Increased endothelial shear stress improves insulin‐stimulated vasodilatation in skeletal muscle

Abstract

Key points

• It has been postulated that increased blood flow‐associated shear stress on endothelial cells is an underlying mechanism by which physical activity enhances insulin‐stimulated vasodilatation.

• This report provides evidence supporting the hypothesis that increased shear stress exerts insulin‐sensitizing effects in the vasculature and this evidence is based on experiments in vitro in endothelial cells, ex vivo in isolated arterioles and in vivo in humans.

• Given the recognition that vascular insulin signalling, and associated enhanced microvascular perfusion, contributes to glycaemic control and maintenance of vascular health, strategies that stimulate an increase in limb blood flow and shear stress have the potential to have profound metabolic and vascular benefits mediated by improvements in endothelial insulin sensitivity.

Abstract

The vasodilator actions of insulin contribute to glucose uptake by skeletal muscle, and previous studies have demonstrated that acute and chronic physical activity improves insulin‐stimulated vasodilatation and glucose uptake. Because this effect of exercise primarily manifests in vascular beds highly perfused during exercise, it has been postulated that increased blood flow‐associated shear stress on endothelial cells is an underlying mechanism by which physical activity enhances insulin‐stimulated vasodilatation. Accordingly, herein we tested the hypothesis that increased shear stress, in the absence of muscle contraction, can acutely render the vascular endothelium more insulin‐responsive. To test this hypothesis, complementary experiments were conducted using (1) cultured endothelial cells, (2) isolated and pressurized skeletal muscle arterioles from swine, and (3) humans. In cultured endothelial cells, 1 h of increased shear stress from 3 to 20 dynes cm⁻² caused a significant shift in insulin signalling characterized by greater activation of eNOS relative to MAPK. Similarly, isolated arterioles exposed to 1 h of intraluminal shear stress (20 dynes cm⁻²) subsequently exhibited greater insulin‐induced vasodilatation compared to arterioles kept under no‐flow conditions. Finally, we found in humans that increased leg blood flow induced by unilateral limb heating for 1 h subsequently augmented insulin‐stimulated popliteal artery blood flow and muscle perfusion. In aggregate, these findings across models (cells, isolated arterioles and humans) support the hypothesis that elevated shear stress causes the vascular endothelium to become more insulin‐responsive and thus are consistent with the notion that shear stress may be a principal mechanism by which physical activity enhances insulin‐stimulated vasodilatation.

Key points

• It has been postulated that increased blood flow‐associated shear stress on endothelial cells is an underlying mechanism by which physical activity enhances insulin‐stimulated vasodilatation.

• This report provides evidence supporting the hypothesis that increased shear stress exerts insulin‐sensitizing effects in the vasculature and this evidence is based on experiments in vitro in endothelial cells, ex vivo in isolated arterioles and in vivo in humans.

• Given the recognition that vascular insulin signalling, and associated enhanced microvascular perfusion, contributes to glycaemic control and maintenance of vascular health, strategies that stimulate an increase in limb blood flow and shear stress have the potential to have profound metabolic and vascular benefits mediated by improvements in endothelial insulin sensitivity.

Introduction

Insulin‐stimulated blood flow and capillary recruitment play a critical role in the delivery of glucose and insulin to skeletal muscle, a primary site for glucose disposal (Baron et al. 1996; Vincent et al. 2002; Zhang et al. 2004; Barrett et al. 2009; Meijer et al. 2012; Eggleston et al. 2013). As such, reduced insulin‐stimulated vasodilatation and poor skeletal muscle perfusion limit glucose uptake and contribute to impaired glucose control (Baron et al. 1991; Laakso et al. 1992; Womack et al. 2009). At the endothelial cell level, the vasodilatory effects of insulin are nitric oxide (NO)‐dependent. Specifically, insulin binds to insulin receptors on endothelial cells leading to PI3K activation, phosphorylation of Akt and subsequent activation of endothelial NO synthase (eNOS) (Eringa et al. 2002, 2007; Kim et al. 2006). As a countercurrent pathway, insulin activates Ras/mitogen‐activated protein kinase (MAPK) signalling resulting in the production of the vasoconstrictor peptide endothelin‐1 (ET‐1) (Eringa et al. 2002, 2007; Kim et al. 2006). Available evidence indicates that aberrant vascular insulin signalling, typically characterized by diminished eNOS activation and/or heightened MAPK activation, is an early contributor to the pathogenesis of type 2 diabetes and vascular disease, emphasizing the importance of identifying strategies that improve vascular insulin sensitivity (Eringa et al. 2002, 2007; Kim et al. 2006; Lteif et al. 2007; Shemyakin et al. 2010).

In this regard, a number of studies from animals and humans demonstrate that increased physical activity is effective at enhancing insulin‐stimulated vasodilatation and blood flow to skeletal muscle (Dela et al. 1995; Bisquolo et al. 2005; Mikus et al. 2011, 2012; Crissey et al. 2014; Padilla et al. 2015; Sjoberg et al. 2017). In fact, recent evidence indicates that even a single bout of leg exercise improves insulin‐stimulated leg blood flow (Sjoberg et al. 2017). The precise mechanisms by which exercise elicits insulin‐sensitizing effects on the vasculature remain unknown. An important and consistent finding is that this effect appears to be restricted to vascular beds that undergo increases in blood flow during exercise and it is not dependent on changes in body composition or reductions in hyperglycaemia (Mikus et al. 2011, 2012; Sjoberg et al. 2017). On the basis of these findings, we have previously speculated that exercise‐induced increases in blood flow and, thus, presumably increased shear stress on the vascular endothelium may be an underlying mechanism by which physical activity enhances insulin‐stimulated vasodilatation (Padilla et al. 2015). However, it remains unknown if increased shear stress, in the absence of muscle contraction, augments the vasodilator actions of insulin. Given that both shear stress and insulin activate eNOS through PI3K signalling pathways (Eringa et al. 2002, 2007; Kim et al. 2006; Sriram et al. 2016), it is plausible that shear stress renders the vasculature more insulin‐responsive. To test this hypothesis, complementary experiments were conducted using (i) cultured endothelial cells, (ii) isolated and pressurized skeletal muscle arterioles from swine, and (iii) humans. In humans, we hypothesized that if shear stress is indeed a signal that stimulates insulin actions in the vasculature, a bout of increased leg blood flow‐induced shear stress, using a non‐exercise stimulus (i.e. lower‐limb heating), would augment insulin‐stimulated skeletal muscle perfusion.

Methods

Protocol 1. Cultured human aortic endothelial cells

Human aortic endothelial cells (HAECs) (Lonza, Walkersville, MD, USA; no. CC‐2535, lot no. 0000316663) were cultured in complete VascuLife® EnGS medium (Lifeline Cell Technologies, Fredrick, MD, USA; no. LL‐0002) with 2% fetal bovine serum. All primary endothelial cells were cultured in a humidified incubator at 37°C and 5% CO₂. Experiments were conducted when cells were between passages 7 and 8.

Shear stress treatments

To induce shear stress on cultured cells, HAECs were seeded into Ibidi μ‐Slide I 0.4 Luer channels (Ibidi, Martinsried/Munich, Germany; no. 80176) that were previously coated with 2% gelatin (Sigma‐Aldrich, St Louis, MO, USA; no. G6650). After cells achieved adherence, a unidirectional laminar flow producing a low level of shear stress (3 dynes cm⁻²) was applied for 40 h to promote the alignment of cells in the direction of flow. A computer‐controlled flow device containing an air‐pressure pump and a two‐way switching valve was used to generate shear stress (Ibidi, no. 10902). The flow device was kept at 37°C in a 5% CO₂ atmosphere inside an incubator. After the initial 40 h exposure to low shear stress, shear stress was changed to 20 dynes cm⁻² (increased shear stress) or maintained at 3 dynes cm⁻² (control condition) for 1 h. Both treatments were followed by a recovery period of 30 min at 3 dynes cm⁻². Thereafter, cells were stimulated with insulin (100 nm, Humulin R, Eli Lilly, Indianapolis, IN, USA) or vehicle (cell culture media) for 30 min under static no‐flow conditions. Each experimental condition was repeated four times (i.e. n = 4/treatment) and the experimental protocol is illustrated in Fig. 1 A.

Figure 1. Increased shear stress in cultured endothelial cells causes a shift in insulin signalling characterized by greater activation of eNOS relative to MAPK.

A, schematic diagram and illustration of experimental protocol 1. Human aortic endothelial cells (HAECs) were seeded and exposed to low shear stress conditions (3 dynes cm⁻²) for 40 h to promote cell alignment. Cells were then either exposed to increased shear stress (20 dynes cm⁻²) or maintained at control conditions (3 dynes cm⁻²) for 1 h. Following a 30 min recovery period (3 dynes cm⁻²) designed to mimic subsequent protocols, cells were stimulated with insulin or vehicle for 30 min and fixed in their chambers. Immunofluorescence and image analysis were performed in all conditions. The cell image displayed shows fluorescence associated with stains for phospho‐eNOS (green) and 4′,6‐diamidino‐2‐phenylindole for nuclei (blue). B, the change in the ratio of phospho‐eNOS/phospho‐MAPK upon insulin stimulation in cells previously exposed to control conditions vs. increased shear stress. The cartoon underneath further illustrates the results. IR, insulin receptor. n = 4/condition. Values are means ± SEM. ^* P < 0.05 vs. control condition.

Immunofluorescence and imaging

Following insulin stimulation, cells were fixed within the flow chambers with 4% paraformaldehyde (Electron Microscopy Sciences, Hatfield, PA, USA; no. 15710) for 60 min at 4°C. Cells were then rinsed twice in phosphate buffered saline (PBS) and once in 0.1 m glycine for 5 min each time. Subsequently, permeabilization and blocking were achieved by exposure to 0.1% Triton X‐100 for 5 min followed by a 1 h incubation in antibody blocking buffer (10% goat serum + 0.05% Triton X‐100 + 1% BSA) at room temperature. Cells were incubated with primary antibodies overnight at 4°C in antibody buffer (2% goat serum + 0.05% Triton X‐100 + 1% BSA) and fluorophore‐conjugated secondary antibodies were applied for 1 h at room temperature. The cells then were washed and prepared for imaging with ProLong Diamond reagent (Life Technologies, Carlsbad, CA, USA; no. P36962). An investigator blinded to characteristics of the treatments captured images in triplicates using a Leica SPE confocal microscope (Leica, Wetzlar, Germany) with a ×20/0.60 numerical aperture oil objective. Imaris software v9.0 (Bitplane, Zurich, Switzerland) was used to quantify the mean fluorescence intensity of the images within a defined region of interest (500 × 500 μm). Fluorescence intensity within the images was normalized to the number of cells as determined by the number of nuclei and the change in the ratio of phospho‐eNOS/phospho‐MAPK intensities upon insulin stimulation was calculated.

Antibodies

Primary antibodies utilized included: anti‐phospho‐eNOS (Ser1177) rabbit polyclonal (1:100; Abcam, Cambridge, MA, USA; no. ab184154) and anti‐phospho‐p44/42‐MAPK rabbit polyclonal (1:250; Cell Signaling Technology, Danvers, MA, USA, no. 9101). Fluorescently tagged secondary antibodies were purchased from Life Technologies and applied at a 1:200 dilution.

Protocol 2. Isolated skeletal muscle arterioles from swine

Ethical approval and animals

All experimental procedures described herein were approved by the Animal Care and Use Committee (protocol no. 8907) at the University of Missouri. Intact female farm swine (n = 9; 3.6 ± 0.2 months of age; 30.9 ± 1.4 kg) were housed under temperature‐controlled conditions, with a 12:12 h light–dark cycle. Intramuscular injection of Telazol (5 mg kg⁻¹)–xylazine (2.25 mg kg⁻¹) mixture was followed by inhalation of isoflurane (5%) for 20 min at which point, after absence of reflexes was confirmed, a sternotomy was performed and the heart was removed. The lateral head of the triceps was harvested immediately thereafter and placed in ice‐cold physiological saline solution (PSS: 145 mm NaCl, 4.7 mm KCl, 2.0 mm CaCl₂, 1.17 mm MgSO₄ with 10 g l⁻¹ albumin added) with a pH of 7.4.

Shear stress and insulin‐stimulated vasodilatation in isolated arteries

Skeletal muscle arterioles (i.e. two sections from the same artery) were dissected from the isolated triceps using an Olympus microscope (Olympus, Tokyo, Japan), transferred to a Plexiglas chamber filled with PSS and cannulated within two resistance‐matched glass micropipettes, as described previously (Woodman et al. 2005). Thereafter, the chambers were transferred to the stage of an inverted microscope (Nikon Diaphot 200, Tokyo, Japan) attached to a video camera (Javelin Electronics, Los Angeles, CA, USA), video micrometer (Microcirculation Research Institute, Texas A&M University, TX, USA) and a PowerLab data acquisition system (ADInstruments, Colorado Springs, CO, USA). The micropipettes were subsequently attached to separate fluid‐filled reservoirs filled with warm PSS supplemented with albumin (1 g/100 ml) and arterioles were pressurized to 60 mmHg under no‐flow conditions and allowed to stabilize for ∼30 min at which point maximal arterial vasoconstriction in response to 80 mm KCl was determined. Luminal diameter was monitored throughout the experiment.

Following the stabilization period, one section of the arteriole remained pressurized and exposed to no‐flow control conditions for 1 h, while the other section was subjected to intraluminal flow (i.e. increased shear stress, 20 dynes cm⁻²) for 1 h. To achieve the target dose of increased shear stress, intraluminal flow was adjusted by changing proximal and distal pressures in equal but opposite directions to establish a pressure gradient across the arteriole while maintaining mean pressure at 60 mmHg (Woodman et al. 2005). Following the experimental intervention, arterioles were returned to no‐flow conditions for 30 min. Arterioles were then pre‐constricted with U46119 (thromboxane A2 analogue; 1 × 10⁻⁷–1 × 10⁻⁴ m) and dose–response curves for insulin (1e⁻⁹–1e⁻⁴ m) and sodium nitroprusside (SNP) (whole‐log doses; 1 × 10⁻⁹–1 × 10⁻⁴ m) were examined. After the dose–response curves, vessels were washed twice with Ca²⁺‐free PSS to determine maximal passive diameter. Vasomotor responses were expressed as the percentage possible dilatation, i.e. (Δdiameter − baseline diameter)/(Δmaximal Ca²⁺‐free diameter − baseline diameter) × 100). The experimental protocol is illustrated in Fig. 2 A.

Figure 2. Increased shear stress augments insulin‐stimulated vasodilation in skeletal muscle arterioles from swine.

A, schematic diagram and illustration of experimental protocol 2. Skeletal muscle arterioles were harvested from the triceps muscle of farm swine, cannulated, pressurized to 60 mmHg, stabilized, and subsequently exposed to 1 h of control conditions (0 dynes cm⁻²) or increased shear stress (20 dynes cm⁻²). Following a 30 min recovery period under no‐flow conditions, dose–response curves for insulin and sodium nitroprusside (SNP) were examined. B, vasomotor responses to insulin and SNP in arteries previously exposed to control conditions or increased shear stress. n = 9/condition. Values are means ± SEM. ^* P < 0.05 vs. control condition.

Protocol 3. In vivo human study

Ethical approval and participants

The study was approved by University of Missouri Health Science Institutional Review Board (IRB, no. 2008181) and registered at ClinicalTrials.gov (NCT03203694). The study was conducted in accordance with the Declaration of Helsinki. Eighteen healthy, recreationally active, subjects (10 men, 8 women; age: 26 ± 1 years; height: 169.6 ± 2 cm; weight: 72.1 ± 4 kg; BMI: 25.1 ± 1 kg m⁻²) were recruited from the University of Missouri campus and surrounding Columbia, MO area to participate in this protocol. Following written consent, subjects were scheduled for a single experimental visit, as illustrated in Fig. 3 A. Experimental visits for female subjects were coordinated to be scheduled during the early follicular phase to minimize the impact of the menstrual cycle on vascular outcomes. Four of eight women used contraceptive medications. All studies were performed in a temperature‐controlled room (∼21°C) following a 2 h fast. Subjects were asked to refrain from caffeine and alcohol for at least 12 h, as well as from exercise for 24 h prior to the study visit.

Figure 3. In vivo human study: experimental design and evidence that leg heating increases blood flow.

A, schematic diagram of experimental protocol 3 and measurements performed throughout the course of the study. Subjects underwent 1 h of single‐leg heating (limb immersion in a 40–42°C water bath) followed by a 60 min cooldown period during which leg blood flow progressively returned to pre‐heating levels. Thereafter, bilateral measures of calf muscle microvascular perfusion (via contrast‐enhanced ultrasound, CEU) and popliteal artery blood flow (via 2D/Doppler ultrasound) were performed at baseline and during systemic infusion of insulin (40 mU m⁻² min⁻¹). Dextrose was co‐infused at varying rates to maintain euglycaemia. B, experimental set‐up during single‐leg heating. C, bilateral popliteal artery blood flows during single‐leg heating; n = 18. D, background‐subtracted colour‐coded frames of the calf muscle captured during end‐diastole at 0 and 60 min during single‐leg heating in the control leg and heated leg of a single subject. Corresponding time (cardiac cycle) vs. background‐subtracted intensity curves are displayed. Values are means ± SEM. ^* P < 0.05 vs. control leg.

Single‐limb heating

Subjects were placed in a semi‐recumbent seated position and baseline bilateral measurements of popliteal artery blood flow were performed utilizing 2D and Doppler ultrasonography. To increase blood flow to a single leg, subjects underwent 1 h of unilateral‐leg heating via limb immersion in a 40–42°C water bath, while the contralateral leg remained outside of the water bath and served as a non‐heated internal control. Ultrasound measures were performed again in the heated and non‐heated legs at 30 and 60 min during heating (Fig. 3 C). This water temperature is below the threshold for pain sensation (Minson, 2010) and has been shown by us and others to evoke robust increases in limb blood flow without producing major systemic cardiovascular effects (Pyke et al. 2004, 2008a, b ; Heinonen et al. 2011; Naylor et al. 2011; Padilla et al. 2011; Restaino et al. 2016; Romero et al. 2017). While a large portion of the flow is directed to the skin (Taylor et al. 1984; Green et al. 2010), there is firm evidence that calf muscle blood flow is also markedly increased with local heating (Heinonen et al. 2011).

Systemic insulin infusion with assessment of leg blood flow and skeletal muscle microvascular perfusion

Heating was followed by a 60 min cooldown period in the lying‐down position, allowing for leg blood flow to gradually return to pre‐heating levels. During the cooldown period, two intravenous catheters were placed in the antecubital veins of both arms, one for infusion and the other for blood draws. Thirty minutes before baseline (i.e. pre‐insulin infusion) measures of blood flow, subjects were moved to the lying‐down prone position. Measures of popliteal artery blood flow via 2D/Doppler ultrasound and calf muscle microvascular perfusion via contrast enhanced ultrasound (CEU) were performed in both the previously heated and non‐heated control legs. Insulin (Humulin R) was prepared via dilution in 250 ml 0.9% saline along with 5 ml of blood taken from the subject to a final concentration of 500 mU ml⁻¹. Insulin was then infused at a steady state delivering insulin at a constant dosage of 40 mU m⁻² min⁻¹ (Reynolds et al. 2017), mimicking postprandial insulin levels, for 1 h. Whole blood glucose was determined at baseline and every 5 min and maintained at euglycaemic concentrations, achieved by variable infusion rates of a 20% dextrose solution. Plasma was also collected every 15 min and stored at −80°C for future analysis. Measures of CEU (in duplicate) and 2D/Doppler ultrasound (2 min segments) were performed again at 15 and 45 min (Fig. 3 A), respectively, after constant insulin infusion in both the previously heated and non‐heated control legs.

Popliteal artery blood flow assessed via 2D/Doppler ultrasound

Popliteal artery diameter and blood velocity were measured using ultrasound (iE33; Philips Medical Systems, Bethesda, MD, USA) as previously described (Morishima et al. 2016; Restaino et al. 2016; Morishima et al. 2017; Walsh et al. 2017). Briefly, an 11 MHz linear array transducer was placed over the popliteal artery just distal to the popliteal fossa. Simultaneous diameter and velocity signals were obtained in duplex mode at a pulsed frequency of 5 MHz and corrected with an insonation angle of 60 deg. Sample volume was adjusted to encompass the entire lumen of the vessel without extending beyond the walls and the cursor was set mid‐vessel. Continuous artery diameter and blood velocity were recorded for 2 min along with measurements of systolic and diastolic blood pressure. Recordings were analysed offline using specialized edge‐detection software (Cardiovascular Suite, Quipu srl, Pisa, Italy). Blood flow was calculated from continuous diameter and mean blood velocity recordings using the following equation: 3.14 × [diameter (cm)/2]² × mean blood velocity (cm s⁻¹) × 60 and reported as ml min⁻¹. Vascular conductance was calculated at each time point as blood flow/mean arterial pressure (MAP); MAP was calculated as [(2 × diastolic pressure) + systolic pressure]/3 and reported in mmHg.

Calf muscle microvascular perfusion assessed via CEU

CEU perfusion imaging was performed in the midline of the widest portion of the calf with a phased‐array transducer (S5‐1, Philips Medical Systems) connected to an iE33 (Philips Medical Systems) imaging system. Imaging was performed with a contrast‐specific multipulse amplitude modulation algorithm at a transmission frequency of 1.8 MHz and a mechanical index (MI) of 0.18. Gain settings were optimized and held constant. Imaging was performed bilaterally using trans‐axial imaging planes for the calf muscle at a depth of 5 cm. A 1.3 ml suspension of octafluoropropane gas‐filled lipid microbubbles (DEFINITY®, Lantheus Medical Imaging, Inc., Billerica, MA, USA) was diluted in normal saline for a total volume of 30 ml and infused at a constant rate of 2 ml min⁻¹ (Womack et al. 2009). When a systemic steady‐state microbubble concentration was achieved (3 min), destruction replenishment kinetics were measured after destroying microbubbles in the ultrasound beam with a five‐frame high MI (1.25) sequence. Up to 20 post‐destruction frames were acquired at end‐diastole trough ECG gating. For image analysis (Narnar App, Narnar LLC, Portland, Oregon, USA), the first post‐destruction frame was selected as background and was digitally subtracted from subsequent end‐diastolic images. Time vs. video intensity data were then fit to the function:

$$y=A(1-{\mathrm{e}}^{-\beta t})$$

where y is background‐subtracted video intensity, A is the plateau intensity after complete replenishment reflecting relative microvascular blood volume, t is time and β is the rate constant of the refill curve reflecting microvascular flux rate. Microvascular flow was calculated as the product of A and β. Only a subset of eight subjects (4 men and 4 women) underwent CEU measures.

Insulin and glucose quantification

Whole blood glucose levels were analysed utilizing YSI 2300 STAT PLUS glucose analyser (YSI Inc., Yellow Springs, OH, USA). Plasma insulin concentrations were quantified by a commercial laboratory (Comparative Clinical Pathology Services, Columbia, MO, USA).

Statistical analyses

In the cell culture experiment, a two‐tailed unpaired Student's t test was utilized to examine the change in the ratio of phospho‐eNOS/phospho‐MAPK in response to insulin stimulation between treatments. A two‐way (treatment × dose) ANOVA with Tukey's post hoc test was performed on the vasomotor responses to insulin and SNP in isolated skeletal muscle arterioles. In the human study, comparisons were made using a two‐way (leg condition × time) ANOVA with Tukey's post hoc test on all dependent variables. No sex differences were detected on primary outcome variables; therefore, data from male and female subjects were pooled for analysis. For all statistical tests, significance was accepted at P < 0.05. Data are expressed as means ± SE. All statistical analyses were performed with SPSS Statistics version 23 (IBM Corp., Armonk, NY, USA).

Results

Treating endothelial cells with increased shear stress alters subsequent insulin signalling

Endothelial cells exposed to 1 h of increased shear stress (20 dynes cm⁻²) did not significantly exhibit an increase in insulin‐stimulated phospho‐eNOS (+10 ± 10%) when compared to cells that were kept at low (3 dynes cm⁻²) shear stress (−23 ± 11%, P = 0.11). Similarly, increased shear stress did not affect the change in insulin‐stimulated phospho‐MAPK (−18 ± 15%) when compared to cells maintained at low shear stress (+39 ± 19%, P = 0.06). However, when expressed as a ratio of phospho‐eNOS to phospho‐MAPK, which is thought to be the primary determinant of biological effect (King et al. 2016), increased shear stress resulted in a greater ratio in response to insulin (P = 0.04, Fig. 1 B).

Insulin‐induced vasodilatation is augmented in skeletal muscle arterioles previously exposed to increased shear stress

Insulin‐induced vasodilatation was greater in skeletal muscle arterioles previously exposed to 1 h of increased shear stress (20 dynes cm⁻²) compared to those exposed to time‐controlled no‐flow conditions (Fig. 2 B, P < 0.01). Dilation to the endothelium‐independent vasodilator SNP was not affected by increased shear stress (Fig. 2 B, P = 0.84). Skeletal muscle arteriole characteristics (i.e. passive and preconstriction diameters, wall thickness, wall:lumen ratio and KCl‐induced vasoconstriction) did not differ between treatments and are presented in Table 1.

Table 1.

Swine skeletal muscle arteriole characteristics

	Control condition	Increased shear stress	P
Passive diameter in Ca2+‐free PSS (μm)	193 ± 11	175 ± 14	0.27
Preconstriction diameter (μm)	76 ± 5	71 ± 6	0.45
Wall thickness (μm)	54 ± 4	46 ± 6	0.28
Wall:lumen ratio	0.39 ± 0.02	0.36 ± 0.03	0.45
KCl‐induced % vasoconstriction	67 ± 9	63 ± 9	0.78

Values are means ± SEM; n = 9/condition.

Insulin increases leg blood flow in the previously heated, but not the control, leg

As expected, 60 min of leg heating increased popliteal artery blood flow by ∼3‐fold, relative to the non‐heated control leg (Fig. 3 C). Shear rate at the popliteal artery was also augmented by ∼3‐fold (heated leg, baseline (BL): 113.14 ± 14.1 s⁻¹, 60 min: 301.17 ± 32.34 s⁻¹, P < 0.01; non‐heated leg, BL: 111.03 ± 14.98 s⁻¹, 60 min: 111.81 ± 16.41 s⁻¹, P = 0.95). Moreover, increased calf muscle microvascular perfusion with heating was confirmed via CEU (Fig. 3 D). Following the 1 h cooldown period, leg blood flow largely returned to basal levels (as shown in Fig. 4 B and C). During systemic infusion of insulin, plasma insulin was elevated to postprandial levels while co‐infusion of dextrose resulted in successful maintenance of euglycaemia throughout the study protocol (Fig. 4 A). Indices of calf microvascular perfusion, popliteal artery blood flow, and leg vascular conductance were significantly elevated during hyperinsulinaemia in the previously heated leg, whereas no significant changes were noted in the control leg (Fig. 4 B and C and Table 2). MAP and popliteal artery diameter in both legs were similarly unchanged during hyperinsulinaemia (Table 2).

Figure 4. In vivo human study: increased leg blood flow induced by unilateral limb heating subsequently augments insulin‐stimulated popliteal artery blood flow and muscle perfusion.

A, plasma insulin and whole blood glucose before and during systemic insulin infusion with co‐infusion of dextrose to maintain euglycaemia. n = 18. B, popliteal artery blood flow at baseline (BL) and at 45 min after insulin infusion (Insulin) in both the control leg and the previously heated leg (ANOVA; leg: P = 0.39, time: P = 0.14, interaction: P = 0.01, n = 18). C, bilateral assessment of microvascular blood flow (A × β) at BL and at 15 min after insulin infusion (Insulin) (ANOVA; leg: P = 0.20, time: P = 0.01, interaction: P = 0.06, n = 8). D, background‐subtracted colour‐coded frames of the calf muscle captured during end‐diastole at BL and during systemic insulin infusion (Insulin) in the control leg and previously heated leg of a single subject. Corresponding time (cardiac cycle) vs. background‐subtracted intensity curves are displayed. Values are means ± SEM. ^* P < 0.05 vs. time 0 or BL.

Table 2.

Popliteal artery haemodynamics and calf muscle microvascular perfusion indices in both control and previously heated legs at baseline and during systemic infusion of insulin

		Baseline	Insulin	ANOVA
β	Control leg	0.34 ± 0.04	0.29 ± 0.04	Time: P = 0.72 Leg: P = 0.31 Interaction: P = 0.38 n = 8
	Previously heated leg	0.26 ± 0.04	0.28 ± 0.04
A	Control leg	7.84 ± 1.86	11.67 ± 1.38	Time: P = 0.02 Leg: P = 0.01 Interaction: P = 0.23 n = 8
	Previously heated leg	14.25 ± 3.12	22.53 ± 4.16*
Popliteal artery diameter (cm)	Control leg	0.49 ± 0.03	0.49 ± 0.03	Time: P = 0.23 Leg: P = 0.73 Interaction: P = 0.14 n = 18
	Previously heated leg	0.49 ± 0.02	0.52 ± 0.02
Leg vascular conductance (mL min−1 mmHg−1)	Control leg	1.05 ± 0.19	0.99 ± 0.10	Time: P = 0.06 Leg: P = 0.14 Interaction: P = 0.04 n = 18
	Previously heated leg	1.01 ± 0.11	1.49 ± 0.18*
Mean arterial pressure (mmHg)		84.4 ± 2.6	85.6 ± 2.5	Time: P = 0.59 n = 18

Values are means ± SEM. ^* P < 0.05 vs. baseline.

Discussion

The salient finding of the present investigation is that unilateral limb heating resulting in increased blood flow, and likely attendant increase in shear stress, causes a subsequent augmentation of the vasodilatory response to insulin in that leg. The data from isolated skeletal muscle arterioles further indicate that exposure to increased shear stress augments succeeding insulin‐stimulated vasodilatation, an effect that our cell culture data suggest may be due to a shift in the balance between insulin signalling pathways that favours activation of eNOS over MAPK. Collectively, these findings are in line with the hypothesis that shear stress primes the vascular endothelium to become more insulin‐responsive.

Vascular insulin signalling and its consequent vasodilatory effects facilitate delivery of glucose and insulin to skeletal muscle, contributing to glucose uptake and homeostasis (Baron et al. 1991; Laakso et al. 1992; Womack et al. 2009). Thus, vasodilator actions of insulin in arteries perfusing skeletal muscle can be viewed as an important target for improving glycaemic control. Insulin signalling in endothelial cells is also critical for maintenance of vascular health (Montagnani et al. 2002; Schulman & Zhou, 2009; Rask‐Madsen et al. 2010; King et al. 2016). Current data indicate that lack of insulin signalling in endothelial cells accelerates atherosclerosis in mouse models susceptible to vascular disease (Rask‐Madsen et al. 2010), whereas selective activation of the insulin receptor/Akt signalling pathway results in protection from atherosclerosis (Kanter et al. 2018). Importantly, herein, we report for the first time that application of increased shear stress subsequently alters insulin signalling and augments insulin‐stimulated vasodilatation, establishing haemodynamic forces as a vital determinant of vascular insulin sensitivity beyond its already recognized role in regulating other attributes of arterial function and structure (Green et al. 2017). Interrogation of the extra‐ and intracellular mechanisms by which increased shear stress improves insulin signalling now becomes a chief area of study. Dissecting out the mechanisms will require substantial efforts as it is likely that shear stress interferes at multiple sites of the insulin signalling cascade.

Evidence from previous studies demonstrates that physical activity profoundly modulates insulin actions in the vasculature, an effect particularly manifested in vascular beds highly perfused during exercise (Bisquolo et al. 2005; Mikus et al. 2011, 2012; Padilla et al. 2015). Based on findings from the present investigation, it can now be deduced that increased shear stress may be a primary signal by which exercise exerts such insulin‐sensitizing effects in the vasculature. In addition, findings from our human experiments indicate that improved leg vascular insulin sensitivity can be evoked by means that are effective at stimulating leg blood flow and associated shear stress, independent of exercise, e.g., local heating. Indeed, here we employed lower‐limb heating as a means to increase leg blood flow and presumably associated shear stress without the concomitant influence of muscle contractions inherent to exercise. However, we cannot rule out the possibility that the reported enhancement in vascular insulin actions may be in part attributed to heating per se and, thus, not solely attributed to the prior increase in blood flow‐induced shear stress achieved with heating. In fact, there is evidence that upregulation of heat shock proteins with chronic intermittent heating improves insulin sensitivity in skeletal muscle (Chung et al. 2008; Gupte et al. 2009; Silverstein et al. 2015). Whether a similar phenomenon occurs in vascular cells remains to be experimentally tested. Nevertheless, current results from the experiments in endothelial cells and isolated arterioles support the notion that increased shear stress alone is sufficient to exert vascular insulin‐sensitizing effects, at least acutely.

The finding that leg heating was effective at subsequently augmenting insulin‐stimulated leg blood flow is strengthened by the fact that it was corroborated by two independent vascular outcome measures, namely calf microvascular perfusion assessed by CEU and popliteal artery blood flow assessed by 2D/Doppler ultrasound. In this regard, prior work has demonstrated that insulin‐stimulated skeletal muscle capillary recruitment (i.e. perfusion) is critical to glucose/insulin transport and can precede changes in vascular conductance and total limb blood flow (Vincent et al. 2002). The reported vascular effects of limb heating have notable therapeutic implications when we consider that a significant fraction of the population exhibiting vascular impairments are also disabled or incapable of performing sustained aerobic exercise. Therapies such as local heating may be an attractive strategy to transduce exercise‐like vascular adaptations in human limbs. Our findings extend those from previous studies demonstrating that local heating also prevents prolonged sitting‐induced leg vascular dysfunction in healthy subjects (Restaino et al. 2016) and improves limb vascular function in young and older individuals (Naylor et al. 2011; Romero et al. 2017). Further highlighting the clinical implications of heat therapy, recent data from a large prospective cohort study revealed that increased frequency of sauna bathing is associated with a reduced risk of fatal cardiovascular diseases and all‐cause mortality (Laukkanen et al. 2015).

Although this is the first study examining popliteal artery and calf muscle blood flow during insulin stimulation and thus we cannot directly compare our findings with findings from other studies, an observation that should be noted is the lack of insulin‐stimulated increased perfusion in the control, previously non‐heated leg. In vivo, the haemodynamic effects of insulin reflect the integrated response of neurohumoral and endothelial vasodilator and vasoconstrictor signals. We speculate that during insulin stimulation the control leg may be subjected to more vasoconstrictor forces compared to the previously heated leg. Specifically, it is plausible that ET‐1‐mediated vasoconstriction is enhanced in the control leg exposed to low shear stress relative to the previously heated leg. Notably, the effects of ET‐1 are not the sole opposing force to insulin‐stimulated NO‐mediated vasodilatation (Cardillo et al. 1999). Insulin also has centrally mediated effects that stimulate sympathetic nerve activity, which can limit insulin‐induced vasodilatation via α‐adrenergic vasoconstriction (Rowe et al. 1981; Creager et al. 1985; Anderson et al. 1992). In this regard, in patients with regional sympathectomy, insulin‐induced vasodilatation occurs more rapidly in the denervated limb than in the innervated limb (Sartori et al. 1999). As such, given prior work indicating that heating and NO can be sympatholytic (Patel et al. 2001; Jendzjowsky & Delorey, 2013; Gifford et al. 2014; Mizuno et al. 2014), it is possible that in our study direct heating and/or increased shear stress‐induced NO produced local sympatholytic effects during insulin stimulation, i.e. the loss of α‐adrenoceptor‐mediated restraint of skeletal muscle blood flow, otherwise existent in the control leg. Another possible explanation for the lack of increase in blood flow upon systemic insulin stimulation in the control leg could be related to the fact that lying down in bed without any leg movement is associated with a decay in leg blood flow over time, although this reduction only reaches statistical significance after 3 h of bed rest (Walsh et al. 2017). This tendency for leg blood flow to decline is likely secondary to the suppressed metabolism of fully inactive skeletal muscle. Thus, the lack of insulin‐stimulated increase in popliteal artery blood flow in the control leg should be interpreted in the context that, under non‐insulin stimulation, leg blood flow has the propensity to decrease over time during bed rest. Nevertheless, the primary finding is that prior exposure to heating and associated increase in leg blood flow unmasks the vasodilatory effects of insulin in that leg.

Some other considerations related to experimental design should be highlighted, including the levels of shear stress utilized in protocols 1 and 2. The precise magnitude and profile of shear stress to which human skeletal muscle arterioles are exposed at rest and during exercise has not been characterized, likely because current imaging techniques do not have the required spatial and temporal resolution. However, studies in isolated arterioles demonstrate that the amount of intraluminal flow required to exert 20 dynes cm⁻² can (1) be elicited by altering the pressure gradient across the vessel within a physiological range, and (2) produce a robust dilatory response (i.e. flow‐mediated dilatation) in healthy vessels. Accordingly, we consider 20 dynes cm⁻² to be a physiological dose of shear stress that skeletal muscle arterioles can experience in the setting of high perfusion as it occurs during exercise. In protocol 2, arterioles treated with this elevated dose of shear stress for 1 h were compared against arterioles exposed to time‐control no‐flow conditions. We opted to use this ‘all or none’ approach (i.e. use of the two ends of the spectrum) as the starting point in our endeavour to test if shear stress is implicated in the regulation of vascular responsiveness to insulin. Undoubtedly, future studies should perform dose–response experiments to better characterize the relationship between shear stress and vascular insulin sensitivity. In our experiments in isolated arterioles, after the 1 h of 20 dynes cm⁻² treatment, intraluminal flow was removed for a defined period of time prior to performing the insulin and SNP concentration–response curves. The rationale for removing flow was to allow for recovery and stabilization of vascular tone, thus to better resemble the time frame of the human experiment where the insulin clamp was initiated 1 h after cessation of heating (i.e. sufficient time for blood flow to return to baseline). Moreover, for accuracy, drug concentration–response curves in isolated arteries should be performed under no‐flow/static conditions in order to prevent the drug from being ‘washed away’. Another aspect of the experimental design that should be acknowledged is that our in vitro (used in protocol 1) and ex vivo (used in protocol 2) systems produce laminar shear stress. Given that blood flow is pulsatile, future research should also interrogate the effects of more physiological profiles of shear stress on insulin signalling in endothelial cells and vascular beds.

In conclusion, we provide evidence supporting the hypothesis that an acute increase in shear stress exerts insulin‐sensitizing effects in the vasculature and this evidence is based on experiments in vitro in endothelial cells, ex vivo in isolated arterioles and in vivo in humans. Given the recognition that vascular insulin signalling, and associated enhanced microvascular perfusion, contributes to glycaemic control and maintenance of vascular health, strategies that stimulate an increase in limb blood flow and shear stress have the potential to have profound metabolic and vascular benefits mediated by improvements in endothelial insulin sensitivity.

Additional information

Competing interests

The authors have no conflicts of interest to report.

Author contributions

L.K.W., T.G., T.D.O., C.A.E., J.R.L., C.M.‐A., L.A.M.‐L. and J.P. conceived and designed the research. L.K.W., T.G., T.D.O., A.M.‐H., J.C.E., P.K.T., C.M.‐A. and J.P. performed the experiments. L.K.W., T.G., T.D.O., A.M.‐H. and J.C.E. analysed the data. All authors contributed to interpretation of the results. L.K.W., T.G., T.D.O. and J.P. prepared the figures. L.K.W. and J.P. drafted manuscript. All authors edited and revised manuscript. All authors have read and approved the final version of this manuscript and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed.

Funding

This work was supported by the APS Research Career Enhancement Award (J.P.), the College of Veterinary Medicine – Committee on Research: Faculty Research Grant (C.A.E.), an Investigator‐Sponsored Research Grant from Lantheus Medical Imaging, Inc (J.P.), the National Science Foundation grant CBET‐1403228 (L.A.M.‐L.), and the National Institutes of Health grants: K01 HL125503 (J.P.), R01 HL137769 (J.P.), R01 HL088105 (L.A.M.‐L.), K08 HL129074 (C.M.‐A.) and R01 HL112998 (C.A.E.).

Biography

Lauren Walsh is a doctoral candidate in the Department of Nutrition and Exercise Physiology at the University of Missouri‐Columbia studying under the mentorship of Jaume Padilla. Her research interests include investigating therapeutic strategies aimed at improving insulin‐stimulated skeletal muscle blood flow in insulin‐resistant and type 2 diabetic patients, with a particular focus on clearly defining the role of haemodynamic forces (e.g. shear stress). She plans to continue pursuing a career in research, with the ambition to continue incorporating integrative and novel translational models to fully characterize the molecular and physiological mechanisms underlying vascular derangements in diabetes.