Inhibition of Mitochondrial Pyruvate Transport by Zaprinast Causes Massive Accumulation of Aspartate at the Expense of Glutamate in the Retina*

Background: Pyruvate transport into mitochondria is a key step in energy metabolism. Zaprinast is a well known phosphodiesterase inhibitor. Results: Zaprinast has a strong influence on pyruvate transport into mitochondria. Conclusion: Inhibition of the mitochondrial pyruvate carrier by Zaprinast causes accumulation of aspartate at the expense of glutamate. Significance: Maintenance of normal amino acid levels in the retina relies on pyruvate transport into mitochondria. Transport of pyruvate into mitochondria by the mitochondrial pyruvate carrier is crucial for complete oxidation of glucose and for biosynthesis of amino acids and lipids. Zaprinast is a well known phosphodiesterase inhibitor and lead compound for sildenafil. We found Zaprinast alters the metabolomic profile of mitochondrial intermediates and amino acids in retina and brain. This metabolic effect of Zaprinast does not depend on inhibition of phosphodiesterase activity. By providing 13C-labeled glucose and glutamine as fuels, we found that the metabolic profile of the Zaprinast effect is nearly identical to that of inhibitors of the mitochondrial pyruvate carrier. Both stimulate oxidation of glutamate and massive accumulation of aspartate. Moreover, Zaprinast inhibits pyruvate-driven O2 consumption in brain mitochondria and blocks mitochondrial pyruvate carrier in liver mitochondria. Inactivation of the aspartate glutamate carrier in retina does not attenuate the metabolic effect of Zaprinast. Our results show that Zaprinast is a potent inhibitor of mitochondrial pyruvate carrier activity, and this action causes aspartate to accumulate at the expense of glutamate. Our findings show that Zaprinast is a specific mitochondrial pyruvate carrier (MPC) inhibitor and may help to elucidate the roles of MPC in amino acid metabolism and hypoglycemia.

Transport of pyruvate into mitochondria by the mitochondrial pyruvate carrier is crucial for complete oxidation of glucose and for biosynthesis of amino acids and lipids. Zaprinast is a well known phosphodiesterase inhibitor and lead compound for sildenafil. We found Zaprinast alters the metabolomic profile of mitochondrial intermediates and amino acids in retina and brain. This metabolic effect of Zaprinast does not depend on inhibition of phosphodiesterase activity. By providing 13 C-labeled glucose and glutamine as fuels, we found that the metabolic profile of the Zaprinast effect is nearly identical to that of inhibitors of the mitochondrial pyruvate carrier. Both stimulate oxidation of glutamate and massive accumulation of aspartate. Moreover, Zaprinast inhibits pyruvate-driven O 2 consumption in brain mitochondria and blocks mitochondrial pyruvate carrier in liver mitochondria. Inactivation of the aspartate glutamate carrier in retina does not attenuate the metabolic effect of Zaprinast. Our results show that Zaprinast is a potent inhibitor of mitochondrial pyruvate carrier activity, and this action causes aspartate to accumulate at the expense of glutamate. Our findings show that Zaprinast is a specific mitochondrial pyruvate

carrier (MPC) inhibitor and may help to elucidate the roles of MPC in amino acid metabolism and hypoglycemia.
Pyruvate is a critical metabolite that links glycolysis with the mitochondrial tricarboxylic acid (TCA) cycle, and it is a hub for synthesis of amino acids, carbohydrates, and fatty acids. Pyruvate enters mitochondria through a recently identified mitochondrial pyruvate carrier (MPC), 3 a 150-kDa complex of MPC1 and MPC2 located on the inner mitochondrial membrane (1,2). Knock-out of MPC1 in Drosophila elevates pyruvate and decreases mitochondrial TCA intermediates (1). Children with mutations in a conserved region of human MPC1 have symptoms of lactic acidosis and hyperpyruvatemia (1,3). MPC inhibitors have been identified in the past 40 years including the classical ␣-cyanocinnamate analogs, UK5099, ␣-cyano-4-hydroxycinnamic acid (4), and insulin sensitizer thiazolidinedione (5)(6)(7).
Zaprinast, a lead compound used for the development of sildenafil (Viagra), is a well established inhibitor of cGMP-specific phosphodiesterase (PDE). PDEs hydrolyze the cyclic phosphate bond in cAMP and cGMP, and they function to inactivate cyclic nucleotide signaling pathways. Zaprinast has been used as a tool to study PDE5 and PDE6 (8 -11). PDE5, PDE6, and PDE9 are cGMP-specific, PDE4, PDE7, and PDE8 are cAMPspecific, and the rest of the members of the PDE family have dual specificity (12). Zaprinast is the only drug that inhibits PDE6 more potently than PDE5 (13). PDE6 is primarily expressed in retinal photoreceptors where it is responsible for light-dependent signal transduction. Mutations in the Pde6a, b, or c genes all cause retinal degeneration in humans (14 -16) and in mouse models (17,18). We performed a study intended to use Zaprinast to simulate the effects of PDE6 dysfunction in an ex vivo retinal degeneration model. We treated cultured mouse retinal explants with Zaprinast, but we found unexpectedly that Zaprinast is a potent inhibitor of mitochondrial pyruvate transport. This led to a novel and important finding, the focus of this report, that inhibition of mitochondrial pyruvate transport triggers dramatic metabolomic changes in neuronal tissues that severely alter the concentrations of glutamate and aspartate.
Animals-C57BL/6 mice (6 -8 weeks old) and C3Sn.BLiA-Pde6bϩ/DnJ (PDE6bϩ) mice were purchased from The Jackson Laboratory (Bar Harbor, ME). C3H retinal degeneration (rd1) mice, which bear a Pde6b mutation, were provided by Dr. Thomas A Reh, University of Washington. Pde6bϩ mice were used as the background control for rd1 mice. Cyclic nucleotide gated channel ␤1 knock-out (Cngb1 Ϫ/Ϫ ) mice on C57BL/6 background were obtained from the Institute for Ophthalmic Research, Tuebingen, Germany (19,20). Aralar/AGC1 ϩ/Ϫ mice (21) were crossed to produce Aralar/AGC1 Ϫ/Ϫ and Aralar/AGC1 ϩ/ϩ control littermates in Sv129/C57BL6 background at Dr. Jorgina Satrústegui's laboratory (Madrid, Spain). Experiments were performed in accordance with the Institutional Animal Care and Use Committee (IACUC) recommendations at the University of Washington guidelines after IACUC approval and with procedures approved in the Directive 86/609/EEC of the European Union and the Ethics Committee of the Universidad Autónoma de Madrid.
Retina Isolation and Culture-Mice were euthanized using CO 2 and cervical dislocation. The eyes were removed, and retinas were separated from retinal pigment epithelium in cold HBSS. The retina was cultured in Krebs-Ringer/HEPES/bicarbonate buffer (22) with glucose (5 mM) or other 13 C-labeled substrate (5 mM) in a 37°C 5% CO 2 incubator.
cGMP Assay-Two mouse retinas were homogenized in 300 l of 0.1 N HCL. After centrifugation, 100 l of the supernatant was assayed for cGMP by using a Direct cGMP ELISA kit from Enzo Life Sciences, Inc. (Farmingdale, NY). Gas Chromatography/Mass Spectrometry (GC/MS) Analysis of Metabolites-Individual mouse retina was rinsed in cold 0.9% NaCl and snap-frozen in liquid nitrogen. The retina was homogenized in a 700:200:50 cold mixture of methanol/chloroform/water with 5 nmol of internal standard methyl succinate. After centrifugation, the supernatant was dried under vacuum, derivatized, and analyzed by GC-MS (Agilent 7890/ 5975C) (22). The peaks were analyzed using Agilent data analysis software. The measured distribution of mass isotopomers was corrected for natural abundance of 13 C metabolite intensity defined with standards and verified by mass after each experiment. Enrichment was calculated by dividing the labeled ions with total ion intensity. The natural abundance from the tracers and derivatization reagents were corrected using IsoCor software (23,24). O 2 Consumption Measurements of Isolated Mitochondria-Brain mitochondria were isolated as previously reported (22). O 2 consumption was measured in an Oroboros Oxygraph-2K. The respiration buffer consisted of 150 mM KCl, 10 mM KH 2 PO 4 , 1 mM MgCl 2 , pH 7.4. 25 l of 5 mg/ml mitochondria was injected into the chamber for each assay. Respiration was measured in response to serial additions of pyruvate/malate, ADP, drug (DMSO, Zaprinast, or UK5099), glutamate/malate, and succinate. The final concentrations of substrates were 100 M EGTA, 50 M CaCl 2 , 1.5 mM pyruvate, 0.5 mM malate, 2.5 mM ADP, 1 mM glutamate/0.5 mM malate, and 5 mM succinate. After the addition of each component, the O 2 consumption rate was allowed to stabilize for 4 -5 min before the slope was quantified.
MPC Activity Assay-Mouse liver mitochondria were prepared as described before (25). Mitochondrial pyruvate transport was measured as reported (4,26) with modifications. The liver mitochondria were suspended in medium containing sucrose (250 mM), Tris-HCl (5 mM), and EGTA (2 mM) with pH 7.6. (ϳ5 mg of mitochondrial protein/ml of medium). An aliquot of the mitochondrial suspension (50 l or 0.5 mg of mitochondrial protein) was added to 150 l of medium containing KCl (125 mM) and Tris-HCl (20 mM), pH 6.8, with or without Zaprinast. After incubation for 5 min at room temperature, [2-14 C]pyruvate (15 M or 0.0225 Ci) was spiked into the 200-l solution to start the assay. After 1 min, the mitochondria-associated radioactivity was separated from the free medium by transferring the mitochondrial suspension to a 0.4-ml centrifuge tube containing a layer of oil consisting of 1:37.5 n-dodecane:bromododecane (Sigma) and spinning at 12,535 ϫ g for 8 s. The portion of the tube containing the mitochondrial pellet was cut off with a razor blade, placed into a scintillation vial, and then counted for radioactivity (26).
Measurement of Mitochondrial Metabolites-Freshly isolated mitochondria had intact mitochondrial membranes. To disrupt them we treated mitochondria with 0.1% Triton and freeze-thawed 5 times. Both membrane-intact and -disrupted mitochondria from separate isolations were incubated with 1 mM [ 13 C 3 ]pyruvate in the buffer used for respiration studies on isolated mitochondria at 30°C for 10 min. The metabolites were analyzed by GC/MS. To test shuttling of metabolites into and out of mitochondria, we incubated freshly isolated mitochondria with 100 M EGTA, 1 mM glutamate, 0.5 mM malate, 2.5 mM ADP, 50 M CaCl 2 , and 1 mM [ 13 C 3 ]pyruvate in respiration buffer at 30°C for 10 min. After centrifugation, the mitochondrial pellet and incubation medium were collected for GC/MS analysis.
Glucose Concentration Assay-Glucose in the culture medium was measured using an Amplex Red Glucose/Glucose Oxidase Assay kit (Invitrogen).
Pyruvate Dehydrogenase Activity (PDH) Activity Assay-PDH activity was measured using an NADH cycling assay in the presence of diaphorase to reduce the generated NADH and MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) was as an electron acceptor (27). Purified PDH protein from pig heart was incubated with 1 mM MgCl 2 , 5 mM CaCl 2 , 5 mM l-carnitine, 0.2 mM thiamine pyrophosphate, 0.1 mM CoA, 2.5 mM NAD ϩ , 1 mM MTT, and 1 unit/ml diaphorase in 50 mM HEPES/KOH buffer, pH 7.5, at 30°C. 5 mM pyruvate was added into the mixture to start the reaction, and the plate was read at 570 nm every 10 s for 10 min. The ⌬570-nm with maximum linear rate was used for calculating the enzyme activity.
Statistical Analysis-Data are expressed as the mean Ϯ S.E. The significance of differences between means was determined by unpaired two-tailed t tests or analysis of variance with an appropriate post hoc test. A p value Ͻ 0.05 was considered to be significant.

Zaprinast Causes Depletion of Glutamate and Accumulation
of Aspartate-We began the experiments with the intention of identifying the metabolomic signature of PDE6 inhibition. We isolated mouse retinas and incubated them with or without Zaprinast in the presence of 5 mM glucose and then extracted metabolites for GC/MS analysis. We then determined the concentration of Zaprinast required to inhibit PDE activity in mouse retinas. The concentration of Zaprinast required to cause accumulation of cGMP in the retina is in the range of 20ϳ200 M (supplemental Fig. 1A). We also found large changes in several metabolite levels at 200 M (Fig. 1A). Glutamate and aspartate had the largest changes: about a 5-fold decrease and increase, respectively. The effects on both glutamate and aspartate were first detectable at 30 min after the addition of Zaprinast, became more apparent at 60 min, and remained constant from 2 to 6 h (supplemental Fig. 1, B and C). We reasoned that the change of these two metabolites might be primarily responsible for the overall change of metabolic profile as aspartate-derived metabolites, malate, methionine, cysteine, and N-acetyl-aspartate increased, whereas glutamate-derived metabolites, glutamine and ␣-ketoglutarate (␣KG), decreased. We considered that glutamate in the retina might decrease because it is released into the medium. However, both glutamate and aspartate were undetectable in the incubation medium. Pyruvate increased by about 2-3-fold in response to Zaprinast in both retina and medium, but lactate did not change (Fig. 1, A and B). Glucose deprivation (22,28) causes similar changes of glutamate and aspartate levels in retina and brain slices. Therefore, we tested the effect of Zaprinast on glucose consumption. The retina consumed about 1.2 mol/h/ retina. Zaprinast did not affect the rate of glucose consumption (Fig. 1C).
The Effect of Zaprinast on Glutamate and Aspartate Is Independent of PDE Inhibition-To investigate the relationship between PDE6 inhibition and the metabolic effects of Zaprinast, we compared the concentrations of Zaprinast required to inhibit PDE in the retina with the concentration required to cause glutamate to decrease. We found that Zaprinast increases cGMP at 50 and 100 M and increases both cGMP and cAMP at 200 M. This indicates that Zaprinast inhibits PDE6 specifically between 50 and 100 M, but it inhibits other PDEs at 200 M ( Fig. 2A). Surprisingly, Zaprinast decreased glutamate and increased aspartate at much lower concentrations beginning at 2 M (Fig. 2B).
To test whether cyclic nucleotide-gated channels and Ca 2ϩ are involved in the decrease of glutamate, we used the retina from Cngb1 Ϫ/Ϫ mice in which the rising cGMP could not activate cyclic nucleotide gated channels to cause depolarization and elevation of intracellular free Ca 2ϩ (19,29). Remarkably, Zaprinast decreased glutamate and increased aspartate in Cngb1 Ϫ/Ϫ retina (Fig. 2C). Furthermore, neither of the cellpermeable cGMP analogues 8-Br-cGMP and 8-(4-chlorophenylthio)-cGMP nor inhibitors of Protein kinase G and PKA changed the levels of glutamate and aspartate in the retina (supplemental Fig. 2). These findings suggested that Zaprinast increases aspartate and lowers glutamate by a different mechanism than we expected, i.e. in a PDE6-independent fashion.
As further confirmation that the metabolic effects of Zaprinast occur independently of PDE6 inhibition, we applied Zapri-nast to the retina from rd1 mice that were 2 months old. The rd1 mutation causes almost all PDE6-expressing photoreceptors to die and disappear by 1 month of age and thus abolishes PDE6 activity. We found that glutamate and aspartate levels in rd1 retina are lower than normal, consistent with photoreceptors being a major source of these amino acids in normal retinas. We also found that Zaprinast potently decreases glutamate and increases aspartate in rd1 retina, suggesting that Zaprinast acted at a site distinct from PDE6 (Fig. 2D). As further confirmation that the effect of Zaprinast on glutamate and aspartate was not through PDE, we evaluated the effect of sildenafil, another PDE5/6 inhibitor. Sildenafil between 0.1-50 M robustly increased cGMP in WT retina but did not have a significant effect on glutamate or aspartate (Fig. 2, E and F). Taken altogether, these results show unambiguously that the effect of Zaprinast on glutamate and aspartate occurs independently of PDE inhibition.
The Effect of Zaprinast on Glutamate and Aspartate Occurs Independently of GPR35 Activation-Zaprinast also has been identified as a potent agonist of G protein-coupled receptor 35 (GPR35) (30,31). GPR35 is expressed predominantly in spleen, immune cells, and gastrointestinal tissues, but expression in eye also has been reported (32,33). To test whether GPR35 contrib- utes to the metabolic effect of Zaprinast, we preincubated the retina with a GPR35 antagonist, CID 2745687, for 30 min followed by an additional 30 min of Zaprinast treatment. CID 2745687 did not block the effect of Zaprinast on glutamate and aspartate. Furthermore, the GPR35 agonist, pamoic acid, did not change glutamate or aspartate levels (supplemental Fig. 3). These results show that GPR35 was not responsible for the observed Zaprinast effects. Zaprinast Blocks Entry of Glucose-derived Pyruvate into the TCA Cycle-To identify the mechanism behind the metabolic effect of Zaprinast, we incubated the retina with [ 13 C 6 ]glucose for 15 min with/without Zaprinast and then analyzed the isotopomer distributions of the intermediates. Under normal conditions, pyruvate and lactate become fully labeled by glycolysis of [ 13 C 6 ]glucose. Pyruvate loses 1 13 C to synthesize acetyl-CoA, and the two-labeled carbons appear in most TCA intermediates in the first round of the TCA cycle (Fig. 3A). In the second cycle citrate gains a total of four labeled carbons, and the rest of the intermediates have either three or four carbons labeled due to the loss of CO 2 at one of two different positions (supplemental Fig. 4A). Zaprinast did not change the 13 C enrichment in pyruvate and lactate. However, the enrichment of glutamate, aspartate, and TCA intermediates was much lower after Zaprinast treatment (Fig. 3B). Zaprinast increased the concentration of M3 (three labeled carbons) pyruvate but not lactate, and it significantly decreased both enrichment and concentration of the M2 isotopomer of citrate (the form with two labeled carbons from the first cycle), M4 (four labeled carbons from the second cycle), and M5/M6 (after three or more cycles) (Fig. 3, C-E; supplemental Fig. 4B). These results suggest that Zaprinast inhibits the formation of citrate from pyruvate. Accordingly, the downstream intermediates ␣KG and glutamate had fewer M2 and M3/4 isotopomers in Zaprinast-treated retinas. Zaprinast decreased M2 in both succinate and malate but increased both M2 and M3 in aspartate (Fig. 3, F--J). Zaprinast increased the total amount of labeled aspartate, but it decreased the overall enrichment of aspartate because Zaprinast causes a very large increase in the amount of unlabeled aspartate. Overall, these results indicate that Zaprinast inhibits the incorporation of pyruvate-derived carbons into the TCA cycle. This causes accumulation of oxaloacetate and aspartate. We also noted that the excess oxaloacetate "spills over" into the M3 isotopomers of malate and succinate by the reversible malate dehydrogenase and fumarate hydratase reactions.
Oxidized Glutamate Is the Source of the Aspartate That Accumulates in Response to Zaprinast Treatment-The decrease of glutamate that occurs with Zaprinast treatment always occurs in parallel with an increase of aspartate. We hypothesized that the increased carbons of aspartate come from oxidation of glutamate through ␣KG dehydrogenase. To test this hypothesis, we labeled the retina with [ 13 C 5 ]glutamine for 5 min and then replaced it with unlabeled glucose for 10 min (Fig. 4)  M5, and the subsequent intermediates have four 13 C carbons. A small amount of M5 citrate also accumulates. This must result from the reversal of the isocitrate dehydrogenase reaction (34) (Fig. 4A). For the second cycle, ␣KG and glutamate begin with three labeled carbons. After oxidative decarboxylation of ␣KG, the downstream intermediates have two labeled carbons, and citrate ends up having either two or three labeled carbons (supplemental Fig. 5A). Enrichment of citrate decreased with Zaprinast treatment, whereas enrichment of malate and aspartate increased (Fig. 4B). In the first TCA cycle, Zaprinast decreases the labeled citrate (M4) but significantly increases labeled succinate, malate, and especially aspartate (Fig. 4, C-I). These results show that Zaprinast prevents glutamate-derived carbons from passing through citrate to the second cycle. Consistent with this, M2 citrate, M3 glutamate, and M3 ␣KG decreased in response to Zaprinast. Zaprinast blocked citrate formation and thereby caused accumulation of oxaloacetate and aspartate produced from oxidized glutamate. To confirm this we incubated the retina with and without glutamine for 2 h. Both glutamate and aspartate increased ϳ5-fold in response to glutamine (relative to glucose alone), and Zaprinast shifted the glutamate-to-aspartate ratio by decreasing glutamate about 5-fold and increasing aspartate about 10-fold (supplemental Fig. 5B).

Zaprinast Inhibits O 2 Consumption in Intact
Retina-To test the effects of Zaprinast and UK5099 on mitochondrial function in ex vivo retina, we measured O 2 consumption from small pieces of intact mouse retina. Zaprinast caused a rapid increase followed by sustained inhibition of O 2 consumption. This biphasic effect can be explained by a mixed effect of Zaprinast on PDE and MPC activities. PDE inhibition increases O 2 consumption by stimulating energy consumption whereas MPC inhibition decreases availability of fuel to mitochondria. Consistent with this we found that sildenafil increases O 2 consumption whereas UK5099 inhibits it (Supplemental Fig. 6). UK5099 causes a slow but strong inhibition, probably due to its ability to inhibit complex II (Results not shown). Consistent with this we found that UK5099 at 100 M but not 10 nM inhibits succinatedriven O 2 consumption.
Zaprinast Inhibits Pyruvate-driven O 2 Consumption in Brain Mitochondria-Pyruvate enters via the MPC into the mitochondrial matrix where PDH transforms it into acetyl-CoA. PDH and the dehydrogenases of the TCA cycle produce NADH that provides reducing power for O 2 consumption. To test the hypothesis that Zaprinast inhibits pyruvate oxidation, we isolated brain mitochondria, and measured rates of O 2 consumption fueled either by pyruvate, glutamate, or succinate. Zaprinast almost completely blocked pyruvate-driven O 2 consumption (Fig. 5, A--D). However, Zaprinast did not block O 2 consumption fueled by glutamate or succinate. This is consistent with our isotopic labeling experiment in which Zaprinast did not inhibit glutamate oxidation. We next tested the hypothesis that Zaprinast blocks pyruvate transport by inhibiting MPC. Similar to Zaprinast, MPC inhibitor UK5099 blocked pyruvate-driven but increased glutamate and succinate-driven oxygen consumption (Fig. 5, E and F, supplemental Fig. 7, A and  B). There was no significant difference in the rate of oxygen consumption between Zaprinast and UK5099 (supplemental Fig. 7, C and D).  Zaprinast Inhibits MPC but Not PDH-To confirm that Zaprinast targets the MPC, we measured MPC activity directly using isolated mouse liver mitochondria. The MPC kinetics we measured are similar to previous reports (2,4). We measured flux of [2-14 C]pyruvate transport into isolated liver mitochondria. We confirmed the specificity of the assay by showing that 10 mM unlabeled pyruvate inhibits uptake of the labeled pyruvate (supplemental Fig. 8, A and B). Zaprinast inhibits pyruvate uptake (Fig. 6A) at a dose (Fig. 6B) that is similar to the dose required for its effect on glutamate and aspartate in the retina ( Fig. 2A).
To exclude the possibility that Zaprinast might inhibit PDH, we incubated pure PDH enzyme with Zaprinast or the known PDH inhibitor 3-fluropyruvate. Zaprinast did not affect the activity of PDH at any concentration tested. In contrast, 3fluropyruvate effectively inhibited PDH activities (Fig. 6C).
To confirm that the metabolic effect of Zaprinast is caused by MPC inhibition and not by inhibition of PDH activity, we bypassed MPC by adding [ 13 C 3 ]pyruvate to mitochondria with disrupted membranes (see "Experimental Procedures"). In this experiment pyruvate is accessible directly to PDH. Both Zaprinast and UK5099 decrease the amount of labeled citrate produced by mitochondria with intact membranes, but they have no effect on citrate formation when mitochondrial membranes are disrupted (Fig. 6D). As expected, disruption of the mito-chondrial membrane blocks the ability of either Zaprinast or UK5099 to stimulate depletion of glutamate (Fig. 6D).
Known Inhibitors of MPC Replicate the Effects of Zaprinast-We then asked whether other known MPC inhibitors cause metabolic effects similar to those of Zaprinast. Two known MPC inhibitors, ␣-cyano-4-hydroxycinnamic acid and UK5099, produce effects nearly identical to the effects of Zaprinast; they are increased aspartate and decreased glutamate, increased pyruvate, and suppressed pyruvate-driven O 2 consumption (Fig. 6,  F-G). Taken together, these results are consistent with our other data showing that Zaprinast blocks pyruvate transport into mitochondria.

Inactivation of Aspartate Glutamate Carrier (AGC1) Does Not Prevent the Decrease of Glutamate Caused by Zaprinast-
Aspartate normally is transported out of mitochondria via AGC1 in exchange for glutamate from cytosol (Fig. 7A). AGC1 allows the carbons from aspartate to bypass the conventional "complete" version of the TCA cycle. In the absence of AGC1, the carbons that would normally be used to make aspartate are forced into the complete TCA cycle, thus decreasing the levels of the amino acid in AGC1 Ϫ/Ϫ retina. In the absence of AGC1, the malate aspartate shuttle becomes impaired, and pyruvate is diverted to produce lactate and not acetyl-CoA. Consistently, the retina from AGC1 Ϫ/Ϫ mice had lower levels of aspartate and pyruvate than WT mice (Fig. 7B) (35).  DECEMBER 13, 2013 • VOLUME 288 • NUMBER 50

JOURNAL OF BIOLOGICAL CHEMISTRY 36135
We investigated whether AGC1 plays a role in the Zaprinast effect and found that it does not (Fig. 7B). When incubated with [ 13 C 6 ]glucose, AGC1 ϩ/Ϫ or AGC1 Ϫ/Ϫ retinas have 13 C enrichment similar to WT except that the overall levels of pyruvate and aspartate are lower than normal (supplemental Fig. 9). AGC1 deficiency impaired production of pyruvate, citrate, glutamate, and especially aspartate, consistent with a previous study of AGC1-deficient neurons (35). Nevertheless, Zaprinast still decreased labeling of glutamate from glucose and increased labeling of aspartate and pyruvate in both AGC1 ϩ/Ϫ and AGC1 Ϫ/Ϫ retinas (Fig. 7, C-F).
Zaprinast Causes Accumulation of Aspartate in Mitochondria and Prevents Generation of Glutamate from Aspartate-To examine the distribution of accumulated aspartate in or outside of mitochondria, we incubated isolated intact brain mitochondria with EGTA, malate, glutamate, Ca 2ϩ , ADP, and [ 13 C 3 ]pyruvate for 10 min and then measured the metabolites from the mitochondria and from the incubation medium. Consistent with previous experiments, Zaprinast blocked incorporation of 13 C from pyruvate into citrate both in and outside of mitochondria. Similarly, Zaprinast also inhibited incorporation of 13 C into aspartate. At the same time, Zaprinast stimulated the accumulation of unlabeled aspartate in mitochondria (Fig.  8, A-C). Some glutamate may enter the mitochondria by a glutamate carrier (36), contributing to aspartate increase in mito-chondria, but not the medium, within 10 min of incubation. Each of these findings is consistent with inhibition by Zaprinast of MPC-catalyzed uptake of pyruvate into mitochondria.
Unlabeled aspartate accumulates in mitochondria during Zaprinast treatment (Fig. 8, A-C). To confirm that Zaprinast decreases utilization of aspartate, we incubated the retina with labeled aspartate in the presence of glucose. During 1 h without Zaprinast, aspartate contributed 13 C to ϳ30% of glutamate and other TCA cycle intermediates (Fig. 8, D-E). However, Zaprinast blocked the incorporation of 13 C from aspartate into citrate and glutamate (Fig. 8, F-H). This reflects diminished acetyl-CoA synthesis in mitochondria, consistent with inhibition of pyruvate carrier activity by Zaprinast. In addition, the decreased lactate/pyruvate ratio (Fig. 8I) might lower cytosolic reducing power, which causes accumulation of aspartate by reversing the malate dehydrogenase reaction in the cytoplasm (Fig. 9). Consistent with this, we found that Zaprinast decreases NADH levels in the retina (Fig. 8J).

DISCUSSION
We demonstrated that the well known PDE5/6 inhibitor Zaprinast blocks pyruvate transport into mitochondria by a mechanism that is independent of PDE inhibition. Inhibition of pyruvate transport either by Zaprinast or by known MPC inhibitors decreases de novo synthesis of glutamate from glucose and aspartate, and it increases the net oxidation of glutamate into aspartate. The result is severe depletion of glutamate and severalfold accumulation of aspartate. These findings are significant because glutamate is important as a fuel for energy production, as a neurotransmitter, and as a substrate for glutathione synthesis.

The Effect of Zaprinast on Glutamate and Aspartate Occurs Independently of Its Previously Known Effects on PDE-
The ability of Zaprinast to inhibit PDE has been characterized (9, 10). Here we report four lines of evidence that Zaprinast also has an independent activity that causes severe depletion of glutamate and accumulation of aspartate. 1) The concentration at which Zaprinast decreases glutamate is about 20 times lower than that required to inhibit PDE. 2) Other PDE inhibitors, sildenafil or cell-permeable cGMP analogues, do not influence glutamate or aspartate. 3) Inactivation of cyclic nucleotide gated channels and inhibition of PKG or PKA do not rescue Zaprinast-induced glutamate depletion. 4) Zaprinast causes depletion of glutamate and accumulation of aspartate even in the retina, which has no PDE6.
Previous studies found that depletion of glutamate correlates with cell death in the retina (22,28). Interestingly, in long term organotypic retinal explant cultures, at concentrations of up to 200 M, Zaprinast induces selective photoreceptor death without affecting other retinal neurons (37). Why does Zaprinast not cause generalized cell death and instead appears to affect only PDE6 under culture conditions (37)(38)(39)(40)? Tissue culture medium, such as R16 (41), used for organotypic retinal culture contains both fatty acids and glutamine. This provides two opportunities to bypass the consequences of MPC inhibition, either via direct acetyl-CoA uptake from fatty acid oxidation or via glutamine to glutamate conversion. Hence, the composition of the culture medium may help to discriminate between PDE6 and MPC effects of Zaprinast in cultured retina.
Zaprinast Inhibits Transport of Pyruvate into Mitochondria-MPC transports pyruvate into mitochondria. UK5099 or ␣-cyano-4-hydroxycinnamic acid are known inhibitors of MPC activity (1,4). In this study we showed that MPC can also be inhibited by Zaprinast. Inhibition or knockdown of MPC prevents pyruvate-dependent acetyl-CoA formation and O 2 consumption (5,6,42) and decreases the lactate/pyruvate ratio (43). Our study shows that Zaprinast and UK5099 not only block pyruvate-dependent mitochondrial O 2 consumption and citrate synthesis in brain mitochondria but also cause increases in pyruvate and aspartate and decrease in glutamate in brain mitochondria and the retina.
We confirmed by direct measurements of pyruvate uptake that Zaprinast dose-dependently inhibits MPC activity in isolated mitochondria. MPC is an integral membrane protein in the mitochondrial inner membrane. To rule out any effect on    DECEMBER 13, 2013 • VOLUME 288 • NUMBER 50 PDH activity we measured the effect of Zaprinast on mitochondria whose membrane structure was disrupted by detergent.

Pyruvate Transport Balances Aspartate and Glutamate
There was no effect of Zaprinast on these preparations. This shows that Zaprinast does not target PDH or citrate synthase. We also confirmed directly that Zaprinast does not affect the activity of purified PDH. In our MPC activity assay some labeled pyruvate may have been metabolized. We did not correct for this or for labeled pyruvate that may have been at the surface of the pellets. This may explain why the pyruvate uptake continued for more than 1 min and neither cold pyruvate nor Zaprinast completely eliminated the labeled pyruvate in mitochondrial pellet. Future studies should address the structural basis for Zaprinast inhibition of MPC and whether Zaprinast directly binds MPC. Zaprinast Causes Oxidation of Glutamate and Accumulation of Aspartate-Hypoglycemia or inhibition of glycolysis leads to decreased glutamate and cellular accumulation or release of aspartate in brain (44 -46), synaptosomes (47), and the retina (22,28). During glucose deprivation, neurons turn to glutamine and glutamate for energy (48). Zaprinast does not affect glucose uptake or glycolysis (as assessed by lactate production), but its block of pyruvate transport mimics hypoglycemic conditions because less glucose-derived pyruvate is available to mitochondria. The lack of two carbons from pyruvate causes oxaloacetate to accumulate at the expense of glutamate. Some pyruvate may enter mitochondria before Zaprinast takes effect or there might be some residual MPC activity, so some citrate, ␣KG, glutamate, and succinate can be labeled by 13  A s p a r t a t e L a c t a t e P y r u v a t e C i t r a t e α K G  glucose. However, once these labeled carbons are incorporated into oxaloacetate, there is so little acetyl-CoA that they cannot enter a second turn of TCA cycle. The result is similar to what happens during glucose deprivation. Glutamine starts to replenish the TCA cycle to rescue the mitochondrial energy crisis, but it can only lead to accumulation of more oxaloacetate. By Le Chatelier's principle, the increased oxaloacetate will shift aspartate transaminase toward producing more aspartate and more ␣KG from glutamate. In the retina, Ͼ60% of glutamate de novo synthesis depends on transaminases such as alanine transaminase, aspartate transaminase, and branched amino acids transaminase (49). However, only aspartate and not alanine or branched amino acids increase with Zaprinast. Glutamate may also decarboxylate into GABA, but Zaprinast does not increase GABA content in the retina. Therefore, the amino group of aspartate most likely comes from glutamate transamination (49).
In conclusion, independently of its known effects on PDE, Zaprinast potently inhibits pyruvate transport into mitochondria, impairing glucose oxidation and mitochondrial function. Inhibition of pyruvate transport mimics the effect of hypoglycemia: accumulation of aspartate, depletion of glutamate and glutamine, and mitochondrial dysfunction. Our findings suggest a new route for pharmacological application and development of Zaprinast, they highlight the physiological influence of MPC on metabolism, and they contribute to understanding of the basic metabolic reaction to hypoglycemia.