Ketogenic diet-produced β-hydroxybutyric acid accumulates brain GABA and increases GABA/glutamate ratio to inhibit epilepsy

Abstract

Ketogenic diet (KD) alleviates refractory epilepsy and reduces seizures in children. However, the metabolic/cell biologic mechanisms by which the KD exerts its antiepileptic efficacy remain elusive. Herein, we report that KD-produced β-hydroxybutyric acid (BHB) augments brain gamma-aminobutyric acid (GABA) and the GABA/glutamate ratio to inhibit epilepsy. The KD ameliorated pentetrazol-induced epilepsy in mice. Mechanistically, KD-produced BHB, but not other ketone bodies, inhibited HDAC1/HDAC2, increased H3K27 acetylation, and transcriptionally upregulated SIRT4 and glutamate decarboxylase 1 (GAD1). BHB-induced SIRT4 de-carbamylated and inactivated glutamate dehydrogenase to preserve glutamate for GABA synthesis, and GAD1 upregulation increased mouse brain GABA/glutamate ratio to inhibit neuron excitation. BHB administration in mice inhibited epilepsy induced by pentetrazol. BHB-mediated relief of epilepsy required high GABA level and GABA/glutamate ratio. These results identified BHB as the major antiepileptic metabolite of the KD and suggested that BHB may serve as an alternative and less toxic antiepileptic agent than KD.

Fig. 1. KD increased GABA, glutamate, and GABA/glutamate ratio in mouse brains.

a Schematic diagram of the timeline of dietary exposure and PTZ treatment of mice and the Racine scale used to assess the intensity of epilepsy in mice. b–e PTZ-induced epilepsy in mice. KD- and ND-fed mice were intraperitoneally administered with a shot of PTZ before being monitored for epilepsy symptoms that were staged by the Racine scale. Latency time (b), the highest level (c), duration of (d) and frequency of seizures (e) were monitored. Data are means ± SD, n = 10 mice in each treatment. f The KD weakened neuronal activity in mice. The representative EEG recordings of WT mice fed the ND or KD with epileptiform activity are shown. g Pathway enrichment analysis showing metabolic pathways affected by the KD. The rich factors represent the ratio of differentially expressed protein in this pathway annotated by number. h The KD increased GABA and glutamate in mouse brains. Levels of GABA, glutamate, glutamine and aspartate were determined in the brains of C57 mice fed with either ND or KD. n = 6 mice. i The KD increased the GABA/glutamate ratio in mouse brains. The ratio of GABA/glutamate was compared between ND- and KD-fed mice. n = 6 mice.

Fig. 2. KD relieved epilepsy by changing GDH, GAD1, and SIRT4.

a Diagram of the enzymes involved in glutamate/GABA metabolism pathways. Enzymes studied in the current study (GDH and GAD1) are highlighted. b, c The KD upregulated GAD1 expression. Protein (b) and mRNA (c) levels of glutamate/GABA metabolism enzymes were determined in the brains of mice fed ND or KD. n = 4 mice. d KD induced SIRT4 and GAD1 expression. The histogram reflects the protein levels of SIRT4 and GAD1. n = 3 mice. e GAD1 had a limited impact on glutamate levels. GAD1 was overexpressed in U87MG cells; GABA and glutamate levels were compared between overexpressed and control U87MG cells. f The KD induced SIRT4 expression. SIRT4 levels in mouse brains were compared between KD- and ND-fed mice. n = 4 mice. g–k SIRT4 was required for the KD to relieve epilepsy in mice. KD- and ND-fed WT and Sirt4^–/– mice were each administered with an intraperitoneal PTZ injection before being monitored for epilepsy symptoms. Latency time (g), duration (h), the highest level (i), and frequency of seizures (j) and EEG (k) were monitored. Data were means ± SD, n = 10 mice in each treatment.

Fig. 3. KD-produced BHB activated SIRT4 and GAD1 transcription.

a, b The KD increased mice ketone body levels. Serum (a) and brain (b) levels of the BHB, AcAc, and acetone were detected in ND- and KD-fed mice. c, d BHB increased SIRT4 and GAD1 expression levels. U87MG cells were treated with ketone bodies and the protein (c) and mRNA (d) levels of SIRT4 and GAD1 were measured. e, f BHB induced SIRT4 and GAD1 expression in a dose-dependent manner. Protein (e) and mRNA (f) levels of SIRT4 and GAD1 were determined in HT22 cells that were treated with BHB at the indicated concentrations. g BDH1 overexpression decreased BHB levels. The BHB levels were measured in HT22, U87MG and Neuro-2a cells and in BDH1-overexpressing HT22 cells, U87MG and Neuro-2a cells. h, i BDH1 overexpression decreased SIRT4 and GAD1 expression. Protein (h) and mRNA (i) levels of SIRT4 and GAD1 were detected in HT22 cells, U87MG, and Neuro-2a cells and in BDH1-overexpressing HT22 cells, U87MG and Neuro-2a cells. j–l BDH1 knockdown increased BHB levels and increased SIRT4 and GAD1. BDH1 was silenced with independent siRNAs in HT22 and U87MG cells. BHB levels (j), protein (k), and mRNA(l) levels of SIRT4 and GAD1 were compared between siRNA-untreated and -treated cells.

Fig. 4. BHB inhibited HDAC1/HDAC2 to activate SIRT4 and GAD1 transcription.

a mRNA degradation did not account for the BHB-promoted SIRT4 and GAD1 mRNA increase. SIRT4 mRNA in U87MG cells was detected at the indicated time points with 10 mM actinomycin D. b BHB increased chromatin accessibility of the SIRT4 promoter. Quantitative PCR was performed on DNase I-pretreated nuclei of U87MG cells that were treated with BHB with the indicated concentrations. c BHB specifically elevated H3K27Ac. Histone acetylation levels were assayed in U87MG cells and in BHB- or trichostatin A (TSA)-treated U87MG cells, employing site-specific and pan-anti-acetyl-lysine (Ac) antibodies. d BHB-treatment elevated H3K27Ac at the promoter regions of SIRT4 and GAD1. The H3K27Ac antibody was used for immunoprecipitation (IP) after BHB treatment, and qRT-PCR of SIRT4 and GAD1 promoter output was performed. e, f The HDAC1/HDAC2-specific inhibitor FK228 increased SIRT4 and GAD1 protein (e) and mRNA (f) levels. The ability of FK228 to upregulate SIRT4 and GAD1 expression was tested in U87MG cells. g, h BHB elevated SIRT4 and GAD1 expression, dependent on HDAC1 and HDAC2 acetylation. The effects of BHB (10 mM) treatment on SIRT4 and GAD1 expression were assayed in HeLa cells with HDAC1, HDAC2, or double knockout of both (g, quantified 4 replicates in right), and in U87MG cells overexpressing HDAC1, HDAC2, or both (h). i HDAC1 and HDAC2 knockout enhanced H3K27Ac at the SIRT4 promoter. ChIP assays were performed for the promoter region of SIRT4, using the H3K27Ac antibody for IP and qRT-PCR for the SIRT4 promoter output.

Fig. 5. Lysine carbamylation activated GDH.

a, b The activity of the ADP-ribosylation-null GDH^C119G mutant was under regulation by SIRT4. Flag-tagged GDH or GDH^C119G was ectopically expressed in HEK293T cells or SIRT4-silenced HEK293T cells (a). Flag-tagged GDH or GDH^C119G was ectopically expressed in HEK293T cells or SIRT4-overexpressing HEK293T cells (b). The specific activities of GDH and the GDH^C119G mutant were determined and compared. c GDH interacted with CPS1. Flag-tagged CPS1 or GDH was ectopically expressed in HEK293T cells together with individual HA-tagged GDH or CPS1 as indicated. Flag bead- or HA bead-purified proteins were subject to western blotting analysis. d Carbamylation of endogenous GDH relied on CPS1 to generate CP. Endogenous GDH of HeLa cells was purified via IP with Protein A beads conjugated with the GDH antibody, and carbamylation of the purified GDH was detected. e KD decreased mouse brain mitochondrial CP-K levels. Mitochondrial CP-K levels were detected in brains of mice fed with either ND or KD. f Carbamylation activated GDH. Flag-tagged GDH was ectopically expressed in HEK293T cells treated with CP (5 mM) and purified via IP with Flag beads. The specific activity of GDH was determined. g GDH K162 was carbamylated. MS/MS spectra from tryptic peptide libraries of HEK293T cells matching MS/MS spectra from synthetic peptide (composed of ¹²C, ¹H, ¹⁴N, and ¹⁶O) identified GDH CP-K162. h Screening of the major regulatory carbamylation sites in GDH. Detected carbamylated residues were each mutated to arginine (R), and the specific activity of each mutant GDH was compared with that of WT GDH. i In vitro formation of CP-K162. The synthetic GDH K162-containing peptide was treated with 10 mM CP, followed by mass spectrometry analysis.

Fig. 6. SIRT4 decarbamylated and inactivated GDH to increase glutamate and GABA and not the GABA/glutamate ratio.

a SIRT4 decarbamylated GDH CP-K162. RP-HPLC analyses were used to detect decarbamylation of the synthetic GDH CP-K162 peptide catalyzed by SIRT4. b SIRT4 decarbamylated GDH. HA-tagged mitochondrial SIRT3, SIRT4, and SIRT5 were each individually expressed in untreated and CP (5 mM)-treated HEK293T cells. The GDH ectopically expressed in these cells were purified and subject to CP-K and specific activity analysis. c, d K162 was the major regulatory carbamylation site in GDH. Ectopically expressed GDH and GDH^K162R mutant from HEK293T cells and SIRT4 knockdown (c) or SIRT4 overexpression (d) HEK293T cells assayed for CP-K and specific activities. e GDH in Sirt4^–/– mouse brain was more carbamylated. The GDH protein was purified using Protein A beads conjugated with the GDH antibody from the brains of WT and Sirt4^–/– mice that were fed either ND or KD. GDH carbamylation was detected for purified GDH. f GDH was inactivated by the KD only in WT mouse brains. The WT and Sirt4^–/– mice were each fed the ND and KD from week 8. The mouse brain GDH purified at week 20 was assayed for specific activity. g SIRT4 knockdown decreased GABA and glutamate and increased α-KG. Culture media was supplemented with 2 mM ¹³C-glutamine. The M + 5 glutamate and α-KG (carbons fully labeled) and M + 4 GABA (carbons fully labeled) were determined in the mitochondria and cytoplasm of WT and SIRT4 knockdown U87MG cells after 6 h of chasing. h Sirt4 ablation caused more pronounced GABA elevation than glutamate elevation. The levels of GABA and glutamate were compared between WT and Sirt4^–/– mouse brains. i Sirt4 ablation did not alter GABA/glutamate ratio in mouse brains. GABA/glutamate ratios were detected in WT and Sirt4^–/– mouse brains.

Fig. 7. BHB relieved epilepsy in mice.

a BHB increased the GABA/glutamate ratio in WT and Sirt4^–/– mouse brains. Data were represented as individual values. b–e BHB alone SIRT4-dependently relieved seizure phenotypes. The ND- and BHB-fed WT and Sirt4^–/– mice were subject to measurements after they were administered with PTZ. Latency time (b), the highest level (c), the frequency (d) and the duration of seizures (e) were monitored. Data were means ± SD, n = 10 mice in each treatment. f–k GDH inhibition potentiated BHB to relieve seizures. BHB or the GDH inhibitor EGCG alone, or BHB together with EGCG were employed to treat Sirt4^–/– mice. The seizure-relieving effects of these treatments were monitored after mice were administered with PTZ. Brain GABA and glutamate (f) and GABA/glutamate ratio (g), latency time (h), the highest level of seizures (i), the frequency of seizures (j) and the duration of seizures (k) were monitored. Data were means ± SD, n = 10 mice in each treatment. l Schematic illustration showing KD upregulates brain GABA levels and GABA/glutamate ratio. Compared to ND, KD promotes BHB, which inhibits HDAC1/HDAC2 to upregulate SIRT4 and GAD1 transcription, results in inactivated GDH and activated GAD1, and consequently higher GABA and GABA/glutamate ratio.