This study presents evidence that the KIC-induced amplification of insulin secretion coincides with the rapid and sustained increase in supply of cytosolic acetyl-CoA. This result was obtained by measuring the islet content of acetyl-CoA after 5 min and 20 min incubations with KIC and omitting the time-consuming washing step thereafter. Measured this way, the acetyl-CoA content of islets was practically twice as high as measured earlier under the same experimental conditions (Panten et al. 2013, 2016). Irrespective of the washing step, normalized (control = 100%) increases in the islet content of acetyl-CoA induced by KIC or glucose within 20 min were pronounced in the presence of glipizide (Panten et al. 2016, Fig. 2A as compared with current Fig. 3B), but moderate for KIC or insignificant for glucose in the absence of glipizide (Panten et al. 2016, Fig. 2E as compared with current Fig. 3D). So, a typical earlier result was reproducible in spite of the omission and, at the same time, meaningful measurements of early changes could be performed.

The metabolic fate of the alpha-keto acids in beta-cell mitochondria is largely known, facilitating conclusions as to the subcellular localization of the changes of the acetyl-CoA content reported here. In mouse islets exposed to glipizide throughout the experiment and pretreated without exogenous fuel, the efficacy of 10 mmol/L alpha-ketoisovalerate (KIV) to amplify the secretion of insulin after 20 min was much lower than the one of 10 mmol/L KIC (see Fig. 2). The following observations explain the differences between KIV and KIC. First, pyruvate dehydrogenase is inhibited by the catabolites of both alpha-ketoacid anions, as was observed in hepatocytes (Walajtjs-Rode and Williamson, 1980). Second, KIV generates acetyl-CoA only via pyruvate dehydrogenase, whereas KIC generates acetyl-CoA also via its strong oxidation (Lenzen and Panten 1980, see also Fig. 1). Therefore, the similar increases in the islet content of acetyl-CoA produced by KIV and KIC after 20 min (Panten et al. 2016) suggest that the mitochondrial acetyl-CoA content is of minor relevance for the observed changes. Furthermore, in mouse islets, mitochondria account only for ca. 4% of the beta-cell volume, whereas the cytosol accounts for ca. 50% (Dean 1973). So, for changes of the mitochondrial acetyl-CoA to significantly affect the measurements of islet contents, they have to be an order of magnitude higher than the cytosolic changes.

Acetyl-CoA is also located in the peroxisomes and the nucleus (Pehar and Puglielli 2013). The peroxisomes are unlikely to take up acetyl-CoA from the cytosol (Antonenkov and Hiltunen 2006) and the nucleus, which makes up to 12% of the mouse beta-cell volume (Dean 1973), forms a common compartment with the cytosol for acetyl-CoA. We previously assumed that increases in the islet acetyl-CoA content after incubation for 20 min reflected acetyl-CoA uptake into the Golgi/ER and not the cytosol (Panten et al. 2016). The following considerations argue against our earlier hypothesis. Uptake of acetyl-CoA into the Golgi/ER from the cytosol is driven by the concentration gradient between the cytosol and the Golgi/ER, the volume of which is in beta-cells about eightfold smaller than that of the cytosol (Dean 1973). If the increase in the total acetyl-CoA content after 20 min reflected acetyl-CoA taken up into the Golgi/ER, increase in consumption of cytosolic acetyl-CoA by production of metabolites (see Fig. 1) should have eliminated the KIC-induced rise in the cytosolic acetyl-CoA concentration nearly completely at min 5, since in islets exposed to glipizide KIC had not decreased the total acetyl-CoA content after 5 min (Fig. 3A, B). But the consumption of cytosolic acetyl-CoA starts later than the strong and continuous supply by KIC (10 mmol/L). Moreover, limited supply of cytosolic NADPH during amplification by KIC (see “Introduction”) is expected to slow down the acetyl-CoA consuming synthesis of non-acetyl-CoA thioesters. Therefore, the period of consumption was too short for sufficient elimination of acetyl-CoA within 5 min. Hence, the increases in acetyl-CoA content by KIC mainly reflected increases in the cytosolic acetyl-CoA.

In islets exposed to glipizide, 5 min exposure to 10 mmol/L KIC significantly decreased the acetyl-CoA plus CoA-SH content (Fig. 4A), suggesting that CoA-SH was consumed by increase in thioester production. CoA-SH can be trapped by a rapid rise in the cytosolic production of 3-hydroxy-3-methylglutaryl-CoA (HMG-CoA), malonyl-CoA, and fatty acyl-CoA, synthesized by fuel-induced supply of cytosolic acetyl-CoA (MacDonald et al. 2005, Prentki et al. 2013, see also Fig. 1). After 20 min exposure to KIC, the acetyl-CoA plus CoA-SH content was no longer decreased (Fig. 4B), presumably due to more rapid consumption of non-acetyl-CoA thioesters than at minute 5. The outcome of the acetyl-CoA plus CoA-SH measurements was not changed by the increases in the acetyl-CoA content (Fig. 3A, B), since these increases involved trapping of corresponding amounts of CoA-SH via citrate lyase and acetoacetyl-CoA synthetase (see Fig. 1). The evidence that KIC promoted non-acetyl-CoA thioester synthesis renders our earlier assumption unlikely that the increases in acetyl-CoA content reflected reduced turnover of acetyl-CoA pools (Panten et al. 2016). However, coincidence with increases in acetyl-CoA content (Fig. 3A, B) indicated that the effects on acetyl-CoA plus CoA-SH content resulted from KIC-induced increase in supply of cytosolic acetyl-CoA.

In islets exposed to glipizide and preincubated without exogenous fuel, the pyruvate carboxylase was inhibited and glucose was unable to amplify the secretion (Panten et al. 2013). Under this condition however, where weak supply of cytosolic acetyl-CoA by glucose is expected, glucose elevated the acetyl-CoA content as intensively as KIC (Fig. 3A, B). This suggests that glucose enabled only slow and weak acetyl-CoA-consuming syntheses of non-acetyl-CoA thioesters.

In islets preincubated without exogenous fuel, the acetyl-CoA content was less elevated after 20 min than after 5 min, both with KIC and with glucose as stimuli (Fig. 3C, D). This may indicate that the supply of cytosolic acetyl-CoA rose more rapidly than the acetyl-CoA consumption by synthesis of non-acetyl-CoA thioesters and uptake into the Golgi/ER. This view is supported by the failure of KIC to cause a significant decrease in acetyl-CoA plus CoA-SH content under this condition (Fig. 4C, D).

In islets pretreated with 3 mmol/L glucose, raising the glucose concentration from 3 to 15 mmol/L failed to induce a significant elevation of the islet acetyl-CoA content (Fig. 3E, F), but coincided with considerable insulin release (Fig. 5). Presumably increase in synthesis of non-acetyl-CoA thioesters and uptake into the Golgi/ER nearly completely consumed the cytosolic acetyl-CoA supplied by glucose. The increase in acetyl-CoA content by raising the glucose concentration from 3 to 30 mmol/L (Fig. 3E, F) may then reflect a more rapidly increasing supply than consumption of cytosolic acetyl-CoA.

The increase in acetyl-CoA content by 30 mmol/L glucose differs from the observation in an earlier study, where the acetyl-CoA content of perifused rat islets pretreated for 30 min with 2.5 mmol/L glucose was not increased after 3 min with 25 mmol/L glucose and was even decreased after 30 min (Liang and Matschinsky 1991). Also, the first-phase insulin secretion in this study was more pronounced than that induced by 30 mmol/L glucose in the present study (see Fig. 5). A straightforward explanation for these phenomena would be that the rate of cytosolic acetyl-CoA consumption by synthesis of non-acetyl-CoA thioesters was higher in rat islets than in mouse islets. The inability of glucose stimulation to change the islet content of acetyl-CoA plus CoA-SH (Fig. 4E) and of CoA-SH (Liang and Matschinsky 1991) does not rule out trapping of CoA-SH, since CoA-SH can be generated from acyl-CoA during glucose-stimulated triacylglycerol synthesis on the ER (Prentki et al. 2013; Lorenz et al. 2013).

Collectively, the previous and the present findings indicate that the islet content of acetyl-CoA did not always correlate with the glucose-induced amplification of insulin secretion, but the findings are consistent with amplification by increased supply of cytosolic acetyl-CoA.

As strengthened by the present findings, the supply of cytosolic acetyl-CoA is well suited to provide signals, which amplify the insulin secretion. The following considerations narrow down these signals: (1) The failure of 15 mmol/L glucose to induce significant increase in the islet content of acetyl-CoA (Fig. 3E, F) despite stimulation of insulin release (Fig. 5) argues against amplification by direct action of acetyl-CoA (without molecular transformation). (2) Protein isoprenylation and cholesterol production are not acute regulatory events (Metz et al., 1993, Zúñiga-Hertz et al. 2015) and stimulation of malonyl-CoA and fatty acyl-CoA synthesis is not always required for amplification (Chakravarthy et al. 2007; Cantley et al. 2019). (3) The numerous lysine-acetylated proteins in the islet cytosol (Zhang et al. 2019) indicate that acetyl-CoA serves as substrate for cytosolic protein acetylation. The rate of acetylation appears to be determined by the concentration of acetyl-CoA relative to the concentration of CoA-SH in the immediate vicinity of the lysine acetyltransferases, since CoA-SH is known to exert a product inhibition on most of the lysine acetyltransferases (Albaugh et al. 2011; Chaudhary et al. 2014; Drazic et al. 2016). 4. The wash-out of fuel secretagogues at stimulatory concentrations abolished the amplification of insulin secretion within 14 min (Panten et al. 2016). This reversibility of the metabolic amplification fits to the rapidly reversible protein acetylation by the action of lysine deacetylases.

In conclusion, the present observations indicate that glucose as well as KIC increase the supply of acetyl-CoA in the beta-cell cytosol during both phases of insulin secretion. This situation enables intensified protein acetylation, whereby the supply of cytosolic acetyl-CoA may function as a critical signal in the pathway of metabolic amplification. Identification of those cytosolic proteins which become acetylated during stimulated insulin secretion will give insight into the mechanisms which specifically promote the amplification of secretion and are not merely permissive or supportive. Clarification of these issues is not only relevant for the physiology of beta-cell function but also of major importance for the pathophysiology of type 2 diabetes (Grespan et al. 2018).