Differential effects of obesity on visceral versus subcutaneous adipose arteries: role of shear-activated Kir2.1 and alterations to the glycocalyx

Abstract

Obesity imposes well-established deficits to endothelial function. We recently showed that obesity-induced endothelial dysfunction was mediated by disruption of the glycocalyx and a loss of Kir channel flow sensitivity. However, obesity-induced endothelial dysfunction is not observed in all vascular beds: visceral adipose arteries (VAAs), but not subcutaneous adipose arteries (SAAs), exhibit endothelial dysfunction. To determine whether differences in SAA versus VAA endothelial function observed in obesity are attributed to differential impairment of Kir channels and alterations to the glycocalyx, mice were fed a normal rodent diet, or a high-fat Western diet to induce obesity. Flow-induced vasodilation (FIV) was measured ex vivo. Functional downregulation of endothelial Kir2.1 was accomplished by transducing adipose arteries from mice and obese humans with adenovirus containing a dominant-negative Kir2.1 construct. Kir function was tested in freshly isolated endothelial cells seeded in a flow chamber for electrophysiological recordings under fluid shear. Atomic force microscopy was used to assess biophysical properties of the glycocalyx. Endothelial dysfunction was observed in VAAs of obese mice and humans. Downregulating Kir2.1 blunted FIV in SAAs, but had no effect on VAAs, from obese mice and humans. Obesity abolished Kir shear sensitivity in VAA endothelial cells and significantly altered the VAA glycocalyx. In contrast, Kir shear sensitivity was observed in SAA endothelial cells from obese mice and effects on SAA glycocalyx were less pronounced. We reveal distinct differences in Kir function and alterations to the glycocalyx that we propose contribute to the dichotomy in SAA versus VAA endothelial function with obesity.

NEW & NOTEWORTHY We identified a role for endothelial Kir2.1 in the differences observed in VAA versus SAA endothelial function with obesity. The endothelial glycocalyx, a regulator of Kir activation by shear, is unequally perturbed in VAAs as compared with SAAs, which we propose results in a near complete loss of VAA endothelial Kir shear sensitivity and endothelial dysfunction. We propose that these differences underly the preserved endothelial function of SAA in obese mice and humans.

INTRODUCTION

Obesity is a complex and multifactorial metabolic disease that imposes well-established deficits to vascular endothelial function (1, 2). However, obesity-induced endothelial dysfunction is not observed in all vascular beds. Subcutaneous adipose arteries (SAAs) from obese humans exhibit a strong response to endothelium-dependent agonists such as flow and acetylcholine. In contrast, visceral adipose arteries (VAAs; e.g., mesenteric arteries) exhibit a relatively blunted response attributed to endothelial dysfunction (3, 4). Although mechanisms of VAA endothelial dysfunction are largely accepted to be mediated through a combination of impaired endothelial nitric oxide synthase (eNOS) function and/or reduced nitric oxide (NO) bioavailability (5–7), the underlying mechanisms driving the dichotomy in SAA versus VAA endothelial function in obesity are poorly understood. Identifying specific differences in adipose artery endothelial function will provide novel pathways to target in obesity-associated cardiovascular disease.

We recently showed that shear sensitive inwardly rectifying K⁺ (Kir) 2.1 channels are essential to flow-induced vasodilation (FIV), a hallmark of endothelial function (8), upstream of eNOS and NO production (9). We also showed that obesity abolishes shear-induced activation of VAA endothelial Kir2.1 channels by altering the biophysical properties of the glycocalyx (10). In the latter study, we identified a mechanism whereby a decrease in glycocalyx thickness, along with a concomitant increase in glycocalyx stiffness, with obesity resulted in loss of a Kir2.1 shear sensitivity in VAAs. These findings point to a potential tethering mode of activation of the channel by the glycocalyx in response to fluid shear, a relationship that is lost in obesity and results in endothelial dysfunction. Therefore, we aimed to identify the roles of Kir2.1 and the glycocalyx in the dichotomy observed between SAAs and VAAs in obesity.

In the present study, we first show that a mouse model of diet-induced obesity recapitulates the previously established human condition (3, 4); in contrast to VAAs that exhibit endothelial dysfunction in obese mice, the function of SAAs is preserved. We further show that endothelial cells isolated from SAAs of obese mice contain shear sensitive Kir channels that contribute to SAA FIV. In contrast, Kir shear sensitivity is abolished in VAA endothelium with obesity and endothelial Kir2.1 does not contribute to FIV in VAA from obese mice. Furthermore, we provide evidence that the SAA and VAA glycocalyx are differentially affected by obesity with VAA, exhibiting exacerbated stiffness and significantly reduced thickness compared with SAAs. We also provide translational relevance by showing that endothelial Kir2.1 channels are functional contributors to FIV in SAAs, but not in VAAs of obese humans. Therefore, we propose that differential alterations to the integrity of the glycocalyx renders endothelial Kir channels insensitive to fluid shear, ultimately contributing to endothelial dysfunction in VAA. These observations unveil a unique difference regarding the effects of obesity on spatially distinct adipose arteries. The identification of such specific differences in endothelial mechanisms of VAA dysfunction versus preserved SAA function will advance our understanding of obesity-induced endothelial dysfunction with the potential to lead to targeted, vascular bed-specific therapeutics.

MATERIALS AND METHODS

Animals

All studies involving animals were approved by the Institutional Animal Care and Use Committees at the University of Illinois at Chicago and the University of Delaware. Ten-week-old male C57BL/6J mice were purchased from Jackson Laboratories (Stock No. 000664) and randomly divided into diet groups. Lean controls were maintained on a normal laboratory diet and an obese group were fed a high-fat Western diet (42% kcal from fat; Envigo, Indianapolis, IN) for 4, 8, or 24 wk. We previously showed that male and female mice were equally affected by obesity regarding the effects on Kir channel function and the glycocalyx (10). Mice were euthanized by CO₂ asphyxiation followed by cervical dislocation. Subcutaneous and mesenteric (visceral) adipose were removed and first-order SAAs and VAAs were cleaned of parenchymal tissue in chilled HEPES-buffered saline before downstream analyses.

Transduction of Adipose Artery Endothelium with Adenoviral Constructs and FIV

Isolated SAAs and VAAs were incubated ex vivo with adenovirus containing either dominant negative (DN)-Kir2.1 or empty constructs [100 multiplicity of infection (MOI)] for 48 h under standard culture conditions (10% FBS DMEM medium, 5% CO₂, 37°C), as previously described (10, 11). Endothelial-specific expression, driven by the hVE-cadherin (human vascular endothelial cadherin) promoter, was confirmed in our earlier studies (9, 10). Following incubation with adenovirus (AV; Vector, Malvern, PA), isolated arteries were cannulated from either side onto glass pipettes in a chamber specialized for use in video microscopy. The chamber was placed on an Olympus CKX41 microscope equipped with a camera for visualizing the artery on a monitor and artery diameters were measured using a model VIA-100 video system (Boekeler). Arteries were gravity perfused from both sides with Krebs buffer from graduated cylinder reservoirs that were maintained at an equal height of 60 cmH₂O (equivalent to ∼44 mmHg) to pressurize the artery for 1 h before preconstriction with endothelin-1 (ET-1). ET-1 was used to constrict the artery between 40% and 50% of the baseline-pressurized diameter. Any arteries that did not constrict in this range to a maximum dose of 200 pmol/L of ET-1 were discarded from the study. Following preconstriction in the accepted range, the reservoirs were moved in equal and opposite directions to generate intraluminal flow through the vessel in a step-wise fashion. Each step increase in flow was administered for 5 min before measuring the lumen diameter. This method administers shear stress in the physiological range (∼10 dyn/cm² at Δ40 cmH₂O up to 50 dyn/cm² at Δ100 cmH₂O) (9). To confirm that vascular smooth muscle function remained intact throughout the duration of the experiment, we assessed dilations to papaverine (100 µmol/L) at the end of each protocol. If the diameter was not ≥80% of the baseline-pressurized diameter after addition of papaverine, no measurements collected from the artery were accepted for analysis. BaCl₂ (30 µmol/L) was added to the circulating bath after measuring the initial response to flow for 30 min before repeating the flow protocol (9). Dilations to flow in each artery were calculated as a percent change in dilation (%dilation) from ET-1 relative to the baseline diameter (Eq. 1):

$$\text{\%Dilation}=\frac{\text{baseline}\text{diameter}-\text{diameter}\text{after}\text{ET}1}{\text{baseline}\text{diameter}-\text{diameter}\text{in}\text{response}\text{to}\text{flow}\mathrm{\Delta }\mathrm{X}}\text{×}100.$$ (1)

Electrophysiological Recordings of Freshly Isolated Endothelial Cells

Endothelial cells from SAAs and VAAs of lean and obese mice were isolated as previously described (12, 13). Arteries were first cleaned of adipose tissue and incubated in digestion solution containing a cocktail of enzymes (0.5 mg/mL neutral protease and 0.5 mg/mL elastase; Worthington Biochemical Corporation) for 1 h at 37°C with gentle shaking every 10 min. Arteries were then incubated with 0.5 mg/mL collagenase type I (Worthington Biochemical Corporation) added to the cocktail for an additional 2–3 min. Arteries were then removed from the enzyme cocktail and placed in 1 mL of fresh digestion solution without enzymes. Endothelial cells were then mechanically liberated from tissue using two syringe needles. The digested tissue was further mechanically disrupted by triturating several times with a glass pipette. Solution (500 μL) containing a single cell suspension was immediately added to the parallel flow chamber and cells were allowed to adhere for 30–45 min at room temperature before electrophysiological recordings. Isolated endothelial cells, which we previously confirmed by CD31⁺ staining (10), were identified by morphology as previously described (10, 13, 14). The remaining 500 μL of solution containing cells was kept on ice until needed. Electrophysiological recordings were performed using soft glass pipettes (SG10 glass, Richland Glass) with resistances between 2 and 4 MΩ. Perforated whole cell patches were generated in a static bath by adding 250 µg/mL of amphotericin B to the pipette solution. After formation of the GΩ seal, perforated whole cell configuration occurred within 5 min. Cells were held at −30 mV, pipette and cell capacitances were automatically corrected, and currents were recorded using Pulsefit software and HEKA EPC10 amplifier. Only patches that maintained a series resistance of ≤50 MΩ were accepted for offline analysis. A ramp protocol of −140 mV to +40 mV over 400 ms was administered and currents were allowed to stabilize in a static bath before the addition of flow to the parallel flow chamber. Flow to the chamber was administered by gravity perfusion and generated a shear stress across patches of 0.7 dyn/cm² (Eq. 2):

$$\text{τ}=\frac{6\text{μQ}}{h^{2}w},$$ (2)

where τ is shear in dyn/cm², µ is the viscosity of the perfused bath solution (0.009 g/cm·s), Q is the flow rate (0.3 mL/s), h is the height of the chamber (0.1 cm), and w is the width (2.2 cm). Accepted recordings were leak subtracted offline and normalized to cell capacitance to compare current densities (pA/pF). SAA and VAA preparations from a single mouse were tested within 10 h post cell isolation. Kir currents from 2 to 4 cells were recorded per preparation (i.e., separate SAA and VAA preparations from a single mouse). Group data compared inward K⁺ current densities at −100 mV.

Atomic Force Microscopy

Isolated SAAs and VAAs were open en face on double-sided tape using microscissors (Fine Science Tools, Foster City, CA) in order to probe the glycocalyx and endothelial cell layer using atomic force microscopy (AFM). Adipose arteries were processed and experiments performed in phosphate buffered saline (Gibco, Gaithersburg, MD), pH 7.4. We previously confirmed that arteries processed in this way have intact endothelial glycocalyx (10). The elastic moduli of the glycocalyx and cell layer of en face arteries were measured with an Asylum MFP-3D-Bio atomic force microscope (Santa Barbara, CA) (15). The use of a highly flexible cantilever (spring constant of 0.01 N/m vs. typically used 0.10 N/m), as previously described for analysis of the glycocalyx using AFM (16, 17). A 4.5-µm polystyrene bead AFM probe (Novascan, Ames, IA) was used to facilitate even force dispersion. A total of 15–20 sites were analyzed per artery from 1 or 2 vessels per mouse (20–50 independent force-distance measurements/mouse). To obtain Young’s elastic moduli of the arterial sites as a measure of stiffness, E, at least five force/distance approach curves were analyzed by fitting the experimental curve to the Hertz model (Eq. 3):

$$\text{F}=4/3\frac{E}{1-{\text{v}}^2}\delta \frac{3}{z}\surd R,$$ (3)

where F is the force applied to the cantilever (loading force); δ is the indentation depth; ν is the cellular Poisson’s ratio (assumed to be 0.5 for incompressible biological material); and R is the radius of the spherical indenter (2.25 µm). Curve fitting was performed using Igor Pro data analysis software v. 6.3.7.2 (WaveMetrics Inc., Portland, OR) along with MFP-3D software v. 14.23.153 (Asylum Research, Santa Barbara, CA). The thickness of the endothelial glycocalyx was calculated as the distance, projected on the x-axis, between the starting points of the glycocalyx indentation curve and the starting point of the second curve denoting the force from the endothelial cell layer. The elastic modulus of the glycocalyx layer was assessed from a pointwise modulus at 50 nm indentation depth, as similarly performed in earlier studies (10, 18). These approaches were previously verified to separate the softer, more pliable glycocalyx from the stiffer cellular surface (16, 17).

Human Studies

Human studies were conducted in accordance with the Institutional Review Board at the University of Illinois at Chicago. Obese human subjects recruited to this study gave informed consent before the planned bariatric surgery date. All subjects had a body mass index of ≥38 and were between 26 and 48 yr old. Subjects were excluded from the study if they had diagnosed hypertension, diagnosed hypercholesterolemia, diabetes mellitus, cancer, heart disease, a history of smoking, kidney or liver disease, gallbladder disease, or autoimmune/inflammatory disease. Six subjects were female and one was male. Two subjects were Hispanic, three were African American, one was mixed ethnicity, and one was White. Subcutaneous and visceral (mesenteric) adipose biopsies were collected on the day of the surgery and arteries immediately cleaned of adipose and incubated with adenovirus, as described earlier. FIV was performed on isolated adipose arteries as described in Transduction of Adipose Artery Endothelium with Adenoviral Constructs and FIV. Only subjects that produced data from each artery type and adenovirus were included in the analysis.

Statistical Analyses

The Anderson–Darling test for normality was first used to identify whether the data set fit the normal distribution. One- or two-way ANOVAs were used and were followed by the Bonferroni post hoc tests, where appropriate. Two-way repeated measures ANOVA was used when comparing paired treatments across multiple groups. Nonparametric data sets were tested for significant differences using the Mann–Whitney U test (unpaired) or the Wilcoxon rank-sum test (paired) for comparing two groups. The Kruskal–Wallis test was followed by the Dunn–Sidak post hoc test for comparing multiple groups. Initial tests were set to P < 0.05 before post hoc corrections.

Solutions and Reagents

HEPES buffer contained (in mmol/L) 145 NaCl, 5 KCl, 2 CaCl₂, 1 MgCl₂, 10 glucose, and 10 HEPES at pH 7.4. Krebs buffer contained (in mmol/L) 123 NaCl, 4.7 KCl, 1.2 MgSO₄, 2.5 CaCl₂, 4 16 NaHCO₃, 0.026 EDTA, 11 glucose, and 1.2 KH₂PO₄ at pH 7.4. Digestion solution contained (in mmol/L) 55 NaCl, 5.6 KCl, 2 MgCl₂, 80 Na-glutamate, 10 HEPES, and 10 glucose at pH 7.3. Electrophysiology bath solution contained (in mmol/L) 80 NaCl, 60 KCl, 10 HEPES, 1 MgCl₂, 2 CaCl₂, and 10 glucose at pH7.4. Electrophysiology pipette solution contained (in mmol/L) 5 NaCl, 135 KCl, 5 EGTA, 1 MgCl₂, 5 glucose, and 10 HEPES at pH 7.2. BaCl₂ was purchased from Sigma-Aldrich.

RESULTS

A Mouse Model of Diet-Induced Obesity Recapitulates the Human Condition

An inherent limitation to study the mechanisms of differential endothelial dysfunction in different vascular beds in obesity is the lack of lean, nonobese controls as SAAs and VAAs are commonly collected from obese subjects undergoing bariatric surgery. Therefore, access to these adipose arteries in lean human subjects, which represent a crucial control to obesity-induced endothelial dysfunction, is difficult to obtain.

To circumvent this limitation, we first determined whether a mouse model of diet-induced obesity, which would allow for the inclusion of lean controls, recapitulates the obese human condition in that VAAs, but not SAAs, exhibit blunted FIV. Mice were randomly placed on a high fat, Western diet to induce obesity or maintained on a normal rodent diet (lean controls) for 4, 8, or 24 wk, and FIV was tested ex vivo in isolated SAAs and VAAs, as previously described (3, 19) (Fig. 1). No differences in the baseline arterial diameter were observed between artery type, diet group, or time on diet (Supplemental Table S1; all Supplemental material is available at https://doi.org/10.6084/m9.figshare.15046338). Despite significant elevations in body weight after 4 wk on the diet (Supplemental Table S1), SAAs and VAAs from both lean and obese mice responded to the flow similarly (Fig. 1, A, B, G, and H). After 8 wk on respective diets, VAAs from the obese mice exhibited blunted FIV, whereas SAAs from obese mice responded comparably to SAAs from lean counterparts (Fig. 1, C, D, G, and H). Similar observations occurred after 24 wk on respective diets with no further impairment of VAAs detected in obese mice (Fig. 1, E–H). This indicates that even a short time on the high fat, Western diet (e.g., 8 wk) results in an equal deficit to VAA endothelial function as prolonged feeding (e.g., 24 wk) in this mouse model of diet-induced obesity, whereas SAA function is fully preserved (Fig. 1, C–H). These findings support the use of the diet-induced obese mouse model in investigating mechanisms underlying the dichotomy in endothelial dysfunction with obesity in adipose arteries. As the 8-wk time point was the earliest in which we detected robust VAA endothelial dysfunction, we elected to conduct our studies at this time point with the goal of investigating early effects of obesity in VAAs versus SAAs.

Figure 1.

A mouse model of diet-induced obesity promotes endothelial dysfunction in VAAs but not SAAs. FIV was tested in pressurized and preconstricted SAAs and VAAs from the same mice after 4 wk (A and B; n = 4 mice/group), 8 wk (C and D; n = 7 mice/group), or 24 wk (E and F; n = 4 mice/group) on a normal rodent diet (lean) or a high fat, Western diet (obese). FIV was analyzed using a two-way ANOVA to identify significant effects of obesity on endothelial function (P < 0.05). A Bonferroni post hoc test was used to detect significant differences between diet groups at each intraluminal flow administered. *P < 0.01, significantly different vs. obese. FIV, flow-induced vasodilation; SAAs, subcutaneous adipose arteries; VAAs, visceral adipose arteries.

The Contribution of Endothelial Kir2.1 to FIV Is Preserved in SAAs of Obese Mice

We previously showed that endothelial Kir2.1 was critical mediator of FIV upstream of the Akt/eNOS/NO signaling axis (9) and that it is drastically impaired in VAAs in obesity (10). Therefore, we tested the role of endothelial Kir2.1 in SAAs versus VAAs from lean and obese mice to determine whether Kir2.1 differentially contributes to FIV in these distinct adipose arteries. To test the role of endothelial Kir2.1 in FIV of SAAs and VAAs of lean and obese mice, we incubated isolated SAAs and VAAs with an AV containing a DN-Kir2.1 construct with expression driven by the vascular endothelial (VE)-cadherin promoter for endothelium-specific downregulation. Expressing the DN-Kir2.1, which contains a substitution of the GYG region of the pore with AAA in the selectivity filter (20), abrogates Kir2.1 function (9, 21). Arteries were incubated for 48 h under standard culture conditions as described in our earlier studies which results in significant inhibition of FIV in VAAs of lean WT mice expressing the DN-Kir2.1 (10, 11). In contrast, arteries from lean mice incubated with the empty adenovirus (Em-AV) control dilated ∼80% or more in response to the maximum flow gradient of Δ100, similarly to what we observed in freshly isolated arteries of mice and humans (9, 22), indicating that the AV transduction and culture conditions do not impact endothelial function. In the present study, we further confirmed that viable endothelial function remains after 48 h incubation under culture conditions in the presence of an empty (Em)-AV control by assessing dilations to acetylcholine (ACh; Supplemental Fig. S1). Similar to our findings that there was no effect of the high-fat diet on FIV over time, dilations to ACh were not affected after 4 wk on the high-fat diet, reaching ≥80% dilation in response to 100 μmol/L ACh in each artery type and diet group following culture conditions (Supplemental Fig. S1, A and B). After 8 wk of feeding, VAA from obese mice exhibited significant reductions in the dilatory response to Ach, whereas SAAs from obese mice responded comparably to SAA of lean controls. These data further support the evidence for obesity-induced VAA endothelial dysfunction which is maintained following culture protocols in the presence of the Em-AV (Supplemental Fig. S1, C and D).

Functional downregulation of endothelial Kir2.1 in VAAs from lean mice resulted in a significant reduction in FIV (Fig. 2, B and F), as expected (10, 11). However, functional downregulation of Kir2.1 in VAA from obese mice did not have an effect on FIV when compared with VAA incubated with empty (Em)-AV (Fig. 2, D and F), which was significantly reduced compared with VAA from lean controls incubated with Em-AV (Fig. 2F). Functional downregulation of Kir2.1 in SAA from lean mice resulted in a mild but significant reduction in FIV compared with SAA incubated with an Em-AV control with no further effect of Ba²⁺, an inhibitor of Kir channels (23, 24) (Fig. 2, A and E). The lack of an additional effect of Ba²⁺, which we showed previously to blunt FIV through inhibition of Kir2.1 channels at low micromolar concentrations (30 μmol/L) (9), in arteries expressing DN-Kir2.1 in endothelium confirms complete downregulation of endothelial Kir2.1 with the DN-Kir2.1-AV. In contrast to VAAs from obese mice, expressing the DN-Kir2.1 in endothelium of SAAs from obese mice significantly blunted FIV, suggesting that endothelial function is preserved in SAA via Kir2.1-mediated signaling (Fig. 2, C and E). As obesity is well established to differentially impair eNOS signaling in VAAs and SAAs of obese humans (3, 4), and we previously identified that flow activation of endothelial Kir2.1 is upstream of eNOS activation (9), we next tested the role of eNOS in FIV in SAAs and VAAs from lean and obese mice exposed to the Em or DN-Kir.21 AVs. Inhibiting eNOS with N^ω-nitro-l-arginine methyl ester (100 μmol/L) significantly blunted FIV in arteries from lean mice and SAAs from obese mice incubated with the Em-AV but did not have an effect on arteries expressing the DN-Kir2.1 (Supplemental Fig. S2, A–C). These data suggest that the contributions of eNOS to FIV is dependent on the functional expression of endothelial Kir2.1, as reported previously (9–11). In contrast, inhibiting eNOS in VAA from obese mice had no effect on FIV in arteries exposed to Em-AV or arteries expressing the DN-Kir2.1 (Supplemental Fig. S2D). Together these findings indicate that, in contrast to VAAs, SAAs from obese mice maintain functional endothelial Kir2.1 that preserves endothelial function.

Figure 2.

Endothelial Kir2.1 contributes to FIV in SAAs, but not VAAs, from obese mice. FIV was tested in pressurized and preconstricted SAAs and VAAs incubated with either an Em-AV (Em) or the endothelium-specific DN-Kir2.1-AV (DN) from the same mice following 8 wk on a normal rodent diet (lean) or a high-fat, Western diet (obese). The effect of expressing DN-Kir2.1 in the endothelium of SAA (A and C) and VAA (B and D) on FIV in arteries isolated from lean or obese mice is shown (n = 5 mice/group for all). A two-way ANOVA was used to compare Em vs. DN vs. DN + Ba²⁺ (Ba; 30 μmol/L) in lean vs. obese mice (P < 0.05). A Bonferroni post hoc test was used to detect significant differences between diet groups at each intraluminal flow administered. *P < 0.01, significantly different vs. arteries expressing DN. A two-way repeated measures ANOVA was used to compare DN vs. DN + Ba²⁺ within diet and artery groups. DN, dominant negative; Em-AV, empty adenovirus; FIV, flow-induced vasodilation; SAAs, subcutaneous adipose arteries; VAAs, visceral adipose arteries.

Shear Sensitivity of Kir2.1 Channels Is Abolished in VAAs, but Not SAAs, Endothelial Cells of Obese Mice

To directly determine a role for endothelial Kir2.1 in SAAs and VAAs of lean and obese mice, we next tested the function of Kir2.1 in freshly isolated endothelial cells using the perforated whole cell patch clamp technique. Endothelial cells were freshly isolated and seeded into a flow chamber designed to give access to cells for recording currents during exposure of cells to fluid shear, as previously described (9–11). Inwardly rectifying K⁺ currents recorded during a voltage ramp protocol (−140 to +40 mV) in a 60-mmol/L K⁺ bath solution and 140-mmol/L K⁺ pipette solution reversed between −25 and −20 mV (Fig. 3), close to the calculated Nernst potential for K⁺ under these conditions. Following leak subtraction, little to no outward current was detected positive to reversal. These observations are in line with the biophysical properties of Kir channels in a high K⁺ bath (24).

Figure 3.

Shear sensitivity of Kir2.1 channels is abolished in VAA, but not SAA, endothelial cells of obese mice. Freshly isolated endothelial cells were isolated from SAA and VAA of lean and obese mice and seeded into the parallel plate flow chamber for electrophysiological recordings under fluid shear. Representative baseline Kir currents (I_Kir) recorded in a static bath from endothelial cells isolated from SAA (A) and VAA (B) of lean and obese mice are shown. C: group data (n = 6–10 cells from 3 to 5 mice/group) reveals significant effects of obesity on the baseline I_Kir densities at −100 mV. A Kruskal–Wallis test (P < 0.05) was followed by Dunn–Sidak post hoc analysis to determine significant differences between groups (*P < 0.0125). Representative recordings from endothelial cells isolated from SAAs (D and F) and VAAs (E and G) of lean and obese mice are shown before (static) and during (shear) application of flow to the chamber. D–G, insets: shear-induced increase in I_Kir (I_Sh-St) from the representative recordings. H: Kruskal–Wallis test for multiple comparisons (P < 0.05) was followed by Dunn–Sidak post hoc analysis to identify significant differences in shear-induced I_Kir between groups. *P < 0.0125, significantly different vs. respective lean control; †P < 0.0125, significantly different vs. SAA within diet group; n = 6–10 cells from 3 to 5 mice/group. Kir current (pA) was normalized to cell capacity (pF). Endothelial cell capacity was not different between artery type or diet groups tested by two-way ANOVA. pA/pF , cell capacitance to current densities; SAAs, subcutaneous adipose arteries; VAAs, visceral adipose arteries; Sh-St, static − shear.

As compared with Kir currents detected in lean cells, both baseline currents recorded in a static bath (Fig. 3, A–C) and the response to shear were significantly reduced in endothelial cells isolated from SAAs and VAAs of obese mice relative to lean counterparts (Fig. 3, D–I), suggesting that obesity impairs endothelial Kir channel function in both adipose artery types. In endothelial cells isolated from SAAs and VAAs of lean mice, robust shear sensitive Kir currents were similarly detected as shown in representative recordings from respective groups (Fig. 3, D and E). The shear-induced Kir currents (I_Sh-St) ranged from −1.1 to −12.0 pA/pF in SAA endothelial cells and −0.1 to −7.3 pA/pF in VAA endothelial cells from lean mice (Fig. 3H). However, as shown in the representative recordings in Fig. 3F, Kir currents detected in endothelial cells isolated from SAAs of obese mice exhibited shear sensitivity ranging from −0.1 to −3.6 pA/pF (Fig. 3, G and H). Despite both SAA and VAA endothelial cells of obese mice having significantly reduced basal level of Kir currents compared with lean controls, each SAA from an obese mouse exhibited an increase in Kir current in response to fluid shear, whereas the shear sensitivity of Kir in VAA endothelial cells of obese mice was virtually abolished with only 1/10 cells exhibiting a response to shear (I_Sh-St = −1.26 pA/pF; Fig. 3, G and H). Overall, when compared with lean control counterparts, there was an average ∼97% decrease versus a 78% decrease in shear-activated Kir currents with obesity in VAAs and SAAs, respectively. These findings suggest that, despite having reduced baseline Kir currents and blunted shear-induced Kir activation, the remaining SAA endothelial Kir channel shear sensitivity is adequate to contribute to FIV as observed in Figs. 1 and 2. In contrast, obesity promotes both a reduction in baseline Kir current and a near-complete loss of Kir shear sensitivity in VAA endothelial cells. Together, these findings suggest that Kir shear sensitivity is a major component in mediating SAA and VAA FIV, the absence of which contributes to obesity-induced endothelial dysfunction in VAAs.

Obesity-Induced Biophysical Alterations Are More Pronounced in VAA Glycocalyx

We previously showed that obesity promotes an increased stiffness and a decreased thickness of the VAA glycocalyx and that these altered biophysical properties negatively impact Kir shear sensitivity (10). Therefore, we compared the biophysical properties of the glycocalyx in SAAs and VAAs to identify whether differences exist between these two artery types in lean and obese mice using AFM, as previously described (10, 25). Briefly, adipose arteries were isolated and open en face so that the glycocalyx could be probed using a specialized AFM cantilever with a low spring constant, which allows for the detection of 1) the glycocalyx elastic modulus that informs on the stiffness of the glycocalyx and 2) the thickness of the glycocalyx. The presence of the glycocalyx manifests as a deviation from the Hertz model, observed as a break in the slope of the force-distance curve, and reveals the effects of obesity on the endothelial glycocalyx and underlying endothelial layer in the same curve (10, 16, 17, 25).

The effect of obesity on glycocalyx stiffness was significantly different in SAAs as compared with VAAs: only a mild increase in stiffness was observed in SAAs of obese mice compared with that of SAAs from lean controls, whereas a drastic increase in glycocalyx stiffness was observed in VAAs of obese mice compared with VAAs from lean controls (Fig. 4A). Histograms further reveal the increased stiffness in VAAs isolated from obese mice observed as a right shift toward a greater elastic modulus detection frequency (Supplemental Fig. S3). Furthermore, a pairwise comparison between SAAs and VAAs from lean mice revealed no significant difference in glycocalyx stiffness between these artery types (Fig. 4A, left). In contrast, VAAs exhibited a significantly stiffer glycocalyx than SAAs from obese mice (Fig. 4A, right). These data indicate that obesity-induced stiffening of the glycocalyx is more pronounced in VAAs. Furthermore, the endothelium elastic modulus, which informs on the stiffness of the endothelial cell layer and is analyzed from the same force-indentation curve from which the biophysical properties of the glycocalyx are assessed (10, 25), was elevated in VAAs of obese mice whereas the stiffness of the SAA endothelium was not significantly altered (Supplemental Fig. S4). As we recently showed that the stiffness of the glycocalyx was positively correlated to the stiffness of the endothelium (10), the increased VAA endothelial stiffness in obesity coincides with the drastic increase in the VAA glycocalyx stiffness but effects not observed in SAAs.

Figure 4.

Obesity-induced biophysical alterations are more pronounced in VAA glycocalyx. Adipose arteries were isolated and open en face so that biophysical properties of the glycocalyx could be assessed using atomic force microscopy. A: significant increases in SAA and VAA endothelial glycocalyx elastic modulus (EM) were detected with obesity following an independent samples Mann–Whitney test. *P < 0.05, compared with arteries isolated from the same adipose tissue of lean mice. A paired Wilcoxon rank-sum test was used to compare the glycocalyx EM of adipose arteries isolated from the same lean or obese mice. †P < 0.05, compared with arteries isolated from different adipose depots within diet group. B: significant decreases in SAAs and glycocalyx thickness with obesity were observed following an independent samples Mann–Whitney test. *P < 0.05 compared with arteries isolated from the same adipose tissue of lean mice. A paired Wilcoxon rank-sum test was used to compare the glycocalyx thickness of adipose arteries isolated from the same lean or obese mice. †P < 0.05, compared with arteries isolated from different adipose depots within diet group. For A and B, 15–20 measurements were averaged from each VAA and SAA from the same mice (n = 6–8 arteries from 4 to 5 mice/group) and are shown as connected dot plots. Bar graphs show average and means ± SE for each artery type from lean and obese mice. ns, not significant; SAAs, subcutaneous adipose arteries; VAAs, visceral adipose arteries.

We next assessed if obesity differentially affected the thickness of the glycocalyx in SAAs versus VAAs. The thickness of the glycocalyx was significantly reduced in SAAs and VAAs (Fig. 4B) from obese mice as compared with respective arteries from lean controls but the effect on VAA glycocalyx thickness was significantly more pronounced as shown by the pairwise comparison of arteries from obese mice (Fig. 4B, right). In contrast, no differences were detected between SAAs and VAAs of lean mice when analyzed in a pairwise fashion (Fig. 4B, left). Together, these findings reveal unequal obesity-induced biophysical perturbations to the SAA and VAA glycocalyx with relatively modest effects of obesity on SAAs.

The Contribution of Endothelial Kir2.1 to FIV Is Preserved in SAAs of Obese Humans

To determine if Kir2.1 has a differential role in SAAs versus VAAs of obese humans, obese human subjects undergoing planned bariatric surgery were recruited to this study. Subcutaneous and visceral (mesenteric) adipose biopsies were collected at the time of surgery and SAAs and VAAs isolated and incubated with Em-AV or AV-DN-Kir2.1 for 48 h under standard culture conditions before testing FIV, as previously described (10). SAAs and VAAs isolated from biopsies of obese humans and exposed to the control Em-AV support prior observations in that SAAs responded to intraluminal flow (∼74% dilation at Δ100 cmH₂O; Fig. 5) in a fashion consistent with healthy endothelial function, whereas VAA exhibited a relatively blunted vasodilatory response (∼48% dilation at Δ100 cmH₂O; Fig. 5) consistent with endothelial dysfunction (3, 4). As shown in Fig. 5, functional downregulation of endothelial Kir2.1 with AV-DN-Kir2.1 in SAAs resulted in a significant reduction in FIV as compared with SAA exposed to Em-AV (∼27% reduction in FIV at Δ100 cmH₂O; Fig. 5, A and B). In contrast, expressing DN-Kir2.1 in endothelium of VAAs from the same subjects did not significantly affect FIV as compared with VAA incubated with Em-AV (∼7% reduction in FIV at Δ100 cmH₂O; Fig. 5, A and B). These findings are in line with our observations in the mouse model of diet-induced obesity (Fig. 2) and provide translational evidence regarding the role of endothelial Kir2.1 in SAAs and VAAs of obese humans.

Figure 5.

Endothelial Kir2.1 contributes to SAA, but not VAA, FIV in obese humans. A: FIV was tested in pressurized and preconstricted SAAs and VAAs incubated with either an Em-AV (Em) or the endothelium-specific DN-Kir2.1-AV (DN) from obese human subjects (n = 7). Subcutaneous and visceral (mesenteric) adipose biopsies were collected at the time of planned bariatric surgery and SAA and VAA were isolated and incubated with respective AVs for 48 h before FIV. The effect of expressing DN-Kir2.1 (DN) vs. Em-AV (Em) in SAAs and VAAs from obese human subjects on FIV is shown. A two-way ANOVA was used to detect significant differences in FIV (P < 0.05). A Bonferroni post hoc test was used to detect significant differences between diet groups at each intraluminal flow administered. *P < 0.01, significantly different vs. arteries expressing SAA-DN; †P < 0.01, significantly different vs. VAA-Em. B: individual data points collected from each obese subject and artery tested show the percent dilation at the maximum flow rate of Δ100. Averages are shown as horizontal bars with means ± SE. DN, dominant negative; Em-AV, empty adenovirus; FIV, flow-induced vasodilation; ns, not significant; SAAs, subcutaneous adipose arteries; VAAs, visceral adipose arteries.

DISCUSSION

Obesity-induced endothelial dysfunction represents an early target in preventing more severe disease states and obesity-associated cardiovascular complications (1, 26). However, not all vascular beds are similarly afflicted by obesity-driven pathological mechanisms. Previous studies identified that arteries that exist in mesenteric visceral adipose exhibit endothelial dysfunction whereas those arteries that reside in subcutaneous adipose retain endothelial function (3, 4). Identifying specific differences in endothelial mechanisms of VAA dysfunction versus preserved SAA function will advance our understanding of obesity-induced endothelial dysfunction with the potential to lead to targeted, vascular bed-specific therapeutics. In the present study, we identified the shear stress sensitivity of endothelial Kir channels as a major factor in the distinct effects of obesity in VAAs versus SAAs. Specifically, we show that obesity imposes an exacerbated perturbation to the biophysical properties of the VAA endothelial glycocalyx, effects that we have found to impair Kir channel function (10), compared with SAAs. SAA endothelial Kir retains shear sensitivity in obesity, whereas VAA endothelial Kir lose shear sensitivity and, consequently, Kir2.1 contributes to FIV in SAAs from obese mice and humans but not in VAAs.

Our previous work demonstrated the critical role that endothelial Kir2.1 channels play in FIV (9–11), a dilatory mechanism implemented by the drag shear force on the endothelium that is composed of NO-dependent and NO-independent signaling pathways (27). We previously showed that the flow-induced Akt/eNOS/NO signaling axis required shear-sensitive endothelial Kir2.1 channels, events that comprised approximately half of the full dilatory response to flow (9). As deficits in eNOS function are well accepted to be a primary underlying cause of obesity-induced endothelial dysfunction (5–7), we hypothesized that obesity impairs Kir2.1 channels upstream of NO production. Furthermore, multiple studies have shown that the integrity of the endothelial glycocalyx, a mechanosensor that is also upstream of eNOS and NO production (28, 29), is reduced in obesity (25, 30, 31). In light of other studies identifying a role for the glycocalyx as a physical tethering point for mechanoactivated endothelial ion channels (32), we further predicted that obesity-induced insult to the glycocalyx disrupts shear-induced mechanoactivation of Kir2.1 channels, ultimately resulting in endothelial dysfunction. Indeed, we showed that Kir shear sensitivity was dependent on the glycocalyx and that increased stiffness and decreased thickness of the glycocalyx, as occurs in VAAs with obesity, disrupt Kir channel function (10).

These prior studies, however, were focused on mechanisms of endothelial dysfunction in arteries that show significant impairment and did not include those arteries that are seemingly unaffected by obesity. Therefore, we aimed to determine whether the function of Kir channels and endothelial glycocalyx are protected from obesity in seemingly fully functional arteries, such as SAAs. By first establishing that a mouse model of diet-induced obesity recapitulates the obese human condition (i.e., VAAs, but not SAAs, exhibit endothelial dysfunction in obesity), we allow for the inclusion of lean controls not readily available in human studies in subsequent analyses. Using a combination of ex vivo pressure arteriography to assess endothelial function via FIV, electrophysiology to assess shear-induced Kir channel activation, and AFM to assess the biophysical properties of the glycocalyx, we aimed to reveal if distinct differences exist between VAAs and SAAs with obesity regarding the shear activation of Kir channels and biophysical properties of the glycocalyx.

Several studies have reported on the deleterious effects of obesity on the endothelial glycocalyx in animal models and in humans (10, 25, 30, 31); however, no prior study examined if differences existed between SAAs and VAAs with obesity. We previously showed that the biophysical properties of the glycocalyx were critical to mediating Kir shear sensitivity: a stiffer and shorter glycocalyx reduced baseline Kir currents and rendered Kir channels insensitive to fluid shear (10). Here we show that in SAA, there was less stiffening of the glycocalyx with obesity as compared with VAA and, in accordance, the sensitivity of Kir to shear stress remains in SAA endothelium. The mild stiffening and reduced thickness of SAA from obese mice relative to that of lean control counterparts observed in the present study may also explain the reduction in the baseline Kir current as well as the reduced Kir shear sensitivity observed in perforated whole cell patch clamp experiments. In line with this, the exacerbated glycocalyx stiffness and reductions in glycocalyx thickness observed in VAAs likely result in abolished VAA Kir shear sensitivity and endothelial dysfunction. The ability to blunt FIV with expression of the DN-Kir2.1 in SAA endothelium of obese mice suggests that the remaining Kir shear sensitivity observed in SAA endothelial cells is adequate to preserve endothelial function in obesity despite being significantly reduced compared with the shear-activated currents detected in lean controls. These findings highlight the extent of alterations to the glycocalyx required to impair Kir channels and induce endothelial dysfunction as both SAAs and VAAs exhibited varying degrees of perturbation to the glycocalyx and reduced Kir function with only VAAs exhibiting overt endothelial dysfunction. Furthermore, no differences were detected between lean and obese SAA endothelium stiffness, an element which may impact the glycocalyx stiffness as we previously showed a positive correlation between endothelial layer and glycocalyx elastic moduli (10). In support of this, we identified a significant increase in VAA endothelial stiffness with obesity compared with VAAs from lean controls, which may further augment glycocalyx stiffness and, therefore, impair Kir and endothelial function. Further studies aimed at elucidating the relationship between an increase in endothelium stiffness and increased glycocalyx stiffness in obesity are warranted.

We recently showed that obesity does not affect the membrane expression of Kir2.1 in VAA endothelial cells, suggesting that a functional relationship between the glycocalyx and endothelial Kir channels is disrupted (10). However, the exact means by which the glycocalyx regulates shear-induced Kir activation has yet to be established. Recent evidence suggests that the glycocalyx acts as a physical tethering point for other mechanoactivated channels (33). In particular, shear activation of ENaC was shown to be dependent on the interaction between glycosylated asparagine of the channel and hyaluronic acids of the glycocalyx (32). Whether a similar biophysical tethering exists between the glycocalyx and endothelial Kir channels remains to be determined; however, we also showed that degrading specific glycocalyx components, including hyaluronic acids, abolished the shear sensitivity of Kir in vitro (10). Because obesity significantly decreased the thickness of the glycocalyx in VAAs and SAAs, the endothelial cells of which exhibited varying degrees of reduced Kir shear sensitivity, this may reflect the loss of glycosaminoglycans resulting in reduced interactions with the channel that are critical to transducing mechanical stimuli. Obesity also induced a severe stiffening of the glycocalyx in VAAs which likely impacts the “lever” function of a proposed tethering mechanism (34) that may also be required to transduce the shear stimulus to Kir channel opening. Studies aimed at determining the molecular and biophysical interactions between Kir, as well as other mechanoactivated ion channels, and the glycocalyx are needed.

The difference in VAA and SAA endothelial function is well established in obese humans (3, 4); however, underlying mechanisms dictating the dichotomy in endothelial function remain unresolved. A prominent hypothesis, termed visceral adiposopathy, or the “sick fat hypothesis,” proposes that the obesity-induced metabolic challenge to the spatially distinct adipose depots is handled quite differently when considering subcutaneous (containing populations of beige and white adipocytes) versus visceral (predominantly white adipocytes) adipose tissue (4, 35). The presence of the beige cell type in subcutaneous adipose may promote metabolic function toward using fat as opposed to storing it, thereby preventing severe lipid accumulation and associated adipocyte dysfunction (35) which may promote endothelial dysfunction via paracrine signaling in VAAs (4). Although both adipose depots expand with obesity, visceral white adipose expansion has been shown to correlate to insulin resistance and reduced endothelial function in the associated vasculature (36, 37).

Several studies have shown that arteries isolated from humans dilate in response to flow via endothelium-dependent mechanisms (38–40). Furthermore, we previously showed that in arteries isolated from human subjects FIV is dependent on Kir channel function (9). Here, we show that in VAAs and SAAs isolated from adipose biopsies of obese human subjects collected during bariatric surgery, downregulation of endothelial Kir2.1 inhibited FIV only in SAAs. In contrast, no effect of endothelial Kir2.1 downregulation was observed in VAAs, which exhibited markedly blunted FIV compared with SAAs regarding adipose arteries exposed to the Em-AV control. These findings provide new insight into the mechanism of endothelial dysfunction in VAAs versus SAAs of obese humans. Furthermore, we established that a diet-induced mouse model of obesity recapitulates the human condition prompting further use of this model to identify cellular and molecular mechanisms of VAAs versus SAAs endothelial function, studies that would otherwise be difficult to conduct exclusively using human tissues. Taken together, our data reveal that there are fundamental differences in endothelial Kir2.1 function in SAAs and VAAs in obese mice and humans, effects that may depend on biophysical alterations to the endothelial glycocalyx.

DATA AVAILABILITY

Data will be made available upon reasonable request. Please see the link to Supplemental Material in supplemental data.

SUPPLEMENTAL DATA

Supplemental Table S1 and Supplemental Figs. S1–S4: https://doi.org/10.6084/m9.figshare.15046338.

GRANTS

This study was supported by National Institutes of Health (NIH) Grants R01 HL-141120 (to I. Levitan and S. A. Phillips), HL-073965 (to I. Levitan and S. A. Phillips), R01 AG044404 (to J. C. Lee), and P20GM113125-6564 (to I. S. Fancher). This project was also supported by the Delaware IDeA Network of Biomedical Research Excellence program, with NIH Grant P20GM103446 from the National Institutes of Health and the State of Delaware (to I. S. Fancher), and a University of Delaware General University Research grant (to I. S. Fancher).

DISCLAIMERS

This content is solely the responsibility of the authors and does not necessarily represent the official views of National Institutes of Health.

DISCLOSURES

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

AUTHOR CONTRIBUTIONS

S.J.A., E.L.M., S.A.P., I.L., and I.S.F. conceived and designed research; S.J.A., E.L.M., and I.S.F. performed experiments; S.J.A., E.L.M., J.C.L., and I.S.F. analyzed data; S.J.A., E.L.M., J.C.L., S.A.P., I.L., and I.S.F. interpreted results of experiments; S.J.A., E.L.M., and I.S.F. prepared figures; S.J.A. and I.S.F. drafted manuscript; S.J.A., E.L.M., J.C.L., S.A.P., I.L., and I.S.F. edited and revised manuscript; S.J.A., E.L.M., J.C.L., S.A.P., I.L., and I.S.F. approved final version of manuscript.