A subset of synaptic transmission events is coupled to acetyl coenzyme A production

Abstract

Biological principles sustain the inference that synaptic function is coupled to neural metabolism, but the precise relationship between these two activities is not known. For example, it is unclear whether all synaptic transmission events are uniformly dependent on metabolic flux. Most synapses use glutamate, and the principal metabolic function of the brain is glucose oxidation, which starts with glycolysis. Thus, we asked how glutamatergic synaptic currents are modified by partial deficiency of the main glycolytic enzyme pyruvate dehydrogenase (PDH), which generates the intermediary metabolism product acetyl coenzyme A (acetyl-CoA). Using brain slices obtained from mice that were genetically modified to harbor a behaviorally relevant degree of PDH suppression, we also asked whether such impact is indeed metabolic via the bypassing of PDH with a glycolysis-independent acetyl-CoA substrate. We analyzed spontaneous synaptic currents under recording conditions that minimize artificial metabolic augmentation. Principal component analysis identified synaptic charge transfer as the major difference between a subset of wild-type and PDH-deficiency (PDHD) postsynaptic currents. This was due to reduced charge transfer as well as diminished current rise and decay times. The alternate acetyl-CoA source acetate rapidly restored these features but only for events of large amplitude as revealed by correlational and kernel density analyses. Application of tetrodotoxin to block large-amplitude events evoked by action potentials removed synaptic event charge transfer and decay-time differences between wild-type and PDHD neurons. These results suggest that glucose metabolic flux and excitatory transmission are intimately coupled for synaptic events characterized by large current amplitude.

NEW & NOTEWORTHY In all tissues, metabolism and excitation are coupled but the details of this relationship remain elusive. Using a brain-targeted genetic approach in mice, reduction of pyruvate dehydrogenase, a major gateway in glucose metabolism, leads to changes that affect the synaptic event charge associated primarily with large excitatory (i.e., glutamate mediated) synaptic potentials. This can be modified in the direction of normal using the alternative fuel acetate, indicating that this phenomenon depends on rapid metabolic flux.

INTRODUCTION

Neural electrical activity, including that arising from synaptic function, depends on glucose metabolism as its principal source of carbon and energy. In most cells of the organism, including the brain, glucose is initially metabolized to pyruvate during glycolysis (1, 2). Subsequent entry of pyruvate into the tricarboxylic acid (TCA) cycle necessitates its conversion into acetyl coenzyme A (acetyl-CoA) in a reaction catalyzed by pyruvate dehydrogenase (PDH) (Fig. 1A). This step constitutes a quantitatively important rate-limiting reaction for the brain, as it steers the transit of the majority of cerebral glucose through metabolic consumption by the TCA cycle (3). As may be expected, in vertebrates absence of PDH is incompatible with life, but less pronounced decreases in PDH activity primarily impact the nervous system because of this unique relevance of glucose metabolism to brain cells. In fact, the critical dependence of neural cells on TCA cycle flux for excitability and even viability is made patent by the severe human neurological diseases that stem from relatively mild pyruvate metabolism or TCA cycle enzyme activity reductions, which lead to epilepsy and profound intellectual disability and, for some of the disorders, to early mortality (4). Murine PDH deficiency (PDHD) and TCA cycle defects are, from a comparative neural point of view, as severe as their counterpart human diseases including their association with exacerbated mortality (5, 6). Nevertheless, a modest degree of PDH deficiency achieved by selective genetic reduction of brain PDH allows for sufficient animal viability to conduct mechanistic investigation before the cell death process ensues (7). Whole cell patch-clamp analysis of synaptic currents is well suited to study the relationship between neural metabolism and synaptic physiology (Fig. 1B), as these currents can report several aspects of neurotransmitter activity. In this experimental configuration, the patch pipette records the natural transmission of synaptic potentials to the cell soma (Fig. 1B) provided that the cell content is not rapidly dialyzed by the pipette solution and that other metabolic conditions are also maintained or approximated. In this context, the spontaneous synaptic release of the TCA cycle byproduct glutamate generates a synaptic current waveform (Fig. 1C) that reflects total synaptic charge transfer during single-synapse activation, a parameter that is dependent on the amount of glutamate available in the synaptic cleft (8, 9). Glutamate is a byproduct of carbon rearrangement directly stemming from TCA cycle reactions and its import into vesicles and release from the synapse consumes ATP. Therefore, reduced flux through the cycle can lead to decreased glutamate abundance and action (7). Glutamate uptake from the cleft is also dependent on ATP use (10–12), such that reduced glutamate uptake results in prolongation of the decay time of synaptic currents (8). Furthermore, the rise-time (or time to peak) for a synaptic transmission event (or current waveform) increases with increasing distance from the soma and may thus reflect neuronal arborization abnormalities. Such structural abnormalities are in fact characteristic of murine PDH deficiency (13). Last, patch-clamp recordings can also quantify reversal of synaptic current deficits under a metabolic intervention such as an alternate TCA cycle substrate, thus helping differentiate between metabolism-independent and modifiable features with potential therapeutic relevance. Therefore, examining the impact of TCA cycle dysfunction and modulation on spontaneous postsynaptic excitatory currents can inform aspects of metabolism-excitability coupling and pave the way for the treatment of neurometabolic disorders associated with excitable dysfunction.

Figure 1.

Spontaneous excitatory postsynaptic currents in pyruvate dehydrogenase deficiency. A: schematic depiction of the metabolism of glucose into pyruvate and its subsequent oxidation in the TCA (or Krebs) cycle. B, top: representation of narrow pipette whole cell patch-clamp recordings from cortical layer 4/5 neurons. The internal solution of pipettes did not contain energy substrates ATP, GTP, or phosphocreatinine. Bottom: example trace of a current step inducing adapting action potential firing from a putative pyramidal neuron. C: schematic spontaneous excitatory postsynaptic current (sEPSC) event. Labels indicate the features measured for each synaptic event. D: representative traces from a wild type (WT) cell and a cell from a mouse with PDHD illustrating large and smaller sEPSC events. Scale bar = 500 ms/10 pA. E, left: example plot of rise time vs. amplitude demonstrating a lack of filtering (i.e., no negative slope or negative correlation-coefficient r) of events with high rise times. Right: group data from a subset of recorded neurons that showed no filtering (WT n = 11 cells from 9 mice, PDHD n = 7 cells from 6 mice) for neurons used for analysis. The y-axis refers to correlation coefficient r derived as in the left panel for each cell recording. F: bar graph depicting cell series resistance (WT n = 11 cells from 9 mice, PDHD n = 7 cells from 6 mice) for neurons that were used for analysis. GTP, guanosine triphosphate; NS, not statistically significant; PDHD, pyruvate dehydrogenase-deficiency.

In this study, we set out to further characterize the metabolism-excitability relationship by studying single-cell excitatory synaptic activity in isolation from concurrent inhibition by GABAergic interneurons. We examine amino-hydroxy-methyl-isoxazolepropionic acid (AMPA)-mediated glutamatergic spontaneous excitatory postsynaptic currents (sEPSC) via whole cell patch-clamp recordings acquired in the presence of the GABA_A blocker picrotoxin and the NMDA blocker AP-5 in the context of PDH deficiency and of metabolic modulation using the alternate metabolic fuel acetate. We find, via principal component analysis, that total synaptic current recorded over time per synaptic event [also known as charge transfer (Fig. 1C)] largely accounts for the difference between wild-type (WT) and PDH-deficiency sEPSCs (Fig. 1D). Larger-amplitude sEPSCs are predominantly impacted in neurons with PDH deficiency. Specifically, spontaneous EPSCs from neurons with PDH deficiency exhibit a smaller charge transfer and decay time constant (τ). Alternate fueling with acetate rapidly ameliorates, for a subset of synaptic events, both charge transfer and event amplitude, with a predominant effect on the large-amplitude events. In support of the TCA cycle’s preferential impact on large-amplitude events, removing the presence of action potential evoked large-amplitude events removed differences in charge transfer and decay-times between WT and PDHD neurons. These results reveal that a subset of excitatory synaptic currents are closely dependent on oxidative metabolism and that synaptic dysfunction in PDH deficiency can be favorably modulated by alternative fuel.

METHODS

All animal studies were approved and followed guidelines set by the Institutional Animal Care and Use Committee (IACUC) of UT Southwestern Medical Center.

PDH-Deficiency Mice

Generation of C57BL6 Pdha1-deficiency mice was described previously (7). In brief, Pdha1^flox/flox females (14) were bred with hGFAP-Cre males to generate hemizygous Pdha1^flox/y male mice used in the experiments. Pdha1^flox/y mice exhibited smaller brains, excess mortality after P25, decreased Pdha1 protein levels and PDH enzymatic activity, and decreased Pdha1 expression in neurons and astrocytes (7).

Acute Brain Slice Preparation

Brain slices were prepared for whole cell patch-clamp as described previously (7), using 16- to 18-day-old mice. In brief, mice were anesthetized with ketamine (100 mg/kg), xylazine (10 mg/kg), and acepromazine (2 mg/kg). Ketamine is an NMDA antagonist and can attenuate excitotoxicity (15). Anesthetized mice underwent transcardiac perfusion with chilled N-Methyl-d-glucamine (NMDG) dissection buffer containing (in mM): 93 NMDG, 2.5 KCl, 1.2 NaH₂PO₄, 30 NaHCO₃, 20 HEPES, 10 d-glucose, 0.5 CaCl₂, and 5 MgSO₄, aerated with 95% O₂/5% CO₂. Coronal 400 µm brain slices were prepared using a Leica VT 1200S vibratome. Slices were transferred to a chamber containing artificial cerebrospinal fluid (ACSF) and incubated at room temperature for 2.5 h. The ACSF contained (in mM): 118 NaCl, 2.5 KCl, 26 NaHCO₃, 1.25 NaH₂PO₄, 2 CaCl₂, 2 MgSO₄, and 10 d-glucose and was aerated with 95% O₂/5% CO₂.

Whole Cell Patch-Clamp Recording

Following incubation (16), the slices were transferred to a recording chamber. ACSF bathing the slices in the chamber was supplemented with the GABA antagonist picrotoxin (100 µM) and with 2-amino-5-phosphonopentanoic acid (AP-5, 100 µM), a glutamate competitive antagonist at the NMDA receptor. For miniature excitatory postsynaptic current (mEPSC) recordings, 1 µM tetrodotoxin was added to the ACSF. Recordings proceeded at room temperature (24°C) since none of the slices from PDHD mice contained surviving cells when maintained at 30°C. The rate of ACSF chamber perfusion was 3 mL/min. A differential interference contrast (DIC) microscope (Eclipse FN1, Nikon) was used to locate the barrel cortex and to visualize the cells located in the center of barrels (cortical layer 4/5). Putative pyramidal neurons were characterized by their characteristic pyramidal shape and the presence of apical dendrites. Their putative pyramidal nature was further corroborated by their electrophysiological properties. A borosilicate glass tube with an outer diameter of 1.5 mm and an inner diameter of 0.86 mm was used to prepare pipettes with a P-97 Sutter Instruments Micropipette puller. Narrow pipette tips (14–20 MΩ resistance) were used to minimize or delay intracellular fluid dialysis on metabolism relevant intracellular currents by as long as 20 min (17). Only the first 12–16 min of recording were analyzed. The pipette internal solution lacked ATP, guanosine triphosphate (GTP), or phosphocreatine and contained (in mM): 125 K-gluconate, 2.6 KCl, 1.3 NaCl, 10 HEPES, 0.1 EGTA, and 15 sucrose.

Data acquisition involved a CV-7B headstage (Axon Instruments), an Axon Multiclamp 700B amplifier, and an Axon Digidata 1440 A digitizer. Recordings were sampled at 10,000 Hz, with a low-pass Bessel filter set at 3,000 Hz. After obtaining a gigaohm seal (>4 GΩ) with the voltage clamp set with holding potential −65 mV, alternating 50–100 µS, 1 V pulses and strong, brief suction allowed for breakthrough of the cell membrane. Subsequently, more gentle suction was applied to avoid resealing. Series resistance was compensated (40%–60%) to remain below 30 MΩ. Initially, action potentials were elicited by 600-ms current steps to ascertained if the cell fired action potentials that adapted over time. Subsequently, sEPSCs were recorded in voltage clamp mode (holding potential −65 mV) for 5 min (baseline period), followed by acetate-containing ACSF [in mM: 113 NaCl, 2.5 KCl, 26 NaHCO₃, 1.25 NaH₂PO₄, 2 CaCl_2, 2 MgSO₄, and 5 C₂H₃NaO₂, which maintained the same electrochemical driving force as ACSF (18)] where indicated, first allowing 2 min for acetate perfusate equilibration before recording for 5 more min. Series resistance was monitored throughout the recording every 2 min with a 600 ms, −10 mV step from holding potential −65 mV. The cell was removed from analysis if series resistance varied more than 25% in the course of a recording. Only one cell was recorded from in each cortical slice. Reagents were procured from Sigma-Aldrich (St. Louis, MO).

The majority of neurons in the barrel cortex include excitatory (∼ 86%) or parvalbumin-positive (PV+) fast-spiking inhibitory interneurons (∼7%) (19). Whole cell patch-clamped cells were classified as putative excitatory pyramidal based on: 1) appearance on DIC microscopy: a pyramidal shape with apical dendrites, rather than an oval or spherical shape without an apical dendrite as is typical for a PV+ interneuron; 2) distinct adaptation of action potential frequency after current injection; 3) lack of the high-frequency firing characteristic of PV+ interneurons; and 4) broader action potential width and prolonged after-hyperpolarization compared with PV+ interneurons.

Data Analysis

Analysis was conducted offline with MiniAnalysis (Synaptosoft), Prism 6.0 (GraphPad software) and with standard MATLAB (MathWorks) functions. Data were low-pass filtered with a Bessel filter at 1,000 Hz before sEPSC or mEPSC detection analysis. Excitatory post synaptic potentials are referred to as sEPSCs if they include action potential-induced events and as mEPSCs (miniature excitatory post synaptic currents) if action potential-induced events are prevented with tetrodotoxin in the ACSF. For detection of sEPSCs in MiniAnalysis, first the root-mean-square (RMS) baseline noise in a 100 ms event-free recording was calculated. A value in pA that was three times RMS baseline noise or 5 pA (whichever was greater) was used as a first-pass amplitude threshold to detect sEPSC events. Each detected event shape (i.e., temporal course) was then scrutinized visually to confirm or rule out sEPSC attribution. The detection threshold was kept constant for each cell recording. Prism 6.0 was used for cumulative distribution function analysis. All other analyses, including principal component analysis, employed publicly available functions in MATLAB including functions “pca” for principal component analysis and “ksdensity” for kernel density estimate using the default parameters provided by MATLAB. Kernel density estimates were set to an area under the curve equal to one. y-Axis values were multiplied by 100 for comparison purposes.

Statistical Analysis

Prism 6.0 or MATLAB function “ttest2” was used for statistical analysis including independent Student’s t test with an assumption of unequal variances between two groups. For comparison of post-acetate parameter values with baseline values obtained from the same mouse, a paired or ratio-paired sample t test was used. For nonparametric data, Prism 6.0 or MATLAB function “ranksum” was used to conduct a Mann–Whitney U test. For comparison of cumulative distribution probabilities, the Kolmogorov–Smirnov test was used. The significance level (α) was set at 0.05, and all tests of significance were two sided. Data given with error ranges represent means ± SE (standard error of mean) values.

RESULTS

Recording from Neurons from Mice with PDH Deficiency

After systematically varying various parameters, the selected recording conditions departed from common technique in brain slice whole cell patch-clamp. To reduce artificial metabolic interference with the recordings (Fig. 1B), the pipette solution was devoid of energy substrates such as ATP, GTP, and phosphocreatine. Furthermore, pipette opening diameters were unusually narrow to limit the replacement of the cellular metabolic milieu by unintended cell dialysis with pipette solution. Pipette resistances were thus increased to allow only a negligible degree of dialysis as established in 20 min recordings that specifically assessed influences on energy metabolism (17). From all recordings performed (WT n = 14 cells; PDH deficiency or PDHD n = 15), we analyzed data from only a subset (WT n = 11 cells; PDHD n = 7 cells) that demonstrated lack of current filtering or distortion by the pipette due to sufficiently high series resistance while maintaining effective clamping (Fig. 1E). Higher pipette series resistance can distort voltage space clamp and artificially reduce synaptic currents arising from more distal locations, resulting in a negative slope when rise-time is plotted against amplitude (20). We thus confirmed that all the cells selected for analysis exhibited a positive slope in this relationship (Fig. 1E). All the series resistances were within the acceptable range and did not differ between WT and PDHD cells (Fig. 1F).

Principal Component Analysis of sEPSCs: Charge Transfer in PDHD and WT Neurons

Because energy metabolism can impact individual synaptic current temporal course (event shape) in several, partially dependent ways (8), we conducted a broad search for potential sEPSC differences between PDHD and WT using principal component analysis (PCA). We treated sEPSCs as multiple-dimensional data containing partially interdependent variables and reasoned that PCA can reduce the number of dimensions (or influences on sEPSCs) of interest by combining contributions from different variables into one dimension. The first principal component obtained in this manner accounts, to a large extent, for any variance in the data. Using averaged values from each cell for a series of physiologically-relevant parameters, we observed that PCA scores for PDHD for the first component were different from WTs (Fig. 2, A and B). The variable that contributed most to this first principal component was sEPSC area (or total current flow or charge transfer for the duration of the synaptic event; Fig. 2C). We also considered the first 25 synaptic events regardless of source cell, which may introduce additional variability, and asked if the differences between WT and PDHD neurons were robust enough to be reflected in the analysis of these pooled data containing all of the individual synaptic events. PCA analysis of the pooled data also demonstrated first principal component values that differentiated PDHD from WT sEPSCs (Fig. 2, D and E), with the variable that contributed the most to the first principal component being again sEPSC area (Fig. 2F).

Figure 2.

Principal component analysis (PCA) of event waveform features. A: scatter plot of scores derived from principal component analysis of average spontaneous excitatory postsynaptic current (sEPSC) frequency, amplitude, decay τ, area, and rise time. Each dot represents the score for a single cell. B: bar graph of the first principal component scores shown in A for each group (WT n = 11 cells from 9 mice, PDHD n = 7 cells from 6 mice). C: bars depict percentage contribution of waveform features to the first principal component. D: scatter plot of pooled scores derived from principal component analysis of the first 25 sEPSC events in a cell. Each dot represents the score for a single sEPSC event. Parameter features employed in PCA were sEPSC slope, amplitude, rise time, decay τ, and area. Each dot represents the score for a single cell. E: cumulative probability distribution of first principal component scores shown in D for each group. The P value reflects the outcome of a Kolmogorov–Smirnoff test for statistical significance. F: bar graph depicting percentage contribution of waveform features to the first principal component. *P < 0.05, unpaired two-tailed Student’s t test. PDHD, pyruvate dehydrogenase deficiency.

Reduced sEPSC Charge Transfer for Large Synaptic Events and Acetate Augmentation in PDHD

EPSC amplitude is directly influenced by the amount of glutamate released (8). Based on the PCA results, we hypothesized that sEPSC area was diminished in PDHD and found this to be the case, as average sEPSC area was reduced (Fig. 3A). Examination of the average distribution of sEPSC area using the Kolmogorov–Smirnoff test for differences in distribution indicated no significant change between the two groups. Because the mean can be influenced by higher values, we examined the data for changes within subset bins of area. A subset of the distribution displayed decreased charge transfer for larger sEPSC events, thus explaining the decrease in area noted when comparing average sEPSC areas.

Figure 3.

Synaptic charge transfer (sEPSC area) in WT and PDHD excitatory events. A, left: Average sEPSC area for all events in each cell (WT n = 11 cells from 9 mice, PDHD n = 7 cells from 6 mice). Middle: average cumulative distribution function of sEPSC area for each group. Right: percent change of cumulative distribution in PDHD mice compared with WT for each cumulative distribution bin across mice. Average WT sEPSC area for each cumulative distribution bin is depicted on the right-sided y-axis. *P < 0.05, unpaired two-tailed Student’s t test. B, left: dot plot showing impact of acetate on average sEPSC area for all events in each WT cell (n = 10 cells from 8 mice). Middle: average cumulative distribution function of sEPSC area for baseline and acetate treated sEPSC areas. Right: acetate-induced percent change of cumulative distribution in PDHD mice compared with baseline for each cumulative distribution bin across mice. Average baseline sEPSC area for each cumulative distribution bin is depicted on the right-sided y-axis. C, left: dot plot showing impact of acetate on average sEPSC area for all events in each PDHD cell (n = 6 cells from 5 mice). Middle: average cumulative distribution function of sEPSC area for baseline and acetate-treated sEPSC areas. Right: acetate-induced percent change of cumulative distribution in PDHD mice compared with baseline for each cumulative distribution bin across mice. Average baseline sEPSC area for each cumulative distribution bin is depicted on the right-sided y-axis. *P < 0.05, paired two-tailed Student’s t test. PDHD, pyruvate dehydrogenase deficiency; sEPSC, spontaneous excitatory postsynaptic current; WT, wild type.

Because acetate is a PDH-independent source of acetyl-CoA, we asked if fueling the TCA cycle with this alternate fuel influenced sEPSC charge transfer. We reason that, if the changes in sEPSCs are due to a simple reduction in metabolic flux through PDH, the provision of an alternative source of PDH metabolic product should restore synaptic event properties in a rapid timeframe compatible with acetate metabolism (s to min). Upon acetate addition, WT littermate neurons did not exhibit changes in average sEPSC area values, average cumulative probability or subset-divided sEPSC areas (Fig. 3B). In PDHD, average charge transfer from all events also exhibited no change (Fig. 3C). However, an sEPSC subset containing the largest events demonstrated a significant increase in area following acetate supply (Fig. 3C). Example sEPSC traces (Fig. 4) illustrate the decrease in area in PDHD under basal recording conditions and its increase with acetate. These findings differentiate the larger sEPSCs from other excitatory synaptic events by virtue of their metabolic dependence, as their associated charge transfer may decrease or increase in relation to TCA cycle activity stemming from the availability of acetyl-CoA.

Figure 4.

sEPSC waveforms for WT and PDHD neurons. A: representative waveforms from a WT (gray) and PDHD (black) neuron: Left: low amplitude waveforms (average of all events less than 14 pA in peak amplitude). Middle: large amplitude waveforms (average of all events greater than or equal to 14 pA in peak amplitude). Right: same waveforms as in the middle, normalized to the largest peak amplitude to discern changes in decay rate (if present). Scale bar: 5 ms/5 mV. B: example waveforms from baseline (solid line) and after administration of acetate (dotted) in a WT neuron: Left: low amplitude waveforms (average of all events less than 14 pA in peak amplitude). Middle: large amplitude waveforms (average of all events greater than or equal to 14 pA in peak amplitude). Right: same waveforms as in the middle, normalized to the largest peak amplitude to better discern changes in decay rate if present. Scale bar: 5 ms/5 mV. C: example waveforms from baseline (solid line) and after administration of acetate (dotted) in a PDHD neuron: Left: low amplitude waveforms (average of all events less than 14 pA in peak amplitude). Middle: large amplitude waveforms (average of all events greater than or equal to 14 pA in peak amplitude). Right: same waveforms as in the middle, normalized to the largest peak amplitude to discern any changes in decay rate. Scale bar: 5 ms/5 pA. PDHD, pyruvate dehydrogenase deficiency; sEPSC, spontaneous excitatory postsynaptic current; WT, wild type.

Decreased sEPSC Amplitude in PDHD and Acetate Restoration

One possible mechanism that explains the observed changes in sEPSC area in PDHD is a difference in event current amplitude. EPSC amplitude has been used to assess glutamatergic activation strength at the synapse and is influenced by the number of glutamatergic receptors (8, 21). We hypothesized that PDHD sEPSC amplitude may parallel the decreased area changes noted when compared with WT neurons. Contrary to this expectation, there was only a slight impact of PDH deficiency on sEPSC amplitude, with no change in average amplitude and with only a decrease in average cumulative probability distribution (Fig. 5A). Analysis of subsets of amplitudes did not show any changes for each subset bin. In WT neurons, administration of acetate did not impact the average cumulative probability distribution or subset data (Fig. 5B). In contrast, in PDHD, although the average values or average cumulative probability was also invariant, multiple subsets of amplitude bins displayed an increase after acetate (Fig. 5C and Fig. 6). These results reveal only a minimal decrease in amplitude for PDHD sEPSCs compared with WT and an enhancement of a subset of amplitudes with alternate fueling.

Figure 5.

sEPSC amplitude in WT and PDHD. A, left: average sEPSC amplitude for all events in each cell (WT n = 11 cells from 9 mice, PDHD n = 7 cells from 6 mice). Middle: average cumulative distribution function of sEPSC amplitude for each group. Right: percent change of cumulative distribution in PDHD mice compared with WT for each cumulative distribution bin across mice. Average WT sEPSC amplitude for each cumulative distribution bin is represented on the right-sided y-axis. B, left: dot plot showing effect of acetate on average sEPSC amplitude for all events in each WT cell (n = 10 cells from 8 mice). Middle: average cumulative distribution function of sEPSC amplitude for baseline and acetate treated sEPSC amplitudes. Right: acetate-induced percent change of cumulative distribution in PDHD mice compared with baseline for each cumulative distribution bin across mice. Average baseline sEPSC amplitude for each cumulative distribution bin is represented on the right-sided y-axis. C, left: dot plot showing effect of acetate on average sEPSC amplitude for all events in each PDHD cell (n = 6 cells from 5 mice). Middle: average cumulative distribution function of sEPSC amplitude for baseline and acetate-treated sEPSC amplitudes. Right: acetate-induced percent change of cumulative distribution in PDHD mice compared with baseline for each cumulative distribution bin across mice. Average baseline sEPSC amplitude for each cumulative distribution bin is depicted on the right-sided y-axis. *P < 0.05 paired two-tailed Student’s t test. PDHD, pyruvate dehydrogenase deficiency; sEPSC, spontaneous excitatory postsynaptic current; WT, wild type.

Figure 6.

Example WT and PDHD neuron recordings. Example traces from a WT cell and a cell from a mouse with PDHD to illustrate a slight increase in the occurrence of large events in a PDHD neuron but not in a WT neuron. Scale bar: 500 ms/10 pA. Black dots and triangles indicate small and large-amplitude events, respectively. PDHD, pyruvate dehydrogenase deficiency; WT, wild type.

sEPSC Decay and Rise Time Are Quicker for Some Event Subsets in PDHD and Do Not Change with Acetate

Our findings indicate that changes in sEPSC area were not simply explained by changes in sEPSC amplitude. Another factor that influences sEPSC area is the decay time of the EPSC. Thus, we asked if the time constant (τ) of decay parallels charge transfer changes in PDHD. Mean decay τ was decreased in PDHD neurons (Fig. 7A). Although the average cumulative probability distribution of decay τ did not change, a subset of bins exhibited decreased decay τ for larger events in PDHD, in keeping with the changes noted for current area. Acetate elicited no changes in WT neurons (Fig. 7B) and, in contrast to our expectations, also no change in PDHD neurons (Fig. 7C).

Figure 7.

sEPSC decay time constant (τ) of PDHD synaptic events. A, left: average sEPSC decay τ for all events in each cell (WT n = 11 cells from 9 mice, PDHD n = 7 cells from 6 mice). Middle: average cumulative distribution function of sEPSC decay τ for each group. Right: percent change of cumulative distribution in PDHD mice compared with WT for each cumulative distribution bin across mice. Average WT sEPSC decay τ for each cumulative distribution bin is represented on the right-sided y-axis. *P < 0.0.05 unpaired two-tailed Student’s t test. B, left: dot plot illustrating the effect of acetate on average sEPSC decay τ for all events in each WT cell (n = 10 cells from 8 mice). Middle: average cumulative distribution function of sEPSC decay τ for baseline and acetate treated sEPSC decay τ. Right: acetate-induced percent change of cumulative distribution in PDHD mice compared with baseline for each cumulative distribution bin across mice. Average baseline sEPSC decay τ for each cumulative distribution bin is represented on the right-sided y-axis. C, left: dot plot showing effect of acetate on average sEPSC decay τ for all events in each PDHD cell (n = 6 cells from 5 mice). Middle: average cumulative distribution function of sEPSC decay τ for baseline and acetate-treated sEPSC decay τ. Right: acetate-induced percent change of cumulative distribution in PDHD mice compared with baseline for each cumulative distribution bin across mice. Average baseline sEPSC decay τ for each cumulative distribution bin is indicated on the right-sided y-axis. PDHD, pyruvate dehydrogenase deficiency; sEPSC, spontaneous excitatory postsynaptic current; WT, wild type.

As sEPSC rise time to peak can also influence event area, we also examined it. There was no change in average values or average cumulative distribution for sEPSC rise times (Fig. 8A). However, a small proportion of shorter rise-time events was decreased in PDHD. Administration of acetate did not influence rise time in WT or PDH deficiency neurons (Fig. 8, B and C).

Figure 8.

sEPSC rise-time in PDHD. A, left: average sEPSC rise time for all events in each cell (WT n = 11 cells from 9 mice, PDHD n = 7 cells from 6 mice). Middle: average cumulative distribution function of sEPSC rise time for each group. Right: percent change of cumulative distribution in PDHD mice compared with WT for each cumulative distribution bin across mice. Average WT sEPSC rise time for each cumulative distribution bin is represented on the right-sided y-axis. *P < 0.0.05 unpaired two-tailed Student’s t test. B, left: dot plot illustrating the impact of acetate on average sEPSC rise time for all events in each WT cell (n = 10 cells from 8 mice). Middle: average cumulative distribution function of sEPSC rise time for baseline and acetate-treated sEPSC rise times. Right: acetate-induced percent change of cumulative distribution in PDHD mice compared with baseline for each cumulative distribution bin across mice. Average baseline sEPSC rise time for each cumulative distribution bin is represented on the right-sided y-axis. C, left: dot plot showing the impact of acetate on average sEPSC rise time for all events in each PDHD cell (n = 6 cells from 5 mice). Middle: average cumulative distribution function of sEPSC rise time for baseline and acetate-treated sEPSC rise times. Right: acetate-induced percent change of cumulative distribution in PDHD mice compared with baseline for each cumulative distribution bin across mice. Average baseline sEPSC rise time for each cumulative distribution bin is depicted on the right-sided y-axis. PDHD, pyruvate dehydrogenase deficiency; sEPSC, spontaneous excitatory postsynaptic current; WT, wild type.

Decay τ Contributes to Changes in Large Amplitude sEPSCs

Our results suggest that larger area synaptic events displayed bidirectional changes upon metabolic changes or modulation. We next asked if changes in area were attributable to bidirectional changes in decay τ, and examined area and decay τ after correcting for amplitude by analyzing amplitude bins. We found that, after correcting for amplitude in PDHD, a small subset of the large amplitude events displayed decreased area and decay τ (Fig. 9A). Administration of acetate resulted in an enhancement of area and decay τ for these large amplitude events (Fig. 9B). There was a correlation between the change in amplitude-adjusted decay τ and area when comparing PDHD with WT neurons, indicating that decay τ decrease contributed to area decrease (Fig. 9C). Similarly, alternate fueling with acetate made manifest a correlated increase in area and decay. These findings indicate that changes in synaptic charge transfer are influenced by current decay rate.

Figure 9.

Relationship between change in sEPSC decay τ and event area. A: for specific amplitude bins, percentage change in area, and decay τ from the mean of the WT group (WT n = 11 cells from 9 mice, PDHD n = 7 cells from 6 mice). Error bars represent means ± SE with mean at the center. *P < 0.0.05 unpaired two-tailed Student’s t test. B: for specific amplitude bins, percentage change in area and decay τ in PDHD sEPSCs from the mean of baseline recordings (n = 6 cells from 5 mice) after administration of acetate. Error bars represent means ± SE with mean at the center. *P < 0.0.05 paired two-tailed Student’s t test. C: scatter plot of average percentage change in decay τ vs. percent change in area for different amplitude bins. Each dot corresponds to an amplitude bin (the darkness of the dot increases with amplitude size) and its position on the graph is determined from the y-axis coordinates in A and B. Note, correlation between change in decay τ and area for comparison between PDHD and WT (Pearson’s r = 0.88) and for comparison between baseline and acetate (Pearson’s r = 0.97) in PDHD neurons. PDHD, pyruvate dehydrogenase deficiency; sEPSC, spontaneous excitatory postsynaptic current; WT, wild type.

Relationship between PDHD sEPSC Size or Event Interval in Relation to WT

We asked whether there was a relationship between the average degree of change and response to alternate fuel and size or frequency of events. We reasoned that for a cumulative probability distribution graphically represented, if every parameter value (e.g., sEPSC area plotted on the x-axis in a cumulative probability plot) in every cumulative distribution bin changed, two rising parallel cumulative probability lines appear, with the percentage difference between these line plots for each area bin exhibiting a flat line in the plot. Conversely, if only a subset of values changed consistently (as noted for decreased sEPSC areas for larger events), the scatter plot depicting the average percentage difference would gravitate toward a line with a negative slope and can be assayed for significance with a P value, depicting an inverse relationship between the magnitude of the parameter value and its difference from WT (Fig. 10A). Importantly, a reversal of this slope with acetate would indicate an opposite effect of TCA cycle fueling, with increased sEPSC area in the larger events. We found an inverse relationship in PDHD between average area or amplitude and change compared with the WT average. For both WT and PDHD, this relationship was reversed by acetate, wherein area and amplitude increase with acetate (Fig. 10, B and C).

Figure 10.

Relationship between sEPSCs amplitude and area and percent change in PDHD and with acetate. A: scatter plot illustrating average cumulative probability value for each probability bin plotted against percentage change in average cumulative distributions between WT and PDHD for that bin (WT n = 11 cells from 9 mice, PDHD n = 7 cells from 6 mice) for sEPSC area and amplitude. B: scatter plot depicting average cumulative probability value for each probability bin plotted against percentage change with acetate in average cumulative distributions for WT cells (n = 10 cells from 8 mice) for sEPSC area and amplitude. C: scatter plot depicting average cumulative probability value for each probability bin plotted against percentage change with acetate in average cumulative distributions for PDHD cells (n = 6 cells from 5 mice) for sEPSC area and amplitude. PDHD, pyruvate dehydrogenase deficiency; sEPSC, spontaneous excitatory postsynaptic current; WT, wild type.

Comparison of average cumulative distributions can reveal broad differences, but is subject to limitations including: 1) the average change might not reflect change in every cell, and 2) cumulative probability distribution values are normalized (with a probability ranging from 0 to 1) and are influenced by preceding smaller parameter values. We further examined sEPSC area and amplitude for their distribution in each cell with kernel density estimates of the distribution (Fig. 11A). We employed kernel density estimates to compensate for relatively irregular histogram peaks, where an inference about the population distribution is made based on a finite sample of data points. PDH-deficiency events exhibited decreased density for some larger sEPSC areas and amplitudes (Fig. 11, B and D). Acetate reduced this density for some larger amplitude events in WT but otherwise had no impact (Fig. 11, E and F). In contrast, in PDH deficiency, acetate augmented the density of area and amplitudes for larger sEPSC events (Fig. 11, F and G).

Figure 11.

sEPSC amplitude and modulation in PDHD and after acetate. A: example of a histogram of sEPSC areas from a WT neuron with a kernel density line (thick black line). B: average Kernel Density Estimate of sEPSC areas across WT (n = 11 cells from 9 mice) and PDHD neurons (n = 7 cells from 6 mice). C: data from the same plot as in B, represented as change in Kernel Density Estimate of sEPSC areas across WT and PDHD neurons. D: average change in Kernel Density Estimate of sEPSC amplitudes across WT and PDHD neurons (WT n = 11 cells from 9 mice, PDHD n = 7 cells from 6 mice). E: average change in Kernel Density Estimate of sEPSC areas for WT neurons after acetate (n = 10 cells from 8 mice). F: average change in Kernel Density Estimate of sEPSC amplitudes for WT neurons after acetate (n = 10 cells from 8 mice). G: average change in Kernel Density Estimate of sEPSC areas for PDHD neurons after acetate (n = 6 cells from 5 mice). H: average change in Kernel Density Estimate of sEPSC amplitudes for PDHD neurons after acetate (n = 6 cells from 5 mice). A–D: *P < 0.05 unpaired two-tailed Student’s t test. E–H: *P < 0.05 paired two-tailed Student’s t test. PDHD, pyruvate dehydrogenase deficiency; sEPSC, spontaneous excitatory postsynaptic current; WT, wild type.

Action Potential Independent Synaptic Activity

Our findings suggest that larger amplitude events, putatively from action potential induced vesicle release, are most impacted by deficits and modulation of the TCA cycle. However, our detected synaptic events (sEPSCs) are a consequence of both spontaneous and action potential evoked vesicle release. To ask if evoked release of vesicles contributes to our observed phenotype, we perfused slices with 1 µM tetrodotoxin (TTX) to block action potentials and examined miniature excitatory postsynaptic currents (mEPSCs). In the presence of TTX, we found that there was no difference in average synaptic charge transfer (or mEPSC area) between WT and PDHD cells or in any mEPSC area subset of synaptic events (Fig. 12A). Two more waveform shape parameters, that of rise time and decay times, too were not different between the two groups (Fig. 12, B and C). Unexpectedly, there was a modest increase in mEPSC amplitudes in PDHD compared with WT neurons (Fig. 12D). Because glutamate synthetic flux is also modestly increased in PDHD (7), we hypothesized that if evoked vesicle release was more sensitive to TCA cycle failure, there could be a shift in glutamate association from evoked vesicle release to spontaneously released vesicle pools as compensation, increasing amplitude of spontaneous mEPSCs. Such a shift would also be reflected in an increased detection of spontaneously released vesicles. In support of this, we found that 1) when all events were considered in the absence of tetrodotoxin, there was no difference in event number or interevent intervals between WT and PDHD neurons, and 2) perfusing with TTX to compare only spontaneously released vesicles showed a greater number of synaptic events for PDHD cells (Fig. 13, A and B).

Figure 12.

mEPSC profile in PDHD compared with WT in tetrodotoxin perfused slices. A–D, left: bar graph depicting average mEPSC area, rise times, decay τ, and amplitudes for all events in each cell (WT n = 9 cells from 3 mice, PDHD n = 8 cells from 3 mice). Right: cumulative distribution in PDHD mice compared with WT mice for each cumulative distribution bin across mice. E, left: example WT (gray) and PDHD (black) neuron recorded traces in tetrodotoxin perfused slices illustrate more events with greater amplitude in PDHD neurons. Scale bar: 200 ms/5 pA. The dots depict events detected by the program. Right: example average mEPSC event waveform for a WT (gray) and PDHD (black) neuron. *P < 0.05 unpaired two-tailed Student’s t test. mEPSC, miniature excitatory postsynaptic current; PDHD, pyruvate dehydrogenase deficiency; WT, wild type.

Figure 13.

Comparison of interevent intervals with and without tetrodotoxin. A, left: average sEPSC interevent interval for all events in each cell (WT n = 11 cells from 9 mice, PDHD n = 7 cells from 6 mice). Right: cumulative distribution of sEPSC interevent intervals in PDHD mice compared with WT mice for each cumulative distribution bin across mice. B, left: average mEPSC interevent interval in tetrodotoxin perfused slices for all events in each cell (WT n = 9 cells from 3 mice, PDHD n = 8 cells from 3 mice). Right: cumulative distribution of mEPSC interevent interval in PDHD mice compared with WT mice for each cumulative distribution bin across mice. *P < 0.05 unpaired two-tailed Student’s t test. mEPSC, miniature excitatory postsynaptic current; PDHD, pyruvate dehydrogenase deficiency; sEPSC, spontaneous excitatory postsynaptic current; WT, wild type.

DISCUSSION

We sought to examine the consequences of reduced TCA cycle activity stemming from PDH deficiency by recording spontaneous excitatory synaptic currents. Through unbiased as well as parameter-driven analyses, we find that synaptic charge transfer is reduced in PDH deficiency. Administration of acetate enhances charge transfer in a subset of synaptic events. The results point both to altered event charge transfer, which reflects net glutamate release, and event current decay, which can parallel the persistence or clearance of glutamate in the synaptic cleft, as contributors to changes in synaptic charge transfer. Larger synaptic events are more impacted in PDH deficiency and undergo greater modulation after alternate fueling by acetate.

Larger Size sEPSC Events in PDH Deficiency

sEPSCs may arise from action-potential dependent as well as independent release of neurotransmitter in the absence of experimental electrical stimulation (22). Furthermore, 15%–20% of sEPCSs are induced by action potentials and are typically characterized by larger amplitudes (23). This phenomenon correlates with synaptic vesicle release processes, which differ for action-potential-stimulated and action-potential-independent postsynaptic currents (24). In this context, one possible explanation for PDH deficiency synaptic event changes is that the reduction of ATP or carbon exerts a greater impact on action-potential-driven vesicular release. In support of this assertion, blocking action-potential-induced synaptic events in our study removed differences in event area and decay τ between WT and PDHD neurons. In further support of the impact of oxidative phosphorylation on large synaptic events, oxygen-glucose deprivation preferentially impacts large-amplitude inhibitory currents (25). Nevertheless, we note that the cumulative experimental evidence available today does not provide unequivocal support for this notion. For example, both action-potential-driven and non-driven events use vacuolar ATPase V0a1 subunit, with nonaction-potential-driven events proving more susceptible to the vacuolar ATPase inhibitor folimycin (26). Another possibility is that reduction of larger sEPSC events in PDHD and their enhancement with acetate merely reflects the generation and replenishing of larger synaptic vesicle volumes. In PDH deficiency, demands for a smaller volume of vesicular content may be more easily satisfied than a larger volume. Similarly, the administration of acetate might more easily satisfy the requirements of the smaller vesicles.

Comparison with Previous Studies

Experimental administration of pyruvate-depleted ACSF with acute poisoning of the TCA cycle using oligomycin to block ATP generation leads to no impact on postsynaptic EPSCs, which appears to contrast with our findings (27). Nevertheless, we believe that our findings are not considerably different from those in that study. On examining average amplitudes, we too detect no impact on sEPSC amplitude in PDH-deficiency neurons, and the difference becomes apparent only on further examination of the average cumulative distribution probability or kernel density estimates. An additional difference is that we examined spontaneous synaptic events, whereas the cited study evaluated EPSCs induced or evoked by artificial stimulation. A robust single (i.e., not paired pulse) electrical stimulation capable of eliciting synaptic vesicular release could potentially overcome a mild deficit in vesicular release via the mobilization of additional vesicles that do not contribute to our analysis of spontaneous synaptic events. Last, the impact of lifelong PDH deficiency in the transgenic mice of our study may differ from the acute poisoning preparation by virtue of (heretofore unidentified) adaptive changes. Our findings nevertheless are coherent with our previous study demonstrating a decrease in glutamate production and release probability in the same mouse model of PDH deficiency (7).

Glutamate synthesis is a fundamental metabolic aspect derived from the TCA cycle, as is compartmentation into synaptic vesicles. Vesicular release also necessitates energy metabolism integrity. Reuptake and reutilization of glutamate released in the synaptic cleft is energy dependent. The ionic gradient maintained by the Na⁺/K⁺-ATP-ase pump, itself energy dependent, is required for the entry of cleft glutamate into astrocytes via glutamate transporters (10). It may be postulated that energy failure will lead to defective glutamate uptake and thus result in prolongation of EPSC decay τ (8). Our results indicate the opposite. One explanatory possibility is that energy allocated to the Na⁺/K⁺-ATP-ase pump derives primarily from anaerobic glycolysis (i.e., the metabolic reactions previous to or upstream acetyl-CoA generation) rather than downstream, acetyl-CoA-dependent oxidative phosphorylation (28). This would imply that the Na⁺/K⁺-ATP-ase pump is relatively resistant to deficits in TCA cycle flux. In support of this contention, we found no change in resting membrane potential in neurons from the same mouse model (7), which indirectly report on the activity of the ion gradient-maintaining Na⁺/K⁺-ATP-ase pump.

Limitations of This Study

One limitation is the age of the mice studied. We assayed mice at postnatal days P16–P18, even though 1) at that age there continues to be a development increase in pyruvate oxidation (29), and 2) the impact of PDH deficiency on mouse behavior is more pronounced at later ages, as the mice visibly deteriorate and die at ∼P25 (7). Less pronounced decreases in PDH brain activity, although compatible with full animal viability, exert little impact on brain metabolism (30). The choice of P16–P18 as the age to conduct our whole cell patch-clamp experiments is due to a lack of brain slices containing viable cells in mice with PDH deficiency at older ages (see Methods in Ref. 7). Thus, our findings in the present work may underestimate the extent of neurophysiological dysfunction possible in more severe forms of the disease or may be reporting neurophysiological processes that are particularly vulnerable or unique to less severe PDH deficiency.

Another limitation is potential distortion of currents measured via narrow pipette electrodes. We employed high resistance pipettes to delay the potential impact of dialyzing the intracellular metabolic milieu. Markedly narrow pipettes used as sharp electrodes (with resistances of 150–180 MΩ) for intracellular recording can dampen or distort current amplitudes (31), potentially affecting the detection of smaller sEPSCs. The electrodes we used had resistances of 14–20 MΩ and, thus, should not attenuate current amplitudes to the degree that sharp intracellular electrodes could, but likely caused a mild attenuation of sEPSC amplitudes. By limiting inferences to differences between groups that were subject to similar recording conditions, we expect to have captured the salient differences between the PDHD and WT experimental groups while minimizing the influence of a distorted cellular metabolic environment.

Conclusions

In summary, within the context of our PDH-deficiency mouse model, the waveform of spontaneous AMPA-mediated glutamatergic currents reveals that PDH deficiency is associated with diminished synaptic charge transfer. Under our experimental conditions, this metabolic deficit preferentially impacts larger synaptic currents, in a manner that is ameliorated or reversed with alternative metabolic fuel.

GRANTS

The study was supported by the Rafael Romanillos Fund and National Institutes of Health Grant NS102588.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

V.J. and J.M.P. conceived and designed research; V.J. and Q.M. performed experiments; V.J. and Q.M. analyzed data; V.J., Q.M., and J.M.P. interpreted results of experiments; V.J. prepared figures; V.J. and J.M.P. drafted manuscript; V.J., Q.M., and J.M.P. edited and revised manuscript; V.J., Q.M., and J.M.P. approved final version of manuscript.