Feasibility of Pathways for Transfer of Acyl Groups from Mitochondria to the Cytosol to Form Short Chain Acyl-CoAs in the Pancreatic Beta Cell*

The mitochondria of pancreatic beta cells are believed to convert insulin secretagogues into products that are translocated to the cytosol where they participate in insulin secretion. We studied the hypothesis that short chain acyl-CoA (SC-CoAs) might be some of these products by discerning the pathways of SC-CoA formation in beta cells. Insulin secretagogues acutely stimulated 1.5–5-fold increases in acetoacetyl-CoA, succinyl-CoA, malonyl-CoA, hydroxymethylglutaryl-CoA (HMG-CoA), and acetyl-CoA in INS-1 832/13 cells as judged from liquid chromatography-tandem mass spectrometry measurements. Studies of 12 relevant enzymes in rat and human pancreatic islets and INS-1 832/13 cells showed the feasibility of at least two redundant pathways, one involving acetoacetate and the other citrate, for the synthesis SC-CoAs from secretagogue carbon in mitochondria and the transfer of their acyl groups to the cytosol where the acyl groups are converted to SC-CoAs. Knockdown of two key cytosolic enzymes in INS-1 832/13 cells with short hairpin RNA supported the proposed scheme. Lowering ATP citrate lyase 88% did not inhibit glucose-induced insulin release indicating citrate is not the only carrier of acyl groups to the cytosol. However, lowering acetoacetyl-CoA synthetase 80% partially inhibited glucose-induced insulin release indicating formation of SC-CoAs from acetoacetate in the cytosol is important for insulin secretion. The results indicate beta cells possess enzyme pathways that can incorporate carbon from glucose into acetyl-CoA, acetoacetyl-CoA, and succinyl-CoA and carbon from leucine into these three SC-CoAs plus HMG-CoA in their mitochondria and enzymes that can form acetyl-CoA, acetoacetyl-CoA, malonyl-CoA, and HMG-CoA in their cytosol.

Mitochondria play two important roles in insulin secretion. One role is ATP production, which in addition to powering numerous cellular processes triggers insulin exocytosis via its acting on the plasmalemmal ATP-sensitive potassium channel. In addition, it is well established that beta cell mitochondria use carbon from glucose-derived pyruvate to synthesize various metabolic intermediates (anaplerosis) (1,2). About one-half of the pyruvate derived from glucose, the most potent insulin secretagogue, is carboxylated by pyruvate carboxylase to oxaloacetate (2)(3)(4)(5)(6)(7). Although most of this oxaloacetate is used for anaplerosis and many studies have suggested that anaplerosis is important for insulin secretion (8 -12), the purpose of anaplerosis in the beta cell is unclear. It seems clear, however, that if the carboxylation of pyruvate increases the levels of citric acid cycle intermediates inside the mitochondria, this would alter mitochondrial function because many citric acid cycle intermediates inhibit or activate citric acid cycle enzymes (as described in biochemistry texts and briefly reviewed in Refs. 2 and 13). Therefore, this indicates that these metabolites must be transferred to the extramitochondrial space where they might have signaling or supporting roles in insulin secretion.
To explain part of the role of anaplerosis in insulin secretion, we recently proposed the "succinate mechanism" (14). This hypothesis was based on our own data and data from the literature and speculated that succinate metabolism supplies both NADPH and short chain acyl-CoA precursors of mevalonic acid that have signaling or supporting roles in insulin secretion (2,14). Although succinate metabolism was instrumental in suggesting the original hypothesis, succinate has become less central to the hypothesis as it has evolved, in part, into a scheme involving the mitochondrial synthesis of multiple short chain acyl-CoAs from the carbon derived from all metabolized insulin secretagogues. Another hypothesis of insulin secretion that has been discussed for 2 decades involves the mitochondrial export of citrate as a carrier of acetyl groups to the cytosol where they are converted to malonyl-CoA, which has a special role in insulin secretion via its inhibition of mitochondrial fatty acid oxidation. This is believed to increase cytosolic long chain acyl-CoA molecules, which might perform signaling roles in insulin secretion (for a review see Ref. 15). Previous reports have described the possible involvement of NADPH in insulin secretion (6,8,10,16,17). In this study we examined the possibility that multiple short chain acyl-CoAs, not just mevalonic acid or malonyl-CoA, might be some of the many products of anaplerosis in the beta cell. We first showed that acetyl-CoA, acetoacetyl-CoA, HMG-CoA, 2 succinyl-CoA, and malonyl-CoA are increased to various extents in INS-1 cells stimulated by many different insulin secretagogues as judged from liquid chromatography-tandem mass spectrometry measurements. We then considered the pathways by which multiple short chain acyl-CoAs might be synthesized from insulin secretagogues in mitochondria and their acyl groups transferred to the cytosol where they could be re-activated with coenzyme A and have roles in insulin secretion. Acyl-CoA molecules themselves cannot be transported across the inner mitochondrial membrane and, therefore, after they are synthesized in mitochondria, the coenzyme A moiety is removed, and their acyl groups are transported as carboxylate ions out of the mitochondria to the extramitochondrial space. Outside the mitochondria they are reconverted to their CoA derivatives. The pathways of the original "succinate mechanism" scheme (14) were proposed on the assumption of pathways known to be present in nonbeta cells without verification of the presence of the enzymes in the beta cell that might synthesize and metabolize short chain acyl-CoAs or, if present, their intracellular location (mitochondrial versus extramitochondrial). Thus, the scheme did not fully address the pathways by which short chain acyl-CoAs could be synthesized in mitochondria and re-formed in the cytosol. The malonyl-CoA hypothesis seemed to consider citrate as the only carrier of short chain acyl groups from mitochondria to the cytosol (15).
To discern the feasibility of the various possible relevant pathways, we determined the presence or absence, as well as the intracellular locations, of short chain acyl-CoA-synthesizing enzymes in rat and human pancreatic islets and in INS-1 832/13 cells with measurements of their enzyme activities, immunoblot analysis of their protein levels, and/or RT-PCR analysis of their cognate mRNAs. The results showed that beta cells possess enzymes that can form all of the short chain acyl-CoAs studied except malonyl-CoA in their mitochondria and all of the CoAs except succinyl-CoA in their extramitochondrial space. Proof that citrate is not the only carrier of acyl groups from mitochondria to the cytosol was obtained by lowering ATP citrate lyase activity with shRNA and showing that this did not inhibit glucose-induced insulin release. Evidence that acetoacetate is a carrier of acyl groups from mitochondria to the cytosol in the beta cell was obtained by lowering acetoacetyl-CoA synthetase activity with shRNA and observing that this partially inhibited insulin release. Thus, we believe the data suggest that multiple short chain acyl-CoAs formed by anaplerosis could play roles in signaling or supporting insulin secretion and that there are at least two pathways, one involving acetoacetate and the other citrate, for translocation of acyl groups from mitochondria to the cytosol in beta cells.

EXPERIMENTAL PROCEDURES
Materials-␤- [3-14 C]Hydroxybutyrate was from American Radiolabeled Chemicals, Inc. Succinic acid monomethyl ester and all other chemicals were from Sigma. Auxiliary enzymes used in analysis of enzyme activity were from Roche Diagnostics. An antibody against rat acetoacetyl-CoA synthetase was a generous gift of Dr. Tetsuya Fukui (18). Rabbit antibodies against the ␤ subunits of rat succinyl-CoA synthetase were from David O. Lambeth (19). A rabbit polyclonal antiserum against the rat peroxisomal acetyl-CoA acetyltransferase (Acaa1a) 3 was from Paul Van Veldhoven (20). An antibody to human HMG-CoA lyase was from Grant Mitchell (21). A rabbit polyclonal antibody to human HMG-CoA reductase was from Upstate Cell Signaling Solutions (Lake Placid, NY). A rabbit polyclonal antibody to human cytosolic acetyl-CoA acetyltransferase (ACAT2) was from Abcam. Human pancreatic islets were from the Islet Isolation Core at Washington University School of Medicine, St. Louis. INS-1 832/13 cells were from Chris Newgard (22).
Short Chain Acyl-CoAs-INS-1 832/13 cells were maintained as monolayers on 150-mm diameter plates in INS-1 medium (RPMI 1640 tissue culture medium (the glucose concentration in this medium is 11.1 mM) supplemented with 10% fetal bovine serum, 1 mM pyruvate, 50 M ␤-mercaptoethanol, and 10 mM Hepes buffer (23)) and penicillin (100 units/ml) and streptomycin (100 g/ml). Twenty four hours before an experiment was to be performed, this medium was replaced with fresh medium modified to contain 5 mM glucose. On the day of the experiment plates were washed twice with 10 ml of phosphate-buffered saline and once with 10 ml of Krebs-Ringer bicarbonate solution modified to contain 15 mM sodium bicarbonate and 15 mM sodium Hepes buffer, pH 7.3. Secretagogues were added to the plates in 10 ml of the modified Krebs-Ringer solution, and after 30 min at 37°C, this solution was quickly withdrawn and replaced with 2 ml of 1% trifluoroacetic acid in 50% methanol. The extracts were then prepared for analysis and analyzed by liquid chromatography-electrospray ionization-tandem mass spectrometry exactly as described previously (24).
Subcellular Fractionation-Subcellular fractionation of pancreatic islets was as described previously (6). Briefly, islets were homogenized in 220 mM mannitol, 70 mM sucrose, 5 mM potassium Hepes buffer, pH 7.5 (MSH). The homogenate was centrifuged at 600 ϫ g for 10 min to precipitate the nuclei and cell debris fraction, and the resulting supernatant fraction was centrifuged at 5,500 ϫ g for 10 min to precipitate the mitochondrial fraction. The resulting supernatant fraction or a 20,000 ϫ g for 20 min supernatant fraction was the cytosol. In some instances, the 5,500 ϫ g supernatant fractions were centrifuged at 20,000 or 50,000 ϫ g for 20 min to obtain a pellet enriched in microsomes and peroxisomes. Liver, kidney, and heart subcellular fractions were prepared as described previously (25). INS-1 cells were fractionated with reagents from the mitochondrial isolation kit (catalog number 8974, Pierce). About 100 l of loosely packed INS-1 832/13 cells were homogenized in 800 l of mitochondria isolation reagent A containing a protease inhibitor EDTA-free mixture (Pierce). After vortexing and a 2-min incubation on ice, 10 l of reagent B was added. The mixture sat on ice for 5 min and was then homogenized with 30 strokes up and down in a Potter-Elvehjem homogenizer. Reagent C (800 l) containing Halt protease inhibitor was added, and the mixture was centrifuged at 700 ϫ g for 10 min. The resulting supernatant fraction was centrifuged at 12,000 ϫ g for 10 min to give the mitochondrial pellet that was resuspended in MSH solution and freeze-thawed three times before use. The resulting supernatant fraction (cytosol) and mitochondrial fraction were used for enzyme assays and immunoblot analysis. In some instances, the resulting supernatant fraction was centrifuged at higher speeds to obtain a pellet enriched in microsomes and peroxisomes as described for islets.
Enzyme Assays-Succinyl-CoA:3-ketoacid CoA transferase was measured in the presence of 50 mM sodium acetoacetate, 0.2 mM succinyl-CoA, 5 mM MgCl 2 , 5 mM iodoacetamide in 50 mM Tris chloride buffer, pH 8.0 (26), at 30°C. The rate of acetoacetyl-CoA formation was followed by measuring the increase in absorbance at 310 nm. Acetyl-CoA acetyltransferase was measured in the presence of 25 M acetoacetyl-CoA, 60 M coenzyme A, 5 mM MgCl 2 in 50 mM Tris chloride buffer, pH 7.6, at 30°C. The disappearance of acetoacetyl-CoA was followed spectrophotometrically by monitoring the decrease in absorbance at 303 nm (27). Acetoacetyl-CoA synthetase was measured with slight modifications of a radiochemical assay (28). The complete assay mixture contained final concentrations of 0.5 mM ␤-[3-14 C]hydroxybutyrate (specific radioactivity, 3 mCi/mmol), 1 mM coenzyme A, 1.5 mM ATP, 3 mM NAD, 10 mM MgCl 2 , 100 mM KCl, and 0.05 units/ml of ␤-hydroxybutyrate dehydrogenase from Rhodobacter sphaeroides in Tris chloride buffer, pH 7.5. The mixture was maintained at 37°C for 10 min to generate [3-14 C]acetoacetate before pancreatic islet cytosol or INS-1 cytosol was added to bring the final volume to 50 l. After 20 more min the enzyme reaction was stopped by the addition of 12 l of acetic acid to the reaction mixture, and part of or the entire volume was spotted on Whatman SG81 chromatography paper. Chromatography was run in a mixture of ethyl ether:formic acid (7:1) for 60 -90 min. Acetoacetyl-CoA remains at the origin and acetoacetate and 3-hydroxybutyrate migrate with the solvent front. Liquid scintillation spectrometry was used to measure radioactivity in squares of paper cut from the origin to estimate synthetase activity. Radioactivity in blanks containing cytosol and reaction mixtures minus ATP was subtracted from the radioactivity in samples containing the complete reaction mixture plus cytosol to calculate activity attributable to the synthetase. The activity with ATP present in the enzyme assay mixture was 17-26-fold higher than with no ATP present or when a blank with no cellular extract was present in the reaction mixture when INS-1 cytosol was used and 5-fold higher when rat islet cytosol was used. ATP citrate lyase activity was measured in a reaction mixture containing 5 mM citrate, 0.3 mM coenzyme A, 3 mM ATP, 0.15 mM NADH, 10 mM MgCl 2 , 10 mM dithiothreitol, and 6 units/ml of malate dehydrogenase from pig heart mitochondria in Tris chloride buffer, pH 8.5, at 37°C (29). The disappearance of NADH, which corresponded to oxaloacetate formation, was monitored spectrophotometrically at 340 nm. Glutamate dehydrogenase activity was measured in the presence of 10 mM ␣-ketoglutarate, 50 mM NH 4 Cl, 100 M NADH, 2 mM ADP, 100 M EDTA, 1 mM KCN, 0.1% Triton X-100, and 50 mM imidazole buffer, pH 7.0 at 30°C (30). NADH disappearance was monitored spectrophotometrically at 340 nm. In each spectrophotometric assay, the background rate measured in the absence of one substrate was subtracted from the rate in the presence of the complete reaction mixture to give the rate attributable to the enzyme. Usually for positive controls, enzyme activities were measured in non-islet tissues known to contain the enzyme of interest (data not shown).
RT-PCR-Total RNA was isolated from cultured INS-1 (832/ 13) cells and from freshly isolated rat islets and liver using TRIzol reagent (Invitrogen) and further purified on an RNeasy Mini column (Qiagen). cDNA was synthesized using Moloney murine leukemia virus reverse transcriptase (Invitrogen) and oligo(dT) primers (Promega, Madison, WI). PCR was performed with the primers shown in supplemental Table 1. cDNAs from tissues known to contain specific transcripts were run as positive controls (data not shown).
Immunoblotting-For immunoblotting of HMG-CoA synthase, rat islets and INS-1 cells, or human islets, were washed twice in phosphate-buffered saline and lysed with M-PER mammalian protein extraction reagent (Pierce) containing protease inhibitors. Rat liver and heart tissue were homogenized, and protein was extracted using T-PER tissue protein extraction reagent (Pierce) containing protease inhibitors. Whole rat islets were either homogenized in MSH and boiled in sample buffer or boiled directly in sample buffer (1% SDS, 5% glycerol, 65 mM Tris chloride, pH 6.8, 0.01% bromphenol blue, and either 100 mM dithiothreitol or 2.5% ␤-mercaptoethanol). Proteins were separated by electrophoresis on 7.5% SDS-polyacrylamide gels and electrotransfered to nitrocellulose membranes. The membranes were blocked with a mixture of 5% nonfat milk and 1% bovine serum albumin and 0.05% Tween 20 in Tris-buffered saline (TBST) and incubated overnight at 4°C with primary antiserum diluted 1:500 to 1:2000 in blocking buffer. Primary antisera were raised in rabbits immunized with internal peptides of the rat cytosolic HMG-CoA synthase (HIPSPA-KKVPRLPAT) and the rat mitochondrial HMG-CoA synthase (CSTIPPAPLAKTDT) conjugated to keyhole limpet hemocyanin (the amino-terminal cysteine was added to the native sequence to facilitate conjugation). The membranes were washed and treated with the secondary antibody labeled with goat anti-rabbit horseradish peroxidase diluted 1:15,000 in blocking buffer, and proteins were visualized with the Supersignal West Pico chemiluminescent substrate kit (Pierce). Luminescence was captured on a Chemi Doc XRS Gel documentation system (Bio-Rad) or on KSR-B Luc Ultra Autorad x-ray film (ISC Bioexpress). To demonstrate equal loading of proteins per lane, the membranes were then washed in TBST and incubated in Restore Western blot stripping buffer (Pierce) for 30 min at 42°C and re-probed with anti-␤-actin antibody (Sigma). Immunoblotting to detect other proteins was done similarly using dilutions of antibody recommended by the supplier. For each enzyme as appropriate, proteins were also measured in subcellular fractions from kidney, liver, or heart as positive or negative controls (data not always shown).
Silencing of ATP Citrate Lyase and Acetoacetyl-CoA Synthetase Genes-Five different 65-bp inserts containing 19-bp sequences targeted against the ATP citrate lyase gene were cloned into the BamHI and HindIII sites of the pRNA-U6.1/ Hygro plasmid downstream of the U6 promoter (Genescript) and transfected into INS-1 832/13 cells with Lipofectamine 2000 reagent (Invitrogen). Hygromycin-resistant cells were selected and maintained in INS-1 tissue culture medium containing 150 g/ml of hygromycin. The target sequence that provided the largest decrease in enzyme activity and enzyme protein was CAACAGACCTATGACTACG (nucleotides 1012-1030 of GenBank TM accession number NM_016987). Three 64 -69-bp inserts containing 19 -22-bp sequences targeted against the acetoacetyl synthetase gene were cloned into the BamHI and HindIII sites of plasmid pSilencer 2.1-U6 Hygro (Ambion) downstream of the U6 promoter and transfected into the INS-1 cells and selected with hygromycin. The target sequence that provided the largest decrease in enzyme activity and insulin release was ACAGCATGTTTCTGGATGA (nucleotides 872-891 of GenBank TM accession number NM_023104). Sequences that gave less knockdown of enzyme activity and insulin release were GGAATTACGATGACTTATA and GGAAG-GCTTACTTCTCCAAATA. As an additional control, cells were also transfected with a vector containing a nontargeting shRNA sequence.
Data Analysis-Statistical significance was confirmed with Student's t test.

RESULTS
Short Chain Acyl-CoA Measurements-The first phase of insulin release normally begins within 3 min of an increase in glucose or other insulin secretagogues that come in contact with the beta cell. The second phase of insulin release, in which the largest amount of insulin is released, is caused by metabolism of insulin secretagogues (31) and lasts for several hours or until the stimulus stops. We stimulated INS-1 832/13 cells with various insulin secretagogues, as well as nonsecretagogues as negative controls, and selected 30 min as a time point to measure short chain acyl-CoAs in beta cells when the cells would likely be showing changes in metabolite levels related to a stimulated state. Relative insulin release caused by the various secretagogues and nonsecretagogue controls is listed in the legend of Fig. 2. In INS-1 832/13 cells used for these studies numerous concomitant measurements of insulin release showed that glucose, pyruvate, and leucine plus glutamine were the most potent stimulants of insulin release from these cells, closely followed by 2 mM ␣-ketoisocaproate plus 10 mM monomethyl succinate (24). 4 Leucine provides a moderate stimulus of insulin release in these cells. 4 Incubation of INS-1 832/13 cells for 30 min under conditions that stimulate insulin release caused large relative increases in several short chain acyl-CoAs. The unstimulated concentration of acetoacetyl-CoA was the lowest of any of the CoAs measured, and the relative increases in acetoacetyl-CoA were the largest (2-5-fold). Succinyl-CoA, HMG-CoA, and malonyl-CoA were present at intermediate levels. Succinyl-CoA and malonyl-CoA were increased 2-3-fold. HMG-CoA was increased 30 -60% and acetyl-CoA, which was present at the highest level of the CoAs studied, was increased 55-80%. Pyruvate, which is as potent an insulin secretagogue as glucose in INS-1 832/13 cells (24,32,33), also caused large increases in the acyl-CoA levels (Fig. 2). In fresh rat pancreatic islets ␣-ketoisocaproate at a 10 mM concentration is as potent an insulin stimulant as glucose, but ␣-ketoisocaproate alone does not stimulate insulin in INS-1 832/13 cells (24). ␣-Ketoisocaproate alone at either 2 or 10 mM also caused large increases in the short chain CoAs. Methyl succinate alone and leucine alone each stimulate about one-third the amount of insulin release of glucose in fresh rat pancreatic islets (24, 34 -36). In INS-1 832/13 cells methyl succinate alone, like ␣-ketoisocaproate alone, does not stimulate insulin release (24). However, when ␣-ketoisocaproate (at only 2 mM) is applied along with methyl succinate to INS-1 cells, strong stimulation of insulin release occurs that is almost as large as that produced by glucose (24). 5 Methyl succinate alone increased succinyl-CoA, as might be expected from succinate's proximity to succinyl-CoA in metabolic pathways (Fig. 1, reactions 3 and 6). When methyl succinate was combined with ␣-ketoisocaproate, each of the five acyl-CoAs measured was increased (Fig. 2). Because ␣-ketoisocaproate alone caused large increases in all of the short chain acyl-CoAs, this may be one part of the mechanism by which ␣-ketoisocaproate interacts with methyl succinate to stimulate insulin release. Glutamine by itself does not stimulate insulin release in islets or INS-1 cells, and it did not increase the short chain acyl-CoA levels in the INS-1 832/13 cells (Fig. 2). In addition to its metabolism, leucine stimulates insulin release by allosterically activating glutamate dehydrogenase, and this enhances endogenous glutamate metabolism. When glutamine, which is a source of glutamate, is applied to islets or INS-1 cells together with leucine, it stimulates insulin release about equal to that produced by glucose (34,37,38). Leucine plus glutamine caused large increases in the CoAs. As shown previously (24), a high concentration of glucose caused the levels of various CoA molecules to increase, but somewhat less so than the other conditions (Fig. 2).
Pathways That Form Acyl-CoAs in Beta Cells-Because all of the known anaplerotic pathways that can form short chain acyl-CoAs or their precursors from glucose or other physiologic insulin secretagogues originate in mitochondria, it was assumed that increases in acyl-CoA levels can only arise from their initial synthesis in mitochondria. The inner mitochondrial membrane is not permeable to acyl-CoAs themselves. Therefore, in order for acyl-CoAs to increase in the cytosol, their acyl groups must be transported as carboxylates from the mitochondria to the cytosol where the coenzyme A is added back to the carboxylates. To discern the pathways by which this might occur, we looked at the intracellular location of relevant enzymes that can interconvert carboxylic acids with acyl-CoA molecules. The general approach was to use measurements of enzyme activity and/or immunoblotting to show the presence of enzymes and RT-PCR to prove the absence or confirm the presence of enzymes.

Multiple Short Chain Acyl-CoAs in the Beta Cell
into HMG-CoA only in the cytosol as shown in Fig. 1. However, mitochondrial HMG-CoA formed from leucine metabolism can be converted to acetyl-CoA and acetoacetate by HMG-CoA lyase (Fig. 1, reaction 4), which was detected in mitochondria of both rat and human pancreatic islets as well as INS-1 cells, as judged from immunoblot analysis (Fig. 4). HMG-CoA lyase has also been reported to be present in peroxisomes (21). Subcellular fractionation of islets and INS-1 832/13 cells showed this enzyme to be present predominantly, if not exclusively, in the mitochondrial fraction (data not shown). Succinyl-CoA Synthetase and Succinyl-CoA:3-Ketoacid CoA Transferase-There are two isoforms of succinyl-CoA synthetase, which is also called succinate thiokinase, a GTP-specific form and an ATP specific form. Succinyl-CoA synthetases catalyze the reversible interconversion of succinate and succinyl-CoA (Fig. 1, reaction 6). These isoforms possess identical catalytic ␣ subunits, and their regulatory ␤ subunits confer the nucleotide specificity. Both isoforms are present in the mitochondria in varying ratios in all body tissues studied (19). With immunoblot analysis with antibodies specific for each ␤ sub-unit, we found that rat pancreatic islets and INS-1 832/13 cells each possess both isoforms of succinyl-CoA synthetase in their mitochondrial fraction. Because the presence of either isoform in the cytosol would indicate that succinate formed in mitochondria and exported to the cytosol could be converted to succinyl-CoA in the cytosol, we attempted to show the presence of either isoform of the enzyme in the cytosol fraction of INS-1 832/13 cells and rat pancreatic islets. Numerous immunoblot analyses such as the one shown in Fig. 5, as well as measurements of enzyme activity (data not shown), failed to demonstrate the presence of either isoform of the enzyme in the cytosol. A previous study of succinyl-CoA synthetase in islets using the ␤ subunit-specific antibodies also found both isoforms of succinyl-CoA synthetase in mitochondria of INS-1 cells and HIT-15 cells (39). As judged from measurements of enzyme activity (Table 1) and RT-PCR analysis (Fig. 4, top  panel), succinyl-CoA:3-ketoacid CoA transferase is present in the mitochondria of islets and INS-1 832/13 cells. This enzyme catalyzes the reversible conversion of succinyl-CoA plus acetoacetate to acetoacetyl-CoA plus succinate (Fig. 1, reaction 3). human islets. This figure shows pathways by which short chain acyl-CoAs can be synthesized from insulin secretagogue carbon in mitochondria and their acyl groups transported across the inner mitochondrial membrane as either acetoacetate or citrate, which are re-converted to short chain acyl-CoAs in the extramitochondrial space. In the mitochondria, both acetyl-CoA acetyltransferase 1 (ACAT1) (previously called acetoacetyl-CoA thiolase) and acetyl-CoA acyltransferase 2 (ACAA2) (also called 3-ketoacyl-CoA thiolase) catalyze the same reaction (reaction 2). In the cytosol acetyl-CoA acetyltransferase 2 (ACAT2) and acetyl-CoA acyltransferase 2 (ACAA2) can catalyze the same reaction (reaction 9). Rat and human islets possess both the mitochondrial and cytosolic isoforms of each of these thiolase enzymes, and INS-1 832/13 cells possess all but the mitochondrial acetyl-CoA acetyltransferase (ACAT1). The scheme also shows that although succinate can be transported out of mitochondria, there is no enzyme in the extramitochondrial space that can convert succinate into succinyl-CoA. PC, pyruvate carboxylase and PDH, pyruvate dehydrogenase.
Acetyl-CoA Acetyltransferases and Acetyl-CoA Acyltransferases-Acetyl-CoA acetyltransferase, previously called acetoacetyl-CoA thiolase, catalyzes the reversible conversion of one molecule of acetoacetyl-CoA to two molecules of acetyl-CoA. An enzyme assay showed that acetyltransferase enzyme activity was present in both mitochondria and cytosol of rat pancreatic islets and INS-1 cells ( Table 1). As judged from RT-PCR analyses (Fig. 6) and immunoblotting (Fig. 4, 2nd panel from bottom), cytosolic acetyl-CoA acetyltransferase (ACAT2) (Fig. 1,   reaction 9) is present in rat and human pancreatic islets and INS-1 832/13 cells. Although transcripts for the mitochondrial acetyl-CoA acetyltransferase (ACAT1) (Fig. 1, reaction 2) were shown to be present in both rat and human pancreatic islets, they could not be identified in INS-1 832/13 cells (Fig. 6). Therefore, the acetyltransferase enzyme activity in INS-1 cell mitochondria could not be attributed to ACAT1 (Fig. 1, reaction 2). However, RT-PCR analysis showed that mitochondrial acetyl-CoA acyltransferase (Acaa2) (also called 3-ketoacyl-CoA thiolase) transcripts were present in INS-1 832/13 cells (Fig. 6). This enzyme can catalyze the same reaction as ACAT1 and can explain the presence of acetyl-CoA acetyltransferase activity in INS-1 cell mitochondria. As judged from immunoblot analysis, peroxisomal acetyl-CoA acyltransferase (ACAA1) (Fig. 1, reaction 9) is present in rat and human pancreatic islets and INS-1 832/13 cells (Fig. 4, bottom panel).
Acetoacetyl-CoA Synthetase-Acetoacetyl-CoA synthetase is found exclusively in the cytosol of cells. It catalyzes the unidirectional conversion of acetoacetate and coenzyme A to acetoacetyl-CoA. The breakdown of ATP to AMP plus pyrophosphate drives the reaction (Fig. 1, reaction 8). As judged from

. HMG-CoA synthase is present in cytosol but not in mitochondria of rat pancreatic islets and INS-1 cells.
A depicts an RT-PCR experiment that shows that transcripts from the gene, which encodes the cytosolic isoform of HMG-CoA synthase but no transcripts from the gene which encodes the mitochondrial isoform of the enzyme, are present in islets and INS-1 832/13 cells. Glutamate dehydrogenase gene transcripts (Glutamate DH) were estimated to show uniform loading of cDNA. B depicts immunoblot experiments that show that cytosolic HMG-CoA synthase protein, but no protein of the mitochondrial isoform of the enzyme, is present in rat islets and INS-1 832/13 cells. ␤-Actin was probed to show uniform loading of protein across lanes. There was 40 g of protein/lane. In the RT-PCR and immunoblot experiments, liver tissue was used as a positive control. measurements of enzyme activity (Table 1), immunoblot (Fig.  4, 3rd panel down), and RT-PCR analyses (data not shown), this enzyme is present in rat and human islets and INS-1 cells, and its level in these cells is as high or higher than in liver.
ATP Citrate Lyase and Acetoacetyl-CoA Synthetase Knockdown-ATP citrate lyase enzyme activity was, of course, found in islets and INS-1 cells (Table 1). Another indication that there is more than one pathway for the export of short chain acyl groups from mitochondria to the cytosol and their conversion to short chain acyl-CoAs in the cytosol comes from studies in which the expression of ATP citrate lyase (Fig. 1,  reaction 12) was almost eliminated with shRNA in INS-1 832/13 cells without inhibiting insulin release. ATP citrate lyase mRNA was decreased 87% versus the parent INS-1 832/13 cells and 80% versus INS-1 832/13 cells transfected with a vector containing no target sequence, as judged by quantitative PCR. Enzyme activity was knocked down 88 Ϯ 1% (mean Ϯ S.E., n ϭ 7) in our stable cell line ACL 940-12 (passages 12-28) versus the control cells transfected with either no vector or a vector containing a nontargeting insert without inhibiting glucoseor pyruvate-stimulated insulin release in four separate insulin release studies spread out over 1 year (data not shown). The activities of control enzymes, aspartate aminotransferase, malic enzyme, isocitrate dehydrogenase, and acetoacetyl-CoA synthetase were not lowered in the ATP citrate lyase-deficient clone. These results are consistent with the idea that molecule(s) other than citrate can exit mitochondria and be converted to acyl-CoAs in the cytosol, most likely  from exported acetoacetate (Fig. 1, reactions 8 -11 and 13). In support of this idea, acetoacetyl-CoA synthetase enzyme activity was lowered 29 -80% with shRNA constructs in three different INS-1 832/13 cell lines versus the parent INS-1 832/13 cell line or a control cell line transfected with a nontargeting shRNA sequence. This caused proportional 30 -57% decreases in glucose-stimulated insulin release in the enzyme-deficient cell lines versus the control cell lines (Fig. 7). Enzyme activities of the enzymes ATP citrate lyase, pyruvate carboxylase, aspartate aminotransferase, malic enzyme, and isocitrate dehydrogenase, measured as controls, were not lowered in the cells with deficient acetoacetyl-CoA synthetase.

Formation of Short Chain Acyl-CoAs in Mitochondria and
Translocation of Acyl Carbon to the Cytosol-There is not a single product or class of products of anaplerosis that mediates insulin secretion. Measurements of short chain acyl-CoAs in secretagogue-stimulated INS-1 832/13 cells indicated that short chain acyl-CoAs may be some of the anaplerotic products formed from anaplerosis in the beta cell. The measurements showed relative increases in acetoacetyl-CoA Ն succinyl-CoA Ϸ malonyl CoA Ͼ acetyl-CoA Ϸ HMG-CoA (Fig. 2). Fig.  1 shows the intracellular locations of enzymes that can form these short chain acyl-CoAs from various insulin secretagogues in rat and human pancreatic islets and INS-1 cells and was developed from the measurements of activities and immunoblot analyses of relevant enzymes as well as RT-PCR analyses of their cognate mRNAs. These enzyme studies by themselves show it is feasible for carbon from insulin secretagogues to be incorporated into acyl-CoAs in mitochondria and their acyl groups translocated to the cytosol not only as citrate but also as acetoacetate. In the cytosol coenzyme A can be added to the acetate component of citrate or to acetoacetate directly (Fig. 1,  reactions 12 or 8), leading to the formation of various short chain acyl-CoAs. The successful shRNA knockdown of ATP citrate lyase (Fig. 1, reaction 12) enzyme activity without a decrease in insulin release indicates that citrate is not the only carrier of acyl groups from mitochondria to the cytosol. The idea that acetoacetate can also transfer acyl groups in insulin secretion was previously suggested by our finding increased acetoacetate levels in pancreatic islets and INS-1 832/13 cells stimulated with insulin secretagogues (24) as well as our finding increased acetoacetyl-CoA in secretagogue-stimulated INS-1 832/13 cells in this study (Fig. 2). To test whether acetoacetate could transfer acyl group precursors of acyl-CoAs from the mitochondria to the cytosol, we used shRNA to knock down acetoacetyl-CoA synthetase, an enzyme that converts acetoacetate to acetoacetyl-CoA in the extramitochondrial space (Fig.  1, reaction 8). The activity of this enzyme was decreased to various extents in three lines of INS-1 cells, and glucose-induced insulin in these lines was partially decreased in proportion to the decreases in enzyme activity (Fig. 7). After acetoacetate is converted to acetoacetyl-CoA by acetoacetyl-CoA synthetase in the cytosol, acetoacetyl-CoA can be converted to acetyl-CoA by either cytosolic acetyl-CoA acetyltransferase (ACAT2) or acetyl-CoA acyltransferase (ACAA1) that each catalyze the same reaction (Fig. 1, reaction 9). 3 Malonyl-CoA and HMG-CoA can each be formed from acetyl-CoA in the cytosol via reactions 13 and 10, respectively (Fig. 1). When only acetoacetyl-CoA synthetase (Fig. 1, reaction 9) is deficient, ATP citrate lyase (Fig. 1, reaction 12) should still be able to form acetyl-CoA, which is a substrate for reactions 10 and 13. Therefore, the partial inhibition of insulin release is somewhat surprising and apparently means that acetoacetyl-CoA synthase is very important for insulin secretion.

TABLE 1 Intracellular location of enzyme activities that influence short chain acyl-CoA metabolism in rat pancreatic islets and in INS-1 832/13 cells
Enzyme activities were measured in mitochondria and cytosol (the extramitochondrial fraction) of both INS-1 832/13 cells and rat pancreatic islets. The percentage of total enzyme activity in a subcellular fraction was determined by multiplying the activity in the fraction assayed by the ratio of total volume of the subcellular fraction to the volume of the fraction assayed. In the case of enzyme activities known to be present in both mitochondria and cytosol in non-islet tissues, glutamate dehydrogenase activity was measured to give an estimate of mitochondrial matrix enzymes from mitochondrial damage contributing to cytosolic enzyme activities. The percentage of an enzyme activity in the cytosol attributable to mitochondria was then subtracted from the total enzyme activity to give the activity attributable to the cytosolic enzyme. Results are the means Ϯ S.E. of duplicate measurements from three or more preparations of islets or INS-1 832/13 cells.

Enzyme
Specific enzyme activity Percentage of total enzyme activity in cytosol nmol product formed or substrate used/min/mg protein % Pathways of Acyl-CoA Formation from Physiologic Insulin Secretagogues- Fig. 1 shows how carbon derived from various insulin secretagogues, such as from the three major secretagogues that stimulate insulin in vivo, glucose, leucine alone, and glutamine in the presence of leucine, can be incorporated into the acyl components of short chain acyl-CoAs in mitochondria and their acyl groups translocated to the cytosol and reconverted to short chain acyl-CoAs. (Glutamine only stimulates insulin secretion when leucine is present to activate glutamate dehydrogenase, which enhances metabolism of glutamine-derived glutamate (34,37,38).) As depicted in Fig. 1, upper left corner, inside the mitochondrion carbon from glucose-derived pyruvate can be decarboxylated to acetyl-CoA catalyzed by pyruvate dehydrogenase and carboxylated to oxaloacetate, catalyzed by pyruvate carboxylase. The acetyl-CoA and oxaloacetate can combine to form citrate in the citrate synthase reaction (Fig. 1, reaction 1). Alternatively, the acetyl-CoA can be converted to acetoacetyl-CoA catalyzed by either acetyl-CoA acetyltransferase or acetyl-CoA acyltransferase, which each catalyze the same reaction (Fig. 1, reaction 2). The acetoacetyl-CoA can be converted to acetoacetate by succinyl-CoA:3-ketoacid CoA transferase (Fig. 1, reaction 3). Thus, carbon in the acetyl group of acetyl-CoA, originally from glucose, can be carried to the extramitochondrial space to form short chain acyl-CoAs in either citrate via the sequential reactions 1,12,9,10, and 13 shown in Fig. 1 or in acetoacetate via the sequential reactions 2, 3, 8 -10, and 13 shown in Fig. 1. HMG-CoA can be formed inside mitochondria from ␣-ketoisocaproate, the first metabolite of leucine, whereas carbon from glucose can be incorporated into HMG-CoA only in the cytosol of beta cells. This is because beta cells possess only cytosolic HMG-CoA synthase (Fig. 1, reaction 10) and lack mitochondrial HMG-CoA The p value that confirms the difference between each glucose-stimulated enzyme-deficient cell line, and the controls are shown near each bar. synthase (Fig. 3). Several pathways can export carbon from leucine and ␣-ketoisocaproate from mitochondria to the cytosol. Carbon from ␣-ketoisocaproate can form both acetoacetate and acetyl-CoA in the mitochondria catalyzed by HMG-CoA lyase (Fig. 1, reaction 4). The acetoacetate can be directly exported to the cytosol to form short chain acyl-CoAs via reactions 8 -10, and 13 in Fig. 1. In addition, the acetyl group of HMG-CoA-derived acetyl-CoA can be converted to acetoacetate in the mitochondria via reactions 4, 2 and 3 shown in Fig. 1, and the acetoacetate exported to the cytosol or the acetyl group can be exported from mitochondria to the cytosol as citrate. The acetoacetate formed by HMG-CoA lyase (Fig. 1,  reaction 4) could also be converted to acetyl-CoA and then citrate in the mitochondria via reactions 3, 2 and 1 shown in Fig.  1 and the citrate exported to the extramitochondrial space.
Stimulation of insulin release by leucine plus glutamine is about as potent as glucose-stimulated insulin release in islets and INS-1 cells, and this combination of amino acids caused large increases in the short chain acyl-CoAs (Fig. 2). Mitochondrial formation of short chain acyl-CoAs from the carbon of leucine plus glutamine and the translocation of acyl carbon as acetoacetate and/or citrate to the cytosol and the re-formation of short chain acyl-CoAs in the cytosol can occur via numerous interconnected pathways (Fig. 1). Leucine can be converted to ␣-ketoisocaproate and ␣-ketoisocaproate converted to HMG-CoA in the mitochondrion, and then HMG-CoA can be metabolized to acetyl-CoA and acetoacetate (Fig. 1, reaction 4), which can be converted to other acyl-CoAs as described above. Leucine also allosterically activates glutamate dehydrogenase, which will increase the rate of conversion of glutamate to ␣-ketoglutarate. The ␣-ketoglutarate can be converted to succinyl-CoA via ␣-ketoglutarate dehydrogenase (Fig. 1, reaction 5) and then to other short chain acyl-CoAs when acetoacetate derived from ␣-ketoisocaproate combines with the succinyl-CoA (Fig.  1, reactions 4, 3, and 2). In addition, ␣-ketoglutarate can be converted to malate via reactions of the citric acid cycle. Malate can be converted to pyruvate by malic enzyme in the cytosol and then pyruvate can enter the mitochondria where it can be converted to acetyl-CoA by pyruvate dehydrogenase (4,9). The acetyl groups can be exported to the cytosol as acetoacetate and citrate as described in the previous paragraph.
Short Chain Acyl-CoAs Are Only Some of the Products of Anaplerosis in the Beta Cell-The data in Fig. 2 show that although multiple short chain acyl-CoAs might be involved in insulin secretion, they are not sufficient to support insulin secretion. Both methyl succinate and ␣-ketoisocaproate by themselves are potent insulin secretagogues in pancreatic islets, but by themselves do not stimulate insulin release from INS-1 cells (24). The reason for this is not known. However, because ␣-ketoisocaproate alone does not stimulate insulin release in INS-1 cells, but it increases all short chain acyl-CoAs and methyl succinate increases succinyl-CoA in INS-1 cells, this indicates that other factors besides generation of short chain acyl-CoAs are necessary for insulin secretion. One of these factors, as suggested in the original succinate mechanism scheme (14), may be the generation of NADPH equivalents by mitochondria and their export to the cytosol (6,10,16,17), and this may be one of the factors that methyl succinate contributes that allows it to synergize with ␣-ketoisocaproate to stimulate nearly as much insulin release as glucose does in INS-1 cells (24). Methyl succinate increases malate up to 30-fold in islets or INS-1 cells (6,40), and malate will increase cytosolic NADPH via the malic enzyme reaction (2,6,10,41,42). Another mechanism by which methyl succinate and ␣-ketoisocaproate may interact synergistically in INS-1 cells is via succinate derived from methyl succinate combining with acetoacetyl-CoA derived from ␣-ketoisocaproate to increase succinyl-CoA and acetoacetate in the succinyl-CoA:3-ketoacid CoA transferase reaction (Fig. 1, reaction 3). The acetoacetate can be transported out of mitochondria to form short chain acyl-CoAs as described above.
Succinate Cannot Be a Direct Precursor of Cytosolic Short Chain Acyl-CoAs-As shown in the lower left of Fig. 1, within the mitochondria, carbon from glucose can be incorporated into succinyl-CoA via the reactions of the citric acid cycle and also carbon from glutamate can be incorporated into succinyl-CoA via the reactions catalyzed by glutamate dehydrogenase and ␣-ketoglutarate dehydrogenase (Fig. 1, reaction 5). The succinyl-CoA can be converted to succinate by succinyl-CoA synthetase, a citric acid cycle enzyme (Fig. 1, reaction 6). Succinate generated in the mitochondria can be transported across the inner mitochondrial membrane on the dicarboxylate transporter to the cytosol. If succinyl-CoA synthetase was present in the extramitochondrial space, succinate could be converted to succinyl-CoA in this compartment. Both the GTP and the ATP isoforms of this enzyme have been described to be present in various ratios in mitochondria of mammals and birds (19); and our immunoblot analyses and measurements of enzyme activity with rat islets and INS-1 832/13 cells demonstrated both isoforms of the enzyme in their mitochondria but not in their cytosol. Because succinyl-CoA levels were increased in the secretagogue-stimulated INS-1 cells and there does not seem to be a mechanism to produce succinyl-CoA in the cytosol, it appears that succinyl-CoA can have only an intramitochondrial role in insulin secretion. This role might be very important. By combining with acetoacetyl-CoA in the succinyl-CoA:3-ketoacid CoA transferase reaction (Fig. 1, reaction 3), succinate will increase the level of acetoacetate that can transfer short chain acyl groups to the cytosol as described above.
Conclusion-The increases in the beta cell concentrations of several short chain acyl-CoAs by insulin secretagogues suggest a scheme in which multiple short chain acyl-CoAs might perform important functions in insulin secretion. In this study we found that rat and human pancreatic islets and INS-1 832/13 cells possess enzymes for at least two redundant pathways for the translocation of short chain acyl carbon from the mitochondria to cytosol. One involves acetoacetate and the other citrate. In this respect, the beta cell may resemble the liver where mitochondrially derived carbon used for lipogenesis is transferred to the cytosol as both acetoacetate and citrate (43). The use of shRNA to knock down the cytosolic enzymes ATP citrate lyase, which does not decrease insulin release, and acetoacetyl-CoA synthetase, which does cause a concomitant decrease in insulin release, supports the idea that acetoacetate may be very important as a carrier of short chain acyl groups from the mitochondria to the cytosol in the beta cell. Our multiple short chain acyl-CoA mechanism additionally hypothesizes that short chain acyl-CoA carbon is derived from insulin secretagogue carbon via anaplerosis and that malonyl-CoA might be more important in supplying its carbon for lipid synthesis 5 than its being an inhibitor of the transport of long chain fatty acids into mitochondria. Possible roles of short chain CoAs are protein acylation (44,45), including protein isoprenylation (46), and lipid synthesis (47)(48)(49) that might modify the lipid composition of intracellular membranes that could influence the trafficking of insulin secretory granules (2,50).