Exogenous pyruvate and ethyl pyruvate (EP), the key ingredient in a new Ringer's solution in clinical trials, are antioxidants as well as metabolic substrates. In vivo studies show both to be protective in oxidative stress, with EP being better. The authors used an acute rat brain slice preparation to compare EP and pyruvate rescue after H(2)O(2) oxidative stress, asking whether EP was again better and whether its actions were exclusively metabolic.
Oxygenated neonatal P7 cerebrocortical slices were exposed for 1 h to 2 mM H(2)O(2), and recovered for 4 h with artificial cerebrospinal fluid having 2 mM glucose and (1) 20 mM EP, (2) 20 mM pyruvate, or (3) 1 mM of the nonmetabolizable radical scavenger N-tert-butyl-alpha-phenylnitrone (PBN). Perchloric acid extracts were studied with 31P/1H nuclear magnetic resonance at 14.1 T. Acute cell injury was assessed by counting terminal deoxynucleotidyl transferase-mediated uridine 5'-triphosphate-biotin nick end labeling (TUNEL)-stained cells.
At the end of recovery, preservation of adenosine triphosphate and N-acetylaspartate was better with EP than with pyruvate. Adenosine triphosphate preservation was best when PBN and EP were coadministered. 1H nuclear magnetic resonance revealed changes in lactate, alanine, gamma-aminobutyric acid, glutamate, glutamine, succinate, taurine, and myoinositol. Two-dimensional [1H-13C] heteronuclear single quantum coherence spectroscopy found that 13C-EP administration produced the same tricarboxylic acid metabolites as C-pyruvate. TUNEL-positive cell percentages with EP were less than half of those for PBN or pyruvate rescue (P < 0.05).
EP enters cells, provides pyruvate as a tricarboxylic acid substrate, and is more protective. Although EP provides metabolic protection of adenosine triphosphate levels, it does not maximize antioxidant protection.
ETHYL pyruvate (EP; Chemical Abstracts Service No. 617-35-6; ethyl 2-oxopropionate; C5H8O3; fig. 1) is an aliphatic ester composed from ethanol and pyruvate. The α-keto carbonyl in pyruvate and EP makes them potent scavengers of reactive oxygen species, particularly H2O2.1,2EP easily enters cells, apparently without the help of monocarboxylate transporters,3which are needed by pyruvate for intracellular entry.4Intracellular ester cleavage of EP makes pyruvate available for tricarboxylic acid (TCA) metabolism. Ringer’s EP solution,5a new intravenous crystalloid for which US Food and Drug Administration approval is being sought, has undergone numerous preclinical studies in which, compared with Ringer’s lactate, it has been found to improve the survival of injured cells and tissues.6
The idea behind exogenous intravenous administration of pyruvate was that it might provide two basic metabolic advantages. First, in the cytosol, pyruvate conversion to lactate would transform NADH into NAD+, thereby helping to maintain the large NAD+/NADH ratio necessary for glycolysis. Second, in the mitochondria pyruvate serves as a substrate for adenosine triphosphate (ATP) production by the TCA cycle. Indeed, preclinical studies with intravenous pyruvate showed impressive protection from ischemia and oxidative stress and have been summarized by others.5Attention turned to EP, however, after serious potential problems associated with exogenous pyruvate were fully appreciated. Pyruvate is unstable in solution, where it rapidly dimerizes into pyruvate hydrate, which then completely transforms itself into parapyruvate, a potent metabolic inhibitor.7
Exogenous EP was then looked at and was found to provide more protection in ischemia and oxidative stress than did equimolar amounts of pyruvate, particularly in ophthalmologic studies of oxidative stress8and in general surgery studies of sepsis after cecal disruption.2,5,9As well, recent animal studies demonstrated a wide neuroprotective window by EP in cerebral ischemia10and H2O2radical stress.11Therefore, at least three mechanisms might be responsible for exogenous EP’s superiority in the early period after ischemia and oxidative stress: the absence of parapyruvate poisoning, better radical scavenging, and better metabolic function from EP being a prodrug that boosts TCA cycle metabolism. Approximately 3–6 h after ischemia or oxidative stress, there is an important fourth mechanism: antiinflammatory actions by EP but not by pyruvate.9,12–14
Studies in human volunteers recently found Ringer’s EP to be a safe intravenous solution.15However, studies in human patients, while also showing no harm, have so far not revealed distinct therapeutic benefits.15However, human studies are still ongoing, and numerous avenues, such as neuroprotection, have not yet been explored. It is quite possible that Ringer’s EP, in one preparation or another, will one day be administered to humans by anesthesiologists.
Nuclear magnetic resonance (NMR) spectroscopy, also referred to as magnetic resonance spectroscopy, is a powerful tool for in vivo and in vitro studies of metabolism.16,17Although 13C precursors used for this purpose originated in the 1940s,18new methods that advance its applications continue to appear, thanks to constant technological progress,19especially in the area of brain metabolism.20We recently completed an NMR brain slice study that found exogenous pyruvate (10 mm) to be protective and metabolically active in superfused brain slices during H2O2-induced oxidative stress.21In this work, we seek to compare neuroprotection from EP versus that from pyruvate, with a focus on radical scavenging and TCA cycle metabolism. Specifically, we asked whether a nonmetabolizable antioxidant, N -tert-butyl-α-phenylnitrone (PBN), could provide protection similar to that of EP. Protection in these studies was defined by sustained levels of ATP, other high-energy phosphates and N -acetyl-l-aspartate (NAA), and by reductions in counts of terminal deoxynucleotidyl transferase-mediated uridine 5′-triphosphate-biotin nick end labeling (TUNEL)–stained cells.
Materials and Methods
Superfused Brain Slice Preparation and Experimental Design
Experimental protocols were approved by the University of California, San Francisco Committee on Animal Research and were very close to those described previously.21–23Briefly, in each experiment, 20 cerebrocortical slices (350 μm thick) were obtained from ten 7-day-old Sprague-Dawley rats and superfused with fresh, oxygenated artificial cerebrospinal fluid (ACSF) that consisted of a modified Krebs balanced salt solution: 124 mm NaCl, 5 mm KCl, 1.2 mm KH2PO4, 1.2 mm MgSO4, 1.2 mm CaCl2, 26 mm NaHCO3, and 10 mm glucose. The superfusion chamber held 3 ml ACSF in addition to the slices. It was immersed in a water bath that was kept at 37°C, and the oxygenated ACSF flow rate was 10 ml/min. Continuous bubbling with a 95% O2–5% CO2gas mixture maintained constant ACSF values of Pco2(40 mmHg), Po2(more than 450 mmHg), and pH (7.4).
In all experiments, slices were superfused for a total of 8 h. One hour of oxidative stress, which consisted of superfusion with 2 mm H2O2in glucose-free ACSF, began at t = 0, which was after 3 h of superfusion with the metabolizable substrate being 10 mm glucose. The hour of oxidative stress was followed by 4 h of recovery (from t = 1 h to t = 5 h), with the recovery superfusate containing 2 mm glucose and one of the following five different metabolic substrates: (1) no additional metabolic substrate, indicated as “no treatment”; (2) EP (20 mm); (3) pyruvate (20 mm); (4) PBN (1 mm); or (5) the combination of PBN (1 mm) with EP (20 mm). Previous studies by others, which were confirmed by us for slice experiments, found that it is always necessary to have some glucose in the ACSF when comparing different metabolites.24Such is why 2 mm glucose was used as described above. Slice sampling was performed at t = 0 h and t = 5 h. To detect any metabolic or histologic deterioration that might arise from the nutrition protocol alone, we studied a sixth group, known as the “control group,” in which slices underwent no H2O2exposure, and had their metabolizable substrate at 10 mm glucose for 3 h, and then at 2 mm glucose for 5 h. Special studies were performed, including EP rescue at 10 and 28 mm; and PBN rescue at 5 mm, to determine whether additional protection would occur at higher concentrations. As well, to determine whether EP metabolism resembled pyruvate metabolism, an experiment was performed using EP labeled with 13C at the site that corresponds to the methyl carbon in pyruvate (arrows in fig. 1). This site was labeled in our previous studies with [3-13C]pyruvate.21 13C-EP was obtained from Sigma-Aldrich Inc. (St. Louis, MO).
Perchloric Acid Extraction of Metabolites and Sample Preparations for NMR Studies
Perchloric acid extraction and NMR tube loading were also performed as in previous studies.21–23In brief, seven frozen slices from each time point were pulverized in liquid nitrogen. The resulting fine powder was placed in 7 ml perchloric acid, 12%, at 4°C. Final extracts were lyophilized (BenchTop 2K lyophilizer; Virtis, Gardiner, NY), and the weight of dry powder was measured. Each lyophilized sample was dissolved in 99.9% D2O, and 1 μl 3-trimethylsilyl-tetradeuterosodium propionate (TMSP), 700 mm, was then added as an internal NMR reference for chemical shifts and signal intensities in 1H and 13C spectroscopy. 1H/13C NMR spectroscopy was performed after neutralizing samples with NaOD. After 1H NMR spectroscopy was completed, EDTA was added to chelate and remove line-broadening cations. At the same time, 2 μl methylene diphosphonate, 330 mm, was added as a 31P NMR reference, and the pH was again adjusted to 7.0 to 7.4.
1D 31P and 1H NMR Spectroscopy
One-dimensional (1D) NMR studies of extracts were performed in the University of California, San Francisco Magnetic Resonance Laboratory using a 14.1-T (600-MHz) Varian INOVA spectrometer and a customized, multinuclear Z-SPECT radiofrequency probe that was optimized for this project (3NG600-8; Nalorac Division of Varian, Martinez, CA). Basic one-pulse, 90° tip-angle radiofrequency sequences were used for obtaining 1H spectra at 599.92 MHz and 31P spectra at 242.86 MHz. The extra homogeneity provided by the 5-mm Shigemi NMR tubes (Allison Park, PA) and the Varian INOVA shimming software typically resulted in spectral line widths less than 0.0025 ppm full width at half maximum (1.5 Hz full width at half maximum) for 1H water protons and 0.01 ppm full width at half maximum for 31P in phosphocreatine. Proton spectra were composed from 64 transients, with the interpulse delay being 15 s.
Metabolites corresponding to different chemical shifts were identified from values published in careful studies,25,26as well as from computer databases at the University of Wisconsin’s Biologic Magnetic Resonance Data Bank (Madison, Wisconsin)#and at the National Institute of Advanced Industrial Science and Technology (Tokyo, Japan).**
Figure 2shows various regions of a typical 1H control spectrum. The chemical shift origin (0 ppm) was assigned to the TMSP resonance, which is off scale to the right. Figure 2Ashows resonances for all metabolites that were easily resolved and quantified: lactate, alanine, NAA, γ-aminobutyric acid, glutamate, glutamine, succinate, phosphocreatine, creatine, taurine, and myoinositol. Figures 2B and Cshow magnifications of spectral regions, enabling appreciation of the high 1H spectral resolution that was achieved. Note that phosphocreatine and creatine were resolvable, and substantial structure was seen for the γ-aminobutyric acid, glutamine, and glutamate resonances.
31P spectra were obtained with proton decoupling and a 2-s interpulse delay during 8-h runs. The chemical shift origin (0 ppm) was assigned to the narrow phosphocreatine resonance. As can be seen in the representative 31P spectral section in figure 3A, which is from a control spectrum, adenosine diphosphate (ADP) could be cleanly quantified from integrating the clearly separated signal for α-ADP. The large triplet component to the β-nucleoside triphosphate resonance peaks in figure 3Bcomes primarily from β-ATP, with a small overlap component coming from guanosine triphosphate at the same chemical shifts.27The small peaks to the right of the triplet show two of the three triplet peaks from a small, combined cytidine triphosphate and uridine triphosphate background.27The ATP signal intensity was obtained by integrating the large triplet resonance.
31P and 1H relaxation time corrections were not introduced, because the same NMR pulse sequence was used for all runs of the same nucleus, and the goal of the analysis was primarily to detect large differences in each metabolite from its control value.
Finally, relative metabolite values were computed relative to methylene diphosphonate signal intensities and then normalized twice: first to the weight of the dry powder that was mixed with D2O before NMR measurements, and then again to the same numerical quantity that was obtained from slices removed at t = 0 samples, just before the H2O2stress.
2D NMR Spectroscopy: Indirect Detection of Metabolism of 13C-labeled Ethyl Pyruvate with HSQC
The same extracts (same NMR tubes that were used for 1D spectroscopy) were also used for two-dimensional (2D) spectroscopy, but on a different 14.1-T Varian UNITY-INOVA spectrometer equipped with a “cold” radiofrequency probe: a 5-mm triple resonance cryogenic probe with an actively shielded Z-axis pulsed field gradient coil (1H at 600 MHz, 13C at 150.9 MHz), and radiofrequency coils that were maintained by an Intelligent Temperature Controller at 25°K while the sample, 1 mm away, was maintained at 25°C. Sensitivity-enhanced heteronuclear single quantum coherence spectroscopy (HSQC) spectra were obtained.28,29Transients were acquired with a 1-s relaxation delay, 8 scans per increment, a 1H spectral width of 7,200 Hz (acquisition time of 142 ms), and a 13C spectral width of 15,000 Hz with a maximum evolution time of 133 ms. The National Institutes of Health programs NMRPipe and nmrDraw 30(updated software versions of August 2006) were used to convert raw data into pure absorption mode spectra. Apodization was applied using sine-squared window functions in the 1H and 13C dimensions. Resulting 2D spectra were then analyzed and displayed with SPARKY (Thomas D. Goddard, M.A., Senior Software Specialist, School of Pharmaceutical Chemistry, University of California, San Francisco and Donald G. Kneller, Ph.D., SPARKY3, University of California, San Francisco). Chemical shifts were calculated relative to those of TMSP, with the lactate methyl group being assigned to 1.33 ppm (1H) and 21.3 ppm (13C).
In Situ Labeling of DNA Fragmentation
Manifestations of cell death were sought using fluorescent in situ TUNEL staining. Sections were first permeabilized in 0.1% Triton X-100 (Sigma Chemical Co, St. Louis, MO) in phosphate-buffered saline for 8 min. TUNEL reaction mixture was obtained by adding terminal deoxynucleotidyl transferase to the nucleotide mixture according to instructions in the manufacturer’s manual (Roche Diagnostics, Mannheim, Germany). Each section was then incubated with 50 μl TUNEL reaction mixture in a humidified, dark chamber at 37°C for 60 min. After rinsing with phosphate-buffered saline, sections were counterstained with 0.5 g/ml propidium iodide to highlight nuclei. The fluorescein wavelength for excitation was 488 nm, with emissions being detected above 515 nm. Propidium iodide excitation was at wavelengths less than 535 nm, with emission being detected at approximately 615 nm.
Quantifications and Statistical Analysis
Calculations of relative metabolite concentrations in 1H/31P NMR spectra began with numerical integration of areas under resonance peaks (IQ-NMR software, version 1.5, Spectrum Research, Madison, WI; and iNMR, version 2, Nucleomatica, Mofetta, Italy). Relaxation time corrections were not applied. To account for variations in slice size so that relative metabolite signal intensities at different times could be compared, corrected relative intensities were obtained via the following a divided by b divided by c calculation: (Integrated area of signal on NMR spectrum)/(total integrated area of the spectrum’s reference compound [TMSP for 1H, or methylene diphosphonate for 31P])/(weight of dry powder after lyophilization). Corrected relative signal intensities of 31P spectra for slices taken at t = 5 h were then compared with values for slices taken at t = 0, the time just before H2O2insults. Experiments were repeated three times, and values of corrected signal intensities were averaged over the repeated experiments.
Quantifications of positive cells in TUNEL staining were performed at a magnification of 400× from images in five slice sections regions below the pial layer but above the injury layer. Cell counts for healthy and injured cells included neurons and glia, these not being histologically distinguishable. Counts of TUNEL-positive cells are given as percentages of total cells counted.
All data were plotted as mean ± SD and assessed by analysis of variance followed by Fisher protected least significant difference post hoc test, with P values less than 0.05 considered to be statistically significant. Statistical software was JMP IN (SAS Institute, Cary, NC). Parametric assumptions underlying these tests were that all slices were equivalent after recovery from decapitation, that all slices had equal probability of being assigned to treatments, that sampling was entirely random, that treatment groups had independent means and variances, and that data for compared treatment groups had the same variance.
High-energy Phosphate Outcomes from 31P NMR Spectra
31P NMR outcomes for the five rescue groups and the control group are shown in figure 4Aas ratios of normalized values at t = 5 h to normalized values at t = 0 h. It is important to recall that slices removed at t = 0 benefited from having 10 mm glucose in their ACSF up until the time they were taken from the superfusion chamber, and that slices taken at t = 5 h had just completed 4 h of superfusion with rescue ACSF that had 2 mm glucose plus one of the following additional metabolizable substrates: no additional metabolizable substrate (no treatment), 20 mm EP, 1 mm PBN, 20 mm pyruvate, or 20 mm EP with 1 mm PBN.
Figure 4Ashows that after oxidative stress, the combination of 20 mm EP and 1 mm PBN produced the highest ATP levels and the smallest increase in the ADP/ATP ratio (P < 0.05). This is interpreted as the best metabolic outcome. As for the preservation of ATP, pyruvate alone and “no treatment” were equally the worst. From a bioenergetic perspective, pyruvate alone ranks lower than “no treatment” because of its substantially larger increase in the ADP/ATP ratio. The second best metabolic outcome was for EP alone. Significant differences found in the multiple comparisons analysis are also indicated in figure 4A. Values for the control group (open white bars in fig. 4Aat the left of each set), which had no H2O2stress but had the ACSF glucose reduced after 3 h from 10 mm to 2 mm, revealed a finite metabolic stress compared with control groups of previous studies, where corresponding values were unchanged after superfusion for 8 h without the glucose reduction.21
We note that although the ADP/ATP ratio increased to 1.8 times its control value in the pyruvate group, the ADP/ATP ratio in all cases was always below 0.50, indicating that there were only small deteriorations intracellular in high-energy phosphates. Figure 4Bcompares high-energy phosphate data for two sets of three experiments, corresponding to rescue superfusates of 1 and 5 mm PBN. The use of 5 mm PBN produced the same results found with 1 mm PBN.
Metabolite Changes from 1H NMR Spectra
1H NMR outcomes for the five rescue groups and the control group, which had no H2O2stress but had ACSF glucose reduced after 3 h, are presented in figure 5, also as ratios, these again being computed by dividing values for t = 5 h by values for t = 0 h. Thus, a y-axis reading of 1.0 would result from slices that were the exactly the same at t = 5 h and t = 0 h.
In figure 5, there are significantly larger alanine peaks from experiments having pyruvate or EP in the superfusate, which is no surprise because an exogenous supply of pyruvate drives the aminotransferase conversion of pyruvate into alanine. A glance at the metabolite levels in figure 5reveals that H2O2stress tended to reduce them all, except for lactate, which is associated with cell injury and understandably was increased. NAA is generally taken to be a marker of neurons,31,32and it was largest in the EP and EP–PBN groups. Statistical significance existed for NAA, being greater in the EP and EP–PBN groups compared with the “no treatment” and pyruvate groups. Myoinositol is generally taken as a glial marker that is also an osmolyte,33,34but within error limits it was unchanged from control in groups. Taurine, an intracellular osmolyte known to decrease during hypotonic stress, particularly cytotoxic edema,35was slightly decreased by the superfusion protocol alone, and except for the group superfused with the EP–PBN combination, taurine further decreased in all groups that underwent H2O2stress. Lactate and alanine were lowest for the PBN group. Glutamate was most reduced in the PBN group and most elevated in the pyruvate and pyruvate–PBN groups (P < 0.05 for each case). γ-Aminobutyric acid was most reduced in the EP and pyruvate groups.
Pyruvate and parapyruvate were detectable in NMR spectra for the pyruvate and EP runs as single sharp peaks. However, these were not quantified because pyruvate in the ACSF could contaminate the extract, despite slice washing during harvest, and transformation to parapyruvate was possible during the interim between NMR tube preparation and NMR data taking.
Metabolites Detected in 2D [1H-13C]HSQC Spectra
Figure 6shows a representative region of two overlaid 2D NMR spectra, one where the substrate consisted of 2 mm glucose with 20 mm [3-13C]pyruvate (green contours), and a second where the substrate was 2 mm glucose with 20 mm 13C-EP (red contours). Contours on an HSQC plot arise only from protons having a chemical bond to 13C atoms, and at coordinates corresponding to the 13C and 1H chemical shifts. In this plot, 13C carbon chemical shifts are plotted along the vertical axis, whereas proton chemical shifts are plotted along the horizontal axis. 12C nuclei are not observable with NMR spectroscopy. Every metabolite seen in the EP data overlapped exactly with the same metabolite’s tracing in the pyruvate data. In a previous study,21we demonstrated a result different from that just found: The ensemble of 13C metabolites seen after superfusion with 20 mm 13C-glucose alone had substantial nonoverlap with all metabolites seen in 13C-pyruvate data. Regarding the 2D NMR results, metabolites seen after the administration of 13C-EP were the same as those seen after the administration of 13C-pyruvate.
Quantifications of Cell Death at 5 h with Different Substrates during Recovery
Figure 7shows representative TUNEL staining of slices removed at t = 5 h. TUNEL staining images consist of an overlay of green fluorescence from the TUNEL stain and red fluorescence from propidium iodide, which is not a vital stain in this figure. Low-intensity clusters inside TUNEL-positive cells arise from shrunken, fragmented nuclei. The slices in the low-magnification upper row of figure 7Abegan the experiment at approximately the same size. The different slice widths arise from cytotoxic (intracellular) brain edema. The high-magnification lower row of figure 7Ashows that substantial TUNEL staining comes from parenchymal regions midway between the surface pia matter and the 50-μm injury layer at the edge where the slice was cut. Cell counts are shown in figure 7B. EP administration significantly gave better protection to all brain slice groups that were exposed to oxidative stress, having much smaller numbers of TUNEL-positive cells (P < 0.05).
We undertook our studies after reviewing numerous in vivo comparisons of pyruvate and EP protection in hypoxia–ischemia and radical stress, and noting that EP was always more protective.10,13,36–38Therefore, it is not surprising that the two treatments having EP were also best in our studies. This outcome might in part arise from the impossibility of exogenously administering totally pure pyruvate, which dimerizes rapidly (approximately 20 min) into pyruvate hydrate, a compound that has little if any metabolic influence, but which autotransforms slowly (approximately 24 h for 100%) to parapyruvate, a potent metabolic inhibitor.7,39
PBN has been used in other studies of brain ischemia and oxidative stress.40–42After we found that PBN protection at 1 mm was not as good as that from 20 mm EP, we addressed the concern that PBN might do better than EP if it were at a higher concentration. Therefore, we performed two additional sets of experiments: one where 5 mm PBN was used instead of 1 mm, and one where 1 mm PBN and 20 mm EP were used together. Figure 4Bshowed that switching the 1 mm PBN concentration to 5 mm made no difference in ATP preservation. This was also true for other metabolite levels. However, superfusion with the mixture of 20 mm EP with 1 mm PBN produced the best protection, as described in the Results. From the perspective of the EP experiments, adding PBN proved that EP did not do maximal radical scavenging. But did EP alone do any radical scavenging? A few points should be considered before answering.
If adding PBN to EP had produced the same outcome as EP alone (i.e. , no effect from adding PBN), we could have cited that as evidence for radical scavenging by EP. This study’s control group (not exposed to H2O2, with glucose during recovery decreased from 10 mm to 2 mm) was different from that of the previous study in that it imposed a metabolic stress: substrate insufficiency, with evidence being the ATP decrease found at the end of the 4-h recovery period. (The height of the “control” column in fig. 4Ais at 0.6 instead of 1.0.) There was no ATP decrease in the previous study’s control group, where ACSF was maintained at 10 mm. Although figure 4Ashows that decreases in phosphocreatine did not, within the accuracy of the measurements, permit a distinction among the treatment groups, final values of phosphocreatine–ATP were lowest with EP and EP–PBN, the groups with highest ATP. This could suggest an important role for the creatine kinase reaction, which is known to participate in phosphocreatine reductions aimed at maintaining ATP.43,44
From the perspective of the PBN experiments, figure 4Asuggests that adding EP helped by addressing a deficit in metabolic support. But did EP also help via radical scavenging? A comparison that might have been helpful is the extent of ATP preservation by the combination of EP with adequate glucose (10 mm), compared with preservation by the combination PBN with adequate glucose (10 mm), because this would have shown whether adding EP was no different from adding PBN. If such were the result, it could be taken as evidence for EP radical scavenging. However, data do not exist for this comparison. Therefore, it is not possible to unassailably conclude from our data that EP reduced radical damage. At the same time, it is not possible to exclude an antioxidant role for EP in this study, because the spectroscopy results had PBN providing protection comparable to that of EP alone. Very likely, EP directly eliminated some radicals and so would fall in line with other experiments that found EP caused smaller reductions in glutathione, malonaldehyde, and protein carbonyls11,45as well as less lipid peroxidation.46
Our 2D NMR data are consistent with the impression that EP’s actions come completely from its intracellular delivery of pyruvate.8Because EP is electrically neutral, it easily crosses cell membranes to reach the cytosol, and once in the cytosol, it encounters numerous esterases that cleave it into ethanol and pyruvate. Therefore, EP seems to be what one should use exogenously when one wants to augment intracellular pyruvate. It is also noteworthy that in the 31P data of figure 4A, the pyruvate rescue (20 mm pyruvate–2 mm glucose) is the least protective, with ATP levels being as low as those of the “no treatment” group, and the increase in the ADP/ATP ratio being even higher. This might be due to the intracellular acidification generated by the H+-monocarboxylate cotransporter that takes pyruvate into the cytosol.3But possibly it is due to 2 mm glucose being inadequate, which is also supported by figure 6of our recent publication,21where a 10 mm pyruvate–10 mm glucose rescue almost fully restored ATP, but a 10 mm pyruvate–0 glucose rescue restored ATP by only 60%.
N -acetylaspartate, a brain intracellular osmoregulator whose metabolism and regulation have recently been further elucidated,47is generally taken to be a neuronal marker, although it is relevant to our studies that some NAA is also present in immature glia.47Evidence for EP being the best protector therefore also came from the NAA data at t = 5 h, where superfusates with EP had the largest NAA level (P < 0.05) after oxidative stress. Myoinositol and taurine, also osmoregulators, are generally taken as glial markers, the former because of preferential cellular uptake48and the latter because of intracellular synthesis.49Taurine decreases during cytotoxic brain edema.35,50Cellular release of this osmolyte pushes the cells toward having less water retention. Therefore, it makes sense that taurine levels in the control group (no H2O2exposure; least edema) were significantly higher than in all groups exposed to H2O2. Although it is clear from figure 7Athat among the treated groups, slice edema must have been lowest in the EP group, whose myoinositol and taurine levels after oxidative stress were not statistically different from control.
Ethyl pyruvate is a common substance, one long used as an additive to pharmaceutical preparations and foods, including candy, beverages, and baked goods. There was, however, a paucity of EP-related publications about intracellular protection until approximately 7 yr ago, when independent collaborations in the laboratories of A. J. Varma and M. Fink started publishing studies that involved ischemia and oxidative stress. The studies of Varma et al. and Fink et al. emphasized pyruvate’s role as an H2O2radical scavenger, while the studies of Swanson et al. emphasized rescue from intracellular energy failure. Indeed, radical scavenging and metabolic improvement are neither exclusive nor independent. H2O2radicals can interfere with brain metabolism by inhibiting three important TCA cycle enzymes, α-ketoglutarate dehydrogenase, succinate dehydrogenase, and aconitase,51,52with the inhibition being reversible if the mitochondria are viable.52
We were not able to detect oxidative injury that was diminished by chemical interactions between radicals and the α-keto carbonyl of pyruvate or EP, e.g. , as in peroxidative decarboxylation: pyruvate + H2O2→ acetate + H2O + CO2. Diminished radical injury can also arise as a secondary benefit of higher ATP production, possibly from decreased radical production or release. Still, in principle, there are metabolites to detect when α-keto acids (R-CO-COO−) scavenge H2O2radicals. In the case of pyruvate, R is CH3, and the products of the reaction are acetate and the formate radical, with the latter easily converting to H2O + CO2.53In the case of EP (CH3-CO-COEt, where Et is the chemical group -C2H5), the products are acetate and ethyl formate. Our 1H NMR spectra contain a large, sharp peak at 8.46 ppm, the known chemical shift for formate. However, this peak is regularly present before the administration of pyruvate or EP, and therefore not usable in this study as a marker of radical quenching. Neither acetate nor ethyl formate was quantifiable in our 1D 1H spectra, the former because the main acetate peak at approximately 2.08 ppm was tightly situated among other known peaks with almost the same chemical shift, and the latter because there simply was no signal in the very-low-noise region where an ethyl formate peak should be seen. However, it was possible in the 2D HSQC data for both pyruvate and EP to clearly identify a lone 2D acetate peak (at 2.08 ppm 1H; 24.6 ppm 13C). Unfortunately, there is not enough information in the NMR data to indicate whether this acetate peak came from radical scavenging or normal metabolism. 1H peaks for ethanol were not identified in the EP data from t = 5 h, most likely because NMR sensitivity is too low. In some runs, there was a small triplet resonance pattern at 1.19 ppm, where ethanol is expected, having approximately one sixtieth the intensity of the lactate intensity.
We have not touched upon any of the numerous antiinflammatory effects attributed to EP, such as reductions in cyclooxygenase-2, interleukin-1β, tumor necrosis factor-α, lipid peroxidation, reactive gliosis, and nuclear factor-κB activation.9Although some inflammatory injury from hypoxia–ischemia can be detected in vivo within hours, no doubt from microglial activation that is stimulated immediately by neuronal and synaptic degeneration, such activation is small on the time scale of this study,54and isolated brain slices do not experience delayed neutrophil infiltrations that become noticeable 12–24 h after brain injury.
In conclusion, this metabolite study showed that compared with pyruvate, EP provides better acute protection in brain tissue after oxidative stress. Our findings are consistent with EP’s primary advantage being its ability to better deliver pyruvate to intracellular compartments without having the problems associated with exogenous pyruvate solutions. But it is also possible that there are one or more additional mechanisms.
The authors thank Ray Swanson, M.D. (Professor), and Weihai Ying, Ph.D. (Assistant Adjunct Professor), of the Department of Neurology, University of California, San Francisco, and also Ralph Hurd, Ph.D. (Chief Scientist), of GE Healthcare, Menlo Park, California, for helpful suggestions and insights. Cristina Angelo (Technical Sales Consultant) of Sigma-Aldrich’s Stable Isotope Division (ISOTEC, Miamisburg, Ohio), was helpful in expediting production and delivery of carbon 13–labeled compounds. Much appreciated as well is the assistance provided in the early stages of the project by Kiyoshi Hirai, M.D. (Postdoctoral Associate at University of California, San Francisco until 2003; current position: Attending Physician, Kyoto Prefectural University, Kyoto, Japan).