Perivascular Adipose Tissue Controls Insulin-Stimulated Perfusion, Mitochondrial Protein Expression, and Glucose Uptake in Muscle Through Adipomuscular Arterioles

The diameter of skeletal muscle resistance arteries and muscle perfusion are regulated by a complex interplay of hemodynamic variables, circulating hormones, the autonomic nervous system, and local factors. Perivascular adipose tissue (PVAT), adipose tissue surrounding most peripheral arteries with an internal diameter >100 μm (1), is a source of adipokines within organs that has been proposed to locally regulate vascular function (2,3) and muscle insulin sensitivity (4).

After a meal, a rise in insulin induces dilation of resistance arteries within 10 min, a mechanism that regulates microvascular muscle perfusion (5,6). As a result, insulin increases microvascular blood volume (MBV) (i.e., insulin-mediated microvascular recruitment [IMVR]) in skeletal muscle (7), the primary target site for insulin-stimulated postprandial glucose uptake (8). IMVR expands the endothelial surface area in direct contact with blood, facilitating extraction of glucose and insulin into the muscle interstitium (7). IMVR relies on endothelium-dependent vasodilation (9), primarily relaxation of precapillary arterioles (7,10). These endothelial effects of insulin have been shown to control ∼50% of muscle insulin sensitivity (11).

PVAT secretes vasodilator hormones, such as adiponectin (12,13), and vasoconstrictor adipokines, such as adipocyte fatty acid–binding protein (FABP) (14,15). PVAT of healthy, lean humans and mice antagonizes sympathetic tone (2) and stimulates insulin-induced vasodilation in isolated resistance arteries (3,13). Loss of these vasodilator effects is a hallmark of PVAT dysfunction in human insulin resistance and type 2 diabetes (2,3,15). Vasodilator effects of PVAT are dependent on adiponectin (2,13), and decreased adiponectin secretion is characteristic of PVAT in type 2 diabetes (13). Adiponectin regulates insulin sensitivity (16) and muscle perfusion (17), and we have proposed that impaired cross talk between PVAT and microvascular endothelium predisposes to type 2 diabetes and cardiovascular disease (4). As PVAT is not present around the precapillary arterioles regulating IMVR, this proposed relationship includes proximal-to-distal transfer of adipokines, or “vasocrine” signaling, within microvascular beds.

In the current study, we investigated in vivo whether PVAT around larger resistance arteries directly communicates with proximal and distal muscle microvessels to control local vasomotor function, perfusion, and glucose uptake in muscle independently from whole-body metabolism. Using a microsurgical approach in mice, we tested this hypothesis by evaluating the effect of physical separation of local PVAT from muscle blood vessels on muscle perfusion. Hereafter, we repeated the measurements after surgically severing the connections between PVAT and the adjacent muscle, leaving PVAT attached to the resistance artery in situ. To assess whether proximal PVAT is directly connected to the distal muscle microcirculation, we studied the microvascular anatomy at the interface between PVAT and the muscle using light and fluorescence microscopy. Our study provides evidence for a role of PVAT in vivo in insulin-induced vasodilation, local regulation of IMVR, and skeletal muscle glucose uptake in vivo. We describe previously unrecognized adipomuscular microvessels that directly transfer PVAT-derived signals to the distal muscle microcirculation, regulating microvascular blood content. Finally, removal of PVAT from healthy muscle induces changes in muscle protein expression that have been shown to contribute to diet-induced insulin resistance.

Animal experiments were performed in accordance with the European Community Council Directive 2010/63/EU for laboratory animal care and Dutch law on animal experimentation. The experimental protocol was approved by the local committee on animal experimentation of Vrije Universiteit Amsterdam. In functional assays, we used C57BL/6 mice (male, age 8 weeks; Charles River Laboratories, Sulzfeld, Germany). VeCadherin-CreERT2; mTmG mice were generated by crossing VeCadherin-Cre ERT2 (Tg(Cdh5-cre/ERT2)CIVE23Mlia) (18) and mTmG (Gt(ROSA)26Sortm4(ACTB-tdTomato,-EGFP)Luo) (19). GFP expression was induced by a single injection of tamoxifen 24 h before sacrifice.

For manipulation of PVAT in vivo, a 2-cm skin incision was made under sevoflurane anesthesia parallel to the inguinal ligament. Four groups were studied: a group where PVAT was removed (PVAT-removed, n = 6) (Fig. 3B), a group where skin and released deep fascia were incised without removing PVAT (sham, n = 5), a group where PVAT was separated from the underlying muscle while leaving it attached to the resistance artery (PVAT-disconnect) (Fig. 3C), and a group without surgery (PVAT-intact) (Fig. 3A) (n = 13). The experimental design is shown in Fig. 1. All mice tolerated the surgical procedures well, with no loss of mice during surgery or in the postoperative period. No PVAT was observed in the operated area 2 weeks after PVAT removal (Fig. 3B). The weight of PVAT was comparable in both hind limbs (Supplementary Fig. 1A), and body weights were similar among mice in all experimental groups (Supplementary Table 1).

Insulin sensitivity was evaluated after an overnight fast by hyperinsulinemic-euglycemic clamp as described (6), using an insulin infusion rate of 7.5 mU/kg/min for 60 min.

Muscle perfusion in the thigh muscles was determined using contrast-enhanced ultrasonography (CEUS) as described (6).

Local muscle glucose uptake was determined by positron emission tomography (PET)-computed tomography (CT) scanning of uptake of ¹⁸F-2-fluoro-2-deoxy-d-glucose (¹⁸FDG) (BV Cyclotron VU, Amsterdam, the Netherlands). In six mice, we removed PVAT in one hind limb (n = 3 in the right hind limb, n = 3 in the left hind limb) and used sham surgery in the contralateral hind limb as an internal control.

During PET-CT (nanoScan; Mediso, Budapest, Hungary) a CT scan was performed for 6 min. After 15 min of hyperinsulinemia, ¹⁸FDG (7 MBq) was administered intravenously, and a dynamic emission scan of 1 h was performed. PET data were normalized and corrected for scatter, randoms, attenuation, decay, and dead time as described (20). After PET scanning, hind limb muscle and blood were obtained for determination of radioactivity in blood and muscle (20).

PET data were analyzed using AMIDE (A Medical Image Data Examiner) version 0.9.2 software (21), and fixed-size ellipsoidal-shaped regions of interest (ROIs) (dimensions 4 × 4 × 4 mm³) were manually drawn over predetermined areas of the left and right medial upper hind limb in the last frame of the image (three regions of the upper hind limb, medial to the femur: proximal, middle, distal corresponding to three different blood supplies to these regions). ROIs were projected onto the dynamic image sequence, and time-activity curve data were extracted. Time-activity curves were expressed as standardized uptake values: mean ROI radioactivity concentration normalized for injected dose and body weight.

The effect of PVAT from intact (n = 14) and sham-operated (n = 5) muscles on insulin-induced vasodilation was analyzed ex vivo as described (13) in gracilis resistance arteries of PVAT-removed mice.

Adiponectin secretion by intramuscular PVAT was measured as described (13) and corrected for PVAT weight.

Hind limb muscle of VeCadherin-CreERT2; mTmG mice was collected and fixed for imaging. Scanning laser microscopy was performed on a Nikon A1R+ system (Nikon Instruments). GFP and mTomatoRed were irradiated at 488 and 561 nm, respectively. A 20× air objective with numerical aperture of 0.5 was used to image the sample. Detection of the fluorescent signal was performed with gallium arsenide phosphide photo-multiplier tubes.

For light sheet fluorescence microscopy, muscle tissue was prepared according to the uDISCO protocol (22). An automated surface was applied using the surface tool to segment the resistance artery.

Protein expression in hind limb muscle tissue of four animals of the sham group and five animals of the PVAT-removed group was analyzed using mass spectrometry (MS) as described (23). Gel lanes were divided into five slices, and peptides were extracted for analysis by nano–liquid chromatography-tandem MS (MS/MS) on a Q Exactive mass spectrometer (Thermo Fisher Scientific, Bremen, Germany). Liquid chromatography-MS/MS, raw data processing, and database searching were performed as described (24–26). MS/MS spectra were searched against a Swiss-Prot Mus musculus reference proteome FASTA file with canonical proteins and isoforms (release August 2017, 25,052 entries).

Data are presented as mean ± SD. Differences among PVAT-removed, sham, and PVAT-intact were determined by one-way ANOVA with Tukey post hoc correction. For comparison between PVAT-removed and sham legs within the same mouse, a paired t test was performed. Analyses were done using GraphPad Prism 6.04 software (GraphPad Software, San Diego, CA). PET images were analyzed using AMIDE version 1.0.1 software (21). Differences were considered significant at P < 0.05.

For proteomics, differential analysis of normalized spectral counts for identified proteins was performed using the β-binominal test, which takes into account within- and between-sample variation (27). Protein-protein association data were retrieved from STRING version 10 (28) and visualized as networks in Cytoscape 3.5.1 (29). Heatmaps with hierarchical clustering (Euclidean distance, complete linkage) were generated in R version 3.4.3, using the ComplexHeatmap package version 1.17.1 (30).

The MS data for this publication have been deposited with the ProteomeXchange Consortium through the Proteomics Identifications (PRIDE) partner repository (www.ebi.ac.uk/pride/archive) and assigned the identifier PXD011179.

To assess the role of PVAT in regulation of insulin-stimulated muscle perfusion in vivo, we unilaterally removed PVAT from the gracilis muscle resistance artery and vein and the proximal part of the femoral artery (Fig. 3A and B). First, we verified the vasodilator properties of PVAT ex vivo by pressure myography. PVAT uncovered insulin-stimulated vasodilation in gracilis resistance arteries ex vivo, and microsurgery did not affect this vasodilator effect of PVAT (Fig. 2). Insulin (1 nmol/L) increased the gracilis artery (GA) diameter in the presence of PVAT (mean ± SD 26 ± 25%, n = 14; P < 0.0001) but not in absence of PVAT (0 ± 12%, n = 9). PVAT from the PVAT-removed and sham groups induced a similar vasodilator effect of insulin (20.2 ± 7%, n = 14; P = 0.64 vs. control PVAT) (Fig. 2).

We examined adiponectin secretion by intramuscular PVAT by measuring adiponectin in PVAT-conditioned medium. In line with previous data showing that PVAT is an intramuscular source of vasodilator adipokines, we found abundant adiponectin in PVAT-conditioned medium but no differences between control and insulin-stimulated conditions (mean ± SEM 2,750 ± 347 vs. 2,608 ± 306 pg/mL/mg; P = 0.7).

After confirming adiponectin secretion, vasodilator properties of PVAT, and lack of lasting effects of microsurgery thereon, we evaluated effects of PVAT manipulation on the muscle microcirculation in vivo using intravital microscopy and CEUS. We determined resistance artery diameters (Fig. 3A) before and during hyperinsulinemia in the PVAT-removed and sham-operated hind limbs and compared them to those of mice without any intervention (PVAT-intact) (Fig. 3). At baseline, resistance artery diameters were comparable among groups (Fig. 3D). In contrast, removal of PVAT fully inhibited insulin-induced vasodilation of gracilis resistance arteries (Fig. 3F). After 30 min of hyperinsulinemia, insulin increased GA diameter of PVAT-intact (14 ± 2 vs. −6 ± 4% in saline-infused mice) and sham (12 ± 3%) hind limbs but not in the PVAT-removed group (1 ± 2%). These differences persisted through 60 min of hyperinsulinemia, as insulin increased GA diameter of the PVAT-intact (15 ± 2%) and sham (19 ± 4%) compared with the PVAT-removed (2 ± 2%) and control mice that received saline (−6 ± 2%; P = 0.001) (Fig. 3F).

To test whether proximally located PVAT controls distal microvascular recruitment in muscle in response to insulin, we used CEUS. At baseline, MBV in muscle was similar among groups (Fig. 3E). Similar to its effects on proximal muscle resistance arteries, the presence of PVAT determined IMVR in muscle. In the sham group, insulin increased MBV by 44 ± 10%, comparable to the PVAT-intact group (36 ± 12%) and different from mice receiving saline infusion (−11 ± 11%). In the PVAT-removed group, IMVR was abrogated (−5 ± 7%) (PVAT-removed vs. sham and PVAT-intact P = 0.007) (Fig. 3G).

PVAT removal from muscle was not accompanied by a change in structural capillary density, which was similar between muscles of sham-operated and PVAT-removed hind limbs (mean ± SEM 539 ± 22 vs. 556 ± 38 capillaries/mm², respectively; n = 5/group; P = 0.7). In summary, removal of local PVAT from hind limb muscle specifically abrogated insulin-induced vasodilation in proximal resistance arteries and IMVR in the distal muscle microcirculation. Baseline resistance artery diameter, capillary density, and perfusion were not dependent on the presence of PVAT.

Using a second microsurgical approach, we unexpectedly discovered that anatomical connections between proximal PVAT and the adjacent muscle are critical to the effect of PVAT on IMVR in the distal capillaries. In six mice, we severed the connections between PVAT and the underlying muscle (PVAT-disconnect group) (Fig. 3C). In these mice, the local vasodilator response to insulin in the GA was preserved (20 ± 3%) (Fig. 3F), yet IMVR was fully inhibited (−2 ± 7%), similar to the PVAT-removed group (Fig. 3G).

We observed microbleeds at the interface between PVAT and muscle during the disconnection of PVAT from adjacent muscle and hypothesized that the effect of PVAT on distal IMVR is mediated by direct microvascular connections between PVAT and muscle (adipomuscular arterioles). In line with this, we observed small branches of the resistance artery extending through the PVAT into the underlying muscle (Fig. 4A). To examine the microvascular connections between PVAT and muscle in detail, we generated reporter mice expressing GFP in vascular endothelium (see research design and methods). Using this technique, we found that GFP-positive microvessels surrounding the perivascular adipocytes and capillaries running in parallel to muscle fibers were readily visible (Supplementary Fig. 2). When visualized in three dimensions, vascular structures extending from the adipocytes within PVAT to muscle fibers were seen (Fig. 4B–D, Supplementary Fig. 2, and Supplementary Movies 1 and 2). These adipomuscular arterioles were 20–40 μm in diameter and appeared to come together proximal to the PVAT-muscle interface and muscle fibers.

After observing that PVAT controls insulin-induced vasodilation and muscle IMVR in vivo, we investigated whether these regulatory effects of PVAT contribute to local insulin-stimulated glucose uptake in muscle. To this end, we compared the R_g in skeletal muscle between the PVAT-removed and sham hind limbs using PET with ¹⁸FDG. Whole-body insulin sensitivity, measured as the glucose infusion rate during the hyperinsulinemic-euglycemic clamp, was comparable among the PVAT-intact (180 ± 20 μmol/kg/min), PVAT-removed (204 ± 10 μmol/kg/min), PVAT-disconnect (216 ± 51 μmol/kg/min), and sham (216 ± 23 μmol/kg/min) groups.

In contrast to whole-body insulin sensitivity, local uptake of ¹⁸FDG in PVAT-removed hind limbs was markedly decreased compared with sham hind limbs (Fig. 5). In muscle proximal to the area where PVAT was removed, there was no significant difference in ¹⁸FDG uptake between the sham-operated and PVAT-removed hind limb (mean difference at 45 min after ¹⁸FDG injection ± SD −20 ± 36%; P = 0.54) (Fig. 5A). In the central area around the surgery site corresponding to the region perfused by the GA, ¹⁸FDG was decreased by a mean ± SD of 38 ± 28% and 23 ± 18% at 45 and 60 min after ¹⁸FDG injection, respectively (P = 0.009 and P = 0.04, respectively; n = 5) (Fig. 5B). In the distal ROIs corresponding to the area perfused by small arteries branching from the saphenous artery, ¹⁸FDG uptake was reduced in the PVAT-removed hind limb after 60 min (−20 ± 12%; P = 0.04) (Fig. 5C). After the hyperinsulinemic clamp, total ¹⁸FDG content in proximal muscle biopsies taken from PVAT-removed and sham hind limbs was determined. In these biopsies, ¹⁸FDG content was similar between muscle from PVAT-removed and sham hind limbs (2.0 ± 0.2 vs. 2.1 ± 0.2% injected dose/g; P = 0.58).

To gain insight into the mechanisms underlying the effects of PVAT removal on muscle physiology, we examined protein expression by MS-based proteomics. In hind limb muscle tissue, a total of 1,720 proteins were identified with at least one MS/MS spectrum (raw MaxQuant protein groups export in Supplementary Tables 2 and 3). Forty-five proteins were significantly upregulated (P < 0.05) in the PVAT-removed group (Fig. 6A), and 63 proteins were significantly downregulated compared with the sham group (Fig. 6B). Abundance of proteins was also visualized with heat maps (Supplementary Figs. 3 and 4). As expected, the adipocyte proteins FABP4 and 9 were reduced in PVAT-removed muscles.

A major protein in the upregulated network (Fig. 6A) was Hsp90ab1, which showed a moderate fold-change of 1.4. Proteins associated with HSP90ab1 included those linked to vesicular trafficking (Rab12, Nsf, Plaa, Pip4kb2, Hspa4, vinculin [Vcl]), albeit several of these were detected at a very low level. In addition, the cytoskeletal proteins Vcl and Arhgdia, a regulator of Rho GTPases such as Rac1 and Cdc42, and Capn2 (calpain 2), were also upregulated. Some of the upregulated proteins can also be involved in cellular stress and immune responses (Hsp90ab1, Hsp70 family member Hspa4, Ywhaq/14-3-3 protein θ, Il33, Serpinb1a). The effect of PVAT removal on expression of the most abundant upregulated muscle protein HSP90 was confirmed by Western blotting (Supplementary Fig. 5).

Several upregulated proteins have a mitochondrial localization (respiratory complex I components Ndufb4, Ndufs5, Ndufa6, Ndufa7; Cbr2, Qdpr, Gfm1, Mtx2, Mut, Bola1, Lrpprc). Surprisingly, three proteins involved in erythrocyte development (Hba-x, Hba-a1, Bpgm) showed an approximately twofold upregulation in muscle deprived of PVAT (Fig. 4A), although total Hb was not affected by PVAT removal. The upregulated Gyg/glycogenin protein is a primer for glycogen synthesis.

The major cluster of downregulated proteins following PVAT removal involved a different set of mitochondrial proteins, including respiratory complex III, IV, and V (ATP synthase) components (Fig. 6B). Furthermore, PVAT removal reduced the abundance of three proteins involved in regulation of glucose metabolism: the regulator complex protein LAMTOR3, the vesicle-associated membrane protein Vamp5 that regulates trafficking of Glut4 in insulin-treated cells (31) (fourfold downregulation; P = 0.02), and Cisd1/CDGSH iron-sulfur domain-containing protein 1 (1.4-fold downregulation; P = 0.01), which binds pioglitazone (32). In summary, removal of local PVAT from the proximal end of the muscle microvascular bed induced changes in protein composition of the connected muscles, highlighted by upregulation of protein components of mitochondrial complex I, downregulation of a cluster of protein components of mitochondrial complexes III, IV, and V, and downregulation of Vamp5.

Here, we demonstrate a role of PVAT in regulation of muscle perfusion, protein expression, and glucose uptake in vivo. We present three main findings (Fig. 7). First, PVAT in muscle is required for insulin-stimulated vasodilation and muscle IMVR. Second, the vasoregulatory function of PVAT is mediated by vascular connections extending from the larger arterioles to the underlying muscle through PVAT. Third, removal of normal PVAT reduces insulin-stimulated skeletal muscle glucose uptake in vivo and changes muscle protein expression, downregulating a cluster of mitochondrial proteins.

Insulin controls NO production through phosphatidylinositol 3-kinase and Akt (33) and endothelin release through extracellular signal-regulated kinase 1/2 (17). We have shown here that muscle resistance arteries do not dilate in response to insulin in the absence of PVAT. These data extend earlier in vitro evidence for vasodilator effects of PVAT (13) (Fig. 2). While this vasodilator effect critically depends on adiponectin in vitro (13), the presence of blood-borne adiponectin in the vascular lumen within muscles without PVAT in vivo is insufficient to permit insulin-induced vasodilation (Fig. 3F). As such, locally rather than systemically generated adiponectin determines insulin-stimulated vasodilation in muscle.

We did not observe differences in basal muscle MBV or basal resistance artery diameter among the PVAT-removed, PVAT-disconnected, sham, and PVAT-intact groups (Fig. 3D). This suggests that PVAT does not control basal tone of resistance arteries in vivo but specifically controls arteriolar responses to insulin. In support, isolated muscle resistance arteries also showed no differences in basal tone or general endothelium-dependent vasodilation, whether the artery was incubated with PVAT or not.

A mechanism we have previously proposed to mediate a regulatory action of PVAT is vasocrine, or vessel-to-vessel, signaling (4). As originally envisioned, PVAT-derived adipokines were secreted into the arteriolar lumen to dilate distal arterioles in response to insulin. Our present data support vasocrine signaling by PVAT yet suggest a more elegant mechanism. Instead of adipokine secretion into the arteriolar lumen and subsequent dilution in plasma before reaching terminal arterioles, adipomuscular arterioles enable PVAT-derived adipokines to reach the muscle tissue directly (Fig. 4). These microvessels may converge with the precapillary muscle arterioles, functioning primarily as arteriolar-capillary connections that deliver adipokines.

To test the function of the adipomuscular arterioles, we severed these microvessels by separating the outer surface of PVAT from the attached muscle tissue, while leaving the PVAT attached to the GA (PVAT-disconnect) (Fig. 3C). This intervention abrogated IMVR in the distal muscle microcirculation (Fig. 3G) but preserved proximal insulin-induced vasodilation (Fig. 3E), arguing that the paracrine effect of PVAT is still intact. The discrepancy between the effects of insulin on the proximal and distal parts of the muscle microcirculation after PVAT-disconnect surgery supports a functional role for adipomuscular arterioles in vasocrine signaling by PVAT.

Impairment of insulin-induced vasodilation and IMVR are commonly observed in insulin resistance and type 2 diabetes (34,35), reducing postprandial insulin delivery to insulin-sensitive tissues. We have shown that removing PVAT from muscle microvessels blunts both IMVR and muscle glucose uptake in vivo. The decrease in insulin-induced glucose uptake was evident in the central and distal ROIs drawn in the hind limb corresponding to the vascular regions of the GA and distal small arteries from the saphenous artery (36). Total hind limb muscle glucose uptake was not significantly different between muscles with and without PVAT, which may be explained by the fact that biopsies were taken from a muscle region supplied by the proximal deep profunda artery (36) that was not affected by the surgery. Indeed, several collateral routes exist in the innate muscle vasculature (37,38). Taken together, local PVAT depots surrounding muscle resistance arteries have a crucial role in IMVR, and when absent, insulin-induced muscle glucose uptake is decreased. The differences in skeletal muscle glucose reported in this study cannot be explained by systemic changes in insulin secretion or liver insulin sensitivity.

Removal of PVAT induced changes in myocellular protein expression typical of insulin resistance, particularly decreased expression of mitochondrial electron transport chain components (Fig. 6B). Mitochondrial dysfunction is a hallmark of muscle insulin resistance in type 2 diabetes (39), and decreased expression of cytochrome C and ATPase has been observed in muscles of insulin-resistant human subjects (40) and mice (41). Furthermore, mitochondrial function has been shown to be regulated by adiponectin and AdipoR1 receptors (42,43). HSP90ab1 (HSP90β), which was upregulated in muscles after PVAT removal (Fig. 6A), is also upregulated in diet-induced insulin resistance (44) and reduces muscle mitochondrial function and glucose uptake (44). These data are consistent with a model wherein intramuscular PVAT regulates muscle glucose metabolism by secreting adiponectin, reducing HSP expression, and increasing expression of mitochondrial proteins. In health, this regulation is independent from oxidative stress, as expression of oxidative stress markers, such as superoxide dismutase and catalase, was normal in muscle without PVAT.

These metabolic effects of PVAT add to the growing body of evidence on endocrine effects of adipose tissue in the pathogenesis of type 2 diabetes. While intra-abdominal adipose tissue has established roles in liver insulin resistance and impaired insulin secretion (15,45), our data suggest that paracrine signaling by intramuscular PVAT contributes to regulation of muscle insulin sensitivity in addition to endocrine regulation by circulating adipokines.

A few limitations of our study should be considered. It is difficult to standardize the amount of PVAT removed by different investigators, and the surgery itself could have influenced local perfusion and metabolism through inflammation. To control for effects of mechanical trauma, we applied sham surgery to the contralateral hind limb (Fig. 3). In addition, we used a 2-week recovery period between surgery and measurements. Finally, we tested GA from the PVAT-removed group ex vivo in the pressure myograph to validate preserved PVAT function and found normal vasodilation in response to insulin (Fig. 2) and acetylcholine. Absence of a chronic inflammatory response after microsurgery is supported by protein expression of the inflammatory proteins macrophage migration inhibitory factor 1 (MIF-1), the mast cell markers chymase and mast cell carboxypeptidase A, and the stress protein MAPK14, which were all similar between muscles of sham and PVAT-removed hind limbs. Classic inflammatory mediators, such as tumor necrosis factor-α, interleukin 6, interleukin 1β, and MCP-1, were undetectable in the muscles studied.

The critical products of PVAT mediating local actions in vivo as well as their targets remain to be determined because nonsurgical interventions to alter PVAT function still have to be developed. As a final limitation, it should be noted that insulin was the only vasoactive agent used, and the effects of PVAT on responses to other vasoactive agents in vivo are uncertain.

In conclusion, our data demonstrate a local regulatory role for PVAT in muscle perfusion, mitochondrial function, and glucose metabolism mediated by dedicated microvascular connections. These data open new avenues for treatment of type 2 diabetes and the associated organ failure.

Acknowledgments. The authors are grateful to Elisa Meinster (Pathology, Amsterdam University Medical Center [UMC]), Mariska Verlaan and Ricardo Vos (Radiology and Nuclear Medicine, VU University), Nanne Paauw (Molecular Cell Biology and Immunology, Amsterdam UMC), and Jeroen Kole, and Zeineb Gam (Physiology, Amsterdam UMC) for expert technical assistance. The authors also thank Professor A.K. Groen (Internal Medicine, Amsterdam UMC, Amsterdam, the Netherlands) for valuable discussion of the manuscript.

Funding. This work was supported by the Netherlands Organisation for Health Research and Development (ZonMW, VIDI grant 917.133.72) and the Netherlands Organisation for Scientific Research (NWO/ALW grant ALWOP.271).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. A.H.T. wrote the manuscript and researched and analyzed data. E.H.S., Y.M.S., C.R.J., and V.W.M.v.H. contributed to discussions, supervised, and edited the manuscript. J.J.K. analyzed data and edited the manuscript. J.K. researched and analyzed data and edited the manuscript. C.F.M.M., H.W.N., M.J.T.H.G., E.M.v.P., and J.S.Y. contributed to discussions and edited the manuscript. E.C.E. contributed to discussions, supervised, and edited the manuscript. E.C.E. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in oral form at the 76th Scientific Sessions of the American Diabetes Association, New Orleans, LA, 10–14 June 2016, and in poster form at the 78th Scientific Sessions of the American Diabetes Association, Orlando, FL, 22–26 June 2018.