Glucose-stimulated insulin secretion in infected and non-infected INS-1E cells

Non-infected as well as AdRIPGFP and AdshRNAcAAT3-infected INS-1E cells were exposed to either 2.5mM or 16.7mM glucose concentrations. At the end of the test incubation insulin release was determined. The data represented in figure 3.17 were obtained from the same experiments performed for α-ketoglutarate and glutamate measurements. In non-infected cells, insulin secretion augmented from 0.45±0.03µg/mg protein to 1.0±0.06µg/mg protein (Δ increase 0.54±0.06µg/mg protein). A similar increase was observed in AdRIPGFP-infected INS-1E cells. Raising glucose concentrations from 2.5mM to 16.7mM caused an increase of insulin secretion from 0.40±0.03 to 0.94±0.1µg/mg protein) (Δ increase 0.54±0.11µg/mg protein). In the presence of AdshRNAcAAT3 glucose-stimulated insulin secretion was significantly (p<0.01) decreased (0.47±0.08µg/mg protein and 0.73±0.08µg/mg protein at 2.5mM and 16.7mM glucose respectively). In absolute terms, insulin secretion increased by only 0.26±0.14µg/mg

protein compared to 0.54µg/mg protein in non-infected and AdRIPGFP-infected cells (p<0.05). Thus, cAAT repression inhibits glucose-stimulated insulin secretion.

Figure 3.17: The impact of _cAAT repression on glucose-stimulated insulin secretion. (A) Insulin secretion was determined after a static incubation of non-infected (-) and infected (ctrl & cAAT) INS-1E cells for 30min at 37°C at the indicated glucose concentrations.

Insulin is expressed as µg/mg protein. (B) The delta increase in insulin amounts [µg/mg protein] after glucose stimulation was determined in non-infected (-) and infected INS-1E cells. Four independent experiments were performed (n=4); *p<0.05 two-tailed student’s t test for paired data.

CHAPTER 4 – DISCUSSION & PERSPECTIVES

The level of glucose in the blood is monitored by pancreatic β- and α-cells which respectively secrete insulin and glucagon depending on the nutritional state. Failure to maintain blood glucose in the normal range leads to conditions of persistently high (hyperglycemia) or low (hypoglycemia) blood sugar. The tight regulation of hormone secretion is not only controlled by glucose but also implicates other factors. One of these additive factors that has been proposed is the amino acid glutamate. However, conflicting data concerning glutamate synthesis and function have been suggested. It is still debated which pathway leads to glucose-dependent glutamate formation and if glutamate exerts either intracellular or extracellular messenger actions. Three main questions were investigated in this thesis work:

1) To study glucose-stimulated glutamate generation pathways 2) To investigate glutamate release mechanisms

3) To clarify glutamate signalling function (autocrine/paracrine or intracellular messenger).

Mitochondrial metabolism plays a central role in the pancreatic β-cell by generating key metabolites that couple glucose oxidation to insulin secretion. Several studies have demonstrated that glutamate is formed from glucose [5, 57, 129, 130, 135, 241, 244, 269]. Nuclear magnetic resonance (NMR) spectroscopy revealed that glutamate is the major glucose metabolic product. Insulinotropic action of glutamate was shown in permeabilized β-cells at permissive [Ca²⁺]i independently of mitochondrial activation.

Furthermore, GSIS was potentiated at intermediate glucose concentrations by the cell-permeant glutamate analogue dimethyl-glutamate [57, 245] whereas it was impaired when glucose-stimulated glutamate generation was repressed in GAD-overexpressing β -cells, conditions under which glutamate is preferentially converted to GABA [241].

Therefore, glucose-derived glutamate has been proposed to participate in the amplifying pathway of GSIS [57, 258].

The results presented in project 3.1 (“Mechanism of glucose-mediated glutamate production in α- and β-cells: Impact on insulin but not glucagon secretion”; submitted article) indicate that glutamate is produced from glucose not only in pancreatic rat islets but also in isolated primary α- and β-cells. Two enzymes have been proposed to play a key role in fuel-mediated glutamate production. Several studies link glutamate dehydrogenase (GDH) activity to insulin secretion and glutamate biosynthesis [57, 246, 258]. In mitochondria, GDH catalyses the reversible reaction αKG + NH3 + NAD(P)H ↔ glutamate + NAD(P)⁺ [296]. GDH flux can be either cataplerotic forming glutamate

through reductive transamination of αKG or anaplerotic generating αKG through oxidative deamination of glutamate. Evidence suggests the implication of GDH in metabolism secretion coupling. It has been demonstrated that GDH overexpression in clonal β-cells and rat but not mouse islets increased GSIS [246], while gene suppression caused impaired insulin release [246, 247]. In agreement with the hypothesis is the finding that the hyperinsulinemia/hyperammonemia (HI/HA) syndrome is caused by loss of GTP regulation of GDH [137, 297]. Although GDH seems to be implicated in the control of insulin secretion its implication in glucose-stimulated glutamate generation is debated [135, 242, 244, 265]. The high Michaelis constant (Km) of GDH for ammonium seems to prohibit glutamate formation under normal conditions in most organisms [298]. The mitochondrial enzyme is allosterically regulated by several ligands [121, 299-305].

Potent inhibitors of the reaction are ATP and GTP which are increased during glucose stimulation. GDH activators include ADP, leucine and the non-hydrolysable form of leucine BCH. A GDH-activating gain of function mutation (in the GTP binding site) was demonstrated to increase the flux through GDH in direction of oxidative deamination of glutamate to αKG in transgenic mice [135]. In normal mouse islets, GSIS is associated with a suppression of the flux through GDH, consistent with the allosteric inhibitory effects of ATP and GTP on the activity of the enzyme [135]. Furthermore, loss of the mitochondrial enzyme SIRT4 that uses NAD to ADP-ribosylate and inhibit GDH activity caused a slight increase in GSIS [262]. In contrast, in the absence of glucose, when ADP concentrations are high, GDH catalyses the oxidative deamination of glutamate to αKG in order to supply the TCA cycle with intermediates, thereby raising the ATP/ADP ratio and triggering insulin release.

The paradox that GDH flux is suppressed during glucose stimulation while glutamate levels are increased makes this pathway unlikely. Therefore, other pathways might be more relevant in glucose-stimulated glutamate generation. An association between malate-aspartate shuttle activity and glutamate generation has been proposed [129, 269, 270]. The malate-aspartate shuttle is a biochemical system for translocating reducing equivalents (NADH) and their electrons produced during glycolysis across the impermeable inner membrane of the mitochondrion for oxidative phosphorylation. The first step is the formation of malate from oxalacetate and NADH catalysed by cytosolic malate dehydrogenase (cMDH). Malate is imported from the cytosol to the mitochondrial matrix in exchange for αKG through the malate-αKG carrier, also known as dicarboxylate carrier (DC). In the mitochondrial matrix malate is re-converted to oxalacetate by mitochondrial MDH (mMDH). During this reaction NAD⁺ is reduced to NADH which can enter the electron transport chain in order to generate ATP. To close the cycle, oxalacetate is transformed to aspartate by mitochondrial aspartate aminotransferase

glutamate exchanger, Aralar1 or Citrin) catalyses the exchange of aspartate for glutamate. Once in the cytosol, aspartate is deaminated to oxalacetate by cytosolic aspartate aminotransferase (cAAT) with the concomitant formation of generate glutamate from αKG. The NAD⁺ in the cytosol can then be reduced again by another round of glycolysis.

Figure 4.1: Malate-aspartate shuttle system. Reactions and transport systems that are implicated in aspartate shuttle activity are depicted in the illustration. 1, malate-α-ketoglutarate exchanger (dicarboxylate carrier); 2, glutamate-aspartate exchanger (Aralar1 or citrin); _c/mAAT, cytoplasmic/mitochondrial aspartate aminotransferase;

c/mMDH, cytoplasmic/mitochondrial malate dehydrogenase; αKG, α-ketoglutarate; OAA, oxalacetate.

The importance of the malate-aspartate shuttle system was tested, using either the blocker of the malate-αKG exchanger phenylsuccinate [306] or the inhibitor of aminotransferases aminooxyacetate [307-309]. AOA was demonstrated to suppress GSIS completely in mouse islets lacking glycerol-phosphate shuttle activity [271]. A dose-dependent partial inhibition of nutrient-stimulated insulin release was also observed in rat pancreatic islets [117, 310]. Furthermore, AOA was shown to attenuate glutamate production from glucose in clonal β-cells [129]. These observations coincide with the results presented in this thesis work. In the presence of elevated glucose concentrations both phenylsuccinate and AOA decreased glutamate release and intracellular glutamate production from isolated α- and β-cells. Furthermore, GSIS in β-cells but not glucagon secretion from α-cells was reduced in the presence of either inhibitor. Interestingly, a decrease in ¹³C labelling of aspartate duringglucose stimulation was detected in different β-cell lines and isolated pig islets [269]. These authors suggest that aspartate is the main

anaplerotic substrate and that GSIS is impaired if aspartate metabolism is inhibited [269]. They observed decreased aspartate levels associated with GSIS. A similar decrease in aspartate in response to glucose was also observed in mouse islets [130].

However, the latter investigators did not observe any change in glutamate and αKG concentrations. A reduction of aspartate levels was also detected in isolated β-cells but not in pancreatic islets where intracellular aspartate level increased (Supplemental table 2 submitted article). A possible explanation for the difference to published data could be that the incubation time in the study by Li et al was much longer (2h versus 30min) and islet incubations were carried out in the presence of an amino acid mixture [130].

In summary, the experiments using the inhibitors phenylsuccinate and AOA to block malate-aspartate shuttle activity suggest that flux of αKG through cAAT is important in glucose-stimulated glutamate production as well as in GSIS. However, AOA-mediated inhibition of cytoplasmic aspartate aminotransferase activity is controversial because the compound acts on all enzymes that use pyrridoxal as co-factor, thereby inhibiting transamination reactions in general [308, 309]. Other cellular processes might be affected by AOA. To investigate the specific role of cAAT in glutamate biosynthesis and GSIS, the activity of the enzyme was suppressed using shRNA technology (project 3.2).

cAAT mRNA and protein expression were successfully downregulated in INS-1E cells. The impact of _cAAT downregulation on both glucose-stimulated metabolite generation (αKG and glutamate) as well as on GSIS was tested in the clonal β-cells. In the absence of the adenovirus, glucose stimulation caused a ~9-fold increase in intracellular αKG concentration. This is in agreement with published data where a similar increase as well as similar absolute αKG amounts were detected [92]. In contrast, glucose-stimulated glutamate production was less prominent in INS-1E cells (~30% increase; Figure), similar to the results in isolated β-cells (project 3.1; submitted article Figure 2).

Remarkably, _cAAT repression caused an increase in intracellular αKG levels at both basal and stimulatory glucose conditions while absolute glutamate amounts formed during glucose stimulation decreased. These data strongly suggest an implication of cAAT in glutamate production at the expense of αKG. However, it should be noted that absolute αKG amounts produced during glucose stimulation are lower than the corresponding glutamate values (ctrl virus: Δ increase (αKG) = 2.3±0.5nnmol/mg protein; Δ increase (glu-) = 18.4±3.1nmol/mg protein). Therefore, other pathways might be implicated in stimulated glutamate generation. A tentative explanation is that glucose-stimulated GABA catabolism is implicated in glutamate generation. GABA catabolism, also referred to as the GABA shunt, occurs in the mitochondria. GABA-transaminase catalyzes the deamination of GABA to succinate semialdehyde (SSA) which is subsequently

stimulate TCA activity, ATP production and insulin exocytosis. During GABA-transaminase reaction glutamate is formed from its precursor αKG.

Figure 4.2: GABA shunt activity during glucose stimulation. Glucose stimulation increases GABA-transaminase activity. During transamination reaction succinate semialdehyde (SSA) is formed from GABA while glutamate (Glu^-) is synthesized from its precursor α -ketoglutarate (αKG). Subsequently, SSA is converted to succinate by succinate semialdehyde dehydrogenase (SSADH) to stimulate TCA activity, ATP production and insulin release. GAD, glutamate decarboxylase.

Glucose stimulation has been shown to enhance GABA transaminase activity, thereby causing a reduction in GABA amounts [135, 257]. Our findings confirm a glucose-mediated suppression in GABA production. We observed a decrease in intracellular GABA level in pancreatic rat islets and purified β-cells as well as GABA release from isolated β -cells (Supplemental table 2, project 3.1, submitted article). Thus, the activation of GABA shunt activity might be responsible for the differences in glutamate and αKG generation during glucose stimulation.

The phosphate-activated enzyme glutaminase which is localized to the outer side of the inner mitochondrial membrane seems to be less important in glucose-stimulated glutamate generation. In our preparation, intracellular glutamine amounts were strikingly higher in isolated β-cells compared to pancreatic rat islets. However, glutamine accumulation did not cause an increase in glutamate levels in isolated β-cells (project 3.1, supplemental table 2, submitted article). Total glutamate amounts are even lower than the corresponding values in pancreatic islets. The reason for the accumulation of glutamine remains unexplained. It can be speculated that the high levels of glutamine contribute to the increased basal insulin secretion rates in the purified cell population. A messenger role of glutamine in insulin secretion has been proposed [130, 131]. It has been reported that glutamine increased basal intracellular calcium level [131].

In summary, the present study not only clearly shows the importance of cAAT and malate-aspartate shuttle activity in glucose-stimulated metabolite generation but also demonstrates that cAAT expression is crucial for GSIS. Both GSIS as well as absolute insulin amounts released during glucose stimulation were decreased when cAAT expression was suppressed. Further studies have to be performed in rat pancreatic islets to confirm the importance of the enzyme in primary β-cells.

The mechanism of glutamate release is still debated [57, 311-315]. It has been proposed that glutamate is produced from the TCA cycle intermediate αKG during glucose stimulation by the mitochondrial enzyme GDH and transported into the cytosol where it is taken up by insulin-containing secretory granules to potentiate GSIS [57]. Glutamate was also suggested to be co-secreted with glucagon to exert auto- or paracrine actions [311, 313-315]. Vesicular glutamate transporters (VGLUT) expression was mainly detected in clonal and native α-cells [238, 311, 313-315]. Therefore, it has been suggested that glutamate accumulates in glucagon-containing secretory granules by VGLUTs at the expense of an electrochemical proton gradient across the membrane which is established by the vacuolar H⁺-ATPase [314, 315]. In agreement with this hypothesis is the finding that inhibition of the vacuolar H⁺-ATPase by bafilomycin A1

cells [243]. Although two different studies reported glutamate and glucagon co-secretion from pancreatic α-cells at low glucose condition [311, 315], they do not agree on glutamate action. On the one hand, released glutamate was proposed to inhibit glucagon secretion. Two distinct pathways were suggested: (1) AMPA-receptor mediated secretion of GABA from β-cells followed by its binding to GABAA-receptors on α-cells (paracrine actions), and (2) autocrine signalling of glutamate through class III metabotropic glutamate receptors (mGluR) on α-cells [227, 315]. On the other hand, glutamate was assumed to function as a positive autocrine signal for human α-cells causing a potentiation of glucagon secretion [311]. In this model glutamate activates ionotropic glutamate receptors of the AMPA/kainate type on α-cells thereby generating a positive feedback for α-cell function and amplifying glucagon secretion [311]. However, this study could not detect any iGluR expression in β-cells whereas other reports did [150, 240].

Glutamate was also suggested to act as intercellular messenger on ionotropic glutamate receptors on δ-cells in order to stimulate somatostatin release from this cell type [150].

Although these reports claimed that vesicular glutamate uptake is restricted to α-cells, VGLUT expression has also been detected in clonal β-cells [232, 233]. Furthermore, application of the vesicular glutamate uptake inhibitor bafilomycin was shown to block glutamate-induced insulin exocytosis in permeabilized INS-1E cells [57]. Indirect evidence of glutamate uptake in insulin-containing secretory granules stems from the observation that methyl-glutamate caused slight alkanalization of insulin granules in intact islets [316]. It has also been proposed that glucose-stimulated glutamate synthesis promotes glutamate uptake into insulin-secretory granules in order to bind the metabotropic glutamate receptor 5 (mGluR5) [233]. Activation of this receptor type would then potentiate GSIS. However, the acidic lumen of secretory granules makes glutamate receptor binding and activation unlikely [317]. Furthermore, it has been demonstrated that glucose stimulation decreases pH of secretory granules [318].

Eventually, extracellular glutamate signaling might play an important role in the regulation of both insulin and glucagon secretion as the various receptor types are expressed in both pancreatic islet cell types [16, 18-20, 234-237, 239, 240]. Several studies showed that glutamate itself or the glutamate receptor agonists AMPA and kainate stimulate insulin secretion from perfused pancreas, isolated islets or clonal islet cells in the presence of stimulatory glucose concentrations [16-19]. This stimulatory effect was blocked by an AMPA but not NMDA receptor antagonist. In contrast, in the presence of low glucose concentrations, glutamate and the glutamate receptor agonist AMPA were shown to stimulate glucagon secretion from perfused pancreas [236]. The implication of metabotropic glutamate receptors (mGluR) in hormone secretion remains to be elucidated. It has been demonstrated that group I and II mGluR agonists increased GSIS whereas a group III agonist inhibited insulin release at high glucose concentration

[235]. In contrast, at low glucose condition, a group III agonist was shown to inhibit glucagon release [238]. This inhibitory effect was diminished in the presence of a group III mGluR antagonist.

In the present study, glutamate release was investigated in isolated rat pancreatic α- and β-cells separately. Simultaneously, hormone secretion was measured and compared to the corresponding glutamate values to elucidate glutamate release mechanisms and its possible function as extracellular signalling molecule. Several findings argue against a possible co-secretion of glutamate and hormones and extracellular glutamate function:

- Glucose-stimulated insulin exocytosis was not accompanied by an increase in glutamate release in pancreatic rat islets and isolated β-cells whereas intracellular glutamate production was stimulated. Other nutrients known to stimulate hormone secretion were ineffective on glutamate release. In the presence of the deamination product of leucine, α-ketoisocaproate (αKIC), there was even decreased glutamate release in both cell populations, suggesting that glutamate acts as amino-group donor in order to generate the secretagogue leucine.

- Stimulation of Ca²⁺-dependent exocytosis using the K_ATP-channel blocker tolbutamide triggered hormone but not glutamate release from isolated α- and β -cells.

- Exposure of α- and β-cells to the catecholamine epinephrine caused a decrease of glutamate release from both cell types whereas the compound had opposing effects on hormone secretion. Epinephrine is known to activate β-adrenergic receptors in α-cells, followed by a rise in cAMP level, PKA activation and an increase in intracellular Ca²⁺. Subsequently, Ca²⁺-dependent exocytosis is stimulated. In contrast, epinephrine caused an inhibition of insulin secretion through the activation of α2-adrenergic receptors and inhibition of adenylate cyclase as well as a distal step in exocytosis [37, 319-321]. In astrocytes, the catecholamine has been shown to stimulate glutamate uptake [322]. Therefore, epinephrine might affect glutamate transport in α- and β-cells resulting in a decrease in glutamate release.

- Glutamate is released in a high and continuous manner from pancreatic islets as well as from purified islet cells. Fractional glutamate release was 5-50fold higher than the corresponding insulin and glucagon values. A similar high glutamate release was also observed in clonal β-cells [312].

These findings argue against a co-release of glutamate and hormone through secretory granule exocytosis. Other release mechanisms seem to be more important. Excitatory

glutamate uptake through EAATs is known to play a major role in the maintenance of extracellular glutamate concentrations in neuronal tissue. However, glutamate can also be released via EAATs by reversal of uptake. The high affinity and Na⁺-dependent glutamate transport system is present in rat pancreatic islets [20]. Weaver and co-workers demonstrated transporter activity in the cell-rich islet mantle. Inhibition of EAATs using the glutamate transport blocker, L-trans-pyrrolidine-2, 4-dicarboxylic acid (trans-PDC) caused a decrease in glutamate release in pancreatic islets as well as in isolated α- and β-cells. However, intracellular glutamate concentrations did not change, suggesting that glutamate is further metabolized. Interestingly, GSIS was potentiated in the presence of the antagonist in pancreatic islets and isolated β-cells whereas glucagon release from purified α-cells was not affected by the inhibitor. Our results are in agreement with earlier presented data showing that in the absence of exogenous glutamate trans-PDC caused a potentiation of insulin secretion at intermediate glucose

glutamate uptake through EAATs is known to play a major role in the maintenance of extracellular glutamate concentrations in neuronal tissue. However, glutamate can also be released via EAATs by reversal of uptake. The high affinity and Na⁺-dependent glutamate transport system is present in rat pancreatic islets [20]. Weaver and co-workers demonstrated transporter activity in the cell-rich islet mantle. Inhibition of EAATs using the glutamate transport blocker, L-trans-pyrrolidine-2, 4-dicarboxylic acid (trans-PDC) caused a decrease in glutamate release in pancreatic islets as well as in isolated α- and β-cells. However, intracellular glutamate concentrations did not change, suggesting that glutamate is further metabolized. Interestingly, GSIS was potentiated in the presence of the antagonist in pancreatic islets and isolated β-cells whereas glucagon release from purified α-cells was not affected by the inhibitor. Our results are in agreement with earlier presented data showing that in the absence of exogenous glutamate trans-PDC caused a potentiation of insulin secretion at intermediate glucose