Effects of Nitric Oxide on Neuromuscular Properties of Developing Zebrafish Embryos

Nitric oxide is a bioactive signalling molecule that is known to affect a wide range of neurodevelopmental processes. However, its functional relevance to neuromuscular development is not fully understood. Here we have examined developmental roles of nitric oxide during formation and maturation of neuromuscular contacts in zebrafish. Using histochemical approaches we show that elevating nitric oxide levels reduces the number of neuromuscular synapses within the axial swimming muscles whilst inhibition of nitric oxide biosynthesis has the opposite effect. We further show that nitric oxide signalling does not change synapse density, suggesting that the observed effects are a consequence of previously reported changes in motor axon branch formation. Moreover, we have used in vivo patch clamp electrophysiology to examine the effects of nitric oxide on physiological maturation of zebrafish neuromuscular junctions. We show that developmental exposure to nitric oxide affects the kinetics of spontaneous miniature end plate currents and impacts the neuromuscular drive for locomotion. Taken together, our findings implicate nitrergic signalling in the regulation of zebrafish neuromuscular development and locomotor maturation.

The aim of the current study was to determine how NO signalling influences anatomical and physiological maturation of zebrafish NMJs. Using histochemical approaches we show that developmental manipulation of NO and cGMP signalling affects the formation of NMJs along the axial swimming muscles of developing zebrafish. In addition, using in vivo patch clamp electrophysiology we show that developmental perturbation of NO affects the kinetics of spontaneous miniature end plate currents (mEPCs) at nascent NMJs. Finally, we provide evidence for NO-dependent effects on the maturation of locomotor network activity. Our data provides in vivo evidence that NO/cGMP signalling affects NMJ and locomotor maturation in zebrafish.

Ethics and Zebrafish Care
Zebrafish were maintained according to established procedures [42] and in compliance with the Animals (Scientific Procedures) Act 1986. Embryos were collected and incubated at 28.5uC in embryo medium until the required developmental stage. Staging was performed in accordance with Kimmel et al. [43]. All experiments were conducted on 2 day old zebrafish embryos. At this immature stage, zebrafish are not considered to be sentient and are unable to register pain. Thus ethics approval was not required for this study. On completion of electrophysiology experiments, embryos were anaesthetised with 0.02% MS-222 and killed with a bleach solution (sodium hypochlorite 6.15%). For histochemistry experiments, embryos were anaesthetised with 0.02% MS-222 and subsequently euthanised with 4% paraformaldehyde (PFA, Fisher Scientific).

Chronic Drug Treatments
To exogenously elevate NO and cGMP levels the NO-donor DETA-NO and the cGMP analog 8-pCPT-cGMP were diluted in embryo medium (pH 7.4) to the desired concentration. Embryos at 24 hours post fertilisation (hpf) were placed in the resulting solution. Inhibition of NO or cGMP synthesis was achieved with L-NAME or ODQ, respectively. As these drugs appeared to exhibit low skin penetrance, they were dissolved in Evans physiological saline [composition (in mM): 134 NaCl, 2.9 KCl, 2.1 CaCl 2 , 1.2 MgCl2, and 10 HEPES, pH 7.8] containing fast green (0.2%; Sigma) and around 50 nl of the resulting solution was injected directly into the yolk. To do this, glass microinjection needles (tip diameter: 10-20 mm) were filled with drug solution and inserted into the yolk of 24 hpf embryos. Drugs were pressureejected using a Picospritzer microinjector (Parker, USA). Subsequent to drug treatment, embryos were incubated at 28.5uC until 48-52 hpf.

Labelling of Putative NMJs
Histochemistry was performed as previously described [41]. Briefly, fish were anaesthetised in 0.02% MS-222 (Acros Organics) and subsequently fixed in 4% PFA for 90 min at room temperature. After thorough rinsing in PBSTX (Phosphate buffered saline containing 0.1% Triton-X 100), fish were placed in PBS containing 1 mg/ml collagenase (Sigma) for 16 min. After rinsing in PBSTX, fish were placed in blocking solution (composition: 3% milk powder; 1% DMSO; 0.1% Triton-X 100 in PBS) containing 10 mg/ml rhodamine-conjugated a-bungarotoxin (Rh-a-BTX; Sigma) which labels postsynaptic ACh receptors. After 30 min the block/Rh-a-BTX solution was rinsed off and fish were placed in blocking solution containing anti-SV2 antibodies (1:200; Developmental Studies Hybridoma Bank (DSHB), University of Iowa), which label presynaptic neurotransmitter vesicles. Fish were incubated in primary antibody overnight at 4uC before thorough rinsing in PBSTX and incubation in fresh blocking solution. After 40 min AlexaFluor 488 secondary antibody (1:500; Invitrogen) was added to the block solution and the fish were incubated overnight (4uC). Fish were subsequently rinsed with PBS and cleared in glycerol prior to mounting for image acquisition.

Image Acquisition and Analysis of NMJ Distribution
Labelled NMJs were imaged with an Olympus FV1000 confocal microscope connected to a PC running Fluoview FV1000 capture software. Images were captured using a 40x objective in z-stacks spaced at 3 mm increments [41]. Staining dorsal to the horizontal myoseptum was excluded from analysis. All experiments were conducted on fish ranging between 48 and 52 hpf and within each experimental replicate fish were carefully age-matched to minimise age-dependent variation. Images were obtained from trunk segments located at the level of the yolk-sac extension ( Figure 1A). Acquired z-stacks were compressed into a single image and deconvoluted using Huygens Essential software prior to analysis with the ImageJ plugin SynaptcountJ. The accuracy of this method was confirmed via visual quantification of NMJ distribution. To analyse total NMJ number per somite, regions of interest (ROIs) were selected such that they spanned a single motor axon fascicle and its associated branches. For fascicle or branch analysis, ROIs were selected to encompass either the fascicular or motor axon branch domains within each somitic region. To quantify the density of branch-associated NMJs, the total number of branchassociated puncta was divided by the sum of axon branch lengths.

In vivo Patch Clamp Electrophysiology
To prepare fish for physiological recording, 48-52 hpf embryos were anaesthetised in MS-222 (0.02%) and secured to a Sylgardcontaining Petri dish via insertion of fine (0.025 mm) tungsten pins through the notochord. The trunk skin was carefully removed with a pair of fine forceps to expose the underlying muscle tissue. Fish were then transferred to a patch clamp microscope (Nikon FN-1) and the MS-222 washed off with Evans physiological saline containing 10 mM glucose. To gain access to embryonic fast twitch fibres, a small number of overlying embryonic slow twitch fibres were removed from a two to four somite region via aspiration with a broken-tipped patch electrode. All recordings were made from ventral fibres in somites adjacent to the tip of the yolk extension; approximately the same position used for imaging of NMJs (see above).
For all physiological experiments, whole cell patch clamp electrodes (resistance = 3-10 MV) were pulled from borosilicate glass (Harvard Apparatus, UK) using a P-80 micropipette puller (Sutter, USA). Electrodes were filled with a K-gluconate recording solution containing (in mM): 10 HEPES, 10 EGTA, 2 MgCl?6 H 2 O, 10 NaCl, 6 KCl and 126 D-gluconic acidpotassium salt (pH adjusted to 7.2 with KOH). During analysis of mEPCs, 1 mM TTX was added to the Evans physiological saline to ensure that recorded cells were synaptically isolated. For experiments in which mEPC kinetics were analysed, the gap junction blocker 18bGA (100 mM) was included in the bath solution in order to reduce contamination from electronically generated currents originating in neighbouring muscle fibres [44].
To study effects of acute DETA-NO/L-NAME application, fish were exposed to a 10-20 min pre-treatment with 2 M formamide prior to experimental recording. This treatment is known to cause effective excitation-contraction uncoupling [45], thus minimising the risk of muscle contraction during prolonged recordings. After 10 min in control saline, fish were exposed to a 10 min perfusion with either L-NAME or DETA-NO.
During analysis of locomotor-related end plate potentials (EPPs), a low concentration of the neuromuscular blocker tubocurarine (3 mM; Sigma) was substituted for TTX in the extracellular saline. Under these conditions muscle ACh receptors are partially blocked [46] such that muscular contractions are inhibited but locomotor-related drive to muscle fibres can be readily monitored with whole cell patch clamp electrophysiology. Fictive motor episodes were evoked with light stimulation.
During current clamp recordings muscle fibres were injected with a small amount of current to maintain a membrane potential of 275 mV and 265 mV for embryonic fast and slow twitch fibres respectively whereas during voltage clamp experiments, fibres were clamped at 275 mV. Fibres with access resistances .10 MV were routinely excluded and series resistance compensated by $70%. Biological signals were amplified using an RK-400 amplifier (Biologic) and digitized with a National Instruments A-D converter. Signals were acquired at a sample rate of 30 KHz and subsequently filtered to 2 kHz offline using a PC running pClamp (Molecular Devices).

mEPC and EPP Analysis
All analysis was conducted offline in Clampfit (Molecular Devices). Events were detected using the template matching function and all captured events were subject to manual examination where erroneous events were excluded from analysis. mEPC frequency was determined by counting the number of events occurring over a 200 s recording period. Rise times (10-90%) and half-widths of mEPCs were analysed offline using Clampfit. For analysis of locomotor EPPs, 30 EPPs taken from each swim episode were captured using the template matching feature. These were pooled and averaged and the mean rise time (10-90%), decay time (10-90%) and amplitude recorded.

Statistical Analysis
For analysis of NMJ puncta and mEPC frequency, statistical significance was determined using an unpaired Student's t-test. Pairwise comparisons of mEPC kinetics were conducted using the Dunnett's test and heteroscedasticity within mEPC datasets was controlled using the method described by Herberich et al. [47]. Results are presented as means 6 standard error and statistical significance is reported as follows: *p,0.05; ** p,0.01; *** p,0.001.

Developmental Effects of Elevated NO/cGMP Levels on NMJ Markers
Previously we had shown that NO signalling suppresses zebrafish motor axon branch development [41] and further demonstrated that chronic blockade of NOS (with L-NAME) increases the number of co-localised pre-and post-synaptic domains within the developing zebrafish musculature at 3 days post fertilisation (dpf). However, the effects of NO elevation and sGC/cGMP manipulation on NMJ formation have not been explored. Additionally, as we previously used voxel co-localisation methods to quantify changes in the number of volumetric pixels that contained overlapping synaptic markers, nitrergic effects on synapse number and distribution were not determined. To address these issues we manipulated NO/cGMP levels during development and labelled NMJs with the presynaptic marker anti-SV2 and the postsynaptic ACh receptor marker rhodamine conjugated a-bungarotoxin (Rh-a2BTX). Subsequently we examined the number and spatial distribution of NMJs with SynaptcountJ, a plugin for ImageJ that permits quantification of synapse number (see methods). Analysis was restricted to the second day of development ( Figure 1A) because by this stage a small, easily quantifiable network of arbours has been established in each somitic region ( [48]; Figure 1B). Since our previous work suggested that NO/cGMP signalling affects motor axon branch formation Mean density of branch-associated puncta in control (black) and DETA-NO treated (grey) fish. E. Left hand panels: lateral trunk views of anti-SV2 (green)/Rh-a-BTX (red) co-staining in control (top) and 8-pCPT-cGMP treated (cGMP, bottom) zebrafish at 2 dpf. Right hand panels: expanded regions showing staining localised to a single somitic region. F. Bar chart depicting the mean (6 SEM) number of synapses located within each somite (som), along motor fascicles (fas) and along branches of control (black) and 8-pCPT-cGMP (cGMP, grey) treated fish. G. Mean density of branch-associated puncta in control (black) and 8-pCPT-cGMP treated (cGMP, grey) fish. Scale bars = 30 mm. ***p#0.001. doi:10.1371/journal.pone.0086930.g001 without impairing motor root growth, we segregated puncta into two domains: those located on the motor axon fascicle and those located on motor axon branches ( Figure 1A). Using this approach, differential effects of NO/cGMP signalling on each compartment could be examined.
To characterise the effects of NO/cGMP signalling on neuromuscular synapse formation, 1 dpf embryos were exposed to embryo medium containing the NO donor DETA-NO or 8-pCPT-cGMP, a cell permeable cGMP analogue. Fish were incubated in drugs until 2 dpf, at which point they were processed for SV2/Rh-a-BTX histochemistry and compared to control fish raised in egg media alone. In agreement with previous reports [48], regions of SV2/Rh-a2BTX co-localisation (putative NMJs) were found scattered throughout the muscle of control embryos ( Figure 1B). When developing embryos were exposed to DETA-NO, a marked decrease in the total number of somitic NMJs was observed ( Figure 1B [41] or from a change in the number of NMJs formed per unit of branch length. To distinguish between these possibilities we calculated the density of branch-specific NMJ puncta, finding that DETA-NO had no effect on this parameter ( Figure  1D; control = 0.1760.01 puncta/mm, DETA-NO = 0.1660.02 puncta/mm, p.0.05). This suggests that the NO-dependent reduction in NMJ number is a consequence of the previously reported reduction in motor axon branch formation [41].
Incubating embryos in 8-pCPT-cGMP had a similar effect to DETA-NO ( Figure 1E

Developmental Effects of Inhibiting NOS/sGC on NMJ Markers
To determine the effects of reducing NO/cGMP levels during development we perturbed either NOS or sGC activity by exposing zebrafish embryos at 24 hpf to L-NAME or ODQ. Fish remained in drugs until 2 dpf, at which point they were processed for immunohistochemistry and compared to age-matched controls.

Developmental Effects of NO Signalling on mEPP Parameters
We next used electrophysiological approaches to determine whether developmental perturbations in NO signalling affected neuromuscular physiology. Two discreet muscle fibre populations, termed 'embryonic fast' (EF) and 'embryonic slow' (ES) are known to exist in early stage zebrafish [44,49,50]. These differ with respect to speed of contraction, position, orientation, extent of electrical coupling and synaptic kinetics: the ES fibres are superficial, run parallel to the body axis, are extensively electrically coupled to multiple neighbouring ES fibres and generate synaptic currents with relatively slow rise and decay kinetics [44]. By contrast, the EF fibres lie medial to the ES population, are arranged in an oblique orientation, exhibit less extensive electrical coupling and generate synaptic currents with relatively fast kinetics [44]. As a first step towards examining the effects of NO signalling on physiological maturation of zebrafish NMJs we examined the effects of NO on mEPCs occurring in both fibre populations. To this end we synaptically isolated muscle fibres of 2 dpf control, DETA-NO treated and L-NAME treated fish with the sodium channel blocker tetrodotoxin (TTX) and recorded mEPCs in the whole cell voltage clamp configuration [44,49,50].
During mEPC recordings, the gap junction blocker 18bGA was added to the extracellular saline in an attempt to block gap junction coupling between muscle fibres [44]. Electrical coupling has several effects on muscle physiology. First, it reduces fibre input resistance; second it decreases mEPC amplitude [44,51] and; third, it introduces a sub-population of small amplitude, slow currents which are believed to be gap junction filtered representations of mEPCs occurring in neighbouring fibres [44,50]. As expected, we found that 18bGA abolished transcellular dye labelling ( Figure 3A,B) ). However, we found that a proportion of small amplitude, slow currents persisted in the presence of this drug ( Figure 3C,D). These were apparent in plots of mEPC rise time against amplitude where events segregated into two distinct populations, one with larger amplitudes and fast rise times and one with smaller amplitudes and slow rise times ( Figure 3E,F). To determine whether the second population represented residual transjunctional events we recorded from pairs of adjacent muscle fibres (n = 1 EF pair, n = 3 ES pairs). We found that even after relatively prolonged (20-30 min) 18bGA exposure, coincident inward currents with small-amplitudes (,100 pA) and slow rise times (.0.6 ms) were observed in recorded muscle fibres ( Figure 3G). Based on these observations, we considered these events to be mediated by passage of current through 18bGA insensitive gap junctions and thus excluded them from analysis.
We next compared mEPCs occurring in EF fibres of control, DETA-NO and L-NAME treated embryos ( Figure 4A-G). Cumulative probability distributions indicated that developmental exposure to either DETA-NO or L-NAME reduced mEPC amplitude in EF fibres ( Figure 4B) and analysis of mean amplitudes confirmed that this was the case (Figure 4C Finally, we also examined mEPC frequency in EF and ES fibres. We found that in the presence of 18bGA, neither DETA-NO or L-NAME significantly affected the frequency of EF (control = 0.1260.02 Hz, n = 47; DETA-NO = 0.1060.03 Hz, n = 17; L-NAME = 0.1160.02 Hz, n = 46, p.0.05, data not shown) or ES (control = 0.0860.01 Hz, n = 24; DETA-NO = 0.0660.01 Hz, n = 33; L-NAME = 0.0860.01 Hz, n = 38; p.0.05, data not shown) mEPCs. However, during current clamp recordings in which 18bGA was excluded from the bathing media, Figure 2. Developmental inhibition of NO/cGMP synthesis increases NMJ numbers. A. Left hand panels: lateral trunk views of control (top) and L-NAME (bottom) treated zebrafish at 2 days post fertilisation (dpf) processed with anti-SV2 (green) and Rh-a-BTX (red) staining. Right hand panels: expanded regions showing staining localised to a single somitic region. B. Bar chart depicting the mean (6 SEM) number of synapses located within each somite (som), along motor fascicles (fas) and along branch-associated domains (branch) of control (black) and L-NAME (grey) treated fish. C. Mean density of branch-associated puncta in control (black) and L-NAME treated (grey) fish. D. Left hand panels: lateral trunk views of control (top) and ODQ (bottom) treated zebrafish at 2 dpf. Right hand panels: expanded regions showing staining localised to a single somitic region. E. Bar chart depicting the mean (6 SEM) number of synapses located within each somite (som), along motor fascicles (fas) and along branchassociated domains (branch) of control (black) and ODQ (grey) treated fish. F. Mean density of branch-associated puncta in control (black) and ODQ treated (grey) fish. Scale bars = 30 mm. **p#0.01, ***p#0.001. doi:10.1371/journal.pone.0086930.g002 Figure 3. Effects of 18bGA on electrical coupling in EF and ES fibres. A, B. Images of embryonic fast (EF, A) and embryonic slow (ES, B) muscle fibres that were dialysed with sulforhodamine during whole cell recording. Note that in control saline, dye spreads to neighbouring muscle fibres whilst 18bGA pre-treatment abolishes this effect. C,D. Representative traces of miniature end plate currents (mEPCs) occurring in EF (C) and ES (D) fibres exposed to 18bGA. Right: example events captured on an expanded time scale. E,F. Amplitude versus rise time plots for mEPCs recorded from EF (E) and ES (F) fibres exposed to 18bGA. Insets show histograms of mEPC rise time in control (black) and 18bGA-treated (grey) fibres. G. Paired recording between ES fibres reveal that coincident events, presumed to be carried by electrical synapses, are occasionally observed in the presence of 18bGA; asterisked events with extended time scale on right. Scale bars = 50 mM. doi:10.1371/journal.pone.0086930.g003  whereas L-NAME increased (control EF = 0.9660.07 Hz, L-NAME treated EF = 1.2460.10 Hz; control ES = 1.2660.09 Hz, L-NAME treated ES = 1.7160.14 Hz, p,0.01, data not shown) the number of synaptic events.

Acute Manipulation of NO Levels does not Affect NMJ Physiology
To test for acute effects of NO signalling on NMJ physiology we obtained patch recordings from muscle fibres of 2 dpf control fish and subsequently exposed them to a brief (10 min) incubation in either DETA-NO (n = 3 EF, n = 3 ES) or L-NAME (n = 4 EF, n = 3 ES). During recordings ( Figure 6A-D), neither bath perfusion of DETA-NO or L-NAME affected the frequency, amplitude, rise time or half-width of EF and ES mEPCs ( Figure 6E-H). Thus, transient manipulation of NO signalling does not affect physiological parameters of the developing zebrafish NMJ.

Developmental Effects of NO on the Locomotor Drive for Swimming
We next asked how developmental perturbation of NO affected the neuromuscular drive for swimming behaviour. To do this we examined locomotor-related end plate potentials (EPPs) in EF ( Figure 7) and ES (Figure 8) fibres that were not exposed to 18bGA. Using this approach we were thus able to study the NOdependent regulation of locomotor ontogeny under more physiologically-relevant conditions. The synaptic drive for locomotion was monitored during sensory-evoked bouts of fictive swimming in 2 dpf fish raised in control saline (n = 10 EF; n = 10 ES), DETA-NO (n = 4 EF; n = 4 ES) or L-NAME (n = 7 EF; n = 9 ES). During EF recordings, developmental DETA-NO exposure ( Figure 7A Figure 7D; control = 0.9460.03 mV; L-NAME = 1.1760.04 mV, p,0.001) and induced a small, yet significant decrease in EPP rise times ( Figure 7E; control = 3.4060.08 ms; L-NAME = 2.9660.08 ms, p,0.05) but did not affect decay times ( Figure 7F). During ES recordings ( Figure 8A-C), neither pharmacological reagent affected the amplitude of locomotor-related EPPs ( Figure 8D). However, marked effects on EPP rise and decay times were observed. Here, DETA-NO dramatically increased EPP rise ( Figure  8E; control = 7.0260.13 ms; DETA-NO = 11.2760.32 ms, p,0.001) and decay ( Figure 8F; control = 20.6660.29 ms; DETA-NO = 31.29605.2 ms, p,0.001) durations whereas L-NAME raised fish exhibited a significant decrease in rise ( Figure 8E; control = 7.0260.13 ms; L-NAME = 4.8060.11 ms, p,0.001) and decay ( Figure 8F; control = 20.6660.29 ms; L-NAME = 17.3060.27 ms, p,0.001) times.
To determine how NO affected the frequency and duration of motor episodes, data from EF and ES recordings were pooled. Thereafter locomotor frequency was estimated by measuring EPP frequency during the beginning, middle and end of each episode. When compared to controls, EPP frequency of DETA-NO exposed fish was consistently lower across all periods of the locomotor episode ( Figure 9A; p,0.001). By contrast, mean EPP frequency was higher throughout episodes of L-NAME treated fish ( Figure 9A; p,0.001). In addition, developmental exposure to both DETA-NO and L-NAME caused a significant reduction in the duration of locomotor episodes ( Figure 9B; control = 15.661.98 s; DETA-NO = 4.5360.60 s; L-NAME = 6.3460.90 s, p,0.001). Taken together, these data strongly suggest that developmental manipulation of NO signalling impacts locomotor performance.

Discussion
The main findings of this study are that developmental manipulation of NO signalling reduces the number of neuromuscular synapses and affects the physiological properties of NMJs within the developing muscle population of zebrafish embryos. When taken together, our observations provide in vivo evidence that endogenous NO signalling influences the formation and functional maturation of zebrafish NMJs.

NO/cGMP Signalling as a Regulator of Neuromuscular Synaptogenesis
Our previous study revealed that NO signalling suppresses and NOS inhibition enhances motor axon branch development in zebrafish. Moreover, voxel analysis revealed that the number of co-localised pre-and post-synaptic NMJ markers is increased by chronic L-NAME exposure. In the current study we have built upon these findings so that the effects of NO and cGMP on NMJ number, density and distribution could be determined. This allowed us to make three important observations. First, we show that developmental elevation of NO/cGMP (via addition of DETA-NO/8-pCPT-cGMP) inhibits, whilst developmental inhibition of NO/cGMP synthesis (via exposure to L-NAME/ODQ) promotes, formation of NMJs within the axial swimming muscles. Second, we show that these effects arise from changes in the number of puncta located on motor axon branches, but not fascicles. Finally, we show that the density of branch associatedsynapses is not affected by NO signalling. Since our previous data demonstrates that NO decreases motor axon branch number [41], we posit that the NO-dependent modifications in NMJ number reported here arise as a consequence of modified branch formation. However, once branches are formed, NO does not appear to alter the rate of synapse addition.
During the current study we found that NO signalling did not affect the number of fascicular NMJs. This is perhaps not surprising as drug treatment was initiated at 24 hpf. By this stage, early periods of motor fascicle extension and the incorporation of pre-patterned ACh receptors into fascicular NMJs is complete [48,52,53]. Whilst we cannot exclude the possibility that pharmacological treatment at earlier time points may influence fascicular synaptogenesis, the observation that NOS1 is first expressed within the spinal cord at around 30 hpf [41] strongly suggests that NO has no physiological role during initial stages of neuromuscular development which begins at , 17 hpf [54]. Thus, unlike amphibian and avian species, where NO affects aneural ACh receptor cluster assembly [2][3][4][5], the earliest stages of zebrafish NMJ development are unlikely to be NO-dependent.

NO Signalling and NMJ Physiology
During mEPC experiments we routinely included 18bGA in the bath solution to block electrical synapses within the zebrafish musculature. This drug has previously been reported to abolish the low-amplitude, filtered events that arise from electrotonic spread of current from electrically coupled neighbouring fibres [44]. Here, we found that 18bGA application abolished transcellular spread of sulforhodamine and increased both muscle fibre input resistance and mEPC amplitude. Although these observations suggest that 18bGA abolished gap junction signalling, we found a population of small, filtered mEPCs persisted in the presence of this drug. Interestingly, Sylvain et al. [55] reported a similar phenomenon following exposure to the 18bGA analog carbenox-olone, reasoning that they were generated by nascent synapses located on the recorded muscle fibre. However, as we found that similar events often occurred synchronously during paired muscle recordings, we favour the hypothesis that they are caused by current flow through 18bGA-resistant gap junctions. Thus, for the purposes of our study, this population of events was excluded from analysis. Our data suggests that chronic manipulation of NO signalling had marked effects on synaptically-mediated mEPCs. Specifically, developmental elevation of NO (DETA-NO exposure) caused a marked slowing of mEPC rise times and half-widths in EF and ES muscle. By contrast, mEPC half-widths of L-NAME treated fish were reduced in EF and ES fibres, although rise times were not significantly affected. In addition, L-NAME reduced the ampli-tude of EF and ES mEPCs whilst DETA-NO reduced EF mEPC amplitudes. Collectively, these observations suggest that NO may regulate the maturation of immature neuromuscular synapses. In support of this hypothesis, previous studies have shown that the kinetics of zebrafish mEPPs and mEPCs change markedly during neuromuscular ontogeny. During early embryonic development, these exhibit prolonged rise and decay times and relatively small amplitudes. As maturation proceeds, rise and decay times shorten whilst amplitude increases [46,50,56,57]. The speeding-up of synaptic currents has been attributed to a combination of changes that include maturation of presynaptic properties, the gradual aggregation and addition of ACh receptors at the endplate and the expression of ACh-esterase, which serves to hydrolyse ACh within the synaptic cleft and thus accelerates relaxation of endplate currents [57][58][59]. In addition, vertebrate neuromuscular ACh receptors often undergo developmentally-related changes in subunit composition which engenders differing kinetic properties during development. For example, immature ACh receptors are typically composed of a2bdc subunits whilst adult receptors contain a2bde subunits [60][61][62][63]. Receptors expressing the c subunit exhibit prolonged open channel times and generate prolonged macroscopic currents whereas those expressing the e subunit exhibit short open channel times and generate shorter macroscopic currents [61]. Whilst a delay in the switch from c to e subunits may contribute to the effects of NO on EF fibres, it is unlikely to explain the effects observed within ES fibres as they do not exhibit a developmentally-related change in subunit composition [59]. Future molecular and biochemical studies are needed to identify whether one or more of these developmental phenomena are influenced by NO signalling.
Interestingly, we found that the NO-dependent effects on mEPC frequency varied as a function of recording conditions. In the absence of 18bGA, event frequency decreased in DETA-NO treated fish but increased in L-NAME treated fish. When considered in light of our anatomical findings, we hypothesise that these effects are caused by NO-dependent regulation of NMJ numbers. By contrast, differences in mEPC frequency were not observed when 18bGA was included in the extracellular saline. These effects could arise because differences in mEPC frequency may only be apparent when inputs to multiple muscle fibres are simultaneously monitored. Addition of 18bGA would reduce electrical coupling between fibres, thus reducing the number of fibres from which events could be sampled.

Developmental Effects of NO on Locomotor Drive
Although transient NO biosynthesis is known to modulate parameters of ongoing locomotor activity [64,65], its relevance to locomotor maturation has remained unclear. Here we show that chronic manipulation of NO signalling can strongly affect locomotor network output. Developmental exposure to DETA-NO prolongs the rise and decay durations of locomotor-related EPPs. Whilst the effects of NOS inhibition (L-NAME incubation) were less pronounced, we did observe a small acceleration in EF/ ES EPP rise times and a reduction in ES EPP decay times during locomotor activity. Given that mEPC rise and decay times are prolonged in DETA-NO and accelerated in L-NAME, the observed changes to the neuromuscular drive may, at least in part, arise from a change in the waveform of macroscopic currents generated by ACh receptor activation. Specifically, DETA-NO dependent slowing of the rise and decay time of ACh receptor currents would be expected to slow rates of membrane depolarisations and repolarisations during swimming. By contrast, L-NAME, which accelerated the rise and decay times of mEPCs would be expected to have the opposite effect. However, other factors may also contribute to this phenomenon. For example, the extensive electrical coupling between zebrafish embryo muscle fibres is known to filter postsynaptic responses within the muscle [44,50,66]. Thus, a change in the extent of electrical coupling could affect the rise and decay times of EPPs recorded during swimming. Additionally, the amplitude, rise and decay time of compound EPPs is known to be influenced by the time course of transmitter release from individual presynaptic terminals [67,68]. A number of factors, including synaptic ACh [69], catecholamines [70] and presynaptic action potential duration [68,71] have been shown to influence release latency at the NMJ and as a consequence affect the rise, decay and amplitude of postsynaptic events. On this note, NO has been reported to directly affect ACh release at the NMJ [72], catecholamine release within the spinal cord [65] and neuronal excitability of interneuron populations [73] and so could potentially alter release latency via a number of different mechanisms. Future motoneuron-muscle recordings will help to determine whether NO does indeed have an effect on the timecourse of evoked ACh release at the developing NMJ.
We also found that when compared to controls, the frequency of locomotor-related EPPs was lower in DETA-NO treated and higher in L-NAME treated fish, an observation that stands in broad agreement with our previous behavioural studies [41]. In Figure 9. Developmental effects of NO on the frequency and duration on the neuromuscular drive for locomotion. A. Line graph of mean (6 SEM) locomotor-related end plate potential (EPP) frequency at the beginning, middle and end of evoked fictive swim episodes in control fish and fish exposed to DETA-NO or L-NAME during development. B. Bar chart showing mean duration of fictive motor episodes in control, DETA-NO and L-NAME treated fish. *** p,0.001. doi:10.1371/journal.pone.0086930.g009 addition, we observed that both DETA-NO and L-NAME exposure reduced the duration of locomotor-related episodes. Although the mechanism of action awaits investigation, this effect is likely to be caused by an NO-dependent effect upon the properties of spinal cord interneurons or descending motor control pathways, rather than NO-dependent regulation of NMJ properties: at the onset of swimming (approximately 26 hpf), the locomotor network contains a small number of immature neurons that can only generate low-frequency, inflexible forms of swimming behaviour [46,74]. As electrical properties of spinal neurons mature and new cells are integrated into the spinal network, fish become capable of swimming at a wide range of frequencies [75][76][77][78][79][80]. Thus, an NO-dependent effect on the development of later-developing spinal interneurons may underlie the observed actions of NO. Alternatively, observed changes in motor output might arise from homeostatic feedback mechanisms that scale the frequency of rhythm generator activity to the biophysical characteristics of the muscle. Thus, the NO-dependent slowing of postsynaptic NMJ kinetics may trigger feedback mechanisms that ensure network activity is limited to frequencies that do not cause EPP summation and tetanus. In contrast, the speeding up of postsynaptic responses in conditions of low NO may allow the locomotor network to drive muscles at slightly higher frequencies. Irrespective of the causes, our observations stand in agreement with previous behavioural studies showing a reduction in swim frequency in fish exposed to NO donors during development and an acceleration in locomotor output in fish lacking the NOS1 isozyme [41].
In summary, our data provide strong evidence that NO serves as a developmental regulator of NMJ maturation in zebrafish. These observations provide further support for the hypothesis that NO is a fundamentally important signalling molecule during periods of NMJ synaptogenesis and locomotor maturation.