larva with a well formed, dense basal lamina in the broad synaptic cleft.

In embryos (30 h), immature nerve endings were occasionally observed at similar points of contact between surrounding muscle fibers (Fig. 4, top) but these lacked an obvious synaptic specialization. The endings were recognized by their larger size (1.8 ± 1.4 µm in diameter, n = 5) compared with the microtubule-containing axons observed in the spinal cord (0.30 ± 0.06 µm in diameter, n = 9; not shown). As expected from the low frequency of mEPCs at this stage (Nguyen et al. 1999), embryonic nerve endings were observed less frequently compared with larval NMJs. Whereas several NMJs were commonly observed in a single cross-section of a given larva (total of 22 NMJs documented in 3 larvae), many cross-sections had to be examined before finding an immature profile in a given embryo (total of 5 documented in 4 embryos). The nerve endings were immature in appearance and could be larger than in larvae (Fig. 4, top) but contained only occasional and sometimes irregularly shaped vesicles, with 13 vesicles observed within one vesicle diameter along 57 µm of presynaptic membrane for an average of 0.2 vesicles/µm. Furthermore, the nerve endings were closely juxtaposed to the muscle cells, with a gap of 0.06 ± 0.01 µm (n = 5), and lacked a synaptic cleft or other specialization. These observations indicate that the nerve-muscle gap in embryos is narrower and more extensive than the wider, less restricted cleft at larval NMJs and could act as a greater diffusion barrier resulting in slower clearance of ACh and prolongation of the time course of embryonic mEPCs.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We examined a number of physiological and morphological parameters at zebrafish NMJs as they developed from an immature embryonic form (on day 1) to become mature larval synapses a few days later. Previous work (Nguyen et al. 1999) indicated an order of magnitude acceleration in time course of the mEPC (both for rise time and decay) during NMJ maturation. We consider how the properties and distributions of key determinants may contribute to shaping neuromuscular transmission in this preparation.

Clustering of AChE is delayed compared with AChRs

The maturation of the zebrafish NMJ was in part due to the delayed appearance of ACh hydrolytic activity as blocking AChE with eserine did not affect embryonic mEPCs but resulted in a twofold slowing of larval mEPCs (Nguyen et al. 1999). An apparent delay of ~1 day occurs between clustering of AChR and AChE in Xenopus (Peng et al. 1999) where blocking AChE activity caused only a minor slowing of the time course of early junctional events (Kullberg et al. 1980). We observed comparable proportions of the collagen-tailed (synaptic) and globular (extrasynaptic) forms of AChE in larvae and adults but detected less intense AChE staining co-localized with AChR staining in larval muscle compared with adult muscle. These results suggest that the increase in synaptic AChE density at the zebrafish NMJ takes many days to occur, longer than for AChR clustering and mEPC maturation. A high level of unclustered AChE could account for the higher level of diffuse staining in larval muscle observed with Fas2.

Release of a high ACh concentration from the onset

Explanations for slow embryonic mEPCs include slow clearance of ACh leading to channel re-openings, a lower concentration of released ACh, slower AChRs or a lower density of postsynaptic AChRs at embryonic NMJs. Due to the lack of significant hydrolysis expected at embryonic NMJs, released ACh should remain at an elevated concentration. The gap between embryonic nerve endings and muscle cells was half that observed over the junctional region of larval terminals and could contribute to slowing ACh clearance and thus trapping released ACh. This would promote channel re-openings and would contribute to both the slow rise time and decay of embryonic mEPCs.

Alternatively, a very low (micromolar) concentration of ACh could be released to slowly activate AChRs. However, the presence of open channel block (K_0.5 = 0.5 mM) (Legendre et al. 2000) at embryonic NMJs indicates that a high concentration of ACh is released from immature nerve endings. Because the embryonic AChR channels have kinetics an order of magnitude faster than the mEPC time course and the latter showed open channel block even after several milliseconds, we presume that the mEPC time course reflects the time course of ACh clearance.

Extrasynaptic vs. synaptic AChRs

During fast-flow application of a saturating concentration of ACh, single AChR channel currents had decay times twofold slower in embryos than in larvae; but in contrast to the mEPCs, the AChR channel rise times were comparable. These results indicate that the channels undergo a minor kinetic modification affecting deactivation and not activation. Recordings of end-plate potentials in adult zebrafish (Westerfield et al. 1986) revealed fast rise times (~1 ms) consistent with the time integral of the AChR currents determined in this study and indicating that fast synaptic AChR kinetics persist in the adult zebrafish. The presence of mature larval junctional currents is consistent with the high contraction rates during swimming both in larvae (Budick and O'Malley 2000; Buss and Drapeau 2001; Eaton and Farley 1973; Kimmel et al. 1974) and in adult zebrafish (Plaut 2000).

The slower deactivation of embryonic AChRs could contribute partially to the shortening of the mEPC decay duration but is not expected to affect the rise time. The slower rise time of the embryonic mEPC is likely due to a low density of extrasynaptic AChRs. It has been estimated that ACh could diffuse several µm to activate extrasynaptic AChRs with the similar slow time course observed at embryonic Xenopus NMJs (Kullberg et al. 1980). This would also account for the smaller amplitude of the embryonic events we observed.

Although tight clusters of AChRs form within a few hours of nerve-muscle contact (Liu and Westerfield 1992), a further increase in the density of AChRs is possible and would be consistent with the formation of the postsynaptic density and the greater mEPC amplitudes observed at the larval stage. The extrasynaptic AChRs are likely reduced in density, perhaps as a consequence of synaptic clustering, to prevent re-openings by ACh diffusing from the cleft which would eventually (after many days) be hydrolyzed by synaptic (and perhaps also by diffusely distributed extrasynaptic) AChE. We conclude that no single feature of AChE or AChR properties or distribution limits maturation of the zebrafish NMJ. Rather it is a combination of convergent changes that determines the amplitude and time course of fast neuromuscular transmission.