Beta‐cell excitability and excitability‐driven diabetes in adult Zebrafish islets

Abstract Islet β‐cell membrane excitability is a well‐established regulator of mammalian insulin secretion, and defects in β‐cell excitability are linked to multiple forms of diabetes. Evolutionary conservation of islet excitability in lower organisms is largely unexplored. Here we show that adult zebrafish islet calcium levels rise in response to elevated extracellular [glucose], with similar concentration–response relationship to mammalian β‐cells. However, zebrafish islet calcium transients are nor well coupled, with a shallower glucose‐dependence of cytoplasmic calcium concentration. We have also generated transgenic zebrafish that conditionally express gain‐of‐function mutations in ATP‐sensitive K+ channels (KATP‐GOF) in β‐cells. Following induction, these fish become profoundly diabetic, paralleling features of mammalian diabetes resulting from equivalent mutations. KATP‐GOF fish become severely hyperglycemic, with slowed growth, and their islets lose glucose‐induced calcium responses. These results indicate that, although lacking tight cell‐cell coupling of intracellular Ca2+, adult zebrafish islets recapitulate similar excitability‐driven β‐cell glucose responsiveness to those in mammals, and exhibit profound susceptibility to diabetes as a result of inexcitability. While illustrating evolutionary conservation of islet excitability in lower vertebrates, these results also provide important validation of zebrafish as a suitable animal model in which to identify modulators of islet excitability and diabetes.


Introduction
Electrical activity is the essential trigger of insulin secretion in mammalian b-cells (Koster et al. 2006). At low plasma [glucose], ATP-sensitive potassium (K ATP ) channels are open, hyperpolarizing the cell membrane and keeping voltage-dependent calcium channels (VDCCs) closed, thereby inhibiting secretion. A rise in plasma glucose results in enhanced b-cell glycolysis and oxidative phosphorylation, and increases the [ATP]/ [ADP] ratio. This in turn results in closure of K ATP channels and, because of electrical coupling via gap junctions (Benninger et al. 2008), uniform depolarization of the b-cell syncytium and calcium influx through VDCCs, which triggers pulsatile insulin release. The essential role of this coupling is illustrated by the fact that gain-of function mutations in K ATP channels cause diabetes, whereas loss of channel function reciprocally causes hyperinsulinism and hypoglycemia (Remedi and Koster 2010).
K ATP channel genes are not present outside the vertebrates, and the evolutionary extent and origins of excitability-dependent insulin secretion, although wellelucidated in mammals, has not been well-studied even in lower vertebrates. This is important, both from an evolutionary perspective, and to provide novel insights to the process. We have recently shown that K ATP channels are expressed in b-cells within the zebrafish (Danio rerio) islet, that they are functionally similar to their mammalian orthologs, and that activation of these channels by the drug diazoxide can similarly alter glucose tolerance (Emfinger et al. 2017). Previous studies have suggested a role for K ATP channels in early islet responses to overnutrition: activation of K ATP , either pharmacologically or with inducible transgenes, increases b-cell growth in response to excess nutrients (Li et al. 2014). However, direct assessment of zebrafish islet function is lacking, and whether alterations in excitability can drive persistent changes in glucose control in zebrafish, remains unknown. We recently showed that intracellular [Ca 2+ ] is glucose-sensitive in embryonic zebrafish islets (Lorincz et al. 2018). Here we characterize the glucose-sensitivity of intracellular [Ca 2+ ] in adult zebrafish islets, and show that it is similar to mammals, although unlike in mammalian islets (Bavamian et al. 2007), b-cells appear to function as independent units, such that Ca 2+ transients are not well-coupled between b-cells. We further show that transgenic expression of K ATP -GOF mutations blocks glucose-dependent [Ca 2+ ] elevations, resulting in severe hyperglycemia, paralleling the consequences of b-cell inexcitability in mammals.

Ethical approval
All animal procedures were approved by the Washington University in St. Louis Instiutional Animal Care and Use Committee.
For each of the constructs above, transgenic fish were created as follows: 2 nL of injection solution containing 25 ng/lL of construct and 25 ng/lL of Tol2 transposase RNA were injected into AB zebrafish embryos at the single-cell stage. The pDestTol2CG2 vector contains eGFP expressed under the cardiac myosin light-chain promoter as a transgenesis marker, permitting detection of subsequent founders by visible green fluorescence in the heart. For heat shock induction in larvae, larvae were placed in 20 mL glass scintillation vials at 40-70 larvae/vial and heated at 37°C from day 1-5 pf in a water bath for 3 h/ day. For adult induction, fish were transferred to glass beakers (7 fish/500 mL) with air stones and placed in a 37°C water bath for 3 hr/day for 2-10 days.

Gene expression analyses
RNA and cDNA were prepared using Qiagen RNeasy mini kit and ThermoFisher High-Capacity cDNA reverse transcription kit respectively. Previously described protocols were used to isolate islet (Emfinger et al. 2017) and brain (Lopez-Ramirez et al. 2016) tissues. Heart tissue was collected with forceps. PCR was performed with previously published primers for connexin 35b (Carlisle and Ribera 2014).

Ex-vivo microscopy of adult zebrafish islet calcium
Islets were isolated as described (Emfinger et al. 2017). Glass-bottomed 35 mmol/L dishes (MatTeK) were coated with 1% agarose, and glass pipette tips were used to remove a section of agarose at the plate center, creating a well. Individual islets were transferred to wells and immersed in pH 7.4 Krebs Ringer's solution buffered with HEPES (KRBH) containing 2 mmol/L glucose. The KRBH base solution consisted of (in mmol/L): NaCl 114, KCl 4.7, MgSO 4 1.16, KH 2 PO 4 1.2, CaCl 2 2.5, NaHCO 3 5, and HEPES 20, with 0.1% BSA. Solutions of varying glucose concentrations were flowed into the plate chamber through lines running into and out of the chamber ( Fig. 2A). Bulk islet data were captured using a Zeiss Axiovert 200M microscope equipped with a Lambda DG-4 illumination system and EM-CCD camera and a Till photonics microscope with PolyChrome V monochromator and cooled CCD camera in the CIMED Live Cell Imaging Core (https://resea rch.wustl.edu/core-facilities/cmed-live-cell-imaging-core/). Time lapse images used 100 msec exposure at an interval of 500 msec. For single-cell comparisons, high resolution images were captured using a Nikon Spinning Disk confocal microscope (a motorized Nikon Ti-E scope equipped with PerfectFocus, a Yokagawa CSU-X1 variable speed Nipkow spinning disk scan head, and Andor Zyla 4.2 Megapixel sCMOS camera) at the Washington University Center for Cellular Imaging (http://wucci.wustl.edu/). Images of ubiquitin-gCAMP6s fish islets were collected on a Zeiss LSM 880 Airyscan confocal microscope equipped with two non-descanned detectors for two-photon imaging, also at the Washington University Center for Cellular Imaging. Time-lapse images used 100 msec exposure at 1 sec intervals. All images were analyzed in Fiji (Schindelin et al. 2012).
To correct for movement in x-and y-planes, images were stack registered (using StackReg, rigid body) in Fiji before analysis. All calcium image data are presented as change in fluorescence intensity relative to baseline fluorescence intensity. Because the maximum excitability of an islet or b-cell can vary, and the intensity of islet fluorescence can vary, glucose responses are shown normalized to the change in fluorescence in response to KCl (showing maximum excitability due to islet depolarization). For determining trace cross-correlation and synchronicity, ROI measurements were analyzed using PeakCaller in MATLAB (Artimovich et al. 2017). The KCl response was excluded from segments in which cross-correlation analysis was performed, to capture the responses to glucose only.

Chemicals
All salts, amino acids, and other compounds were purchased from Sigma, except where indicated above.

Statistics
Statistical analyses were made in GraphPad prism. Except as noted, each data set was tested for deviation from normal distribution (D'Agostino-Pearson). For multiple group column data comparisons, data were analyzed by ANOVA, followed by Tukey's post-tests where normality assumptions were met. In comparisons of two groups, Student's T test with Welch's correction was used. In cases of non-normal distributions, the Kruskal-Wallis (more than two groups) or Mann-Whitney (two groups) tests were used. All values are indicated as mean AE SEM, except where noted. For dose-response curves, data were fitted using nonlinear regression.

Results
Zebrafish islets express L-type calcium channels The fish genome contains orthologs of the calcium channels found in mammalian islets, and RNA encoding these channels is present in islets (Sidi et al. 2004;Zhou et al. 2008;Sanhueza et al. 2009;Tarifeño-Saldivia et al. 2017), but functional demonstration of Ca 2+ channels in fish islet cells is lacking. Whole-cell voltage-clamp recordings from isolated zebrafish b-cells reveal nifedipine-sensitive L-type calcium currents (Kuryshev et al. 2014;Striessnig et al. 2015) (Fig. 1A), with current/voltage profiles ( Fig. 1B) very similar to those of mammalian L-type calcium currents (Lipscombe 2002;Mangoni et al. 2006), the observed major VDCCs in mammalian islets. The zebrafish pancreas develops from the endodermal germ layer comprising endocrine and exocrine tissue and is conserved from mammals to fish. An early 'primary' islet forms within the first day of development (Argenton et al. 1999;Biemar et al. 2001), and additional smaller duct-related "secondary" islets form as development progresses (Chen et al. 2007;Parsons et al. 2009). Similar currents were identified in both primary islet and secondary islet cells (Fig. 1B).

Adult Zebrafish islet calcium is glucose responsive
To probe intracellular calcium levels and potential responsivity to glucose in zebrafish b-cells, we generated Tg (-1.0ins:gCAMP6s) stl441 fish expressing cytosolic gCAMP6s driven by the zebrafish insulin promoter. Ex vivo isolated islets from these cGCAMP6s fish show colocalization of cGCAMP6s fluorescence and insulin staining ( Fig. 2A). Figure 2B shows representative snapshot images and timecourses of cGCAMP6s fluorescence in the presence of low glucose and after switch to high (20 mmol/L) glucose, and then after addition of KCl. The response is specific to metabolizable D-glucose, since there was no response to 20 mmol/L L-glucose ( Fig. 2C and D), and is sensitive to diazoxide ( Fig. 2C and D), indicating that it involves closure of K ATP channels (see below). The sigmoidal [glucose]-dependence (Fig. 3A), obtained from similar experiments with switches to intermediate [glucose], is similar to that seen in mammalian islets, with EC 50 of~10 mmol/L glucose, slightly higher than the typically reported 5-9 mmol/L for mouse and rat (Antunes et al. 2000), or human (Henquin et al. 2006) islets. Consistent with the nifedipine sensitivity of calcium currents, the glucose-induced calcium responses are abolished by the addition of 10 lmol/L nifedipine ( Fig. 3B and C). In contrast to mammalian islets (Henquin et al. 2006;Liu et al. 2008) and embryonic zebrafish islets  2019 | Vol. 7 | Iss. 11 | e14101 Page 5 (Lorincz et al. 2018), the amino acids glutamine, alanine, and leucine caused no activation in the presence of threshold (8 mmol/L) glucose (Fig. 3D). Finally, islets did not show any response to sucrose (Fig. 3E, representative trace), which further indicates a specific response to metabolizable glucose and not to osmotic shock or other stress from the added sugars.

Ca 2+ transients are not well-coupled between b-cells in adult Zebrafish islets
In mammals, individual b-cells vary in their expression of metabolite transporters, metabolic enzymes, and ion channels involved in the insulin secretion response, and thus exhibit variable sensitivities to glucose Silva et al. 2014). However, in intact islets, gap-junction coupling between b-cells ensures synchronous electrical and calcium signals, and hence secretory responses, across the islet (Farnsworth and Benninger 2014). In contrast to mammals, uncorrelated gCAMP6s fluorescence oscillations were observed in individual b-cells within the zebrafish islet. In high-resolution confocal images (Fig. 4A), individual cells clearly become active at very different glucose levels. Even cells that are physically close together lack synchronicity in their calcium spikes and glucose sensitivity (Fig. 4B), evidenced by very weak correlation coefficients between signals from individual cells (Fig. 4C). This is in stark contrast to the strong cell-cell correlation in isolated mouse islets under comparable conditions (Kenty and Melton 2015;Johnston Natalie et al. 2016). Connexin 36 is the primary gap junction protein in mammalian islets (Farnsworth and Benninger 2014). Connexin 35b is the major ortholog of mouse connexin 36 in fish (Jabeen and Thirumalai 2013; Carlisle and Ribera 2014; Watanabe 2017). It has been well-characterized in zebrafish brain (Jabeen and Thirumalai 2013; Carlisle and Ribera 2014), and both cDNA (Fig. 4D) and protein (Fig. 4E) were readily detected in brain, but not in zebrafish islets or heart.

Islets expressing K ATP -GOF are inexcitable, resulting in profound diabetes
In mammals, excitability is dramatically suppressed by gain-of-function mutations in K ATP channels, resulting in profound neonatal diabetes (Koster et al. 2000;Gloyn et al. 2004). To examine susceptibility of zebrafish glycemia to b-cell membrane excitability, we generated additional transgenic fish (Tg(-1.0ins:LoxP_mCherry_polyA_ LoxP,Kir6.2(K185Q,ΔN30)-GFP) stl443 , K ATP -fish) which conditionally express the same GFP-tagged gain-of-function Kir6.2 subunit that has been extensively used to demonstrate and study neonatal diabetes in mice (Koster et al. 2000) and previously shown to increase glucose levels in zebrafish larvae (Li et al. 2014). We crossed these K ATP -fish to zebrafish expressing HSP-16 inducible Crerecombinase, to generate K ATP -GOF animals which express the K ATP -GOF transgene only in b-cells, following heat shock induction (schematic in Fig. 5A). Robust transgene expression is evident in double transgenic bcells after 5-days heat-shock (by visualizing tagged GFP, Fig. 5B). Following 5 days of heat shock induction, blood glucose was unaltered in non-GOF control or single transgenic fish/islets (Fig. 5C), but K ATP -GOF fish rapidly developed severe hyperglycemia (>600 mg/dL), and this was then maintained >400 mg/dL for weeks (Fig. 5D). When the transgene was activated at the larval stage, glucose levels in adult fish were similar to those resulting from adult induction (Fig. 5E) Excised inside-out patch-clamp experiments (Fig. 6A) confirm that, in fluorescent cells, the K ATP -GOF transgene was incorporated into b-cell K ATP channels, resulting in the expected loss of ATP-sensitivity (Fig. 6B). Even though overall channel density was if anything reduced (excised patch current in zero ATP was 51.8 AE 7.4 pA in control, n = 4, c.f. 9.6 AE 2.4 pA in K ATP -GOF, n = 4), whole-cell voltage-clamp currents (Fig. 6C) show that, under basal conditions following break-in, voltage-gated K currents were unaffected in K ATP -GOF islets, but K ATP currents were already basally activated ( Fig. 6C and D). To examine the consequences for glucose-dependent Cahandling, we crossed K ATP -GOF fish to cGCAMP6s fish. Isolated islets from heat-shock induced K ATP -GOF fish show essentially no glucose-induced calcium responses, even at high (20 mmol/L) glucose levels, although they still respond appropriately to direct depolarization by 30 mmol/L KCl (Fig. 7A and B). These results are consistent with K ATP -GOF inhibiting electrical activity at high [glucose] by hyperpolarizing cells, an effect that is overcome by direct KCl-induced depolarization.
K ATP -GOF mice with untreated diabetes develop significant secondary consequences, including growth limitation (Girard et al. 2009;Remedi et al. 2009). Larvae-induced K ATP -GOF fish also showed dramatically reduced body length and weight at 10 weeks of age, when compared to Cre-negative littermate controls from the same clutches ( Fig. 7C and D). These data indicate that not only does b-cell K ATP -GOF induce similar inexcitability-dependent hyperglycemia in zebrafish as in mammals, but also similar secondary diabetic consequences.

Conservation of islet function between mammals and fish
In mammals, excitability-dependence of intracellular [Ca 2+ ] is well-established and is shown to be critical for regulation of insulin secretion. By contrast, islet excitability in lower vertebrates remains essentially unaddressed. We previously reported the expression and function of K ATP channels in zebrafish islet b-cells and showed that pharmacological K ATP channel activators worsened glucose tolerance in adult fish (Emfinger et al. 2017). Larval activation of K ATP -GOF transgene under tetracycline and tebufenozide-driven promoter control in b-cells, or treatment of normal larvae with diazoxide, raises glucose levels and inhibits overnutrition-induced b-cell expansion (Li et al. 2014), and intracellular [Ca 2+ ] is glucose-dependent in larval zebrafish islets (Lorincz et al. 2018). In the present study we have now characterized glucose-dependence of intracellular calcium in adult zebrafish islets, and show that the glucose dependence of calcium oscillations is similar, although not identical, to mammalian islets. We further show that islet calcium response to glucose can be blocked by b-cell specific induction of GOF mutations in K ATP , resulting in profound diabetes. This indicates that key components of excitability-dependent insulin secretion are well conserved between mammals and fish, although unfortunately, we do not have a suitable assay for secreted insulin or C-peptide in zebrafish. However, we observe several potentially important differences. First, amplification of glucose signals by amino acids, which is observed in mammalian islets (Henquin et al. 2006;Liu et al. 2008) and embryonic zebrafish islets (Lorincz et al. 2018), was not seen in adult fish islets (Fig. 3D). Given that zebrafish may be more dependent on protein than carbohydrate in the diet, this is teleologically surprising, but may reflect a difference in cellular expression of essential carriers or enzymes. Second, isolated zebrafish islets are not well-coupled electrically (Fig. 4). Tight coupling of b-cells to one another, evidenced by tight cell-cell coupling of Ca 2+ transients, is critical for normal mammalian glucose tolerance (Klee et al. 2008;Farnsworth and Benninger 2014): in mice lacking connexin 36, which forms the primary gap junctions in islet b-cells, basal insulin is elevated, and glucose responses of the overall islet are slowed, worsening glucose tolerance in otherwise healthy animals (Head et al. 2012). Because of the lack of cell-cell coupling in the adult fish islets, glucose-sensitivity of Ca 2+ is quite variable between cells within the intact islet (Fig. 4). Variable glucose-sensitivities of b-cells ex vivo have also been observed in larval and early juvenile zebrafish (Singh et al. 2017;Lorincz et al. 2018); ex vivo recordings of these younger zebrafish islets clearly show cells activating independently, at thresholds between 5 and 20 mmol/L glucose. Absence of cell-cell electrical coupling in zebrafish b-cells may explain, in part, the relatively lower glucose tolerance of zebrafish compared to mammals (Eames et al. 2010;Emfinger et al. 2017), with higher peak blood glucose and slower return to baseline glucose after a glucose injection. Figure 6. Islet K ATP -GOF expression results in basal K ATP and ATP-insensitive channels. (A) Representative excised inside-out patch-clamp recordings (at À50 mV) from b-cells isolated from control (black) or K ATP -GOF (red) islets, in the presence of ATP at concentrations (in micromolar) as indicated. (B) Steady-state dependence of membrane current on [ATP] (relative to current in zero ATP (I rel )) for control and K ATP -GOF channels. Data points represent the mean AE SEM. (n = 4 patches in each case). The fitted lines correspond to least squares fits of a Hill equation (see Methods). (*) P < 0.01 compared to wild-type K ATP (controls) by unpaired Student's t test. (C) In whole-cell mode basal conditions, voltage-clamp ramps from À120 to À40 mV (over 1 sec) activates similar Kv currents above À20 mV in both K ATP -GOF and control cell. However, basal K ATP channel activation is evident in K ATP -GOF cells as additional~linear current reversing at À80 mV (boxed current is amplified in insert). (D) Averaged basal currents at À120 and À40 mV from experiments as in C (n = 5 control cells, n = 7 K ATP -GOF cells). K ATP -GOF zebrafish recapitulate major features of mammalian neonatal diabetes Transgenic mice expressing K ATP -GOF at birth, as well as human K ATP -dependent human neonatal diabetic patients, exhibit reduced growth rates (Koster et al. 2000;Polak and Cave 2007;Girard et al. 2009;Remedi et al. 2009). Mammalian and Danio K ATP genes are very similar (>90% identity at the amino acid level) (Emfinger et al. 2017). K ATP -GOF zebrafish, expressing the identical transgene that we originally used in mice (Koster et al. 2000), results in basal activity of K ATP channels in b-cells and in the fish becoming profoundly hyperglycemic, and also showing growth limitation when induced as larvae. Other secondary consequences of hyperglycemia can occur directly in the islet, with loss of b-cell identity and insulin content (Brereton et al. 2014;Wang et al. 2014). Whether this occurs in zebrafish islets exposed to high glucose for long periods is currently unknown but future lineage tracing studies using models such as ours may now be used to clarify these questions. Zebrafish are transparent as larvae, reproduce frequently and in large clutches, have a fully sequenced genome, and can be efficiently genetically mutated, making them well-suited to large-scale drug or genetic screens. We have shown that zebrafish b-cells exhibit many similarities to mammals in glucose responsivity, including the profound hyperglycemia that results from electrical glucose-unresponsivity. We thus provide a zebrafish model of b-cell inexcitability-dependent diabetes that may be useful for drug and genetic modifier screens.