Early suppression of the endocannabinoid degrading enzymes FAAH and MAGL alters locomotor development in zebrafish

The endocannabinoids (eCBs) anandamide (AEA) and 2-arachidonoylglycerol (2-AG) have been shown to regulate the perception of pain, neuroprotection, motor movement, memory and cognition (Van Der Stelt et al., 2001; Cajanus et al., 2016; Blankman and Cravatt, 2013; Di Marzo, 2008). The eCB system is composed of the eCBs AEA and 2-AG, the enzymes that produce them, the enzymes that break them down [fatty acid amide hydrolase (FAAH) and monoacylglycerol lipase (MAGL)] and the receptors that bind them (Van Der Stelt et al., 2001). AEA and 2-AG are synthesized by N-acyl phosphatidyl ethanolamine phospholipase D (NAPE-PLD) and diacylglycerol lipase (DAGL), respectively. Following a rise in synaptic activity, AEA and 2-AG are rapidly metabolized by the serine hydrolase enzymes FAAH and MAGL, respectively (Van Der Stelt et al., 2001; Freund et al., 2003; Day et al., 2001). Although eCBs can interact with receptors, such as CB₁R, CB₂R, GPR55, TRP channels and serotonin receptors (5-HTRs), the two most abundant cannabinoid receptors and the ones most studied in terms of eCB binding are the CB₁Rs and CB₂Rs (Pertwee, 2006). CB₁Rs and CB₂Rs are G-protein coupled receptors (GPCRs) that, when activated, initiate the downregulation of adenylyl cyclase activity and the depletion of intracellular cAMP levels (Starowicz et al., 2013). eCBs are rapidly synthesized following neuronal demand (Piomelli, 2005) and although their synthesis typically occurs in the postsynaptic compartment, they readily cross cell membranes and act on presynaptic receptors to inhibit presynaptic transmitter release (Chevaleyre et al., 2006).

Even though eCBs bind to both CB₁R and CB₂R, most of the neurobehavioral effects are due to their activation through CB₁Rs (Zimmer et al., 1999). Moreover, the beneficial effects of phytocannabinoid treatment as analgesic molecules or for the treatment of seizures may primarily occur through the binding and activation of CB₁Rs (Maroon and Bost, 2018). However, this may also lead to undesirable side effects, such as locomotor and cognitive impairments (Freund et al., 2003; Bridges et al., 2003). To avoid unwanted psychotropic side effects, indirect manipulation of the eCB pathway is becoming more common to treat pain and other central nervous system (CNS)-related issues (El-Alfy et al., 2010). One such widely used strategy focuses on inhibition of the FAAH and MAGL enzymes (Long et al., 2009), but little is known about the effects of inhibiting these enzymes during early development.

Pharmacological inhibition of FAAH and MAGL enzymes can rapidly increase the bioavailability of the eCBs AEA and 2-AG, respectively (Long et al., 2009; Cravatt et al., 2001; Kinsey et al., 2009). Even though acute FAAH and MAGL inhibition may be beneficial for pain attenuation (Kinsey et al., 2010), a prolonged suppression of the enzymes may negatively affect signaling processes that are required for normal brain development (Anavi-Goffer and Mulder, 2009; Harkany et al., 2008). For instance, knockdown of the 2-AG synthesizing enzyme, Daglα, via morpholino technology can lead to altered axonal growth, and the development of abnormal motor movement (Martella et al., 2016a,b). Also, the upregulation of AEA and 2-AG by bisphenol A, a chemical used in the production of plastics and resins, increases hepatosteatosis in zebrafish (Martella et al., 2016b). The eCB system has been shown to play roles in neuronal proliferation (Harkany et al., 2008), axonal growth (Mulder et al., 2008), synaptic formation (Mulder et al., 2008) and movement (Martella et al., 2016a,b) during early development. Even though some studies have attempted to elucidate the roles and functions of eCBs and their receptors, the full spectrum of actions of the FAAH and MAGL enzymes during embryogenesis is not fully understood.

In this study, we set out to determine whether an imbalance or a perturbation of the eCB degradation mechanism would alter embryonic development because key eCB signalling components, including FAAH and MAGL, are present from as early as egg fertilization in zebrafish embryos (Oltrabella et al., 2017). We used the specific FAAH and MAGL inhibitors JZL195, URB597 and JZL184 for the dual or selective inhibition of the enzymes in the first 24 h following egg fertilization, and then examined the developmental effects on motor neurons, the neuromuscular junction (NMJ) and the locomotor system. To address specificity, we determined whether the effects of inhibiting these enzymes could be abrogated by blocking CB₁Rs and CB₂Rs. Our results start to discern the functional roles played by both FAAH and MAGL enzymes in determining motor neuron and locomotor development in zebrafish embryos through the eCB pathway.

We used the Tubingen Longfin (TL) strain of wild-type zebrafish [Brachydanio rerio (Hamilton 1822)] in this study. Adult fish for breeding were maintained in an aquatic facility at the University of Alberta, and all experimental procedures were approved by the Animal Care and Use Committee (AUP 00000816) and adhered to the Canadian Council on Animal Care guidelines for animal use. For breeding, five adult zebrafish consisting of three females and two males were randomly selected and placed in a breeding tank the evening before eggs were required. The following morning, eggs were collected from the breeding tanks, usually within 30 min of release by the females. A 12 h:12 h light:dark cycle and 28.5°C temperature was set for housing the embryos and larvae in incubators to ensure a consistent growth and development.

JZL195, URB597, JZL184, AM630 (Adooq Bioscience, Irvine, CA, USA) and AM251 (Selleck Chemicals, Houston, TX, USA) were dissolved in dimethylsulfoxide, which never exceeded a final concentration of 0.1%. Embryos were exposed to chemicals diluted in egg water (60 μg ml⁻¹ Instant Ocean) in the first 24 h post-fertilization (0–24 hpf). The selected doses for each drug treatment in this study was chosen based upon testing a range of concentrations that are similar to those used in previously published studies (Fish et al., 2019; Martella et al., 2016a,b; Song et al., 2015; Tran et al., 2016; Akhtar et al., 2016; Sufian et al., 2019; Fraher et al., 2015). Fresh solutions were prepared on the day of the experiment immediately before use. The drug stocks were diluted in egg water (60 μg ml⁻¹ Instant Ocean). After drug exposure, eggs were washed several times with egg water to remove the drugs and were then kept in normal egg water. The egg water was changed every morning and evening until further experimentation at 2 or 5 days post-fertilization (dpf). For immunohistochemical studies, pigmentation was blocked by a single time application of phenylthiourea at an optimal concentration of 0.003% in egg water at 24 hpf.

Photographs of embryos and larvae were taken using a Lumenera Infinity2-1R color camera mounted on a Leica DM2500 stereomicroscope under 2.5× (embryo full-length images) magnification. Embryos were placed in a 16-well plate with a single embryo per well and were anaesthetized using 0.02% MS-222. A dissecting microscope was used to quantify the rates of embryo malformation.

An in vitro 96-well plate-based assay was carried out to assess FAAH and MAGL enzyme activity. The enzyme inhibitor screening assay kit was purchased from Cayman Chemical (Ann Arbor, MI, USA) and the assay was performed according to the manufacturer's instructions to measure enzyme activity on total protein extracts collected from treated embryos at 24 hpf. Briefly, the FAAH enzyme assay measures the fluorescence from 7-amino-4-methylcoumarin (AMC), upon FAAH enzyme cleavage of the non-fluorescent substrate AMC arachidonoyl amide (AMC-AA) (Wilson et al., 2003). Fluorescence was detected and quantified at an excitation wavelength of 360 nm and an emission wavelength of 460 nm. For the MAGL enzyme assay, the substrate 4-nitrophenylacetate was hydrolyzed by MAGL, yielding a yellow product, 4-nitrophenol, and the reading was recorded at an absorption wavelength of 410 nm. Each well contained a reaction mixture of 30 μg embryo protein extract, the enzyme assay buffer (either FAAH assay buffer: 125 mmol l⁻¹ Tris-HCl with 1 mmol l⁻¹ EDTA, pH 9.0, or MAGL assay buffer: 10 mmol l⁻¹ Tris-HCl with 1 mmol l⁻¹ EDTA, pH 7.2), and was incubated for 5 min at 37°C. Final reactions were then initiated by adding 10 μl of either FAAH or MAGL enzyme substrate (AMC-AA: 400 µmol l⁻¹, or 4-nitrophenylacetate: 4.25 mmol l⁻¹) and incubated for another 30 min at 37°C before plate reading. Controls were run in every experiment. Control samples contained human recombinant FAAH or MAGL enzyme or protein extracts collected from untreated embryos. Also, blank wells were run in every experiment to monitor any background fluorescence from each reading. All of the sample reactions were run in triplicate.

Treated embryos were manually dechorinated at 2 dpf and then fixed with 2% paraformaldehyde for approximately 2 h. After fixation, samples were washed every 15 min for 2 h in 0.1 mol l⁻¹ phosphate buffered saline (PBS; 150 mmol l⁻¹ NaCl, 8 mmol l⁻¹ Na₂HPO₄, 2 mmol l⁻¹ NaH₂PO₄·2H₂O, pH 7.2). Embryos were then permeabilized using 4% Triton X-100 containing 2% bovine serum albumin and 10% normal goat serum for 30 min. Sample preparations were then incubated at 4°C with their corresponding primary antibodies for 48 h.

Primary motor neurons and secondary motor neurons were visualized using the mouse primary antibodies anti-znp1 and zn-8, respectively (Developmental Studies Hybridoma Bank). Anti-znp1 (1:250, University of Iowa, deposited by B. Trevarrow) targets the synaptotagmin-2 isoform, a protein highly localized to primary motor axons (Fox and Sanes, 2007; Trevarrow et al., 1990), whereas anti-zn8 (1:250, University of Iowa, deposited by B. Trevarrow) targets DM-GRASP, a protein highly present on the surface of secondary motor axons (Fashena and Westerfield, 1999; Sylvain et al., 2010). After incubation for 48 h in the primary antibody, embryos were washed in PBS every 15 min for 3 h. After washes, samples were incubated in Alexa Fluor ^® 488 goat anti-mouse IgG (1:1000, Molecular Probes, Life Technologies) secondary antibody for 4 h at room temperature. For nAChRs labelling, embryos were incubated with Alexa-488 conjugated α-bungarotoxin (α-BTX, 100 nmol l⁻¹, Molecular Probes, Invitrogen) for 4 h at room temperature.

Following antibody or α-BTX treatment, embryos were washed with PBS every 30 min for 7 h and mounted in MOWIOL mounting medium. A confocal Zeiss LSM microscope was used to obtain immunofluorescent images of embryos using a 40× objective. Multiple image stacks were photographed every 1 μm z-section throughout the entire thickness of the embryo. Primary motor neuron branches, secondary motor axon branches and the number and diameter of α-BTX puncta were compiled and tracked using ImageJ software (National Institutes of Health, Bethesda, MD, USA). Motor neuron branch patterns were analysed using the ImageJ plugin software, neurite tracer. Images were loaded into neurite tracer and every stack was individually reviewed to ensure that the branches were appropriately documented. The neurite tracer software automatically detects each branch, but occasionally the tracing path was manually guided when the software made an obvious error.

For anti-znp1 labelling, primary, secondary and tertiary branches emanating from a single primary motor axon per sample were counted. For anti-zn8 labelling, the percentage of normal (non-misshapen or non-truncated) axonal branches emanating from three secondary motor axons per sample were counted. For α-BTX labelling, the total number of puncta and the diameter of the puncta (per 2500 μm² area) were counted and averaged from three different boxed regions that were evenly placed along the trunk (dorsal, mid and ventral regions).

To track locomotor activity, individual 5 dpf larvae were placed in a single well of a 96-well plate and then video-recorded. The data were analyzed following previously published procedures (Baraban et al., 2005; Leighton et al., 2018). Larvae were gently positioned in the centre of wells containing 150 µl egg water, pH 7.0, and the central 48 wells were used from a 96-well plate (Costar 3599). Prior to video recording, larvae were acclimated in the well plate for 60 min. Plates were placed on top of an infrared backlight source, and a Basler GenlCaM (Basler acA 1300-60) scanning camera with a 75 mm f2.8 C-mount lens (Noldus, Wageningen, The Netherlands) was used for individual larval movement tracking. EthoVision^® XT-11.5 software (Noldus) was used to quantify activity (% of time active over a 1 h time period), velocity (mm s⁻¹, average velocity from active time period), swim bout frequency and cumulative duration of swim bouts for 1 h. To exclude background noise, ≥0.2 mm was defined as an active movement. The percentage of pixel change within a corresponding well between samples were defined as activity (motion was captured by taking 25 frames s⁻¹) as reported previously (Leighton et al., 2018).

GraphPad Prism software (Version 7, San Diego, CA, USA) was used to perform all the data analysis and the values are reported as means±s.e.m. here. For statistical analysis, one-way ANOVA followed by Dunnett's or Tukey's post hoc multiple comparison test was run to determine the significance of the results (at P<0.05). Dunnett's multiple comparisons test was used to identify the significance between every treatment group with the control. Tukey's post hoc test was chosen for comparing every treatment group with every other one. For locomotion experiments, outliers owing to off-tracking (undetected larval tracing) were identified using the ROUT method at Q=0.1 (Q=maximum desired false recovery rate) and were removed objectively.

In this study, we investigated the effects of enhanced eCB signalling during early development in zebrafish embryos after an indirect manipulation of the eCB pathway. We exposed zebrafish embryos to egg water that contained inhibitors of the eCB degrading enzymes FAAH and MAGL, from 0 to 24 hpf, as shown in Fig. 1A. The chemical compounds used in our study were: JZL195 (dual inhibitor of FAAH and MAGL) (Long et al., 2009), URB597 (inhibitor of FAAH) (Piomelli, 2005) and JZL184 (inhibitor of MAGL) (Seillier et al., 2014). The concentrations of the drugs that we used were based on a detailed analysis of full dose–response studies on zebrafish hatching and malformation rates in Fig. 1C–H. For the remainder of the study, we chose to use the following concentrations: 2 µmol l⁻¹ JZL195, 5 µmol l⁻¹ URB597 and 5 µmol l⁻¹ JZL184. These concentrations of JZL195, URB597 and JZL184 are also within a similar range reported in other studies (Fish et al., 2019; Song et al., 2015; Martella et al., 2016a,b).

Previous studies have demonstrated the actions and efficacy of JZL195, URB597 and JZL184 toward the FAAH and MAGL enzymes with a subsequent increase in AEA and 2-AG bioavailability in different animal models (Long et al., 2009; Piomelli, 2005). To confirm that JZL195, URB597 and JZL184 were indeed blocking the activity of FAAH and MAGL in zebrafish embryos, we performed an in vitro enzyme inhibitory screening assay on protein extracts obtained from 24 hpf zebrafish embryos (n=80–120 embryos, N=4 batches) (Fig. 2).

Newly fertilized eggs at the 1–4 cell stage were treated with either vehicle control solutions, JZL195, URB597 or JZL184 for 24 h and the total protein extracts were collected at the 24 h time point. We found that embryos treated with JZL195 and URB597 had a significant reduction in FAAH activity (Fig. 2A). For example, embryos treated with JZL195 and URB597 exhibited FAAH activity levels of 50±7% (N=4, P<0.01) and 55±7% (N=4, P<0.01), respectively, compared with 93±8% activity of the vehicle controls. Exposure to JZL184 did not alter FAAH activity, which remained at a value of 84±8% at 24 hpf (N=4, P=0.71; Fig. 2A).

With respect to the MAGL enzyme assay (Fig. 2B), JZL195 and JZL184 induced a significant reduction in MAGL activity, which decreased to 51±5% (N=4, P<0.01) and 55±7% (N=4, P<0.01), respectively. However, URB597 did not result in a significant change in MAGL activity when compared with vehicle controls (N=4, P=0.21). These data confirmed the specificity of the inhibitors on FAAH and MAGL activity.

To study how perturbation of the eCB system alters neuronal development, we examined how a reduction in FAAH and MAGL enzyme activity would regulate axonal branching patterns of developing primary and secondary motor neurons (Fig. 3).

Anti-znp1 immunolabelling of primary motor neurons in zebrafish embryos (Fig. 3B–E) revealed that exposure to JZL195 significantly reduced the number of primary branches per hemisegment from 14.3±1.2 (n=6) in vehicle controls to 8±1.1 (n=6) branches in treated animals (P<0.001). Secondary+tertiary neuronal branches were also reduced from 15.3±1.2 (n=6) in controls to 6.7±1.0 (n=6) branches in treated animals (P<0.001). A similar effect was observed on neuronal branch numbers after URB597 treatment, where the number of primary branches was 9.8±0.8 (n=6) and the number of secondary+tertiary branches was 8.2±0.9 (n=6) in treated animals. Both of these values were significantly different from vehicle controls (P<0.004). In contrast, the selective MAGL inhibitor JZL184 had no significant effect on neuronal branching (n=6, P<0.6) compared with vehicle controls (Fig. 3F,G). This result indicates that either dual inhibition of FAAH and MAGL or a selective inhibition of FAAH can interrupt primary motor neuronal branching growth, whereas inhibition of MAGL was ineffective.

Immunohistochemical analysis of secondary motor neuronal branches using the anti-zn8 antibody (Fig. 3H–K) showed that only the dual inhibition of FAAH and MAGL by the JZL195 enzymes resulted in abnormally shaped lateral and ventral branches. For instance, embryos exposed to JZL195 showed a drastic reduction in the proportion of normal lateral branches (39±16%, n=6, P<0.01) and ventral branches (56±11%, n=6, P<0.001) emanating from secondary motor axons, compared with vehicle controls (lateral branches, 94±5.6%; ventral branches, 100±0%). In contrast, embryos treated with URB597 (n=6, P=0.55) or JZL184 (n=6, P=0.90) displayed no significant effects upon inhibition of FAAH or MAGL during secondary motor axonal growth (Fig. 3L,M).

The effects on motor neuron branch patterns prompted us to examine whether FAAH and MAGL inhibition early in development could alter the expression of nAChRs at NMJs (Fig. 4). To do this, we fluorescently labelled the postsynaptic membrane at NMJs, using an Alexa-488 conjugated α-BTX label that irreversibly binds to nAChRs. Counts of the total number of α-BTX puncta per 2500 μm² box showed that the JZL195-treated embryos expressed a lower number of nAChR puncta (n=6, P<0.001) in the dorsal and mid-trunk regions, but not in the ventral region. The total number of α-BTX puncta in URB597- and JZL184-treated embryos were not different from vehicle controls in any region of the trunk (n=6, P=0.68 and P=0.25, respectively; Fig. 4E). We measured the diameter of the nAChR puncta and found that embryos treated with JZL195 expressed a smaller mean puncta diameter in the dorsal trunk region (n=6, P<0.01), but curiously, a greater mean diameter in the mid trunk region (n=6, P<0.05) and no change in the ventral trunk region (n=6, P>0.05). Animals treated with URB597 had smaller diameter nAChR puncta in the dorsal trunk but nowhere else (Fig. 4C,H).

Once again, JZL184-treated animals (n=6) showed no significant difference compared with controls (P<0.4 and P<0.1, respectively) when quantifying the mean nAChR cluster size (Fig. 4F). These findings suggest that changes in the development and expression of synaptic NMJs occur if the activity of both FAAH and MAGL enzymes are altered during early development.

Because we observed a deficit in motor neuron morphology and nAChR expression in embryos in which both FAAH and MAGL were inhibited, we asked whether locomotor activity was altered in these animals. To examine this, we placed 5 dpf larva in 96-well plates, allowed them to acclimate and then recorded swimming activity for an hour.

Inhibition of FAAH by URB597 (n=24), MAGL by JZL184 (n=24), or both enzymes together resulted in larvae that exhibited a significant reduction in swimming compared with vehicle controls (n=24, P<0.001; Fig. 5). Embryos exposed to JZL195, URB597 and JZL184 experienced a decrease in the total distanced moved, a reduction in activity and a decrease in velocity of around 40–60% compared with their vehicle controls (P<0.001; Fig. 5B–D).

We used specific receptor blockers of CB₁Rs (AM251; 50 nmol l⁻¹) and CB₂Rs (AM630; 1 µmol l⁻¹) to determine whether the cannabinoid receptors CB₁R and CB₂R were involved in the effects of blocking FAAH and MAGL. We found that inhibition of CB₁R with AM251 largely prevented the JZL195-mediated reduction in primary motor neuron branches (n=6, P<0.04), whereas co-incubation of the enzyme inhibitor with AM630, a CB₂R inhibitor, could not ameliorate the JZL195-induced reduction in primary branches emanating from the primary motor axon (Fig. 6J,K) (n=6, P=0.70). The CB₁R blocker AM251 also partially prevented the effects of JZL195 on secondary+tertiary branches of the primary motor neuron axon (n=6, P<0.04; Fig. 6L). Again, CB₂R blocking by AM630 had no effect on the branching deficits induced by JZL195 (n=6, P=0.84; Fig. 6M). We did not test the effects of JZL184 because as we had previously found, it did not alter the branching patterns of primary and secondary motor neurons, as shown in Figs 3 and 4.

Interestingly, AM251 also prevented the effects of URB597 on motor neuron branching (n=6, P<0.04; Fig. 6J,L), suggesting an involvement of CB₁Rs in motor neuron development. In contrast, we found that AM630 inhibition of CB₂R was unable to prevent the URB597-mediated effects (n=6, P=0.91; Fig. 6K,M). Taken together, these results suggest that inhibition of the FAAH enzyme can alter development of primary motor neuronal morphology, through a mechanism involving CB₁Rs, but not CB₂Rs.

To investigate the effects of the eCB system on secondary motor neuron development, either AM251 (CB₁R inhibitor) or AM630 (CB₂R inhibitor) was co-administered with the dual FAAH and MAGL inhibitor, JZL195 (n=6; Fig. 7A–F). We did not test URB597 and JZL184 because these enzyme inhibitors did not alter secondary motor neuron branching (Figs 3 and 4). Importantly, AM251 (50 nmol l⁻¹) and AM630 (1 µmol l⁻¹) treatment alone (i.e. in the absence of the enzyme inhibitors) did not show any effect on neuronal branching. Our findings indicate that only AM251 treatment could inhibit JZL195-induced branching anomalies associated with secondary motor neuron branches (n=6, P<0.03; Fig. 7G,I). For instance, treatment with JZL195 resulted in a roughly 50–60% reduction in the proportion of normal lateral and ventral branches (Fig. 7G, I), and inhibition of CB₁R completely prevented this reduction (n= 6, P<0.01; Fig. 7G).

In contrast, AM630 inhibition was ineffective in blocking the actions of JZL195 on secondary motor axons (n=6, P<0.5). Taken together, we found that FAAH and MAGL enzyme inhibition, and the subsequent changes in eCB levels working through CB₁R activation, are important for secondary motor neuronal development in zebrafish embryos.

Lastly, to determine whether CBRs were involved in the effects of FAAH and MAGL inhibition on locomotion, we examined the effects of co-treatment of the enzyme inhibitors with AM251 or AM630. As noted previously (in Fig. 5), all three treatments – JZL195, URB597 and JZL184 – showed a significant reduction in movement-related parameters, such as total distance moved, mean activity and mean velocity (Fig. 8). Co-administration with AM251 partially prevented the reduction in locomotor activity induced by JZL195 (n=24, P<0.04) and URB597 (n=24, P<0.03). However, AM251 could not prevent the JZL184-induced effects on locomotion (n=24, P=0.99 for Fig. 8A,C,D), suggesting a role of MAGL in the development of locomotion that is independent of the eCB signalling pathway (Fig. 8A,C,E). Finally, we found that co-treatment with AM630 was completely ineffective in preventing the effects of JZL195, URB597 and JZL184 on larval locomotor activity (n=24, P=0.90–0.99; Fig. 8B,D,F).

Our study shows that disruption of the eCB signalling pathway via inhibition of FAAH and MAGL activity can lead to improper development of primary and secondary motor neurons and locomotion in developing zebrafish. We found that either dual inhibition of FAAH/MAGL or inhibition of FAAH alone can reduce axonal branching of primary motor neurons at 2 dpf through a CB₁R-related mechanism. Furthermore, only the effects of the dual inhibitor of FAAH/MAGL (JZL195) appeared to act via a CB₁R-mediated mechanism on secondary motor neurons and post-synaptic nAChRs at NMJs. Interestingly, MAGL inhibition had no effects on motor neuron development or nAChR expression at 2 dpf. Notably, all three enzyme inhibitors reduced larval swimming at 5 dpf, but only the CB₁R selective inhibitor (AM251) was able to partially rescue the JZL195- and URB597-induced locomotor changes.

The eCB system and its key components are developmentally expressed in different animal models (Maccarrone et al., 2004; Schuel et al., 2002), and in zebrafish both FAAH and MAGL are expressed as early as 1 h after egg fertilization (Oltrabella et al., 2017). In adult zebrafish, FAAH appears to be present at high levels in the brain, and a moderate expression has been identified in skin and testis, whereas MAGL is highly expressed in the brain, kidney, eye and spleen (Oltrabella et al., 2017).

In our study, we confirmed the efficacy of the drugs, by testing an in vitro 96-well plate-based FAAH and MAGL enzyme activity assay. The tests suggested that the concentrations used for JZL195, URB597 and JZL184 were potent enough to preferentially block the activity of FAAH and MAGL enzymes, although their activity was only blocked by approximately 50%. Nonetheless, an effective and cumulative block of 50% activity in the first 24 h was enough to impact development. The dose that we chose to use was based on the efficacy of FAAH and MAGL, and resulted in minimal embryo malformation rates at 2 dpf. It is reported that a reduction in MAGL activity, induced by either genetic deletion or pharmacological inhibitors, can lower 2-AG hydrolysis activity by more than 80% (Blankman et al., 2007; Dinh et al., 2004). Similarly, mice treated with FAAH inhibitors or FAAH (−/−) mice experience an accumulation of 10-fold higher levels of AEA in brains with no change in 2-AG content (Kathuria et al., 2003; Saghatelian et al., 2004; Long et al., 2009). Therefore, a perturbation of FAAH and MAGL will very likely result in an upregulation of AEA and 2-AG, respectively.

Other studies have shown that agonists and antagonists that act on CB₁Rs can impair the growth of motor neuron axons (Williams et al., 2003), and knockout or knockdown of CB₁Rs alters the relative balance between cortical projection neurons (Diaz-Alonso et al., 2012). eCB signalling itself may also play both chemo-attractive and chemo-repulsive roles during cortical development. In fact, neurite outgrowth of cerebellar neurons is altered upon over-activation of CB₁R in the presence of fibroblast growth factor activity (Berghuis et al., 2007, 2005). Further, DAGLα (2-AG synthesizing enzyme) and CB₁Rs can be localized to the growth cones of developing hippocampal neurons in rats, indicating a possible role of 2-AG in early development (Oudin et al., 2011). In contrast, our study found no effects of MAGL interruption on motor neuron development when counting axonal branch numbers from primary motor neurons. Here, we found that pharmacological inhibition of FAAH can affect the growth of primary motor neurons after the first 2 days of development. Although not much is known about the role of FAAH in neurogenesis, FAAH activity does affect both embryo development and fetal viability. For example, reduced AEA hydrolysis activity in peripheral lymphocytes can induce higher abortion rate in women, suggesting an adverse effect of AEA accumulation during pregnancy (Maccarrone et al., 2000).

Even though there were variable effects of JZL195, URB597 and JZL184 during primary motor neuron, secondary motor neuron and NMJ development, all three drugs reduced zebrafish larval motor activity at 5 dpf. The locomotor deficits were more pronounced in animals treated with the dual inhibitor JZL195 (Long et al., 2009). Dual inhibition of FAAH/MAGL (JZL195) or specific inhibition of MAGL (JZL184) in mice can result in hypomotility, but there were little to no effects when FAAH was inhibited (PF-3845), suggesting that in mice, locomotion can be modulated exclusively by a 2-AG/MAGL mechanism (Long et al., 2009). In contrast, our study revealed that an inhibition of not only MAGL but also FAAH can affect locomotor activity, indicating a role of both AEA/FAAH and 2-AG/MAGL in zebrafish locomotor development. The difference between these two findings might be organism specific, or could be attributed to the different FAAH inhibitors in the two different studies. Additionally, our experimental model specifically focused on the very earliest developmental phases in zebrafish during neurogenesis, whereas the study of Long and co-workers (2009) focused on 6-month-old mice.

In our study, exposure to JZL195 and URB597 resulted in aberrant branching of primary motor neurons at 2 dpf, and was followed by a deficit in larval locomotion at 5 dpf. Both the motor neuron morphological and locomotor abnormalities were prevented when CB₁R was inhibited, suggesting that aberrant motor neuron development can occur through CB₁R overactivation. No effects on axonal branching or postsynaptic development of the NMJ were observed following exposure to JZL184 and subsequent downregulation of MAGL activity, suggesting that the JZL184-associated changes in locomotion may not be related to motor neuron or NMJ synaptic development. Furthermore, neither CB₁R nor CB₂R inhibition prevented the JZL184 linked locomotor deficits. Thus it is possible that the effects of JZL184 occur through a non-eCB mechanism. For instance, FAAH and MAGL inhibition can partly affect arachidonic acid (AA) concentration in the CNS (Hu et al., 2017; Kerr et al., 2013), and there is supporting evidence that AA levels can importantly drive postnatal neurogenesis in the brain (Maekawa et al., 2009). The FAAH enzyme can also target N-palmitoyl ethanolamine, which is an anti-inflammatory and analgesic compound (Hamtiaux et al., 2012). However, the inhibitors of the main enzymes FAAH and MAGL affect the eCB pathway such that the eCB signalling is enhanced. It remains to be determined whether the changes in locomotion are a direct consequence of an effect on motor neuron branching or alterations in eCB signalling in the CNS. Indeed, there is clear documentation that perturbations to the eCB system will result in changes to synaptic activity or function (Martella et al., 2016a,b; Mulder et al., 2008; Williams et al., 2003), and this may be the driving force behind the significant reduction in locomotion at 5 dpf.

Because the targets of eCBs are the CB receptors (CB₁ and CB₂), GPR55, peroxisome proliferator-activated receptors (PPARs) and vanilloid receptors (TRPV1), FAAH and MAGL enzymes are becoming a very promising target to treat a wide range of pathological conditions (Sakin et al., 2015; Cisneros et al., 2012; Bari et al., 2006). Although some exogenous cannabinoids have already received market approval to be used as a therapeutic option for anorexia, neuropathic pain and multiple sclerosis treatment, they may induce neurological side effects, such as cognitive and motor dysfunctions. Therefore, exogenously administered cannabinoids are not recommended for long-term treatment (Martín-Sánchez et al., 2009; Rice, 2008). Rather, increasing endogenous cannabinoids (eCBs) via inhibition of their catabolic enzymes, FAAH and MAGL, are suggested to minimize cannabinoid-like side effects (Cravatt et al., 1996; Cao et al., 2013). For instance, the FAAH inhibitor URB597 can mitigate THC-like side effects, such as catalepsy and hyperthermia (Kathuria et al., 2003). However, chronic ablation of FAAH and MAGL have unwanted consequences (Mallet et al., 2016) and permanent elevation of AEA combined with early life stress has severe negative impacts on the stress response pathway through the downregulation of CB₁R expression during brain neurodevelopment (Lazary et al., 2016).

It is clear that FAAH and MAGL enzyme inhibition during early development has significant consequences ranging from impaired motor neuron development to locomotor abnormalities. Despite considerable effort to understand the impacts of the eCB pathway in early development, a comprehensive view of the roles of FAAH and MAGL has not yet been realized. The zebrafish preparation offers an excellent opportunity to focus on the physiology of identifiable neurons (reticulospinal and motor neurons) involved in locomotion during perturbation of these enzymes.