Trehalose Treatment in Zebrafish Model of Lafora Disease

Mutations in the EPM2A gene encoding laforin cause Lafora disease (LD), a progressive myoclonic epilepsy characterized by drug-resistant seizures and progressive neurological impairment. To date, rodents are the only available models for studying LD; however, their use for drug screening is limited by regulatory restrictions and high breeding costs. To investigate the role of laforin loss of function in early neurodevelopment, and to screen for possible new compounds for treating the disorder, we developed a zebrafish model of LD. Our results showed the epm2a−/− zebrafish to be a faithful model of LD, exhibiting the main disease features, namely motor impairment and neuronal hyperexcitability with spontaneous seizures. The model also showed increased inflammatory response and apoptotic death, as well as an altered autophagy pathway that occurs early in development and likely contributes to the disease progression. Early administration of trehalose was found to be effective for rescuing motor impairment and neuronal hyperexcitability associated with seizures. Our study adds a new tool for investigating LD and might help to identify new treatment opportunities.


Introduction
Lafora disease (LD, OMIM #254780) is a rare autosomal recessive neurodegenerative disorder with a worldwide prevalence of almost four cases per million [1]. LD is associated with mutations in three genes: EPM2A [2], encoding laforin, EPM2B [3], encoding malin, and PRMD8 [4], which codes for a protein responsible for the translocation of laforin and malin to the nucleus. However, diagnostic confirmation shows mutations in the EPM2A or EPM2B gene in more than 95% of diagnosed patients [5,6].
LD manifests between eight and 19 years of age as a form of progressive myoclonic epilepsy (PME) [1] characterized by multiple types of seizures, including tonic-clonic seizures, drop attacks, and action-sensitive and stimulus-sensitive myoclonus [1,7]. Progressive worsening of the seizures typically leads to severe, drug-resistant epilepsy, with possible episodes of status epilepticus [1,5]. In LD patients, EEG recordings typically show slowing of the background activity, diffuse or predominantly posterior paroxysmal abnormalities, and photosensitivity [1,7,8]. Neurological deterioration leads to progressive neuropsychiatric and motor impairments that include behavioral disturbances, cognitive

Regional and Developmental Expression of the epm2a Gene
We investigated the regional and developmental expression of epm2a in zebrafish. Developmental expression of the epm2a transcript was assessed by qRT-PCR analysis at a series of developmental stages-0 h post-fertilization (hpf), six hpf, 24 hpf, 48 hpf, 72 hpf and five days post-fertilization (dpf)-in WT zebrafish larvae. We observed high levels of epm2a expression at 0 hpf that decreased over the next few hours before increasing gradually from 48 hpf ( Figure 1A). A similar trend was observed in mouse brains, where levels of Epm2a transcript increased gradually after birth, reaching the highest level in adults [44]. Regional expression of epm2a tested in WT zebrafish embryos at 24 hpf showed laforin to be distributed almost ubiquitously within the zebrafish central nervous system ( Figure 1B).

Generation of epm2a −/− Mutant Zebrafish
The zebrafish epm2a gene (ENSDARG00000059044), which maps to chromosome 23, consists of five coding exons, and encodes laforin, a 310-amino acid protein. Performing a bioinformatics analysis of the amino acid sequence of Danio rerio and of Homo sapiens, we observed high conservation of the zebrafish protein compared with its human counterpart (75% homology and 62% identity) (Supplementary Figure S1). By means of CRISPR/Cas9 editing, we engineered zebrafish epm2a-null mutants in which a 5-bp insertion mutation in exon 2 of the epm2a gene (Figure 2A) led to a frameshift mutation and premature stop codon at residue 74 (p.Thr54Asnfs*74) ( Figure 2B). The homozygous epm2a −/− F2 zebrafish line was bred to adulthood, and its progeny were used to study the phenotype of the mutant strain, and to determine whether it replicates the characteristics of LD. Given the lack of a reliable antibody against zebrafish laforin, it was not possible to test protein abundance in the mutant zebrafish strain by western blotting. However, qRT-PCR analysis showed a significant decrease in the amount of epm2a transcript in mutants compared with WT siblings at 5 dpf ( Figure 2C). Homozygous epm2a −/− larvae showed grossly normal morphology ( Figure 2D) but were significantly different in length from their control siblings (n = 83) at 4 dpf. The red rectangle shows the premature stop codon at residue 74 (p.Thr54Asnfs*74) caused by the 5-bp insertion mutation in exon 2. (C) qRT-PCR analysis revealed a decrease in the level of epm2a mRNA expression, normalized to the expression of β-actin mRNA. Three independent RNA samples (each obtained from about 30-40 larvae) from epm2a −/− larvae at 120 hpf and from controls were analyzed. The values are expressed as mean ± standard deviation (SD). **, p ≤ 0.01, calculated by Student's t-test. (D) Lateral view photographs of representative control and epm2a −/− specimens. At 4 dpf we observed a significant reduction in body length in epm2a −/− larvae (n = 83) compared to WT controls (n = 83), **** p ≤ 0.0001, calculated by Mann-Whitney test.

Glycogen Accumulation and Lafora Body Formation
The hallmark of LD is the accumulation of glycogen and subsequent formation of LBs in several tissues, including the nervous tissue. To assess whether these features are already present in the early stages of the disease, we first tested glycogen accumulation in whole zebrafish larvae at 5 dpf, observing significantly increased glycogen concentration (measured as ug/mL) in epm2a −/− mutants (n = 30) compared with WT larvae (n = 30) (p < 0.01) ( Figure 3A). Then, by means of PAS staining of epm2a −/− mutants and WT controls, we looked for LBs in zebrafish larvae at 5 dpf. Histological examination with PAS staining did not detect LBs in the brains of the epm2a −/− mutants; however, a slight increase in positive PAS staining was observed in mutants compared with WT larvae at 5 dpf ( Figure 3B).

Mutant Zebrafish Displays Neuronal Hyperexcitability and Spontaneous Seizures
Forebrain local field potential (LFP) recordings at five dpf showed a higher frequency of high-power electrophysiological events in epm2a −/− larvae compared with WT siblings. In particular, the duration and power (i.e., RMS power of the signal measured in the 30-95 Hz band, see Section 4) of high-power electrophysiological events were statistically significantly increased in epm2a −/− larvae compared to WT siblings, likely accounting for their spontaneous seizures ( Figure 4A). On further analysis, the high-power events observed in the epm2a −/− recordings were found to differ in morphology compared with the activity recorded in controls ( Figure 4B). Both experimental groups exhibited frequent events characterized by low duration and low power, likely associated with normal brain activity. However, only the mutants showed the seizure-like events associated with multiple high-power, long-duration (>1 s) discharges ( Figure 4C,D), as well as smaller events, sometimes associated with interictal-like discharges. The presence of these spontaneous high-power events is an important pathological sign, given that the larvae were recorded in normal conditions and were not challenged with any pro-epileptic stimuli. These events were never observed in the control group. Furthermore, these data were supported by behavioral analysis at 30 hpf, which showed an increase in spontaneous burst activity in knockout embryos compared with WT controls (p < 0.0001) ( Figure 4E).

Motor Impairment in epm2a −/− Mutant Zebrafish
Whilst their increased burst activity at 30 hpf likely indicates a higher frequency of "seizures", epm2a −/− larvae, compared with WT controls, showed significantly reduced motor activity (slowing and reduced distance traveled) at five dpf (p < 0.0001) ( Figure 5). To rule out morphological alteration of spinal motor neurons as the cause of these events, we used a motor axon marker, synaptotagmin-2, to label and study the motor axons; however, no alterations in arborization or length were seen in epm2a −/− larvae compared with WT controls (Supplementary Figure S2).    As well as glycogen accumulation, epm2a −/− larvae also showed changes in other physiological pathways. On studying apoptosis through an in vivo acridine orange (AO) assay, we found an increased number of apoptotic cells in epm2a −/− embryos compared with WT controls at 24 hpf ( Figure 6A). Oxygen consumption rate measurements did not reveal any major impairment of mitochondrial respiratory chain function in epm2a −/− compared with control larvae at the same stage of development (five dpf), and baseline respiration and ATP production were normal (Supplementary Figure S3). However, epm2a −/− larvae showed a slight reduction in maximal respiratory capacity compared with controls ( Figure 6B). Furthermore, epm2a −/− larvae at five dpf showed greater activation of the inflammatory response compared with WT controls as shown by altered transcript levels of all the proinflammatory cytokines studied in qRT-PCR analysis, namely, cox2b (p < 0.0119), il1b (p < 0.0018), and tnfα (p < 0.0001) ( Figure 6C). However, upregulation of anti-inflammatory cytokine il10 (p < 0.0258) was also observed ( Figure 6C). Analysis of genes specifically expressed in microglia showed upregulation of p2ry12 (p < 0.0101), hex-b (p < 0.0260) ( Figure 6D), and csf1ra (p < 0.0052) ( Figure 6D). Analysis of genes specifically expressed in astrocytes showed the upregulation of gfap (p < 0.0439) and kir4.1b (p < 0.0325) ( Figure 6E) in laforin-deficient larvae compared with WT controls at five dpf.
Since we found an autophagy impairment in mutant larvae at five dpf, we decided to test whether administration of trehalose, from the early stages of embryonic development (within four hpf) up to five dpf, might at least partially reverse the pathological phenotype of epm2a −/− larvae. Trehalose is known for its neuroprotective effects [51] and for its capacity to induce autophagy flow [52], although the underlying mechanisms are only partially known [51,53]. Trehalose added to the egg water to reach a final volume of 100 uM was used to treat four hpf larvae. Locomotor behavior ( Figure 8A) improved in treated epm2a −/− larvae compared with untreated mutants, both in terms of distance traveled (p ≤ 0.0001) and speed (p ≤ 0.0001), suggesting rescue of the motor impairment. The analysis of neuronal hyperexcitability, associated with burst activity, and electroencephalographic activity reached similar results. Indeed, the analysis of 30 hpf tail flick data ( Figure 8B) revealed attenuation and reduction of burst activity in trehalose-treated epm2a KO embryos (p ≤ 0.0001), whereas LFP recordings from five dpf larvae ( Figure 8C) showed reduced duration and power of seizure-like events in trehalose-treated epm2a −/− larvae compared with untreated mutants. Overall, our results indicate that trehalose administered in the early stages of development can reverse the pathophysiological phenotype of epm2a −/− larvae. in epm2a −/− mutant larvae treated (n = 213) and untreated with trehalose (n = 228). Statistical analysis (**** p ≤ 0.0001) was performed using the Kruskal-Wallis test. Dunn's test was used to perform post-hoc analysis for multiple comparisons after the Kruskal-Wallis test. (B) Tail flick analysis performed at 30 hpf in controls (WT) treated (n = 91) and untreated with trehalose (n = 87) as well as in epm2a −/− mutant larvae treated (n = 106) and untreated with trehalose (n = 111). Statistical analysis (**** p ≤ 0.0001) was performed using the Kruskal-Wallis test. Dunn's test was used to perform post-hoc analysis for multiple comparisons after the Kruskal-Wallis test. ns, not significant (C) Violin plot of the effect of trehalose on the power and duration of detected events, 11 larvae per group were analyzed at five dpf. Statistical analysis (**** p ≤ 0.0001) was performed using the Mann-Whitney test.

Discussion
Other than palliative care and the repurposing of a few drugs, current therapeutic approaches in LD include antibody-enzyme fusion (VAL-0417) to degrade LBs [54][55][56], strategies to reduce glycogen synthase [19,57], and new vectors for gene therapy, such as the cationic lipoplexes containing plasmid DNA [58], all developed in vitro or in murine models, and not yet translated into clinical trials. Even in more common monogenic neurodegenerative diseases, there is a growing trend to combine "repair approaches" with therapy based on "downstream approaches" where the aim is not to repair or substitute the defective gene, but rather to focus on experimentally accessible targets in the disease pathogenesis. Thus, identification, with a view to their repurposing, of EMA-/FDAapproved drugs or small molecules or nutrients that act downstream of the gene mutation might accelerate trial readiness in LD. In order to repurpose approved drugs and identify new small molecules for LD, it is necessary to have rapid and efficient screening tools available. In this perspective, we think the zebrafish model of LD could be a valuable and useful aid in drug research.
Loss of laforin function in zebrafish was found to recapitulate the human disease, showing locomotor impairment and neuronal hyperexcitability with spontaneous recurrent seizures (i.e., epilepsy). Indeed, in epm2a −/− larvae we observed the variability of LFP recordings, identifying spontaneous electrophysiological events and both interictal and seizure-like discharges. This may to an extent reflect what occurs in clinical practice in patients with LD in whom, especially initially, the frequency of the seizures may be relatively low. The course of the disease is progressive and characterized by the increasing frequency and intractability of seizures [59]. The ability of the new zebrafish model to mimic human disease in terms of electrophysiological events makes it an ideal tool for investigating drug response and new opportunities to treat LD.
Furthermore, the epm2a −/− larvae showed increased glycogen concentration, similar to what is seen in laforin-deficient mice [12], but without evidence of LBs. The absence of LBs has also been seen in epm2a KO mice, where polyglucosan aggregation occurs on average at approximately two months of age (or at one month in some LD mouse models) [23], and gradually becomes widespread [60][61][62]. Our data recapitulate major characteristics of the Epm2a KO mouse, particularly in its earlier stages of development.
In the zebrafish model of LD, we observed increased apoptotic cell death in mutant larvae compared with controls, in agreement with previous studies showing increased susceptibility to apoptosis of laforin-deficient cells [29], probably caused by the induction of endoplasmic reticulum stress linked to proteasome dysfunction [63]. In addition, laforin appears to have a possible anti-apoptotic function [64]. It is possible that increased susceptibility to apoptosis contributes to neurodegeneration in LD. This apoptotic death occurs in zebrafish larvae independently of the formation of LBs. "Dark cell death", a unique form of non-apoptotic cell death that precedes the formation of LBs, has been described in other models of LD [60,65]; indeed, many degenerated neurons do not show visible LBs [60,65]. These data suggest multiple causes of neuronal death in LD.
We did not observe any major impairment of mitochondrial bioenergetics, although epm2a −/− larvae at 5 dpf seemed to show a weak response to stress associated with increased energy demand when more ATP is required to maintain cellular function. In contrast, increased oxidative stress is a feature of different models of LD [22], and could be related to the alteration of intracellular proteolytic systems-and thus of mitophagy [32]-but also to mitochondrial dysfunction [22] and to the reduction of enzymes involved in ROS detoxification [22]. We cannot exclude that oxidative stress could become evident at later developmental stages in epm2a −/− zebrafish larvae and contribute to cell death.
As seen in KO mice [23,24], we observed the activation of inflammatory responses in epm2a −/− larvae. Analysis of the inflammatory genes in our model suggests that microglial populations with different phenotypes may coexist in the early stages of the disease. The upregulation of pro-inflammatory cytokines genes (tnfα, cox2b and il1b) and of csfr1a suggests a pro-inflammatory activation of microglia, leading to a detrimental phenotype, sometimes called disease-associated microglia (DAM), which is described in several neurodegenerative diseases [66,67]. CSF1R has been found to be increased in disease-associated microglia in the Alzheimer's disease brain [68]. Upregulation of the CSF1/CSF1R pathway may, via a positive feedback mechanism, help to support the DAM signature [66,69]. The early activation of neuroinflammation observed in the pathophysiology of our laforin-deficient zebrafish model recalls other neurodegenerative disorders characterized by intracerebral accumulations [70] and suggests new targets for alleviating symptoms in preclinical studies. On the other hand, the upregulation of anti-inflammatory cytokines genes (e.g., il10) suggests that some microglial cells, at least, have an anti-inflammatory phenotype [71]. We also observed the upregulation of the purinergic receptor gene (e.g., p2ry12), whose density is modified depending on the state of microglial activation [72]; the first step of microglial activation in response to chemo-attractive signals released by damaged cells [73,74] is stimulation of P2RY12, subsequently followed by its downregulation [75]. We analyzed the homeostatic gene hexb, a lysosomal hydrolase regulating the metabolism of gangliosides and involved in autophagy [76], whose expression does not generally change in the transition between homeostatic and harmful microglia [77,78]. The hexb upregulation we observed in laforin-deficient larvae supports the possibility that dysregulation of lysosomal function and autophagy in microglia may contribute to the neuropathology of LD [16][17][18]21,30,31,45,46]. These results are also in line with those described in microglia from a mouse model of Alzheimer's disease [79]. Expression analysis of some astrocytic markers showed upregulation of gfap and kir4.1b genes in laforin-deficient larvae at five dpf, suggesting a role for these cells in the epm2a −/− zebrafish model; early astrocytic activation has also been observed in mouse models of the disease [23,24,80]. Moreover, we previously observed the upregulation of kir4.1b, which is associated with autism-epilepsy phenotypes in humans [81], and we cannot exclude its possible participation in the epileptic phenotype observed in epm2a −/− . Indeed, both upregulation and downregulation of Kir4.1 seem likely to produce neuronal hyperexcitability [82,83]. It must be underlined that zebrafish possess no star-shaped glial cells, but rather radial glia [84,85] whose functions have yet to be precisely defined. It seems, however, that these glial cells carry out many of the functions of classical mammalian astroglia, such as neurogenesis and homeostatic roles in neural circuits and brain barriers [85]. The involvement of this cell type has also been described in other neurodegenerative diseases modelled in zebrafish, in which astrogliosis [86] or induction of neuronal regeneration [87] has been observed. However, the possible role of glial cells in LD and their role in teleost fish compared with mammals remain to be studied in more detail.
Given the evidence pointing to primary or secondary involvement of the autophagy pathway in LD [16][17][18]21,30,31,45,46], our study focused on the upregulation of several autophagy genes and on increased protein levels of ATG5, LC3-II and the LC3-II/LC3-I ratio from early development. These results could indicate a compensatory increase in autophagic flux [88], similar to what has been observed in Batten disease [89], another form of PME. This compensation could explain in part the absence of LBs at this stage of the disease, with ensuing loss of autophagic compensation as the disease progresses and LBs become evident. As an alternative hypothesis, there might occur a decreased flux and the subsequent accumulation of LC3 due to a block in fusion or degradation, as documented by increased LC3-II [88] and increased LC3-II/LC3-I ratio [90], followed by a compensatory increase upstream. More robust studies in the new model of LD are warranted.
Early dietary supplementation with the autophagy flux modulator trehalose ameliorated PTZ-induced seizure susceptibility in laforin-deficient mice [38] without acting on LBs. In zebrafish we corroborated these data: Early administration of trehalose reduced neuronal excitability and the frequency of spontaneous seizures and rescued the motor impairment, offering further support to its clinical use in children with LD. The mechanism by which trehalose induces autophagy is still unknown, but it appears to act as an inducer of autophagy flux and lysosomal biogenesis by activating TFEB [52]. In our laforin-deficient larvae we observed high levels of tfeb mRNA, which has also been documented in other neurodegener-ative diseases [91]. The increased tfeb expression could be a sign of a possible compensatory mechanism activated to overcome the impaired lysosomal function [92] through autophagy induction. It is also possible that TFEB is sequestered in the cytosol in the inactive form and is therefore unable to act at the nuclear level to stimulate autophagy genes. In fact, our model also showed the upregulation of mtor, which phosphorylates TFEB and blocks it in the cytosol in the inactive form [93]. Our hypothesis is that trehalose may act by activating TFEB and promoting its translocation to the nucleus [52]. Trehalose also appears to act on autophagy in an mTOR-independent manner [53], favoring its chaperone function [53], preventing inflammatory responses [94], and inducing de novo autophagosome formation [95]. Given the good safety profile of trehalose in children and adults, and the multiple ongoing clinical trials of the treatment (ClinicalTrials.gov; Available online: https: //clinicaltrials.gov/ct2/results?cond=trehalose&term=&cntry=&state=&city=&dist=; accessed 3 June 2022), it is tempting to anticipate its use in LD patients before too long.
In conclusion, the laforin-deficient zebrafish is a valuable new model for the study of LD, recapitulating the main features described in the mouse models and patients (Supplementary Table S1), providing a unique platform for studying the early stages of the disease and for more robust and cheap drug screening. Further research will be needed on adult fish, since the nervous system of larvae is incomplete and can limit the translation of complex neurological disorders. However, we believe that epm2a −/− larvae will likely help to speed up the drug discovery studies in vivo ahead of more expensive studies in mice.
Adults were housed in tanks at a density of no more than five zebrafish per liter at a constant temperature of 28 • C with a 14-h light/10-h dark cycle. Zebrafish eggs and embryos were collected and grown at 28.5 • C in egg water (60 µg/mL of "Instant Ocean", Sea Salts, (Aquarium Systems, Sarrebourg, France)) and E3 medium [292.2 mg NaCl, 12.6 mg KCl, 48.6 g mM CaCl2 2H2O, and 39.8 mg MgSO4, per 1 L of deionized water] respectively using established procedures [96] and staged in either hours post fertilization (hpf) or days post fertilization (dpf). All compounds used for E3 medium solution preparation were purchased from Sigma (St. Louis, MO, USA). The generation of the mutant, using CRISPR/cas9 technology, was performed with the ethical approval of the Italian Ministry of Health (approval n • 258/2021-PR) in accordance with the European Union Directive 2010/63/EU on the protection of animals used for scientific purposes, under the supervision of the Institutional Animal Care and Use Committee of the University of Pisa, and in compliance with the 3R principles.

Establishing Mutant Lines
The selected sgRNA was chosen among the top targets identified by CHOPCHOP software (CHOPCHOP. Available online: www.chochop.rc.fas.harvard.edu/index.php; accessed 8 January 2022) set with NGG PAM sites and zero predicted off-targets (fewer than three mismatches in the epm2a-targeting 20-mer). The sgRNA was designed against exon two of the epm2a transcript (ENSDARG00000059044) (Supplementary Table S2) and generated as already described [97]. The sgRNA was transcribed using the Megascript T7 Transcription Kit (Invitrogen, Heidelberg, Germany). The optimized Cas9 mRNA, for genome editing in zebrafish, was transcribed from a linearized template plasmid pCS2-nCas9n using the mMESSAGE mMACHINE™ SP6 Transcription Kit (ThermoFisher Scientific, Waltham, MA, USA). RNA concentration was quantified using a NanoDrop spectrophotometer (Optosky, Xiamen, China) and diluted to 500 ng/µL. About 100 ng of epm2a-sgRNA and 500 ng of Cas9 mRNA were co-injected into 1-cell stage embryos to ensure high-efficiency delivery of the injected mRNA to the embryo. The volume of solution injected was~1 nL. At least three independent injection experiments were performed with spawns from different founder fish to control for the batch effect. CRISPR/Cas 9 was into wild-type (WT) AB genotype.

Genotyping
For mutation screening, sgRNA-injected F0 embryos, raised to adulthood and outcrossed with the Tg(neurod1:GCaMP6F) strain in the nacre (mitfa −/+ ) background (kindly donated by Claire Wyart from Institut du Cerveau et de la Moelle Épinière, Paris, France) [98], were used for the study to obtain F1 heterozygous embryos. Adults potentially carrying mutations were identified by PCR and fragment analysis using genomic DNA from 16 randomly selected F1 embryos. The sequence of the primers used is listed in Supplementary  Table S2. F1 heterozygous fish carrying a 5-bp insertion mutation in the targeted site were selected and inter-crossed to generate the F2 homozygous epm2a −/− line.

In Situ Hybridization
Whole-mount in situ hybridization was performed in 24 hpf WT embryos as described previously [99]. The sequences of the primers used can be found in Supplementary Table S3.

Analysis of Larval Morphology
Live zebrafish were mounted on glass depression slides with 1% low-melting agarose. Images were obtained using a Leica M205FA stereomicroscope (Leica Microsystem, Wetzlar, Germany). Zebrafish body length was measured using Danioscope software (Noldus Information Technologies, Wageningen, The Netherlands) [100].

Quantitative (q)RT-PCR
Gene expression levels of epm2a mRNA were examined using RNA pooled from WT and epm2a −/− larvae. Total RNA was extracted from 30 larvae at 120 hpf using the Quick RNA Miniprep Kit (Zymo Research, Irvine, CA, USA) according to the manufacturer's instructions and quantified with a NanoDrop™ ND-1000 spectrophotometer (Thermo Scientific, Waltham, MA, USA). Extraction of cDNA and qRT-PCR were performed as described elsewhere [101]. Relative expression levels of each gene were calculated using the 2 −∆∆CT method [102]. The results obtained in at least three independent experiments were normalized to the expression of the housekeeping gene, β-actin (ENSDARG00000037746). The expression analysis of epm2a mRNA in mutant larvae was calculated setting the mean of the controls at one, and the p-value was calculated using GraphPad Prism 6 software (GraphPad Software, Inc., San Diego, CA, USA). The primers used are listed in Supplementary Table S4.

Glycogen Assay
The amount of glycogen was determined in dechorionated zebrafish larvae (minimum 10 per experiment) at five dpf, using the Glycogen Assay Kit II (Abcam, Cambridge, UK) according to the manufacturer's instructions and as previously described [105].

Histological Preparation for Periodic Acid-Schiff (PAS) Staining
Zebrafish larvae at 120 hpf were sacrificed by overdose of anesthesia (0.25 mg/mL, MS-222, Sigma-Aldrich, St. Louis, MO, USA) and fixed in cold 10% neutral buffered formalin at 4 • C overnight. Prior to processing, to optimize sectioning, the larvae were embedded in 1% agarose blocks, as previously described [106]. Tissue processing of agarose blocks was then performed in a controlled automatic processor (Shandon TP 1020, Leica Microsystem), followed by paraffin embedding. Paraffin blocks were sliced into 5 µm sections and stained with periodic acid-Schiff (PAS) stain under standard protocol. Sections were then examined using a light microscope (Eclipse 80i, Nikon, Tokyo, Japan).

Detection of Apoptotic Cells
To visualize apoptotic cells, in vivo staining was carried out in live, PTU-treated, manually-dechorionated larvae at 24 hpf using the vital dye acridine orange (AO, #235474; Sigma-Aldrich). Zebrafish embryos were incubated with 10 µg/mL AO solution for 15 min in the dark; the larvae were then washed three times with E3 medium. We counted AOpositive cells within a pre-defined area and a quantitative analysis was performed as described elsewhere [100,108].

Mitochondrial Respiratory Analysis
Mitochondrial respiration was analyzed in control (WT) and homozygous epm2a −/− larvae at 120 hpf using the XF24 extracellular flux analyzer (Seahorse Bioscience, North Billerica, MA, USA). The dual analyte sensor cartridges were soaked in an XF calibrator solution (Seahorse Bioscience) and placed in 24-well cell culture microplates overnight at 28 • C to hydrate. About 30 min before the trial period, the injection ports on the sensor cartridge were filled with oligomycin (0.6 µmol/L), FCCP (6 µmol/L), and rotenone (5 µM) plus antimycin A (0.182 µmol/L). The 120 hpf larvae were staged and placed in 20 of the 24 wells of an islet microplate. The islet plate acquisition screens were placed on the measurement area to hold the larvae in place. Four wells were left empty as a control. Each well was filled with 500 µL of E3 medium (pH 7.4). A standard approach was used to measure basal respiration, ATP production, maximal respiration rate, and spare respiratory capacity [109].

Locomotor Behavior
Locomotor behavior (distance traveled and velocity) was measured in homozygous epm2a −/− F2 larvae and WT at five dpf using the Daniovision device (Noldus Information Technology, Wageningen, The Netherlands). Briefly, single larvae were transferred into 96-well plates containing 300 µL of E3 medium per well. Then, one plate at a time was placed in the DanioVision system, and larval locomotor activity was recorded for 30 min and analyzed using EthoVision XT software (Noldus Information Technology, Wageningen, The Netherlands). Coiling behavior was measured to assess burst activity in homozygous epm2a −/− and WT embryos at 30 hpf; to this end, the number of tail flicks was measured in 30-s time frames using Danioscope software (Noldus Information Technology, Wageningen, The Netherlands).

LFP Recordings and Analysis
Electrophysiological forebrain recordings were performed in 120 hpf larval zebrafish to characterize the epilepsy phenotype. The larvae were placed on a drop of 1.2% low-melting point agarose and LFPs were recorded with an AgCl electrode inside a glass micropipette backloaded with 2M NaCl. Electrophysiological signals were amplified 500-fold (EXT-02F, NPI, Tamm, Germany), band pass filtered (0.3-1300 Hz), and digitized at a rate of 5 kHz (Axon Digidata 1550B, Clampex 10.7.0.3, Molecular Devices, Berkeley, CA, USA). The microelectrode was positioned by advancing its tip until it punctured the skin, and then it was carefully advanced into the forebrain [110]. Data analyses of the LFP recordings were performed using a suite of custom Matlab scripts. Burst activity analysis was performed on the LFP signals filtered in the 30-95 Hz band as already described [111]. The root mean square (RMS) power of the signal was computed on the recording on rolling windows of 250 ms and in steps of 50 ms. The distribution of the logarithm of the RMS power measurements was fitted with a single gaussian distribution. In some recordings, the distribution was characterized by an asymmetry due to a high-energy population. Those distributions were fitted with double gaussians56, then the ROC curves and their area under the curve (AUC) were calculated; values > 0.9 were considered acceptable and events were extracted as described in Supplementary Figure S5. For normally distributed recordings, bursts were identified by extracting all events with a power higher than three standard deviations (SD) from the mean. Violin plots were produced using the online-available function 'violin.m' [112]. Statistical analysis was performed using GraphPad Prism version 9 software. (GraphPad Software, San Diego, CA, USA).

Pharmacological Treatments
Fifty mutant and fifty WT 4 hpf embryos were randomly placed in 60 mm × 15 mm Petri dishes containing trehalose diluted in egg water. Trehalose stock solutions were prepared in Milli-Q water (Merck-Millipore, Milan, Italy) and diluted in egg water to the final administered concentrations. The 100 µM concentration was chosen as the working dilution. To test treatment efficacy, we performed tail coiling and locomotor behavioral assays and LFP recordings.

Statistics
All data in the manuscript represent three or more independent experiments. Statistical analysis was performed using GraphPad Prism 6. All quantitative variables were analyzed applying either parametric or non-parametric methods, depending on the distribution shown by the Shapiro-Wilk test. For multiple comparisons (post-hoc analysis), Dunn's test was performed after the Kruskall-Wallis test, since the data of the various groups examined did not follow the gaussian distribution. Statistical significance is reported as: * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001, or **** p ≤ 0.0001.