Compounds were purchased from Tocris Bioscience, except for the 5-HT_2A-antagonist, ketanserin (Sigma-Aldrich) and (±)FA that was a gift from Prof. Berten Ceulemans (Child Neurology, University Hospital Antwerp, Belgium). Agonists and antagonists were selected by their high and selective affinity for a specific receptor (Tables 1, 2). In addition, all compounds have a cLogP value higher than one (cLogP > 1.0) ensuring a good bioavailability in ZF larvae (Milan, 2003; García-Pedraza et al., 2013; Watry and Lu, 2013) (Tables 1, 2). These LogP values were calculated by an interactive calculator after conversion of the chemical names to SMILES (cLogP range between 1.21 and 4.59). Compounds were dissolved in DMSO (99.9% spectroscopy grade, Acros Organics) (stock solution), and immediately before use diluted in embryo medium to achieve a final DMSO concentration of 0.1% w/v that also served as VHC control.

Before performing pharmacological experiments, the MTC of each compound was determined. Six dpf WT scn1Lab^+/+ ZF larvae (hereinafter referred to as WT larvae) were incubated in a 96-well plate, one larva per well (tissue culture plate, flat bottom, FALCON^®, USA), with different concentrations of compound or VHC at 28°C on a 14/10 h light/dark cycle (medium was replenished daily). Each larva was checked under the microscope during a period of 48 h for the following signs of toxicity: decreased or no touch response upon a light touch of the tail, loss of posture, body abnormalities, edema, anomalies in heart rate or circulation and death. The MTC was determined using 12 WT ZF larvae (6 dpf) and defined as the highest concentration at which none of these larvae showed signs of toxicity within 48 h of exposure. Compounds were used at their MTC (Tables 1, 2), whereas FA was used at a final concentration of 25 μM as described before (Sourbron et al., 2016).

GraphPad Prism 5 software (GraphPad Software, Inc.) was used for statistical analyses.

The locomotor activity was analyzed by the methods described above [see Locomotor Behavior (Single and Combined)]. The results obtained with method A were analyzed by One-way ANOVA and subsequent Dunnett’s multiple comparison tests. The results obtained with method B were performed for the individual experiments by contingency tables, followed by Fisher’s exact tests.

Electrographic brain activity was analyzed by a Student’s t-test [if data passed the normality test (D’Agostino & Pearson omnibus normality test)]. Mann–Whitney U tests were used if the data did not pass the normality test, as described before (Sourbron et al., 2016). To determine if the genotype (scn1a mutant vs. WT) and/or FA treatment affected the neurotransmitter content, data were analyzed by a two-way ANOVA for genotype and FA treatment as the inter-subject factors. Regarding the analyses of single vs. triple experiments [whether or not combined with FA; see Anti-epileptiform Activity Induced by Sigma₁-Antagonism and 5-HT_1D&2C-Agonism (Triple Treatment) and Counteraction of FA’s Anti-epileptiform Activity by Sigma1-Agonism and 5-HT_1D&2C-Agonism (Triple Treatment), respectively] Mann–Whitney U tests were applied because the data did not pass the normality test.

For all analyses, differences between a treatment group and the equivalent control groups were considered statistically significant if the p-value was below 0.05 (p < 0.05).

Blocking the β₂-adrenergic receptor (by ICI 118551) or the σ₁ receptor (by NE 100) resulted in a significant decrease of epileptiform locomotor activity of the scn1a mutant larvae, as identified by one of the two methods assessing locomotor activity (i.e., method A and method B, respectively) (Figures 2A,B). The β₂-adrenergic receptor antagonist, however, also reduced the locomotor activity of WT larvae, whereas the σ₁-antagonist did not (Figure 2A1). Subsequently, treated scn1a mutant larvae were subjected to brain activity recordings, showing that only the σ₁-antagonist (NE 100) was able to significantly reduce epileptiform brain activity (Figures 3A,B). Representative traces of brain activity during 5 min recordings are depicted in Figure 3C. σ₂-antagonism (by SM 21) did not decrease epileptiform locomotor or brain activity in scn1a mutant larvae but lowered the locomotor activity of WT larvae (Figures 2, 3).

Previous work showed evidence for a role of 5-HT_1D, 5-HT_2A, and 5-HT_2C receptors in seizure modulation (Sourbron et al., 2016). Here we reconfirmed the anti-epileptiform effect of a distinct set of 5-HT-agonists (Figures 2A,B and Table 1). Agonism at the 5-HT_1D (by PNU 109291) and 5-HT_2A receptor (by NBOH-2C-CN) also decreased epileptiform brain activity of scn1a mutant larvae (Figures 3A,B) and did not affect the locomotor activity of WT larvae (Figure 2A1). Similarly, the 5-HT_2C-agonist (Ro 600175) decreased locomotor activity of scn1a mutant but also of the WT larvae (Figure 2A1). However, consecutive brain recordings proved the selective anti-epileptiform activity of this compound in scn1a mutant larvae (Figures 3A,B).

In order to explore the mechanisms of action of FA, we combined FA with the σ₁-agonist (PRE 084) or 5-HT subtype selective antagonists (Figure 1 and Table 2). First, we reconfirmed the anti-epileptiform activity of FA in both assays (locomotor and brain recordings) (Figures 4, 5) as shown before (Dinday and Baraban, 2015; Sourbron et al., 2016). The σ₁-agonist was unable to counteract the anti-epileptiform activity of FA in the locomotor assay (Figures 4A,B), although brain recordings revealed a statistically significant counteraction (Figures 5A,B). Treatment with the 5-HT_1D-antagonist (LY 310762) or the 5-HT_2C-antagonist (SB 242084) significantly counteracted the decrease in locomotor activity as elicited by FA in scn1a mutant larvae (Figures 4A,B). This was not the case for the 5-HT_2A-antagonist (ketanserin). The locomotor behavior was not significantly altered in age-matched WT larvae by combinatorial treatments (Figure 4A2), pointing to a selective effect on scn1a mutant larvae. Interestingly, quantification of the corresponding electrograms confirmed the significant counteraction of the 5-HT_1D-antagonist and 5-HT_2C-antagonist (Figures 5A,B). Representative traces of brain activity during 5 min recording are depicted in Figure 5C. Finally, we performed experiments with a selective 5-HT_2B-antagonist (SB 204741, negative control). This antagonist was unable to counteract the anti-epileptiform activity of FA in the locomotor (Figures 4A,B) and the brain recording assays (Figures 5A,B).

Subsequently, we performed single treatment experiments to investigate whether the compounds, used in combinatorial treatment with FA [see Anti-epileptiform Activity of FA and the Counteraction by Sigma₁-Agonism or 5-HT_1D&2C-Antagonism (Combinatorial Treatment)], affected the locomotor activity of ZF larvae in the absence of FA (Figure 6). Treatment with the σ₁-agonist did not significantly influence the locomotor activity of ZF larvae, although a slight statistically insignificant increase in the epileptiform behavior of scn1a mutant larvae was present (p > 0.05, Figure 6A). The pharmacological inhibition of the 5-HT_1D, 5-HT_2A, and the 5-HT_2C receptor did not alter epileptiform locomotor activity in scn1a mutant larvae (Figures 6A,B). The 5-HT_1D-antagonist and the 5-HT_2B-antagonist significantly affected locomotor activity of age-matched WT larvae (Figure 6A1).

Single treatment (σ₁-antagonist, 5-HT_1D- or 5-HT_2C-agonist) decreased the abnormal (epileptiform) locomotor activity (range 17–34% decrease compared to VHC; Figure 2A1). Triple treatment (combination of the aforementioned three compounds) decreased the epileptiform locomotor activity to a greater extent (58% decrease compared to VHC, Figure 2A1), which is similar to the decrease elicited by FA (>55% decrease compared to VHC; Figure 4A1). Subsequent brain recordings confirmed these anti-epileptiform effects, though no major differences were observed between single and triple treatment (Figures 3, 5). Consistently, statistical analyses did not show any significant differences between triple treatment and single treatments (p < 0.05).

Combinatorial treatment of FA with single treatment of the σ₁-agonist, 5-HT_1D- or 5-HT_2C-antagonist counteracted FA’s decrease in locomotor activity (range 0–41% increase relative to FA; Figure 4A1). Triple treatment (combination of the aforementioned three compounds) counteracted FA’s decrease in locomotor activity more profoundly (89% increase relative to FA; Figure 4A1). Consistently, brain recordings confirmed the counteraction of FA’s anti-epileptiform effect and showed that triple treatment counteracted FA’s effect more profoundly (116–144% increase relative to FA for the frequency and cumulative duration of epileptiform events, respectively; Figure 5), in comparison to the single treatment counteraction experiments (74–91% relative to FA for the frequency and cumulative duration of epileptiform events, respectively; Figure 5). Overall, these more pronounced counteractive effects by triple treatment compared to single treatments were not statistically significant (p > 0.05).

Our data indicate an overall reduction of two out of three monoamines after FA treatment, i.e., DOP and NAD (Figure 7). The most pronounced and significant decrease was shown for NAD (Figure 7 and Supplementary Table S6).

The only statistical significant difference between both genotypes was the GLUT content which was significantly higher in scn1a mutant larvae (-/-) when compared to WT larvae (+/+). However, GLUT content was not significantly altered after FA treatment (Figure 7 and Supplementary Table S6).

In previous experimental work using scn1a mutant larvae, which mimic well the epileptiform features of DS in ZF, we have shown that selective pharmacological agonists of the 5-HT_1D, 5-HT_2A and 5-HT_2C receptor subtypes (GR 46611, TCB 2 and lorcaserin, respectively) substantially decreased the epileptiform activity. Of importance, stimulation of other 5-HT receptor subtypes was not met with success. Using an identical approach here, we explored whether selective antagonists of 5-HT_1D, 5-HT_2A and 5-HT_2C receptor subtypes were able to counteract the FA-induced inhibition of epileptiform locomotor and brain activity observed in scn1a mutant larvae. We show that antagonists of 5-HT_1D and the 5-HT_2C receptor subtypes were able to do so (LY 310762 and SB 242084, respectively), but notably, the 5-HT_2A-antagonist (ketanserin) was not.

Of importance, and in line with our previous findings (Sourbron et al., 2016), a 5-HT_2B-antagonist (SB 204741) did not counteract the anti-epileptiform activity of FA. Consequently, we can conclude that the 5-HT_2B receptor subtype, that is proven to cause cardiotoxic effects after prolonged use of FA (Rothman et al., 2000), is not involved in the anti-epileptiform effect of FA.

We also examined the anti-epileptiform activity of the subtype selective 5-HT_1D-, 5-HT_2A- and 5-HT_2C-antagonists without combinatorial FA treatment (i.e., single treatment) as other studies have shown that blocking these receptor subtypes can protect against seizures in other seizure/epilepsy models. For example, the 5-HT_2A-antagonist used in this study, ketanserin, protected against seizures induced by hippocampal kindling in cats (Wada et al., 1992). The same antagonist was also involved in increasing the latency to audiogenic seizures in DBA/2 mice (Semenova and Tiku, 1997). Consistently, the 5-HT_2C-antagonist, mesulergine, inhibited myoclonic jerks induced by 5-HTP in guinea pigs. This latter antagonist primarily blocks the 5-HT_2C receptor and displays affinity for the 5-HT_2A receptor subtype (Pappert et al., 1998). Selective pharmacological blockade of the 5-HT_1D receptor also inhibited myoclonic jerks in the same model of generalized myoclonus (Hagan et al., 1995). However, our data clearly show that antagonists for these 5-HT receptor subtypes did not show any anti-epileptiform activity in our ZF model of DS.

The activity of FA has always been labeled as 5-HT related, as confirmed in this study. Nonetheless, recent in vitro data suggested antagonism of sigma (σ) and beta₂-adrenergic (β₂) receptors as part of its pharmacological spectrum (Parthena et al., 2016). Hence, in this study we used the scn1a mutant larvae to explore the anti-epileptiform activity of σ- and β₂- antagonism in in vivo conditions.

A selective antagonist of the β₂-adrenergic receptor (ICI 118551) was able to significantly reduce the locomotor activity of scn1a mutant larvae. This is in agreement with the findings of Luchowska et al. (2001), that showed an anti-epileptiform effect against maximal electroshock-induced seizures in mice by antagonizing the β-adrenergic receptor. Additionally, β-adrenergic receptors have been implicated in audiogenic seizures and pindolol, a potent β-antagonist, was found to attenuate audiogenic seizures in DBA/2 mice (Lints and Nyquist-Battie, 1985). Nevertheless, in this work we show that the percentage of scn1a mutant larvae with decreased epileptiform activity was not significantly changed. In addition, the locomotor activity of WT larvae was decreased, suggesting a non-specific alteration of locomotor activity (i.e., potential sedative or muscle-relaxant effect). In line with these latter findings, the β₂-antagonist did not decrease epileptiform brain activity. Thus, our in vivo data imply that the β₂-adrenergic receptor is not involved in modulating epileptiform activity in the ZF DS model.

As far as the sigma receptors are concerned, the σ₁-antagonist (NE 100) decreased both epileptiform behavior and especially brain activity in scn1a mutant larvae. On the other hand, the σ₁-agonist (PRE 084) did not counteract FA’s decrease in abnormal locomotor activity (behavioral assays, Figure 4), whereas brain activity recordings clearly demonstrated a statistically significant counteraction (Figure 5). Locomotion is regulated in a complex way in different parts of the brain (Dunn et al., 2016) and it is possible that these neuronal activities are not always consistent with the brain activity alterations recorded by a single electrode positioned at the forebrain. Alternatively, it is also conceivable that the σ₁-agonist (PRE 084) does not uniformly distribute over the brain, and that especially the peripheral areas probed by the electrode accumulated more compound resulting in a more pronounced effect.

Altogether, the data therefore suggest an involvement of the σ₁ receptor in the mechanism of FA, beyond its serotonergic profile. In contrast, treatment with the σ₂-antagonist (SM 21) did not reduce epileptiform activity in scn1a mutant larvae. This result was somewhat anticipated as most studies underline the role of the σ₂ subtype receptor in cancer-related diseases, rather than in brain-related disorders (Rousseaux and Greene, 2015).

Of interest, a number of studies demonstrate the role of σ₁ receptors in neurological diseases like epilepsy (Rousseaux and Greene, 2015) and overstimulation of σ₁ receptors has been associated with GLUT-related excitatory effects (via NMDA receptors) (Borlot et al., 2014; Nicita et al., 2014; Rousseaux and Greene, 2015). In addition, blockade of σ₁ receptors was associated directly with anti-epileptiform activity (Matsumoto et al., 2002). In contrast, also σ₁-agonists have also been shown to exert clear anti-seizure effects (Meurs et al., 2007; Guo et al., 2015), and possibly σ₁-agonist/antagonists might exert different activities depending on the seizure/epilepsy model. Anyhow, our data show that σ₁-agonism was ineffective in reducing epileptiform events, whereas σ₁-antagonism likely plays a role in the mechanism of FA in the ZF DS model.

The neurotransmitter content analysis of scn1a mutants compared to WT only demonstrated a statistically significant increase in GLUT (p = 0.0036). This finding is in line with the neuro-excitatory properties related to GLUT and its role in epilepsy (Meldrum, 1994). Hence it is not surprising that decreasing glutamatergic effects is one of the main targets for currently available AEDs (Landmark, 2007). Nonetheless, the anti-epileptic compound FA did not alter GLUT levels. Interestingly, the concentration of other neurotransmitters (i.e., NAD and DOP) in head homogenates was altered after FA treatment. Our data indicate a significant DOP decrease by almost one third compared to the VHC treated group (p = 0.0335). The NAD content was even nearly halved when comparing the FA vs. the VHC treated group (p = 0.0035). The latter result is therefore in line with previous work that demonstrated a FA-related decrease of the NAD content in rat brain (Calderini et al., 1975).

As 5-HT-agonism (Astorne Figari et al., 2014) but also σ-antagonism (Monnet et al., 1992) can result in decreased NAD signaling, we hypothesize that the reduction of NAD content after FA treatment as observed in this study is due the effects of FA at 5-HT_2C and σ₁ receptors. Of interest, elevated NAD transmission has been associated with some cases of epilepsy (Fitzgerald, 2010) and although some studies show contradictory results (Svob Strac et al., 2016), also evidence exists supporting the use of NAD-decreasing drugs in the treatment of neurological diseases, including epilepsy (Fitzgerald, 2015).

Of importance, the concentrations measured encompass the total intra- and extracellular levels, so inter-synaptic level alterations were not assessed.

The following, yet incomplete, hypothesis emerges regarding the pharmacological profile of FA that underlies the treatment of DS and possibly other SCN1A-related epilepsies. Of importance, the present study used only one agonist or antagonist at their respective MTCs for the exploration of the involvement of some type of receptors, and further experiments using alternative methods (e.g., a genetic approach using receptor gene knock-down or overexpression) are needed to confirm the present results.

SCN1A mutations, as a major monogenic cause of DS, lead to a malfunction of the encoded voltage-gated sodium channels 1.1 (Na_v1.1). In mammalian (rodent) DS models these channels are mainly affected in inhibitory (GABAergic) interneurons and to a lesser extent in excitatory pyramidal neurons (Dutton et al., 2013; Mistry et al., 2014). As a result, neuronal inhibition is impaired, which is assumed to cause seizures (Wong et al., 2016). Our data suggest that FA reduces epileptiform activity by stimulating 5-HT_2C and 5-HT_1D receptors, leading to enhanced GABAergic neurotransmission (as proven for 5-HT_2C receptors) (Shen and Andrade, 1998; Higgins et al., 2014), and by a σ₁-antagonistic action that can reduce the excitatory neurotransmission by modulating NMDA responses (Rousseaux and Greene, 2015). In addition, FA also significantly decreases the NAD brain content that might contribute to its anti-epileptic effects.