Effects of prior exercise on insulin-mediated and noninsulin-mediated glucose uptake in horses during a hyperglycaemic clamp


  • R. J. GEOR,

    Corresponding author
    • Present addresses: Dr Geor, Department of Large Animal Clinical Sciences, Michigan State University, D202 Veterinary Medical Center, East Lansing, Michigan 48824, USA. Dr McCutcheon, Department of Pathobiology and Diagnostic Investigation, Michigan State University, F130G Veterinary Medical Center, East Lansing, Michigan, 48824, USA. Email: geor@cvm.msu.edu

    Search for more papers by this author

    1. Departments of Biomedical Sciences and Pathobiology, Ontario Veterinary College, University of Guelph, Ontario, Canada
    Search for more papers by this author

    1. Departments of Biomedical Sciences and Pathobiology, Ontario Veterinary College, University of Guelph, Ontario, Canada
    Search for more papers by this author


Reasons for performing study: There is limited information about factors regulating glucose utilisation post exercise in horses.

Objectives: To determine the effects of a single bout of moderate intensity exercise on measures of insulin-mediated (IMGU) and noninsulin-mediated (NIMGU) glucose uptake during a hyperglycaemic clamp (HC).

Methods: Hyperglycaemic clamps were administered in random order to 8 Standardbreds under 4 conditions: 1) rest, insulinopenia (R-L); 2) rest, hyperinsulinaemia (R-H); 3) post exercise (45 min at ∼50% VO2peak), insulinopenia (Ex-L) and 4) post exercise, hyperinsulinaemia (Ex-H). In the R-L and Ex-L trials, somatostatin was infused to suppress insulin secretion and induce insulinopenia. After 30 min, a 2 h HC was initiated with plasma glucose concentrations maintained at ∼10 mmol/l by variable glucose infusion. In R-H and Ex-H, regular insulin (1.0 mu/kg bwt/min) was also administered to induce physiological hyperinsulinaemia. Serum insulin and C-peptide concentrations were measured in samples obtained at 10 min intervals. Glucose uptake was calculated from mean glucose infusion rate (GIR) during the last 60 min of the HC.

Results: In all HCs C-peptide remained below baseline concentrations, evidence of suppression of insulin secretion by somatostatin. Overall, mean ± s.e. insulin concentrations during the final 60 min of the HC in R-L and Ex-L were 5.7 ± 1.1 and 6.9 ± 1.9 mu/l respectively, and corresponding values in R-H and Ex-H were 64.1 ± 11.1 and 61.2 ± 10.9 mu/l. Prior exercise affected IMGU but not NIMGU. Over the final 60 min of the HC, mean GIR was higher (P<0.001) in Ex-H (5.6 ± 1.1 mg/kg bwt/min) than in R-H (3.3 ± 0.9 mg/kg bwt/min), whereas mean GIR did not differ (P = 0.26) between R-L (1.2 ± 0.3 mg/kg bwt/min) and Ex-L (1.8 ± 0.5 mg/kg bwt/min).

Conclusions: A single bout of moderate intensity exercise increased glucose uptake during a hyperglycaemic clamp under hyperinsulinaemic conditions but not under somatostatin-induced insulinopenia.


Exercise leads to adaptations in the post exercise state that facilitate replenishment of glycogen stores in skeletal muscle (Hamilton et al. 1996; Richter 1996) and liver (Galassetti et al. 1999). Studies in rodents (Richter et al. 1982; Cartee et al. 1989) and man (Wojtaszewski et al. 2002; 2003) have demonstrated that prior exercise or muscle contraction increases the sensitivity of skeletal muscle to insulin stimulation, an adaptation that allows for enhanced glucose uptake and repletion of glycogen stores (Richter et al. 1982). One previous study in horses reported that whole-body insulin sensitivity, measured by the hyperinsulinaemic-euglycaemic clamp, was not increased during a 24 h period after a single bout of exercise (Pratt et al. 2007), potentially explaining the slow rate of post exercise replenishment of muscle glycogen observed in horses compared with other species (Hyyppa et al. 1997; Lacombe et al. 2004). This observation notwithstanding, there is limited information about factors regulating glucose metabolism in horses during the post exercise state.

Glucose uptake occurs via 2 mechanisms: 1) insulin-mediated glucose uptake (IMGU), which occurs in insulin-sensitive tissues; and 2) noninsulin-mediated glucose uptake (NIMGU), which occurs in both insulin-sensitive and noninsulin-sensitive tissues (Baron et al. 1988). The objective of the study reported here was to determine the effects of a single bout of moderate-intensity exercise on measures of IMGU and NIMGU in horses during hyperglycaemia. Hyperglycaemic clamps were performed during physiological hyperinsulinaemia (insulin concentration approximately 60–70 mu/l) and during somatostatin-induced insulinopenia (<10 mu/l) to measure IMGU and NIMGU, respectively (Baron et al. 1988; Araújo-Vilar et al. 1997). It was hypothesised that prior exercise would increase both IMGU and NIMGU under hyperglycaemic conditions.

Materials and methods


All experimental procedures were approved by the University of Guelph Animal Care and Use Committee. Eight matureStandardbred horses (4 mares and 4 geldings; mean ± s.d. 436 ± 29 kg bwt, age 4–6 years) that had undergone 6 weeks of physical conditioning by running on a treadmill1 before the experimental trials were used in the study. Horses were housed in box stalls (3.5 × 4 m) and fed a daily ration of 5–6 kg mixed grass hay and 3.5–4 kg of a pelleted feed (12.5% CP, 8.1% fat, 26.5% NSC on a DM basis). Water and loose salt were provided ad libitum. Horses were turned out into a dry lot for ∼3 h daily. Before the experiments, horses completed an incremental treadmill (3° incline)1 exercise test for measurement of the peak rate of oxygen consumption (VO2peak), as previously described (Pratt et al. 2007). For each horse, linear regression analysis was used to determine the treadmill belt speed corresponding to 50% VO2peak and this speed was used for the 2 exercise protocols.

Experimental design

A replicated 4 × 4 Latin square design was used to determine the effects of prior exercise and insulin state on whole-body glucose uptake. A 2 h hyperglycaemic (10 mmol/l) clamp (HC) was administered under each of 4 conditions: 1) resting state, insulinopenia (R-low insulin [L]); 2) rest, hyperinsulinaemia (R-high insulin [H]); 3) post exercise, insulinopenia (Ex-L) and 4) post exercise, hyperinsulinaemia (Ex-H). For each horse, the order of trials was randomised and there was a 7 day interval between trials.

Experimental protocol

All trials were preceded by 48 h of rest (no turnout or treadmill exercise). Feed was withheld starting at 20.00 h the day before each trial. At 07.00 h, catheters (Angiocath, 14 g, 11.5 cm)2 were inserted into both jugular veins. Prior to catheter insertion, the overlying skin was aseptically prepared and desensitised by the subcutaneous injection of 2% mepivacaine (Carbocaine)3. One of the catheters was subsequently used for infusion of somatostatin, glucose and insulin and the other for collection of blood samples. The period of rest (R-L and R-H trials) or exercise (Ex-L and Ex-H trials) was undertaken between 07.45 and 08.30 h. The exercise protocol was completed on a treadmill set at a 3° incline and consisted of a 5 min warm-up at 4 m/s, followed by 45 min at a speed that elicited 50% VO2peak, and a 5 min cool down (0° incline) at 1.6 m/s. Horses undergoing the rest treatments stood quietly on the treadmill for an equivalent period of time. The horses were offered fresh water after completion of the exercise protocol. Mean ± s.e. room temperature and relative humidity were, respectively, 20.5 ± 1.3°C and 53.2 ± 5.6% and did not differ among the trials.

Ten minutes (t=−40 min) after exercise or rest, horses were positioned in padded stocks and 2 blood samples (t=−40 and −30 min) were collected for measurement of baseline plasma glucose and serum immunoreactive insulin and C-peptide concentrations. Subsequently (t=−30 min), an infusion (Vet IV 2.2 peristaltic pump)4 of somatostatin ([D-Trp8]-somatostatin-14)5 was initiated (500 µg bolus administered in <30 s, followed by a constant-rate infusion of 1 µg/kg bwt/h for 150 min) to suppress pancreatic insulin secretion before and during the HC procedures (Tóth et al. 2010). At t=−10 min in R-H and Ex-H, insulin (Humulin-R)6 was infused4 (priming dose of 9 mu/kg bwt followed by a constant-rate infusion at a rate of 1.0 mu/kg bwt/min) to induce moderate hyperinsulinaemia (serum insulin ∼60–70 mu/l) during the HC. Further blood samples were collected at t=−20, −10 and −1 min for measurement of glucose, insulin and C-peptide concentrations. At t= 0 min, after 30 min of somatostatin infusion, hyperglycaemia was acutely achieved (target blood glucose concentration of 10 mmol/l) by a priming infusion (KD Scientific Precision Syringe Pump)7 of glucose (50% wt/vol dextrose). The priming dose (0.1 g glucose/kg bwt) was given i.v. within 1 min. Subsequently, blood glucose concentration was maintained at the target concentration with a variable-rate dextrose infusion using the glucose clamp method. Additional blood samples (∼3 ml) were drawn at 5 min intervals throughout the HC for determination of whole blood glucose concentration by the glucose oxidase method using a YSI 2300 Glucose-Lactate Analyzer8. The rate of dextrose infusion was adjusted if blood glucose concentration deviated by >0.5 mmol/l from target hyperglycaemia. Blood samples (∼15 ml) for subsequent determination of plasma glucose and serum immunoreactive C-peptide and insulin concentrations were obtained every 10 min during the HC. Blood was transferred to Vacutainers9 containing sodium heparin anticoagulant or no additive. The samples were centrifuged at 1500 g for 15 min and the harvested plasma or serum stored at −80°C until analysis. Infusions of somatostatin, insulin and dextrose were terminated at the end of the 120 min HC. The horses were closely monitored for a further 45 min, including regular measurement of blood glucose concentrations. Thereafter, horses were returned to their stall and given feed and water.

Sample analysis

Plasma glucose concentrations were measured in duplicate by use of the YSI glucose analyser. Serum immunoreactive insulin concentrations were determined in duplicate by use of a radioimmunoassay (Coat-a-Count)10 previously validated for equine insulin (Freestone et al. 1991). The mean intra- and interassay coefficients of variation (CV) were 4.5 and 7.1%, respectively. Serum immunoreactive C-peptide concentrations were determined in duplicate by use of a human double antibody radioimmunoassay (Human C-Peptide RIA Kit)11 validated for use with equine sera (Tóth et al. 2010). For each horse, samples from all 4 trials were analysed in a single assay session. Mean intra-assay CV was 7.0%.


The first hour of the HC was considered an equilibration period with data from the final 60 min used to calculate the mean glucose infusion rate (GIR) required for maintenance of target hyperglycaemia during the 60–90 min and 90–120 min periods. It was assumed that hepatic glucose output was completely suppressed by hyperglycaemia or hyperinsulinaemia (DeFronzo et al. 1979; Rijnen and van der Kolk 2003). Under these conditions, the amount of glucose infused equals the amount of glucose utilised and therefore mean GIR can be taken as a measure of whole-body glucose uptake. During somatostatin-induced insulinopenia (R-L and Ex-L), it was assumed that mean GIR represented NIMGU. During the hyperinsulinaemic HCs (R-H and Ex-H), mean GIR was assumed to represent IMGU, although under these conditions a small component of glucose uptake may be noninsulin-mediated (Baron et al. 1988). Calculations of glucose uptake were not corrected for any urinary glucose excretion.

In R-H and Ex-H, an index of insulin sensitivity was obtained by dividing the mean GIR during the final 60 min of the HC by the mean insulin concentration during the same period. This index of insulin sensitivity represents the combined effects of hyperglycaemia and hyperinsulinaemia on glucose uptake (Henriksen et al. 2000).

Statistical analysis

The effects of treatment, time within treatment and the interaction of treatment and time on plasma glucose, serum immunoreactive insulin and C-peptide and mean GIR were assessed by mixed model analysis of variance with repeated measures using a statistical software program (SAS Version 9.1)12. Paired students t test was used to compare insulin sensitivity values from R-H and Ex-H. Normality was tested using the Shapiro-Wilk statistic. Log transformation was used to correct non-normally distributed data sets. Differences were considered significant at P<0.05 and when fixed effects were significant, pre planned comparisons of the means were made using the pdiff option. Data are expressed as mean ± s.e.


The mean VO2peak of the horses was 123.3 ± 15.1 ml/kg bwt/min and mean running speed corresponding to 50% VO2peak during the Ex-L and Ex-H trials was 5.7 ± 0.3 m/s. All horses completed each of the 4 trials with no adverse effects observed during the experimental procedures.

Basal plasma glucose and serum immunoreactive insulin and C-peptide concentrations did not differ among trials (Fig 1a–c). In samples collected 10 min after completion of exercise or rest (t=−40 min in Fig 1), mean serum insulin (treatment × time: P<0.01) and C-peptide (treatment × time: P<0.001) were lower in Ex-L and Ex-H when compared to R-L and R-H. Conversely, at t=−40 min plasma glucose concentration was significantly higher in Ex-L and Ex-H than in R-L and R-H. In all trials, mean C-peptide decreased (treatment × time: P<0.001) during the first 30 min after initiation of somatostatin infusion and remained significantly below baseline/pre-infusion values throughout the HC procedures (Fig 1b). In R-L and Ex-L, mean serum insulin concentrations were at or below baseline values throughout the HC, whereas, in R-H and Ex-H, serum insulin concentrations increased linearly to reach a plateau of ∼60–65 mu/l between 40 and 120 min of the HC (Fig 1a). Overall mean insulin concentrations during the final 60 min of the HC in R-L and Ex-L were 5.7 ± 1.1 mu/l and 6.9 ± 1.9 mu/l respectively, and corresponding values in R-H and Ex-H were 64.1 ± 11.1 mu/l and 61.2 ± 10.9 mu/l. In all trials, plasma glucose concentration was successfully clamped near the target of 10 mmol/l during the final 60 min of the HC (Fig 1c). Overall mean plasma glucose concentrations during the final 60 min of the HC were 10.5 ± 0.2 mmol/l, 10.6 ± 0.1 mmol/l, 10.2 ± 0.2 mmol/l and 10.3 ± 0.2 mmol/l for R-L, R-H, Ex-L and Ex-H, respectively, with no difference among trials.

Figure 1.

Mean ± s.e. (n=8 horses/treatment) serum immunoreactive insulin (A), serum immunoreactive C-peptide (B), and plasma glucose (C) concentrations in the 4 treatments (R-L=inline image; Ex-L=▴; R-H=○; Ex-H=●). R/Ex=rest vs. exercise period; SRIF=period of somatostatin infusion at rest.

Insulin state and prior exercise affected whole-body glucose uptake as estimated by the mean GIR required for maintenance of hyperglycaemia (Fig 2). In each trial, there was no significant effect (P>0.7) of time during the HC (60–90 min vs. 90–120 min) on GIR. Over the final 60 min of the HC, mean GIR was higher (P<0.001) in Ex-H (5.6 ± 1.1 mg/kg bwt/min) than in R-H (3.3 ± 0.9 mg/kg bwt/min), whereas mean GIR did not differ (P = 0.26) between R-L (1.2 ± 0.3 mg/kg bwt/min) and Ex-L (1.8 ± 0.5 mg/kg bwt/min). The insulin-stimulated increase in glucose uptake, calculated as the difference between GIR during insulinopenia and GIR during elevated insulin (i.e. L vs. H trials), was higher (P<0.01) after exercise (3.9 ± 0.6 vs. 2.1 ± 0.5 mg/kg bwt/min). Insulin sensitivity was higher (P<0.001) in Ex-H (9.22 ± 1.3 × 10−2 mg/kg bwt/min per mu/l) than in R-H (5.05 ± 0.93 × 10−2 mg/kg bwt/min per mu/l).

Figure 2.

Mean±s.e. (n=8 horses/treatment) glucose infusion rate (GIR) during the hyperglycaemic clamp (HC) in the 4 treatments. For each treatment, mean GIR did not differ between the 2 periods of assessment (60–90 min and 90–120 min).*Significant difference (P<0.001) between R-H and R-L. aSignificant difference (P<0.001) between Ex-H and Ex-L.Significant difference (P<0.001) between Ex-H and R-H.


The present study examined the impact of a single bout of moderate intensity exercise and insulin state on whole-body glucose uptake in horses during hyperglycaemia. The experimental design, with infusion of somatostatin to suppress endogenous insulin secretion, allowed measurement of glucose disposal at low (<10 mu/l) and physiologically high (∼60–65 mu/l) insulin concentrations and therefore enabled estimation of both IMGU and NIMGU. The results of the present study provide evidence that an acute bout of moderate intensity (50% VO2peak) exercise enhances the ability of physiological hyperinsulinaemia to stimulate glucose uptake during the early post exercise period. However, we did not detect a significant effect of exercise on NIMGU. Previous studies in man have demonstrated that both IMGU and NIMGU occur predominantly in skeletal muscle under hyperglycaemic (>10 mmol/l) conditions (DeFronzo et al. 1981, 1983; Baron et al. 1988), while research in dogs has shown that prior exercise does not alter the percent of whole body glucose uptake ascribed to hepatic extraction (Pencek et al. 2003). Although muscle glycogen concentrations were not measured in the present study, our findings may be relevant to the understanding of factors affecting the availability of glucose for muscle glycogen synthesis in horses after exercise.

The interpretation of the present data is dependent on a number of assumptions that deserve comment. First, during the R-L and Ex-L trials it was assumed that insulin action was absent. However, consistent with previous observations in man (Hwu et al. 2001) and horses (Tóth et al. 2010), insulin secretion was not completely blocked by somatostatin infusion as evidenced by continued detection of immunoreactive C-peptide throughout experimental periods. Nonetheless, C-peptide concentrations were suppressed by >75% in all trials and in R-L and Ex-L serum insulin concentrations were below baseline values at most time points during the hyperglycaemic clamp. A human study reported that glucose uptake during a hyperglycaemic clamp did not differ when prevailing insulin concentration was approximately 10 mu/l vs. 5 mu/l, suggesting that basal insulin concentrations have little impact on glucose uptake under these conditions (del Prato et al. 1995). Therefore, we considered the measurements of whole-body glucose uptake in R-L and Ex-L to primarily reflect NIMGU.

The current approach was also based on the assumption that somatostatin exerts no influence on the glucoregulatory system other than that mediated by the suppression of insulin. Somatostatin is known to inhibit the secretion of a number of other hormones, including glucagon, growth hormone (GH) and various gastrointestinal hormones (Low 2004). Therefore, studies in man (Henriksen et al. 2000) and dog (Pencek et al. 2003) have often employed replacement infusions of glucagon and GH to maintain basal concentrations of these hormones during somatostatin infusion. However, there is evidence in both species (Shulman et al. 1978; Sacca et al. 1982) that short-term (<3 h) deficiencies of glucagon and GH have minimal impact on glucose production and disposal. Therefore, the available evidence supports the validity of the current approach for examination of the role of insulin in regulation of glucose disposal in horses.

There is robust evidence in man and rodents that a single bout of exercise causes an increase in insulin sensitivity to glucose uptake, measured by use of glucose clamp or minimal model techniques (Cartee et al. 1989; Wojtaszewski et al. 2002; Hayashi et al. 2005; Bordenave et al. 2008). Similarly, in the horses of the present study, prior exercise was associated with an increase in insulin-mediated glucose disposal with a >70% increase in mean GIR in Ex-H when compared to R-H (Ex-H vs. R-H; Fig 2) and almost 2-fold higher insulin sensitivity in Ex-H than in R-H. Treiber et al. (2006) observed markedly increased minimal model insulin sensitivity in horses during low-intensity exercise, while other studies in horses have reported higher insulin sensitivity 24 h after 5–7 consecutive days of physical conditioning (Powell et al. 2002; Stewart-Hunt et al. 2006). In contrast to the findings of the present study and reports in other species, Pratt et al. (2007) found no increase in insulin sensitivity measured at 0.5, 4 and 24 h after a single bout of exercise in trained horses. Several factors, including differences in the technique for measurement of insulin sensitivity, level of insulin stimulus (hyperinsulinaemia) and exercise workload and the magnitude of glycogen depletion could account for the apparent discrepancy between studies. Pratt et al. (2007) assessed insulin action during a euglycaemic-hyperinsulinaemic clamp (EHC) with a steady-state serum insulin concentration of approximately 275–290 mu/l. In contrast, similar to previous studies in man (Araújo-Vilar et al. 1997; Arciero et al. 1999), the present study employed a hyperglycaemic clamp at physiological hyperinsulinaemia (60–65 mu/l). In man, there is an approximately linear increase in glucose uptake from plasma as steady-state insulin concentration is increased from 20 mu/l to about 100 mu/l, reaching an asymptote at supraphysiological insulin concentrations (150–350 mu/l). Furthermore, sensitivity for the detection of exercise-associated alterations in insulin action appeared to be optimal at physiological hyperinsulinaemia, i.e. 20 mu/l and 100 mu/l (Mikines et al. 1988, 1989). Further studies are warranted to determine whether the same is true in horses.

The physiological basis for the effect of acute exercise on insulin-stimulated glucose uptake and metabolism has not been fully elucidated. Enhanced insulin sensitivity is restricted to the previously contracted muscle rather than a consequence of changes in systemic factors (Wojtaszewski et al. 2002). Increased insulin sensitivity to glucose transport after acute exercise is associated with increased recruitment of GLUT-4 molecules to the plasma membrane (Hansen et al. 1998) but activation of proximal components in the insulin signalling cascade is unchanged (Wojtaszewski et al. 2002; 2003). Recent human studies have provided evidence that TBC domain family, member 4 (TBC1D4; also known as AS160) may play a central role in the mobilisation of GLUT-4 to the plasma membrane in skeletal muscle after exercise (Treebak et al. 2009).

An unexpected finding of the present study was the lack of detectable change in NIMGU after exercise (R-L vs. Ex-L; Fig 2). Lack of statistical power (small sample size) or the delay between the end of exercise and the period of measurement may have limited the ability to detect a significant effect of exercise on IMGU. Most studies in man have reported enhanced glucose-mediated glucose disposal after a single bout of exercise, measured either by the hyperglycaemic clamp technique (Marin et al. 1993; Brun et al. 1995) or by estimation of glucose effectiveness (SG) from minimal model analysis of a frequently-sampled i.v. glucose tolerance test (Sakamoto et al. 1999; Hayashi et al. 2005; Bordenave et al. 2008). Several factors may contribute to an increase in NIMGU early after exercise, including increased skeletal muscle blood flow (Sakamoto et al. 1999), increased glucose transporter number and/or activity in the plasma membrane (Goodyear et al. 1990) and a contraction-associated activation of α2-AMP-activated protein kinase (AMPK) that induces translocation of GLUT-4 to the plasma membrane and enhances cellular glucose uptake independent of insulin (Hayashi et al. 2005). Further study is needed to examine these factors in the context of glucose metabolism in horses during recovery from exercise.

In conclusion, the present study demonstrated that a bout of moderate intensity exercise in horses enhances the ability of physiological hyperinsulinaemia to stimulate whole-body glucose uptake but does not alter the rate of glucose-mediated (noninsulin-dependent) glucose disposal. These findings may be relevant to the understanding of factors affecting the availability of glucose for muscle glycogen synthesis in horses after exercise.


This study was supported by grants from Equine Guelph and the Natural Science and Engineering Research Council of Canada. The authors acknowledge Leah Larsen and Louise Waterfall for assistance with the study.

Manufacturers' addresses

1 Sato, BIAB Industrial, Uppsala, Sweden.

2 Abbott Laboratories, Chicago, Illinois, USA.

3 Deseret, Sandy, Utah, USA.

4 Heska, Denver, Colorado, USA.

5 Bachem Inc., Los Angeles, California, USA.

6 Eli Lily, Indianapolis, Indiana, USA.

7 KD Scientific, Kansas City, Missouri, USA.

8 Yellow Springs Instruments, Yellow Springs, Ohio, USA.

9 Becton-Dickinson, Parsippany, New Jersey, USA.

10 Diagnostic Products Corporation, Los Angeles, California, USA.

11 Diagnostic Systems Laboratories Inc., Webster, Texas, USA.

12 SAS Institute Inc., Cary, North Carolina, USA.