Microvascular Recruitment Is an Early Insulin Effect That Regulates Skeletal Muscle Glucose Uptake In Vivo

Male Sprague-Dawley rats weighing 250–350 g (Hilltop Laboratory Animals, Scotdale, PA) were housed at 22 ± 2°C and maintained with a 12-h/12-h light/dark cycle. Animals were provided with food and water ad libitum and then fasted overnight before study. The University of Virginia Animal Care and Use Committee approved the experimental protocols. Rats were anesthetized with sodium pentobarbital (55 mg/kg body wt i.p.), and polyethylene cannulae were inserted into the carotid artery for blood sampling and measurement of blood pressure and into both jugular veins for intravenous infusions. A tracheostomy was performed to facilitate spontaneous respiration during the experiment. Both femoral vessels were exposed via a 1.5-cm incision, and the left femoral artery was carefully separated from the femoral vein and nerve, the epigastric vessels were ligated, and a flow probe (Transonic Systems, VB series 0.5 mm) was positioned over the artery. The probe was interfaced through a flow meter to an IBM-compatible computer. Femoral artery blood flow, blood pressure, and heart rate were measured continuously using WINDAQ data acquisition software (DATAQ Instruments). Anesthesia was maintained by a continuous infusion of aqueous sodium pentobarbital (0.6 mg · min⁻¹ · kg⁻¹) via the carotid artery. Rats were then maintained for a 1-h basal stabilization period before beginning each of three infusion protocols.

After a 1-h basal period, two groups of animals were studied using each of three infusion protocols. The first protocol involved infusion of saline only, the second involved regular insulin (6 mU · min⁻¹ · kg⁻¹ × 5 min, then 3 mU · min⁻¹ · kg⁻¹ × 25 min, euglycemic insulin clamp), and the third involved the same insulin infusion with a superimposed infusion of the NOS inhibitor N^ω-l-nitro-arginine-methyl ester (l-NAME) (50 μg · min⁻¹ · kg⁻¹). The infusion rate for l-NAME was selected after preliminary experiments demonstrated that it caused an ∼20-mmHg rise in mean arterial pressure, indicating effective antagonism of NOS (18). The l-NAME infusion was begun with the 30-min basal period and persisted throughout insulin infusion. The volume of saline infused during protocol 1 was comparable to that used during insulin/glucose infusion in the other two protocols. In one of the two groups of animals studied with each of these three infusion protocols, we measured microvascular blood volume (CEU) and glucose metabolism (glucose infusion rate to maintain euglycemia) in vivo at baseline and serially during a 30-min experimental period. In the second group, we positioned an ultrasonic flow probe around the femoral artery (Transonic Systems, Ithaca, NY) to measure total limb blood flow and measured arterial-venous glucose differences at 0, 7, 15, and 30 min. Calf muscle samples were obtained at 0, 7, and 30 min to measure phosphorylation of Akt. Two separate groups were required for each protocol because Doppler measurements of femoral blood flow cannot be obtained during microbubble infusion. Four to six rats were studied in each protocol in both groups.

Using CEU imaging, we measured microvascular blood volume continuously throughout each infusion. This technique relies on the ultrasound detection of a microbubble contrast agent composed of a lipid shell and a perfluorocarbon gas core. These microbubbles average 4 μm in diameter, remain entirely within the vascular space, and have similar rheology to erythrocytes (19). As a result, they are an excellent perfusion tracer. CEU allows imaging of specific regions within muscle and is not influenced by contributions from nonmuscle tissue (skin, bone, and adipose). The acoustic signal that is generated from the microbubbles exposed to ultrasound is proportional to the concentration of microbubbles within the volume of tissue being imaged. Essentially, all microbubbles within the ultrasound beam are simultaneously imaged and destroyed in response to a single pulse of high-energy ultrasound (20,21). As the time between successive pulses is prolonged, the beam becomes progressively replenished with microbubbles. Eventually, the beam will be fully replenished and further increases in the time between each pulse will not affect the microbubble signal in tissue. The rate of microbubble reappearance within the ultrasound beam provides an indication of microvascular flow velocity, and the plateau video intensity reached at long pulsing intervals provides a measurement of microvascular volume, a measure of microvascular recruitment (22). In addition, by using background subtraction, the signal from the tissue as well as from larger vessels (which fill rapidly) can be eliminated. We have previously reported that the time required to reach plateau video intensity is ∼10 s in rat muscle (23).

To acquire images, a linear array, a transducer interfaced with an ultrasound system (HDI-5000; Philips Medical Systems, Andover, MA), was positioned over the right hindlimb of the rat to image the proximal adductor muscle group (adductor magnus and semi-membranous) and was secured for the duration of the experiment. Pulse inversion imaging was performed at a transmission frequency of 3.3 MHz and a mechanical index (peak negative acoustic pressure divided by the square root of the frequency) of 0.8. The acoustic focus was set at the mid-portion of the muscles. Gain settings were optimized and held constant. Data were recorded on a 1.25-cm videotape using an S-VHS recorder. Phospholipid-coated perfluorocarbon gas microbubbles were manufactured at the University of Virginia as reported previously (24). Microbubble concentration and size distribution were determined by electrozone sensing with a Coulter Multisizer IIe (Beckman-Coulter). The microbubbles were infused continuously into the jugular vein at 30 μl/min. Microbubbles were infused for 10 min before beginning data acquisition to allow the bubbles to reach steady state within the blood pool. Data acquisition was started 6 min before the start of saline, insulin, or insulin + l-NAME infusion. Images were acquired continuously at a pulsing interval of 20 s. Data were analyzed offline. Frames were aligned by cross-correlation, and several frames during a short pulsing interval (0.5 s) were averaged and subtracted from frames at a pulsing interval of 20 s. This background subtraction served to largely eliminate contributions from vessels with blood flow velocity faster than 5 mm/s (vessels >20 μm in diameter).

Samples of calf muscle, freeze-clamped at the end of the protocols, were powdered under liquid nitrogen. Approximately 20 mg of the tissue was added to cold 25 mmol/l Tris HCl buffer (26 mmol/l potassium fluoride and 5 mmol/l EDTA, pH 7.5). Tissue was then disrupted by sonication with the use of a microtip probe, 0.5 s on and 0.5 s off for 45 s total at a 3.0 power setting on the Fisher XL2020 sonicator. The homogenate was then centrifuged at 5,000 rpm for 5 min, the protein content of the supernatant was determined, and one aliquot containing 120 μg protein was heated to 100°C with an equal volume of SDS sample buffer and electrophoresed on an 8% polyacrylamide gel. Proteins were electrophoretically transferred to nitrocellulose membranes. After being blocked with 5% low-fat milk in Tris-buffered saline plus Tween 20, membranes were incubated with rabbit anti-phospho-Akt (Cell Signaling Technology, Beverly, MA) overnight at 4°C. This was followed by a 1-h incubation with anti-rabbit IgG coupled to horseradish peroxidase (Amersham Biosciences, Piscataway, NJ) at room temperature, and then the blot was developed using an enhanced chemiluminescence Western blotting kit (Amersham Biosciences). Autoradiographic film was scanned densitometrically (Molecular Dynamics, Piscataway, NJ) and quantitated using TotalLab software.

Two-way repeated-measures ANOVA was used to assess the effects of infusion protocols on microvascular volume. Post hoc testing was used as indicated in the results.

The plasma glucose concentrations, the glucose infusion rate during the clamp, and the femoral artery flow are shown in Fig. 1 for rats studied using each of the three protocols. Results for plasma glucose and glucose infusion rate were pooled for the two groups studied with each protocol because there were no differences between the two groups. Femoral flow was only measured in the rats not receiving microbubbles (see research design and methods). There were no significant time-dependent changes in either plasma glucose or femoral artery blood flow during the 30-min infusions (repeated-measures ANOVA). Femoral flow did appear slightly higher in the insulin + l-NAME, and this occurred together with an increment in systemic blood pressure of ∼30 mmHg (compared with insulin alone), whereas vascular resistance of the hindlimb was not increased significantly over the other two study groups. The exogenous glucose infusion rate was similar in protocols 2 and 3 where insulin was given. At 30 min, the plasma insulin concentrations were as expected higher (P < 0.01) with insulin (530 ± 56 pmol/l) than with saline (∼260 ± 37 pmol/l). There was no difference in plasma insulin between insulin and insulin + l-NAME infusions.

Figure 2 illustrates the time course for change in microvascular volume (presented as the fold-change over baseline) for the three study protocols. This was calculated by measuring video intensity every 20 s for 5 min before beginning insulin, saline, or insulin + l-NAME to obtain an average baseline video intensity. The average video intensity for subsequent 5-min blocks was divided by this baseline value. Insulin infusion provoked a time-dependent increase in the microvascular volume (P < 0.001, repeated-measures ANOVA) that was significant (55 ± 22%, P < 0.05) by 5–10 min after beginning the infusion. There were lesser and slower time-dependent increases in microvascular volume in the saline-infused rats (P < 0.01, repeated-measures ANOVA). Microvascular volume also rose slowly in rats given l-NAME in addition to insulin (P < 0.01, repeated-measures ANOVA), and the pattern was not different than that seen with saline, possibly reflecting a response to the volume of fluid infused with the microbubbles. The increments in video intensity were significantly greater in the rats given insulin when compared with either of the other two groups (Fig. 2). By the end of the insulin infusion (25 min), microvascular volume in the insulin group had risen by 127 ± 33%.

We also measured the time course for the insulin-induced changes of arterial glucose concentration minus venous glucose concentration (A_G − V_G) across the rat hindleg. Because total limb blood flow did not change during the infusion period, A_G − V_G directly indexes glucose disposal. Measurements were made at baseline (time = 0 min) and 7, 15, and 30 min. We observed that despite the increase in insulin-mediated microvascular blood volume between 5 and 10 min, the corresponding leg A_G − V_G glucose disappearance did not significantly differ among saline, insulin, or insulin + l-NAME at baseline, 7 min, or even 15 min (Fig. 3). In addition, it was not necessary to infuse exogenous glucose to maintain arterial glucose constant until after 7 min into the insulin infusion (Fig. 1). By 30 min, A_G − V_G during insulin infusion was 14 ± 3 mg/dl, which was significantly (two-way ANOVA) higher (P < 0.05) than baseline, 30-min saline (4 ± 1 mg/dl), or 30-min insulin + l-NAME (7 ± 1 mg/dl). These findings suggest that microvascular recruitment preceded increases in glucose disposal during insulin infusion. There was no significant difference in the glucose infusion rate required to maintain euglycemia between the insulin alone or insulin + l-NAME groups at either 15 or 30 min.

Before insulin can increase muscle glucose uptake, it must first activate insulin-signaling pathways in muscle (25). To address whether insulin-mediated capillary recruitment preceded or followed the activation of muscle insulin signaling, we measured Akt phosphorylation in calf muscle freeze-clamped at 7 or 30 min of saline, insulin, or insulin + l-NAME infusion. Western analysis revealed (Fig. 4) that at 7 min, Akt phosphorylation was not significantly different between any of the groups. However, by 30 min, insulin, compared with saline, significantly (P < 0.05) enhanced Akt phosphorylation. Akt phosphorylation in insulin + l-NAME at 30 min was intermediate and not statistically different than insulin or saline alone. The nitrocellulose membranes were stripped and blotted for total Akt, and there were no differences among the three groups (data not shown). As mentioned previously, Akt phosphorylation in endothelial cells precedes activation of eNOS (12,26). However, endothelial tissue contributes negligibly to the mass of tissue analyzed here, and the Akt measured in these whole muscle homogenates overwhelmingly reflects that of the myocyte.

The findings reported here demonstrate that, within minutes, insulin significantly increases the microvascular volume perfused in resting skeletal muscle. This is a distinctly different time course from that seen for the increase in total hindlimb blood flow in the rat (17) or for increases in leg or forearm blood flow in response to physiological hyperinsulinemia in humans (2). In both cases, these increases in total flow require 1 to several hours. The vessels involved in these responses are distinct. Capillary blood volume increases when precapillary arterioles dilate, allowing flow to previously unperfused or underperfused vessels. Total flow increases when larger resistance vessels relax. It has been previously demonstrated that these two classes of arterial vessels can be regulated independently in response to electrical stimulation (27). Low-frequency (four cycles per minute) electrical stimulation of the obturator nerve relaxes precapillary arterioles in canine skeletal muscle and recruits previously unperfused capillary beds in the absence of any changes in total blood flow (27). Higher-frequency stimulation both recruits microvasculature and increases total blood flow. With electrical stimulation, both responses occur promptly after stimulation. Likewise, we have observed that low-intensity forearm exercise increases microvascular volume measured by CEU without changing total forearm blood flow, whereas higher-intensity exercise increases both forearm flow and microvascular volume (28).

Considerable other data indicate that relaxation of precapillary arterioles and resistance vessels can be regulated independently. Previously, we reported that in human skeletal muscle, insulin infused locally into the brachial artery increased forearm microvascular volume without affecting total forearm flow (29). We have also recently measured the dose-response characteristics for insulin’s action on total blood flow and on microvascular volume using both the CEU method used here (30) and measurements of 1-methylxanthine removal by hindlimb skeletal muscle (31). In those studies, regardless of the method used, microvascular recruitment was more sensitive than total blood flow to 2 h of steady-state hyperinsulinemia. Conversely, in the rat hindlimb in vivo, we observed that low doses of epinephrine increase total flow without affecting the capillary volume perfused (measured using either 1-methylxanthine [15] or laser Doppler flowmetry [14]).

In the current study, we show that the microvascular action of insulin, like that involving resistance vessels, can be blocked by inhibition of NOS. We had previously found that l-NAME blocked the increases in both total and hindlimb flow provoked by 2 h of marked hyperinsulinemia (10 mU · min⁻¹ · kg⁻¹) in the rat hindlimb (18). However, in that study, both total flow and microvascular volume were increased by insulin; as a result, we could not discount the possibility that the decline in capillary recruitment caused by l-NAME was in some manner secondary to an effect to decrease total flow. Clearly, findings in the current study using lower insulin doses and shorter observation times when total flow is unchanged supports a direct effect of NOS inhibition on microvascular recruitment.

It has not been addressed previously whether the vasodilatory effects of insulin on either resistance vessels or precapillary arterioles was a direct effect of insulin on the vascular cells or arose indirectly, mediated by either a metabolic action of insulin within the muscle, or perhaps neural action of insulin. Insulin receptors are present on endothelial cells (32), and insulin can activate NOS via a specific phosphorylation mediated by Akt (10). The fact that NOS inhibition was able to block both vascular actions of insulin is consistent with a direct endothelial effect of insulin but does not eliminate the possibility that an indirect action is contributing. This issue is particularly pertinent to the effect of insulin on resistance vessels simply because of the time that appears to be required to relax these vessels. Insulin can activate its receptor and downstream signaling to Akt and NOS within seconds in isolated endothelial cells (11,12) but requires 30 min to several hours to dilate resistance vessels (2,33). In contrast, the rapid time course for insulin’s action to increase microvascular perfusion argues for a direct effect on the vasculature.

It is of interest to compare the time course of insulin’s microvascular action reported here to its time course for its appearance in muscle interstitium, activation of insulin receptor kinase activity, and stimulation of glucose disposal. This has been carefully studied by Miles et al. (34), who estimated the half-times for insulin transfer from the vasculature to the muscle interstitium (∼28 min), for activation of the receptor tyrosine kinase in muscle (∼32 min), and for increases in muscle glucose disposal (∼20 min). From Fig. 2, it is clear that the time course for insulin’s recruitment of microvasculature in the rat temporally precedes these other effects and is consistent with a direct microvascular action of insulin. These findings do not eliminate the possibility of a neural-mediated insulin effect. However, the observation that locally infused insulin recruits microvasculature in human forearm muscle is somewhat against a central neural action.

A potential limitation of the current work is the use of anesthetized animals. Because any movement distorts the ultrasound image, it is not possible to conduct these studies in conscious animals. However, whereas the absolute estimates of muscle microvascular perfusion may be altered from what would be seen in the conscious animal, this should not impair our ability to compare the three groups of animals in this study.

The fact that inhibition of NOS blunted glucose uptake by muscle supports a relationship between insulin’s microvascular and metabolic actions. Bonadonna et al. (35) previously pointed out (based on the analysis by Renkin [36]) that in as much as the extraction ratio for glucose across skeletal muscle is relatively low under basal conditions and during modest hyperinsulinemia, changes in total flow would be expected to minimally affect glucose disposal. Conversely, when insulin concentrations are high and muscle glucose uptake strongly stimulated (a high extraction ratio for glucose across the muscle), then increments in total blood flow would significantly affect muscle glucose disposal (2,4).

We have observed that several factors that antagonize insulin-mediated glucose disposal (e.g., infusion of free fatty acids [37], tumor necrosis factor-α [38], or α-methyl serotonin [16]) each blunt the effect of insulin to recruit muscle microvasculature. Likewise, we have observed that insulin’s action to recruit microvasculature is impaired in the insulin-resistant Zucker rat (39). Thus, understanding the relationship between insulin sensitivity and its action to recruit microvasculature may be important to the overall understanding of insulin resistance.

The relation of insulin’s vascular to its metabolic actions has also been explored using genetically altered mice. The eNOS knockout mouse is insulin resistant (8,9), as is the nNOS animal. A vascular endothelial cell-specific insulin receptor knockout mouse has been reported to have a normal phenotype with regard to vascular development but to express lower amounts of both eNOS and endothelin 1 (13). Significant insulin resistance in these latter animals was seen when they were on a low-salt diet but not a regular diet. Inasmuch as the specific changes in insulin sensitivity seen in these gene deletion studies may be affected by developmental changes, it is difficult to interpret these findings vis-à-vis normal physiological regulation. At present, the status of insulin-mediated capillary recruitment, or indeed vasodilation, generally in the muscle of these mice is unknown.

In conclusion, physiological hyperinsulinemia rapidly recruits microvascular elements in rat skeletal muscle in vivo. This vascular action of insulin precedes insulin’s effects to increase muscle glucose uptake or activate downstream kinases involved in mediating skeletal muscle insulin action. Inhibition of NOS specifically blocks this action of insulin and diminishes muscle glucose uptake, reaffirming that this vascular action of insulin is partly responsible for insulin’s overall metabolic action in muscle.

This work was supported by National Institutes of Health Grants DK-38578 and DK-54058 (to E.J.B.) and KO8 HL-03810 (to J.R.L.) and grants from the National Health and Medical Research Council Australia (to M.G.C. and S.R.).