Characterization of β-Hydroxybutyrate as a Cell Autonomous Fuel for Active Excitatory and Inhibitory Neurons: β-Hydroxybutyrate as a Fuel for Active Neurons

Abstract

The ketogenic diet is an effective treatment for drug-resistant epilepsy, but the therapeutic mechanisms are poorly understood. Although ketones are able to fuel the brain, it is not known whether ketones are directly metabolized by neurons on a time scale sufficiently rapid to fuel the bioenergetic demands of sustained synaptic transmission. Here, we show that nerve terminals can use the ketone β-hydroxybutyrate in a cell- autonomous fashion to support neurotransmission in both excitatory and inhibitory nerve terminals and that this flexibility relies on Ca²⁺ dependent upregulation of mitochondrial metabolism. Using a genetically encoded ATP sensor, we show that inhibitory axons fueled by ketones sustain much higher ATP levels under steady state conditions than excitatory axons, but that the kinetics of ATP production following activity are slower when using ketones as fuel compared to lactate/pyruvate for both excitatory and inhibitory neurons.

Keywords: BHB; GABA; ketone; metabolism; mitochondria; synapse.

Figure 1.. Neurons cell autonomously process the ketone β-hydroxybutyrate as fuel.

A) Synaptic vesicle recycling following 100 action potential (AP), 10 Hz stimulus in excitatory neurons expressing CaMKII synaptophysin-pHluorin (physin-pHluorin), while fueled by 5 mM glucose or 5 mM β-hydroxybutyrate (BHB) (n=9, error bands SEM). B) Synaptic vesicle cycling in inhibitory neurons expressing hDlxI56i physin-pHluorin and fueled by glucose or BHB during 100 AP, 10 Hz stimulus (n=4, error bands SEM). C) Schematic of ketone catabolism in neurons by β-hydroxybutyrate dehydrogenase 1 (BDH1), succinyl-CoA:3-ketoacid CoA transferase (SCOT), and thiolase II (T2). D) Representative Western blot against BDH1 in WT cortical culture or culture treated with BDH1 shRNA virus for 10 days (full blots in Extended Data Figure 1–1, −2, −3, −4 ,−5, −6 , −7, −8). E) Quantification of BDH1 knockdown efficiency in Western blot, normalized to average signal of uninfected controls (N=4 cultures, n=4 technical replicates, mean=0.6238. Error bars SEM, *p=0.011, Wilcoxon matched-pairs signed rank test). F) Synaptic vesicle recycling following 100 AP, 10 Hz stimulus in excitatory neurons expressing physin-pHluorin and shRNA against BDH1, while fueled by 5 mM glucose or BHB (n=4, error bands SEM). G) Synaptic vesicle cycling in inhibitory neurons expressing physin-pHluorin with and without shRNA against BDH1, while fueled by glucose or BHB during 100 AP, 10 Hz stimulus (n=4, error bands SEM).

Figure 2.. Inhibitory neurons using lactate/pyruvate as fuel perform better than those relying on β-hydroxybutyrate.

A) Quantification of iATPSnFR2.0/HALO ratio in excitatory (CaMKII) and inhibitory (hDlxI56i) neurons prior to stimulation in β-hydroxybutyrate (n=12 excitatory, mean=0.3371; n=10 inhibitory, mean=1.288; Error bars SEM, *p=0.0358, Mann-Whitney test). B) Fluorescence ratio traces of excitatory and inhibitory neurons in β-hydroxybutyrate during stimulation with 600 action potentials (AP) at 10 Hz (n=12 excitatory, n=10 inhibitory, error bands SEM). C) Fluorescence ratio traces of excitatory and inhibitory neurons in lactate/pyruvate during stimulation with 600 AP at 10 Hz (n=5 excitatory, n=5 inhibitory, error bands SEM). D) Fluorescence ratio recovery in excitatory neurons following stimulation in lactate/pyruvate or β-hydroxybutyrate (n=6 lactate/pyruvate, mean=1.736 %; n=12 β-hydroxybutyrate, mean=2.554 %; Error bars SEM, n.s., p=0.682, Mann-Whitney test). E) Fluorescence ratio recovery in inhibitory neurons following stimulation (n=6 lactate/pyruvate, mean=6.949 %; n=9 β-hydroxybutyrate, mean=1.463 %; Error bars SEM, *p=0.036, Mann-Whitney test).

Figure 3.. Excitatory and inhibitory neurons perform vesicle recycling better in lactate/pyruvate than in β-hydroxybutyrate.

A) Representative traces of excitatory and inhibitory neurons expressing synaptophysin-pHluorin (physin-pHluorin) stimulated with 100 action potentials (AP) every minute at 10 Hz, fueled by lactate and pyruvate (LP; 1.25 mM each) in presence of glycolytic inhibitor koningic acid (KA). B) Representative traces of excitatory and inhibitory neurons fueled by β-hydroxybutyrate (BHB; 5 mM racemic mixture) with 100 AP, 10 Hz stimulation in KA and glucose. C) Survival analysis of number of 100 AP, 10 Hz stimulations before reaching endocytic block of > 50% at the specified timepoint of 3 endocytic time constants after stimulation (n=18 inhibitory KA, median survival=4.5 stimulations; n=16 excitatory KA, median survival=3 stimulations; n=16 inhibitory BHB/gluc+KA, median survival=11.5 stimulations; n=17 excitatory BHB/gluc+KA, median survival=9 stimulations; n=10 inhibitory LP/gluc+KA, median survival=undefined/16+ stimulations; n=11 excitatory LP/gluc+KA, median survival=undefined/16+ stimulations; **p<0.005, Kaplan-Meier analysis).

Figure 4.. Pyruvate dehydrogenase activation by Ca²⁺ is a key point of mitochondrial metabolic regulation in active excitatory and inhibitory neurons.

A) Diagram of activity dependent upregulation of mitochondrial metabolism. B-C) Fluorescence traces of excitatory neurons expressing synaptophysin-pHluorin (physin-pHluorin) and MCU shRNA stimulated with 100 action potentials (AP) at 10 Hz, in presence of 5 mM glucose followed by 10 μM koningic acid (KA) and either B) 1.25 mM lactate and 1.25 mM pyruvate (n=7, error bands SEM. Data pooled with neurons in lactate/pyruvate without glucose and KA) or C) 5 mM β-hydroxybutyrate (n=7, error bands SEM). D-E) Fluorescence traces of inhibitory neurons expressing physin-pHluorin and MCU shRNA, stimulated with 100 AP at 10 Hz in 5 mM glucose followed by 10 μM KA and either D) 1.25 mM lactate and 1.25 mM pyruvate (n=6 MCU KD, error bands SEM. Data pooled with neurons in lactate/pyruvate without glucose and KA) or E) 5 mM β-hydroxybutyrate (n=8 MCU KD, error bands SEM).