Optogenetic regulation of insulin secretion in pancreatic β-cells

Pancreatic β-cell insulin production is orchestrated by a complex circuitry involving intracellular elements including cyclic AMP (cAMP). Tackling aberrations in glucose-stimulated insulin release such as in diabetes with pharmacological agents, which boost the secretory capacity of β-cells, is linked to adverse side effects. We hypothesized that a photoactivatable adenylyl cyclase (PAC) can be employed to modulate cAMP in β-cells with light thereby enhancing insulin secretion. To that end, the PAC gene from Beggiatoa (bPAC) was delivered to β-cells. A cAMP increase was noted within 5 minutes of photostimulation and a significant drop at 12 minutes post-illumination. The concomitant augmented insulin secretion was comparable to that from β-cells treated with secretagogues. Greater insulin release was also observed over repeated cycles of photoinduction without adverse effects on viability and proliferation. Furthermore, the expression and activation of bPAC increased cAMP and insulin secretion in murine islets and in β-cell pseudoislets, which displayed a more pronounced light-triggered hormone secretion compared to that of β-cell monolayers. Calcium channel blocking curtailed the enhanced insulin response due to bPAC activity. This optogenetic system with modulation of cAMP and insulin release can be employed for the study of β-cell function and for enabling new therapeutic modalities for diabetes.


Expression of bPAC and [cAMP] i modulation in MIN6 β-cells. We set out to investigate if [cAMP] i and insulin release can be modulated via light in β-cells upon expression of the bPAC from
Beggiatoa. An adenoviral vector, AdbPAC, was constructed for delivering the human codon-optimized bPAC to MIN6 β-cells. The myc tag (fused to the C-terminus of bPAC) and fluorescent protein mCherry were included in the cassette to facilitate the detection of bPAC (Fig. 1A,B). Based on immunostaining and fluorescence, the expression of mCherry was seemingly lower (Fig. 1A), which is typical for cistrons downstream of an IRES element 22,23 . Indeed, the fraction of cells expressing bPAC was determined at different multiplicities of infection (MOI) either by immunostaining for the myc tag and microscopy or by flow cytometry for mCherry (Suppl. Fig. S1). Transduction levels increased to almost 55% for MOI of 100 with greater than 70% cell viability. When [cAMP] i was measured, we noted an increase that was commensurate with the bPAC expression based on the MOI (Fig. 1C). Cells infected at MOI of 100 displayed the highest [cAMP] i level with a 10-min photoinduction (8.59 ± 2.36 vs. 2.03 ± 0.53 (dark) ng cAMP/mg total protein, p = 0.035, n = 3).
We also probed the relationship between the duration of illumination and increase in [cAMP] i ('on' kinetics). For this purpose, cultures transduced with AdbPAC were stimulated with continuous or pulsed light (5 s on/10 s off) for various durations between 1-20 min in 2.5 mM glucose (Fig. 1D). Compared to bPAC-expressing cells kept in dark, cells exposed to blue light for 5 min or longer exhibited a 6-to 8-fold increase in [cAMP] i . There was also no difference between the two modes of photostimulation in this setup. Conversely, the ('off ') kinetics were determined of [cAMP] i reduction post-illumination (Fig. 1E). After a 10-min illumination, [cAMP] i levels returned to a stable baseline within 12 min albeit higher than for cells kept in dark (4.75 ± 0.78 vs. 1.99 ± 0.14 ng cAMP/mg total protein, p = 0.023, n = 3). Incubation of bPAC-expressing MIN6 β-cells with an AC inhibitor led to a reduction in the [cAMP] i (Suppl. Fig. S2) further illustrating that the observed increase in the secondary messenger is due to the action of bPAC upon photoactivation. Taken together, our data show that heterologous expression of bPAC in β-cells allows the on-demand alteration of [cAMP] i with light following physiologically relevant kinetics similar to [cAMP] i changes in islet β-cells exposed to changing concentrations of glucose 24 .

Insulin secretion in β-cells expressing bPAC. The light-mediated modulation of [cAMP] i in
bPAC-expressing β-cells prompted us to investigate if insulin release could be regulated in an analogous manner. AdbPAC-transduced MIN6 β-cells at 2.5 mM of glucose secreted 131.65 ± 15.14 ng insulin/mg total protein/h ( Fig. 2A) while the rate for non-transduced MIN6 β-cells was 34.97 ± 6.99 ng insulin/mg total protein/h (p = 2.74 × 10 −3 , n = 4). Similarly, bPAC-expressing cells at 25 mM glucose released 296.6 ± 63.25 vs. 87.57 ± 31.8 ng insulin/mg total protein/h (p = 6.91 × 10 −3 , n = 4) for control cells. Thus, blue light triggered an analogous increase (~3.5-fold) in bPAC-positive cells at low and high concentrations of glucose. In contrast, this increase was not observed in bPAC-expressing cells that were kept in dark. Of note, cells infected with AdGFP (control) did not differ in their hormone secretion from uninfected cells, suggesting that the adenoviral transduction did not affect the secretory function of β-cells. Additionally, the expression of bPAC and photostimulation augmented the rate of insulin release in another β-cell line (βTC cells; Suppl. Fig. S3), supporting the notion that the enhanced insulin secretion that we observed upon photoexposure of bPAC was not applicable only to MIN6 β-cells. The ratio of hormone secretion rates at 25 mM to 2.5 mM glucose was similar (~2.2-2.5) between cells expressing bPAC and those that did not suggesting that light-stimulated bPAC causes a surge in the rate of insulin release but at a constant ratio.
These data show that elevation of [cAMP] i amplifies the secretion of insulin in the presence of glucose, even at 2.5 mM glucose, in bPAC-expressing MIN6 β-cells after activation with light. This behavior parallels the response of AdbPAC-transduced primary islets (see results in the next section) after photostimulation and is in support of a previous report indicating the resemblance in the regulation of insulin secretion between particular insulinoma lines and normal β-cells 25 .
We also compared the enhanced hormone release due to treatment with blue light of bPAC-expressing cells with that resulting from incubation with various secretagogues (Fig. 2C). The responses were statistically similar even when cells were cultured with forskolin (AC activator) and IBMX (phosphodiesterase inhibitor), indicating that bPAC expression and activation by light mimics the amplifying effects of secretagogues on the insulin release by β-cells. Of note, the [cAMP] i levels of AdbPAC-transduced cells exposed to light were higher than for β-cells treated with forskolin or IBMX (Suppl. Fig. S4), suggesting that there may be an upper threshold in [cAMP] i above which there is no further significant increase in insulin secretion.
Transduced cells were also subjected to repeated cycles of light stimulation to assess the secretory capacity of β-cells over successive activations of bPAC and the potential effects on viability (Fig. 3A). Cells were stimulated with light every 12 h for 2 days. The rate of insulin release for bPAC-expressing cells exposed to light at 2.5 mM glucose remained constant and higher than for control cultures. After 36 h, insulin release was similar to that of the control cells most likely because of the 'dilution effect' caused by cell proliferation on the expressed transgene that was delivered with an adenoviral vector. Indeed, proliferation did not differ among non-transduced cells and bPAC-expressing cells stimulated with light or kept in the dark (Fig. 3B). Even the fractions of viable and apoptotic cells in these populations were similar after four repeated stimulations with blue light (Fig. 3C,D) suggesting that there were no adverse effects of cAMP modulation on the cell viability and proliferation.
Light-triggered enhancement of insulin release in murine islets and MIN6 β-cell pseudoislets expressing bPAC. In addition to β-cells lines, which represent relatively homogeneous β-cell populations, we determined the effect of bPAC activation in murine islets. Islets expressing bPAC and exposed to blue light for 30 min exhibited higher [cAMP] i both at 2.8 mM (1.82 ± 0.36 ng cAMP/mg total protein) and 20.2 mM (21.66 ± 5 ng cAMP/mg total protein) of glucose compared to islets not carrying the bPAC gene (0.28 ± 0.08 (2.8 mM) and 2.13 ± 0.62 (20.2 mM) ng cAMP/mg total protein; Fig. 4A). This translated to augmented insulin secretion for islets transduced with the AdbPAC (Fig. 4B). The insulin secretion rate of primary islets at 20.2 mM glucose was  Results are shown as mean ± SD from at least three independent experiments. almost 5-fold higher than the corresponding rate at 2.8 mM glucose. This was not different (~4.5-fold) for islets transduced with AdGFP but was higher than for AdbPAC-expressing islets, which displayed greater insulin secretion rates both at 20.2 mM and 2.8 mM of glucose, i.e. 19.14 and 10.83 ng insulin/islet/h (p = 6.7 × 10 −3 , n = 3), respectively. The higher hormone release by AdbPAC-transduced islets vs. control islets at 2.8 mM was aligned with the higher [cAMP] i (Fig. 4A) similar to our observations with MIN6 β-cells described above. These findings also underline the amplifying role of cAMP on insulin secretion in the presence of glucose.
Dispersed islet cells including primary and insulinoma β-cells re-organize into aggregates or 'pseudoislets' (PIs) exhibiting ultrastructural characteristics of native islets and higher GSIS 27, 28 than monolayer cultures. We have shown that insulinoma PIs, which can be formed either in static or stirred-suspension cultures, respond to glucose and secretagogues with enhanced insulin secretion 29 . Thus, we questioned if this response could be further enhanced with photostimulation of PIs comprising β-cells expressing bPAC. To address this, MIN6 β-cells transduced with AdbPAC were cultured to form PIs (Fig. 4C). These PIs showed greater rates of insulin release upon illumination than PIs of control MIN6 β-cells both at 2.5 mM (439.32 ± 66.43 vs. 267.47 ± 41.05 ng insulin/ mg total protein/h, p = 0.018, n = 3) and 25 mM of glucose (1048.82 ± 75.3 vs. 612.37 ± 31 ng insulin/mg total protein/h, p = 7.5 × 10 −4 , n = 3) (Fig. 4D). Of note, the ratio of rates at 25 mM over 2.5 mM of glucose was similar for PIs with (~2.4) or without (~2.3) the expression of bPAC analogous to our observations in monolayer cultures. The enhanced secretion of insulin at either glucose level by PIs of bPAC β-cells exposed to light was not different than that by PIs treated with IBMX, tolbutamide or KIC (Fig. 4D) but were lower compared to that of PIs incubated with forskolin (2.5 mM: 796.64 ± 148.92, p = 0.019; 25 mM: 1652.52 ± 264.88 ng insulin/mg total protein/h, p = 0.019, n = 3). Hence, insulin secretion can be modulated optogenetically in primary islets and β-cell PIs. We found in the latter that the enhancement achieved through blue light illumination is equivalent to that gained by treating PIs with known chemical secretagogues.

Optogenetic regulation of insulin release is Ca
to extracellular glucose elicits events leading to plasma membrane depolarization and opening of the L-type voltage-dependent Ca 2+ channels that mediate ion influx, subsequent fusion of secretory vesicles and the secretion of the hormone. In fact, agents (e.g. KCl) that elevate [Ca 2+ ] i in β-cells can trigger insulin secretion even in the absence of glucose. We thus wondered if the Ca +2 dependence is maintained during insulin secretion in our system given that photoactivation of bPAC induces an increase in [cAMP] i . Released insulin was measured in monolayer or PI cultures of bPAC-expressing cells upon incubation with nifedipine, a blocker of L-type Ca +2 channels (Fig. 5A). In monolayers, the inclusion of the Ca 2+ -channel blocker reduced the secretion rate by 47% and 40% at 2.5 and 25 mM glucose, respectively. Nifedipine treatment led to a similar decrease in PIs at 2.5 mM (45%) but the decline was less pronounced at 25 mM (27%). We also obtained evidence of the light-triggered changes in [Ca 2+ ] i by analyzing ion influx after incubation with CaSiR-1. A greater increase in [Ca 2+ ] i was noted in cells irradiated with blue light than in those cells kept in dark (Fig. 5B,C). Therefore, the insulin secretion is

Discussion
Exogenous regulation of [cAMP] i , which is an amplifier rather than an inducer of insulin secretion, is a promising avenue for enhancing β-cell function particularly for clinical applications.
[cAMP] i levels are dictated mainly by cAMP synthesis and degradation by ACs and PDEs, respectively. In vivo AC activation with incretins to elevate [cAMP] i is hampered by their rapid clearance by dipeptidylpeptidase-4 (DPP4) and PDE inhibition is challenging because islet cells express multiple PDE types 30,31 . Here, we describe the optogenetic regulation of [cAMP] i and thus of insulin secretion in β-cells. To our knowledge, control of insulin release via optogenetic modulation of [cAMP] i using a photoactivatable AC has not been demonstrated to date. In recent reports, ectopic expression of ChRs in MIN6 β-cells and islets led to changes in hormone secretion through light-triggered variations in Ca 2+ influx 9,10 . Interestingly, illumination induced the release of insulin by these cells in a manner that was independent of glucose metabolism. An almost 40% increase in insulin was noted after ChR stimulation in glucose-free medium 9 , and secretion was enhanced at low but not at high glucose concentrations 9,10 . Moreover, ChRs require a 'co-factor' (retinal) for light-triggered activity, and the sequence of processes involving ChR-mediated [Ca 2+ ] i increase, activation of gene transcription (e.g. GLP-1 8 ) and translation may result in a slower overall response. Furthermore, high-fat diet and fatty acids enhance [Ca 2+ ] i 32, 33 making plausible the hypothesis that an exaggerated Ca 2+ response could contribute to the initial hyperinsulinemia seen in T2D. This could confound the prospective regulation of Ca 2+ signaling as a way to potentiate insulin production. Hence, targeting cAMP directly may be advantageous because its increase either by treatment with agents such as forskolin and IBMX or through bPAC activation with light amplifies GSIS. We observed that illumination of bPAC-expressing β-cells (MIN6, βTC) and primary islets resulted in increased secreted insulin even at low glucose concentrations but not in glucose-free medium. The observed response is aligned with the role of cAMP as an amplifier of insulin release in the presence of glucose. Indeed, no change in basal insulin release by MIN6 cells was noted in buffer devoid of glucose despite a raise in [cAMP] i due to bPAC activation. Therefore, this mode of amplification of the secretory function does not abolish control of glucose on insulin release in β-cells.
The augmented insulin production due to [cAMP] i elevation also points to a reduced number of β-cells or islets required -for instance, in a transplantation setting -to generate sufficient hormone amounts for glucose homeostasis. In fact, the secretion rate of bPAC-expressing β-cell monolayers or PIs was enhanced evenly and proportionally at 2.5 mM and 25 mM maintaining approximately the same constant ratio (2.2-2.5) as regular MIN6 cell monolayers or PIs. Such even increase in insulin secretion was not observed in primary islets. This insulin secretion ratio was lower after photoactivation of bPAC (~1.8) in primary islets compared to non-transduced islets (~5). This can be attributed to the pronounced enhancement of insulin release at 2.8 mM glucose with bPAC activation. Other factors however, may also need to be factored in such as a potential upper limit in the release of insulin at high glucose despite the significantly elevated [cAMP] i and the presence of other endocrine cells, which are also exposed to AdbPAC upon transduction of the islets.
When exposed to blue light for 30 minutes every 12 hours, AdbPAC-transduced MIN6 β-cells exhibited consistently higher insulin secretion over 36 hours but a slight decline was noted subsequently. Lower viability and enhanced apoptosis were ruled out as reasons for this decline. The reduction in insulin secretion, which was still on par with that from non-transduced MIN6 β-cells, was most likely due to a 'dilution effect' associated with the continuous cell proliferation and transient transgene expression. This issue can be addressed with the generation of β-cell lines stably expressing genes encoding for photoactivatable ACs.
The system also exhibited favorable kinetics with the [cAMP] i increasing within 5 minutes of photoactivation. This timeframe is comparable with that of [cAMP] i increase in murine β-cells within intact pancreatic islets following elevation of glucose concentration from 3 to 20 mM 24 . The [cAMP] i returned to a stable level within 8 minutes after illumination was terminated but was marginally higher than the baseline, i.e. for cells kept in dark. This did not affect the proliferation and viability of the cells over multiple cycles of stimulation. In fact, elevated [cAMP] i (e.g., after incretin stimulation) promotes β-cell proliferation through PKA-dependent mechanisms and suppresses apoptosis 34,35 partly due to the increased expression of the anti-apoptotic proteins Bcl-2 and Bcl-xL 36,37 .
Further studies are warranted about the potential long-term effects of optogenetic modulation of [cAMP] i on the β-cells.
To that end, β-cells engineered to express bPAC can be used as a tool for studying the role of [cAMP] i and its interaction with other signaling pathways in normal islets and those from diabetes patients. For instance, a bidirectional relation between [cAMP] i and [Ca 2+ ] i fluctuations has been hypothesized although the exact mechanism remains unclear 38 . Also, maladaptation of cAMP signaling has been reported in several animal models of diabetes including decreased glucose-induced production of [cAMP] i 39 and lower expression of PDEs 40 . In vitro, prolonged culture of β-cells in high concentrations of glucose results in alterations in [cAMP] i activity 41 . Extensive β-cell juxtaposition seen in PIs increases cAMP responsiveness 42 resulting in augmented GSIS compared to monolayer β-cell cultures as we and others have observed. These observations raise important questions, many of which can be potentially addressed using an optogenetic platform such as the one described here.
In addition to their value as a research tool, bPAC-expressing β-cells can be utilized, for example, as PIs, in a retrievable encapsulation device. This could be integrated into an autonomous bioartificial pancreas device with a low-power light source and existing glucose sensors (e.g. a wireless sensor fitted in a contact lens 43 ) to control illumination. The blue light's relatively poor tissue penetration is not expected to hinder the efficient stimulation of cells in this setup. Yet, the sensitivity of PACs can be shifted toward the infrared spectrum through protein engineering 44 . Additionally, available human β-cell lines 45,46 which are clinically more relevant than their rodent counterparts, can be modified for optogenetic regulation of cAMP. Beyond β-cell lines, human pluripotent stem cell (hPSC)-derived insulin-secreting cells may also be engineered to express PACs aiming at correcting defective GSIS, which is a major challenge in the generation of functional β-cells from hPSCs. Overall, optogenetic approaches such as the one presented here are expected to enable technologies which will enhance our understanding of β-cell physiology and contribute to biomedical solutions for pancreas pathologies.
Murine islet isolation and culture. An abdominal incision was performed to 8-week old male C57BL/6 mice under anesthesia to access the pancreas, which was inflated and harvested. Digestion was carried out with collagenase (CIzyme RI from Vitacyte, Indianapolis, IN). Gradient centrifugation and gravity sedimentation of islets were carried out in lymphocyte separation medium (Corning) and Hank's balanced salt solution (ThermoFisher). Islets were handpicked under a microscope, counted and cultured in RPMI 1640 (MP Biomedicals, Santa Ana, CA) supplied with 10% FBS (ThermoFisher) in Petri dishes at 37 °C and 5% CO 2 . Intact islets underwent two rounds of adenoviral transduction (MOI as stated, ~1,500 cells/islet) 48 h apart. bPAC gene and adenovirus construction. A human-codon optimized bPAC gene was synthesized (GeneArt, ThermoFisher) from the available bPAC sequence (accession number: GU461306.2) 21 . A cMyc-derived epitope (myc tag: EQKLISEEDL) was added to the C-terminus of bPAC followed by an internal ribosomal entry site (IRES) sequence, the mCherry gene and the SV40 polyA tail. The bPAC-cMyc-IRES-mCherry cassette was inserted into the pShuttle-CMV vector for generation of an adenovirus (AdbPAC) in 293AD cells with the AdEasy system (Agilent Technologies, Santa Clara, CA) 49 . The AdGFP adenovirus carrying the GFP gene under the CMV promoter was obtained from the Baylor College of Medicine Vector Development Core (Houston, TX). Purification and titer determination of the adenoviral particles were carried out using the ViraBind and QuickTiter kits (Cell Biolabs), respectively. For AdbPAC propagation, 293AD cells were transduced at 70% confluence.
Arduino-driven LED arrays for blue light stimulation. Arrays were constructed with 6 or 12 LEDs (470 nm; Mouser Electronics, Mansfield, TX) for stimulation of cells cultured in 6-or 12-well plates. For control of its operation, each array was connected to an Arduino board (Adafruit, New York, NY) that was programmed accordingly. In this configuration, the light power on the culture surface was 0.18 mW/mm 2 measured with a S120C sensor connected to PM100D light meter (Thorlabs Inc., Newton, NJ). For pulsed stimulation the light cycle was 5 sec on/10 sec off.
[cAMP] i and secreted insulin assays. [cAMP] 29 . Insulin and [cAMP] i measurements were normalized with the total protein content of the corresponding cell lysates determined via the Bradford method (ThermoFisher). Of note, insulin secretion for primary islets was normalized with respect to the number of islets utilized.
Experiments involving multiple cycles of stimulation with light, were initiated 48 h post-transduction. For this purpose, the following procedure was repeated every 12 h: The culture medium in all wells of 12-well plates was replaced with KRB containing 2.5 mM glucose and cells were incubated for 2 h. Then, KRB was removed and fresh KRB with 2.5 mM glucose was added for 30 min along with or without blue light illumination. Subsequently, supernatants and cells were collected from at least 3 wells for determination of secreted insulin and total protein (as described above). The buffer in the remaining wells was replaced with fresh culture medium and the procedure was repeated 12 h later for up to 2 days.
In these experiments, cell numbers and the fractions of live (Trypan Blue; see above) and apoptotic cells (TUNEL assay; see below) were determined from samples collected before the exchange of medium with KRB. TUNEL assay. DNA fragmentation-related apoptosis was assessed by terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) using the FlowTACS apoptosis detection kit (Trevigen) according to the manufacturer's instructions as we reported 29 . Briefly, cells in aliquots were fixed with 3.7% formaldehyde for 10 min and stained with 10 μg/ml propidium iodide in PBS for 5 min. Then, the samples were permeabilized with cytonin for 30 min, labeled with terminal deoxynucleotidyl transferase (TdT) for 40 min at 37 °C and incubated with a streptavidin-FITC antibody in the dark for 10 min. After two washes with PBS, the samples were analyzed via flow cytometry. Appropriate gating was applied to screen out cellular debris and cell clumps. Positive (cells treated with nuclease) and negative (cells without TdT primary antibody incubation) controls were used to set up bins for apoptotic, and non-apoptotic ranges.
Immunocytochemistry. Cells were fixed with 4% paraformaldehyde (Sigma) in PBS for 20 min, permeabilized/blocked in PBS with 0.1% Triton X-100 (Mallinckrodt Baker, Phillipsburg, NJ) and 1% bovine serum albumin (BSA; Sigma) for 30 min and incubated overnight at 4 °C with a rabbit antibody against the myc epitope (cat. no. 2272, Cell Signaling Technology, Danvers, MA). After three washes with PBS, cells were incubated with a donkey anti-rabbit secondary antibody conjugated to Alexa 488 (Jackson Immunoresearch Laboratories Inc., West Grove, PA) for 1 h at room temperature. Nuclear DNA was stained with DAPI (Vectashield, Vector Laboratories, Burlingame, CA). Images were acquired with a Leica TCS SPE confocal microscope (Leica Microsystems Inc., Buffalo Grove, IL) and quantitative analysis was performed with the Leica LAS X software. Images from at least 10 separate fields per experiment were used for analysis.
Western blot analysis. Cells were lysed in radioimmunoprecipitation (RIPA) buffer supplemented with a protease inhibitor cocktail (EMD Biosciences, CA) for 30 min and total protein was determined (Bradford; ThermoFisher). Cell lysates were boiled for 5 min at 95 °C, loaded to a sodium dodecyl sulfate-polyacrylamide gel (30 μg of total protein/lane) and after gel electrophoresis and protein transfer to polyvinylidene difluoride (PVDF) membranes (Thermo Fisher Scientific), the membranes were blocked with 5% milk in Tris-buffered saline with 0.05% Tween-20 (TBS/T) for 1 h at room temperature. After 3 washes with TBS/T, the membranes were incubated with primary antibodies against myc-tag (Cell Signaling Technology) and β-actin (cat. no. A1978, Sigma) in a shaker at 4 °C overnight. Following 3 more TBS/T washes, secondary horseradish peroxidase (HRP)-conjugated antibodies (Jackson ImmunoResearch Laboratories Inc.) were added for 1 h at room temperature. Membranes were washed 3 times and proteins were detected in a C-DiGit blot scanner after adding WesternSure ECL substrate (Li-Cor Biotechnology, Lincoln, NE). Intracellular Ca 2+ measurement. Cells were washed once with Hank's Balanced Salt Solution (HBSS) and loaded with the calcium indicator CaSiR-1 acetoxymethyl ester (Goryo Chemical Inc., Sapporo, Japan) (5 μM final concentration) at 37 °C for 20 min in the dark 51 according the manufacturer's instructions. The loading solution was then replaced with KRB containing 2.5 mM or 25 mM glucose. Cells were placed on a Leica TCS SPE confocal microscope (Leica Microsystems Inc., Buffalo Grove, IL) and irradiated with a 488-nm laser light for 20 s immediately followed by the recording of fluorescence intensity in the range of 650-750 nm (emitted light) after excitation with a far-red (635 nm) laser light. Cells in each field were selected based on mCherry fluorescence (excitation with a 532-nm laser) and analyzed using the Leica LAS X software. Between 20-30 cells were analyzed per sample. A region of interest was defined for each cell and the average fluorescent intensity was calculated. These intensities were averaged over all mCherry-positive cells. Average intracellular Ca 2+ levels were quantified as the change in average fluorescence intensity at different time points above the initial fluorescence (ΔF = F − F o ) divided by the initial intensity (F o ). The area under the curve (AUC) was calculated from temporal data of ΔF/F o via a script in MATLAB (Mathworks, Natick, MA).

Statistical analysis.
Results are expressed as mean ± standard deviation (SD) from at least 3 independent experiments with each experimental point analyzed in triplicates. Analysis of variance and the posthoc Tukey test were performed using Prism 7 (GraphPad Software, Inc., La Jolla, CA). Values of p < 0.05 were considered as significant.