Introduction

Congenital hyperinsulinism (HI) is the leading cause of persistent hypoglycemia in infants. Currently, more than 30 genetic loci have been identified to cause hyperinsulinism. Inactivating mutations in ABCC8 or KCNJ11, which encode the subunits of the beta cell KATP channel SUR1 and KIR6.2, respectively, account for close to 70% of cases.1 The phenotype of KATPHI is characterized by fasting and protein-induced hypoglycemia, resulting from dysregulated insulin secretion.2 Islets isolated from the pancreas of infants with KATPHI exhibit high basal insulin secretion, lack of response to stimulation with glucose but increased insulin secretion in response to stimulation with a mixture of amino acids.3 Children with KATPHI are typically unresponsive to diazoxide, the only Food and Drug Administration (FDA) approved treatment for HI, and may require pancreatectomy for intractable hypoglycemia.4 Thus, there is a significant unmet need for better treatments for KATPHI.

KATPHI models are limited to two main methodologies: primary isolated human pancreatic islets and the Sur1-/- (aka Abcc8-/-) mouse model. The Sur1-/- mouse demonstrates fasting hypoglycemia, high cytosolic Ca2+ in pancreatic islets, and increased amino acid stimulated insulin secretion similar to that seen in humans with HI.3 5 However, use of primary islets relies on pancreatectomy, which occurs sporadically, thus limiting consistent access to islets for research. The Sur1-/- mouse model is restricted by the inability to study pancreatic islets in vivo, low fecundity, and the increasing cost of colony maintenance. Given these limitations, a new animal model would serve to advance the field with potential for novel breakthroughs in the treatment of HI.

Zebrafish (Danio rerio) are a burgeoning option for modeling human endocrinological diseases.6 7 Zebrafish are an advantageous animal model for numerous reasons: high fecundity, early organ development,8 ease of genetic manipulation, zebrafish larvae transparency, and significant overlap of the zebrafish and human genome.9 It has been demonstrated that zebrafish pancreatic structure and islet architecture are comparable to that found in mammals.10 Importantly, as in humans and mice, zebrafish beta cells express KATP channels composed of SUR1 and KIR6.2 subunits that are encoded by the abcc8 and kcnj11 genes, respectively.11 Additionally, it has been shown that the mechanisms of zebrafish glucose homeostasis are similar to that seen in mammals.10 12 In this study, we demonstrate that the abcc8-/- zebrafish larvae demonstrate the hyperinsulinemic phenotypes seen with inactivating KATP channel mutations in humans and will serve to expand our understanding of HI and discovery of better therapeutics.

Research design and methods

Animal Research: Reporting of in Vivo Experiments (ARRIVE) reporting guidelines were used for this study.13

Data from every zebrafish for which a sample was collected were included during the analysis, there were no exclusions. The study was not blinded during the experiment, outcome assessment, or data analysis. The research question, key design features, and analysis plan were prepared before the study, but not registered.

Fish lines and maintenance

Zebrafish were maintained at the Children’s Hospital of Philadelphia Aquatic Zebrafish Core which performed all daily care and feeding. Zebrafish were monitored daily for normal feeding and swimming behavior, but no adverse events were observed. Larva (age 7–15 dpf) are fed two times per day with GEMMA Micro 75 (Skretting Zebrafish, USA) and supplemented with spirulina. Juvenile zebrafish (age 15–90 dpf) are fed two times per day (GEMMA 150, Skretting Zebrafish) and brine shrimp once daily (grown/ maintained in the CHOP Aquatic Zebrafish Core). Adult zebrafish (aged 90 dpf and greater) are fed two times per day with Zeigler Adult Zebrafish Diet (Zeigler Bros, USA) and once daily with brine shrimp. All experiments were non-survival procedures and completion of the study was the endpoint.

Both male and female (~1:1) zebrafish were used for experiments and allocation of zebrafish to the experimental groups (control abcc8+/+ and abcc8-/-) was determined by genotyping and driven by Mendelian inheritance. Experiments were completed the same time during the day (08:00 to 13:00) and tanks were maintained/pulled from the same rack system.

Adult zebrafish

The abcc8-/- fish line (sa15863) was acquired from the Zebrafish International Resource Center (ZIRC) (http://zfin.org). This zebrafish line was generated by mutagenesis via n-ethyl-n-nitrosourea (ENU) as part of the Zebrafish Mutation Project14 and contains a point mutation that results in a premature stop and therefore truncation of abcc8. This fish line was maintained through abcc8+/- x abcc8+/- crosses and abcc8-/- and abcc8+/+ offspring from these heterozygous crosses were used for adult experiments. With 61 adult zebrafish per group, there is greater than 95% power to detect a difference of 18% in plasma glucose levels (using alpha 0.05).

Zebrafish larva

For larvae experiments, embryos and larvae were maintained at 28°C throughout the experiments. To generate larvae, adult zebrafish were set up in undivided mating tanks and resultant embryos were collected/sorted on the same day (0 day post-fertilization (0 dpf)). The embryos were placed in embryo water (E3) in an incubator overnight (28°C). Viable embryos were sorted the next day (1 dpf), sanitized with bleach, and then treated with pronase to promote uniform hatching. At 2 dpf, pronase was removed and the embryos were maintained in E3.

The beta cell-specific gCaMP6s-expressing transgenic fish (ins:gCaMP6s) that was used for cytosolic Ca2+ imaging studies was described previously15 16 and obtained from the Nichols lab (Washington University, Missouri, USA). To generate the abcc8-/-; ins:gCaMP6s and abcc8+/+; ins:gCaMP6s larva, we initially crossed abcc8-/- x abcc8+/+; ins:gCaMP6s zebrafish, which resulted in 50% abcc8+/- and 50% abcc8+/-; ins:gCaMP6s zebrafish. The green fluorescent protein (GFP) positive (and therefore transgene expressing) embryos were separated. The abcc8+/- and abcc8+/-; ins:gCaMP6s zebrafish were crossed generating abcc8+/-, abcc8+/-; ins:gCaMP6s, abcc8+/+, abcc8+/+; ins:gCaMP6s, abcc8-/-, and abcc8-/-; ins:gCaMP6s. The abcc8; ins:gCaMP6s fish line was maintained through abcc8+/- x abcc8+/-; ins:gCaMP6s crosses and experimental fish were generated from sibling knock-out (KO) (abcc8-/- x abcc8-/-; insgCaMP6s) and wild-type (WT) (abcc8+/+ x abcc8+/+; ins:gCaMP6s) crosses that were direct results of the heterozygous crosses. The KO and WT zebrafish lines were not maintained as separate lines to mitigate genetic differences between the genotypes.

Larval whole body glucose, insulin, protein, and length measurements

Using a protocol modified from that described by Jurczyk et al,12 a single larvae (5 dpf) was collected into an Eppendorf tube, excess media was removed by pipette, and then the larvae were flash frozen on dry ice. Tubes were removed from dry ice and KREBS buffer (12 µL per fish) was added. The samples were homogenized manually with a pestle (~20 turns) and the supernatants were cleared by centrifugation. The cleared supernatant was collected into a new labeled tube and evaluated for glucose content via fluorometric assay (Abcam, Cat. No. ab169559) per manufacture instructions. With 19 zebrafish larvae per group, there is greater than 95% power to detect a difference of >20% on whole body glucose levels (using alpha 0.05).

For measurement of whole body insulin, larva were prepared as described for the whole body glucose assay (single larva collected, excess media removed, flash frozen, homogenized, centrifugation of lysate) and the supernatant was evaluated using a homogeneous time resolved fluorescence (HTRF) assay (Cisbio, Cat. No. 62IN1PEH) per manufacture instructions. Validation of the HTRF assay for use with larval zebrafish insulin can be found in online supplemental figure 1A and B.

For measurement of whole body protein, larva were prepared as described for the whole body glucose and insulin assays above, except using RIPA buffer. The cleared supernatant was assayed using the Pierce BCA Protein Assay Kit (Thermo Scientific, Cat. No. 23225) as per manufacture instructions.

For larval length measurements, larva were anesthetized using tricaine, visualized with a dissecting scope, and measured (tip of head to tip of caudal fin) in the plate by arranging the larva over a ruler.

In vivo cytosolic Ca2+ imaging: abcc8+/+; ins:gCaMP6s and abcc8-/-; ins:gCaMP6s larvae were generated via homozygous sibling offspring crosses directly from heterozygous crosses (abcc8+/-x abcc8+/-; ins:gCaMP6s). Larvae were placed in E3/phenylthiourea (PTU) (0.03 µg/L) from day 2 dpf forward to prevent pigment formation. In vivo calcium imaging of zebrafish larvae was completed using a modified version of Delgadillo-Silva et al.17 Embryos (5 dpf) were anesthetized in tricaine (MS-222) and placed onto channels in an agarose gel (2% agarose, SeaPlaque, Lonza, Cat. No. 50101) to minimize movement. A fluorescent imaging dissecting microscope (Olympus MVX10) was used to focus onto the GFP positive primary islet (×1 objective, MV PLAPO Olympus) on a single optical plane and a microinjector (Harvard apparatus, PL1-90) was used to inject (100–120 nm bubble of either glucose (1 M) or glutamine (10 mM)) into the space near the cardinal vein and proximal to the developing swim bladder. GFP fluorescence (488 nm excitation; 510 nm emission) was recorded (Olympus DP73 color camera) for 60 s after injection. Videos were recorded at 20 frames/s and GFP fluorescence exposure (60% intensity) was continuous during the entire video. Screenshots of GFP fluorescence was captured at every 1 s interval of the video and saved as individual TIFF image files. Alexafluor 568 (12.5 µg) (578 nm excitation; 603 nm emission) (Invitrogen, Cat. No. A11011) was used as a red fluorescent marker in the injection substrate to verify injection into the zebrafish larva after the collection of the GFP video was finished (60 s). ImageJ18 was used to open the individual TIFF files and the fluorescent primary islet was manually outlined using the ImageJ freehand ROI tool and background fluorescence was measured by outlining an area containing no cells. Measurements for ‘area integrated intensity’ and ‘mean gray value’ were collected for each image. Corrected total islet fluorescence (CTIF) was calculated using these averaged measurements collected from the TIFF images. Corrected total islet fluorescence was calculated using the following equation:

CTIF=Integrated Density − (Area of Selected Islet×Mean Fluorescence of Background Readings).

Visualization of the ImageJ procedure found at: https://theolb.readthedocs.io/en/latest/imaging/measuring-cell-fluorescence-using-imagej.html

Adult zebrafish blood collection for plasma insulin and glucose

Adult zebrafish were fasted overnight and euthanized by cold shock in water ≤4°C. The tail was excised and plasma glucose was measured directly on a glucose strip after excision via handheld glucose meter (Freestyle Lite, Abbot). Remaining blood was collected into tubes with a pipette. Blood from six zebrafish was pooled and spun down to collect the plasma. Plasma samples (5 µL in duplicate) were evaluated by insulin ELISA (Mercodia, Cat. No. 10-1249-01). Validation of the Insulin ELISA for use with zebrafish insulin can be found in previous work19 and online supplemental figure 1C and D.

Adult zebrafish glucose tolerance test (GTT)

Adult zebrafish were fasted overnight and anesthetized using the gradual cooling method, which involves slowly chilling water from room temperature to ≤10°C over the course of several minutes by adding ice to the tank. Glucose (1 mg/g) was administered via intraperitoneal injection as previously described.20 Zebrafish were returned to room temperature water to recover. Plasma glucose measurements were collected via tail excision after euthanasia by cold shock of individual zebrafish per time point (0, 30, 60, 90, 240 min) via handheld glucose meter.

Data analysis

Statistical analyses were performed on Excel or GraphPad Prism software V.8 or V.9. Results are presented as mean±SE of the mean (SEM). Single time end points were checked for normal distribution using the Shapiro-Wilk test and if passed for normality, the data were analyzed using a one-way analysis of variance test. If the data did not pass normality in the Shapiro-Wilk test, then data were analyzed using the non-parametric Kruskal-Wallis test. For GTT and Ca2+ fluorescence measurements, data were analyzed using area under the curve (AUC). Differences were considered significant at p<0.05.

ResultsLoss of abcc8 expression in zebrafish larvae results in decreased plasma glucose with increased plasma insulin and insulin secretion

Until zebrafish reach the adult stage (4 months of age), enough blood volume is not present for assessment via traditional means such as handheld glucose meter or insulin ELISA. Therefore, we developed methods to assess whole body glucose and insulin in zebrafish larvae. Comparison using these methods revealed that abcc8-/- zebrafish larvae (5 dpf) have lower whole body glucose levels compared with abcc8+/+ zebrafish (figure 1A) (n=100 for each genotype). Consistent with these findings, whole body insulin is significantly higher in abcc8-/- compared with abcc8+/+ zebrafish larva (figure 1B) (n=100 for each genotype). No difference in body size (length or whole body protein) was identified when comparing abcc8+/+ and abcc8-/- larva (5 dpf) (online supplemental figure 2) (n=100 for each genotype). Taken together, these results indicate excessive unstimulated insulin production in abcc8-/- zebrafish during the larval stage.

Figure 1

Zebrafish larvae whole body glucose and insulin. (A) Whole body glucose measurements of abcc8+/+ control (95% CI 259.4, 294.2) and abcc8-/- (95% CI 226.9, 259.3) larvae (5 days post-fertilization (dpf)) (p=5.5×10-3, n=100). (B) Evaluation of whole body insulin in abcc8+/+ control (95% CI 0.009, 0.01) and abcc8-/- (95% CI 0.013, 0.014) zebrafish larvae (5 dpf) (p=4.4×10-11, n=100). **p=5.5×10-3; ***p=4.4×10-11. Data represent mean±SEM.

Both human KATPHI and Sur1-/- mouse pancreatic islets demonstrate high basal cytosolic Ca2+ concentrations and amino acid stimulated insulin secretion.2 3 Cytosolic Ca2+ concentrations are used as a surrogate for insulin secretion in mouse and human pancreatic islets but can only be assessed in isolated islets. The transparency of zebrafish larvae presents a unique opportunity to evaluate cytosolic Ca2+ in pancreatic islets in vivo using ins:gCaMP6s zebrafish, which express a genetically encoded GFP-labelled calcium sensor under the control of the insulin promoter. We crossed the ins:gCaMP6s line with our abcc8 zebrafish line to create an abcc8-/-; ins:gCaMP6s line and examined cytosolic Ca2+ levels in the pancreatic islets of the larvae. We found that abcc8-/-; ins:gCaMP6s larvae have higher baseline (unstimulated) cytosolic Ca2+ compared with their abcc8+/+; ins:gCaMP6s counter parts (figure 2A) (n=9–10 for each genotype). The elevated cytosolic Ca2+ along with decreased glucose in the abcc8-/-; ins:gCaMP6s zebrafish larvae demonstrate inappropriate insulin secretion akin to that seen in human patients. In zebrafish larvae stimulated with glucose, we did not observe a significant difference (figure 2B) (n=9 for each genotype). However, when stimulated with the amino acid glutamine, only the abcc8-/-; ins:gCaMP6s zebrafish larvae demonstrated an increased cytosolic Ca2+ response (figure 2C) (n=10 for each genotype). This insulin secretion response to glutamine is in line with the established prominent role of glutamine in amino acid stimulated insulin secretion found in Sur1-/- islets.5 AUC calculations confirm the significantly increased baseline cytosolic Ca2+ and response to glutamine in the abcc8-/-; ins:gCaMP6s zebrafish larvae (figure 2D).

Figure 2

Zebrafish larvae pancreatic islet cytoplasmic Ca2+ measurements. (A) Measurement of zebrafish pancreatic islet cytosolic Ca2+ fluorescence in abcc8+/+ control (95% CI 8.4×104, 8.8×104) and abcc8-/- (95% CI 1.1×105, 1.2×105) zebrafish larvae (5 dpf) (p=0.0216, n=8–10). (B) Zebrafish larvae pancreatic islet cytosolic Ca2+ measurements in response to glucose stimulation in abcc8-/- (95% CI 1.6×105, 1.7×105) and control abcc8+/+ (95 % CI 1.4×105, 1.6×105) larvae (p=0.51, n=8–10). (C) Assessment of zebrafish larvae pancreatic islet cytosolic Ca2+ in response to glutamine in abcc8+/+ control (95% CI 9.7×104, 1×105), and abcc8-/- (95% CI 1.32×105, 1.38×105) larvae (p=0.0486, n=8–10). (D) Area under the curve calculations for cytoplasmic Ca2+ experiments as denoted. *p<0.05. Data represent mean±SEM.

Adult Abcc8-/- Zebrafish do not exhibit a hyperinsulinemic phenotype

We evaluated adult zebrafish (aged 4–6 months post fertilization) and observed no difference in plasma glucose levels between the abcc8-/- and the abcc8+/+ zebrafish (figure 3A) (n=123 for both genotypes). Additionally, we identified no difference in plasma insulin levels (figure 3B) or the insulin to glucose ratios (figure 3C) between the two genotypes (n=14 for abcc8+/+; n=13 for abcc8-/-). This contrasts with the significant difference we observed in larvae but is in accordance with results found in adult zebrafish by the Nichols lab.21 Our final metabolic interrogation was an intraperitoneal glucose tolerance test (IPGTT) performed on abcc8+/+ and abcc8-/- adult zebrafish and we observed no difference in glucose tolerance (figure 3D and E).

Figure 3

Evaluation of adult zebrafish plasma glucose, insulin, and glucose tolerance. (A) Plasma glucose measurements in adult abcc8+/+ control (95% CI 58, 66) and abcc8-/- (95% CI 63, 75) zebrafish (p=0.059, n=123). (B) Plasma insulin values in adult abcc8+/+ control (95% CI 0.88, 1.49) and abcc8-/- (95% CI 0.91, 1.58) zebrafish (p=0.777, n=13–14 pools). (C) Insulin to glucose ratios in adult abcc8+/+ control (95% CI 0.32, 0.72) and abcc8-/- (95% CI 0.40, 0.97) zebrafish (p=0.3, n=13–14 pools). (D) Glucose tolerance test (GTT) in adult abcc8+/+ control (n=10 zebrafish per time point) and abcc8-/- zebrafish (n=9 zebrafish per timepoint). (E) Area under the curve (AUC) calculations of the GTT in adult zebrafish (p=0.44 comparing abcc8+/+ glucose vs abcc8-/- glucose). Data represent mean±SEM.

Discussion

We aimed to develop a novel model of KATPHI with potential for higher throughput and in vivo islet assessment capabilities that the currently available models lack, in order to facilitate drug development efforts. In this study, we demonstrate that abcc8-/- zebrafish larvae exhibit the definitive phenotypes found in human HI islets and Sur1-/- mice: fasting hypoglycemia, increased islet cytosolic Ca2+, increased insulin secretion in the context of low glucose, and amino acid induced insulin secretion. Interestingly, we discovered that these phenotypes do not persist into adulthood in zebrafish. There is evidence that this is the course for some human patients as well. Previous studies have shown that for some KATPHI patients, symptoms lessen over time and for some others progression from hyperinsulinism with hypoglycemia to hyperglycemia has been reported.22 23 Understanding of the mechanisms regulating progression of KATPHI has remained elusive and this zebrafish model could provide insight on the evolution of KATPHI over the patient’s lifespan. Further, it reinforces the benefit to long-term outcomes of treatments that would allow preservation of the pancreas.

Notably, traditional methods of assessing plasma glucose and insulin cannot be used for zebrafish larva due to the minuscule volume of blood available per larvae. We thus developed methods to measure whole body glucose and whole body insulin at the larval stage. It is important to acknowledge that plasma insulin and glucose levels in adult zebrafish represent circulating levels, while whole body measurement of insulin and glucose in larvae includes circulating as well as that contained/produced within cells but not yet secreted or circulating. Whole body measurement confers a complete assessment of the totality of pathways involving both production and utilization that will detect changes as it would result in correlating alterations of the upstream production pathways and thus in changes to the total level. The KATPHI mouse model, Sur1-/- (aka Abcc8-/-) mice are an example of these upstream alterations. It was previously identified that in addition to increased plasma insulin in the context of low plasma glucose, that beta cells in the pancreatic islets of Sur1-/- mice also demonstrate changes upstream of secretion with ~50% more insulin granules docked at the plasma membrane ready for exocytosis compared with controls.24 This infers an increased production of insulin in response to secretion demand and thereby a change in the overall total insulin levels. Therefore, both measurement of circulating plasma and whole body levels are valuable tools in the assessment of hyperinsulinism and alterations in insulin secretion.

There are three main areas for expanded study that can be derived from the abcc8-/- zebrafish model. First is the opportunity to investigate the progression of KATPHI and the drivers leading to glucose intolerance or the reduced need for medication seen in some patients. Second, this model provides a platform to evaluate novel therapeutics for hyperinsulinism using the Ca2+ sensor for direct evaluation of effects on insulin secretion in vivo. Additionally, higher throughput is possible, leading to faster identification of novel effective pharmaceuticals. Third, genetic manipulation of zebrafish is relatively straight forward, allowing for the potential to test novel genetic etiologies of HI.25 Our characterization of the abcc8-/- zebrafish line both establishes its capability as a model to study HI and as well as methods necessary for detection of HI phenotypes in zebrafish. Overall, abcc8-/- zebrafish have the potential to expand the understanding of KATPHI progression across a lifespan and establishes a new model for the study of HI with distinct advantages that do not exist in current models of HI.