Unexpected effect of insulin on glucose disposal explains glucose intolerance of rainbow trout

Abstract

The physiological reasons why salmonids show glucose intolerance are unclear. In mammals, rapid clearance of a glucose load is mainly achieved through insulin-mediated inhibition of hepatic glucose production (R_a) and stimulation of glucose disposal (R_d), but the effects of insulin on R_a and R_d glucose have never been measured in fish. The goal of this study was to characterize the impact of insulin on the glucose kinetics of rainbow trout in vivo. Glucose fluxes were measured by continuous infusion of [6-³H]glucose before and during 4 h of insulin administration. The phosphorylated form of the key signaling proteins Akt and S6 in the insulin cascade were also examined, confirming activation of this pathway in muscle but not liver. Results show that insulin inhibits trout R_d glucose from 8.6 ± 0.6 to 5.4 ± 0.5 µmol kg⁻¹ min⁻¹: the opposite effect than classically seen in mammals. Such a different response may be explained by the contrasting effects of insulin on gluco/hexokinases of trout versus mammals. Insulin also reduced trout R_a from 8.5 ± 0.7 to 4.8 ± 0.6 µmol·kg⁻¹·min⁻¹, whereas it can almost completely suppresses R_a in mammals. The partial inhibition of R_a glucose may be because insulin only affects gluconeogenesis but not glycogen breakdown in trout. The small mismatch between the responses to insulin for R_d (−37%) and R_a glucose (−43%) gives trout a very limited capacity to decrease glycemia. We conclude that the glucose intolerance of rainbow trout can be explained by the inhibiting effect of insulin on glucose disposal.

INTRODUCTION

Glucoregulation is generally considered essential to ensure adequate fuel supply to the brain and to working muscles (44, 51). However, fish do not control glycemia as well as birds or mammals, and they do not rely on stable circulating glucose levels (11, 36). Carnivorous fish like salmonids are known to normalize glycemia very slowly in glucose tolerance tests (20) and seem to show limited sensitivity to insulin (23). These observations are based on measurements of blood glucose concentration that depend on changing rates of glucose disposal (R_d) and hepatic glucose production (R_a). The hormonal regulation of in vivo glucose fluxes has been thoroughly characterized in mammals (50) but remains unexplored in fish, except for the effects of epinephrine (52, 53). In humans, insulin can triple glucose disposal and almost completely suppress glucose production (17, 21). It also lowers glycemia in fish (36), but this response could be mediated through a decrease in R_a, an increase in R_d, or both. Present evidence suggests that fish glucose production is probably inhibited by insulin because gluconeogenic enzymes are downregulated (12, 33, 37). The effects of insulin on the R_d glucose of fish are more difficult to predict. It is possible that R_d is stimulated because insulin increases the expression of fish glucose transporters (GLUTs) in liver and muscle (36, 37). However, it is unclear how the glycolytic enzymes of these two key tissues respond because some studies show inhibition (33, 37), while others report the opposite (11, 34). Using continuous tracer infusion to quantify the respective impacts of the hormone on R_a and R_d glucose would allow the determination of what specific changes in glucose fluxes are responsible for the observed variation in glycemia.

The effects of insulin on glucose metabolism are mediated through several signaling cascades, but the phosphatidylinositol 3-kinase (PI3K)/Akt pathway is the most important among them (5). Insulin binding causes the phosphorylation of tyrosine residues in the β-subunit of the receptor, thereby activating a series of adaptor proteins (4). In turn, these adaptor proteins cause the downstream activation of Akt and S6: two key elements of the PI3K/Akt signaling pathway. Phosphorylated forms of Akt and S6 are associated with the regulation of downstream metabolic processes that include glycogen synthesis, gluconeogenesis, and cell growth (4, 33, 37, 46). Several previous studies have shown that Akt and S6 are activated by insulin in fish (30, 33, 34, 37, 42), but how intravascular insulin infusion affects these proteins has never been measured. Glucagon is another key endocrine signal involved in the regulation of glucose metabolism, and it increases glycemia in fish (36). Therefore, the hypoglycemic state induced by exogenous in vivo administration of insulin could also trigger a glucagon response aiming to restore normoglycemia.

The main goal of this study was to test the hypothesis that insulin reduces glycemia in fish by inhibiting glucose production and stimulating glucose disposal as it does in mammals. However, we anticipated that insulin would have a weaker effect on the glucose fluxes of fish than mammals because fish have a lower capacity for glucoregulation. Our other goals were 1) to measure the levels of phosphorylated Akt and S6 in muscle and liver to determine whether the PI3K/Akt pathway is activated by insulin in these tissues, and 2) to monitor circulating glucagon levels to see if insulin triggers a counterregulatory response.

METHODS

Animals

Rainbow trout (Oncorhynchus mykiss) of both sexes with a Fulton’s condition factor K of 1.15 ± 0.03 (n = 22) [K = (10⁵ × M_b)/L³), where M_b is body mass in grams and L is total body length in mm (3)] were purchased from Linwood Acres Trout Farm (Campbellcroft, Ontario, Canada). No differences were found between females and males and all results are therefore presented as pooled values for both sexes. Two groups of fish used were the following: 1) for in vivo measurements of glucose kinetics by continuous tracer infusion, and 2) for measurements of insulin signaling protein activation by Western blots (physical characteristics for each experimental group are presented in Table 1). The fish were held in a 1,200-liter flow-through tank supplied with dechloraminated Ottawa tap water at 13°C on a 12 h:12 h light-dark photoperiod and were fed Profishnet floating fish pellets (Martin Mills, Elmira, ON, Canada) 5 days/wk. They were acclimated to these conditions for a minimum of 2 wk before experiments. All the procedures were approved by the Animal Care Committee of the University of Ottawa and adhered to the guidelines established by the Canadian Council on Animal Care.

Table 1.

Mean physical characteristics and hematocrit of the 2 groups of catheterized rainbow trout used for in vivo measurements of glucose kinetics or for tissue measurements of insulin signaling proteins

	Glucose Kinetics Experiments (10)	Signaling Proteins Experiments (12)
Body mass, g	452 ± 23	322 ± 29
Body length, cm	33 ± 0.4	30.8 ± 0.7
Hematocrit, %	20.1 ± 0.6	19.6 ± 0.4

Values are means ± SE (sample sizes are in parentheses).

Catheterization and Respirometry

Fish were anesthetized with ethyl 3-aminobenzoate methanesulfonate (60 mg/l MS-222 buffered with 0.2 g/l sodium bicarbonate) and doubly cannulated (for glucose kinetics experiments) or singly cannulated (for signaling protein experiments) with BTPE-50 catheters (Instech Laboratories, Plymouth Meeting, PA) in the dorsal aorta (15). The catheters were kept patent by flushing with Cortland saline containing 50 U/ml heparin (Sigma-Aldrich, St. Louis, MO). Fish were left to recover overnight in a 90-liter swim-tunnel respirometer (Loligo Systems, Tjele, Denmark) where all measurements were carried out in resting animals at a water velocity of 0.5 body length/s. This weak current reduces stress and enhances the flow of water over the gills but does not require swimming to maintain body position while resting at the bottom of the chamber (6). The respirometer was supplied with the same quality water as the holding tank and kept at 13°C. Metabolic rate (Ṁo₂) was measured by intermittent flow respirometry using galvanic oxygen probes connected to a DAQ-PAC-G1 instrument controlled with AutoResp software (Loligo Systems). The probes were calibrated before each experiment using air saturated water (20.9% O₂) (see Ref. 47 for more details on the respirometry measurements).

Glucose Kinetics Experiments

Both catheters were made accessible through the respirometer lid by channeling them through a water-tight port. The rates of glucose production (R_a) and glucose disposal (R_d) were measured by continuous infusion of [6-³H]glucose (222 GBq/mmol; Perkin Elmer, Boston, MA). This tracer method has been validated to quantify glucose kinetics in fish (14). The infusate was freshly prepared immediately before each experiment by drying an aliquot of the solution obtained from the supplier under N₂ and resuspending in Cortland saline. A priming dose of tracer equivalent to 3 h of infusion was injected as a bolus at the start of each infusion to reach isotopic steady state in <45 min. The infusate was then administered continuously at ~1 ml/h (determined individually for each fish to account for differences in body mass) using a calibrated syringe pump (Harvard Apparatus, South Natick, MA). Infusion rates for labeled glucose averaged 2,180 ± 107 Bq·kg⁻¹·min⁻¹ (n = 10), and these trace amounts accounted for 0.00001% of the baseline rate of hepatic glucose production. Blood samples (~100 µl each) were drawn 50, 55, and 60 min after starting the tracer infusion to determine baseline glucose kinetics and every 20 min thereafter during bovine insulin administration (1.5 μg insulin·kg⁻¹·min⁻¹ for 4 h). Insulin from bovine pancreas was purchased from Sigma-Aldrich (lot no. I1882) and solubilized in water acidified to pH 2 with glacial acetic acid. Bovine insulin has been used repeatedly to investigate the effects of teleost insulin in vivo and in vitro (27, 33, 34, 37, 48). The blood sampling schedule is indicated by arrows in Fig. 1. The total amount of blood sampled from each fish accounted for <10% of total blood volume. Samples were collected in tubes containing heparin and aprotinin (500 KIU/ml to stabilize glucagon). They were centrifuged to separate plasma (5 min; 12,000 rpm), which was stored at −20°C until analyses.

Fig. 1.

Experimental design and metabolic rate (Ṁo₂) of resting rainbow trout during the measurement of glucose kinetics. Insulin administration started at time 0 and lasted 4 h. Glucose kinetics were quantified before and during insulin administration by continuous infusion of [6-³H]glucose that started at −1 h. Arrows indicate when blood samples were collected. Ṁo₂ values are means ± SE (n = 10). Blood sampling and insulin had no effect on Ṁo₂ (P = 0.3; one-way repeated-measures ANOVA).

Signaling Protein Experiments

To avoid having to measure signaling proteins in radioactive tissues, these experiments were carried out on different fish than those used for glucose kinetics, but they received the same infusions: saline (control group) or insulin (treatment group; 1.5 μg·kg⁻¹·min⁻¹), which were administered at 1 ml/h through the catheter for 4 h. The animals were then euthanized by a sharp blow on the head before collection of the liver and ~4 g of white muscle anteriorly to the dorsal fin. The tissue samples were stored at −80°C until analyses.

Sample Analyses

Glucose kinetics experiments.

Plasma glucose and glucagon concentrations were measured spectrophotometrically using a Spectra Max Plus384 Absorbance Microplate Reader (Molecular Devices, Sunnyvale, CA). Glucose concentration was quantified using a NAD⁺/NADH-coupled enzymatic assay at 340 nm with hexokinase and glucose-6-phosphate dehydrogenase. Glucagon was measured using a commercial ELISA kit (Crystal Chem, Downers Grove, IL). This kit uses a particular COOH-terminal anti-glucagon fragment that has been previously validated for fish glucagon (26). Unfortunately, fish insulin cannot be measured accurately (25). A radioimmunoassay was developed decades ago (31), but it also measures multiple proinsulins and, therefore, greatly overestimates true insulin concentration. Glucose activity was quantified by drying plasma under N₂ to eliminate tritiated water and resuspending in distilled water. Radioactivity was then measured by scintillation counting (Beckman Coulter LS 6500, Fullerton, CA) in Bio-Safe II scintillation fluid (RPI, Mount Prospect, IL).

Insulin signaling proteins experiments.

Frozen livers and muscle (control: n = 6; insulin: n = 6; 200 mg) from the control and insulin infused rainbow trout were homogenized on ice with a sonicator (Fisher Scientific Sonic Dismembrator model 100, San Diego, CA) in 400 µl of buffer per 100 mg of tissue. During homogenization, samples were kept in a buffer containing 150 mmol/l NaCl, 10 mmol/l Tris, 1 mmol/l EGTA, 1 mmol/l EDTA (pH 7.4), 100 mmol/l sodium fluoride, 4 mmol/l sodium pyrophosphate, 2 mmol/l sodium orthovanadate, 1% (vol/vol) Triton X-100, 0.5% (vol/vol) NP40-IGEPAL, and a protease inhibitor cocktail (Roche, Basel, Switzerland). Homogenates were centrifuged at 15,000 g for 5 min at 4°C, and the resulting supernatants were recovered and stored at −80°C. Protein concentrations were determined using a Bio-Rad protein assay kit (Bio-Rad Laboratories, Munich, Germany) with BSA as standard. A denaturing, nonreducing SDS-PAGE was used to separate proteins. Lysates were diluted in the previously described buffer containing protease inhibitor for a total of 30 µg of total protein for liver and 50 µg of total protein for muscle in 15 μl before 15 μl of 2× Laemmli buffer were added for a total loading volume of 30 μl. The prepared samples were denatured at 95°C for 2 min and quick chilled on ice before loading on the gel. Gels were cast as 10% resolving gel consisting of 5 ml ddH₂0, 2.5 ml buffer B pH 8.8 (1.5 M Tris base, 0.04% SDS; both BioShop, Burlington, ON Canada) dissolved in dH₂O, 2.5 ml 40% acryl/Bis (Bio-Rad, Mississauga, ON, Canada) and polymerized with 50 μl 10% APS (Sigma-Aldrich Oakville, ON, Canada) and 20 μl TEMED (Life Technologies Burlington, ON, Canada), and a 4% stacking gel [consisting of 3.25 ml ddH₂O, 1.25 ml buffer C pH 6.8 (0.5 M Tris, 0.04% SDS dissolved (BioShop) in dH₂O], 0.5 ml 40% acryl/bis polymerized with 25 μl 10% APS, and 10 μl TEMED. Gels were immersed in 1× Tris glycine SDS (TGS) running buffer, consisting of Tris base 2.5 mM, glycine 0.192 M, and 0.1% SDS (all BioShop Canada) dissolved in dH₂O, and samples were loaded with 5 μl of Page Ruler prestained protein ladder (Thermo Fisher, Ottawa, ON, Canada).

Proteins were migrated in the gel at 100 V. After migration, they were blotted onto nitrocellulose 0.45-mm pore size membrane paper (Millipore, Etobicoke, ON, Canada) by wet transfer using the Mini TransBlot system (Bio-Rad) with blotting buffer (250 mM Tris base, 1,920 mM glycine; all BioShop Canada) dissolved in dH₂O, by applying 100 V for 2 h. Membranes were incubated with Odyssey blocking buffer (LI-COR Biosciences Lincoln, NE) for 1 h at room temperature using an orbital shaker. After the blocking step was completed, membranes were cut based on the molecular weight marker to allow separate development of p-Akt (S473) and p-S6 (S235/236) proteins with specific primary antibodies. Partial membranes containing the relevant molecular weight range of proteins were incubated with rabbit raised primary p-Akt (no. 9271) or p-S6 (no.2211) antibodies (Cell Signaling Technology Ozyme, Saint Quentin Yvelines, France), respectively, at a concentration of 1:10,000 on an orbital shaker at 4°C overnight. Membranes were washed four times for 5 min with PBS + 0.1% Tween 20 (Sigma-Aldrich) then incubated with a IRDye Infrared dye (680 nm coupled) secondary goat anti-rabbit IgG antibody (LI-COR Biosciences). Bands were visualized by infrared fluorescence using the Odyssey Imaging System (LI-COR Biosciences) and quantified by Odyssey Infrared imaging system software (v.3.0; LI COR Biosciences).

Both p-Akt and p-S6 protein intensity were normalized to rabbit β-tubulin (no. 2146; Cell Signaling Technologies) intensity and expressed as relative-fold change compared with control groups for liver and muscle. The phosphorylated proteins p-Akt and p-S6 were targeted because they represent the active form of the proteins and have shown activation in response to bovine insulin in trout in vitro (hepatocytes and myocytes) and in vivo (18, 19, 24, 35, 42). Such measurements of phosphorylated protein do not take into account potential changes in total protein, and hence changes in the overall capacity to respond to insulin. The PI3K-dependent phosphorylation of Akt-p in rainbow trout has been previously validated using the PI3K inhibitor LY294002 (38), while mammalian target of rapamycin dependency of p-S6 in rainbow trout has been previously demonstrated using rapamycin (8).

Calculations and Statistics

Glucose fluxes were calculated in two different ways: 1) using either the steady-state, or 2) the nonsteady-state equations of Steele (45). Glucose turnover (R_t) was calculated using the steady-state equation where R_t glucose (in μmol·kg⁻¹·min⁻¹) = labeled glucose infusion rate (F in Bq·kg⁻¹·min⁻¹)/glucose specific activity (SA in Bq/μmol). The nonsteady-state equations were used to calculate R_a and R_d glucose separately after changes in specific activity over time were curve-fitted by second-degree polynomial regression for each animal (54):

$$R_{\text{a}}=\frac{F-\text{pV}\left(\frac{{\text{C}}_2+{\text{C}}_1}{2}\right)\left(\frac{{\text{SA}}_2-{\text{SA}}_1}{t_2-t_1}\right)}{\left(\frac{{\text{SA}}_2+{\text{SA}}_1}{2}\right)}$$

and

$$R_{\text{d}}=R_{\text{a}}-\text{pV}\left(\frac{{\text{C}}_2-{\text{C}}_1}{t_2-t_1}\right)$$

where F is tracer infusion rate (in Bq·kg⁻¹·min⁻¹), pV is pool volume of glucose [50 ml/kg (16)], C₁ and C₂ are circulating glucose concentrations at times t₁ and t₂ (in μmol/ml), SA₁ and SA₂ are glucose specific activities at times t₁ and t₂ (in Bq/μmol), and t₁ and t₂ are the sampling times for two consecutive blood samples (in min) (54). Statistical comparisons were performed using one-way repeated-measures ANOVA with the Dunnett’s post hoc test to determine which means were significantly different from baseline (SigmaPlot v12; Systat Software, San Jose, CA). When the assumptions of normality or equality of variances were not met, Friedman’s nonparametric repeated-measures ANOVA on ranks was used in cases where data transformations failed to normalize the data. Signaling protein levels were analyzed using Mann-Whitney rank sum test. Values are presented as means ± SE, and a level of significance of P < 0.05 was used in all tests.

RESULTS

Metabolic Rate

Ṁo₂ was measured before and during insulin administration (Fig. 1). It remained stable at baseline values throughout the experiments (P > 0.05) and averaged 52.5 ± 1.6 µmol O₂·kg⁻¹·min⁻¹. Therefore, repeated blood sampling (indicated by arrows in Fig. 1) and the infusion of insulin had no stimulating effect on overall oxidative metabolism.

Glycemia and Glucose Kinetics

Changes in plasma glucose concentration and glucose specific activity over time are presented in Fig. 2. The administration of insulin caused a steady decrease in glucose concentration that became significant after 2 h (P < 0.05; Fig. 2A). Glucose specific activity increased progressively from a mean baseline value of 240 ± 20.8 to 452 ± 58.8 Bq/µmol after 4 h of insulin infusion (P < 0.05; Fig. 2B). The effects of insulin on R_t, hepatic R_a, and R_d are shown in Fig. 3. R_t glucose decreased progressively over the 4 h of insulin infusion (P < 0.05; Fig. 3A). The same decrease in glucose fluxes was observed when R_a and R_d glucose were calculated separately using nonsteady-state equations (P < 0.05; Fig. 3B). Differences between initial (baseline) and final values (after 4 h of insulin infusion) for glucose concentration and fluxes are summarized in Table 2. Final values for R_t, R_a, and R_d glucose were much lower than the corresponding initial fluxes (P < 0.001). Figure 4 compares the effects of insulin on the glucose kinetics and glucoregulation capacity of fish and humans (human data from Ref. 21). Insulin inhibits R_d glucose in trout (−37%) but stimulates it in humans (+304%) (Fig. 4A). Insulin inhibits R_a glucose in trout (−43%) and suppresses it almost completely in humans (−99%) (Fig. 4B). Insulin weakly increases the capacity to lower glycemia (R_d glucose – R_a glucose) in trout to a maximal value of 0.5 µmol·kg⁻¹·min⁻¹. It stimulates this capacity much more strongly in humans where it reaches 43 µmol·kg⁻¹·min⁻¹ (Fig. 4C). The R_d – R_a difference ratio between humans and trout increases steadily during insulin administration (Fig. 4D). After a few hours of insulin infusion, the capacity to lower glycemia is 97 times higher in humans than in trout.

Fig. 2.

Effects of insulin on plasma glucose concentration (A) and glucose specific activity (B). Values are means ± SE (n = 10). *P < 0.05, means significantly different from baseline by one-way repeated-measures ANOVA.

Fig. 3.

Effects of insulin on the glucose fluxes of rainbow trout. Fluxes were either calculated with the steady-state equation [turnover rate (R_t); A] or the nonsteady-state equations of Steele [glucose disposal (R_d) and glucose production (R_a); B] (45). Values are means ± SE (n = 10). *P < 0.05, means significantly different from baseline by one-way repeated-measures ANOVA.

Table 2.

Initial (baseline) and final values after 4 h of insulin administration for various parameters of glucose metabolism and for circulating glucagon in rainbow trout

	Baseline	Final (After 4 h of Insulin Infusion)
Glucose, µmol/ml	5.5 ± 0.5	3.9 ± 0.3*
Rt glucose, µmol·kg−1·min−1	9.6 ± 0.8	5.5 ± 0.7†
Ra glucose, µmol·kg−1·min−1	8.5 ± 0.7	4.8 ± 0.6†
Rd glucose, µmol·kg−1·min−1	8.6 ± 0.6	5.4 ± 0.5†
Glucagon, pg/ml	26.5 ± 12.7	216.4 ± 97.6*

Values are means ± SE (n = 10). Glucose turnover rate (R_t) was obtained with the steady-state equation, whereas the rates of appearance (R_a) and disposal (R_d) were obtained with the nonsteady-state equations of Steele (45).

^*P < 0.01, effects of insulin;

^†P < 0.001, effect of insulin by paired t-test.

Fig. 4.

Comparative effects of insulin on the glucose fluxes of rainbow trout and humans. Fish values are compared with human results adapted from Lucidi et al. (21): a similar study where insulin was administered continuously for 3 h. A: relative changes in R_d glucose. B: relative changes in R_a glucose. C: effects of insulin on the capacity to lower glycemia expressed as R_d glucose – R_a glucose. D: changes in the R_d – R_a ratio between humans and trout during insulin administration.

Glucagon

Changes in circulating glucagon concentration are presented in Fig. 5. Insulin caused a counterregulatory increase in glucagon that became significant after ~2 h (P < 0.05; Fig. 5). Initial and final glucagon concentrations are in Table 2.

Fig. 5.

Effects of insulin on circulating glucagon levels in rainbow trout. Values are means ± SE (n = 10). *P < 0.05, means significantly different from baseline by one-way repeated-measures ANOVA.

Insulin Signaling Cascade

The effects of insulin on the active (phosphorylated) form of Akt and S6 in muscle and liver are shown in Fig. 6. In muscle, Western blots reveal that both Akt (+4.6-fold; P = 0.002) and S6 (+7.2-fold; P = 0.009) are strongly activated by insulin. In the liver, fish receiving insulin were unable to elicit a significant activation of p-Akt and p-S6 compared with control animals (P > 0.05 Fig. 6).

Fig. 6.

Relative effects of insulin on the levels of key signaling proteins. For each mean, Western blots are given for the phosphorylated protein (top) and β-tubulin (bottom). Values are means ± SE (n = 6 for each group). *P < 0.01, significant insulin activation by Mann-Whitney rank sum test).

DISCUSSION

This study is the first to show that insulin has the opposite effect on the rate of glucose disposal of rainbow trout compared with mammals (Fig. 4A). Instead of stimulating R_d glucose to reduce glycemia rapidly, insulin inhibits glucose clearance from the circulation in trout. This explains why normalizing glycemia in glucose tolerance tests is ~10 times slower in trout than in mammals where insulin can triple R_d glucose (20). Results also show that insulin inhibits hepatic glucose production in trout, as it does in mammals. However, only partial reduction of R_a glucose was observed here in trout, whereas virtually complete insulin-mediated suppression of glucose production can be achieved by mammals (44). In trout, insulin reduces R_a glucose slightly more than R_d glucose (−43 vs. −37%), and this small mismatch only allows a very slow reduction of glycemia. Hyperinsulinemia also causes a counterregulatory response in fish by raising glucagon levels, and it activates the signaling proteins Akt and S6 in white muscle.

Inhibition of Glucose Disposal

Contrary to expectation, insulin inhibits glucose disposal in trout and, therefore, induces the opposite response classically seen in mammals (Figs. 3B and 4A). What mechanism could explain this striking difference? Glucose disposal can be modulated by altering intracellular glucose metabolism (glycolysis, glucose oxidation, and glycogen synthesis) and/or transmembrane glucose transport that feeds these various pathways. Intracellular glucose phosphorylation by hexokinases and glucokinase plays an important role in the regulation of glycolytic flux (28, 51, 56). Therefore, the decrease in glucose disposal seen here in trout could be mediated through the reduced expression of muscle hexokinases and (liver) glucokinase, as previously reported in trout after insulin administration (33, 37). The fact that insulin increases hexokinase and glucokinase expression, concentration, and activity in mammals (28, 49) could explain why R_d glucose responds so differently between the two groups of animals. How the concentration and activity of these enzymes respond to insulin in trout has not been assessed directly but present results suggest that they probably decrease. Slowing down phosphorylation causes an increase in intracellular free glucose that reduces the transmembrane concentration gradient driving inward glucose transport via GLUTs. Surprisingly, reported responses for other aspects of glucose metabolism fail to explain why R_d glucose responds so differently between trout and mammals. They support the notion that trout R_d glucose should be stimulated by insulin because glycogen synthase activity (muscle) (37, 41) and the expression of GLUTs (liver and muscle) (33, 37, 41) are stimulated by the hormone in both groups of animals. Unfortunately, the limited volume of blood that we could sample was insufficient to be able to assess potential responses from other hormones such as glucagon-like-peptide or growth hormone that might help to explain R_d glucose inhibition in trout. Overall, the information presently available suggests that the contrasting effects of insulin on the R_d glucose of trout and mammals depend on the opposite actions of the hormone on hexokinases.

Inhibition of Glucose Production

Insulin inhibits R_a glucose in trout and mammals, although less so in trout (Figs. 3B and 4B). Downregulating hepatic glucose production can be achieved by reducing glycogen breakdown, gluconeogenesis, or both (6, 7). The weaker R_a response of trout could be explained by the fact that insulin impairs glycogen breakdown in mammals but may not do so in fish. The reciprocal activities of glycogen phosphorylase and glycogen synthase determine the rate of net glycogen synthesis or breakdown (29). In mammals, insulin strongly inhibits glycogen phosphorylase and stimulates glycogen synthase (9, 44), but these enzymes seem to be unresponsive in trout (37). Therefore, trout probably maintain baseline glycogen breakdown when insulin is elevated, making the full suppression of R_a glucose impossible.

The weaker inhibiting effect of insulin on the R_a glucose of trout (compared with mammals) could also be linked to differences in the regulation of gluconeogenesis. Insulin clearly downregulates key gluconeogenic enzymes like phosphoenolpyruvate carboxykinase, fructose-1,6-bisphosphatase, and glucose-6-phosphatase in mammals (41), and there is evidence that the same enzymes could be inhibited in trout (37). However, many gene duplication events have happened in the evolutionary history of trout, resulting in multiple forms of these enzymes. A recent study suggests that different gluconeogenic isoforms could be regulated differently (22). If the results of Polakof et al. (37) only apply to a single isoform, they may not express the net effect of insulin on gluconeogenic flux. Overall, therefore, the weak inhibition of glucose production shown by trout can be attributed to the lack of response by glycogen phosphorylase and glycogen synthase and, possibly, to the atypical regulation of gluconeogenic isoforms.

Capacity to Clear Glucose

Insulin decreases glycemia in both trout and mammals but at vastly different rates (21, 36). The capacity of insulin to lower blood glucose can be expressed as the difference between disposal and production (R_d glucose – R_a glucose), and it is shown in Fig. 4 that compares humans with trout. Humans are rapidly able to achieve a R_d − R_a difference of 43 µmol·kg⁻¹·min⁻¹ (21), whereas the maximal value measured here in trout was only 0.5 µmol·kg⁻¹·min⁻¹ (Fig. 4C). After 90 min of insulin administration, the capacity to clear glucose was 97 times higher in humans than in trout (Fig. 4D), and this value is probably an underestimation of the true ratio because less insulin was given in the human experiments.

Such differences in the capacity to clear glucose may exist because carnivorous salmonids naturally consume a low-carbohydrate diet and only experience hyperglycemia very rarely. However, many omnivore or herbivore fish species show the same limited capacity for clearing glucose as rainbow trout. Also, persistent hyperglycemia may not be as harmful to trout as mammals. In diabetic humans, it can cause protein glycosylation that leads to retinopathy, neuropathy, renal failure, and atrial fibrillation (1, 55), but it is unclear whether similar outcomes occur in trout. Chronic hyperglycemia can cause retinopathy in zebrafish (Danio rerio) (13) or hemoglobin glycosylation and insulin resistance in Indian perch (Anabas testudineus (2). However, cave-dwelling fish populations of Astyanax mexicanus are hyperglycemic throughout their life and do not show any health complications (39). Overall, the capacity to clear glucose rapidly is most likely not needed in rainbow trout because they normally eat little glucose and/or easily tolerate hyperglycemia.

Activation of Insulin Signaling Pathway

Insulin was able to activate Akt and S6 in muscle (Fig. 6), but this stimulation of the PI3K/Akt-signaling pathway did not increase R_d glucose as it does in mammals. Instead, trout R_d glucose decreased, and it is unclear why this should be the case. Most of the detailed information on this signaling pathway comes from nonpiscine models, and some but not all of its components are known in fish (4). The uncharacterized parts of the insulin signaling cascade of trout could therefore be quite different from mammals. The target genes or downstream enzymes regulated by the cascade could also have evolved differently. The opposite effects of insulin on gluco/hexokinases of trout (inhibition) and mammals (activation) (37) are somewhat puzzling. This is because these enzymes are modulated by sterol regulatory element-binding protein-1 (SREBP-1) (40) and insulin stimulates SREBP-1 in both trout and mammals (19, 40). SREBP-1 activation may have opposite effects on the expression of these kinases in trout versus mammals. Unknown components of the insulin signaling cascade or nonmammalian regulation of gluco/hexokinases by SREBP-1 could therefore explain why trout activate the muscle PI3K/Akt cascade while decreasing R_d glucose.

Insulin did not affect the liver PI3K/Akt cascade in our experiments (Fig. 6), but several other studies have reported activation (30, 34, 37). It is unclear why the liver failed to respond, but these discrepancies might be explained by varying modes of hormone administration: intraperitoneal bolus injection (30), 11-day-osmotic pump intraperitoneal infusion (34), intravascular bolus injection (37), and a 4-h intravascular infusion (this study). Therefore, differences in insulin concentrations around the liver and in the time course of their changes may have affected whether the PI3K/Akt cascade responds or not.

Stimulation of Glucagon Release

Insulin causes a large increase in circulating glucagon concentration (Fig. 5), and this response could be mediated by hypoglycemia. Glucosensing neurons and A cells in the Brockmann bodies detect declining glucose levels (32) and could stimulate glucagon release to counteract the effects of insulin. Additionally, Glucagon increases the expression and release of somatostatins 14 and 25 (10): hormones that indirectly promote hyperglycemia by inhibiting insulin (43). Therefore, counterregulation of hyperinsulinemia probably also includes somatostatins in rainbow trout.

Perspectives and Significance

This study is the first to characterize the integrated in vivo effects of insulin on fish glucose metabolism. It shows that rainbow trout have a unique response to the hormone: the inhibition of glucose disposal (Fig. 3). This unexpected effect of insulin has not been documented in other animals, especially mammals that stimulate R_d glucose by severalfold instead (Fig. 4). These interspecific differences may be explained by the contrasting effects of insulin on the gluco/hexokinases of trout (inhibition) versus mammals (activation) or to other, yet unknown, reasons. The results also show that insulin reduces hepatic glucose production in trout, whereas mammals can achieve complete suppression. This partial reduction of R_a glucose may be because insulin does not affect glycogenolysis in trout and only inhibits gluconeogenesis, whereas mammals shut down both pathways. The integrated actions of insulin that lead to reducing glucose fluxes in trout (R_a slightly more than R_d) only provide them with a very limited capacity to decrease glycemia because R_d − R_a remains extremely small. We conclude that the glucose intolerance classically exhibited by rainbow trout can be explained by the inhibiting effect of insulin on glucose disposal.

GRANTS

This work was supported by Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery Grant 105639-2012, NSERC Research Tools and Instruments Grant 315429-05 (to J.-M. Weber), NSERC Discovery Grant 2114456-2017 (to J. Mennigen), and Canadian Foundation for Innovation John Evans Leader’s Fund Grant 148035.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

J.L.F. and J.-M.W. conceived and designed research; J.L.F. and D.J.K. performed experiments; J.L.F., D.J.K., and J.-M.W. analyzed data; J.L.F., D.J.K., J.A.M., and J.-M.W. interpreted results of experiments; J.L.F. and J.-M.W. prepared figures; J.L.F. and J.-M.W. drafted manuscript; J.L.F., J.A.M., and J.-M.W. edited and revised manuscript; J.L.F., D.J.K., J.A.M., and J.-M.W. approved final version of manuscript.