SLC13A5 encodes a citrate transporter highly expressed in the brain and is important for regulating intra- and extracellular citrate levels. Mutations in this gene cause rare infantile epilepsy characterized by lifelong seizures, developmental delays, behavioral deficits, poor motor progression, and language impairments. SLC13A5 individuals respond poorly to treatment options; yet drug discovery programs are limited due to a paucity of animal models that phenocopy human symptoms. Here, we used CRISPR/Cas9 to create loss-of-function mutations in slc13a5a and slc13a5b, the zebrafish paralogs to human SLC13A5. slc13a5 mutant larvae showed cognitive dysfunction and sleep disturbances, consistent with SLC13A5 individuals. These mutants also exhibited fewer neurons and a concomitant increase in apoptosis across the optic tectum, a region important for sensory processing. Further, slc13a5 mutants displayed hallmark features of epilepsy, including an imbalance in glutamatergic and GABAergic excitatory-inhibitory gene expression, increased fosab expression, disrupted neurometabolism, and neuronal hyperexcitation as measured in vivo by extracellular field recordings and live calcium imaging. Mechanistically, we tested the involvement of NMDA signaling and zinc chelation in slc13a5 mutant epilepsy-like phenotypes. Slc13a5 protein co-localizes with excitatory NMDA receptors in wild-type zebrafish and NMDA receptor expression is upregulated in the brain of slc13a5 mutant larvae. Additionally, low levels of zinc are found in the plasma membrane of slc13a5 mutants. NMDA receptor suppression and ZnCl₂ treatment in slc13a5 mutant larvae rescued neurometabolic and hyperexcitable calcium events, as well as behavioral defects. These data provide empirical evidence in support of the hypothesis that excess extracellular citrate over-chelates the zinc ions needed to regulate NMDA receptor function, leading to sustained channel opening and an exaggerated excitatory response that manifests as seizures. These data show the utility of slc13a5 mutant zebrafish for studying SLC13A5 epilepsy and open new avenues for drug discovery.

slc13a5 is expressed in the brain and slc13a5^−/− larvae display a smaller midbrain and increased mortality

Due to an ancestral gene duplication event, zebrafish possess two paralogs of SLC13A5: 5a and 5b. Using antisense RNA probes, we performed in situ hybridization for 5a and 5b expression and found that both paralogs were highly expressed in the zebrafish brain. Widespread expression was observed across the central nervous system during the early developmental stages of 25 hours post fertilization (hpf) and 3 days post fertilization (dpf); however, this expression perhaps became slightly stronger in the midbrain (MB) by 5 dpf (Fig 1A–1F). We further confirmed the specificity of the above-mentioned probes by performing in situ hybridization with sense RNA (control) probes at 3 dpf and found negligible expression of both 5a and 5b (S1A and S1B Fig). To examine the neural cell types expressing Slc13a5, we performed co-immunostaining for Slc13a5 with HuC/D (pan neuronal marker), Gfap (astrocyte marker), vGlut1 (excitatory neuron marker) or Gad67 (inhibitory neuron marker). We observed that Slc13a5 is expressed both in the neurons (including excitatory and inhibitory) and astrocytes in the MB region of 3 dpf zebrafish larvae, with a significantly higher percentage of neurons (85.5%) expressing Slc13a5 compared to astrocytes (14.2%) (Fig 1G–1K).

Next, we injected CRISPRs targeting the fifth exons of 5a and 5b genes to generate mutant alleles, 5a^−/− and 5b^−/−. 5a^−/− allele mutations comprised an 11-nucleotide deletion and 5b^−/− allele mutations comprised one-nucleotide substitution followed by 13-nucleotide insertion, both predicted to encode truncated proteins due to a premature stop (Fig 1L–1O). These mutations were targeted in the sodium-binding site of NaCT, conserved among humans and zebrafish (Fig 1N and 1O). Notably, several variants of SLC13A5 epilepsy have mutations in the sodium-binding site of NaCT, suggesting that sodium binding deficiency causes impaired citrate transport [7]. To assess gene expression levels and potential gene compensation, we used quantitative PCR (qPCR) at 5 dpf and showed that 5a transcript levels were significantly reduced in 5a^−/− and 5a^−/−;5b^−/− larvae compared to wild-type siblings (WTs), indicating active mRNA degradation of 5a^−/− transcripts. Notably, 5a transcript levels were unchanged in 5b^−/− single mutant larvae showing an absence of genetic compensation by paralog upregulation (S1C Fig). Similarly, qPCR analysis showed that 5b transcript levels were significantly reduced in 5b^−/− and 5a^−/−;5b^−/− larvae compared to WTs at 5 dpf, also showing active mRNA degradation of 5b^−/− transcripts, and unchanged in 5a^−/− single mutant larvae, likewise indicating an absence of genetic compensation by paralog upregulation (S1D Fig). We next obtained neuronal population by performing fluorescence-activated cell sorting (FACS) on the heads of 5 dpf WTs (Tg(elavl3:ubci-Cer-sv40)). The significantly enriched expression of elavl3 (also known as huc) in the sorted cells compared to astrocytic gfap confirmed the purity of the neuronal population (S1E Fig). Our qPCR analyses on these sorted neurons showed that both 5a and 5b transcript levels are significantly reduced in 5a^−/−;5b^−/− larvae, further indicating that both the paralogs are highly expressed in the neurons compared to astrocytes (S1F and S1G Fig). Additionally, a significant reduction in the Slc13a5 fluorescence intensity, as measured by calculating corrected total cell fluorescence (CTCF) in the brains of Slc13a5 immunostained 5a^−/−;5b^−/− larvae compared to WTs at 5 dpf, was observed, indicating the loss of Slc13a5 expression also at protein level (S1H–S1J Fig).

No significant gross morphological differences were observed between slc13a5 mutants and WTs at 5 dpf (Fig 1P–1S). To examine neural changes, we performed immunostaining for the neuronal marker HuC/D and quantified the size of all three brain regions at 5 dpf by measuring the width of brain regions. slc13a5^−/− larvae exhibited −5.5% (5a^−/−), −4.9% (5b^−/−) and −6.3% (5a^−/−;5b^−/−) reduction in the MB size compared to WTs, with the forebrain (FB) and hindbrain (HB) size unchanged (Fig 1T-W). We further observed −29.8% (5a^−/−), −17.8% (5b^−/−) and −28.1% (5a^−/−;5b^−/−) reduction in MB volume of 5 dpf slc13a5 mutants compared to WTs (S1K Fig). Consistently, the effect of slc13a5 mutations on MB size correlates with the perhaps slightly stronger expression of both paralogs in the MB at a later zebrafish developmental stage. Next, we performed a survival curve analysis (0–30 dpf) and observed a drastic mortality of slc13a5^−/− larvae, with only 15% (5a^−/−), 10.7% (5b^−/−) and 5.4% (5a^−/−;5b^−/−) mutants surviving past the juvenile stages and into adulthood at 30 dpf (Fig 1X).

slc13a5^−/− larvae exhibit behavioral deficits, reduced neuronal numbers, and increased neuronal apoptosis

Individuals suffering from SLC13A5 epilepsy exhibit comorbidities, including cognitive impairments and sleep challenges [12,25]. To determine the effect of slc13a5 loss-of-function on zebrafish behaviors, we performed a 10-minute acoustic startle protocol by exposing 5 dpf larvae to 440 Hz vibration pulses in the light and calculated the total distance traveled (Fig 2A). The slc13a5^−/− larvae moved longer distances compared to WTs, both during baseline and acoustic stimulation periods, with a significant increase in total distance moved during the complete protocol (413.2% 5a^−/−; 1213.3% 5b^−/−; 285.9% 5a^−/−;5b^−/−), exhibiting hyperactive behavior of slc13a5 mutants in light without any external stimuli as well as their hypersensitivity to acoustic startle and a lack of short-term habituation, which can be a measure of learning potential and might resemble the cognitive dysfunction in SLC13A5 individuals (Figs 2B and S2A). The locomotion heatmaps further illustrate a hyperactivity phenotype of slc13a5^−/− larvae compared to WTs when exposed to acoustic startle (Fig 2C). Next, we assessed the circadian behavior of slc13a5^−/− larvae compared to WTs from 4 dpf to 6 dpf by tracking the total distance moved during 10 hours of darkness in the night and 6 hours of light in the daytime. The slc13a5^−/− larvae traveled longer distances compared to WTs during night (4 dpf: 79.8% 5a^−/−; 192% 5b^−/−; 71.3% 5a^−/−;5b^−/− and 5 dpf: 153.2% 5a^−/−; 231.3% 5b^−/−; 142.1% 5a^−/−;5b^−/−), showing disruptions in nighttime behaviors (Fig 2D). The slc13a5^−/− larvae also traveled longer distances compared to WTs during daytime (5 dpf: 177.3% 5a^−/−; 194.6% 5b^−/−; 157.2% 5a^−/−;5b^−/− and 6 dpf: 209% 5a^−/−; 236.8% 5b^−/−; 211.4% 5a^−/−;5b^−/−), further confirming a hyperactive phenotype during daytime (Fig 2D), which is consistent with other epileptic zebrafish models [26]. Approximately 20% of individuals with epilepsy suffer from anxiety [27,28], thus we tested the “wall-hugging” thigmotaxis behavior of our zebrafish mutants, a validated measure of anxiety in animals and humans [29–31]. We exposed the 5 dpf larvae to 6 min of light and 4 min of darkness and measured the distance they traveled and time they spent close to the wall, i.e., in the outer well zone (S2B Fig). There were no significant changes in the distance traveled and time spent close to the wall in mutants compared to WTs (S2C and S2D Fig). Some individuals with epilepsy are more sensitive to stimuli causing reflex seizures [32]. We thus challenged the 5 dpf larvae with an acute treatment of 5 mM pentylenetetrazol (PTZ) to induce instantaneous convulsions, followed by exposure to thigmotaxis protocol as described above. Under this condition, the slc13a5^−/− larvae swam significantly longer distances (101.2% 5a^−/−; 75.8% 5b^−/−; 85% 5a^−/−;5b^−/−) and spent significantly more time hugging the wall (110% 5a^−/−; 63.6% 5b^−/−; 70.3% 5a^−/−;5b^−/−) compared to WTs, indicating anxiety-like behavior in the presence of a chemical stimulus (S2E and S2F Fig). Combined, slc13a5^−/− larvae display a range of behavioral deficiencies.

We further studied whether these slc13a5^−/− larval behavioral deficits correlated with any molecular neural changes. We focused our assessments on the optic tectum region of the MB as its size was significantly affected in our mutants. We performed immunostaining for HuC/D to quantify the number of neurons and observed a reduction in 5 dpf slc13a5^−/− larvae compared to WTs (−27.5% 5a^−/−; −11.6% 5b^−/−; −22.3% 5a^−/−;5b^−/−; Fig 2E and 2F). To determine when neuronal numbers started to decline, we quantified HuC/D positive-cells at 3 dpf and observed a decrease in slc13a5 mutants at this earlier stage (−37.7% 5a^−/−; −29.1% 5b^−/−; −19.9% 5a^−/−;5b^−/−; Figs S3A and 2G), causing us to question if fewer neurons were born in slc13a5^−/− to start. We quantified the number of neurons (HuC/D positive) around peak neurogenesis (28 hpf) in the region that gives rise to optic tectum and found no significant differences between slc13a5^−/− larvae and WTs at this stage (S3B and S3C Fig). We further checked whether any alteration in cell proliferation at later stages is a possible cause of reduction in neuronal numbers and observed no significant changes in the number of pHH3⁺ cells in 3 dpf and 5 dpf slc13a5^−/− larvae compared to WTs (S3D-G Fig). We next questioned if cell death was a potential cause of the decreased neuronal numbers. We performed acridine orange (A.O.) staining at 5 dpf to label apoptotic cells and showed that slc13a5^−/− larvae exhibited an increase in cell death (42.3% 5a^−/−; 33.5% 5b^−/−; 84.2% 5a^−/−;5b^−/−) compared to WTs (Fig 2H and 2I). This result was further validated by performing co-immunostaining for HuC/D and cleaved-Caspase 3 (CC3), a marker for apoptotic cells, at 5 dpf where we observed an increase in the percentage of CC3⁺ neurons in slc13a5^−/− larvae compared to WTs (111.9% 5a^−/−; 120% 5b^−/−; 167.5% 5a^−/−;5b^−/−), signifying that the reduction in neuronal numbers is a consequence of increased cell death (Figs S3H and 2J).

slc13a5^−/− larvae experience excitation/inhibition imbalance, enhanced fosab expression, strong brain hyperexcitability, and compromised mitochondrial function

Next, in slc13a5 mutant zebrafish we tested for disruption of neuronal excitatory/inhibitory (E/I) balance and synchronous hyperexcitation as found in mammalian [33,34] and zebrafish [23,35,36] epilepsies. Using qPCR at 5 dpf, we quantified excitatory glutamatergic (vglut2a) and inhibitory GABAergic (gad1b) neuronal marker expression in larval brains and showed that vglut2a transcripts were significantly upregulated, whereas gad1b transcripts were significantly downregulated in slc13a5^−/− larvae compared to WTs, suggestive of an E/I imbalance (Fig 3A and 3B). Our group and others have shown previously that seizure induction in epilepsy animal models leads to an upregulated expression of immediate early genes (IEG) in the brain [20,23,37]. We used qPCR at 5 dpf to quantify the expression of an IEG, fosab, and observed a significant upregulation in fosab transcripts in the slc13a5^−/− larvae compared to WTs, further suggestive of brain hyperexcitability (Fig 3C). This result was further validated by performing co-immunostaining for HuC/D and c-Fos (also known as Fosab), at 5 dpf where we observed a 193.9% increase in the percentage of Fosab⁺ neurons in the optic tectum in 5a^−/−;5b^−/− mutants compared to WTs (Fig 3D and 3E).

Next, to characterize seizure-like events, we measured extracellular field potentials from the optic tectum of agarose-immobilized 6 dpf WT and slc13a5^−/− larval brains. WTs showed no evidence of abnormal brain activity (Fig 3F); however, extracellular field recordings in 30% (5a^−/−), 10% (5b^−/−) and 7.4% (5a^−/−;5b^−/−) of slc13a5^−/− larval brains revealed repetitive inter-ictal like discharges (<1s duration) with high-frequency, large-amplitude, above threshold (>0.2mV) spikes (1.7 ±  0.7 Hz; 0.5 ±  0.2 mV, mean ±  SEM), consistent with a spontaneous epileptic phenotype (Figs S4A-E and 3G-I). Notably, 40% (5a^−/−), 50% (5b^−/−) and 18.5% (5a^−/−;5b^−/−) of slc13a5^−/− larvae also displayed repetitive but below threshold (<0.2mV) brain activity (Figs 3I and S4E). No longer duration (>1 s) discharges characterized as ictal-like events [38] were identified in the mutants. Thus, to increase the sensitivity of the technique, we employed a 16-channel NeuroProbe to measure extracellular field potentials from the optic tectum of agarose-immobilized 5 dpf WT and 5a^−/−;5b^−/− larval brains. During the protocol, 91.7% WTs and 100% 5a^−/−;5b^−/− larvae occasionally twitched, producing large-amplitude (>0.2 mV) deflections in the recordings. These deflections were of short duration (<1 s), and thus they were easily distinguishable from epileptic activities. Besides these movement-induced deflections, WTs showed no evidence of abnormal brain activity (Fig 3J). In contrast, 16.7% 5a^−/−;5b^−/− larvae showed a distinct epileptic pattern, consisting of a series of short bursts, with a total duration of> 10s (Figs 3K and S5A-C), and 8.3% of 5a^−/−;5b^−/− larvae showed repetitive twitches (>5 s) at consistent frequency (~1.8 Hz) (Figs 3L and S5D), an indication of epileptic seizures [39]. Combined, these data support that our slc13a5^−/− larvae exhibit seizure-like events, with the low percentage of slc13a5 mutants showing such events (Fig 3M) potentially due to spatio-temporal restrictions of this electrophysiological approach.

Dysregulation of mitochondrial function is observed across various types of epilepsies, including those modeled in zebrafish [23,40–42], and here we used the Seahorse XF Flux Bioanalyzer to assess bioenergetics in our slc13a5^−/− larvae at 6 dpf (Fig 4A). We observed a dampening in basal respiration (−20.1%), ATP-linked respiration (−36.6%), total mitochondrial respiration (−32.9%), and non-mitochondrial respiration (−6.1%) in 5a^−/−;5b^−/− larvae compared to WTs (Fig 4B-4E). In contrast, maximum respiratory capacity and reserve capacity were upregulated by 28.2% and 351.2%, respectively (5a^−/−;5b^−/−), whereas proton leaks were unchanged (Fig 4F-4H), consistent with other reports whereby the loss of citrate uptake by cells due to SLC13A5 mutations causes changes in cellular metabolism [43–46].

Transcriptomics analysis reveals differentially regulated pathways in slc13a5^−/− larvae

Further, to unveil the transcriptomics signature of slc13a5 mutants, we performed bulk RNA sequencing on pooled 5 dpf WT, 5a^−/−, 5b^−/− or 5a^−/−;5b^−/− heads. Interestingly, principal component analysis (PCA) revealed that 5a^−/− and 5a^−/−;5b^−/− larvae are more transcriptomically similar than 5b^−/− larvae and WTs (S6 Fig). Further, we observed that 400 genes were significantly upregulated and 380 genes were significantly downregulated (padj < 0.05 and log2FC >  1.5 or < −1.5) in 5a^−/− and 5a^−/−;5b^−/− larvae compared to WTs (S7 and S8 Figs). In particular, we found genes involved in apoptosis and metabolism to be upregulated in 5a^−/− and 5a^−/−;5b^−/− larvae (Fig 5A), validating our data showing increased neuronal death (Figs 2J and S3H) and disrupted neurometabolism (Fig 4). Additionally, we found genes involved in inflammation and immune response to be upregulated in 5a^−/− and 5a^−/−;5b^−/− larvae (Fig 5A), indicative of possible neuroinflammation in these mutants due to seizure-induced neuronal apoptosis, as shown by others [47,48]. Our 5a^−/− and 5a^−/−;5b^−/− larvae further showed downregulation of genes affecting circadian rhythm (Fig 5B), consistent with circadian behavioral deficits in these mutants (Fig 2D). Some genes whose loss of function is associated with epilepsy or other neurological disorders also were downregulated, as were genes regulating calcium ion transport (Fig 5B), pointing towards disturbed calcium homeostasis as a plausible cause of seizures in these mutants. Also, genes involved in the negative regulation of apoptosis were downregulated (Fig 5B), further validating previously shown enhanced apoptosis in these mutants (Figs 2J and S3H).

Next, we observed that 304 genes were significantly upregulated and 444 genes were significantly downregulated (padj < 0.05 and log2FC >  1.5 or < −1.5) in 5b^−/− larvae compared to WTs (S9 and S10 Figs). Specifically, genes involved in inflammation and immune response were upregulated in 5b^−/− larvae (Fig 5A), indicative of activated neuroinflammatory pathway in these mutants perhaps due to seizure-induced neuronal apoptosis (Figs 2J and S3H). Further, genes involved in apoptosis were upregulated (Fig 5A) and those that negatively regulate apoptosis were downregulated (Fig 5B) in 5b^−/− larvae, consistent with our data showing increased neuronal death in these mutants (Figs 2J and S3H). Some genes whose loss of function is associated with neurological disorders were also downregulated (Fig 5B). Finally, these studies revealed that 5a^−/− and 5a^−/−;5b^−/− larvae exhibit similar transcriptomic signatures and thus, we focused our further study on 5a^−/−;5b^−/− larvae.

A mechanistic link between high extracellular citrate and seizure susceptibility in slc13a5^−/− larvae

The underlying mechanism linking citrate transporter mutations to seizures remains debated [11], with one hypothesis being that the accumulation of extracellular citrate chelates the ions important for regulating neuronal membrane depolarizations and E/I balance. N-methyl-D-aspartate (NMDA) receptors are glutamate-gated channels that allow the timely influx of calcium (Ca⁺²) to mediate excitatory signaling. Zinc is a critical NMDA receptor inhibitor that helps regulate Ca⁺² uptake and prevent prolonged excitation. Citrate is a potent zinc chelator, which when present in excess might bind all the zinc and prevent it from modulating NMDA receptor function, causing unremitting Ca⁺² influx and excitatory signaling.

First, we tested whether slc13a5 mutants display increased Ca⁺² events, as would be predicted if NMDA receptor signaling was overactive. We crossed WT and 5a^−/−;5b^−/− larvae with the Tg(elavl3:Hsa.H2B-GCaMP6s) zebrafish that express a neuron-specific calcium biosensor, and performed live calcium imaging at 3 dpf. 5a^−/−;5b^−/− larvae displayed a 19% increase in calcium amplitude and 469% in the frequency of calcium events compared to WTs (Fig 6A and 6B, S1 and S2 Movies). The representative single neuron calcium traces illustrate the extent of increased Ca⁺² signaling in the 5a^−/−;5b^−/− mutants (Figs 6C and S11A-F). Next, we assessed phosphorylated ERK (pERK) levels, an indirect downstream reporter of calcium signaling. We co-stained with pERK- and total ERK- (tERK) selective antibodies and imaged 5 dpf larval brains. Using a standard zebrafish reference brain to calculate normalized pERK levels as described previously [49,50], whole-brain activity maps were generated with voxels exhibiting significantly higher (green) and lower (magenta) pERK levels in the different brain regions. We found a significant increase in pERK signaling (green) in the FB, MB, and HB in 5a^−/−;5b^−/− mutants compared to WTs, signifying upregulation of ERK phosphorylation possibly due to enhanced calcium influx in the neurons (Fig 6D). Notably, the lower pERK intensity observed in the eyes (magenta) is likely due to autofluorescence while imaging and therefore we do not consider it further. Owing to an increased pERK signaling in the FB and HB of 5a^−/−;5b^−/− mutants, we performed live calcium imaging at 3 dpf (as per above) in these brain regions as well. 5a^−/−;5b^−/− larvae displayed an increase of 8.3% in the amplitude and 82.3% in the frequency of calcium events compared to WTs in the FB (S11G and S11H Fig, S3 and S4 Movies), with representative single neuron calcium traces showing the extent of increased Ca⁺² signaling in the 5a^−/−;5b^−/− mutants (S11I Fig). Interestingly, 5a^−/−;5b^−/− larvae displayed an increase of 258.5% in the frequency of calcium events with no significant change in the amplitude compared to WTs in the HB (S11J and S11K Fig, S5 and S6 Movies), with representative single neuron calcium traces depicting the extent of increased Ca⁺² signaling in the 5a^−/−;5b^−/− mutants (S11L Fig).

Next, we wanted to determine the involvement of NMDA receptor signaling in slc13a5 mutants. NMDA receptor subunit grin1 (also known as glun1) is reported to be responsible for calcium permeability of NMDA receptors [51,52]. Both grin1a and grin1b paralogs are expressed in the zebrafish brain and their loss of function affects behaviors [53,54]. We confirmed Grin1 expression in Slc13a5⁺ MB neurons by immunostaining 3 dpf larvae and observed their co-expression in several MB cells (Fig 6E-G). Further, we co-immunostained 3 dpf and 5 dpf larvae with Grin1 and NeuN (mature neuronal marker) and observed a significant increase in the Grin1 fluorescence intensity, as measured by calculating CTCF in the optic tectum of 5a^−/−;5b^−/− larvae compared to WTs, indicating active NMDA receptor signaling in slc13a5 mutant brains (Fig 6H-K).

To determine whether excessive extracellular citrate causes zinc chelation in the slc13a5^−/− larvae, we incorporated a fluorescent zinc sensor, Palm-ZP1 that detects zinc in the extracellular side of the plasma membrane [55]. Fifty hpf larvae were co-incubated in 5 μM Palm-ZP1 and 5 μg/ml CellMask Orange (plasma membrane marker dye) for 15 min at 28°C, followed by live imaging the brain region (Fig 6L). We observed a −53.3% reduction in the ratio of Palm-ZP1:Cell Mask Orange fluorescence intensity, as measured by calculating Palm-ZP1:Cell Mask Orange CTCF ratio in the plasma membrane of 5a^−/−;5b^−/− larvae compared to WTs (Fig 6M, 6M’, 6N, 6N’ and 6Q), suggesting possible zinc chelation due to elevated extracellular citrate. Notably, between untreated WTs (e.g., in E3 water) and DMSO-treated WTs (vehicle for Palm-ZP1), there were no significant differences in the Palm-ZP1:CellMask Orange fluorescence intensity ratios, as measured by calculating CTCF ratios in the plasma membrane (Fig 6R). Further, prior incubation in 25 μM ZnCl₂ for 2 hours led to an increase of 302.2% in Palm-ZP1:CellMask Orange CTCF ratio in WTs compared to untreated WTs (Fig 6M, 6M’, 6O, 6O’ and 6Q). Similarly, ZnCl₂-treated 5a^−/−;5b^−/− larvae also exhibited an increase of 409.5% in Palm-ZP1:CellMask Orange CTCF ratio compared to untreated mutants (Fig 6N, 6N’, 6P, 6P’ and 6Q).

Collectively, these data suggest that slc13a5 mutants exhibit overactivated NMDA receptor signaling due to zinc chelation by high extracellular citrate.

Suppression of NMDA signaling and zinc treatment rescue dysregulated calcium activity, neurometabolism, and startle behaviors in slc13a5^−/− larvae

Next, we used pharmacological and genetic approaches to investigate whether NMDA receptor signaling influenced 5a^−/−;5b^−/− larvae phenotypes. Pharmacologically, we exposed WT and mutant zebrafish larvae to memantine, an established NMDA antagonist used in other zebrafish studies [56]. Genetically, we designed morpholinos to knock down both the paralogs of NMDA receptor subunit grin1. We co-injected grin1a;grin1b morpholinos (MOs) in the WT and 5a^−/−;5b^−/− zebrafish, expressing the Tg(elavl3:Hsa.H2B-GCaMP6s) transgene and performed live calcium imaging in the MB at 3 dpf. Fifty μM memantine-treatment for 2 hours prior to imaging and grin1a;grin1b MO injection in 5a^−/−;5b^−/− larvae caused a reduction of −17.5% (memantine-treated) and −18.6% (grin1a;grin1b MO injected) in the amplitude and −72.7% (memantine-treated) and −83.3% (grin1a;grin1b MO injected) in the frequency of calcium events compared to vehicle-treated (0.5% DMSO) or control MO-injected 5a^−/−;5b^−/− larvae (Fig 7A and 7B, S7-S9 Movies). The representative single neuron calcium traces illustrated that 5a^−/−;5b^−/− larvae exhibited fewer above-threshold events after memantine treatment or grin1a;grin1b MO injection (Fig 7C). Notably, there were no significant changes in the amplitude or frequency of calcium events in WT larvae after treatment with memantine or grin1a;grin1b MO injections (S12A and S12B Fig). Importantly, a significant reduction in the Grin1 fluorescence intensity, as measured by calculating CTCF in the brains of Grin1 immunostained grin1a;grin1b MO-injected WTs compared to control MO-injected WTs at 3 dpf, was observed, confirming efficacy of grin1a;grin1b MOs (S12C Fig). These results demonstrate a role for NMDA receptors in slc13a5 mutant dysregulation of Ca⁺² signaling.

We further assessed the involvement of zinc chelation in 5a^−/−;5b^−/− larvae phenotypes by using a pharmacological approach. We treated WT and 5a^−/−;5b^−/− larvae, expressing the Tg(elavl3:Hsa.H2B-GCaMP6s) transgene, with 25 μM ZnCl₂ or vehicle (E3 water) for 2 hours, followed by live calcium imaging in the MB at 3 dpf. ZnCl₂ treatment in 5a^−/−;5b^−/− larvae led to a reduction of −18.3% in the amplitude and −62.9% in the frequency of calcium events compared to vehicle-treated 5a^−/−;5b^−/− larvae (Fig 7D and 7E, S10 and S11 Movies). The representative single neuron calcium traces illustrated that 5a^−/−;5b^−/− larvae exhibited fewer above-threshold events after ZnCl₂ treatment (Fig 7F). Of note, we did not observe any significant changes in the amplitude or frequency of calcium events in WT larvae after treatment with ZnCl₂ (S12D and S12E Fig). These data support the role of zinc chelation in perturbed Ca⁺² signaling in slc13a5 mutants.

An alternative mechanistic hypothesis is that the loss of intracellular citrate disrupts neuronal homeostasis, leading to hyperexcitability. If this alternative hypothesis is true, then blocking NMDA receptors or treating with ZnCl₂ should have no additional effect in the mutants on a key measure of neuronal homeostasis: bioenergetics. 6 dpf 5a^−/−;5b^−/− larvae were treated with 50 μM memantine, 50 μM MK801 (a second potent NMDA antagonist [54]) or vehicle (0.5% DMSO) for 2 hours prior to initiating the Seahorse assay (as per above). Blocking NMDA receptors significantly rescued several bioenergetics parameters dysregulated in the 5a^−/−;5b^−/− larvae, including basal respiration, ATP-linked respiration, total mitochondrial respiration, and non-mitochondrial respiration (Fig 7G-7J). Further, 6 dpf 5a^−/−;5b^−/− larvae were treated with 25 μM ZnCl₂ or vehicle (E3 water) for 2 hours, followed by performing Seahorse assay as described previously. ZnCl₂ treatment significantly rescued basal respiration, ATP-linked respiration, total mitochondrial respiration, and non-mitochondrial respiration in the 5a^−/−;5b^−/− larvae (Fig 7K-7N). Combined, these data suggest that an interplay between extracellular citrate, zinc chelation and NMDA receptors mediates the seizure-like phenotype in slc13a5 mutants. Notably, memantine, MK801 and ZnCl₂ had no effect on these metabolic parameters in the WTs (S13A-H Fig).

Finally, to link NMDA receptor function and zinc chelation to circuit-level behaviors of slc13a5 mutant zebrafish, we examined whether the impaired startle behavior of our mutants was rescued by NMDA antagonism and ZnCl₂ treatment. We treated 5 dpf larvae with memantine (50 μM), ZnCl₂ (25 μM) or their respective vehicles for 2 hours, followed by the acoustic startle protocol (Fig 2A). Treatment with memantine and ZnCl₂ rescued the impaired startle response in 5a^−/−;5b^−/− larvae, causing a significant reduction in the total distance swam during the complete protocol (Fig 7O, 7P). No effect was observed of memantine and ZnCl₂ exposure on the swimming activity of the WTs in this acoustic startle protocol; however, MK801 exposure significantly reduced the swimming activity of WTs, making it unsuitable for further analyses herein (S13I and S13J Fig).

Taken together, we propose a model whereby deficiencies in Slc13a5 cause accumulation of extracellular citrate that excessively chelates zinc, which fails to efficiently buffer NMDA receptor activity leading to increased Ca⁺² uptake inside the neurons and excessive excitatory firing that manifests as seizures. We further show that blocking NMDA receptors or treatment with zinc prevents enhanced cytoplasmic Ca⁺² influx into the neurons, thereby rescuing epileptic phenotypes in our zebrafish slc13a5 mutants (Fig 8).

Zebrafish maintenance

Adult zebrafish (TL and AB strains) were maintained at 28°C in a 14-hour light/10-hour dark cycle under standard aquaculture conditions, and fertilized eggs were collected via natural spawning. The animals were fed twice daily with Artemia. Zebrafish embryos and larvae were maintained in E3 water in a non-CO₂ incubator (VWR) at 28°C on the same light–dark cycle as the aquatic facility. All protocols and procedures were approved by the Health Science Animal Care Committee (protocol number AC22–0153) at the University of Calgary in compliance with the Guidelines of the Canadian Council of Animal Care.

Generation of zebrafish mutants by CRISPR/Cas9

5a^−/− (slc13a5a^ca204/+) and 5b^−/− (slc13a5b^ca205/+) zebrafish were generated by using CRISPR/Cas9 mutagenesis. The identified founders carried 11-nucleotide deletion in the fifth exon for 5a mutations and one-nucleotide substitution followed by 13-nucleotide insertion in the fifth exon for 5b mutations. The gRNA sequences are CCACACTCACAGGCACTGGACCA for 5a mutation and GTTTGCTATGCTGCCAGTGTTGG for 5b mutation. WTs, single homozygous (5a^−/− and 5b^−/−) and double homozygous (5a^−/−;5b^−/−) larvae were used for the experiments (by breeding heterozygous adults). Genotyping to distinguish 5a^−/−, 5a^+/− and WTs was carried out by performing high-resolution melt analysis (HRMA) using primers listed in S1 Table. Genotyping to distinguish 5b^−/−, 5b^+/− and WTs was carried out by performing HRMA using primers listed in S1 Table, followed by running the non-heterozygous PCR products on agarose gel, based on different sizes of PCR product bands (WTs- 118 bp and 5b^−/−- 131 bp). Survival of animals was analyzed by recording their mortality rate every other day, beginning from 0 dpf until 30 dpf and percentage survival was plotted (GraphPad Prism).

Morpholino injections

The grin1a (5′-CCAGCAGAAGCAGACGCATCGT-3′) and grin1b (5′-CGAACAGAACCAGGCGCATTTTGCC-3′) MOs were purchased from Genetools (Philomath, OR) and co-injected at the one-cell stage at 0.75 ng concentrations in all experiments described. The standard control MO (5′-CCTCTTACCTCAGTTACAATTTATA-3′) purchased from Genetools was also injected at 0.75 ng concentration. The doses of all MOs were determined as optimal by titration (no toxic effects were observed).

Gene expression analysis by in situ hybridization and qPCR

Digoxigenin-labeled sense and antisense RNA probes were synthesized for 5a and 5b as described previously [66], using the primers listed in S2 Table. The embryos and larvae were fixed at desired stages in 4% paraformaldehyde (Sigma-Aldrich) overnight at 4°C. The whole mount in situ hybridization protocol was performed as described previously [66]. qPCR was performed for 5a, 5b, vglut2a, gad1b, and fosab on cDNA obtained from the heads of 5 dpf larvae. Tg(elavl3:ubci-Cer-sv40)^y342 animals imported from Zebrafish International Resource Center were outcrossed with 5a^+/−;5b^+/− to raise adults and subsequently 5a^−/−;5b^−/− and WT larvae were obtained for FACS. FACS-sorted neurons were obtained from the pool of heads of 5 dpf larvae (Tg(elavl3:ubci-Cer-sv40) and the cDNA obtained was used to perform qPCR for elavl3, gfap, 5a, and 5b. rpl13 was used as internal control. The primers used for qPCR are listed in S3 Table.

Behavioral assays

Zebrafish larvae maintained in 48-well plates were habituated for 20 min, under ambient light. This was followed by behavioral assessment to measure distance moved in 100% light and during acoustic startle, using Zebrabox (Viewpoint Life Sciences). Similarly, larvae were habituated in 20 min of darkness, followed by tracking their movement in 100% darkness. For thigmotaxis protocol, we used 24-well plates and generated a protocol to divide each well into inner well zone and outer well zone. The larvae were habituated for 20 min in ambient light (or treated with 5 mM PTZ prior to this). This was followed by tracking the distance traveled and time spent close to the wall under 6 min of 100% light and 4 min of 100% darkness. Tracking of total distance moved and time taken as a measure of swimming behavior was analyzed using Zebralab V3 software (Viewpoint Life Sciences). Locomotion heatmaps depicting the movement of larvae were also generated by using Zebralab V3 software. All the behavioral assessments were performed on 4 dpf to 6 dpf larvae.

Drug treatments

Memantine (Cayman Chemical) and MK801 (Sigma-Aldrich) stock solutions were prepared in DMSO. ZnCl₂ (Sigma-Aldrich) stock solution was prepared in water. The drugs were assessed for toxicity and the highest concentration which did not induce toxicity, was used in subsequent assays. On the day of experiments, the stock solutions were diluted to a final concentration of 50 μM (for memantine and MK801) and 25 μM (for ZnCl₂) in E3 water. The zebrafish larvae were treated with drugs for 2 hours, followed by behavior assessment, Seahorse assay and calcium imaging. The final DMSO concentration was 0.5% and was used as vehicle control in the memantine- and MK801-treatment experiments. Larvae incubated in E3 water alone were used for control experiments for ZnCl₂-treated experiments.

Electrophysiological measurements

Electrophysiological recordings in zebrafish larvae were performed as described previously [41,67]. Briefly, 6 dpf larvae were paralyzed using α-bungarotoxin (1 mg/ml, Tocris) and embedded in 1.2% low melting agarose (LMA). The dorsal side of the larvae was exposed to the agarose gel surface and accessible for electrode placement. Larvae were placed on an upright stage of an Olympus BX51WI upright microscope, visualized using a 5X MPlanFL N Olympus objective, and perfused with embryo media. A glass microelectrode (3–8 MΩ) filled with 2 M NaCl was placed into the optic tectum of zebrafish, and a recording was performed in current-clamp mode, low-pass filtered at 1 kHz, high-pass filtered at 0.1 Hz, using a digital gain of 10 (Multiclamp 700B amplifier, Digidata 1440A digitizer, Axon Instruments) and stored on a PC computer running pClamp software (Axon Instruments). Baseline recording was performed for ≥  five minutes. The threshold for detection of spontaneous events was set at three times noise, as described previously [68]. The frequency and amplitude of spontaneous events were analyzed using the Clampit software 11.0.3 (Molecular Devices, Sunnyvale, CA).

Electrophysiological recordings with NeuroProbe were performed using 5 dpf larvae. Individual larva was transferred to 50 μL E3 water on a clean slide. 2 μL Tricane (0.02%) and 4 μL α-bungarotoxin (1 mg/ml, Tocris) were added to the water to anesthetize and paralyze the larva for 5 min. The larva was then removed from the anesthetizing solution, embedded in 1.5% low-melting-point agarose with the dorsal side facing upwards, and transferred to the recording stage. The anesthetizing solution was added to the recording bath, in addition to 1,500 μL of E3 water. A 16-channel neural probe (Cambridge NeuroTech, Cambridge, United Kingdom), mounted on a micromanipulator (Luigs and Neumann, Ratingen, Germany), was inserted into the optic tectum of larva. A silver wire was placed in the recording bath as grounding. Physiological signal was amplified using a RHD2132 headstage (Intan Technologies, Los Angeles, United States) and digitized by a USB Acquisition Board (Open Ephys, Altanta, United States) at 30k samples/s. The high-pass was set at 1 Hz and the low-pass set at 7,000 Hz. Data was stored on a computer running the Open Ephys software v 0.5.5.3 [69]. A digital camera (IMX265, Sony Group Corporation, Tokuo, Japan) was used for monitoring and video recording. Each trial lasted ≥ 20 min. Data was subsequently imported into Spike2 v10.21 (Cambridge Electronic Design Limited, Cambridge, United Kingdom) and filtered with a band-pass second order IIR filter, set at 1-1k Hz. Filtered data was used for subsequent analysis and quantification.

Metabolic measurements

Oxygen consumption rate (OCR) measurements were performed using the XF24 Extracellular Flux Analyzer (Seahorse Biosciences). Individual 6 dpf zebrafish larvae were placed in 24 wells on an islet microplate and an islet plate capture screen was placed over the measurement area to maintain the larvae in place. Seven measurements were taken to establish basal rates, followed by treatment injections and 18 additional cycles [70]. Rates were determined as the mean of two measurements, each lasting 2 min, with 3 min between each measurement [71]. Three independent assays were performed to establish metabolic measurements. A script was generated on R (using dplyr, reshape2, tidyr, purr and tidyverse libraries) to calculate the metabolic measurements for all the parameters. The code for this analysis was uploaded to Zenodo and is available at https://zenodo.org/records/14853182.

Immunofluorescence

To perform whole mount immunostaining (for HuC/D and pHH3), 3 dpf and 5 dpf larvae were fixed in 4% paraformaldehyde overnight at 4°C, depigmented with a 1% H₂O₂/3% KOH solution, followed by antigen retrival with 150mM Tris HCl (pH 9.0) for 15 min at 68°C, permeabilization (PBS/0.3%Triton-X/1%DMSO/1%BSA/0.1%Tween-20) for 2 hours at room temperature (RT) and blocking (PBS/1%DMSO/2%fetal bovine serum (FBS)/1%BSA/0.1%Tween-20) for 1 hour at RT. Later, the larvae were incubated in primary antibody overnight at 4°C, followed by washing and secondary antibody incubation overnight at 4°C. Finally, the immunostained larvae were washed and counterstained with Hoechst (Thermo Fisher Scientific). pERK/tERK whole mount immunostaining on 5 dpf larvae was performed as described previously [50]. The larvae were mounted in 0.8% LMA for imaging. To perform immunostaining (for HuC/D, Slc13a5, Gfap, vGlut1, Gad67, cleaved-Caspase3, Grin1, NeuN and c-Fos) on 12 µm thick cryosections of 28 hpf embryos, 3 dpf and 5 dpf larvae, antigen retrival was carried out with citrate buffer (pH 6.0) for 15 min at 95°C. This was followed by permeabilization with 1% Triton-X for 30 min at RT and blocking in 5% normal donkey serum or 5% normal goat serum for 1 hour at RT. Later, the sections were incubated in primary antibody overnight at 4°C, followed by washing and secondary antibody incubation for 3 hours at RT. Finally, the immunostained sections were washed, counterstained with Hoechst and mounted for imaging. The primary antibodies (and their working concentrations) used in this study are listed in S4 Table. Secondary antibodies were used at 1:200 concentration (Life Technologies).

Live acridine orange staining

The larvae were raised in embryo medium containing 1-phenyly-2-thiourea (PTU) to prevent pigment formation. 5 dpf larvae were incubated for 30 min in embryo medium containing 0.002% A.O. solution (Sigma-Aldrich) at 28°C. The stained fish were washed with embryo medium several times and mounted in 0.8% LMA for live imaging.

Palm-ZP1 and CellMask Orange staining

Palm-ZP1 (dissolved in DMSO) was received as a gift from Stephen J. Lippard lab and used in 5 μM working concentration. CellMask Orange (C10045, Invitrogen) was used as plasma membrane marker in 5 μg/ml working concentration. Fifty hpf embryos were treated with Palm-ZP1 and CellMask Orange for 15 min at 28°C, followed by mounting in 0.8% LMA for live imaging.

Imaging and quantification

Bright-field images of in situ hybridization-stained embryos and larvae (whole mount and 12 µm thick cryosections) and live animals were obtained using Stereo Discovery.V8 microscope (ZEISS). The live A.O. stained/Palm-ZP1 and CellMask Orange stained/immunostained whole mount samples and cryosections were imaged at 20X magnification using LSM 880/LSM 900 confocal microscopes (ZEISS). Larvae immunostained with pERK/tERK were imaged at 10X magnification in the Airyscan FAST mode of LSM 880. After imaging, the acquired confocal z-stacks were processed, cell counting, and brain width measurements were performed with ZEN software. MB volume measurements were performed using Imaris software. Fluorescence intensity measurements (for Slc13a5, Palm-ZP1, CellMask Orange and Grin1 signal) were performed by calculating CTCF using Fiji software. The ratio of Palm-ZP1:CellMask Orange CTCF was used in data generation.

Brain activity analysis using pERK/tERK immunostaining

The processing and analysis pipeline is performed using the protocol described previously [49]. Briefly, images were registered to a standard zebrafish reference brain using Computational Morphometry Toolkit (CMTK). The pERK images were normalized with the tERK staining. In MATLAB (MathWorks), the Mann-Whitney U statistic Z score is calculated for each voxel, comparing the mutant and WT groups using MapMAPPING. The significance threshold was set based on a false discovery rate (FDR) where 0.05% of control pixels would be called as significant.

Calcium imaging and data analysis

Tg(elavl3:Hsa.H2B-GCaMP6s)^jf5Tg animals imported from Zebrafish International Resource Center were outcrossed with 5a^+/−;5b^+/− to raise adults and subsequently 5a^−/−;5b^−/− and WT larvae were obtained for calcium imaging. The larvae were raised in E3 medium containing PTU to prevent pigment formation. Briefly, 3 dpf larvae were immobilized using α-bungarotoxin and tricaine and embedded in 0.8% LMA. Larvae were imaged at 10X magnification the Airyscan FAST mode of LSM 880 and single-plane movies were acquired at 9.6 frames per second for ~ 5 min. Mean grey value measurements of individual neurons in the FB, MB and HB were performed in Fiji and grey values were normalized against a basal mean grey value. A script was generated on R to calculate fluorescence variation measurements (ΔF/F) in each frame of all the neurons, followed by calculating the frequency and amplitude of calcium events (with a threshold value of ΔF/F > 2) for each larvae recorded. The code for this analysis was uploaded to Zenodo and is available at https://zenodo.org/records/14853182. We have used a threshold amplitude value of ΔF/F > 2, considering the pattern that most of the 3 dpf WT larvae stayed within this threshold.

Bulk RNA sequencing

The heads of 5 dpf larvae were pooled to generate RNA using TRI Reagent (Sigma-Aldrich). Three biological RNA samples were prepared from each genotype, i.e., WTs, 5a^−/−, 5b^−/− and 5a^−/−;5b^−/−. RNA samples (RIN values > 7) were used to generate Stranded poly (A) RNA library and Bulk RNA sequencing (NextSeq2000 P2 100 cycle) was carried out. For RNASeq alignment and differential gene expression analysis, paired end reads were aligned to Danio rerio genome reference GRCz11 release 112 along with annotated transcript GTF file (version 4.3.2) [72] downloaded from Lawson Lab using STAR (version: STAR_2.5.2b) [73]. Gene counts obtained from STAR were processed to provide an input to DESeq2 (version 1.36.0) for differential gene expression analysis [74]. We excluded those transcripts with zero counts in more than four samples. We used log normalized counts to construct heatmaps for prioritized candidate genes using ComplexHeatmap (v2.12.1) [75,76].

Scripts for differential gene expression analysis and to create heatmaps were uploaded to Zenodo and are available at https://zenodo.org/records/14853182. Sequencing raw and analyzed data are publicly available at NCBI’s Gene Expression Omnibus (GEO accession: GSE275235).

Statistical analysis

GraphPad Prism software was used to perform statistical analysis. Data are represented as mean ±  S.E.M. and mean ±  S.D. p-values were calculated by unpaired t test. All raw data underlying figures can be found in S1 Data and S2 Data.

S1 Fig. Expression analysis of slc13a5 zebrafish paralogs (5a and 5b), transcript and protein level detection in slc13a5 mutants and MB volume measurements.

(A, B) In situ hybridization for 5a and 5b expression using sense RNA (control) probes. Negligible expression of 5a and 5b indicates that control probes used are reliable. n = 3 for each probe. (C) qPCR analysis for relative 5a mRNA expression in 5 dpf 5a^−/−, 5b^−/− and 5a^−/−;5b^−/− larvae compared to WTs. 5a is downregulated in 5a^−/− and 5a^−/−;5b^−/− larvae compared to WTs, indicating active mRNA degradation of 5a^−/− transcripts. Unchanged 5a transcript levels in 5b^−/− larvae points to the absence of genetic compensation by paralog upregulation. WT, 5a^−/−, 5b^−/− and 5a^−/−;5b^−/−, n =  3 ×  10 larvae pooled. (D) qPCR analysis for relative 5b mRNA expression in 5 dpf 5a^−/−, 5b^−/− and 5a^−/−;5b^−/− compared to WTs. 5b is downregulated in 5b^−/− and 5a^−/−;5b^−/− larvae compared to WTs, indicating active mRNA degradation of 5b^−/− transcripts. Unchanged 5b transcript levels in 5a^−/− larvae points to the absence of genetic compensation by paralog upregulation. WT, 5a^−/−, 5b^−/− and 5a^−/−;5b^−/−, n =  3 ×  10 larvae pooled. (E) qPCR analysis for relative mRNA expression of elavl3 and gfap in FACS-sorted neurons obtained from the heads of 5 dpf WTs (Tg(elavl3:ubci-Cer-sv40)). elavl3 is upregulated compared to gfap, confirming the purity of the neuronal population. WT, n =  3 ×  100 heads pooled. (F and G) qPCR analysis for relative 5a and 5b mRNA expression in FACS-sorted neurons obtained from the heads of 5 dpf larvae (Tg(elavl3:ubci-Cer-sv40)). 5a and 5b transcript levels are significantly reduced in 5a^−/−;5b^−/− larvae compared to WTs, further validating that both the paralogs are highly expressed in the neurons compared to astrocytes. WT and 5a^−/−;5b^−/−, n =  3 ×  100 heads pooled. (H-J) 5 dpf WT and 5a^−/−;5b^−/− larvae; α-Slc13a5 (green). Quantification of the fluorescence intensity (CTCF) of Slc13a5 in the brain. 5a^−/−;5b^−/− larval brains show a significant reduction in the expression of Slc13a5 compared to WTs. WT, n = 6; 5a^−/−;5b^−/−, n = 6. (K) MB volume quantification at 5 dpf; α-Slc13a5. slc13a5 mutants show a significant reduction in the MB volume compared to WTs. WT, n = 7; 5a^−/−, n = 5; 5b^−/−, n = 5; 5a^−/−;5b^−/−, n = 5. Data are Mean ±  S.D., ns: no significant changes observed, * P ≤  0.05, **P ≤  0.01, ***P ≤  0.001- Unpaired t test. The data underlying this figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3002499.s001

(TIF)

S2 Fig. Analysis of startle response and thigmotaxis behavior in slc13a5 mutants.

(A) Quantification of distance traveled during complete acoustic startle protocol for 10 min at 5 dpf. slc13a5 mutants move higher distances compared to WTs. WT, n = 21; 5a^−/−, n = 8; 5b^−/−, n = 8; 5a^−/−;5b^−/−, n = 24. (B) Schematic representation of thigmotaxis protocol. (C) Quantification of distance traveled close to the wall in 10 min. No significant changes were observed in the distance traveled between slc13a5 mutants and WTs. WT, n = 8; 5a^−/−, n = 6; 5b^−/−, n = 5; 5a^−/−;5b^−/−, n = 7. (D) Quantification of time spent close to the wall in 10 min. No significant changes were observed in the time spent between slc13a5 mutants and WTs. WT, n = 8; 5a^−/−, n = 6; 5b^−/−, n = 5; 5a^−/−;5b^−/−, n = 7. (E) Quantification of distance traveled close to the wall in 10 min post PTZ exposure. slc13a5 mutants swam significantly higher distance hugging the wall compared to WT. WT, n = 18; 5a^−/−, n = 8; 5b^−/−, n = 10; 5a^−/−;5b^−/−, n = 18. (F) Quantification of time spent close to the wall in 10 min post PTZ exposure. slc13a5 mutants spent significantly more time hugging the wall compared to WT. WT, n = 18; 5a^−/−, n = 6; 5b^−/−, n = 10; 5a^−/−;5b^−/−, n = 16. Data are Mean ±  S.D., ns: no significant changes observed, * P ≤  0.05, **P ≤  0.01, ***P ≤  0.001- Unpaired t test. PTZ, pentylenetetrazol. The data underlying this figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3002499.s002

(TIF)

S3 Fig. Neuron population, cell proliferation and apoptosis analysis in slc13a5 mutants.

(A) 3 dpf WTs and slc13a5 mutants; α-HuC/D (green). (B) 28 hpf WTs and slc13a5 mutants; α-HuC/D (green), Hoechst (blue). (C) Quantification of HuC/D⁺ cells in the region that gives rise to optic tectum at 28 hpf. HuC/D⁺ cell numbers are unaffected in the slc13a5 mutants compared to WTs. WT, n = 7; 5a^−/−, n = 4; 5b^−/−, n = 4; 5a^−/−;5b^−/−, n = 5. (D-G) 3 dpf and 5 dpf WT and 5a^−/−;5b^−/− larvae; α-pHH3 (red), Hoechst (blue). Quantification of pHH3⁺ cells in the optic tectum. Cell proliferation is unchanged in the 5a^−/−;5b^−/− larvae compared to WTs. WT, n = 7; 5a^−/−;5b^−/−, n = 7 at 3 dpf. WT, n = 34; 5a^−/−;5b^−/−, n = 23 at 5 dpf. (H) 5 dpf WTs and slc13a5 mutants; α-HuC/D (green), CC3 (red). Data are Mean ±  S.D., ns: no significant changes observed- Unpaired t test. pHH3, Phosphohistone H3. The data underlying this figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3002499.s003

(TIF)

S4 Fig. Electrophysiological analysis in 6 dpf.

5a^−/− and 5b^−/− larvae. (A-E) Representative extracellular recordings obtained from optic tectum of 6 dpf 5a^−/− and 5b^−/− larvae, and pie charts of proportion of 5a^−/− and 5b^−/− larvae showing different patterns of activity. The repetitive inter-ictal like discharges (<1s duration) with above threshold (>0.2mV), high-frequency, large-amplitude spikes seen in slc13a5 mutants are indicative of increased network hyperexcitability. slc13a5^−/− larvae also displayed below threshold (<0.2mV) brain activity.

https://doi.org/10.1371/journal.pbio.3002499.s004

(TIF)

S5 Fig. Electrophysiological analysis in 5 dpf 5a^−/−;5b^−/− larvae.

(A-D) Representative extracellular recordings obtained using NeuroProbe from optic tectum of 5 dpf 5a^−/−;5b^−/− larvae, showing epileptic activity, consisting of a series of short bursts, with a total duration of> 10s and repetitive twitches (>5s) at consistent frequency (~1.8 Hz) as another indication of seizures.

https://doi.org/10.1371/journal.pbio.3002499.s005

(TIF)

S7 Fig. List of upregulated genes in 5a^−/− and 5a^−/−;5b^−/− larvae compared to WTs.

Heatmap showing a list of 400 genes significantly upregulated (padj < 0.05 and log2FC >  1.5) in 5a^−/− and 5a^−/−;5b^−/− larvae compared to WTs. The data underlying this figure can be found in S2 Data.

https://doi.org/10.1371/journal.pbio.3002499.s007

(TIFF)

S8 Fig. List of downregulated genes in 5a^−/− and 5a^−/−;5b^−/− larvae compared to WTs.

Heatmap showing a list of 380 genes significantly downregulated (padj < 0.05 and log2FC < −1.5) in 5a^−/− and 5a^−/−;5b^−/− larvae compared to WTs. The data underlying this figure can be found in S2 Data.

https://doi.org/10.1371/journal.pbio.3002499.s008

(TIFF)

S11 Fig. Calcium event analysis in slc13a5 mutants.

(A-F) Representative single neuron calcium traces illustrating that 5a^−/−;5b^−/− larvae exhibit above-threshold events in the MB. (G-I) Quantification of the amplitude and frequency of calcium events (ΔF/F > 2) and representative single neuron calcium traces at 3 dpf in the FB. 5a^−/−;5b^−/− larvae show significant increase in calcium events compared to WTs. WT, n = 6; 5a^−/−;5b^−/−, n = 7. (J-L) Quantification of the amplitude and frequency of calcium events (ΔF/F > 2) and representative single neuron calcium traces at 3 dpf in the HB. 5a^−/−;5b^−/− larvae show significant increase in the frequency of calcium events compared to WTs and the amplitude of calcium events remains unchanged. WT, n = 6; 5a^−/−;5b^−/−, n = 6. Data are Mean ±  S.E.M., ns: no significant changes observed, * P ≤  0.05, **P ≤  0.01- Unpaired t test. The data underlying this figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3002499.s011

(TIF)

S12 Fig. Assessment of the effect of suppressing NMDA receptor signaling and zinc treatment on calcium influx in WTs, and Grin1 expression after grin1a;grin1b MO injection.

(A-B) Quantification of the amplitude and frequency of calcium events (ΔF/F > 2) at 3 dpf in the MB. Memantine treatment and grin1a;grin1b MO injections have no significant effect on the calcium events in WTs. Vehicle-treated/control MO-injected WT, n = 5; memantine-treated WT, n = 5; grin1a;grin1b MO injected WT, n = 5. (C) Quantification of the fluorescence intensity (CTCF) of Grin1 (α-Grin1). grin1a;grin1b MO injections lead to significant reduction in the expression of Grin1 compared to control MO-injected WTs at 3 dpf. control MO-injected WT, n = 5; grin1a;grin1b MO-injected WT, n = 6. (D-E) Quantification of the amplitude and frequency of calcium events (ΔF/F > 2) at 3 dpf in the MB. ZnCl₂ treatment has no significant effect on the calcium events in WTs. Vehicle-treated WT, n = 4; ZnCl₂-treated WT, n = 4. Data are Mean ±  S.E.M. and Mean ±  S.D., ns: no significant changes observed, * P ≤  0.05- Unpaired t test. The data underlying this figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3002499.s012

(TIF)

S13 Fig. Effect of NMDA receptor antagonists and zinc treatment on the bioenergetics parameters of WTs and WT behavior under acoustic startle.

(A) Quantification of basal respiration at 6 dpf. Memantine and MK801 treatments do not affect basal respiration in WTs. Vehicle-treated WT, n = 6; memantine-treated WTs, n = 6; MK801-treated WTs, n = 6 (individual values plotted from five cycles). (B) Quantification of ATP-linked respiration at 6 dpf. Memantine and MK801 treatments do not affect ATP-linked respiration in WTs. Vehicle-treated WT, n = 5; memantine-treated WTs, n = 4; MK801-treated WTs, n = 6 (individual values plotted from three cycles). (C) Quantification of total mitochondrial respiration at 6 dpf. Memantine and MK801 treatments do not affect total mitochondrial respiration in WTs. Vehicle-treated WT, n = 10; memantine-treated WTs, n = 4; MK801-treated WTs, n = 4 (individual values plotted from three cycles). (D) Quantification of non-mitochondrial respiration at 6 dpf. Memantine and MK801 treatments do not affect non-mitochondrial respiration in WTs. Vehicle-treated WT, n = 8; memantine-treated WTs, n = 6; MK801-treated WTs, n = 6 (individual values plotted from three cycles). (E) Quantification of basal respiration at 6 dpf. ZnCl₂ treatment does not affect basal respiration in WTs. Vehicle-treated WT, n = 5; ZnCl₂-treated WTs, n = 6 (individual values plotted from five cycles). (F) Quantification of ATP-linked respiration at 6 dpf. ZnCl₂ treatment does not affect ATP-linked respiration in WTs. Vehicle-treated WT, n = 5; ZnCl₂-treated WTs, n = 6 (individual values plotted from three cycles). (G) Quantification of total mitochondrial respiration at 6 dpf. ZnCl₂ treatment does not affect total mitochondrial respiration in WTs. Vehicle-treated WT, n = 5; ZnCl₂-treated WTs, n = 6 (individual values plotted from three cycles). (H) Quantification of non-mitochondrial respiration at 6 dpf. ZnCl₂ treatment does not affect non-mitochondrial respiration in WTs. Vehicle-treated WT, n = 5; ZnCl₂-treated WTs, n = 6 (individual values plotted from three cycles). (I) Quantification of total distance traveled in 10 min in acoustic startle before and after treatment with memantine and MK801 at 5 dpf. Memantine treatment exhibited no effect on the behavior of WTs whereas MK801 treatment led to a significant reduction in the distance traveled by WTs. Vehicle-treated WT, n = 16; memantine-treated WT, n = 19; MK801-treated WT, n = 11. (J) Quantification of total distance traveled in 10 min in acoustic startle before and after treatment with ZnCl₂ at 5 dpf. ZnCl₂ treatment exhibited no effect on the behavior of WTs. Vehicle-treated WT, n = 9; ZnCl₂-treated WT, n = 6. Data are Mean ±  S.D., ns: no significant changes observed, ***P ≤  0.001- Unpaired t test. The data underlying this figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3002499.s013

(TIF)