Skeletal muscle insulin resistance in zebrafish induces alterations in β-cell number and glucose tolerance in an age- and diet-dependent manner

Abstract

Insulin resistance creates an environment that promotes β-cell failure and development of diabetes. Understanding the events that lead from insulin resistance to diabetes is necessary for development of effective preventional and interventional strategies, and model systems that reflect the pathophysiology of disease progression are an important component toward this end. We have confirmed that insulin enhances glucose uptake in zebrafish skeletal muscle and have developed a zebrafish model of skeletal muscle insulin resistance using a dominant-negative IGF-IR. These zebrafish exhibit blunted insulin signaling and glucose uptake in the skeletal muscle, confirming insulin resistance. In young animals, we observed an increase in the number of β-cells and normal glucose tolerance that was indicative of compensation for insulin resistance. In older animals, the β-cell mass was reduced to that of control with the appearance of impaired glucose clearance but no elevation in fasting blood glucose. Combined with overnutrition, the insulin-resistant animals have an increased fasting blood glucose compared with the control animals, demonstrating that the β-cells in the insulin-resistant fish are in a vulnerable state. The relatively slow progression from insulin resistance to glucose intolerance in this model system has the potential in the future to test cooperating genes or metabolic conditions that may accelerate the development of diabetes and provide new therapeutic targets.

type 2 diabetes (T2D) is a global health concern that affects 300 million people worldwide (8). The primary risk factor for development of T2D is excess weight and obesity from lack of physical activity and overnutrition. This creates a milieu that interferes with the function of insulin (3), a hormone from pancreatic β-cells that promotes adsorption of blood glucose to muscle, fat, and liver and suppresses liver glucose production (22). To compensate for this prediabetic insulin resistance and maintain euglycemia, β-cells increase insulin production by increasing cell number and secretion (32). However, unresolved insulin resistance can instigate β-cell decompensation, leading to diabetes (40). Although candidates have been proposed, the molecular mechanisms underlying the prediabetes-to-diabetes progression have not been well understood.

Zebrafish are an emerging model to address molecular mechanisms of glucose homeostasis. Although largely used to study β-cell development (13, 15, 20), zebrafish have recently been used to investigate the regulation of β-cell regeneration (1) and as a platform for drug discovery targeting β-cell proliferation (39). We have shown that the overnutrition also induces β-cell compensation in zebrafish (25) and have used this as a model to uncover pathways involved in this process (24). Importantly, regulation of gluconeogenesis is conserved at the molecular level in zebrafish (11, 14, 17). Additionally, zebrafish have been shown to be able to clear exogenous glucose, albeit at a reduced rate compared with mammals (10). But it is unknown whether insulin stimulates glucose uptake in zebrafish skeletal muscle.

A prerequisite for using zebrafish to uncover mechanisms involved in the progression from insulin resistance to diabetes is a model of insulin resistance-driven diabetes. Given its impressive regenerative capacity in many tissues, including β-cells (1, 30, 39), it is unknown whether zebrafish will succumb to persistent insulin resistance and develop T2D. Here, we show that insulin stimulates glucose uptake in zebrafish skeletal muscle, which is inhibited by transgenic tissue-specific expression of a dominant-negative IGF-I receptor (IGF-IR). We show that young transgenic fish with muscle insulin resistance maintain glucose tolerance with an increase in the number of β-cells. As they age, these fish develop glucose intolerance, which can be exacerbated with overnutrition. This study demonstrates that, although skeletal muscle is an insulin target tissue in zebrafish, chronic impairment of insulin signaling in the skeletal muscle results only in slow development of glucose intolerance.

MATERIALS AND METHODS

Zebrafish strains and maintenance.

Zebrafish were raised in an Aquatic-Habitats system on a 14:10-h light-dark cycle. Embryos were obtained from natural crossing and raised at 28.5°C in an incubator with lights on a 14:10-h light-dark cycle. Animals were staged by days postfertilization (dpf) or in months of age. At 5 dpf, fish were placed on the normal rearing diet and maintained as per standard protocols until they were of the appropriate age. All fish used in this study were in an background. All procedures were approved by the Vanderbilt University Institutional Animal Care and Use Committee.

Establishment of transgenic lines.

The Tol2 transposon system (36) was used to generate the Tg(acta1:dnIGF1R-EGFP) transgenic line. A 3.9-kb fragment of the α-actin promoter (16) was used to drive expression of the dominant-negative IGF-IR. Initial analyses were performed in two independent lines for each transgene, and similar results were obtained. All results reported here were from F2 or F3 fish.

Glucose uptake.

To determine the efficiency of zebrafish skeletal muscle to uptake glucose, a nonradioactive method to determine uptake of 2-deoxyglucose was used (33). Six-month-old Tg(acta1:dnIGF1R-EGFP) and nontransgenic sibling fish were fasted overnight, anesthetized using ice water, and injected intraperitonially with 0.5 mg 2-deoxyglucose/g fish wt with and without 0.0075 U insulin (bovine)/g fish wt (Sigma). Fish were allowed to recover for 15 min and then euthanized in ice water. Skeletal muscle was isolated from the posterior trunk and separated from skin, major blood vessels, and bones. The muscle was snap-frozen in liquid nitrogen and stored at −80°C until use. The muscle was solubilized in 10 mM Tris, pH 8, sonicated, and centrifuged to remove debris. Total protein was determined (Bio-Rad), and equivalent amounts of protein were used for the assay. Efficiency of the assay was confirmed using a commercially available kit for 2-deoxyglucose (CosmoBio) that was based on the same chemistry as that used in our assay.

Western blot.

Six-month-old Tg(acta1:dnIGF1R-EGFP) and nontransgenic sibling fish were used to determine insulin receptor activity. All fish were fasted for 16 h. One group remained unfed, one group was fed with artemia and had muscles harvested 60 min after the meal, and one group was injected with insulin. For insulin injection, fish were anesthetized using ice water and injected intraperitonially with 0.0075 U insulin (bovine)/g fish weight (Sigma). Fish were allowed to recover for 15 min and then euthanized in ice water. For all groups, skeletal muscle was isolated from the posterior trunk and separated from skin, major blood vessels, and bones. The muscle was snap-frozen in liquid nitrogen and stored at −80°C until use. The muscle samples were homogenized in RIPA buffer (Sigma) with protease (Complete; Roche) and phosphatase inhibitors (PhosStop; Roche). Fifty micrograms of total protein was used for the Western blot, and membranes were probed for phospho-Akt Ser⁴⁷³ (Cell Signaling Technology), Thr³⁰⁸ (C31E5E; Cell Signaling Technology), and unphosphorylated Akt (Cell Signaling Technology). IR-conjugated secondary antibodies (Rockland) were used in conjunction with membrane scanning on an Odyssey system (LiCor) and analysis using ImageJ software.

Quantification of β-cell number or β-cell area.

For fish <28 dpf, Tg(−1.2ins:H2BmCherry) or Tg(−1.2ins:tagRFP) was used to mark β-cells and the number of β-cells counted as described (25). For adult fish, Tg(−1.2ins:H2BmCherry) was used to mark the β-cells when possible, but immunofluorescence for insulin was used to mark the β-cells where fish did not carry the Tg(−1.2ins:H2BmCherry) transgene. Dissection of the pancreas and quantification of β-cell area was performed as described previously (23). Specifically, the gastrointestinal (GI) tract was removed, and the entire pancreas was dissected away from the intestine and flat-mounted on glass slides. Pairs of images for the entire pancreas of each animal were captured, with one under fluorescence showing fluorescently labeled β-cells and one under transmitted light. The pancreatic tissue has an image density distinct from the intestine and liver, the two potential types of contaminating tissue (not shown). The area of the pancreatic tissue was determined using ImageJ. To convert fluorescent area to the number of β-cells, a conversion equation was determined by counting the absolute number of β-cells per area in a subset of samples. This resulted in more accurate estimation than simply dividing the area by an average β-cell size. For each image, the β-cell fluorescent area was determined using ImageJ, converted to the number of β-cells, and compared with the pancreatic tissue present in that image.

Glucose tolerance testing.

Glucose tolerance was determined in adult fish, as described previously (23). Briefly, after a 16-h fast, fish were anesthetized using ice water, injected intraperitonially with 0.5 mg glucose/g fish wt, and allowed to recover for the appropriate time. For blood collection, fish were fully anesthetized in ice water, and blood was collected in heparanized hematocrit tubes (IRIS Sample Processing) by cutting through the gill operculum and puncturing the heart. Glucose concentration in the samples was determined using Amplex Red Glucose Assay (Life Technologies), and blood volume determined by hemoglobin content was assayed using Drabkin's reagent (Sigma).

Overfeeding study.

One-year-old Tg(acta1:dnIGF1R-EGFP) and age-matched control fish were separated into 3-liter tanks at a density of eight fish per tank with mixed males and females. For the first 2 wk, the fish were fed under the standard feeding protocol of four feedings per day: two live artemia meals (∼100/fish) and two meals of Tetra-Min flake diet (∼10 mg/fish) (Tetra). For the second 2 wk, the standard feeding protocol continued, but the fish were supplemented with two additional meals per day: one with 500–1,000 artemia/fish and one with 50 mg/fish of the flake diet. Weights were recorded weekly. Blood samples were collected after a 16-h fast, following a published procedure that had minimal adverse effects on the fish (41). Briefly, blood was collected from the dorsal aorta in the posterior trunk by collecting blood into heparinized glass needles with a bore size of 100 μm. To minimize reduction in circulating blood volume, 1–3 μl of blood was collected from each animal. Blood glucose was determined as outlined above for the glucose tolerance testing.

Immunofluorescence and 5-ethynyl-2-deoxyuridine staining.

For analysis of proliferation, 28 dpf animals were incubated in 500 μM 5-ethynyl-2-deoxyuridine (EdU) with 1% DMSO in fish water for 24 h. Following the labeling period, animals were euthanized in ice water and fixed in 4% paraformaldehyde for 24 h at 4°C. The GI tract was removed and the liver dissected away from the remainder of the tissue. EdU was detected using the Click-iT EdU Alexa Fluor 488 Imaging Kit (C10337; Life Technologies) according to published protocols (26). All images were collected using a Zeiss LSM510 or Zeiss LSM710 (Carl Zeiss). For quantification, the images were examined through the entire image stack of individual confocal slices. Only cells with fully overlapping signals were scored as double positive.

For immunostaining, 2-mo-old zebrafish were euthanized, the abdominal cavity was injected with 4% parafomaldehyde, and the entire fish was fixed in 4% paraformaldehyde for 24 h at 4°C. The fish was subsequently washed in 1× PBS, equilibrated in 30% sucrose for 2 days, and cryosectioned at 12-μm thickness. Sections were rehydrated in 1× PBS + 0.1% Tween-20, and nonspecific binding was blocked by incubating in 5% normal goat serum and 1.0% BSA plus 0.2% Triton X-100. A rabbit polyclonal antibody against GFP (Abcam) was used at 1:500 diluted in 2% normal goat serum and 1% BSA plus 0.2% Triton X-100. Anti-GFP was detected either using an Alexa Fluor 488-labeled anti-rabbit antibody (1:2,500; Life Technologies) or with a biotinylated anti-rabbit antibody (1:2,500; Life Technologies) followed by avidin-biotin-peroxidase complex amplification and 3,3′-diaminobenzidine detection (Vector Laboratories).

Quantitative RT-PCR.

The entire GI tract was isolated from 10-wk-old control or zMIR fish after a 16-h fast. RNA was isolated from Trizol (Life Technologies) and cDNA generated using MMLV reverse transcriptase (Promega). Real-time PCR was performed on a Bio-Rad CFX96 machine using primers for insulin, glucagon, and amylase, with primers for β-actin and 18S used as quality control. Primer sequences are available upon request.

Statistics.

Data are represented as means ± SE. Data were analyzed using Student's t-test or ANOVA with Tukey-Kramer HSD, depending on the number of comparisons (JMP; SAS Institute, Cary, NC). Significance was accepted at P < 0.05.

RESULTS

Generation of a skeletal muscle insulin-resistant zebrafish (zMIR).

To generate a zebrafish with impaired insulin signaling, we adopted an approach that was shown to be effective for generating insulin resistance in mice (2, 12). We employed a mutated form of the IGF-IR where the intracellular signaling domain has been replaced by enhanced green fluorscent protein (EGFP) (34), which inhibits both IGF-I and insulin signaling. We chose this approach rather than manipulating only the insulin receptor since conditional deletion of insulin receptor (4) or expression of a dominant-negative insulin receptor (29) does not lead to abnormalities in glucose homeostasis in mice. The normal glucose homeostasis in these mice is thought to be due to insulin's ability to activate IGF-IR (35). Skeletal muscle-specific expression of the mutated IGF-IR was driven by the α-actin promoter (16). EGFP expression is visible in these transgenic fish throughout all ages, including adulthood (Fig. 1A). Expression was specifically in skeletal muscle (Fig. 1, B and C) and not in the heart (not shown). No expression was observed in liver (Fig. 1, C and D) or in the brain (not shown), two other potential insulin target tissues. Henceforth, these Tg(acta1:dnIGF1R-EGFP) fish will be referred to as zMIR (zebrafish muscle insulin resistance) for simplicity.

Fig. 1.

Impaired insulin signaling in zMIR (zebrafish muscle insulin resistance) fish. A: expression of the dominant-negative IGF-I receptor (IGF-IR) indicated by enhanced green fluorescent protein (EGFP) fluorescence in a live 3-mo-old fish. B: longitudinal section through the trunk of a 3-mo-old fish highlighting GFP immunostaining (brown) with hematoxylin counterstaining. C: GFP immunofluorscence of highlighted individual muscle bundles. D and E: GFP immunofluorscence in zMIR (D) or nontransgenic (E) fish. GFP is present in skeletal muscle (m) but has no specific signal in liver (li) or intestine (i). F: Western blot for phosphorylated and total Akt in skeletal muscle samples of fasted fish after both feeding and insulin injection. G: analysis of Akt phosphorylation compared with total Akt levels in fasted animals. No difference in basal phosphorylation is observed. H: quantification of Akt phosphorylation relative to total Akt levels in the 3 treatment conditions. Level of Akt phosphorylation in fasted control animals was set to 1. A significant increase in Akt phosphorylation was observed in the control animals with each treatment, but no increase in phosphorylation was observed in the zMIR animals; n = 8–10 animals/group. I: uptake of 2-deoxyglucose (2-DG) into skeletal muscle with and without coinjection of insulin. Level of 2-DG was set to 1 for the control glucose-injected group. A significant increase in 2-DG uptake was seen in control animals with coinjection of insulin but no change in the zMIR animals; n = 8–13 animals/group. All values are means ± SE. *P < 0.05; **P < 0.01; ***P < 0.001.

To determine whether expression of this construct impairs insulin signaling in the skeletal muscle, we examined phosphorylation of Akt, which is used frequently as an indication of insulin receptor activity (Fig. 1F). The levels of Akt phosphorylation were not statistically different between the control and zMIR animals in the fasted state (Fig. 1G). In control animals, feeding increased the Akt phosphorylation compared with fasting levels (p473, P < 0.01; p308, P < 0.05; ANOVA), but no difference was observed in the zMIR animals (Fig. 1H). Like with feeding, insulin injection significantly increased Akt phosphorylation at Ser⁴⁷³ and Thr³⁰⁸ over fasting in control animals (P < 0.001, ANOVA) but not in zMIR animals (Fig. 1H). The data indicate that insulin signaling in the muscle of zMIR fish is compromised but not completely abolished.

Although zebrafish have been shown to be responsive to glucose, it has been unclear whether glucose uptake in skeletal muscle is insulin responsive, in part because an ortholog of the insulin-sensitive glucose transporter GLUT4 has not been clearly identified (38), although it has been identified in brown trout (6). We sought to determine whether glucose uptake into skeletal muscle would be enhanced by insulin. Using a 2-deoxyglucose (2-DG) analog, we found that in nontransgenic control fish there was a significant increase in glucose uptake upon coinjection with insulin (Fig. 1H). However, in zMIR fish there was no significant enhancement of glucose uptake in response to coinjection with insulin. These data support the hypothesis that glucose uptake in zebrafish skeletal muscle is insulin sensitive, which is blunted in the zMIR fish.

Progressive glucose intolerance in zMIR zebrafish.

Given the decreased effectiveness of insulin in the skeletal muscle of zMIR fish, we sought to determine whether the glucose tolerance was compromised. We and others have demonstrated that intraperitoneal glucose tolerance testing is possible in adult zebrafish (10, 23), and no differences between male and female fish have been observed (10), and we have similar results (not shown). However, due to the terminal nature of the blood collection protocol, glucose clearance in an individual animal cannot be determined. We found that at 3 mo of age the zMIR fish had a glucose tolerance similar to that of the sibling controls (Fig. 2A). There also was no significant difference in fasting blood glucose between the zMIR and the control fish (Fig. 2B). Body weight also was not different between the control and zMIR fish (not shown). At 12 mo of age, body weight was similar between the two groups (not shown). However, the zMIR fish had impaired glucose tolerance (Fig. 2C) compared with sibling controls, with a significant difference (P < 0.05, t-test) in blood glucose levels at 90 min postinjection. Again, the fasting blood glucose was not different between the control and zMIR fish (Fig. 2D). These data demonstrate that zebrafish with skeletal muscle insulin resistance can develop glucose intolerance, but the progression is slow, and fasting blood glucose is not altered. The normal glucose tolerance in the young fish suggested a compensatory response in the form of enhanced insulin secretion, increased β-cell number, or both, which is not maintained in the older animals.

Fig. 2.

Progressive glucose intolerance in zMIR fish. A: intraperitoneal (ip) glucose tolerance test in 3-mo-old zMIR and control fish with blood glucose levels determined after fasting (0 min) and 30, 90, and 180 min after injection of 0.5 mg glucose/g body wt; n = 10–14 fish/time point. No difference in glucose tolerance between genotypes was observed. B: fasting blood glucose in 3-mo-old fish. The increase in fasting blood glucose was not statistically significant; n = 10 fish/genotype. C: ip glucose tolerance test in 1-yr-old zMIR and control fish following injection of 0.5 mg glucose/g body wt; n = 8–13 fish/time point. D: fasting blood glucose in 1-yr-old fish. The increase in fasting blood glucose was not statistically significant; n = 47 fish/genotype. All values are means ± SE. *P < 0.05.

β-Cell compensation in zMIR zebrafish.

To determine the number of β-cells, we take advantage of a transgenic line we generated that has the β-cells marked with TagRFP throughout the cell, Tg(−1.2ins:TagRFP), or with mCherry localized to the β-cell nuclei, Tg(−1.2ins:H2BmCherry) (Fig. 3, B–F). In juvenile fish, it is possible to count the absolute number of β-cells in an individual animal. In 7-, 14-, and 21-dpf fish, the β-cell number throughout the large principal islet and any secondary islets was not different in the zMIR fish compared with control siblings (7 dpf, P = 0.7; 14 dpf, P = 0.09; 21 dpf, P = 0.24; t-test; Fig. 3A) and reflects the variability in β-cell numbers at these ages. By 28 dpf, the β-cell number in the zMIR fish was significantly elevated over the sibling controls (P < 0.001; t-test; Fig. 3, A–C). Consistent with the increase in the number of β-cells, we observed an increase in EdU-positive β-cells in the zMIR fish compared with the controls (Fig. 3, H and I), with a significant increase in EdU-positive β-cells in both the principal islet and secondary islets (P < 0.05; t-test) (Fig. 3J). This suggests that the increase in β-cells is through increased β-cell proliferation.

Fig. 3.

Increased no. of β-cells in young zMIR fish. A: quantification of the no. of β-cells during juvenile growth of zMIR (open bars) and control (gray bars) siblings. The no. of β-cells at 7 (n = 18 and 17 zMIR and controls, respectively), 14 (n = 87 and 84), and 21 days postfertilization (dpf) (n = 48 and 47) was not statistically different. At 28 dpf (n = 46 and 52), there was a statistically significant increase in the no. of β-cells in the zMIR animals (P < 0.001, t-test). B and C: principal islet of 28 dpf control (B) and zMIR (C) fish using Tg(−1.2ins:H2BmCherry) to label the β-cells. Scale bars, 20 μm. D and E: examples of pancreatic tissue from 3-mo-old control (D) and zMIR (E) animals. The nuclei of β-cells were labeled by Tg(−1.2ins:H2BmCherry), and the margins of the pancreatic tissue are indicated by the dotted line. Scale bars, 100 μm. F: quantification of β-cells in 3-mo-old fish (n = 10 control, 22 zMIR). The no. of β-cells is normalized to the total pancreatic tissue. A significant increase in β-cells was observed in the zMIR fish (P < 0.01, t-test). G: quantification of β-cells in 1-yr-old fish. No significant difference between genotypes was observed. H and I: 5-ethynyl-2-deoxyuridine (EdU) labeling of β-cells in 28 dpf control (H) and zMIR (I) fish. Each image is a single slice of a confocal stack. Arrows indicate EdU-positive β-cells. Scale bars, 20 μm. J: quantification of β-cell proliferation. An increase in β-cell proliferation was observed in zMIR fish in both the principal and secondary islets; n = 12 for each genotype. All values are means ± SE. *P < 0.05; **P < 0.01.

When quantifying the number of β-cells in the adult fish, we chose to exclude the principal islet from the analysis. The relative contribution of the principal islet and the secondary islets in terms of insulin secretion is unknown. In the adult animal, the secondary islets are embedded in the exocrine tissue (Fig. 3, D and E), similar to mammalian pancreas. In contrast, the principal islet is not embedded in the exocrine tissue and is anatomically separated from the remainder of the pancreas (not shown). By quantifying the total number of β-cells in the secondary islets, we observed a statistically significant increase in the number of β-cells in 3-mo-old fish (P < 0.01, t-test; Fig. 3, D–F), and there was no change in pancreas area between groups (not shown). Using quantitative RT-PCR on RNA isolated from the GI tract or isolated pancreas of 3-mo-old zMIR and control animals, we found that insulin transcript levels were increased in zMIR animals (Fig. 4). This is consistent with the increase in the number of β-cells observed in animals of this age. The increase in β-cells in the 3-mo-old zMIR fish likely explains the normal glucose tolerance seen in these fish. Interestingly, glucagon mRNA is also significantly increased, albeit to a lesser extent than insulin. In fish 1 yr of age we found a similar number of β-cells in the control and zMIR fish (Fig. 3G) again with no difference in pancreas area between groups (not shown). Taking into consideration the elevation in β-cell number in younger animals, this would suggest a progressive decline in the number of β-cells as the zMIR fish age. This decrease in β-cells is also consistent with the abnormal glucose tolerance observed in the 1-yr-old fish.

Fig. 4.

Increased insulin mRNA in zMIR fish. RNA was isolated from the gastrointestinal tract or isolated pancreas of fasted 10-wk-old control and zMIR fish. The transcript levels for insulin (A) and glucagon (B) were compared with amylase as a marker for the endocrine pancreas. A statistically significant increase in the insulin transcripts (***P < 0.001, t-test), as well as a significant increase in glucagon mRNA (*P < 0.05, t-test), was observed; n = 10 fish for each genotype.

Overnutrition increases fasting blood glucose in zMIR zebrafish.

Although zMIR fish at 1 yr of age have impairments in glucose tolerance, the fasting blood glucose is not significantly higher than normal. It is important to note that for all of these studies the fish were maintained on the normal feeding regimen, and the fish were not under any additional nutrition pressure. Under normal feeding protocols in our facility, the average body weight for 3-mo-old fish was 280 ± 57 mg for male fish and 307 ± 62 mg for female fish. At 1 yr of age, average body weight was 349 ± 87 mg for male fish and 364 ± 86 mg for female fish. No differences were observed in body weight between control and zMIR fish at either age. We investigated whether exposure of these fish to overnutrition would uncover differences in fasting blood glucose in zMIR fish compared with normal controls. Using a recently developed method that allows weekly sampling of blood from zebrafish (41), we monitored the effect of overfeeding on fasting blood glucose in control and zMIR fish at 1 yr of age. These fish were housed at a density of six fish per 3-liter tank in mixed groups of males and females with 18 fish per group. For this study group, body weight averaged 331 ± 75 mg for female fish and 324 ± 96 mg for male fish. To establish baseline measures, fish were fed with the normal frequency of two artemia and two flake feedings per day for the first 2 wk. After an overnight fast, fish were weighed, blood was collected, and blood glucose was determined. Subsequently, the fish continued to receive the normal meals, but two additional meals per day were provided for 2 wk: one artemia meal of 500 freshly hatched artemia per fish and one flake meal of 50 mg of flake diet per fish. This represents a five- to 10-fold increase in meal size per fish. After an overnight fast, fish were weighed, blood was collected, and blood glucose was determined. With normal feeding, the body weight did not increase; however, with overfeeding the fish gained significant weight each week (P < 0.001, ANOVA), and there was no difference in weight gain between control and zMIR fish (Fig. 5A). Female fish gain more weight than male fish, but there was no difference in weight gain between control and zMIR fish (Fig. 5B). Final total body weight averaged 596 ± 103 mg for female fish and 434 ± 86 mg for male fish. Fasting blood glucose increased with overfeeding in both groups, consistent with the data from Zang et al. (41). However, by the 2nd wk of overfeeding the zMIR fish had significantly elevated blood glucose compared with the controls (P < 0.01, t-test; Fig. 5C). The fasting blood glucose is greater than that of all zebrafish we have studied so far (not shown). These data suggest that overnutrition exacerbates the glucose intolerance in these fish.

Fig. 5.

Overfeeding increases fasting blood glucose in zMIR fish. A: weight gain. For the 1st 2 wk, fish were maintained on a normal feeding schedule, and no change in body weight was observed. For the 2nd 2 wk, feedings were increased 5- to 10-fold, which induced an increase in body weight. No difference between genotypes was observed. B: weight gain in males and females. Female fish gained more weight than male fish on the overfeeding protocol, but no difference between genotypes was observed. C: fasting blood glucose. An increase in fasting blood glucose was observed with the increased feeding and body weight. After the 1st week of overfeeding there was no difference between genotypes, but in the 2nd week the zMIR fish had a significant elevation of fasting blood glucose compared with controls. All values are means ± SE. **P < 0.01.

DISCUSSION

More than 86 million Americans are prediabetic (7). Although increased physical activity and a healthy diet can delay or even prevent the progression to type 2 diabetes, it has proven to be difficult to achieve sustained change of lifestyle. With the current standard of care, it is projected that 15–30% of the prediabetics will develop type 2 diabetes within 5 years (37). New treatments will require a deeper understanding of the molecular mechanisms underlying the progression. Given the complexity of the process, it is imperative to generate models that recapitulate the progression from insulin resistance to diabetes to unravel the pertinent molecular and cellular events.

We established a transgenic zebrafish model with skeletal muscle insulin resistance that exhibits hallmarks of type 2 diabetes pathogenesis. This initial model exhibits β-cell compensation detectable at 28 days of age. At 3 mo of age, these fish have a 60% increase in β-cell number. This increase in β-cell mass is sufficient for normal glucose tolerance at 3 and 6 mo of age (not shown). We have thus far been unsuccessful in reliably measuring circulating insulin levels (not shown), although these data would suggest an increase in insulin secretion in the zMIR animals. By 1 yr of age, however, the number of β-cells in these insulin-resistant fish declined to control levels. Consequently, they became glucose intolerant, and yet their fasting blood glucose levels are not statistically higher than their sibling controls, indicating that their β-cell mass can sufficiently maintain glucose homeostasis under the regular feeding regime of our facility. However, the transgenic fish have significantly higher fasting glucose after 2 wk of overfeeding, suggesting that their β-cells are compromised and vulnerable.

Our study demonstrates that glucose uptake in zebrafish skeletal muscle is insulin sensitive. Prior to our study, it was unknown whether glucose uptake in teleost skeletal muscle was insulin sensitive. On the one hand, insulin receptor has been found in skeletal muscle, albeit at lower density than in mammals (31), suggesting that the tissue is responsive to insulin. On the other hand, ortholog of the insulin-sensitive glucose transporter GLUT4 has not been identified, with limited exceptions. An extensive search of the zebrafish genome identified 18 GLUT loci, but no GLUT4 (38). A putative GLUT4 has been identified in brown trout (6) and coho salmon (5). The cell surface localization of brown trout GLUT4 is enhanced by insulin in L6 muscle cells, but less so than mammalian GLUT4 (9). A putative glut4 locus was identified in Fugu, Tetraodon, Stickleback, and Medaka, but expression has not been demonstrated (27). To our knowledge, our study is the first to directly demonstrate insulin-stimulated glucose uptake in a Teleost/zebrafish.

The phenotype of transgenic muscle insulin resistance zebrafish is very different from the MKR mice that express dominant-negative IGF-IR. The MKR mice progress from β-cell compensation to diabetes within 10 wk of age (12). One potential reason for the difference is due to species. In our assay, insulin-stimulated glucose uptake is not as pronounced as in mice (19, 21), although there are differences in methodologies. This suggests that although skeletal muscle is insulin sensitive, the role of insulin in glucose uptake is less prominent in zebrafish. This is consistent with the slower glucose clearance (10) and lower density of insulin receptor in teleosts (31). Another possibility is different degrees of insulin resistance in the two models. Although insulin-stimulated Akt phosphorylation is significantly suppressed in our model, it is completely abolished in the MKR mice (12). There are also differences that may underlie the nature of the β-cell dysfunction between these two models. In our zebrafish model, there is a decline in β-cell mass concomitant with the impairment in glucose clearance. In the MKR model, the β-cell mass remains elevated, but the glucose-stimulated insulin secretion from the islets is impaired (2). Determining insulin release from zebrafish islets has not yet been possible, so we are unable to address whether glucose-stimulated insulin secretion is impaired. Despite the differences between the MKR mouse and our zebrafish model, it is interesting to consider that many humans with insulin resistance do not develop diabetes (3). This zebrafish model with the slow progression of glucose intolerance gives us a tremendous opportunity to help understand the genetic susceptibility to diabetes. For example, genome-wide association studies have identified many single nucleotide polymorphisms associated with type 2 diabetes (18, 28). Given the genetic amenability of zebrafish, it is possible to generate loss of function and mimicking of alleles to determine whether these mutations exacerbate the progression to type 2 diabetes in our model.

In conclusion, we show here development of a zebrafish model of skeletal muscle insulin resistance that initially displays a compensatory increase in the number of β-cells and normal glucose tolerance. Over time, the number of β-cells declines, and the fish become glucose intolerant and have elevated fasting blood glucose in conjunction with overnutrition. This model sets up a system where the pathways involved in the progression of insulin resistance to diabetes can be investigated and may lead to better interventional or therapeutic strategies.

GRANTS

This work was supported by the Vanderbilt Diabetes Research and Training Centers and National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) Grant DK-088686 (W. Chen) as well American Diabetes Association Grant 1-13-BS-027 (W. Chen). We utilized the core(s) of the Vanderbilt Diabetes Research and Training Center funded by NIDDK Grant DK-02593, and confocal imaging was performed in the Vanderbilt University Medical Center Cell Imaging Shared Resource (supported by National Institutes of Health Grants CA-68485, DK-20593, DK-58404, HD-15052, DK-59637, and EY-08126).

DISCLOSURES

The authors declare that they do not have any competing or financial interests.

AUTHOR CONTRIBUTIONS

L.A.M. and W.C. conception and design of research; L.A.M., K.E.J., R.M.K., and W.C. performed experiments; L.A.M., K.E.J., R.M.K., and W.C. analyzed data; L.A.M. and W.C. interpreted results of experiments; L.A.M. prepared figures; L.A.M. drafted manuscript; L.A.M. and W.C. approved final version of manuscript; W.C. edited and revised manuscript.