Transient anticonvulsant effects of time-restricted feeding in the 6-Hz mouse model

,


Introduction
Epilepsy affects approximately 1 % of the global population with an estimated economic burden of $119 billion each year [1,2].Despite there being many anti-seizure medications available, they remain ineffective in about 30 % of individuals and can give rise to considerable side effects that can reduce patients' health-related quality of life [3][4][5].In both humans with epilepsy and rodent epilepsy models, impaired brain glucose metabolism has been observed [6][7][8][9][10][11].In the fed state, glucose is the predominant energy source of the brain generating ATP, which is largely used to maintain ion homeostasis and regulate cellular excitability [12,13].It is conceivable that a shortage of ATP production can result in deficiencies of Na + /K + -ATPase activity leading to ion imbalances, increased neuronal excitability, and increased seizure susceptibility [14].In animal models, there is evidence that impaired energy metabolism in epileptogenic brain areas is a result of decreased Abbreviations: ADF, Alternate-day fasting; MEST, maximal electroshock seizure threshold; TRF, time-restricted feeding.activity of mitochondrial enzymes and reduced amounts of tricarboxylic acid (TCA) cycle intermediates (Fig. 1) [7].Furthermore, reactive oxygen species (ROS) derived from increased oxidative stress in the epileptic brain can impair cellular energy production and promote neuroinflammatory pathways leading to increased neuronal excitability [15][16][17].Therefore, therapies targeting brain glucose metabolism and oxidative stress in the brain are warranted [8,9,16,18].
Dietary therapies can provide an alternative and/or additional approach to current medications.Changes to food composition and timing can alter brain fuel sources and modulate key regulatory enzyme activity to improve metabolic impairments and reduce oxidative stress in the brain that is associated with epilepsy [8,11,17,19].For example, anticonvulsant effects have been found with the ketogenic diet for certain paediatric [20,21] and adult epilepsies [22], although full adherence is difficult and side effects such as gastrointestinal-related disturbances and hunger are common [21,22].Other therapies such as medium chain triglyceride supplementation have demonstrated protection in rodent seizure models, some dogs, and people with epilepsy [23], and a new dietary supplement containing decanoate-rich medium chain triglycerides is now available in the UK [24].Intermittent fasting is another potential approach for epilepsy management that is less strict than ketogenic diets and inherently easier to follow than other dietary regimens as it requires only changes to the times of when food is consumed, not diet composition.Although not fully understood, intermittent fasting has been found to promote a vast array of whole-body physiological and biochemical changes that may be beneficial in the context of epilepsy.These effects can promote metabolic and transcriptional changes leading to increased neural bioenergetics, improved antioxidant capacity, and decreased excitotoxicity which may collectively reduce neuronal excitability and confer seizure protection [25][26][27].Lowered and stable blood glucose levels can contribute to antiseizure properties of dietary therapies and alterations in gene transcription and neuronal activity have been found to play a role [6,28].
The two most common intermittent fasting variations studied in rodents and people are time-restricted feeding (TRF) and alternate-day fasting (ADF) [29].Compared to 24 h ad libitum feeding, Wistar rats subjected to TRF fed 2 h daily showed increased latency and reduced severity of pilocarpine-induced status epilepticus [30].ADF was also reported to provide some protection against repetitive pentylenetetrazole injections (kindling) [31] and attenuated kainate-induced status epilepticus [32,33] and excitotoxicity in Wistar rats [32][33][34][35].In mouse models, ADF increased kainate-induced seizure latency and reduced seizure severity in NIH Swiss mice [36], but was proconvulsant in the 6-Hz and maximal electrical shock threshold (MEST) tests and had no effect on pentylenetetrazole-induced seizures [36].When kainate was injected intrahippocampally in C57BL/6 mice, Anson et al. reported no change in seizure severity and duration after ADF, but hippocampal cell death was reduced 24 h later [37].Another paper from the same group showed that the increased resistance to kainate-induced hippocampal neuronal damage provided by ADF was mediated by increased levels of brain-derived neurotrophic factor (BDNF) [38].
Given the mixed effects of intermittent fasting regimens reported in rodent models, further investigation into this therapy is required.Although ADF is commonly used in animal models, the 24 h periods of fasting in between food intake mimics starvation in mice, equivalent to fasting for multiple days in humans due to their seven-fold higher basal metabolic rate [39,40].The severe weight loss associated with ADF in rodents (over 20 %) also renders this approach unviable for long-term epilepsy therapy.Other studies investigating neurological diseases have used a less intensive, 8 h TRF regimen in rodents with feeding during the light phase [41][42][43][44] that may be regarded as a more appropriate way of intermittent fasting when considering translation to humans.Our study is the first to investigate the effects of this restricted feeding regimen for seizure prevention.We aimed to evaluate the impact of one month of uninterrupted TRF on seizure susceptibility in acute electrical seizure models, in which ADF was previously found to decrease seizure thresholds [36].Additionally, a 2:5 TRF regimen (5 days TRF per week) was investigated in a separate cohort to elucidate the effectiveness and robustness of TRF with ad libitum feeding during the weekend.After one month of the dietary regimens, we assessed metabolic and antioxidant alterations in the cortex, hippocampus, liver, and plasma that may contribute to protection against seizures or neuronal damage.

Materials
All reagents were bought from Sigma-Aldrich Pty Ltd (Sydney, NSW, Australia) unless stated otherwise.

Study design, animals, and diet regimens
The ARRIVE guidelines [45] for conducting and reporting animal research were followed.Our study was a randomised blinded study, designed with the main aim to investigate whether 8 h TRF alters seizure thresholds in electrical seizure tests compared to ad libitum-fed control.The use of 15-16 mice per treatment group was determined based on power analysis (GraphPad Statmate 2.0).This was calculated from variability found previously in seizure and blood glucose tests with an average standard deviation of 20 %, allowing for the detection of 20 % differences with 80 % power at alpha 0.05.The secondary aim was to investigate changes in brain protein and mRNA expression, as well as antioxidant and metabolic enzyme activities which may be altered by TRF to confer potential protection against seizure or neuronal damage.With an average standard deviation of 10-20 %, we were able to detect differences in these tests of 14-29 % with 80 % power at the 5 % significance level.These tests were performed with fewer mice (control n = 8, TRF n = 9) per group due to the terminal nature of the MEST test.
All experiments were approved by the University of Queensland's Animal Ethics Committee (2022/AE000376) and followed the guidelines of the Queensland Animal Care and Protection Act 2001 to minimize the suffering of the animals.Eight-week-old male CD1 mice weighing 35-42 g were obtained from the Animal Resources Centre (Perth, Western Australia) and acclimatised for one week.During this time, mice were provided ad libitum access to water and standard rodent chow (Specialty Feeds, Western Australia, Supplementary Table S1).All mice were housed individually under a 12:12 h light-to-dark cycle (lights on between 06:00 and 18:00).After the acclimatisation period, two independent batches of mice with 15-16 mice per group were randomly assigned by body weight to control (24 h ad libitum access to chow) or TRF (16 h fasting, 8 h ad libitum access to chow from 09:00 to 17:00) groups such that starting body weights were similar between groups (P = 0.38-0.88).In experiment one, mice in the TRF group (n = 16, 40.3 ± 1.7 g) were subject to TRF every day for a total of 28-31 days (depending on MEST survival, see later).In the second experiment, TRF mice (n = 15, 37.1 ± 1.9 g) underwent restricted feeding for only 5 days per week for a total of 26 days and were given ad libitum access to food during the weekend (referred to as "2:5 TRF").Control mice in experiment one (n = 15, 39.7 ± 2.0 g) and two (n = 15, 37.2 ± 1.8 g) always had unrestricted access to food.All groups had unrestricted access to water.
Mice were numbered based on their randomised location on the rack.Mice were then weighed and allocated into two groups with average weights less than or equal to 0.6 g different between the groups.Experimenters were blinded to treatment groups throughout all experiments, biomolecular measurements, and analyses.A summary of the experimental outline is presented in Fig. 2.

Blood glucose analysis
After ~ 3 weeks of treatment, random and fasted blood glucose levels were measured using a glucometer (Accu-Check Performa II, Roche, Fig. 2. Experimental study design and timeline to compare one-month time-restricted feeding (TRF) regimen vs. ad libitum-fed mice regarding seizure susceptibility and blood glucose levels.[A] 24 h feeding schedule.Eight-week-old male CD1 mice were randomly assigned, and body weight matched into timerestricted feeding (TRF), fasted for 16 h and fed 8 h between 09:00 to 17:00, or control, provided 24 h ad libitum access to food.Both feeding regimens were permitted 24 h ad libitum access to water.Switzerland) in blood collected from the tail tip.Random blood glucose levels were measured when both groups had access to food, occurring ~ 5 h after the food had been returned to the TRF group.Fasting state blood glucose measurements were taken after both groups had fasted 15-16 h overnight.

Seizure tests
The 6-Hz acute seizure test was performed using established methodology [46,47].On day 1, TRF and control mouse groups were tested after 15 h fasting between 8 and 9 am to test the effects of acute fasting.Thereafter, all seizure tests were carried out during feeding times between 13:00 and 15:00 to minimize potential diurnal or feeding statedependent variations in seizure susceptibility.Briefly, 1-2 drops of 0.5 % tetracaine (w/v) in 0.9 % saline were applied topically to each cornea 15-30 min before the test to anesthetize the corneas.Corneas were stimulated with electrodes pre-wetted with 0.9 % saline using 0.2 ms pulses of square-wave alternating current at 6-Hz for 3 sec with a constant current stimulator (ECT Unit 57800, Ugo Basile, Lombardy, Italy).Seizures were identified by forelimb clonus and/or loss of postural tone with generalised clonic activity.The current required to induce seizures in 50 % of the mice (CC 50 ) was determined by varying the current by 2 mA intervals (beginning from 12 mA) using the "up and down" method [48].
The MEST test was performed as previously described [47].Briefly, 1-2 drops of 0.5 % tetracaine in 0.9 % saline were applied topically to each cornea 15-30 min prior to the test.Corneas were stimulated with electrodes pre-wetted with 0.9 % saline using sine-wave pulses at 50-Hz for 200 ms with a constant current stimulator (HSE-HA Rodent Shocker, Harvard Apparatus, Germany).The CC 50 was calculated as above using the "up and down" method [48], using 1 mA intervals (beginning from 10 mA).Mice that experienced seizures classified by tonic extension of the hind limbs (n = 14) were immediately euthanized by cervical dislocation to avoid MEST-induced death [49].Mice that did not seize during the MEST test (n = 17) resumed their feeding regimens and were left to recover for approximately 3 days until killed (described below).

Tissue collection and plasma extraction
Mice that experienced seizures in the MEST test were excluded from all molecular analyses to avoid any confounding metabolic changes induced by seizures [50,51].After 3 days of recovery from the MEST test, the remaining mice were rapidly killed by cervical dislocation and decapitated.Trunk blood was collected into heparinised Microvette® CB 300 tubes (Sarstedt Inc, Nümbrecht, Germany) and centrifuged at 2000 x g at 4 • C for 10 min to separate plasma.The liver, hippocampal formations, and cerebral cortical hemispheres were rapidly dissected and weighed.A portion of the left lateral lobe of the liver, the hippocampus, and both hemispheres of the cortex were snap-frozen in liquid nitrogen and stored at − 80 • C until required.

Brain cytosolic and mitochondrial isolation
Mitochondrially enriched and cytosolic fractions were isolated from both the cortex and hippocampus.One hemisphere of the cortex was homogenised using a glass-Teflon homogeniser on ice in 800 µL of chilled mitochondrial isolation buffer (210 mM Mannitol, 81 mM Sucrose, 4.6 mM HEPES) and 10 µL per 15 mL buffer of protease inhibitor cocktail (Abcam, ab65400) for 50-60 strokes.50 µL of the total homogenate was reserved and the remaining sample was centrifuged for 1000 x g at 4 • C for 10 min.The pellet containing nuclei and debris was discarded, and the supernatant was re-centrifuged at 12,000 x g at 4 • C for 10 min.200 µL of the supernatant (cytosolic fraction) was aliquoted.The pellet was resuspended in 250 µL of mitochondrial isolation buffer with the addition of 1 mM EGTA and 0.5 % (w/v) BSA, then centrifuged at 10,500 x g at 4 • C for 10 min.The supernatant was discarded, and the pellet (mitochondrially enriched fraction) was resuspended in 100 μL of mitochondrial isolation buffer and aliquoted.The isolation protocol was then repeated for hippocampal formations, using half the volumes as for the cortex.Protein concentrations of tissue fractions were determined using the Pierce bicinchoninic acid (BCA) protein assay kit (Thermo Fisher Scientific, Australia) with a bovine serum albumin (BSA) protein standard (0-2.0mg/mL).All fractions were stored at − 80 • C until further analysis.

Biochemical assays
All absorbance measurements were performed via spectrophotometric measurements using a SpectraMax M Series Microplate Reader on clear 96-well plates (Molecular Devices, CA, USA).Fluorescent measurements were performed using a CLARIOstar Plus Microplate Reader (BMG Labtech, Germany) with 96-well, black bottom plates (NUNC, Denmark).All assays were performed in duplicate in a final reaction volume of 200 μL (unless otherwise stated), corrected for background activity, and quantified within the range of the standard curve (where applicable).

Maximal enzyme activities
The maximal activities of key regulatory enzymes in the cytoplasm and mitochondria were investigated by adding 25 μL of tissue homogenate into the reaction mixtures detailed in Supplementary Table S2.The Spectramax Microplate Reader (Molecular Devices, CA, USA) or CLARIOstar Plus Microplate Reader (BMG Labtech, Germany) were used to record kinetic changes in absorbance or fluorescence respectively for 10-20 min.The rates of change of absorbance or fluorescence were measured from the linear portion of the reaction and corrected for background activity.The maximal activities of hexokinase (HK), phospho-glucose isomerase (PGI), glucose-6-phosphate dehydrogenase (G6PDH), and lactate dehydrogenase (LDH) were measured as changes in absorbance at 340 nm, indicating the reduction of NAD(P)+ to NAD (P)H.Pyruvate carboxylase (PC) and citrate synthase (CS) activities were measured at 412 nm assessing the reaction of DTNB to its 2-nitro-5thiobenzoate anion TNB 2-.Maximal activities of pyruvate dehydrogenase (PDH) and 2-oxoglutarate dehydrogenase (OGDH) were assayed using continuous fluorescence measurements (Ex.255 nm, Em. 460 nm) indicating reduction of NAD+ to NADH.

SDS-PAGE and quantitative Western blotting
Tissue fractions were diluted to 2 mg/ml in Milli-Q water and denatured in 5X sodium dodecyl sulphate (SDS) loading dye solution.Cytosolic fractions were boiled at 95 • C for 5 min to denature proteins.5 µL (10 µg) protein samples were randomly loaded (due to the experimenter being blinded) into a 4 % stacking and 12 % resolving gel along with a protein molecular weight standard (Precision Plus Protein Dual Colour Standards, Bio-Rad Laboratories, Australia) and subjected to SDS-PAGE.Gels were run at 75 V through the stacking gel, then 90-120 V at 4 • C until appropriate separation of the proteins.Proteins were transferred at 100 V to nitrocellulose membranes (0.22 μm, Bio-Rad Laboratories, Germany) for one hour at 4 • C. Membranes were washed using TBS or PBS and blocked for one hour at room temperature with Intercept Blocking Buffer (LI-COR Biosciences, NE, USA).Membranes were incubated overnight at 4 • C with primary antibodies (Supplementary Table S3) followed by one hour at room temperature in IRDye 800CW donkey anti-mouse IgG (LI-COR Biosciences, RRID: AB_621847, 1:15,000) or IRDye 680LT goat anti-rabbit IgG (LI-COR Biosciences, RRID: AB_2713919, 1:15,000) secondary antibodies.Membranes were imaged on an Odyssey CLx imaging system (LI-COR Biosciences, NE, USA).Band intensity was quantified in Image Studio Lite v5.2 (LI-COR Biosciences, NE, USA).To determine total protein amounts, membranes were stained with 0.1 % Ponceau S solution (Sigma-Aldrich, MO, USA), imaged on a Chemidoc M4 imager (Bio-Rad Laboratories, Australia), and quantified in Image Studio Lite v5.2 software as above.To calculate relative protein expression, the band intensity of the protein of interest was normalised to total protein levels calculated from the entire lane.Protein expression levels were reported as fold-change vs the control group.

RNA extraction, cDNA synthesis, and quantitative PCR
The cortex and liver were crushed into powder using liquid nitrogen and a mortar and pestle.Approximately 10-30 mg of tissue was extracted using QIAzol (QIAGEN, Australia) as per the manufacturer's instructions and homogenised with beads using a FastPrep-24 5G homogeniser (MP Biomedicals, Australia).RNA yield and purity were determined using a NanoDrop spectrophotometer (Thermo Fisher Scientific, USA).0.5 μg (cortex) or 1 μg (liver) of total RNA was reverse transcribed into cDNA using the iScript gDNA clear cDNA synthesis kit (Bio-Rad Laboratories, Australia) in a 10 μL reaction according to the manufacturer's instructions.cDNA samples were diluted 1:10 in ultrapure water, then stored at − 20 • C until analysis.Real-Time quantitative PCR (qPCR) was performed using the QuantiNova SYBR Green PCR Kit (QIAGEN, Australia).Primer sequences were obtained from previous studies or predesigned KiCqStart SYBR Green primers (Sigma-Aldrich, MO, USA) were used (Supplementary Table S4).10 μL reactions were amplified in duplicate using the QuantStudio 6 Flex Real-Time PCR System (Applied Biosystems, Thermo Fisher Scientific, USA).Samples were heated to 95 • C for 2 min, then cycled between 10 s at 95 • C and 20 s at 60 • C for 40 cycles.Melt curves for all genes were generated and produced only single peaks.No product was detected in the no-template controls.Relative mRNA expression was calculated using the 2 -ΔΔCt method and normalised to the geometric mean of Tbp and Top1 in the cortex and Tbp and B2m in the liver.These genes were stably expressed and not affected by TRF (data not shown).

Mitochondrial DNA content
DNA was extracted from the cortex, hippocampus, and liver using the DNeasy blood and tissue spin-column kit (QIAGEN, Australia) as per the manufacturer's instructions.Owing to the low tissue weight of the hippocampus, only DNA was extracted from mice that had seizures in the MEST test to prioritise tissue for enzyme assays and Western blotting.DNA was diluted to 50 ng/μL and amplified in duplicate using quantitative PCR (see 2.10.) with primer sets specific to two mitochondrial DNA markers (D-loop and Mt-Nd2) and two nuclear DNA markers (Actb and Gapdh) as described [52] (Supplementary Table S4).To calculate relative mitochondrial DNA content, the geometric mean of the two mitochondrial DNA markers was normalised to the geometric mean of the two nuclear DNA markers using the 2 -ΔΔCt method.
The ferric-reducing ability of plasma (FRAP) was measured by the change in absorbance from the reduction of Fe[III]-TPTZ (Fe 3+ -2,4,6tripyridyl-S-triazine) to Fe[II]-TPTZ (Fe 2+ -2,4,6-tripyridyl-S-triazine) [54].15 μL sample was added with Fe(II)SO 4 -7⋅H 2 O standards (30-1000 μM) to a reaction mixture (Supplementary Table S5).Background absorbance was measured at 593 nm, then the final absorbance was measured after 30 min of incubation at 37  S5).Superoxide dismutase (SOD) activity was measured using a SOD Assay Kit (Cayman Chemical, Item no.706002, USA) by measuring the change in absorbance per minute of the amount of enzyme required to cause 50 % dismutation of the superoxide radical (one unit).10 μL of the sample was added to a reaction mixture and the assay was performed as per the manufacturer's instructions.Samples were incubated for 30 min at room temperature before the final absorbance at 440 nm was measured.
Catalase (CAT) activity was measured using a CAT Assay Kit (Cayman Chemical, Item no.707002, USA) as per the manufacturer's instructions.20 μL of the sample or CAT standards were added to the reaction mixture.The final absorbance at 540 nm was measured after various steps totalling 35 min of incubation at room temperature.

Statistical analysis
All statistical analyses were performed in GraphPad Prism 9.5.1 (GraphPad Software, California, USA).Body weights and blood glucose levels were analysed by two-way repeated measures analysis of variance (ANOVA), followed by Fisher's LSD post hoc multiple comparisons test if a significant main effect was detected.All other data were assessed for normality (Shapiro-Wilk normality test) and analysed using two-tailed unpaired t-tests, with Welch's correction applied when an F test indicated that variances were unequal (denoted by ‡ in the text).Nonnormally distributed data were analysed using a Mann-Whitney U test (denoted by ø in the text).All data are presented as mean ± SEM, with statistical significance determined at P < 0.05.In all analyses, values that exceeded ± 2 standard deviations from the mean of normally distributed data were excluded as outliers.

Results
The present study assessed the effects of a one-month TRF regimen in eight-week-old male mice, with the TRF experimental group fed for 8 h and fasting for the remaining 16 h of every day in experiment 1.In experiment 2, the treated mice had similarly restricted access to food for 5 days during the week (2:5 TRF) but ad libitum feeding during weekends.Control mice were fed ad libitum throughout each experiment.

No alterations in body weight and stabilised fed and fasted blood glucose levels during one-month TRF
All mice increased their body weight over time (P time < 0.0001) with no significant difference in body weight found between TRF and ad libitum-fed mice (P trt = 0.90, Fig. 3A).This was despite a 12 % decrease in their average 24 h food consumption, resembling caloric restriction (P = 0.005, Fig. 3B).TRF also did not affect the wet weights of the cortex, hippocampal formation, or the liver (Supplementary Table S6).Fed and fasted blood glucose levels were measured at week 3.A significant interaction between feeding state and treatment in the two-way repeated measures ANOVA (P int < 0.0001, Fig. 3C) indicated that TRF affected blood glucose levels in a feeding state-dependent manner.Post hoc analysis showed that this interaction was based on TRF mice exhibiting 19 % lower random blood glucose levels in the fed state (P < 0.0001) but 16 % higher blood glucose levels in the fasted state (P = 0.003, Fig. 3C) compared to ad libitum-fed mice.In ad libitum-fed mice, 15-16 h of fasting reduced blood glucose levels by 30 % (P < 0.0001), while TRF mice interestingly did not show any fasting-induced changes in blood glucose levels (P = 0.90, Fig. 3C).

Weekend binge-like eating and stabilised fed and fasted blood glucose levels during one-month 2:5 TRF
Ad libitum-fed and 2:5 TRF mice also increased their body weight over time (P time < 0.0001) and in a similar fashion (P trt = 0.69, Fig. 3D).The 2:5 TRF group showed a 20 % decrease in their weekday food consumption (P < 0.0001) and a 35 % increase in weekend consumption (P < 0.0001).This resembled binge-like feeding behaviour and resulted in no statistical difference in the overall average weekly food consumption among the two groups (P = 0.053, Fig. 3E).During week 3, blood glucose levels differed significantly with treatment (P trt < 0.01) and fasting (P time < 0.001, Fig. 3F), but random blood glucose levels after 5 h refeeding were not different between ad libitum-fed and 2:5 TRF groups (P = 0.30 ø , Fig. 3F).However, we found a 17 % increase in 15-16 h fasting blood glucose levels in 2:5 TRF mice compared to the ad libitum-fed group (P < 0.001, Fig. 3F).This corresponded to a 19 % reduction in blood glucose levels in ad libitum-fed mice with fasting (P < 0.0001), whereas blood glucose levels were unaltered after fasting in the 2:5 TRF group (P = 0.81, Fig. 3F).

Transient anticonvulsant effects in the 6-Hz seizure test after onemonth TRF
In a 6-Hz seizure test, there was no change in the seizure threshold in fed mice vs those that had experienced a single overnight (15-16 h) fast on day 1 (P = 0.52, Fig. 3G), suggesting no differences in electrical seizure thresholds between the fed and fasted state.Thereafter, all seizure thresholds were assessed in fed states.In mice undergoing continuous TRF for one month, we found a significantly increased seizure threshold of 12.7 mA compared to 11.0 mA in the ad libitum-fed group at day 15 (P = 0.04), however, no differences were seen at day 8 (P = 0.25) or day 22 (P = 0.24, Fig. 3H).The seizure threshold in the MEST acute seizure test conducted on day 28 was also unchanged by TRF (P = 0.37, Fig. 3I).We saw similar results with the 2:5 TRF regimen, during which mice had an increased seizure threshold (17.3 mA) compared to 14.4 mA in the ad libitum-fed group at day 19 (P = 0.02), but no changes were observed at day 26 (P = 0.56, Fig. 3J).

Absence of considerable differences in mitochondrial DNA content, Ppargc1a gene expression, or levels of proteins associated with mitochondrial fission and fusion after one-month TRF
After one-month of TRF there were no differences compared to the ad libitum-fed group regarding mitochondrial DNA content in the cortex (P = 0.29), hippocampus (P = 0.46), or liver (P = 0.10, Fig. 7A).This was accompanied by the lack of changes in the gene expression of peroxisome proliferator-activated receptor gamma coactivator 1-alpha (Ppargc1a) in the cortex (P = 0.75), or liver (P = 0.22, Fig. 7B), a master regulator of mitochondrial biogenesis, as well as fission and fusion protein expression, namely mitofusin-1 (MFN1, P = 0.14, Fig. 7C) and dynamin-related protein 1 (DRP1, P = 0.78, Fig. 7D).Interestingly, there was a 21 % increase in the long form of OPA1 mitochondrial dynamin-like GTPase (L-OPA1, P = 0.03) and a 17 % increase in total OPA1 protein expression (P = 0.04).However, there were no differences in short form OPA1 levels (S-OPA1, P = 0.12) or the ratio of long to short   [K-T] the hippocampus of mice subjected to 8 h of time-restricted feeding (TRF, n = 9) for one month.All proteins of interest were normalised to the total protein levels (10-250 kDa) measured using Ponceau staining, with a representative portion of the Ponceau stained membrane displayed from 37 to 50 kDa to save space.Data were analysed by an unpaired t-test with Welch's correction where appropriate.Non-normally distributed data were analysed using a Mann-Whitney U test.Data are presented as mean ± SEM, with statistical significance determined at P < 0.05.*P < 0.05, **P < 0.01.form OPA1 (L-OPA1/S-OPA1, P = 0.34 ‡ , Fig. 7E).In the hippocampus, there were no differences in protein expression of MFN1 (P = 0.58, Fig. 7F), DRP1 (P = 0.92, Fig. 7G), L-OPA1 (P = 0.36), S-OPA1 (P = 0.14), L-OPA1 to S-OPA1 ratio (P = 0.92), or total OPA1 protein expression (P = 0.27, Fig. 7H).

Discussion
The major findings of this study are summarised here.Compared to mice with ad libitum access to food, in the mice subjected to continuous TRF and weekday 2:5 TRF, we found (1) transient protection in the 6-Hz acute seizure test on day 15 and day 19 respectively, as well as (2) similar random vs fasting blood glucose concentrations suggesting greater stabilisation of blood glucose levels without differences in body weight gain.After 4 weeks of TRF we also found, notably in the absence of an anticonvulsant effect in the acute electrical seizure tests, (3) increased MPC1 protein levels in the cortex and hippocampus, accompanied by increased PDH maximal enzyme activity in the cortex, (4) small reductions in maximal enzyme activities of G6PDH and PGI in the cortex and hippocampus respectively, but (5) lack of substantial differences in AMPK/Akt/GSK3β signaling pathways, markers of mitochondrial dynamics, or antioxidant capacity and defence.These findings are discussed below.

Anticonvulsant effects in the 6-Hz model
This study is the first to investigate the effects of 8 h TRF regarding seizure protection in acute electrical seizure models in CD1 mice as opposed to the more extreme fasting duration during ADF.Given the beneficial effects of ADF in mice against kainate-induced seizures and other chemically induced seizures in rodents [30][31][32][33][34][35][36][37][38], our more  of mice fed ad libitum (control, n = 8), or subjected to 8 h of time-restricted feeding (TRF, n = 9) for one month.Data were analysed by an unpaired t-test with Welch's correction where appropriate.Non-normally distributed data were analysed using a Mann-Whitney U test.Data are presented as mean ± SEM, with statistical significance determined at P < 0.05.moderate fasting regime with TRF for 8 h was hypothesised to also provide seizure protection in acute electrical seizure models.Opposite to Hartman et al. who assessed the effect of ADF in the 6-Hz test (24 h after feeding), we found no proconvulsant effects in the 6-Hz and MEST model at any time point (assessed after 4-5 h feeding) [36].We found that in two experiments compared to control groups, the TRF groups had transiently higher seizure thresholds between 2 and 3 weeks in the 6-Hz seizure test, including the 2:5 TRF group which was fed ad libitum during weekends.Interestingly, transient seizure protection has also been observed in male NIH Swiss mice following a ketogenic diet, where the CC 50 in the 6-Hz test was elevated at 12 and 16 days, but not at 2, 5, or 21 days [57].We also found no difference in seizure threshold when we compared ad libitum-fed mice to mice that had been fasted for 15-16 h prior to the 6-Hz test, indicating that there are no anticonvulsant effects associated with acute fasting.Although we did not assess seizure thresholds during the fasting period of TRF, Landgrave-Gómez et al. have found protection against pilocarpine-induced status epilepticus following 20 days of 2 h TRF which was assessed after 22 h of fasting [30].Furthermore, while we found that TRF provided only transient seizure protection in the fed state, other studies using rats have reported anticonvulsant effects after longer durations of intermittent fasting.This includes protection against kainate-induced seizures following 24 weeks of ADF [32], or protection against pentylenetetrazole-kindling after 4 weeks of ADF (assessed 1 h after feeding) [31].
Protection against kainate-induced seizures in the ADF group in NIH Swiss mice has been reported in as little as 11-13 days [36] and a longterm study found no change in kainate-induced seizure severity or duration after 24 weeks of ADF in C57BL/6 mice [37].In these older studies, the fasting durations (22-24 h) were much longer between feedings than in our TRF regimen (16 h) and in mice were proconvulsant in the 6-Hz and MEST acute seizure tests at two weeks [36].Thus, our findings suggest that abstaining from food for shorter intervals by adopting an 8 h TRF regimen may be a more promising approach towards obtaining seizure protection, but further optimisation of our TRF regimen is needed to achieve longer-lasting effects in electrical seizure tests.
The 6-Hz focal seizure threshold model appears to be highly responsive to metabolic alterations and has been used to detect anticonvulsant effects of various metabolic treatments, such as ketogenic diets [46,57], D-leucine [58], octanoic and decanoic acids, as well as medium chain triglycerides containing almost exclusively decanoic acid [59][60][61].Some of the treatments discovered to be anticonvulsant in the 6-Hz threshold test have also been found to be effective in other seizure models and to some extent in humans with medication-resistant epilepsies, such as ketogenic diets and MCT [6,22,24,62,63].Thus, the effects of TRF in the 6-Hz threshold model found here are promising.

Epilepsy and seizure models
Our study only used acute electrical seizure models, but chronic and chemical seizure models may be more likely to reveal the anticonvulsant efficacy of metabolic treatments.Compared to acute electrical models, seizure development is not instantaneous in chemical and chronic models.Particularly, chronic models show low FDG-PET signals and impairments in energy metabolism which are similar to those found in human epilepsies [6][7][8][9][10].Impairments in glucose metabolism can contribute towards seizure generation [7,64,65].Thus, improving energy availability by metabolic interventions is more likely to promote glial and neuronal energy metabolism to support ATP-dependent potassium and glutamate uptake which could delay seizure generation.In contrast, electroshock-induced convulsions in acute seizure models result in instantaneous neuronal activation of various brain regions and the spinal cord [66].The hyperexcited neural tissue rapidly increases its use of brain energy, glucose, and glycogen, to support neural seizure activity [67][68][69].In some instances, rather than conferring protection, improved bioenergetics, increased energy storage or utilization during metabolic therapies may inadvertently promote an environment that can support high energy-demanding hyperexcitation in acute seizure tests [70].This suggests that acute electrical seizure models may not be ideal for detecting the anticonvulsant activity of intermittent fasting, supported by Hartman et al. (2010) who found proconvulsant effects after ADF in the electrical 6-Hz and MEST tests, yet anticonvulsant activity against more gradually evolving kainate-induced seizures in mice [36].In addition, the lack of prolonged protection in acute electrical models in our study may not necessarily indicate a lack of efficacy in other models or humans.Although most current medications used in the clinic have some efficacy in electrical seizure tests [71][72][73], this could be explained by the fact that electrical seizure tests have been the first checkpoint within preclinical anticonvulsant screening programs [74].Thus, to comprehensively assess the anticonvulsant nature of our TRF regimen in the future, investigating TRF in acute chemical models and chronic models of epilepsy with spontaneous recurrent seizures is required.

Blood glucose levels
We found that during week 3, (1) random blood glucose levels measured after 5 h of refeeding were lower in the continuous TRF group compared to ad libitum-fed mice, (2) fasting blood glucose levels were higher in both TRF feeding regimens compared to ad libitum feeding, and (3) both continuous TRF and 2:5 TRF mouse groups showed similar blood glucose levels in the fed and fasted state.Regarding our first finding, it is important to note that decreases in random blood glucose have been previously reported in TRF mice [30,75] and other intermittent fasting regimens in rodents [76] suggesting improved glycaemic control.The second finding of higher blood glucose levels found in TRF compared to the ad libitum-fed group during fasting may indicate stressinduced increases in blood glucose levels from periodic changes in food supply with light phase feeding (discussed later in 4.8.).However, these higher fasting blood glucose levels could also be explained by metabolic adaptations to repetitive fasting, such as increased liver glycogenolysis and/or gluconeogenesis to maintain stable blood glucose levels during intermittent fasting [77].Consistent with the latter explanations, continuous and 2:5 TRF have been associated with increased glucose tolerance and reduced insulin resistance [78].The absence of detected fluctuations in blood glucose levels between feeding and fasting times during TRF, our third finding, may contribute to seizure protection by maintaining a continuous and similar glucose supply to the brain [13], as extreme increases or decreases in peripheral glucose concentrations are proconvulsant in rats [79,80].Future studies are needed to include multiple measurements of blood glucose including the troughs and peaks expected around 1 h before and after start of feeding.Stable low, but still normal blood glucose levels are found with ketogenic diets and have been discussed as one mechanism by which ketogenic diets may confer protection against seizure generation [6,28,81].

Pyruvate metabolism
Despite the absence of a detected anticonvulsant effect during week 4 of TRF, a novel finding of our study was increased MPC1 protein expression in both the hippocampus and cortex of the TRF group.MPC1 is widely expressed in neurons, astrocytes, and microglia and forms a heterodimer transporter complex with MPC2 that is critical for transporting pyruvate into the mitochondria [82].Interestingly, we found no change in mRNA expression of Slc16a1 and Slc16a2 in the cortex and liver, which transport ketone bodies, but also pyruvate across the plasma membrane [83].Deletion of Mpc1 has been found to disrupt mitochondrial pyruvate uptake in yeast and Drosophila cells, impair pyruvate oxidation in human fibroblasts [84], and increase sensitivity to pentylenetetrazole kindling and kainate-induced seizures in mice [85].Overexpressing the MPC complex in cultured mammalian cells has been found to increase movements of metabolites across the mitochondrial membrane and enhance maximal respiration [86], however, there are few studies where there is overexpression of MPC1 alone.Liu et al. (2023) found that MPC1 protein levels were reduced after bilateral internal carotid artery embolization in rats and deficiency of MPC1 aggravated oxidative stress and mitochondrial dysfunction via the PI3K/ Akt/mTOR pathway [87].Overexpression of MPC1 prevented cerebral ischemia-induced cell death [87], as well as ROS formation in cultured cells with defective peroxisomal membrane proteins (pox1Δ and pex34Δ) undergoing endoplasmic reticulum stress [88].Although we did not find any changes in AKT protein expression or phosphorylation, markers of mitochondrial biogenesis, and antioxidant capacity (discussed below in 4.5.) in our study, the MPC1 increase found may contribute to neuroprotective and anticonvulsant effects.The upregulation of MPC1 is expected to be most important in chronic epilepsy models where impaired mitochondrial function and increased oxidative stress have been found.On the other hand, it is unknown to which extent upregulation of MPC1 alone without alterations in MPC2 protein levels would increase pyruvate transport, as both MPC1 and MPC2 are needed to form a functional transporter [82].
The increase in MPC1 protein expression was accompanied by a 20 % increase in maximal activity of PDH in the cortex.As there were no changes in total protein levels of the PDH E1α subunit, increased PDH maximal activity is likely linked to changes in the phosphorylation of the regulatory E1α subunit which could be measured in future studies [89].PDH decarboxylates pyruvate and forms acetyl-CoA which enters the TCA cycle.Acetyl-CoA-derived carbons can contribute to the production of lipids and amino acids and/or after their oxidation within the TCA cycle, high amounts of ATP can be generated via the electron transport chain (ETC).Our laboratory reported that PDH maximal activity was decreased after flurothyl-induced seizures [90] and also that hippocampal PDH activities were lower interictally in the chronic stages of the mouse pilocarpine model [9,89].These alterations are expected to reduce oxidative ATP production in the brain [7].Thus, the increased PDH activity found 4 weeks after our 8 h TRF regimen here may be beneficial to counteract interictal PDH activity losses in epilepsy and to improve ATP production needed for prevention of and recovery after seizures.Increased PDH activity was also observed with triheptanoin supplementation, where restored PDH maximal activity together with increased entry of glucose-derived carbons into the TCA cycle were found in the chronic stage of the pilocarpine mouse model as well as protection against second hit seizures [91,92].Collectively, future studies are needed to functionally determine whether elevations in MPC1 protein expression and PDH maximal activity increase pyruvate uptake and acetyl CoA production in the brain in vivo and to which extent these changes could elicit anticonvulsant effects in other seizure and chronic epilepsy models.

Cytosolic and oxidative glucose metabolism
Several small changes in maximal enzyme activity were found in the two brain regions assessed, namely a 6 % decrease of PGI in the hippocampus and a 7 % decrease of G6PDH in the cortex, but no differences in HK, PGI, or LDH maximal activities after 4 weeks of TRF.The small decreases in PGI maximal activity found in the hippocampus are likely to have little effect as this enzyme is not rate-limiting in glycolysis [93].Since G6PDH is a rate-limiting enzyme that regulates the entry of glucose-6-phosphate (G6P) into the pentose phosphate pathway, reduced G6PDH activity may reduce flux through this pathway and could result in decreased NADPH production which may hamper antioxidant defence.However, G6PDH is in itself regulated by the NADPH/ NADP ratio, such that when the ratio decreases there is increased activity to provide more NADPH [94].Furthermore, G6PDH maximal activity has not previously been found to be changed following one month of ADF in mice measured in the cortex [77,95], liver [77], or skeletal muscle [96].Thus, we prefer not to speculate on the small changes in maximal G6PDH activity we report here.Future studies that assess the metabolism of injected 13 C-glucose and NADPH/NADP ratios using liquid chromatography with tandem mass spectroscopy could clarify whether TRF affects the activity of these enzymes in a functionally relevant manner in vivo.

AMPK, AKT and GSK3β
Contrary to the literature, TRF in the current study did not alter total or phosphorylated protein levels of AMPKα, AKT, or GSK3β in the cytosolic or mitochondrial enriched fractions of the hippocampus or cortex.This was accompanied by a lack of changes in mRNA levels of genes including Foxo1, Foxo3, Sirt1, and Sirt3, gene transcripts within the network of AMPK and AKT signaling pathways induced by calorie restriction [97].In other studies, decreased AKT and GSK3β phosphorylation have been found in the brains of ADF mice [98], as well as increases in AMPK phosphorylation and decreases in AKT phosphorylation in the cortex and liver of rats after 20 days of 2 h TRF [30].These changes have been thought to be connected to increased latency and reduced severity of pilocarpine-induced status epilepticus found after ADF [30].In a similar manner to the ketogenic diet [99] and calorie restriction [100], the protective mechanisms of AMPK/AKT/GSK3β signaling in preventing seizure generation may be partly explained by the inhibition of mammalian target of rapamycin complex 1 (mTORC1).mTORC1 inhibitors such as rapamycin and everolimus are anticonvulsant in tuberous sclerosis and tuberous sclerosis animal models, as overactivation of mTOR signaling is associated with neuronal excitability and decreased seizure thresholds [101,102].Although we did not measure mTOR activity, the absence of changes in AMPK and AKT phosphorylation suggests a lack of alterations in the mTOR signaling pathway influencing neuronal excitability and cellular homeostasis [102,103].Taken together, the lack of persistent seizure protection in our study may be partly explained by the absence of changes in AMPK, AKT, or GSK3β signaling.

Mitochondrial dynamics and content
Given that mitochondrial structure and content are important for maintaining cellular respiration, it was important to investigate whether TRF was associated with metabolic changes through alterations in mitochondrial dynamics.Increased mitochondrial biogenesis and improved mitochondrial dynamics are thought to be among the mechanisms underlying the neuroprotective effects of intermittent fasting [104].This contributes to reduced oxidative stress and inflammation which may occur similarly to the ketogenic diet, where one of the proposed anticonvulsant mechanisms includes increased mitochondrial biogenesis [104,105].Our study found a 17 % increase in the total levels of pro-fusion OPA1 protein in the cortex.This was associated with a 21 % significant increase only in the long form of this protein, although the ratio of long to short OPA1 was unchanged.In the hippocampus, there were no changes in OPA1 protein amounts.We also found no changes in the levels of pro-fusion MFN1 protein or pro-fission DRP1 protein in either cortex or hippocampus.This is unlike a previous study where mice after 40 % calorie restriction for six months showed increased protein expression of the pro-fission mitochondrial fission 1 protein (FIS1) and DRP1, but no changes in the pro-fusion proteins MFN1, MFN2, and OPA1 in the liver [106].Another study that maintained mice on an ADF schedule for 3 or 12 months found increased brain and liver mRNA levels of Ppargc1a, Mfn1, Mfn2, Nrf1, and mitochondrial transcription factor A (Tfam), suggesting increased mitochondrial biogenesis [107].In contrast, in our study with a TRF regimen of shorter overall duration and less extreme fasting periods, Ppargc1a mRNA expression was unchanged in the cortex and liver.This, combined with only slight changes seen in the levels of total and long-form OPA1 in the cortex, is consistent with our data that showed no alterations to mitochondrial DNA content in the hippocampus, cortex, or liver.Together these data indicate no changes to mitochondrial dynamics or biogenesis in our model.

Antioxidant defence
Despite evidence that intermittent fasting can augment antioxidant defence [26,108], our study found no alterations of total antioxidant capacity measured in three different assays in plasma, liver, and the cortex.Antioxidant capacity was assessed using the TEAC, FRAP, and ORAC assays, as these three methods measure the levels of numerous different antioxidant molecules in biological samples, which reduce free radicals or scavenge reactive molecules [109].We also did not detect any mRNA expression changes of antioxidant-associated genes, or changes in superoxide dismutase and catalase activity, which have been amongst the most common alterations found with intermittent fasting in rodents [110].On the other hand, our findings are similar to many other intermittent fasting studies in rodents, in which there were little or no changes in antioxidant enzyme protein and mRNA levels or enzymatic activities [110].Interestingly, this review of several intermittent fasting studies reported that increased activities of antioxidant enzymes were commonly seen in peripheral tissues, but not as much in brain tissue, while increases in brain glutathione levels occurred in the majority of studies [110].
It is also possible that a more intensive intermittent fasting regimen is needed to increase antioxidant defence.One study by Bhoumik et al. (2020) on the differences between ADF and TRF methods found that healthy Wistar rats maintained on 8 h TRF for one month did not improve antioxidant capacity in plasma as measured by FRAP assay [111].Instead, restricting food with ADF resulted in a 34 % increase in antioxidant capacity accompanied by a 67 % increase in intracellular glutathione levels [111].It is also possible that the lack of antioxidant and little anticonvulsant benefits of TRF in our study are due to the use of healthy mice.Increased oxidative stress arises from prolonged seizures and may contribute to epilepsy development and seizure generation through neurodegenerative pathways [15,[112][113][114][115][116].Rajeev et al. found that 8 h TRF during the light phase in C57BL/6NTac mice reduced chronic cerebral hyperfusion-induced increases in the oxidative stress marker malondialdehyde in the cortex, hippocampus, and cerebellum, and increased antioxidant markers glutathione and SOD protein levels in the cortex and cerebellum after 30 days [44].Thus, future investigations in chronic epilepsy models that have pathological features associated with intractable epilepsy in humans may reveal the potential antioxidant and anticonvulsant effects of TRF.

Limitations
The primary limitation of this study was that the mice in our TRF regimen were provided with food during the light (sleep) phase, rather than the dark (active) phase, which has been a common methodology for restricted feeding studies in rodents [41][42][43][44].The timing of food consumption serves as an important synchronizer of mammalian circadian rhythms [117], which is primarily driven by the circadian clock in the liver rather than the suprachiasmatic nucleus of the hypothalamus [118].Animals forced to eat during their resting phase demonstrated increased food intake and adiposity [41] compared to ad libitum feeding.These mice showed desynchronizations of numerous clock and metabolic genes between the brain, liver, and muscle that resemble a model of shift work conditions that can disrupt glucose homeostasis [119,120].Therefore, it is likely that the findings from our study are confounded by the effects of altered eating times, which may result in changes to circadian rhythm, sleep loss, and/or desynchronization of metabolic genes in the brain and periphery.Any of these conceivable changes alone, or a combination thereof may have attenuated the potential benefits of TRF.However, 20 days of TRF with 2 h daily feeding during the light phase was found to increase latency to seizure onset and reduce the severity of pilocarpine-induced status epilepticus in male Wistar rats [30].Furthermore, a study on 321 people with active epilepsy who underwent Ramadan fasting and abstained from food and fluid intake during their active period from sunrise to sunset reported a significant reduction in focal, myoclonic, and absence seizures during the month of fasting, and a significant reduction in focal and myoclonic seizures during the month after [121].Future investigations with fasting during the light cycle are needed to reveal the full potential of an 8 h TRF regimen in seizure models.Also, it would be of interest to include a study of the rhythmic mRNA expression of circadian clock genes such as transcriptional factors and nuclear hormone receptors in peripheral and central tissue to determine the extent to which circadian rhythm was impacted in our study.As TRF could restore disturbances of circadian rhythm genes in healthy mice [122] and in a mouse model of Alzheimer's disease [123], understanding these mechanisms may be beneficial in the context of epilepsy and circadian rhythm disruption.
It is unlikely that intermittent fasting is used without anti-seizure medications, but it could potentially augment the effectiveness of medications.Apart from valproate and everolimus, current anti-seizure medications reduce seizures by altering neurotransmission without other effects thought to play a major role in their mechanisms of action.Instead, metabolic therapies have multiple beneficial mechanisms, including alteration of brain energetics as well as reduction of oxidative stress and inflammation [6,124,125].Future studies are needed to investigate TRF in combination with anti-seizure medications to identify synergistic combinations that may allow reduction of medication use and side effects.
We acknowledge that our work and that of others investigating intermittent fasting in rodents [30][31][32][33][34][35][36][37][38] only included males and studies using female animals are needed.Interestingly, a low glycemic diet delayed epileptogenesis in synapsin II-deficient female mice, but not males [126].This effect was correlated to an increase in cortical allopregnanolone levels, which is likely to contribute to protection against the development of seizures and the reduced seizure severity found.

Conclusions
This study is the first to investigate the anticonvulsant potential of an 8 h TRF regimen in acute electrical seizure models in mice.Opposite to the proconvulsant effects of ADF that have been previously reported in electrical seizure models, our TRF regimen with and without weekend interruptions provided transient protection in 6-Hz seizure tests at 15 and 19 days.These anticonvulsant effects could be driven by stabilisation of blood glucose levels between fed and fasted states, increased cortical and hippocampal MPC1 protein expression, and/or increased cortical maximal PDH activity.The findings of our study support further investigations into electrophysiological and functional metabolic effects in vivo, as well as the anticonvulsant effects of TRF with feeding during the active period in chronic epilepsy models.

Declaration of competing interest
In the past, K.B. has received consulting fees from Nestlé Purina PetCare and Ultragenyx Pharmaceuticals Inc. as well as research support from Ultragenyx Pharmaceuticals Inc. interests or personal relationships that could have appeared to influence the work reported in this paper.

Fig. 1 .
Fig. 1.Simplified schematic of brain glucose metabolism with key regulatory enzymes.Glucose, ketone bodies, lactate, and amino acids are transported into the cytosol and undergo subsequent reactions into the tricarboxylic acid (TCA) cycle and the electron transport chain in the mitochondria where they are metabolised into adenosine triphosphate (ATP) via oxidative phosphorylation.Key regulatory enzymes may offer a therapeutic target for metabolic dysfunctions and oxidative stress that is associated with epilepsy.Abbreviations: Alpha-ketoglutarate (αKG), Beta-hydroxybutyrate (BHB), Electron transport chain (ETC) complexes 1-5 (I-V), Fructose-1,6-biphosphate (FBP), Fructose-6-phosphate (F6P), Glucose transporter (Glut), Glucose-6phosphate (G6P), Glucose-6-phosphate dehydrogenase (G6PDH), Glutamic acid (Glu), Glutamine (Gln), Hexokinase (HK), Lactate dehydrogenase (LDH), Mitochondrial pyruvate carrier (MPC), Monocarboxylate transporter (MCT), 2oxoglutarate dehydrogenase (OGDH), Phosphofructokinase (PFK), Phosphoglucose isomerase (PGI), Pyruvate dehydrogenase (PDH), ribose 5-phosphate (R5P).Red text indicates key regulatory enzymes.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig.2.Experimental study design and timeline to compare one-month time-restricted feeding (TRF) regimen vs. ad libitum-fed mice regarding seizure susceptibility and blood glucose levels.[A] 24 h feeding schedule.Eight-week-old male CD1 mice were randomly assigned, and body weight matched into timerestricted feeding (TRF), fasted for 16 h and fed 8 h between 09:00 to 17:00, or control, provided 24 h ad libitum access to food.Both feeding regimens were permitted 24 h ad libitum access to water.[B] Weekly feeding schedule.Treated mice were subjected to continuous TRF (Experiment 1, n = 16), or TRF during the week and ad libitum feeding during weekends (2:5 TRF, Experiment 2, n = 15).All control mice were fed ad libitum for 7 days per week (n = 15 per experiment).[C] Experimental timeline.During the TRF regimen, four 6-Hz acute seizure tests and one maximal electrical shock threshold (MEST) acute seizure test were performed in Experiment 1, and two 6-Hz acute seizure tests in Experiment 2 to assess the anticonvulsant potential of TRF over time.Random and fasting blood glucose levels were measured after 3 weeks.Mice that did not experience seizures after the MEST test were cervically dislocated and then decapitated immediately on day 31.The trunk blood plasma, cerebral cortex, hippocampus, and liver were extracted and stored at − 80 • C until subsequent analysis (control n = 8, TRF n = 9).Figure created with BioR ender.com.

Fig. 3 .
Fig. 3. Stabilised blood glucose levels and transient seizure protection in the 6-Hz seizure test during one-month time-restricted feeding (TRF).[A] Body weight measurements at days 0 and 28-31 after continuous TRF or ad libitum feeding, [B] average 24 h food consumption measured during weeks 3-4; [C] blood glucose measurements at week 3 after 5 h refeed, or after 15-16 h of fasting; [D] body weight measurements on days 0 and 26 after 2:5 TRF or ad libitum feeding; [E] 2:5 TRF weekday, weekend, and total average 24 h food consumption measured during weeks 3-4; [F] 2:5 TRF blood glucose measurements after 5 h refeed, or 15-16 h of fasting during week 3; [G] 6-Hz seizure test following an overnight fast on day 1 (15-16 h); [H] 6-Hz acute seizure test performed on days 8, 15, and 22; [I] maximal electrical shock threshold (MEST) test performed on day 28; and [J] 2:5 TRF 6-Hz acute seizure test performed on days 19 and 26 in mice fed ad libitum (control, n = 15 per experiment), or subjected to eight hours of continuous (TRF, n = 16) or weekday (2:5 TRF, n = 15) time-restricted feeding for one month, respectively.The critical current (CC 50 (mA)) required for 50 % of the mice to seize was calculated using the up-and-down method.[A], [C], [D], [F] were analysed by two-way repeated measures ANOVA with Fisher's LSD post hoc multiple comparisons test.All other parameters were analysed by an unpaired t-test with Welch's correction where appropriate.Non-normally distributed data were analysed using a Mann-Whitney U test.Data are presented as mean ± SEM, with statistical significance determined at P < 0.05.*P < 0.05, **P < 0.01, ***P < 0.001, ### P < 0.0001.

Fig. 4 .
Fig. 4. Lack of differences in indicators of mitochondrial and cytosolic energy signaling or regulatory pathways after one-month time-restricted feeding (TRF).Relative protein levels of total, phosphorylated, and the phosphorylated ratio of [A] AMP-activated protein kinase alpha (AMPKα, p-AMPKα Thr172 , p-AMPKα Thr172 /AMPKα); [B] total, phosphorylated, and the phosphorylated ratio of protein kinase B (AKT, p-AKT Ser473 , p-AKT Ser473 /AKT); [C] total, phosphorylated, and the phosphorylated ratio of glycogen synthase kinase-3 beta (GSK3β, p-GSK3β Ser9 , p-GSK3β Ser9 /GSK3β) in mitochondrial enriched and cytosolic fractions of the cortex; [D] Western blots and Ponceau staining to indicate total protein levels; and [E-F] mRNA expression of genes forkhead box O1 (Foxo1), forkhead box O3 (Foxo3), sirtuin 1 (Sirt1), sirtuin 3 (Sirt3), solute carrier family 16 (monocarboxylate transporter) member 1 (Slc16a1) encoding MCT1 protein, solute carrier family 16 (monocarboxylate transporter) member 7 (Slc16a7) encoding MCT2 protein, and brain-derived neurotrophic factor (Bdnf) in the cortex [E] and liver [F] in mice fed ad libitum (control, n = 8), or subjected to eight hours of time-restricted feeding (TRF, n = 9) for one month.All proteins of interest were normalised to the total protein levels (10-250 kDa) measured using Ponceau staining, with a representative portion of the Ponceau stained membrane displayed from 37 to 50 kDa to save space.Data were analysed by an unpaired t-test with Welch's correction where appropriate.Non-normally distributed data were analysed using a Mann-Whitney U test.Data are presented as mean ± SEM, with statistical significance determined at P < 0.05.

Fig. 5 .
Fig. 5. Decreased maximal cortical G6PDH activity and hippocampal PGI activity after one-month time-restricted feeding (TRF).Maximal enzymatic activity of hexokinase (HK), glucose-6-phosphate dehydrogenase (G6PDH), phospho-glucose isomerase (PGI), lactate dehydrogenase (LDH) (mU/mg protein) in the cortex [A-D] and hippocampus [E-H] were measured via continuous spectrophotometric assays in mice fed ad libitum (control, n = 8), or subjected to 8 h of timerestricted feeding (TRF, n = 9) for one month.Data were analysed by unpaired t-tests with Welch's correction where appropriate.Non-normally distributed data were analysed using a Mann-Whitney U test.Data are presented as mean ± SEM, with statistical significance determined at P < 0.05.*P < 0.05.

Fig. 7 .
Fig. 7. Lack of sizeable differences in markers of mitochondrial dynamics in the cortex or hippocampus after one-month time-restricted feeding (TRF).[A] Relative mitochondrial DNA content in the cortex, hippocampus, and liver; [B] relative mRNA expression of peroxisome proliferator-activated receptor gamma coactivator 1-alpha (Ppargc1a); [C] relative protein levels of mitofusin-1 (MFN1), [D] dynamin-related protein 1 (DRP1), and [E] mitochondrial dynamin-like GTPase long form (L-OPA1), short form (S-OPA1), long to short form (L-OPA1/S-OPA1), and total OPA1 in the cortex [C-E]; and hippocampus [F-H]; as well as [I, J] Western blots and Ponceau staining to indicate total protein levels of signaling proteins in the cortex [I] and hippocampus [J] of mice subjected to 8 h of time-restricted feeding (TRF, n = 9) for one month.All proteins of interest were normalised to the total protein levels (10-250 kDa) measured using Ponceau staining, with a representative portion of the Ponceau stained membrane displayed from 37 to 50 kDa.Data were analysed by an unpaired t-test with Welch's correction where appropriate.Non-normally distributed data were analysed using a Mann-Whitney U test.Data are presented as mean ± SEM, with statistical significance determined at P < 0.05.*P < 0.05.

Fig. 8 .
Fig. 8. Lack of differences in indicators or mRNA of antioxidant capacity and defence after one-month time-restricted feeding (TRF).[A] Trolox equivalent antioxidant capacity (TEAC) and [B] ferric reducing ability of plasma (FRAP) measured in the plasma [A-B] and cortex [C-D]; [E] oxygen radical absorbance capacity (ORAC) in the cortex; [F, G] superoxide dismutase (SOD) enzyme activity measured in the cytosolic fraction and mitochondrial enriched fraction of the cortex; [H] cytosolic catalase (CAT) enzyme activity in the cortex; [I, J] relative mRNA expression of antioxidant genes heme oxygenase 1 (Hmox1), glutathione peroxidase 1 (Gpx1), superoxide dismutase 1 (Sod1), superoxide dismutase 2 (Sod2), catalase (Cat) and kelch-like ECH-associated protein 1 (Keap1) in the cortex [I]; and liver [J] of mice fed ad libitum (control, n = 8), or subjected to 8 h of time-restricted feeding (TRF, n = 9) for one month.Data were analysed by an unpaired t-test with Welch's correction where appropriate.Non-normally distributed data were analysed using a Mann-Whitney U test.Data are presented as mean ± SEM, with statistical significance determined at P < 0.05.