Introduction
Arterioles, capillaries and venules make up the microcirculation. Important functions of the microcirculation are to dynamically optimise nutrient and oxygen supply to tissues, and to regulate peripheral resistance [
1,
2]. Arterioles regulate flow towards different sites by changing tone [
1]. One of the hallmarks of obesity is reduced vasoreactivity, increasing BP and insulin resistance through increased peripheral resistance and decreased delivery of insulin and glucose [
3,
4].
Insulin regulates perfusion in the muscle microcirculation, stimulating vasodilation through activation of the phosphoinositide 3-kinase–Akt–endothelial nitric oxide synthase (PI3k–Akt–eNOS) pathway, and concomitantly enhancing vasoconstriction through activating the extracellular signalling regulated kinase 1/2–endothelin 1 (ERK 1/2–ET1) pathway [
3]. In insulin-sensitive individuals, activation of the vasodilator pathway dominates, increasing muscle perfusion during hyperinsulinaemia, so-called ‘microvascular recruitment’. In muscle, this augments insulin-stimulated glucose uptake [
4‐
6]. Insulin-induced microvascular recruitment is blunted in insulin-resistant states such as obesity, and in turn contributes to insulin resistance [
5,
7,
8]. Why insulin-induced microvascular recruitment is blunted in obesity is unclear.
We recently identified perivascular adipose tissue (PVAT) around skeletal muscle resistance arteries in mice as a new depot of ectopic adipose tissue, and proposed a regulatory role of PVAT in muscle perfusion and insulin sensitivity [
9,
10]. In mice, we demonstrated that ex vivo, PVAT exerts paracrine effects on muscle resistance arteries [
11]. These paracrine effects enhance insulin-induced vasodilation in lean mice, are mediated by adipokines and are abolished in obesity [
11]. The anticontractile effect of healthy PVAT extends previous findings that PVAT controls vascular smooth muscle contractility [
11‐
15]. Changes in PVAT function in obesity may be caused by inflammation [
13,
16,
17]. Whether PVAT in the muscle microcirculation enhances insulin-induced vasodilation and microvascular recruitment in insulin-sensitive humans is unknown.
In this study, we hypothesised that PVAT influences insulin-induced vasodilation in the skeletal muscle microcirculation and that its effect differs between lean and obese individuals.
Discussion
The relation between microvascular PVAT and microvascular vasomotor responses in vivo was hitherto unknown. This study demonstrates a direct relation between PVAT characteristics and insulin’s effects on muscle perfusion. More specifically, perivascular adipocyte size mediates the difference in insulin-induced microvascular recruitment between lean and obese women. These results were extended by ex vivo evidence that PVAT from lean women potentiates the vasodilator effect of insulin, whereas PVAT from obese women causes insulin-induced vasoconstriction. These findings suggest that PVAT regulates insulin-induced vasodilation, and insulin-induced microvascular recruitment in skeletal muscle.
We studied PVAT which abuts the microcirculation, and provide direct evidence for a functional role of PVAT in the regulation of human skeletal muscle perfusion. As others have found, flow-mediated microvascular vasodilation is related to PVAT around the brachial artery [
23]. In PVAT and vessels obtained from subcutaneous adipose tissue of lean humans, PVAT shows an anticontractile effect ex vivo in the absence of insulin, which is lost in obesity [
12], and can be restored by bariatric surgery [
24]. In the latter study, a reduced macrophage count in obese PVAT after bariatric surgery was found. We did not find a difference in PVAT macrophage content between lean and obese women, possibly because our obese participants were less extremely obese and were healthy. A difference in macrophage content in PVAT was also not found by others in high fat diet fed mice, despite altered PVAT function [
25]. Macrophage count may therefore not adequately reflect the pro-inflammatory potential of PVAT [
13,
26].
Our data demonstrate that insulin-induced microvascular recruitment is a significant mediator in the relationship between obesity and metabolic insulin sensitivity (Fig.
2b). Moreover, our results show PVAT adipocyte size to be a major determinant of the difference in the magnitude of insulin-induced microvascular recruitment between lean and obese women, even though one of the component analyses was of borderline significance (
p = 0.065; Fig.
3c). These observations suggest that perivascular adipocyte size is more important than being lean or obese per se. Perivascular adipocyte size itself seems an unlikely direct cause of altered PVAT phenotype. More likely, larger adipocyte size is a proxy for altered PVAT characteristics (e.g. hypoxia, inflammation) and therefore relates to an altered secretory adipokine profile [
27]. Because we did not find any difference in macrophage infiltration, we did not further explore inflammation as a mediator in the relationship between obesity and microvascular recruitment. Indeed, others have also shown a relationship between adipocyte size and insulin resistance, irrespective of inflammation [
28], but also between adipocyte size and adipokine gene expression [
29]. Nevertheless, the results show clear mediating effects of microvascular recruitment on metabolic insulin sensitivity, and of PVAT adipocyte size on microvascular recruitment. This means that in order to normalise microvascular responses in obesity, normalising PVAT properties could be of key importance. This is further supported in the study describing PVAT effects after bariatric surgery, where BMI was still in the obese range, but microvascular responses to PVAT were comparable with those of lean healthy participants [
24]. Novel ways to decrease adipocyte size, and in particular perivascular adipocyte size, are therefore worth investigating.
The mediation model by Preacher and Hayes was originally designed to study mediation effects in large datasets, but has recently been applied in smaller studies as well. When we studied the same relations by solely looking at the change in β coefficient, this yielded similar results, demonstrating the robustness of the data. We decided to report the mediation model results because these provide a more insightful analysis of the mediating effect, together with an estimate of significance.
The results described in Fig.
4 show that PVAT has vasoactive effects in conjunction with insulin. PVAT helps explain the differences in microvascular recruitment between lean and obese participants. As demonstrated, PVAT is necessary for insulin to enhance vasodilation, and therefore insulin-induced microvascular recruitment in vivo. In the absence of PVAT, the microvessels of lean and obese women respond to insulin identically, i.e. they do not change diameter. This suggests that, even though insulin signalling might be different in the endothelium of lean and obese women, this is not sufficient to affect insulin-induced vasoreactivity. However, in the presence of PVAT, a different behaviour of the microvessels is revealed with insulin-induced vasodilation in lean participants, and insulin-induced vasoconstriction in the obese. The divergent responses in the presence of PVAT also demonstrate the dual activation of vasoactive signalling cascades by insulin [
30‐
32]. The importance of PVAT is also demonstrated by previous studies that may not always have removed PVAT properly, thereby potentially influencing their results [
33]. Our present results support the hypothesis that PVAT is a functional determinant of microvascular recruitment in skeletal muscle, and therefore of insulin resistance.
Mechanistically, hypoxia and inflammation can alter the adipokines secreted by PVAT. The effects of the hypoxia in obese PVAT can be inhibited with free radical scavengers, improving the effect on microvascular vasodilation [
12]. Hypoxia may induce c-Jun N-terminal kinase (JNK) activation in PVAT of obese individuals [
11,
34], inhibiting the vasodilator effect of PVAT. On the other hand, adiponectin R1 agonists such as adiponectin have been shown to propagate the vasodilator effects of lean PVAT through signalling via AMP-activated protein kinase α2 (AMPKα2) [
11,
12,
35], and its secretion decreases when fat is inflamed. Others have described communication pathways from the endothelium to PVAT, where PVAT function changes in response to endothelial stress in obesity, in order to negate this stress [
36,
37]. Our current results cannot confirm or refute that hypothesis, but they at least show that if such a response occurred here, it is incomplete and fails to normalise the microvascular response to insulin.
It is worth mentioning some limitations to this study that need to be considered in conjunction with the results. To the best of our knowledge, this is the first study in which isolated human skeletal muscle microvessels were directly examined in an ex vivo vascular function experiment. Preconstriction was established through 25 mmol/l potassium, which is high compared with interstitial concentrations in vivo, but low compared with other studies examining ex vivo vasoreactivity [
12,
24]. PVAT was physically separated from all microvessels to prevent concerns about damaging microvessels during surgery, or mechanical restrictions of PVAT on vasoreactivity. Despite our best efforts, there may be some degree of selection bias of the microvessels inherent in these experiments, possibly favouring larger microvessels. Although diameters did not differ between the two groups, higher orders of microvessels may have been selected in participants with inward remodelling. Different orders of microvessels might be regulated differently during microvascular recruitment, although no evidence exists for that with regard to insulin-induced microvascular recruitment [
38]. Moreover, microvascular recruitment measured by CEU is impervious to this theoretical bias, leading us to deem this theoretical bias negligible in this study. Furthermore, as most obese participants had long-standing obesity, they might exhibit long-standing endothelial dysfunction, so failing the quality 10% vasodilation to ACh 10
−6 may be due to experimental circumstances, or established endothelial dysfunction.
The insulin resistance in the obese group was not extreme, probably due to the exclusion of women with diabetes and hypertension. But despite that, they performed worse with regards to insulin-induced microvascular recruitment and perivascular adipocyte size, compared with our lean women. This shows that even in the phase before the development of obesity-associated complications, PVAT is an important factor, which assumingly would only become stronger were obese women with obesity-associated complications to be included. Despite these reservations, the results align with our own and others’ previous results [
11,
12,
15].
Summarising, we have found that PVAT adipocyte size partly explains the relationship between obesity and blunted insulin-induced microvascular recruitment through direct regulation of insulin’s microvascular effects. Therefore, PVAT may be an important target for the treatment of obesity-associated microvascular dysfunction.
Acknowledgements
We thank I. Knufman (Department of Internal Medicine, VU University Medical Center, Amsterdam, the Netherlands) for excellent technical assistance.