Reactive oxygen and nitrogen species disturb Ca²⁺ oscillations in insulin-secreting MIN6 β-cells

Abstract

Disturbances in pulsatile insulin secretion and Ca²⁺ oscillations in pancreatic β-cells are early markers of diabetes, but the underlying mechanisms are still incompletely understood. Reactive oxygen/nitrogen species (ROS/RNS) are implicated in reduced β-cell function, and ROS/RNS target several Ca²⁺ pumps and channels. Thus, we hypothesized that ROS/RNS could disturb Ca²⁺ oscillations and downstream insulin pulsatility. We show that ROS/RNS production by photoactivation of aluminum phthalocyanine chloride (AlClPc) abolish or accelerate Ca²⁺ oscillations in the MIN6 β-cell line, depending on the amount of ROS/RNS. Application of the sarcoplasmic/endoplasmic reticulum Ca²⁺ ATPase (SERCA) inhibitor thapsigargin modifies the Ca²⁺ response to high concentrations of ROS/RNS. Further, thapsigargin produces effects that resemble those elicited by moderate ROS/RNS production. These results indicate that ROS/RNS interfere with endoplasmic reticulum Ca²⁺ handling. This idea is supported by theoretical studies using a mathematical model of Ca²⁺ handling adapted to MIN6 cells. Our results suggest a putative link between ROS/RNS and disturbed pulsatile insulin secretion.

Abbreviations

AlClPc

Aluminum Phthalocyanine Chloride

CICR

calcium-induced calcium release

CuFL

selective turn-on fluorescent NO reporter formed by a Cu(II) complex of a fluorescein modified with an appended metal-chelating ligand (FL)

ER

endoplasmic reticulum

IP₃R

inositol triphospate receptor

MitoSOX

mitochondrial superoxide indicator

NO

nitric oxide

PMCA

plasma membrane Ca²⁺ ATPase

RNS

reactive nitrogen species

ROS

reactive oxygen species

RyR

ryanodine receptor

SERCA

Sarcoplasmic/endoplasmic reticulum Ca²⁺ ATPase

TEA

tetraethylammonium

Tg

thapsigargin.

Introduction

Insulin is secreted from the pancreatic β-cells in distinct pulses^1,2 and disturbed pulsatile insulin release has been suggested to be an early marker of diabetes.³ These bursts of insulin are driven by underlying oscillations in electrical activity and cytosolic Ca²⁺ levels, which periodically trigger Ca²⁺-regulated exocytosis.⁴ Importantly, secretory pulses are more effective than constant insulin levels in lowering plasma glucose levels.^3,5,6 In particular, hepatic signaling is highly sensitive to whether insulin is released in pulses,⁷ leading to the suggestion that insulin resistance in the liver is secondary to disturbed pulsatile insulin release.⁸ Thus, a deeper understanding of oscillatory β-cell activity and its disturbance has clinical relevance.

Disturbed insulin secretion and diabetes have been associated with β-cell damage caused by reactive oxygen species (ROS).⁹ However, ROS have also been shown to exert a stimulating short-term effect on insulin release.¹⁰ Hence, ROS may act as a double-edged sword: moderate and short-term elevations play a positive signaling role augmenting insulin secretion, but prolonged exposure to elevated ROS levels, characteristic of oxidative stress, are detrimental for the β-cells.^9,11 Similarly, reactive nitrogen species (RNS), such as nitric oxide (NO), have been shown to elevate Ca²⁺ levels and stimulate insulin secretion at low concentrations,¹² whereas high NO concentrations exert an inhibitory effect on β-cell function¹² and survival.¹³

ROS/RNS target various cellular mechanisms underlying insulin secretion,⁹ and in particular components involved in Ca²⁺ handling.^14,15 For example ROS/RNS activate voltage-gated Ca²⁺ channels and endoplasmic reticulum (ER) release channels, such as the inositol triphospate receptors (IP₃Rs) and ryanodine receptors (RyRs). In addition, sarcoplasmic/endoplasmic reticulum Ca²⁺ ATPases (SERCAs) and plasma membrane Ca²⁺ ATPases (PMCAs) are inhibited by ROS/RNS. All these effects contribute to elevating the cytosolic Ca²⁺ concentration, which may trigger cell dysfunction.¹⁶ Moreover, ROS/RNS effects on ER Ca²⁺ handling could lead to ER stress and consequent protein misfolding.¹⁷ In addition, and of particular interest for the present study, ROS/RNS effects on the Ca²⁺ handling machinery might disturb Ca²⁺ oscillations and downstream pulsatile insulin release.

Aluminum phthalocyanine chloride (AlClPc) is a photosensitizer commonly used in photodynamic therapy experiments and clinical protocols; photoactivation of AlClPc induces chemical changes into neighboring molecules, which lead to ROS production,¹⁸ as well as NO/RNS release.^19-21 Here we show that ROS/RNS production by AlClPc photoactivation accelerates or abolishes Ca²⁺ oscillations in the widely used insulin-secreting MIN6 β-cell line. To guide the interpretation of our results, we used a mathematical model of Ca²⁺ handling adapted to MIN6 cells. The model results support the notion that ROS/RNS effects on ER Ca²⁺ are insufficient to interrupt Ca²⁺ oscillations, and that inhibition of PMCA by high ROS/RNS levels could underlie the cessation of the periodic Ca²⁺ pattern seen at maximal AlClPc photoactivation.

Materials and Methods

Experiments

Cells

MIN6 β-cells (²²; passages 19–40; a gift from Prof. L. Eliasson, LUDC, Malmö, Sweden) were grown in Dulbecco’s modified Eagle’s medium (DMEM) High glucose (4 g/L – 23 mM) and L-glutamine (0.58 g/L – 4 mM), supplemented with 15% heat-inactivated Fetal Bovine Serum (FBS), 75 mg/L penicillin, 50 mg/L streptomycin and maintained in a humidified incubator with 95% air and 5% CO2 at 37°C. Medium was changed every 2–3 d and cells were passed every 5–7 d when they were ∼60% confluent.^22,23

Imaging

Culture media, compounds and chemicals were all from Sigma Aldrich, except Fura-2 AM (Life Technologies, F-1221), MitoSOX (Life Technologies, M36008) and CuFl (see below).

Experiments were conducted at room temperature under constant flow perfusion with extracellular medium (ECM) at pH 7.4 (adjusted with NaOH). The ECM was composed of (in mM): 150 NaCl, 5 KCl, 1 MgCl, 10 Hepes, 2 NaPyr, 2 CaCl₂, and either 3 or 25 d-Glucose. In most experiments 15 mM Tetra Ethyl Ammonium (TEA, a K⁺ channel blocker) Chloride was added to the ECM containing 25 mM d-Glucose. Where indicated, SERCA pumps were inhibited by the addition of 200 nM Thapsigargin.

Intracellular free Ca²⁺ concentration was tracked by fluorescence imaging using Fura-2 acetoxymethyl (AM) ester. MIN6 pseudoislets were placed on poly-L-lysine (0.1% for 2h) treated coverslips and incubated at 37°C in DMEM High Glucose (D5796) for a total time of 20 minutes with 5 μM Fura-2 AM, PluronicF-127 (0,01% w/v) and Sulfinpyrazione (250 μM). Pluronic (P2443) and Sulfinpyrazone (S9509) were used to prevent the complexation and secretion of dye. ROS production was obtained by photoactivation at 660 nm of AlClPc (363530). Min6 pseudoislets were incubated at 37°C in DMEM High Glucose for a total time of 1 hour with 10 μM AlClPc.

NO was monitored by fluorescence imaging using CuFL. The probe was prepared using the Nitric Oxide Sensor (Intracellular) Kit (Strem Chemicals, part No. 96–0293) by reacting the fluorescein-based ligand 2-{2-Chloro-6-hydroxy-5-[2-methylquinolin-8-ylamino)methyl]- 3-oxo-3H- xanthen-9-yl}benzoic acid (FL) with CuCL₂ in a 1:1 ratio in buffered aqueous solution. Min6 pseudoislets were incubated in ECM solution for a total time of 15 min at room temperature with 20 μM CuFl and then washed for 15 min with ECM solution.²⁴ ROS signaling was monitored with fluorescence imaging using MitoSOX. Min6 pseudoislets were incubated at 37°C in DMEM High glucose (D5796) without FBS for a total time of 15 minutes with 2.5 µM MitoSOX.

In experiments with Fura-2, specimens were excited alternately by 365 nm and 385 nm high power LEDs (Thorlabs, M365L2 - M385L2). In experiments with CuFl [24] (Lim et al., 2006), specimens were excited with 460 nm high efficacy 5W Dental Blue LED (Led Engine, LZ1–00DB00). In experiments with MitoSOX, specimens were excited with 530 nm high power LED (Thorlabs, M530L3). Phthalocyanine was photoactivated by 660 nm high power LED (Thorlabs, M660L3) filtered through a pair of polarizers to modulate the LED intensity. The fluorescence emission was selected by an emission filter (HQ 520/40, Chroma, Rockingham, VT) and the emitted fluorescence images were collected by a complementary metal-oxide semiconductor (CMOS) cooled camera (Pco.Edge) using a water immersion objective (40´, Lumplan FL, N.A.= 0.80; Olympus). To generate each frame, the 365 nm and 385 nm LEDs were activated in rapid sequence for 200 ms each and a fura-2 ratio image was generated every second. The 460 nm LED was activated for 100 ms and a CuFL intensity image was generated every second. The 530 nm LED was activated for 200 ms and a MitoSOX intensity image was generated every second. For pulsed-light experiments, the 660 nm LED was activated for 100 ms/pulse and 1 pulse/second.

Data analysis

Images were analyzed with software developed using the Matlab platform (Release 14, MathWorks, Inc., Natick, MA, USA). Ca²⁺ signals were measured as fura-2 emission ratio changes, ΔR = R(t) − R(0), where t is time, R(t) is fura-2 emission intensity excited at 365 nm divided by the intensity excited at 385 nm, and R(0) indicates pre–stimulus ratio.

Fura-2, CuFL and MitoSOX traces were generated by averaging pixel signals within regions of interest (ROIs) corresponding to individual cells. Traces were analyzed using an ad-hoc developed Matlab function that identifies local maxima. Function parameters were adjusted in order to discard noise fluctuations and retain only true Ca²⁺ oscillation peaks. The temporal occurrence of the peaks was used to calculate the time interval between 2 consecutive peaks (period). For each cell (ROI), peak period and amplitude values were averaged to obtain a single-cell value.

Statistical comparisons were done with unpaired t tests for samples with unequal variance.

Mathematical modeling

We used a mathematical model of oscillatory behavior in β-cells²⁵ to investigate the effects of TEA and modifications in ER Ca²⁺ handling on Ca²⁺ oscillations. The model reproduces many observed patterns in mouse β-cells, and was here adapted to MIN6 cells.

The model includes 4 voltage-gated currents: a Ca²⁺ current (I_Ca), a delayed rectifier K⁺ current (I_K), a Ca²⁺ - dependent K⁺ current (I_K(Ca)), and an ATP-sensitive K⁺ current (I_K(ATP)). The membrane potential V evolves in time according to Eq. A1. All currents are modeled as in.²⁵ The ratio ADP/ATP is allowed to vary (Eq. A3) according to evidence of ATP oscillations in MIN6 cells.²⁷ The model also includes Ca²⁺ dynamics in the cytosol and in the ER of MIN6 cells as described by Eq. A6. Influx into the ER, via SERCA pumps, and ER efflux are modeled according to Eq. A7. The ER Ca²⁺ release rate p_ER can be either constant, or an increasing function of cytosolic [Ca²⁺] (c) to simulate calcium-induced calcium release (CICR), which is operating in MIN6 cells.²⁸

The main mechanism resulting in oscillatory behavior in the model is negative feedback of Ca²⁺ onto the ATP concentration, for example because of increased ATP consumption by Ca²⁺ pumps. Lower ATP levels lead to K(ATP)-channel opening and, eventually, to cessation of electrical activity and Ca²⁺ influx. Calcium is then removed from the cytosol, allowing ATP levels to rise, resulting in reduced K(ATP)-current, reappearance of electrical activity and increasing cytosolic Ca²⁺ levels due to Ca²⁺ influx.

For completeness, all expressions and parameters are given in the Appendix. Simulations were performed in the XPPAUT software²⁶ using the cvode solver.

Results

Glucose and TEA concur to evoke Ca²⁺ oscillations in MIN6 cells by blocking excessive K⁺ currents

The MIN6 β-cell line²² is widely used to investigate mechanisms underlying insulin secretion due to its good glucose responsiveness and relatively easy experimental handling.^22,23,29,30 It is well known that MIN6 cells exhibit irregular, small-amplitude Ca²⁺ oscillations when exposed to glucose, and that the addition of the K⁺ channel antagonist tetraethylammonium (TEA) changes these fluctuations into large-amplitude and more regular oscillations (Fig. 1A; see also^23,31). The period of the induced oscillations was typically 20–30 sec, similar to published results for MIN6 cells^23,30,31 and so-called fast Ca²⁺ oscillations observed in mouse islets.^32,33

Figure 1.

The K⁺ channel antagonist TEA induces large amplitude Ca²⁺ oscillations. (A) Recording of Fura-2 fluorescence reporting Ca²⁺ concentrations in a MIN6 cell stimulated by glucose and TEA, as indicated. (B) Model simulation of the experiment in panel a). The increase in glucose from 3 mM to 25 mM was simulated by lowering g_KATP to 500 pS from 650 pS; TEA application was simulated by decreasing the K_v conductance (g_Kv) by ∼80 %, from 41 nS to 8 nS.

To obtain insight into this behavior, we used a mathematical model of oscillatory behavior in β-cells.²⁵ We assumed that, compared to pancreatic β cells in islets of Langerhans, MIN6 cells have larger K⁺ currents,³¹ which prevent large-amplitude oscillations. Our model suggests that this mechanism is plausible. Small-amplitude, high-frequency fluctuations were transformed into slower large-amplitude oscillations when the conductance of the K_v current was reduced by ∼80% (Fig. 1B; see also³¹).

Phthalocyanine photoactivation leads to ROS and RNS production and increases cytosolic Ca²⁺ levels

To investigate the short-term effects of ROS/RNS on oscillatory Ca²⁺ dynamics in MIN6 cells, we loaded the cells with AlClPc. When photoexcited at 660 nm, AlClPc produces ROS in amounts depending on the intensity of the LED illumination as revealed by the ROS sensor MitoSOX (Fig. 2A). LED illumination per se, i.e., in the absence of AlClPc, did not produce ROS, nor result in any artifactual MitoSOX signal (Fig. 2A). Control experiments with HyPer, another ROS sensor, confirmed these results (Fig. S1). In addition, AlClPc photoactivation increased NO levels,²¹ as reported by the selective fluorescent probe CuFL²¹ (Fig. 2B; see also¹⁹). We found that ROS/RNS production by photoactivation of AlClPc at maximal (100%) LED intensity abolished Ca²⁺ oscillations in MIN6 cells. The Ca²⁺ concentration showed a peak and then settled at an intermediate level (Fig. 2C). When SERCA pumps were inhibited by thapsigargin (Tg) to prevent ER Ca²⁺ filling (Fig. 2D), the Ca²⁺ peak was substantially reduced (Ratio of peak to oscillation amplitude; ctrl: 0.98 ± 0.04 (n = 38), Tg: 0.19 ± 0.02 (n = 62), p<0.001, Fig. 2F). Control experiments in the absence of AlClPc showed that maximal LED illumination did not per se influence the Ca²⁺ oscillations (Fig. 2E). These observations suggest that the surge of cytosolic Ca²⁺ in response to ROS/RNS production is caused primarily by release of Ca²⁺ from the ER. Indeed, ROS/RNS are known to inhibit SERCA pumps and activate RyRs and IP₃Rs.^14,15

Figure 2.

Photoactivation of phthalocyanine (AlClPc) produces ROS and RNS that interfere with Ca²⁺ oscillations in MIN6 cells. (A) ROS production by AlClPc photoactivation measured with the superoxide sensor MitoSOX (illumination from t = 60 s). The LED intensity was 100% (dashed curve) or 10% (grey curve). The full, black curve shows the ROS response to 100% LED illumination in the absence of AlClPc. (B) NO levels monitored with the sensor CuFL (illumination from t = 30 s) at 30% (full, black), 70% (grey) or 100% (dashed) LED intensity. Curves in panels a) and b) show the average signal, error bars indicate standard errors (n = 10-69 cells in 2-5 experiments per group). (C) Photoactivation of AlClPc with 100% LED intensity abolishes Ca²⁺ oscillations and results in a peak of [Ca²⁺]_c. (D) In the presence of 200 nM Tg, the Ca²⁺ peak in response to 100% LED photoactivation is reduced. (E) In the absence of AlClPc, 100% LED illumination did not interfere with Ca²⁺ oscillations (representative of 37 cells in 4 experiments). (F) Summary of the Fura-2 (Ca²⁺) peak normalized to oscillation amplitude in cells in the absence (ctrl, n = 38) and presence of Tg (n = 62).

Moderate levels of ROS/RNS modify Ca²⁺ oscillations

Having established that massive ROS/RNS production by AlClPc photoactivation abolishes Ca²⁺ oscillations in MIN6 cells, we turned our attention to the effects of moderate ROS/RNS levels. When photoactivating AlClPc with 10% LED intensity we observed that Ca²⁺ oscillations typically persisted, but with a modified profile.

Under control conditions, a step from 3 mM glucose to 25 mM glucose plus 15 mM TEA elicited a biphasic response consisting of a 1st phase lasting ∼4–5 min with rapid oscillations followed by a 2nd phase typically with slower oscillations (Fig. 3A). In contrast, the oscillations typically did not slow down when the second phase was perturbed by application of Tg (Fig. 3B) or photoactivation of AlClPc at 10% LED intensity (Fig. 3C) at t = 240 sec. Compared to control conditions, Tg or AlClPc photoactivation at 10% LED intensity resulted in a heterogeneous population of Ca²⁺ amplitude responses, which in some cells decreased substantially but in other showed behavior similar to the control patterns. This heterogeneity was observed when studying cells in different experiments (Fig. 3B) as well as for synchronized cells in the same experiment (Fig. 3C).

Figure 3.

Thapsigargin (Tg) and moderate levels of ROS/RNS accelerate Ca²⁺ oscillations. (A) Example of Ca²⁺ trace in a MIN6 cell stimulated with 25 mM glucose and 15 mM TEA. (B) SERCA inhibition by Tg accelerates Ca²⁺ oscillations, and often reduces their amplitude (black trace). Two cells from 2 independent experiments with different amplitude responses are shown. (C) Moderate ROS/RNS production by 10% LED photoactivation of AlClPc result in modifications in Ca²⁺ oscillations similar to Tg application. Two synchronized cell from a single experiments with different amplitude responses are shown. (D) Summary of 2nd phase (after 240 sec) Ca²⁺ oscillation periods normalized to 1st phase (before 240 sec) periods in control conditions (n = 50), or in the presence of Tg (n = 102) or moderate ROS/RNS levels (n = 50) during the 2nd phase. (E) Summary of 2nd phase (after 240 sec) Ca²⁺ oscillation Fura-2 amplitudes normalized to 1st phase (before 240 sec) amplitudes in control conditions (n = 50), or in the presence of Tg (n = 102) or moderate ROS/RNS levels (n = 50) during the 2nd phase.

To handle cell heterogeneity, we normalized the single-cell 2nd phase oscillations period and amplitude to the 1st phase period and amplitude, respectively, of the same cell. This analysis revealed that the 2nd phase period increased by ∼70% (ratio of 2nd to 1st phase periods 1.69 ± 0.14) under control conditions (Fig. 3D). In contrast, the period was unchanged when Tg or 10% LED photoactivation of AlClPc were present during the second phase (ratio of periods: 0.93±0.05 and 0.99±0.12, respectively; in both cases p< 0.001 vs. Control; p=0.22 Tg vs. 10% LED; Fig. 3D). The amplitude of the Fura-2 ratio signal was lower in the 2nd than in the 1st phase under control conditions (ratio of amplitudes 0.76±0.12). The reduction in 2^nd phase amplitude was greater in the presence of Tg or with 10% LED AlClPc photoactivation (ratio of amplitudes: 0.53±0.02 and 0.49±0.03, respectively; in both cases p<0.001 vs. control; p=0.20 Tg vs. 10%LED; Fig. 3E). However, there was substantial cell-to-cell heterogeneity in the response to both Tg and to 10% LED AlClPc photoactivation, in particular with respect to the relative amplitude, which in many cases did not decrease more than in the control case (Fig. 3B, C, E). Nonetheless, the fact that Tg and 10% LED AlClPc photoactivation accelerate the Ca²⁺ oscillations and reduce their amplitude to a similar degree, suggest that their mechanism of action might be shared, i.e., inhibiting SERCA pumping, a well-known target of ROS/RNS,^14,15 and in accordance with our findings using 100% LED activation (Fig. 2).

Mathematical modeling suggests that modified ER Ca²⁺ handling and inhibition of PMCAs by ROS/RNS underlie acceleration and disappearance of Ca²⁺ oscillations

Altogether, the results described above indicate that ROS/RNS production by AlClPc and SERCA inhibition by Tg accelerate Ca²⁺ oscillations to a similar degree (Fig. 3). Moreover, the presence of Tg results in a reduction in the Ca²⁺ peak seen in response to high ROS/RNS levels (Fig. 2). These observations suggest that ROS/RNS modify Ca²⁺ oscillations via a reduction of ER Ca²⁺ levels. To test whether this interpretation is compatible with known β-cell Ca²⁺ handling mechanisms, we performed simulations with the model used to obtain insight into TEA-induced large-amplitude oscillations (Fig. 1). Consistent with our working hypothesis, the oscillatory behavior of the mathematical model accelerated when SERCA activity was lowered (Fig. 4A; see also²⁵). Similarly, increasing Ca²⁺ release from the ER in the model (mimicking ROS/RNS activation of e.g. RyRs and IP3Rs;^14,15) increased the frequency of the Ca²⁺ oscillations (Fig. 4B). However, the simulated amplitude was nearly unchanged by the presence of Tg or increased Ca²⁺ release. We note that some MIN6 cells showed amplitude responses to Tg or 10% LED AlClPc photoactivation that were similar to the control case (Fig. 3). Hence, these model simulations are more similar to this subgroup of MIN6 cells with respect to the effect on oscillation amplitude.

Figure 4.

Mathematical modeling supports the experimental findings. (A) Model simulation showing acceleration of Ca²⁺ oscillations in response to Tg (k_SERCA=0 ms⁻¹ as indicated by the bar “Tg”). (B) As in panel (A) but with increased ER Ca²⁺ release (p_leak multiplied by a factor 10, indicated by the bar). (C) Simulation of the combined effect of inhibition of PMCAs (k_PMCA lowered from 0.20 to 0.12 ms⁻¹), Tg application (as in panel A) and increased ER leak (as in panel B), which abolish Ca²⁺ oscillations as seen experimentally (Figure 2B). d) As in panel c), but with Tg present throughout as in Figure 2C, which removes the Ca²⁺ peak when ER leak and PMCA inhibition is simulated.

Of interest, the simulated modifications of ER Ca²⁺ handling were insufficient to abolish Ca²⁺ oscillations in the model. This result suggests that high levels of ROS/RNS, as produced by 100% LED AlClPc photoactivation (Fig. 2), also act on other components involved in the generation of rhythmic Ca²⁺ concentrations. For example, ROS/RNS are known to inhibit not only SERCA pumps, but also plasma membrane Ca²⁺ ATPases (PMCAs).^14,15 Our simulations confirmed that a reduction in the PMCA rate, together with reduced SERCA activity and increased ER Ca²⁺ leak, could abolish Ca²⁺ oscillations (Fig. 4C). In agreement with experiments (Fig. 2), the presence of Tg before increasing ROS/RNS levels reduced the Ca²⁺ peak in our simulations (Fig. 4D). Thus, the model suggests that moderate ROS/RNS levels mainly act on ER Ca²⁺ handling, while high ROS/RNS concentrations might also inhibit PMCAs and thus abolish Ca²⁺ oscillations.

Discussion

It has been proposed that reactive oxygen species (ROS) act as a double-edged sword in β-cells.¹¹ Similarly, reactive nitrogen species (RNS) have beneficial or detrimental effects on β-cell function depending on their levels.¹² In agreement with this tenet we show here that massive ROS/RNS production by photoactivation of AlClPc at maximal (100%) LED intensity abolished Ca²⁺ oscillations in MIN6 β-cells (Fig. 2), while moderate ROS/RNS production (10% LED) accelerated the periodic Ca²⁺ excursions (Fig. 3). Since regular Ca²⁺ oscillations are a hallmark of healthy β-cell function,⁴ our results suggest that massive ROS/RNS production lead to immediate β-cell dysfunction, whereas moderate ROS/RNS levels might have a slightly positive effect on the Ca²⁺ behavior since oscillations are more frequent. An interesting possibility is that dynamic control of ROS/RNS levels contributes to shaping physiological Ca²⁺ oscillations in β-cells.

Based on our experimental results, we suggest that moderate ROS/RNS levels could influence Ca²⁺ oscillations via inhibition of SERCA pumps and/or by increasing activity of ER release channels (IP3Rs/RyRs), all known targets of ROS/RNS.^14,15 The molecular mechanisms involved in ROS/RNS action include amino acid redox modifications; the typical targets are thiol groups of cysteine residues, albeit other amino acids such as arginine, lysine and methionine residues can also react with ROS/RNS.^14,34 Current hypotheses about the inactivation of PMCAs by ROS/RNS propose an alteration of PMCA Tyr589, Met622 and Met831 residues.¹⁴ SERCA activity has been shown to be inhibited by ROS/RNS modification of cysteine (and tyrosine) residues.¹⁴

Model simulations support this hypothesis. In addition, the model suggests that high amounts of ROS/RNS also act on Ca²⁺ handling components at the plasma membrane, e.g., by inhibiting PMCAs – another ROS/RNS target,^14,15 which abolishes Ca²⁺ oscillations. Interestingly, it has been reported that SERCA is more sensitive to ROS than PMCA³⁵ in line with our findings.

The limited knowledge of MIN6 cell electrophysiology prevents the construction of a bottom-up, data-driven model, in contrast to was has been accomplished for example for human β-cells.^36-38 We included calcium-induced calcium release (CICR) in the model, since MIN6 cells are known to have RyRs.³⁹ The reduced Ca²⁺ amplitude seen in experiments with Tg also indicates that CICR is playing a role in MIN6 cells. In contrast, in mouse islets Tg increases the amplitude of cytosolic Ca²⁺ oscillations⁴⁰ since the ER no longer takes up Ca²⁺. In MIN6 cells, the ER likely functions both as a sink (due to SERCA) and a source (due to CICR) when [Ca²⁺]_c increases. The varying strengths of these 2 mechanisms likely underlie the heterogeneity seen in the amplitude responses to Tg and moderate ROS/RNS levels (Fig. 3e). Indeed, augmenting the role of CICR in our model can lead to reduced Ca²⁺ amplitude in response to SERCA block (Fig. 5). Further, the conclusions drawn from Figure 4 remain valid also when CICR is active (Fig. 5).

Figure 5.

When CICR is playing a major role, causing a peak at the beginning of each Ca²⁺ rise, SERCA block (k_SERCA = 0 ms⁻¹ as indicated by the bar “Tg”) reduces Ca²⁺ oscillation amplitude. Simulations and panels as in Figure 4, but with K_cicr = 0.3 µM, p_min = 0.0018 ms⁻¹, p_max = 0.1 ms⁻¹, g_k(Ca) = 10 pS, and K_D = 0.2 µM. All other parameters were changed as described for Figure 4.

In summary, we show that ROS/RNS can abolish or accelerate Ca²⁺ oscillations in insulin secreting cells, which might have important consequences for the disturbances of oscillatory β-cell function seen in early stages of diabetes. Our experiments and model simulations suggest that moderate ROS/RNS levels act on ER Ca²⁺ handling mechanisms, whereas higher levels of ROS/RNS also target other components involved in creating oscillatory behavior in insulin secreting cells.

Appendix: Model equations and parameters

For completeness we report all expressions and parameters of the mathematical model here, for more details see Bertram and Sherman.²⁵ The equations of the model for the membrane potentials V is

$${\displaystyle \frac{dV}{dt}=-\frac{\left(I_{Ca}+I_K+I_{K(Ca)}+I_{K(ATP)}\right)}{C_m}}$$ (A1)

where C_m is the membrane capacitance. The currents are modeled as

$${\displaystyle \begin{array}{l}{I}_{Ca}=g_{Ca}{m}_{{}_{\infty }}(V)(V-V_{Ca})\\ I_K=g_{K}n(V-V_{Ca})\\ I_{K(Ca)}=g_{K(Ca)}\omega (V-V_K)\\ I_{K(ATP)}=g_{K(ATP)}(V-V_K)\end{array}}$$ (A2)

where the delayed rectifier activation, n, and the nucleotide ratio a are given by

$${\displaystyle \begin{array}{l}\frac{dn}{dt}=\frac{n_{\infty }(V)-n}{{\tau }_n}\\ \frac{da}{dt}=\frac{a_{\infty }(c)-a}{{\tau }_a}\end{array}}$$ (A3)

The equilibrium functions are described with Boltzmann functions,

$${\displaystyle \begin{array}{l}{m}_{\infty }(V)=\frac{1}{1+e^{\left(\frac{v_m-V}{s_m}\right)}}\\ n_{\infty }(V)=\frac{1}{1+e^{\left(\frac{v_n-V}{s_n}\right)}}\\ a_{\infty }(V)=\frac{1}{1+e^{\left(\frac{r-c}{s_a}\right)}}\end{array}}$$ (A4)

The variable ω is the fraction of K(Ca) channels activated by cytosolic Ca²⁺.

$${\displaystyle \omega =\frac{c^5}{c^5+k_D{}^5}}$$ (A5)

where k_D is the Ca²⁺ level giving half-maximal activation of the K(Ca) current. The differential equations for the cytosolic Ca²⁺ and the ER Ca²⁺ are

$${\displaystyle \begin{array}{l}\frac{dc}{dt}=f_{cyt}(J_{mem}+J_{ER})\\ \frac{d{c}_{ER}}{dt}=f_{er}(V_{cyt}/V_{ER})J_{ER}\end{array}}$$ (A6)

Here, J_mem is the Ca²⁺ flux through the plasma membrane, f_cyt is the fraction of free to total Ca²⁺, and f_ER is the fraction of free Ca²⁺ in the ER. J_ER is the total efflux from the ER. These fluxes are given by

$${\displaystyle \begin{array}{l}{J}_{mem}=-(\alpha I_{Ca}+k_{PMCA}c)\\ J_{ER}=J_{leak}-J_{SERCA}\\ J_{SERCA}=k_{SERCA}\\ J_{leak}=p_{ER}(c_{ER}-c)\end{array}}$$ (A7)

where α converts units of current to unit of flux, k_PMCA and k_SERCA are the rate constants for the plasma membrane Ca²⁺ ATPase pump (PMCA) and the ER membrane Ca²⁺ ATPase pump (SERCA), respectively. The release rate p_ER is

$${\displaystyle p_{ER}=p_{\min }+\frac{p_{\max }-p_{\min }}{1+{\left(\frac{K_{cicr}}{c}\right)}^2}}$$ (A8)

where the parameter K_cicr determines the amount of calcium-induced calcium release (CICR), which is absent for K_cicr = 0 µM. All parameters are given in Table A1.

Table A1.

Default parameter values of the mathematical model used to simulate MIN6 cells in the presence of 25 mM glucose and 15 mM TEA

Parameter			Parameter
gCa	1200	pS	sa	0.1	µM
VCa	25	mV	τa	50000	ms
vm	−20	mV	α	4.5 × 10−6	fA−1 µM ms−1
sm	12	mV	fcyt	0.01
gK	8000	pS	fer	0.01
VK	−75	mV	kPMCA	0.2	ms−1
vn	−16	mV	kSERCA	0.4	ms−1
sn	5	mV	Vcyt/Ver	5
τn	16	ms	pmin	5 × 10−4	ms−1
gK(Ca)	100	pS	pmax	5 × 10−3	ms−1
kD	0.3	µM	Kcicr	0.0	µM
gK(ATP)	500	pS	Cm	5300	fF
r	0.14	µM

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

We thank Prof. Lena Eliasson, Lund University Diabetes Center, Malmö, Sweden for the gift of MIN6 cells, Prof. Fabio Di Lisa, University of Padua, Italy, for the gift of HyPer, and Prof. Fabio Mammano, Venetian Institute of Molecular Medicine, Padua, Italy, for use of laboratory and microscopy facilities and technical assistance.

Authors' Contributions

SA performed experiments, analyzed data, prepared figures, drafted and corrected the manuscript; AT analyzed data, developed the mathematical model, performed simulations and prepared figures; MGP designed and supervised model development, simulations and experiments, analyzed data, and wrote the manuscript. All authors gave final approval for publication.

Supplemental Material

Supplemental data for this article can be accessed on the publisher’s website.

Funding

This work was supported by the University of Padova (PRAt 2012, and the Strategic Research Project 2012 “DYCENDI” to MGP).