Experimental zebrafish were obtained from the Institute of Hydrobiology, Chinese Academy of Sciences (Wuhan, Hubei, China). Animal experiments and treatments were performed according to the Guide for Animal Care and Use Committee of the Institute of Hydrobiology, Chinese Academy of Sciences (IHB, CAS, Protocol No. 2016–018).

All embryos were obtained by natural fertilization and were incubated in hatching water (4 L water + 6 mL sea salt + 200 μL methylene blue saturated solution) at 28.5°C with no more than 50 embryos per dish. Developmental stages were determined by days post-fertilization (dpf) or months post-fertilization (mpf) (27). Dead embryos were promptly removed at 0–3 dpf, and membranes were removed at 3 dpf. At 5 dpf, all larvae were transferred to standing water aquariums (50 larvae/aquarium/L) and fed with paramecia. From 9 to 15 dpf, they were fed with paramecia and a small amount of newly hatched brine shrimp (Artemia cysts) (Tianjin Fengnian Aquaculture Co., Ltd., Tianjin, China). From embryos to 15 dpf larvae, half of the rearing water was replaces daily. At 15 dpf, all larvae (60 fish/aquarium/10L) were transferred to a circulated water system and fed with newly hatched brine shrimp twice a day at 28.5°C under a 14-h light and 10-h dark photocycle.

Six wild-type zebrafish (3 mpf) were exposed to ice bath anesthesia and their tissues were immediately separated into RNAlater™ solution (AM7020, Invitrogen, USA). The glut2 mRNA tissue distribution analysis was then performed according to the previously described method (28).

According to a previous study, glut2-knockout zebrafish were created using CRISPR/Cas9 technology (29). Briefly, male zebrafish with mosaic mutations (F0) were mated with wild-type zebrafish to create heterozygous F1 offspring; female fish were further purified with another control male to generate an F2 population. F3 fish, with glut2^+/+(control), glut2^+/−, and glut2^−/− genotypes, were obtained from crossing glut2 F2 heterozygotes at 3 mpf. Examination of fish genotypes from the population was carried out as previously described (25). Two effective mutant lines were obtained, and the knockout efficiency was further verified using qPCR analysis, which was performed on 5 dpf F3 population zebrafish (control and glut2^−/−) as previously described (30).

Six pairs of male and female heterozygote F2 zebrafish (6 mpf) from the same parent were selected as parent fish to ensure that the genetic background of the experimental fish was similar. Then, 100 eggs of each pair were randomly selected and incubated to calculate the hatching rate: hatching rate = 100 × (100 – number of dead embryos – number of unbroken embryos until 3 dpf)/100. The survival rate of zebrafish was analyzed from 6 to 21 dpf. The dead zebrafish were timeously collected in PCR tubes and stored at −80°C. All live fish were killed in an ice bath and genotyped together with the dead fish. Here, 236 fish were successfully identified for survival analysis.

Juvenile zebrafish at 30 dpf were anesthetized using MS-222 (Sigma, St Louis, MO, USA), and each fish was immediately photographed and genotyped. The images were imported into ImageJ software for body length measurement, which was standard with a known length bar. Each genotyped zebrafish was then weighed. Here, 105 fish were used for growth performance analysis.

The homozygous fish (F3) could partially survive and self-cross to produce the maternal zygotic mutant glut2 (MZglut2) zebrafish (F4). To ensure that the genetic background of the next experimental fish was similar to the previous experimental fish, the F4 zebrafish (including the control group) used the same pair of male and female heterozygotes F2 zebrafish offspring as their parent fish (F3). The hatching rate and survival analysis were then performed as described above (200 larvae per genotype were used for survival analysis). Only male fish were analyzed in the following study to avoid ovulation cycle bias on lipid and glucose metabolism (31).

Until their adult stage (3 mpf), fish (21 fish for each genotype) were randomly selected and anesthetized using MS-222, and the body length was measured, and the corresponding weight was weighted.

After ice bath anesthesia, the caudal fin was severed with scissors and whole blood collected from the wound was immediately used for blood glucose measurement using a blood glucose meter (Nipro Diagnostics, Inc., Florida, USA). Liver and muscle tissues (two fish for one sample, eight replicates for each genotype) were then immediately harvested and their glucose content was measured according to the instructions of the assay kit (MS2601, Shanghai Cablebridge Biotechnology Co., Ltd., China).

Hepatocyte preparation was performed according to the previous study (32). The control and MZglut2 zebrafish primary hepatocyte (12 fish per genotype at 3 mpf) were separated and incubated in DMEM/F12 medium at 28.5°C in 5% CO2 for 24 h. The medium was then discarded, and PBS was added to slowly wash the cells; the PBS was removed for the glucose uptake assay. The glucose uptake assay in hepatocytes was performed according to assay kit instructions (J13114, Promega, USA). Briefly, the cells were treated with 10 mM 2-deoxyglucose (2-DG) for 10 min at 28.5°C, and their luminescence intensity was measured according to the manufacturer's instructions.

Glucose tolerance tests were performed on the control and MZglut2 fasting (16 h) zebrafish at 3 mpf. Their dynamic blood glucose levels were measured using a glucose meter after ice bath anesthesia using two methods (six fish for each time point). First, fasting zebrafish were immersed in a 3% glucose solution for 3 h, and following hyperglycemic induction, blood glucose concentrations at different time points were detected. Second, zebrafish were fed a high carbohydrate diet (40% dextrin) for 6 days before the experiment was started, and after the diet was consumed for 7 days, blood glucose concentrations were measured at different time points (calculated from the first bite of food). The caudal fin was severed with scissors after ice bath anesthesia, and whole blood collected from the wound was used for blood glucose measurement. The high carbohydrate diet used in this experiment is shown in Supplementary Table 1.

The qRT-PCR analysis was performed as per the detailed steps in our previous study (33). Briefly, total mRNA was extracted using TRIzol reagent according to the manufacturer's instructions (Invitrogen, Carlsbad, USA), and cDNA was synthesized using a cDNA synthesis kit (TransGen Biotech, AE311-03). A real-time quantitative PCR was performed using SYBR Green I Master Mix (Roche, Germany) on a Light-Cycler 480 system (Roche). All mRNA levels were calculated as fold expression relative to the housekeeping gene rpl7. The primers used for qPCR are listed in Table 1. In the present study, since glut2 is thought to be a high Michaelis constant (Km) transporter, intestinal and liver tissue samples were taken at 2 h of high carbohydrate feeding to monitor glucose transporter expression. However, the difference is that we sampled lipid-related gene expression at 6 h after a normal meal, while protein-related gene expression was sampled at 24 h after a normal meal.

Tg (insulin:EGFP) (34) was used to mark β-cells for imaging and/or counting as previously described (35). The β-cells of control and MZglut2 zebrafish were monitored at 5 dpf. Fisetin was considered a GLUT2 inhibitor (36), and a concentration of 120 μm was used to incubate 3–5 dpf Tg (insulin:EGFP) embryos to monitor the β-cells at 5 dpf. Proliferation was analyzed as previously described (35). Briefly, the Click-iT EdU Alexa Fluor 594 Imaging kit (C10339, Invitrogen) was used to identify proliferating β cells. At 4 dpf, 1–2 nL of 0.1 mM 5-ethynyl-2-deoxyuridine (EdU) was injected into the fish heart. After 24 h, the fish were euthanized and fixed in 4% paraformaldehyde to detect the signals. WISH was performed as previously described (37). Antisense digoxigenin-labeled insulin cRNA was synthesized and used in this study.

Control and MZglut2 zebrafish livers were sampled (three livers per sample, six samples per genotype) in a 1.5-mL tube, which immediately froze in liquid nitrogen. These samples were then sent to the Beijing Genomics Institution for HM350 metabolome analysis.

Lipid and crude protein content was measured according to our previous descriptions (33). Briefly, zebrafish were freeze-dried and then ground using a mortar. Next, the chloroform/methanol (V/V, 2:1) extraction technique was used to measure the lipid content of zebrafish. Crude protein content (N × 6.25) was determined after acid digestion using an auto Kjeldahl system (Kjeltec-8400, FOSS Tecator, Haganas, Sweden).

Neutral lipid accumulation was visualized using fluorescent dye staining, Nile red, in live fish as previously described (25, 38). Nile red (N3013; Sigma) was dissolved to a concentration of 0.1 g/mL. The fish were immersed in Nile red at 28.5°C overnight in the dark. Images were taken using an Olympus SZX16 FL stereomicroscope (Olympus, Tokyo, Japan) at an excitation wavelength of 488 nm. Liver triglycerides were measured using commercially available kits (A110-1, Nanjing Jiancheng Bioengineering Institute, China). Fatty droplet accumulation in the liver was visualized using Oil Red O staining as previously described (25).

A Western blot analysis was performed on the liver and muscle tissues of 12 zebrafish from each genotype (one sample mixed with tissues from four fish). The Western blot analysis protocols were performed using the methods described in our previous study (39). The primary antibodies of P-AMPK (1:1000; #2535S; Cell Signaling Technology, Danvers, MA, USA) and β-ACTIN (1:1000; #4970S; Cell Signaling Technology) were used in this study.

For liver and muscle ATP content analysis, 12 individual fish from each genotype were sampled (two fish tissues for one sample, six replicates for each genotype) and measured according to the instructions of the assay kit (A095-1-1, Nanjing Jiancheng Bioengineering Institute, China). The results were standardized using a protein concentration kit (A045-3, Nanjing Jiancheng Bioengineering Institute, China).

For the muscle histological analysis, six individual fish from each genotype were fixed in 4% paraformaldehyde at 4°C for 24 h, followed by routine paraffin sectioning and H&E staining. The cross-section at the base of the cloaca was selected to quantify the total muscle area. Individual total muscle area was determined using the CaseViewer software.

The locomotion tracking analyses were tracked and analyzed as previously described (40, 41). Briefly, after 6 h postprandial, each zebrafish was kept in a tank, and locomotion tracking was recorded for 5 min; the large speed distance and slow-mild speed distance were calculated in each tank for evaluation using the ZebraBox system (ViewPoint Life Sciences, Montreal, QC, Canada).

Oxygen consumption was measured and analyzed as previously described (40, 41). Briefly, four control or MZglut2 male fish were placed in an 1,150-mL conical flask and sealed for 6 h. Dissolved oxygen before and after sealing was measured using a dissolved oxygen respirator (41). Oxygen consumption was calculated as oxygen consumption per unit body weight. The oxygen consumption of 6 h postprandial (hpp) was calculated from 6 h postprandial fish fed with regular food. Basic oxygen consumption was calculated from the fish after starvation and lasted for 2 days. Four groups of each genotype were set as replicates.

Statistical significance was determined using a two-tailed unpaired Student's t-test. Statistical analyses were performed using GraphPad Prism 8.0.1 (GraphPad Software, San Diego, CA, USA). All results are presented as mean ± SEM (standard error of the mean). A P-value of < 0.05 was considered statistically significant.

We first examined glut2 transcriptional levels in different zebrafish tissues. The abundant transcriptional expression of glut2 was observed in the liver, intestine, and kidney, suggesting its physiological role associated with glucose uptake in target organs where glut2 is highly expressed (Figure 1A). The putative mRNA of glut2 is 1515 base pairs in length and encodes 504 amino acids. Using CRISPR/Cas9 technology, glut2 knockout in zebrafish with target site mutation was established by microinjection of recombinant Cas9 protein and gRNA from glut2 target sites. The two knockout lines with 2 and 4 base pair deletions were obtained (Figure 1B). PCR products with genomic DNA and cDNA were used as templates for mutation validation. Premature stopping occurs in glut2^−/− fish from both knockout lines, which both retain 43 correct amino acids (Figure 1C). Transcriptional expression of glut2 in the 5 dpf control and glut2^−/− fish in both line one and line two F3 populations was examined. The downregulated transcriptional expression of glut2 in glut2^−/− fish compared to control fish indicates knockout efficacy (P < 0.05) (Figure 1D); this is usually considered due to nonsense-mediated mRNA decay. In the F3 population of line two, comparable hatching rates of glut2^+/+, glut2^+/−, and glut2^−/− were observed from 0 to 3 dpf (P > 0.05) (Figure 1E). Subsequently, the F3 population survival rate was statistically analyzed, and a significant decrease was observed in glut2^−/− fish, as they died sharply from 9 to 14 dpf (Figure 1F). Intriguingly, approximately 30% of glut2^−/− fish survived to adulthood and could reproduce. Compared to glut2^+/+ and glut2^+/− zebrafish at 30 dpf, the surviving glut2^−/− fish exhibited apparent growth retardation, shortened body length, and decreased body weight observed (P < 0.05) (Figures 1G–K).

Subsequently, the hatching rate and growth performance of control (offspring from natural mating of glut2^+/+ males and females) and MZglut2 fish were compared and analyzed. Hatching rates of MZglut2 fish decreased significantly compared to control fish (P < 0.05) (Figure 2A). This result suggested that maternal glut2 is essential and critical for hatching. Similarly, more than 60% lethality occurred in MZglut2 fish up to 21 dpf, as well as arrested growth performance at 3 mpf (e.g., decreased body length and weight (P < 0.05) (Figures 2B–F).

Adult control and MZglut2 (3 mpf) zebrafish were used for the following analyses. Compared to control fish, MZglut2 fish at 2 hpp showed decreased glucose levels in the blood, muscle, and liver (P < 0.05) (Figures 3A–C). The in vitro 2-DG uptake assay demonstrated that the 2-DG uptake decreased in the liver of MZglut2 fish compared to the control fish (P < 0.05) (Figure 3D). After administration of a 3% glucose solution for 3 h, significant increases in blood glucose were observed in the control and MZglut2 fish at 20 min after 3% glucose deprivation; however, MZglut2 fish blood glucose decreased before and after 3% glucose administration at all time points examined (P < 0.05) (Figure 3E). Dynamic blood glucose was also assessed after a high carbohydrate diet. Blood glucose peaked at 2 hpp in both control and MZglut2 fish, but the MZglut2 fish blood glucose decreased at all time points (P < 0.05) (Figure 3F). The compensatory effect of glucose transporters in the intestine of the MZglut2 fish at 2 hpp of the high carbohydrate diet was detected, as intestinal glut1 transcriptional expressions were upregulated compared to control fish at 2 hpp (P < 0.05) (Figure 3G). Compensatory expression of other glucose transporters in the liver was not observed, as expressions of glut2, glut1, glut8, and glut12 were downregulated in MZglut2 fish at 2 hpp of the high carbohydrate diet (P < 0.05) (Figure 3H). These results suggest that glucose uptake was effectively inactivated in MZglut2 fish.

To visualize and analyze the number of β-cells and insulin content in glut2-deletion fish, the transgenic line Tg (insulin:EGFP) was bred with the glut2^−/− fish. This enables us to obtain the Tg (insulin:EGFP);glut2^+/− fish, which were inbred to generate Tg (insulin:EGFP);glut2^−/− fish. The Tg (insulin:EGFP);MZglut2 fish from the natural mating of Tg (insulin:EGFP);glut2^−/− males and females were used for the β-cells visualization and analysis. Compared with the control fish at 5 dpf, the number of β-cells significantly decreased in MZglut2 fish (P < 0.05) (Figures 4A, B). Fisetin, which has a hypoglycemic effect (42), was administered to wild-type embryos. In wild-type fish at 5 dpf, the Fisetin-treated fish also displayed decreased number of β-cells (P < 0.05) (Figures 4C, D), mimicking a reduced β-cells number of MZglut2 fish. The EdU staining in control fish and MZglut2 fish at 5 dpf was carried out for 24 h, and the statistical analysis of EdU stained β-cells in MZglut2 larvae was decreased (P < 0.05) (Figures 4E, F). These observations were also supported by the WISH results, from which, the decreased insulin signal in MZglut2 fish at 5 dpf was detected (Figure 4G).

A downregulated insulin receptor gene (insra) was found in the liver of MZglut2 fish at 3 mpf (P < 0.05) (Figure 4H). Meanwhile, fatty acid synthesis (chrebp, srebf1, fasn, fads2, and scd) and triglyceride synthesis (dgat1a) were also downregulated (P < 0.05), but upregulated lipolysis genes (atgl and lpl) and fatty acid β-oxidation (FAO) (cpt1aa and cpt1ab) were observed in MZglut2 fish liver at 3 mpf (P < 0.05) (Figures 4H–J). We performed a metabolomic analysis of the liver to obtain an overview of their energy metabolism due to impaired glucose uptake. Carnitines are effective factors to lower lipid content, as they function as key transporters of fatty acids to mitochondria for FAO (43). For metabolite measurements, increased carnitine, adipoylcarnitine, malonylcarnitine, and 3-hydroxylisovalerylcarnitine metabolites were observed in the liver of MZglut2 fish (P < 0.05) (Figure 4K). We observed a significant decrease in lipid content in MZglut2 fish, as evidenced by lipid measurements (P < 0.05) (Figure 4L) and observations of neutral lipids with Nile red staining (P < 0.05) (Figures 4M, N). The biochemical measurement to examine the quantity of liver triglyceride was performed. The MZglut2 fish liver triglyceride content significantly decreased compared with the control fish (Figure 4O). Pathological features of the control and MZglut2 fish liver were examined with histological analysis and Oil Red O staining, from which, the significantly decreased fatty droplet accumulation was observed in MZglut2 fish liver at 3 mpf (P < 0.05) (Figures 4P, Q). These results suggest that glut2-deletion impaired insulin-mediated anabolic metabolism in MZglut2 fish.

Through metabolomics analysis, we found reduced amino acids, such as L-alanine, L-lysine, and L-proline, in the liver of MZglut2 fish (P < 0.05) (Figure 5A). AMPK is known as an energy sensor for activating lipid and protein catabolic metabolism when glucose is deficient. P-AMPK protein levels in the liver and muscle were significantly increased in MZglut2 fish at 24 hpp compared to control fish (P < 0.05) (Figures 5B–D), suggesting a low energy status of MZglut2 fish. Significantly reduced ATP levels were found in the liver and muscle of MZglut2 fish (P < 0.05) (Figures 5E, F). These results provided direct evidence that when glucose uptake was inactivated, insufficient energy supply from glucose activated catabolic metabolism. The bckdk, glud1b, and murf1a genes are known to activate amino acid catabolism and promote ubiquitin-mediated protein degradation (44, 45). All these genes were upregulated in MZglut2 fish muscle (P < 0.05), suggesting an increase in protein and amino acid degradation (Figure 5G). The downregulated expression of mtor in MZglut2 fish muscle not only reflects its attenuated protein synthesis but also agrees with the rise in proteolytic genes. These observations correlated with the decreased crude protein content (P < 0.05) (Figure 5H) and muscle mass (P < 0.05) (Figures 5I, J), which ultimately accounted for growth retardation.

The swimming activity of the control and MZglut2 fish was recorded for 5 min at 6 hpp. The distance traveled by MZglut2 fish was significantly reduced compared to the control fish (Figures 6A, B). Statistical analysis revealed that the slow-mild movement of MZglut2 fish was upregulated, while the large movement of MZglut2 fish was significantly downregulated (P < 0.05) (Figures 6C, D). To assess energy expenditure, we tested oxygen consumption in control and MZglut2 fish. Both basic and 6 hpp oxygen consumption was significantly increased in MZglut2 fish (P < 0.05) (Figures 6E, F).

Together, we found that glut2-deletion caused a decrease in the number of β-cells, which was related to the decrease of insulin-associated anabolic metabolism in surviving zebrafish. However, surviving zebrafish may adapt to blocked glucose uptake by remodeling the metabolic pattern in vivo by reducing insulin-mediated anabolism and enhancing AMPK-mediated catabolism (Figure 7).