Transamination of 3-Phenylpyruvate in Pancreatic B-cell Mitochondria*

High aminotransferase activities catalyzing the reaction between L-glutamate and the aromatic ketomon- ocarboxylic acid, 3-phenylpyruvate, were observed in the mitochondria from pancreatic B-cells. At very low concentrations of 3-phenylpyruvate, L-glutamine was an effective amino group donor. The aminotransferase activities for the aliphatic ketomonocarboxylic acids, pyruvate and 2-ketoisovalerate, were lower in B-cell mitochondria. High rates of transamination of 2-keto-isocaproate with L-glutamine were observed and may be an important prerequisite for the insulin secretory potency of this 2-keto acid. Since B-cell mitochondria are well supplied with L-glutamine and L-glutamate, 3-phenylpyruvate-in- duced 2-ketoglutarate production may explain the insulin secretory potency of 3-phenylpyruvate which is not a fuel for pancreatic islet cells.


EXPERIMENTAL PROCEDURES
Animals and Methods-24-h fasted ob/ob mice (40-60 g, body weight) were used for the experiments. Pancreatic islets were isolated from the pancreas following collagenase digestion (10). Islets were selected under a stereomicroscope and maintained in Krebs-Ringer bicarbonate medium containing 10 mM Hepes,' 3 mM glucose, and 0.1% bovine albumin. Pancreatic islets were homogenized in ice-cold Tris/sucrose buffer (10 mM Tris, 250 mM sucrose, pH 7.4) using a Potter-Elvehjem homogenizer with a Teflon pestle and maintained on ice. Sediments obtained after centrifugation for 10 min at 100 X g and another 10 min at 500 X g at 4 "C were discarded. The mitochondrial pellet was obtained by centrifugation of the supernatant for 10 min at 8,700 X g at 4 "C and resuspended in Tris/sucrose buffer. The cytoplasmic fraction was obtained by centrifugation of the supernatant for another 20 min at 100,000 x g. Protein was determir,ed according to McKnight (11). Monoamine oxidase was determined using a recently described sensitive modification (12) of the method of Wurtman and Axelrod (13) with phenylethylamine as the substrate. Activities of this mitochondrial enzyme marker, expressed as nmol of phenylethylamine/mg of protein/20 min, were 19.8 ? 2.2 in total tissue homogenates, 43.8 ? 6.0 in the mitochondrial fractions, and 2.4 zk 0.3 in the cytosolic fractions. Rates of transamination were determined in these fractions. Oxidation of 2-ket0[U-'~C]glutarate by isolated pancreatic islets was performed essentially as described for other keto acids elsewhere (14).
Determination of Transamination-Transamination rates were measured as described previously (15) using modifications of the methods of Ichihara and Koyama (16), Cooper and Meister (17,18), and Cooper and Gross (19). Samples (20 pl, 5-10 pg of protein) of the tissue homogenates, the mitochondrial fractions, or the cytoplasmic fractions from pancreatic islets were incubated for 30 min at 37 "C in 40 pl of Tris/sucrose buffer containing varying concentrations of L-[U-"Clglutamate (0.5-42.0 Ci/mol) or ~-[U-"C]glutamine (0.4-40.0 Ci/mol) and supplemented with 0.1 mM pyridoxal phosphate and 5 mM of the keto acid (2-ketoisovalerate, 2-ketoisocaproate, 3-phenylpyruvate, or pyruvate). ~-[U-"C]Glutamate was dried at 56 "C with a constant stream of N2, while ~-[U-"C]glutarnine was dried at 37 "C only. This was necessary to avoid generation of high blank values in the presence of 3-phenylpyruvate due to nonenzymatic conversion of labeled glutamine to [U-14C]2-pyrrolidone-5-carboxylate (19). In control experiments, it was confirmed that transamination rates in 0.1 M phosphate buffer instead of Tris/sucrose buffer yielded similar rates of transamination. Transamination rates in pancreatic islet tissue were not dependent on pH in the range between 7.0 and 8.6 and were linear for incubation times up to 60 min. Freezing and thawing the samples three times in Tris buffer (10 mM, pH 7.4) devoid of sucrose did not affect the transamination rates in tissue homogenates, mitochondrial fractions, or cytoplasmic fractions, indicating that transport of 2-ketoisovalerate, 2-ketoisocaproate, 3phenylpyruvate, pyruvate, glutamate, or glutamine was not ratelimiting. Blank values were obtained by incubating the medium without homogenate or fractions. At the end of the incubation period,

The ability of various keto acids to act as a partner in the transamination of ~-[U-'~C]glutamate to 2-keto[U-l4C]g1utarate and of ~-[U-'~C]glutamine to 2-ket0[U-'~C]glutaramate
and/or 2-keto[U-14C]glutarate was examined in tissue homogenates, mitochondrial fractions, and cytoplasmic fractions of pancreatic islets from ob/ob mice. The rates of transamination are expressed as nmol of 2-keto acid/mg of protein/ 30 min. In Table I When ~-[U-'~C]glutarnate was the transamination partner, the rates of transamination of 2-ketoisovalerate and 2-ketoisocaproate in pancreatic islet homogenates were very low compared to the high rates observed with 3-phenylpyruvate and pyruvate; rates obtained with pyruvate were three times higher than rates obtained with 3-phenylpyruvate (Table I).
In mitochondrial and cytosolic fractions of pancreatic islets, rates of transamination of 2-ketoisovalerate and 2-ketoisocaproate were also low ( Table 1). The rates of transamination of 3-phenylpyruvate and pyruvate were significantly higher (Table I). However, in contrast to total homogenates of pancreatic islets, the rates of transamination of 3-phenylpyruvate were 2.5 times higher than those of pyruvate in mitochondrial fractions (Table I). On the other hand, in cytoplasmic fractions, the rates of transamination of pyruvate were 5.5 times higher than those of 3-phenylpyruvate (Table I). Only in the presence of 3-phenylpyruvate were the rates of transamination higher in the mitochondrial fractions than in the cytosolic fractions of pancreatic islets (Table I). When ~-[U-"C]glutamine was the transamination partner, the rates of transamination of 2-ketoisocaproate, 3-phenylpyruvate, and PYNvate were similar both in homogenates and in cytoplasmic fractions of pancreatic islets from ob/ob mice, while rates of transamination of 2-ketoisovalerate were significantly lower (Table I). In mitochondrial fractions, rates of transamination of ~-[U-'~C]glutamine were significantly higher in the presence of 2-ketoisocaproate or 3-phenylpyruvate than in the presence of 2-ketoisovalerate or pyruvate (Table I) (Table I).
TO further elucidate the importance of transamination of 3-phenylpyruvate for the initiation of insulin secretion, concentration dependencies were studied. Both the concentra-  (Fig. 1). Thus, in contrast to transamination with pyruvate, rates of transamination of 3-phenylpyruvate in pancreatic islet homogenates and mitochondrial fractions were higher in the presence of L-[U-14C]glutamine than in the presence of ~-[U-'~C]glutamate when amino acid concentrations were 5 mM or higher (Fig.   1).
Rates of transamination of L-[U-'4C]glutamate (1 mM) steadily increased with increasing concentrations of pyruvate or of 3-phenylpyruvate in homogenates, mitochondrial fractions, and cytoplasmic fractions from pancreatic islets from oblob mice, but with one noticeable exception (Fig. 2). At concentrations of 3-phenylpyruvate above 10 mM, rates of transamination of ~-[U-'~C]glutamate (1 mM) reached a plateau in the mitochondrial fraction (Fig. 2). Maximal rates of transamination of ~-[U-'~C]glutarnine were lower than those of ~-[U-"C]glutamate (Fig. 2). Maximal rates of L-[U-'~C] glutamine transamination were obtained with micromolar concentrations of both pyruvate and 3-phenylpyruvate in homogenates as well as in mitochondrial and cytoplasmic fractions from pancreatic islets from ob/ob mice (Fig. 2). This   Values shown are the means -C SE. for 4-6 experiments.
was especially striking in the mitochondrial fractions, resulting in considerably lower transamination rates of L-[U-'~C] glutamine than of L-[U-'4C]glutamate with millimolar concentrations of keto acids (Fig. 2).
In a control experiment, the rate of oxidation of 2-keto[U-14C]glutarate (5 mM) by isolated incubated ob/ob mouse pancreatic islets was found to be as low as 0.39 & 0.27 mmol/kg dry weight (n = 8).

DISCUSSION
The results provide support for the existence of a high aminotransferase activity catalyzing the reaction between Lglutamate and the aromatic ketomonocarboxylic acid, 3-phenylpyruvate, in the mitochondria from pancreatic islets ( Table  I). The aminotransferase activity for the ketomonocarboxylic acids, pyruvate, 2-ketoisovalerate, and 2-ketoisocaproate was significantly lower in islet cell mitochondria in the presence of L-glutamate (Table I). Therefore, transamination with glutamate cannot be the sole cause of the strong insulinreleasing capacity of 2-ketoisocaproate (1,2,14,15,20). Our mitochondrial fraction also contained insulin secretory granules. But it is very unlikely that these organelles are sites of transamination. We are not aware of reports describing secretory granules as sites of transamination. Since we used ob/ob mouse pancreatic islets in the present study, which contain more than 90% B-cells (21), the results are representative for B-cell mitochondria. In total homogenates of pancreatic islets, we observed, in accordance with results from Malaisse et al. (7), higher rates of transamination of pyruvate which apparently originate from cytosolic transamination ( Table I) present observation of a high rate of transamination between L-glutamate and 3-phenylpyruvate in mitochondria from pancreatic islets is in accordance with the description of an intramitochondrial phenylalanine aminotransferase with high capacity in pig brain and heart (22). An aminotransferase catalyzing the reaction between L-glutamate and 3-phenylpyruvate has been extracted from mitochondria (23) and there is evidence for the existence of a specific aromatic amino acid aminotransferase (23).
There are apparently also other aminotransferases in the mitochondria which can form 2-ketoglutarate in the presence of 3-phenylpyruvate (24,25). Glutamine is transaminated (19,24,25) yielding 2-ketoglutaramate (26) followed by an wdeamidation reaction in which 2-ketoglutaramate is converted to 2-ketoglutarate (25). In our investigation, we found transamination rates of glutamine in the presence of millimolar concentrations of 3-phenylpyruvate (the optimal range for initiation of insulin secretion by this insulin secretagogue (3)), which were significantly lower in homogenates, mitochondrial fractions, and cytoplasmic fractions from pancreatic islets from ob/ob mice when compared to glutamate transamination (Table I). Maximal rates of pancreatic islet mitochondrial glutamine transamination in the micromolar concentration range of 3-phenylpyruvate (Fig. 2) are in accordance with very low apparent K, values for 3-phenylpyruvate as reported for rat kidney, brain, and liver mitochondrial glutamine transaminase (19,24). Our results support the recent conclusion by Cooper and Meister (24) that apparent K,,, values for 3-phenylpyruvate exhibited by the glutamine transaminase K are several orders of magnitude lower than those reported for other 3-phenylpyruvate-utilizing transaminases. Thus, at 5 mM 3-phenylpyruvate, which is a concentration in the optimal range for initiation of insulin secretion, mitochondrial transamination of L-glutamate seems to be more important for initiation of insulin secretion by 3-phenylpyruvate than transamination of L-glutamine. On the other hand, in vivo where 3-phenylpyruvate plasma concentrations are normally around 5 p M and up to 100-200 pM in phenylketonuria (27), a possible potentiating effect of 3-phenylpyruvate on glucose-induced insulin secretion should be mainly dependent on L-glutamine transamination. With 2-ketoisocaproate, the rate of transamination in the mitochondria from pancreatic B-cells was higher with L-glutamine than with Lglutamate (Table I). Glutamine aminotransferase activity was very low with 2-ketoisovalerate as substrate in pancreatic islet homogenates and subcellular fractions (Table I) which is in agreement with findings in liver tissue reported by Meister (26). Therefore, the high rates of L-glutamine transamination in pancreatic B-cell mitochondria in the presence of %ketoisocaproate as compared to virtually negligible rates in the presence of 2-ketoisovalerate (Table I) may help to explain the ability of 2-ketoisocaproate in contrast to 2-ketoisovalerate to induce insulin secretion from pancreatic islets (1,2,14,20).
Since B-cell mitochondria are well supplied with L-glutamine and L-glutamate (7), 3-phenylpyruvate-induced 2-ketoglutarate production may explain the insulin secretory potency of this aromatic ketomonocarboxylic acid. Thus, the two fuel analogues, 3-phenylpyruvate and BCH, may induce insulin secretion by reactions started by enhanced intramitochondrial availability of %ketoglutarate. One such reaction may be active metabolism of 2-ketoglutarate through the segment of the Krebs' cycle leading to the formation of malate and thus providing more reducing equivalents as recently documented in HeLa cells (28) and in several other mammalian cells such as those from the small intestine (29). Exogenous 2-ketoglutarate administered to intact pancreatic B-cells apparently cannot replace %ketoglutarate produced intramitochondrially. This keto acid is excluded apparently not only from liver cells (30) but also from pancreatic islet cells as shown by the extremely low rates of oxidation of radioactively labeled 2-ketoglutarate (see "results").