The Mitochondrial Ca2+ Uniporter MCU Is Essential for Glucose-Induced ATP Increases in Pancreatic β-Cells

Glucose induces insulin release from pancreatic β-cells by stimulating ATP synthesis, membrane depolarisation and Ca2+ influx. As well as activating ATP-consuming processes, cytosolic Ca2+ increases may also potentiate mitochondrial ATP synthesis. Until recently, the ability to study the role of mitochondrial Ca2+ transport in glucose-stimulated insulin secretion has been hindered by the absence of suitable approaches either to suppress Ca2+ uptake into these organelles, or to examine the impact on β-cell excitability. Here, we have combined patch-clamp electrophysiology with simultaneous real-time imaging of compartmentalised changes in Ca2+ and ATP/ADP ratio in single primary mouse β-cells, using recombinant targeted (Pericam or Perceval, respectively) as well as entrapped intracellular (Fura-Red), probes. Through shRNA-mediated silencing we show that the recently-identified mitochondrial Ca2+ uniporter, MCU, is required for depolarisation-induced mitochondrial Ca2+ increases, and for a sustained increase in cytosolic ATP/ADP ratio. By contrast, silencing of the mitochondrial Na+-Ca2+ exchanger NCLX affected the kinetics of glucose-induced changes in, but not steady state values of, cytosolic ATP/ADP. Exposure to gluco-lipotoxic conditions delayed both mitochondrial Ca2+ uptake and cytosolic ATP/ADP ratio increases without affecting the expression of either gene. Mitochondrial Ca2+ accumulation, mediated by MCU and modulated by NCLX, is thus required for normal glucose sensing by pancreatic β-cells, and becomes defective in conditions mimicking the diabetic milieu.


Introduction
Glucose-induced insulin secretion from pancreatic b-cells is essential to ensure the normal control of blood glucose concentrations [1]. Defects in b-cell glucose sensitivity [2,3] as well as a decrease in b-cell mass [4] are cardinal aspects of type 2 diabetes mellitus (T2D). A key event in glucose-induced insulin release is the stimulation of mitochondrial oxidative metabolism [5,6]. Enhanced ATP synthesis [7] results in the closure of ATP-sensitive K + (K ATP ) channels [8], membrane depolarisation and Ca 2+ influx via voltage-gated Ca 2+ channels, which triggers insulin release [1,9].
In most mammalian cells, mitochondrial oxidative metabolism is thought to be stimulated by Ca 2+ [10,11] through the activation of intramitochondrial dehydrogenases [12]. This stimulates the supply of reducing equivalents to the respiratory chain [13], and hence ATP synthesis [14]. The above process is thought also to be important in pancreatic b-cells [15] and recent analyses using a mitochondrial Ca 2+ buffer [14] have suggested that mitochondrial Ca 2+ accumulation is important for sustained insulin secretion.
The interplay between cytosolic Ca 2+ , mitochondrial Ca 2+ and ATP synthesis has nonetheless remained enigmatic in the b-cell. In particular, Ca 2+ entry into the cytosol, triggered by elevated ATP, is expected to enhance ATP hydrolysis, for example by activating granule exocytosis [16] and Ca 2+ ATPases which pump the cation out of the cytosol [17]. The Ca 2+ -induced drop in ATP is then predicted to open K ATP channels, thereby arresting Ca 2+ influx [18]. In addition, Ca 2+ has been suggested to induce repolarisation of the plasma membrane by opening Ca 2+ -activated K + channels [19] or depolarising the mitochondrial inner membrane, which decreases the driving force for ATP synthesis by the F 1 F o ATPase [20].
Until very recently, the molecular entities responsible for catalysing mitochondrial Ca 2+ uptake have remained unclear in any mammalian cell type. However, two reports in 2011 identified a Ca 2+ -selective mitochondrial uniporter, MCU, encoded by the Ccdc109a gene [21,22], in a complex with a Ca 2+ sensing subunit MICU1 [23], as the likely Ca 2+ transporting entity. Conversely, mitochondrial Ca 2+ efflux was proposed to be mediated by the Na + -Ca 2+ exchanger NCLX [24]. Whether these transporters catalyse mitochondrial Ca 2+ transport in the b-cell, and may thus modulate insulin secretion, is currently unknown.
In the present study, we have sought to explore (a) the molecular mechanisms responsible for Ca 2+ transfer across the mitochondrial membrane in b-cells and (b) the impact of these changes on cytosolic ATP dynamics and electrical excitability. To these ends, we have deployed a recently-developed, molecularly-addressed GFP-based recombinant probe for mitochondrial Ca 2+ ([Ca 2+ ] mit ), 2mt8RP [25], alongside a trappable cytosolic Ca 2+ probe (Fura Red) allowing us to image [Ca 2+ ] cyt simultaneously with [Ca 2+ ] mit in individual primary mouse b-cells. These measurements have been combined with perforated patch electrophysiology to allow plasma membrane potential (V m ) to be recorded or controlled without perturbing cellular composition or metabolism [26]. Critically, this approach permits the ready and rapid control of [Ca 2+ ] cyt via voltage-gated Ca 2+ channels [27] and thus an analysis of the interplay between [Ca 2+ ] cyt and [Ca 2+ ] mit in real time. In parallel, the novel ATP sensor Perceval [28], based on the bacterial regulatory protein, GlnK1, has been used to monitor the cytosolic ATP/ADP ratio ([ATP/ADP] cyt ). These combined approaches have allowed us to characterise the roles of MCU and NCLX as regulators of mitochondrial ATP synthesis in the b-cell.

Results
Glucose induces a monophasic increase in cytosolic Ca 2+ but a biphasic increase in cytosolic ATP/ADP ratio We sought first to determine whether increases in [Ca 2+ ] cyt and/or [Ca 2+ ] mit might influence glucose-induced increases in [ATP/ADP] cyt . The latter parameter was therefore imaged in single mouse b-cells expressing the GFP-based probe Perceval [28], which was chiefly localised to the cytosol as expected (Suppl. Fig. S1A). Changes measured with this probe were shown to be unrelated to small alterations in cytosolic pH, and thus largely to reflect [ATP/ADP] cyt (Suppl. Fig. S2A). [Ca 2+ ] cyt was imaged simultaneously in the same cell using the trappable cytosolic/ nuclear probe Fura-Red (Suppl. Fig. S1A) whilst V m was monitored using patch-clamp in current-clamp mode [3].
b-Cells maintained at low (3 mM) glucose exhibited a resting V m of 26861 mV (n = 30, from 12 separate islet preparations; point i in Fig. 1A). An increase in glucose concentration to 17 mM led to a rapid elevation in [ATP/ADP] cyt (Fig. 1A, point ii) and an increase in input resistance, followed by depolarisation of the plasma membrane and a [Ca 2+ ] cyt rise, as expected. This was closely followed by a drop in [ATP/ADP] cyt (Fig. 1A, iii). The 3364% drop (''trough'' in Fig. 1B) was, however, transient and [ATP/ADP] cyt quickly recovered and displayed a steady further increase (Fig. 1A, iv). The increase was not associated with any significant decrease in [Ca 2+ ] cyt , and thus was not likely to reflect a lowering demand for Ca 2+ extrusion or other ATP-consuming processes. Furthermore, setting V m to 270mV via the patch pipette, thus closing voltage-gated Ca 2+ channels, led to a prompt decrease in [Ca 2+ ] cyt (Fig. 1A, v). The application of the mitochondrial uncoupler carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP) resulted in an abrupt decrease of [ATP/ ADP] cyt , as expected (Fig. 1A, vi), and an elevation of [Ca 2+ ] cyt , presumably due to a compromise in Ca 2+ pumping across the plasma and ER membranes.
Combining data from multiple experiments (n = 30 single cells, Fig. 1B) we were able to observe that high glucose induced an [ATP/ADP] cyt elevation in b-cells in two distinct phases (Fig. 1B). A rapid first phase preceded membrane depolarisation and electrical activity, whilst a slower second phase resulted in a larger increase of [ATP/ADP] cyt (Fig. 1B). These changes contrasted with the essentially monophasic (albeit oscillatory) increases in [Ca 2+ ] cyt (Fig. 1A).
Cytosolic Ca 2+ influx is essential for the second phase of cytosolic ATP/ADP ratio increase To dissect the dependence of the observed ATP increases on cytosolic Ca 2+ increases prompted by depolarisation in response to glucose, we measured the changes in [ATP/ADP] cyt in response to the sugar while keeping the cell hyperpolarised (V m = 270mV) using the patch pipette in voltage-clamp mode (as in point v, Fig. 1A). This prevented extracellular Ca 2+ from entering the cytosol even at high extracellular glucose.
An increase in glucose from 3 mM to 17 mM resulted in a rapid elevation of [ATP/ADP] cyt , followed by a saturation of the [ATP/ ADP] cyt level (ii, Fig. 2). Notably, in the absence of Ca 2+ influx, neither a trough, nor an increase in [ATP/ADP] cyt (see e.g. points iii and iv in Fig. 1A) were observed, suggesting that Ca 2+ influx is involved in the latter changes. To test this possibility, we imposed forced changes in [Ca 2+ ] cyt with a train of 10 depolarisations (as given in Suppl. Fig. S2B) and then setting V m back to 270mV (as indicated in the V m trace in Fig. 2 followed by its recovery. The two phases of the glucose-induced increase in [ATP/ADP] cyt can therefore be classified as Ca 2+ -independent (the one that precedes) and Ca 2+ -dependent (the one that follows) Ca 2+ entry.
We next sought to determine whether the apparent increases in cytosolic ATP/ADP ratio reported with Perceval were associated with the closure of ATP-sensitive K + channels, as expected. This seemed an important question since fluctuations in ''global'' cytosolic ATP/ADP differ in some circumstances from those immediately beneath the plasma membrane, as recorded with a targeted luciferase-based probe [7]. The electrophysiological configuration used here allowed us to address this point as follows.
While keeping the cell hyperpolarised, at 270mV (Fig. 2), we applied small pulses between 265 and 280 mV, to monitor slow whole-cell current, I m . These pulses were too small to trigger any voltage-gated Ca 2+ conductance and therefore had no effect on Ca 2+ entry. The addition of 17 mM glucose decreased I m during the Ca 2+ -independent phase of [ATP/ADP] cyt increase (Fig. 2, inset), most likely due to the inhibition of K ATP channels, the main providers of the b-cell conductance (G m ) [29]. G m thus was found to decrease from the initial value of 0.4360.09 nS/pF to 0.0960.02 nS/pF (n = 12) during the Ca 2+ -independent phase. A strong and significant correlation (Pearson's r = 20.8460.05, p,0.05, n = 12) between the elevation of [ATP/ADP] cyt as recorded with Perceval, and the closure of K ATP changes as measured above, (Suppl. Fig. S2C) indicated that the optical measurements with the GFP-based probe provided a useful guide to [ATP/ADP] cyt changes in the physiologically-relevant domain beneath the plasma membrane. Interestingly, half-maximal inhibition of G m coincided with the increase of [ATP/ADP] cyt of 2068% (n = 12, Suppl. Fig. S2C), while earlier data [29] suggest that half-maximal G m is likely to be reached at around 2864% of the [ATP/ADP] cyt increase. Thus, the increase in [ATP/ADP] cyt was reported with a 32621 s delay after the drop in G m measured using patch-clamp. This small delay may reflect the propagation of the glucose-induced ATP increase from the sub-membrane compartment to the bulk cytosol [7,30].

Glucose induces a sequential increase in [Ca 2+ ] cyt and [Ca 2+ ] mit
We next explored the possibility that the uptake of Ca 2+ by mitochondria may be related to the second phase of [ATP/ ADP] cyt increase, as suggested by earlier experiments in b-cell populations [14]. To explore the temporal relationship between

MCU mediates mitochondrial Ca 2+ increases and the second phase of glucose-induced [ATP/ADP] cyt increases
In experiments using an identical configuration to those above, the maximal rate of [ATP/ADP] cyt decrease was observed 106622 s after the first action potential (between points ii and iii in Fig. 1A). This observation, and those described for the time course of mitochondrial Ca 2+ increases (Fig. 3B, C), are thus consistent with the possibility that mitochondrial Ca 2+ accumulation (and hence an activation of oxidative metabolism) plays a role in the regulation of the [ATP/ADP] cyt increase that follows an initial and small Ca 2+ -induced drop. To test this possibility directly we therefore reduced the expression of the recently-identified mitochondrial Ca 2+ uniporter, MCU [21,22], in b-cells by .80% (as assessed by qRT-PCR, not shown) using a lentivirally-delivered shRNA (Fig. 4). Silencing of MCU caused a substantial impairment of apparent Ca 2+ entry into mitochondria, whilst the imposed cytosolic Ca 2+ increases were unaffected (Fig. 4A, B). Importantly, this manipulation also resulted in an alteration of the glucose-induced [ATP/ADP] cyt changes (Fig. 5A, B). Thus, MCU silencing had no effect on the first phase of the glucose-induced [ATP/ADP] cyt increase, the rise of [Ca 2+ ] cyt or subsequent electrical spiking (Fig. 5A). However, the second (Ca 2+ -dependent) phase of the [ATP/ADP] cyt increase, i.e. the [ATP/ADP] cyt recovery, was significantly impaired in the b-cells where MCU expression was reduced (Fig. 5A, B).
To determine whether MCU knock-down might affect mitochondrial membrane potential (Y m ) independently of a Ca 2+ increase, we explored the glucose-induced changes in this parameter prior to [Ca 2+ ] cyt elevation using tetramethyl rhodamine, ethyl ester (TMRE). The resting Y m (measured as 212764 mV in control vs 213365 mV in MCU 2 cells) and the kinetics of the glucose-induced change (Fig. 5C) were not affected by the knock-down of MCU.

NCLX modulates mitochondrial Ca 2+ changes
Pharmacological inhibition of mitochondrial Na + -Ca 2+ exchange has been reported to elevate the basal ATP levels in INS-1 cells and primary rat islets [31]. However, the agent used (CGP37157) was likely to affect cellular Ca 2+ homeostasis by targeting plasma membrane voltage-gated Ca 2+ channels, as reported by Luciani et al [32]. NCLX was recently identified as an essential component of the mitochondrial Na + -Ca 2+ exchanger [24], responsible for Ca 2+ efflux from mitochondria, thereby providing an opportunity for a specific inhibition of Ca 2+ efflux from mitochondria through RNA interference. In the present study, silencing of NCLX significantly potentiated depolarisationinduced increases in [Ca 2+ ] mit (Fig. 6A, B). NCLX silencing also slightly accelerated the onset of the first phase of the [ATP/ ADP] cyt response to glucose (Fig. 6C, D), but had no significant effect on the amplitude of the [ATP/ADP] cyt changes (Fig. 6C, E).

Chronic glucolipotoxicity inhibits mitochondrial Ca 2+ increases and delays [ATP/ADP] cyt recovery
Previous studies [33] have indicated that the structure and localisation of mitochondria are altered in b-cell dysfunction, including glucolipotoxicity, i.e. exposure to high levels of free fatty acids (FFA) and glucose. Importantly, glucose-induced ATP increases in the b-cell are impaired in this model of T2D [34]. We therefore sought to determine whether these changes were also associated with defective mitochondrial Ca 2+ increases or altered expression of mitochondrial Ca 2+ transporters.
To this end, we cultured primary mouse b-cells under glucolipotoxic conditions (''FFA + '' cells) and studied the impact on the dynamics of [Ca 2+ ] cyt and [Ca 2+ ] mit in response to V m manipulation. FFA + cells displayed slower dynamics of [Ca 2+ ] mit increase (Fig. 7A, B). This resulted in a slower onset of the second phase of glucose-induced ATP increase (Fig. 8A, B) in FFA + bcells. This effect was not likely to be caused by changes in resting Y m (213564 mV in control vs 213764 mV in FFA + cells) or the kinetics of the glucose-induced change in Y m (Fig. 8C). We also failed to observe any significant change of either MCU or NCLX mRNA levels under these conditions (Fig. 8D). The expression of the transcription factor pancreatic duodenum homeobox-1 (Pdx1), in contrast, was significantly reduced by the chronic glucolipotoxicity, in line with earlier observations [35].

Multiparametric analysis of glucose signalling in single primary b-cells
We dissect here the role of mitochondrial Ca 2+ transport in the stimulation of single primary pancreatic b-cells with glucose using a combined imaging and electrophysiology approach. This has allowed us to monitor or manipulate up to four key parameters simultaneously in the same individual cell. Earlier studies in these cells combined the use of a microelectrode [36] or patch-clamp [37] with [Ca 2+ ] measurements to report a close association of [Ca 2+ ] cyt and V m signals during glucose-induced depolarisation.
Furthermore, the control of V m using perforated-patch was shown to be a very efficient means of rapid and precise control of [Ca 2+ ] cyt [19,38]. The latter strategy provided a powerful tool here to explore the inter-relationships between Ca 2+ changes in discrete compartments and with the control of ATP synthesis. Thus, a key technical advantage over earlier studies [14] has been the ability to resolve the exact sequence in which signalling events occurred within the same individual cell. Moreover, possible artefacts resulting from the progressive recruitment of cells within a population were also excluded.
These studies also represent the first use of the novel ATP/ADP probe Perceval [28] in an excitable cell, and provide significant advances over the previous use of less sensitive luciferase-based reporters [7,39]. Although the affinity of Perceval for ATP is relatively high, competition with ADP lowers its sensitivity to a range appropriate for the b-cell cytosol (,1 mM ATP at 3 mM glucose) [7,29]. Importantly, pH changes appeared not to interfere with the probe (Suppl. Fig. S2A).

MCU mediates mitochondrial Ca 2+ uptake and enhanced ATP synthesis in pancreatic b-cells
We demonstrate here firstly that both cytosolic and mitochondrial Ca 2+ increases are essential for the sustained (second) phase of [ATP/ADP] cyt increase in response to high glucose. Interestingly, we show (Fig. 2) that a transient imposed increase in [Ca 2+ ] cyt is sufficient to lead to a progressive and sustained increase in [ATP/ADP] cyt . This finding is consistent with the possibility that mitochondrial uptake of Ca 2+ in response to high glucose (which is slow compared to increases in cytosolic Ca 2+ ; Fig. 3B, C) may then allow a sustained activation (i.e. ''plasticity'' or ''memory'') of oxidative metabolism [39,40].
Recent studies [21,22], have provided convincing evidence for a role of MCU in mitochondrial transport in mammalian fibroblasts. However, no evidence currently exists demonstrating a role for this protein in this process in a more differentiated cell type. We report here firstly that MCU is critical for mitochondrial Ca 2+ accumulation in pancreatic b-cells in response to depolarisationinduced Ca 2+ increases. Likewise, we show that the Na + -Ca 2+ exchanger NCLX [24] regulates [Ca 2+ ] mit increases and may thus be involved in regulating the responses to glucose, consistent with earlier findings using the pharmacological inhibitor CGP37157 [31]. Specifically, NCLX silencing affected the kinetics of the glucose-induced ATP/ADP changes but had no significant effect on the steady-state ATP/ADP level. Although the mechanisms underlying this unexpected observation are presently unclear, they may involve glucose-dependent changes in cytosolic [Na + ] (unpublished observation of I.S.). Future studies are required to address this question and the role of NCLX in the b-cell.
Overall, our data support a two-phase model (Fig. 9), in which an initial increase in cytosolic [ATP/ADP] (first phase) occurs independently of any increase in cytosolic (or mitochondrial) Ca 2+ concentration. In the second phase, the elevation of cytosolic Ca 2+ concentration leads to a gradual increase in mitochondrial Ca 2+ (Fig. 3B). This, in turn, is likely to activate intramitochondrial dehydrogenases [10] (and perhaps other mitochondrial enzymes) [41], stimulating respiratory chain activity and hence mitochondrial ATP production. In line with this view, the initial rapid glucose-induced increase in [ATP/ADP] cyt (first phase) was not affected by the MCU silencing whereas the second phase of [ATP/ADP] cyt increase was essentially eliminated.
A recent study [14] also described biphasic increases in cytosolic ATP/ADP in b-cell populations in response to glucose, and indicated that mitochondrial Ca 2+ accumulation may be essential for increases in cytosolic ATP/ADP in response to the sugar. However, this earlier study relied on the over-expression in the mitochondrial matrix of a high affinity (and high capacity) calcium-binding protein, S100G. Whether the presence of this protein within the mitochondrial matrix may interfere with normal mitochondrial function (for example by leading to a decrease in mitochondrial pH as a result of Ca 2+ binding) is unclear. Mitochondrial Ca 2+ accumulation, catalysed by MCU, is revealed here to be essential for the second phase of glucoseinduced ATP synthesis by glucose. What may be the consequences for electrical activity and insulin secretion? Increases in ATP are believed to be involved in both ''K ATP -dependent'' and ''K ATPindependent'' regulation of exocytosis by glucose [16,42]. Importantly, we obtained no evidence for a role for mitochondrial Ca 2+ accumulation in the regulation of plasma membrane electrical activity (Fig. 5) suggesting that an involvement of mitochondrial Ca 2+ in the regulation of insulin secretion, as implied by earlier studies [14], is likely to involve the latter (K ATP -independent) action on secretory granule movement or fusion, perhaps powered by ATP increases [43]. Further studies, using larger cell populations, will be necessary to explore the impact of MCU on phasic insulin secretion.
A role for mitochondrial Ca 2+ transport in b-cell glucolipotoxicity?
We show here that glucolipotoxic conditions impair Ca 2+ transport into mitochondria (Fig. 7) and the second phase of glucose-induced ATP/ADP increases (Fig. 8). The expression of both MCU and NCLX was unaltered under these conditions (Fig. 8D), in line with previous studies in models of diet-induced bcell dysfunction in rodents [44]. It is therefore likely that changes in the intracellular distribution of mitochondria induced by the diabetic milieu [33] are involved in this impairment in mitochondrial Ca 2+ transport. These changes in mitochondrial architecture, and hence localisation at sites of Ca 2+ entry into the cytosol [45], may consequently interfere with mitochondrial Ca 2+ transport and ATP production.

Conclusions
We show here that mitochondrial Ca 2+ uptake in the excitable b-cell is mediated by MCU and modulated by NCLX. Changes in Ca 2+ in the mitochondrial matrix are shown to be critical for increases in cytosolic ATP/ADP ratio, and may thus be required for glucose-stimulated insulin secretion [14]. Manipulation of MCU activity, in particular, may thus provide potential strategies to improve defective insulin secretion in some forms of diabetes.

Islet isolation and culture
Female CD1 mice were sacrificed by cervical dislocation as approved by the United Kingdom Home Office (HO) Animal Scientific Procedures Act, 1986 and designated as ''Schedule 1'' procedure. Animals were maintained under HO Licence PPL 70/ 7349 (Holder Dr I Leclerc), which received local ethical committee approval, and all participants received approved local training at Imperial College. Pancreatic islets were isolated by collagenase digestion [46], pre-cultured for 5 h in RMPI-1640 medium, containing 11 mM glucose, 10% FCS, 100 U penicillin, 100 mg streptomycin, at 37uC, 5%CO 2 , infected with an appropriate adenovirus encoding cDNA for the required probe, split into single b-cells and plated on glass coverslips. The cells were then cultured for .24 h in absolute humidity for 2-4 days and assayed as described below. Glass-attached single cells or 2-3-cell clusters displayed an infection efficiency of ,90%. b-Cells were identified morphologically and according to their electrophysiological characteristics (membrane capacitance, V m , K ATP current, lack of Na + current, response to glucose).
Chronic glucolipotoxicity was modelled by culturing the cells in medium containing 0.5 mM Na + -palmitate and 17 mM glucose for 72 h. Palmitate was prepared as a 150 mM stock in ethanol; the working solution also contained 0.67% fatty-acid free BSA (Sigma). Control medium contained, respectively, 0.67% FFA-free BSA and 0.17% ethanol.
MCU was silenced in primary b-cells by 24h incubation with shRNA-bearing lentiviral particles (sc-142052-V, Santa-Cruz Biotechnology), at 1610 6 infectious units/ml. Cells infected with the GFP + control particles (sc-108084) at the same titre displayed a multiplicity of infection of two, 36 hours after infection. Particles delivering non-target shRNA (sc-108080) were used as a negative control.

Molecular biology and generation of adenoviruses
cDNA encoding Perceval [28] was excised from pGW1CMV-Perceval plasmid (kindly provided by Prof Gary Yellen, Yale University) by restriction first with EcoRI, then extension using T4 DNA-polymerase and finally by restriction with HindIII to liberate the insert. The HindIII/blunt insert was cloned into pShuttleCMV previously digested with EcoRV and HindIII.
cDNA encoding 2mt8-ratiometric pericam (2mt8RP) was kindly provided by Prof Tullio Pozzan (University of Padua). ''Mt8'' refers to the first 36 amino acids of subunit VIII of human cytochrome c oxidase (COX) while the targeting efficiency was improved by using two tandem repeats of the addressing sequence [25]. Adenoviral particles were produced as in [47].

Gene expression measurement by qRT-PCR
RNA was purified from islet samples using Trizol. RNA was quantified by Nanodrop spectrophotometer then reverse transcribed using a High Capacity cDNA Reverse Transcription kit (Applied Biosystems). mRNA abundance was quantified by qPCR using Sybr Green PCR Master Mix (Applied Biosystems) on a 7500 Fast Real-time PCR machine. Expression of each gene was normalised to cyclophilin A (Ppia), and FFA treatment effect as fold change with 95% confidence intervals was calculated using the DDC T method on 7500 Software (Applied Biosystems, v2.0.5).

Single cell epifluorescence imaging
Simultaneous imaging of [Ca 2+ ] in mitochondria and the cytosol was performed using the mitochondrial pericam 2mt8RP, and Fura-Red (Invitrogen) respectively. 2mt8RP, Fura-Red and Indo-1 were used at single excitation and emission wavelengths. Either dye was dissolved in DMSO (4mM) containing 4% F127-Pluronic. Cells were loaded with Fura-Red by incubation with 4mM of the dye in the extracellular solution for 30 min. Imaging experiments were performed on an Olympus IX-71 microscope with UPlanFL N 640 magnification objective. For acquisition, an F-View-II camera and MT-20 excitation system equipped with a Hg/Xe arc lamp were used, under control of CellˆR software (Olympus). Excitation/emission wavelengths were (nm): 410/535 (2mt8RP), 490/630 (Fura-Red), 490/535 (Perceval). Images were acquired at a frequency of 0.2 Hz with typical excitation times of 10 ms. The acquisition of the fluorescence and electrophysiological data was synchronized using TTL pulses. Imaging data was background-subtracted, analysed and presented as F/F 0 (Perceval) and F 0 /F (Fura-Red, 2mt8RP). Whole cells were selected as regions of interest (ROI) to minimize the effect of the cell drift. For cell clusters, only the patched cell was included in the ROI. Every [Ca 2+ ] recording was subjected to the dynamic range control by applying, at the end of the trace, solutions containing 10 mM ionomycin: ''Ca 2+ -free'' (0.5 mM EGTA), ''Ca 2+ -max'' (5 mM Ca 2+ ). For the [ATP/ADP] cyt recordings the dynamic range was controlled by high glucose (maximum after .30 min of exposure) and 2mM carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP; minimum).

Measurements of TMRE fluorescence
Cells were loaded with 7 nM TMRE for 60min at 3 mM glucose. Confocal imaging was performed in bath solution (see below) initially containing 3mM glucose, using a Zeiss microscope fitted with a Plan Apochromat x63 n. a. 1.4 oil immersion objective and equipped with Yokogawa CSU22 spinning disk module. The TMRE fluorescence signal was excited at 563 nm using a solid-state laser. Emission at 600 nm was registered using Hamamatsu ImagEM EM-CCD camera. The calculations of Y m were done on the basis of the ratio of mitochondrial and cytosolic fluorescence, as was outlined in [48].

Electrophysiology
Electrophysiological recordings and stimulation were done in whole-cell perforated-patch configuration, using an EPC9 patchclamp amplifier controlled by Pulse acquisition software (HEKA Elektronik). The pipette tip was dipped into pipette solution, and then back-filled with the same solution containing 0.17 mg/ml amphotericin B. Series resistance and cell capacitance were compensated automatically by the acquisition software. Record- ings, triggered by the TTL pulse, were started in current-clamp mode, and the depolarization of the plasma membrane was monitored simultaneously with [Ca 2+ ] and [ATP/ADP] cyt , in response to a glucose step from 3 to 17 mM. To monitor the input resistance, the protocol included 10-ms injections of repolarising 10-pA current applied every 20s. The parameters of the current injections were chosen to minimise their effect on the glucoseinduced electrical activity. To control V m and impose electrical stimulations, the mode was periodically switched to voltage-clamp [49]. V m was held at the value of 270 mV, with 0.5 Hz +5/ 210 mV pulses to monitor the K ATP conductance (see Suppl. Fig. S2B). The electrical stimulation was deemed to mimic the naturally occurring bursts of action potentials and comprised of 5-s depolarization trains to 230 mV containing 25 ramps of 100 ms to 0 mV and back (Suppl. Fig. S2B). Data were filtered at 1 kHz, and digitised at 2 kHz. G m was normalized to cell capacitance to account for cell size.

Data analysis
Imaging data was analysed using CellˆR (Olympus) and ImageJ (Wayne Rasband, NIMH). The simultaneous recordings were combined together and analysed using Igor Pro (Wavemetrics). The results are presented as mean6SEM. Mann-Whitney U-test and Wilcoxon's paired test were used to assess the statistical significance of the differences between the independent and dependent samples, respectively. Effects of pH, analysis of kinetics. A: Comparison of the effects of glucose and pH on the Perceval fluorescence. 17 mM glucose was applied to the cell, followed by 140 mM K + plus 10 mM nigericin solutions of the indicated pH. B: Schematic of the depolarisation protocol (single burst). C: The first phase of glucoseinduced [ATP/ADP] cyt increase and the decrease in G m were closely associated in time. G m was calculated from I m traces (Fig. 2B, inset). The pairs of signals (n = 12) were normalised by the range of change during the first phase of ATP elevation. (TIF) Figure 9. Proposed scheme of interplay between Ca 2+ , ATP and V m in the b-cell. The oxidation of glucose that enters the b-cell hyperpolarises the mitochondrial membrane (DY m ) thereby leading to the elevation of cytosolic ATP/ADP ratio, closing of K ATP channels, depolarisation of the plasma membrane (V m ) and Ca 2+ entry. Elevated cytosolic [Ca 2+ ] triggers a number of ATP-dependent processes including insulin secretion and Ca 2+ removal into the ER and extracellular medium. By entering mitochondria via MCU, Ca 2+ potentiates oxidative metabolism to counter-balance ATP expenditure. Ca 2+ exits mitochondria via NCLX. doi:10.1371/journal.pone.0039722.g009