BetaSys - Systems biology of regulated exocytosis in pancreatic β-cells
II. Systems biology approach to β-cells
4. Mitochondria and metabolic signals in β-cells
Department of Cell Physiology and Metabolism, University of Geneva Medical Centre, rue Michel-Servet 1, CH-1211 Geneva 4, Switzerland
Tel: +41 22 379 55 54
Pancreatic beta-cells are able to sense glucose and other nutrient secretagogues to regulate insulin exocytosis, thereby maintaining glucose homeostasis. This systems biology of insulin secretion controls translation of metabolic signals into intracellular messengers recognized by the exocytotic machinery. Central to this metabolism-secretion coupling, mitochondria integrate and generate metabolic signals, connecting glucose recognition to insulin exocytosis. In response to a glucose rise, nucleotides and metabolites are generated by mitochondria and participate, together with cytosolic calcium, in the stimulation of insulin release. This chapter describes the mitochondrion-dependent systems of regulated insulin secretion.
Key words: pancreatic beta-cell, insulin secretion, diabetes, mitochondria, amplifying pathway, glutamate, reactive oxygen species.
Word count (total) : 8011
Glucose homeostasis depends on the normal regulation of insulin secretion from the beta-cells and the action of insulin on its target tissues. Such equilibrated balance requires tight coupling between glucose metabolism and insulin secretory response. The exocytotic process is tightly controlled by signals generated by nutrient metabolism, as well as by neurotransmitters and circulating hormones. In a systems biology fashion, the beta-cell is poised to rapidly adapt the rate of insulin secretion to fluctuations in the blood glucose concentration. This chapter describes the molecular basis of metabolism-secretion coupling. In particular, we will see how mitochondria function both as sensors and generators of metabolic signals.
4.2. OVERVIEW OF METABOLISM SECRETION COUPLING
In the consensus model of glucose-stimulated insulin secretion (Fig. 1), glucose equilibrates across the plasma membrane and is phosphorylated by glucokinase, thereby initiating glycolysis (1). Subsequently, mitochondrial metabolism generates ATP, which promotes the closure of
+ATP-sensitive K channels (K-channel) and, as a consequence, depolarization of the plasma ATP
2+2+membrane (2). This leads to Ca influx through voltage-gated Ca channels and a rise in
2+cytosolic Ca concentrations, which triggers exocytosis of insulin (3).
Additional signals are necessary to sustain the secretion elicited by glucose. They participate in the amplifying pathway (4), formerly referred to as the K-channel independent stimulation of ATP
insulin secretion. Efficient coupling of glucose recognition to insulin secretion is ensured by the mitochondrion, an organelle that integrates and generates metabolic signals. This crucial role
2+goes far beyond the generation of ATP necessary for the elevation of cytosolic Ca (5). The
2+additional coupling factors amplifying the action of Ca (Fig. 1) will be discussed in this chapter.
4.3. MITOCHONDRIAL NADH SHUTTLES AS METABOLIC SENSORS
In the cytosolic compartment, glycolysis produces reducing equivalents in the form of NADH.
+Then, maintenance of glycolytic flux requires reoxidation of NADH to NAD. In most tissues,
lactate dehydrogenase ensures NADH oxidation to avoid inhibition of glycolysis secondary to
+the lack of NAD (Fig. 2). In beta-cells, which exhibit low lactate dehydrogenase activity (6), high rates of glycolysis are maintained through the activity of mitochondrial NADH shuttles,
thereby transferring glycolysis-derivedelectrons to mitochondria (7). Therefore, NADH shuttles
couple glycolysis to activationof mitochondrial energy metabolism, leading to insulin secretion.
The NADH shuttle system is composed essentially of the glycerolphosphate and the malate/aspartate shuttles (8), with its respective key members mitochondrial glycerol phosphate dehydrogenase and aspartate-glutamate carrier (AGC). Mice lacking mitochondrial glycerol phosphate dehydrogenase exhibit a normal phenotype (9), whereas general abrogation of AGC results in severe growth retardation, attributed to the observed impaired central nervous system function (10). Islets isolated from mitochondrial glycerol phosphate dehydrogenase knockout mice respond normally to glucose regarding metabolic parameters and insulin secretion (9). Additional inhibition of transaminases with aminooxyacetate, to non-specifically inhibit the malate/aspartate shuttle in these islets, strongly impairs the secretory response to glucose (9). The respective importance of these shuttles is indicated in islets of mice with abrogation of NADH shuttle activities, pointing to the malate/aspartate shuttle as essential for both mitochondrial metabolism and cytosolic redox state.
2+Aralar1 (or aspartate-glutamate carrier 1, AGC1) is a Ca sensitive member of the
malate/aspartate shuttle (11). Aralar1/AGC1 and citrin/AGC2 are members of the subfamily of 2+Ca-binding mitochondrial carriers and correspond to two isoforms of the mitochondrial
2+aspartate-glutamate carrier. These proteins are activated by Ca (12), acting on the external side
of the inner mitochondrial membrane (11; 13). Adenoviral-mediated overexpression of Aralar1/AGC1 increases glucose-induced mitochondrial activation and secretory response, both in insulinoma INS-1E cells and rat islets (14). This is accompanied by enhanced glucose oxidation and reduced lactate production. Recently, we conducted the mirror experiment by down regulating Aralar1/AGC1 in the same cell models (15). In INS-1E cells, Aralar1/AGC1 knockdown reduced glucose oxidation and the secretory response, although rat islets were not sensitive to such a maneuver (15). Taken as a whole, aspartate-glutamate carrier capacity appears to set a limit for NADH shuttle function and mitochondrial metabolism, exhibiting cell-specific dependence. The importance of the NADH shuttle system also illustrates the tight coupling between glucose catabolism and insulin secretion.
4.4. GETTING IN AND OUT OF THE TRICARBOXYLIC ACID CYCLE
In pancreatic beta-cells, high NADH shuttle activity favors transfer of the glycolysis-product pyruvate into mitochondria. Pyruvate import into the mitochondrial matrix is associated with a futile cycle that transiently depolarizes the mitochondrial membrane (16). After its entry into
mitochondria, pyruvate is converted to acetyl-CoA by pyruvate dehydrogenase or to oxaloacetate by pyruvate carboxylase (Fig. 2). The pyruvate carboxylase pathway ensures the provision of carbon skeleton (i.e. anaplerosis) to the tricarboxylic acid (TCA) cycle, a key pathway in beta-cells (17-20). Noteworthy, inhibition of the pyruvate carboxylase reduces glucose-stimulated insulin secretion in rat islets (21). The very high anaplerotic activity suggests important loss of TCA cycle intermediates (i.e. cataplerosis), compensated for by pyruvate carboxylation to
synthesize de novo oxaloacetate. In the control of glucose-stimulated insulin secretion, TCA cycle intermediates might serve as substrates leading to the formation of mitochondrion-derived coupling factors (5).
Importance of TCA cycle activation for beta-cell function is illustrated by stimulation with substrates bypassing glycolysis. This is the case for the TCA cycle intermediate succinate, or its cell permeant methyl-derivatives, that has been shown to efficiently promote insulin secretion in pancreatic islets (22; 23). Succinate induces hyperpolarization of the mitochondrial membrane,
2+2+resulting in elevation of mitochondrial Ca and ATP generation, while its catabolism is Ca-
The mitochondrion in general, and the TCA cycle in particular, is the key metabolic crossroad enabling fuel oxidation as well as provision of building blocks, or cataplerosis, for lipids and proteins (24). In beta-cells, approximately 50% of pyruvate is oxidised to acetyl-CoA by pyruvate dehydrogenase (18). Pyruvate dehydrogenase is an important site of regulation as,
2+among other effectors, the enzyme is activated by elevation of mitochondrial Ca (25; 26) and,
conversely, its activity is reduced upon exposures to either excess fatty acids (27) or chronic high glucose (28). Oxaloacetate condenses with acetyl-CoA forming citrate, which undergoes stepwise oxidation and decarboxylation yielding α-ketoglutarate. The TCA cycle is completed
via succinate, fumarate, and malate, in turn producing oxaloacetate (Fig. 2). The fate of α-
+ketoglutarate is influenced by the redox state of mitochondria. Low NADH to NAD ratio would
+favour further oxidative decarboxylation to succinyl-CoA as NAD is required as co-factor for
+this pathway. Conversely, high NADH to NAD ratio would promote NADH-dependent
reductive transamination forming glutamate, a spin-off product of the TCA cycle (24). The latter
+situation, i.e. high NADH generated at the expense of NAD, is a physiological consequence of
glucose stimulation in beta-cells (29; 30).
Although the TCA cycle also oxidises fatty acids and amino acids, carbohydrates are the most important fuel under physiological conditions for the beta-cell. Upon glucose exposure, mitochondrial NADH elevations reach a plateau after approximately 2 min (31). In order to maintain pyruvate input into the TCA cycle, this new redox steady state requires continues
+reoxidation of mitochondrial NADH to NAD, primarily by complex I of the electron transport
chain. However, as complex I activity is limited by the inherent thermodynamic constraints of proton gradient formation (32), excess NADH contributed by this high TCA cycle activity must be reoxidized by other dehydrogenases, i.e. through cataplerotic reactions. Indeed, significant
cataplerotic activity in beta-cells was suggested by the quantitative importance of anaplerotic pathways employing pyruvate carboxylase (17; 18), as confirmed by use of NMR spectroscopy (19; 20; 33).
4.5. MITOCHONDRIAL CONTROL OF THE GLUTAMATE DEHYDROGENASE
The enzyme glutamate dehydrogenase (GDH) is a key enzyme in the control of the secretory response (Fig. 2). GDH is a homohexamer located in the mitochondrial matrix and catalyses the
+reversible reaction α-ketoglutarate + NH + NADH ? glutamate + NAD; inhibited by GTP and 3
activated by ADP (34; 35). In the beta-cell, allosteric activation of GDH has received most of the attention over the last three decades (36). Numerous studies have used the GDH allosteric activator L-leucine or its non-metabolized analogue beta-2-aminobicyclo[2.2.1]heptane-2-carboxylic acid (BCH) to address the role of GDH in the control of insulin secretion (36-39). Alternatively, GDH activity can be increased by means of overexpression, an approach that we combined with allosteric activation of the enzyme (40). To date, the specific role of GDH in beta-cell function remains unclear. GDH participates in the glucose-induced amplifying pathway through generation of glutamate (41-43). The enzyme is also an amino acid sensor triggering insulin release upon glutamine stimulation under conditions of GDH allosteric activation (37; 39; 44).
More recently, the importance of GDH has been further highlighted by studies showing that SIRT4, a mitochondrial ADP-ribosyltransferase, downregulates GDH activity and thereby modulates insulin secretion (45; 46). Clinical data and associated genetic studies also revealed GDH as a key enzyme for the control of insulin secretion. Indeed, mutations rendering GDH more active are responsible for a hyperinsulinism syndrome (47). Mutations producing a less active, or even non active, GDH enzyme have not been reported, leaving the question open if such mutations would be either lethal or asymptomatic. We recently generated and characterized
-/-transgenic mice (named ßGlud1) with a beta-cell specific deletion of GDH (48). Data show that loss of GDH in beta-cells is associated with a ~40% reduction in glucose-stimulated insulin
-/-secretion and that the GDH pathway lacks redundant mechanisms. In ßGlud1 mice, the reduced
secretory capacity resulted in lower plasma insulin levels in response to both feeding and glucose load, while body weight gain was preserved (48). This demonstrates that GDH is essential for the full development of the secretory response in beta-cells, operating in the upper range of physiological glucose concentrations.
4.6. MITOCHONDRIAL ACTIVATION
TCA cycle activation induces transfer of electrons to the respiratory chain resulting in hyperpolarization of the mitochondrial membrane and generation of ATP (Fig. 2). The electrons are transferred by the pyridine nucleotide NADH and the flavin adenine nucleotide FADH. In 2
the mitochondrial matrix, NADH is formed by several dehydrogenases, some of which are
2+activated by Ca (25), while FADH is generated in the succinate dehydrogenase reaction. 2
Electron transport chain activity promotes proton export from the mitochondrial matrix across the inner membrane, establishing a strong mitochondrial membrane potential, which is negative on the inside. The respiratory chain comprises five complexes, the subunits of which are encoded by both the nuclear and mitochondrial genomes (49). Complex I is the only acceptor of electrons from NADH in the inner mitochondrial membrane and its blockade abolishes glucose-induced insulin secretion (32). Complex II (succinate dehydrogenase) transfers electrons to coenzyme-Q from FADH, the latter being generated both by the oxidative activity of the TCA cycle and the 2
glycerolphosphate shuttle. Complex V (ATP synthase) promotes ATP formation from ADP and inorganic phosphate. The synthesized ATP is translocated to the cytosol in exchange for ADP by the adenine nucleotide translocator (ANT). Thus, the actions of the separate complexes of the electron transport chain and the adenine nucleotide translocator couple respiration to ATP supply.
Mitochondrial activity can be modulated according to the nature of the nutrients, although glucose is the chief secretagogue as compared to amino acid catabolism (50) and fatty acid beta-
2+oxidation (51). Additional factors regulating ATP generation include mitochondrial Ca levels
(25; 52), mitochondrial protein tyrosine phosphatase (53), mitochondrial GTP (54), and matrix alkalinisation (55).
Mitochondrial activation also involves changes in organelle morphology and contacts. Mitochondria form dynamic networks, continuously modified by fission and fusion events under the control of specific mitochondrial membrane anchor proteins (56). Mitochondrial fission/fusion state was recently investigated in insulin secreting cells. Altering fission by down regulation of fission-promoting Fis1 protein impairs respiratory function and glucose-stimulated insulin secretion (57). The reverse experiment, consisting in overexpression of Fis1 causing mitochondrial fragmentation, results in a similar phenotype, i.e. reduced energy metabolism and
secretory defects (58). Fragmented pattern obtained by dominant-negative expression of fusion-promoting Mfn1 protein does not affect metabolism-secretion coupling (58). Therefore, mitochondrial fragmentation per se seems not to alter insulin secreting cells, at least not in vitro.
4.7. THE AMPLIFYING PATHWAY OF THE SECRETORY RESPONSE
2+The Ca signal in the cytosol is necessary but not sufficient for the full development of sustained insulin secretion. Nutrient secretagogues, in particular glucose, evoke a long-lasting
2+second phase of insulin secretion. In contrast to the transient secretion induced by Ca-raising
agents, the sustained insulin release depends on the generation of metabolic factors (Fig. 1). The
2+elevation of cytosolic Ca is a prerequisite also for this phase of secretion, as evidenced among
2+others by the inhibitory action of voltage-sensitive Ca channel blockers. Glucose evokes K-ATP
channel independent stimulation of insulin secretion, or the amplifying pathway (4), which is
2+unmasked by glucose stimulation when cytosolic Ca is clamped at permissive levels (59-61).
This suggests the existence of metabolic coupling factors generated by glucose.
4.8. MITOCHONDRIA-DERIVED NUCLEOTIDES AS COUPLING FACTORS
ATP is the primary metabolic factor implicated in K-channel regulation (62), secretory ATP
granule movement (63; 64), and the process of insulin exocytosis (65; 66).
Among other putative nucleotide messengers, NADH and NADPH are generated by glucose metabolism (67). Single beta-cell measurements of NAD(P)H fluorescence have demonstrated
2+that the rise in pyridine nucleotides precedes the rise in cytosolic Ca concentrations (30) and
that the elevation in the cytosol precedes the one in mitochondria (29). Cytosolic NADPH is generated by glucose metabolism via the pentose phosphate shunt (68), although mitochondrial
shuttles appear to be the main contributors in beta-cells (69). The pyruvate/citrate shuttle has received some attention over the last years and has been postulated as the key cycle responsible for elevation of cytosolic NADPH (69). As a consequence of mitochondrial activation, cytosolic
+NADPH is generated by NADP-dependent malic enzyme and suppression of its activity was
shown to inhibit glucose-stimulated insulin secretion in insulinoma cells (70; 71). However, such effects have not been reproduced in primary cells in the form of rodent islets (72), leaving the question open concerning its regulatory role.
Regarding the action of NADPH, it was proposed as a coupling factor in glucose-stimulated insulin secretion based on experiments using toadfish islets (73). A direct effect of NADPH was reported on the release of insulin from isolated secretory granules (74), NADPH being possibly bound or taken up by granules (75). More recently, the putative role of NADPH, as a signalling molecule in beta-cells, has been substantiated by experiments showing direct stimulation of insulin exocytosis upon intracellular addition of NADPH (76).
Glucose also promotes the elevation of GTP (77), which could trigger insulin exocytosis via GTPases (65; 78). In the cytosol, GTP is mainly formed through the action of nucleoside diphosphate kinase from GDP and ATP. In contrast to ATP, GTP is capable of inducing insulin
2+exocytosis in a Ca-independent manner (65). An action of mitochondrial GTP as positive regulator of the TCA cycle has been mentioned above (54).
The universal second messenger cAMP, generated at the plasma membrane from ATP, potentiates glucose-stimulated insulin secretion (79). Many neurotransmitters and hormones, including glucagon as well as the intestinal hormones glucagon-like peptide 1 (GLP-1) and gastric insulinotropic polypeptide (GIP), increase cAMP levels in the beta-cell by activating adenyl cyclase (80). In human beta-cells, activation of glucagon receptors synergistically amplifies the secretory response to glucose (81). Glucose itself promotes cAMP elevation (82) and oscillations in cellular cAMP concentrations are related to the magnitude of pulsatile insulin secretion (83). Moreover, GLP-1 might preserve beta-cell mass, both by induction of cell proliferation and inhibition of apoptosis (84). According to all these actions, GLP-1 and biologically active related molecules are of interest for the treatment of diabetes (85).
4.9. FATTY ACID PATHWAYS AND THE SECRETORY RESPONSE
The metabolic profile of mitochondria is modulated by the relative contribution of glucose and lipid products for oxidative catabolism. Carnitine palmitoyltransferase I, which is expressed in
thepancreas as the liver isoform (LCPTI), catalyzes the rate-limitingstep in the transport of fatty
acids into the mitochondria fortheir oxidation. In glucose-stimulated beta-cells, citrate exported
from the mitochondria (Fig. 2) to the cytosol reacts with coenzyme-A (CoA) to form cytosolic acetyl-CoA necessary for malonyl-CoA synthesis. Then, malonyl-CoA derived from glucose
metabolismregulates fatty acid oxidation by inhibiting LCPTI. The malonyl-CoA/long-chain
acyl-CoA hypothesis of glucose-stimulatedinsulin release postulates that malonyl-CoA derived
fromglucose metabolism inhibits fatty acid oxidation, thereby increasingthe availability of long-
chain acyl-CoA for lipid signals implicated in exocytosis (17). In the cytosol, this process promotes the accumulation of long chain acyl-CoAs such as palmitoyl-CoA (86; 87), which
2+enhances Ca-evoked insulin exocytosis (88).
In agreement with the malonyl-CoA/long-chain acyl-CoA model, overexpression of native
LCPTI in clonal INS-1E beta-cellswas shown to increase beta-oxidation of fatty acids and to
decreaseinsulin secretion at high glucose (51), although glucose-derived malonyl-CoA was still
able toinhibit LCPTI in these conditions. When the malonyl-CoA/CPTI interaction is altered in
cells expressing a malonyl-CoA–insensitiveCPTI, glucose-inducedinsulin release is impaired
The malonyl-CoA/long-chain acyl-CoA model has been challenged over the last years, essentially by modulating cellular levels of malonyl-CoA, either up or down. Either approach resulted in contradictory results, according to the respective laboratories performing such experiments. First, malonyl-CoA decarboxylase was overexpressed to reduce malonyl-CoA levels in the cytosol. In disagreement with the malonyl-CoA/long-chain acyl-CoA model, abrogation of malonyl-CoA accumulation during glucose stimulation does not attenuate the
secretory response (90). However, overexpression of malonyl-CoA decarboxylase in thecytosol
in the presence of exogenous free fatty acids, but not intheir absence, reduces glucose-stimulated
insulin release (91). The second approach was to silence ATP-citrate lyase, the enzyme that forms cytosolic acetyl-CoA leading to malonyl-CoA synthesis. Again, one study observed that such a maneuver reduces glucose-stimulated insulin secretion (70), whereas another group concluded that metabolic flux through malonyl-CoA is not required for the secretory response to glucose (71).
The role of long chain acyl-CoA derivatives remains a matter of debate, although several studies indicate that malonyl-CoA could act as a coupling factor regulating the partitioning of fatty acids
into effector moleculesin the insulin secretory pathway (92). Fatty acids, mobilized from intracellular triglyceride stores, might also play a permissive role in the secretory response (93;
94). Moreover, fatty acids stimulate the G-protein-coupled receptor GPR40/FFAR1 that is highly expressed in beta-cells (95). Activation of GPR40 receptor results in enhancement of glucose-
2+induced elevation of cytosolic Ca and consequently insulin secretion (96).
4.10. MITOCHONDRIA-DERIVED METABOLITES AS COUPLING FACTORS
Acetyl-CoA carboxylase catalyzes the formation of malonyl-CoA,a precursor in the biosynthesis
of long-chain fatty acids. Interestingly, glutamate-sensitiveprotein phosphatase 2A-like protein
activates acetyl-CoA carboxylase in beta-cells (97). This observation might link two metabolites proposed to participate in the control of insulin secretion. Indeed, the amino acid glutamate is another metabolic factor proposed to participate in the amplifying pathway (41; 42; 98). Glutamate can be produced from the TCA cycle intermediate α-ketoglutarate or by
transamination reactions (35; 50; 99). During glucose stimulation total cellular glutamate levels have been shown to increase in human, mouse and rat islets as well as in clonal beta-cells (19; 40; 41; 43; 100-102), whereas one study reported no change (103).
The finding that mitochondrial activation in permeabilised beta-cells directly stimulates insulin exocytosis (5) initiated investigations that identified glutamate as a putative intracellular messenger (41; 42). In in situ pancreatic perfusion, increased provision of glutamate using a cell permeant precursor results in augmentation of the sustained phase of insulin release (104). The glutamate hypothesis was challenged by overexpression of glutamate decarboxylase (GAD) in beta-cells to reduce cytosolic glutamate levels (100). In control cells, stimulatory glucose concentrations increased glutamate concentrations, whereas the glutamate response was significantly reduced in GAD overexpressing cells. GAD overexpression also blunted insulin secretion induced by high glucose, showing direct correlation between glutamate changes and the secretory response (100). In contrast, it was reported by others that glutamate changes may be dissociated from the amplification of insulin secretion elicited by glucose (101). Recently, we abrogated GDH, the enzyme responsible for glutamate formation, specifically in the beta-cells of transgenic mice. This resulted in a 40% reduction of glucose-stimulated insulin secretion (48). Export of glutamate out of the mitochondria is mediated by a newly identified protein, namely the glutamate carrier GC1 located in the inner mitochondrial membrane (105). Silencing of GC1 in beta-cells inhibits insulin exocytosis evoked by glucose stimulation, an effect rescued by the provision of exogenous glutamate to the cell (105).