Perivascular Adipose Tissue Control of Insulin-Induced Vasoreactivity in Muscle Is Impaired in db/db Mice

This investigation conforms to the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health (publication no. 85-23, revised 1996).

All animal work was approved by the local ethics committee for animal experiments of the VU University Amsterdam and complied with Dutch government guidelines. Male C57BL/6NCrl (further indicated as C57BL/6) (Charles River, Amsterdam, the Netherlands), male db/db (Harlan, Zeist, the Netherlands), and AMPKα2^−/− and AMPKα2^+/+ mice 10 weeks of age were killed by isoflurane overdose. To study the role of AMPK in insulin-induced vasoreactivity, male C57BL/6N mice (back-crossed for >10 generations; Taconic, Ejby, Denmark) with a deletion of the AMPKα2 (AMPKα2^−/−) catalytic subunit gene, together with their AMPKα2^+/+ littermates, were generated as previously described (26).

PVAT was dissected in one piece, free of surrounding connective tissue, from C57BL/6 (n = 6) and db/db (n = 6) mice and was quantified on a microbalance. Figure 1A gives an impression of the stretch of the dissected PVAT.

To study the insulin-induced vasoreactivity of resistance arteries in the absence of PVAT, first-order resistance arteries from the gracilis muscle were isolated from lean C57BL/6 mice. After dissection, resistance arteries were placed in a pressure myograph, cannulated, and studied at a pressure of 80 mmHg and a temperature of 37°C in K-MOPS buffer, as described previously (16). In addition to the resistance artery itself, PVAT that surrounds the resistance artery was isolated from the origin of the gracilis resistance artery at the femoral artery to its first major side branch (Fig. 1A). PVAT derived from the same stretch of the resistance artery in C57BL/6 mice was also isolated from db/db mice. To study the interaction between PVAT and insulin-induced vasoreactivity of isolated resistance arteries, PVAT was coincubated with resistance arteries in the pressure myograph for 45 min (Fig. 1B). Resistance arteries were randomly assigned to incubation without PVAT (n = 9), C57BL/6 PVAT (n = 10), or db/db PVAT (n = 8) of approximately equal amounts. After incubation, the inner diameter of resistance arteries was recorded to determine the baseline diameter after preconstriction to ∼50% with KCl (25 mmol/L), and diameter changes induced by four concentrations of insulin (0.02, 0.2, 2.0, and 20 nmol/L; Actrapid; Novo Nordisk) were examined for 30 min (16). The three lowest concentrations of insulin are within the physiological range, with the third concentration (2.0 nmol/L) corresponding to postprandial levels and the fourth concentration to pharmacological levels. Endothelial integrity was determined by measuring responses to the endothelium-dependent vasodilator acetylcholine (0.1 μmol/L) after each experiment, with a minimum dilation of 10% used as proof of endothelial integrity.

Moreover, it was investigated whether reducing concentrations of adiponectin would affect insulin-induced vasoreactivity in resistance arteries from C57BL/6 mice with PVAT from C57BL/6 mice (n = 6). A soluble receptor fragment (Acrp-30 [N-20]-P adiponectin receptor blocking peptide; Santa Cruz Biotechnology) (0.4 μg/mL), which binds adiponectin, was preincubated in the pressure myograph for 30 min. Additionally, globular adiponectin (Enzo Life Sciences, Antwerpen, Belgium) was preincubated for 45 min in the vessel bath of the pressure myograph at a concentration of 1 μg/mL with C57BL/6 resistance arteries without PVAT (n = 5) to investigate whether adiponectin can affect insulin-induced vasoreactivity when added from outside the resistance artery. Globular adiponectin was also incubated in the presence of db/db PVAT (n = 7) to study whether it could restore insulin-induced vasodilation. As a model to inhibit downstream adiponectin signaling, the effects of the AMPK inhibitor Compound C (1 μmol/L; product number 171260; Calbiochem) on insulin-induced vasoreactivity were examined in C57BL/6 resistance arteries in the presence of C57BL/6 PVAT (n = 7).

To further study the role of AMPKα2 in the interaction between PVAT and insulin-induced vasoreactivity, resistance arteries from AMPKα2^+/+ and AMPKα2^−/− mice were collected using the same techniques. Resistance arteries of AMPKα2^−/− (n = 8) and AMPKα2^+/+ (n = 6) were mounted in the pressure myograph and coincubated with PVAT from AMPKα2^+/+ mice for 45 min. After incubation, inner diameter and responses to insulin and acetylcholine were studied as described above.

To study whether inflammatory adipokines are involved in the regulation of insulin-induced vasoreactivity by PVAT, we tested whether one phenotype of PVAT would prevail during dual incubation with both C57BL/6 and db/db PVAT (n = 6).

Pressure myography was used to test whether acute inhibition of inflammation using L-JNKi (L-JNKi 1 trifluoroacetate salt, 5 µmol/L; Sigma Aldrich, Zwijndrecht, the Netherlands) could restore vasodilator properties of db/db PVAT (n = 6). Furthermore, the soluble adiponectin receptor fragment (Acrp-30 [N-20]-P adiponectin receptor-blocking peptide, 0.4 μg/mL; Santa Cruz Biotechnology) was coincubated with L-JNKi in experiments with db/db PVAT to test whether the restored vasodilation by L-JNKi was mediated by adiponectin (n = 6).

In order to investigate the effects of C57BL/6 and db/db PVAT on downstream insulin signaling in resistance arteries, Western blot analysis was performed as previously described (16). After 45 min incubation in the presence of PVAT from C57BL/6 or db/db mice, segments of gracilis arteries from C57BL/6 mice were exposed to solvent or insulin (2 nmol/L) for 30 min at 37°C. To investigate whether JNK inhibition restored insulin-mediated Akt phosphorylation in gracilis arteries exposed to db/db PVAT, other arterial segments, exposed to C57BL/6 or db/db PVAT, were exposed to insulin or solvent after pretreatment with L-JNKi. The protein lysates of resistance artery segments were stained with a specific primary antibody against Ser 473–phosphorylated Akt and threonine 172–phosphorylated AMPKα (antibodies from Cell Signaling Technology, Boston, MA) and were visualized with a chemiluminescence kit (GE Healthcare, Diegem, Belgium). As protein yields from gracilis muscle resistance arteries are low (<2 µg), and we were not able to stain AMPKα after stripping, we avoided stripping and reprobing blots. To obviate this, we used ERK1 as a loading control instead of total Akt or AMPKα. ERK1 correlates well with total Akt levels (27). As all experiments were paired and the duration of experiments was 1.5 h in total, differences in protein levels of Akt and AMPKα between experimental conditions are unlikely.

To study whether PVAT from lean C57BL/6 mice secretes more adiponectin than PVAT from db/db mice, the amount of secreted adiponectin was quantified. PVAT isolated from lean (n = 7) and db/db (n = 7) mice, in amounts of comparable size, was incubated with 2 nmol/L insulin for 45 min at 37°C in order to study the release of adiponectin. Adiponectin concentration in conditioned medium was measured with a mouse adiponectin ELISA kit (Millipore, Amsterdam, the Netherlands) and detected using the Luminex system.

Steady-state responses are reported as mean change from baseline (percentages) ± SEM. Baseline diameter was determined as the diameter just prior to the addition of the first concentration of insulin. Statistical analyses for differences in insulin-induced vasoreactivity were performed using one-way ANOVA with Bonferroni post hoc tests, or Mann-Whitney U test where appropriate. Differences in protein phosphorylation, as found by Western blot, were determined using the Wilcoxon signed rank test. Phosphorylation was expressed as the fold increase over the unstimulated controls, assigning a value of 1 to the control. Differences with P values <0.05 were considered statistically significant.

PVAT was located around the resistance arteries, as well as A1 and A2 arterioles, in the muscle microcirculation (Fig. 1A). The weight of PVAT was higher in the db/db mice (1,365 ± 451 μg) than in lean C57BL/6 mice (33.3 ± 6.6 μg; P < 0.01) (Fig. 2B).

Baseline vessel tone was not affected in resistance arteries by either C57BL/6 PVAT or db/db PVAT, nor was acetylcholine-induced vasodilation (Fig. 3A and B). Resistance arteries from C57BL/6 mice in the absence of PVAT did not show vasodilation or vasoconstriction in response to insulin (2 nmol/L; 0 ± 9%). After incubation with PVAT from C57BL/6 mice, insulin induced significant vasodilation (2 nmol/L; 52 ± 21%; P < 0.02 compared with baseline and P < 0.05 vs. resistance arteries without PVAT) (Fig. 3C).

The vasodilation to insulin observed in the presence of C57BL/6 PVAT was completely abrogated when PVAT from db/db mice was used (2 nmol/L; −5 ± 3%; P < 0.02 compared with C57BL/6 PVAT) (Fig. 3D). To further study the insulin-induced vasodilator pathway, which is mediated through Akt and eNOS phosphorylation, phosphorylation of Akt was studied in the presence of C57BL/6 or db/db PVAT. The phosphorylation of Akt was significantly increased after insulin stimulation in resistance arteries of C57BL/6 mice in the presence of C57BL/6 as well as db/db PVAT. In the presence of C57BL/6 PVAT, the increase in Akt phosphorylation was significantly higher (P < 0.01 vs. db/db PVAT) (Fig. 3E).

To elucidate the mechanisms involved in the interaction of intramuscular PVAT with insulin-induced vasoreactivity, we examined the role of adiponectin in this interaction. Adiponectin was secreted more abundantly by PVAT from C57BL/6 mice (502 ± 115 pg/mL medium) than by PVAT from db/db mice (178 ± 33 pg/mL; P = 0.02) (Fig. 4A). When adiponectin was scavenged with a soluble adiponectin receptor fragment, Acrp-30 [N-20]-P, so that the db/db PVAT situation was mimicked, insulin-induced vasodilation was abolished (2 nmol/L; −19 ± 10%; P < 0.01 vs. C57BL/6 resistance arteries in the presence of PVAT). This indicates the importance of adiponectin from PVAT in insulin-induced vasodilation (Fig. 4B). Adiponectin, in the absence of PVAT, stimulated insulin-induced vasodilation in C57BL/6 resistance arteries (2 nmol/L; 21 ± 4%; P = 0.02 vs. C57BL/6 resistance arteries without PVAT) (Fig. 4C). Moreover, we incubated globular adiponectin in the presence of db/db PVAT (n = 7) to study whether supplementation of adiponectin could restore insulin-induced vasoreactivity. Adiponectin, in the presence of db/db PVAT, stimulated insulin-induced vasodilation 17 ± 4% (P < 0.01 compared with db/db PVAT alone) (Fig. 4D).

As AMPK is an important effector of adiponectin in vascular endothelium, we next examined the role of AMPK in the interaction of PVAT with insulin-induced vasoreactivity in isolated resistance arteries. Resistance arteries from C57BL/6 incubated with both PVAT from lean mice and compound C showed (9 ± 4% at 2 nmol/L of insulin) vasodilation, a nonsignificant reduction (P = 0.28 vs. C57BL/6 resistance arteries with C57BL/6 PVAT) (Fig. 5A). However, there was no significant vasodilation compared with resistance arteries without PVAT.

AMPK phosphorylation was significantly increased in vessels exposed to both C57BL/6 PVAT (P = 0.006) and db/db PVAT (P = 0.028), compared with C57BL/6 resistance arteries without PVAT. There was, however, no difference in AMPK phosphorylation between vessels exposed to C57BL/6 PVAT and db/db PVAT (Fig. 5B).

To further examine the role of AMPK in insulin-induced vasodilation induced by PVAT, resistance arteries from AMPKα2^−/− mice were exposed to PVAT from AMPKα2^+/+ mice. Resistance arteries from AMPKα2^+/+ littermates served as controls. In the latter group, unlike the parallel studies using arteries and PVAT from C57BL/6 mice, insulin did not induce a significant vasodilation (2 nmol/L; −3 ± 6%). There was no difference in insulin-induced vasoreactivity between AMPKα2^+/+ and AMPKα2^−/− resistance arteries; both exhibited a neutral response (Fig. 5C). However, AMPKα2^−/− resistance arteries with AMPKα2^+/+ PVAT exhibited vasoconstriction (2 nmol/L; −26 ± 6%; P = 0.010) (Fig. 5D), pointing to strain differences with C57BL/6 mice. This indicates the importance of AMPK in the interaction between PVAT and insulin-induced vasodilation. Tone development and acetylcholine-induced vasodilation were not different between the two groups (Fig. 5E and F).

Because Akt phosphorylation in the presence of db/db PVAT was lower (Fig. 3E) and AMPKα phosphorylation is not different between C57BL/6 and db/db PVAT (Fig. 5B), adipokines other than adiponectin may also play a role in the change of insulin-induced vasodilation with db/db PVAT. To study whether PVAT from the muscle microcirculation produces additional vasoactive adipokines, incubation of a C57BL/6 resistance artery with C57BL/6 PVAT as well as db/db PVAT was performed (Fig. 6A). Dual incubation led to insulin-induced vasodilation (26 ± 9%; P < 0.05 vs. db/db PVAT). The vasodilation with dual PVAT incubation was smaller, however, than in the presence of C57BL/6 PVAT alone (P = NS) (Fig. 6A), suggesting interaction between the effects of C57BL/6 and db/db PVAT.

The vasodilator effect that was absent with db/db PVAT was restored by inhibiting inflammation using the JNK-inhibitor L-JNKi (2 nmol/L insulin; 48 ± 17%; P < 0.001 compared with db/db PVAT) and overlapping with the line of C57BL/6 PVAT, indicating that inflammation in either db/db PVAT or the acceptor resistance artery plays a role (Fig. 6B). Therefore, JNK opposes adiponectin-dependent vasodilator properties of PVAT in db/db mice. When adiponectin was blocked during incubation with db/db PVAT and L-JNKi, the insulin-induced vasodilation was completely inhibited (2 nmol/L; 1 ± 6%; P < 0.01 vs. L-JNKi with db/db PVAT) (Fig. 6C). Treatment with L-JNKi did not restore insulin-stimulated Akt phosphorylation in the gracilis resistance arteries exposed to db/db PVAT (P = NS vs. the non-pretreated resistance arteries) (Fig. 6D).

The principal findings of this study are that 1) PVAT uncovers insulin-induced vasodilation in muscle resistance arteries by secreting adiponectin and activating AMPKα2 in the vessel wall, 2) in obese db/db mice, PVAT dramatically increases in muscle and loses its ability to induce insulin-mediated vasodilation, 3) this impaired interaction of PVAT with muscle resistance arteries is caused by a decrease in adiponectin release by PVAT and impairment of insulin-stimulated Akt activation, and 4) the blunted insulin-induced vasodilation with db/db PVAT can be restored by inhibition of JNK.

We show a direct regulatory effect of PVAT on insulin-induced vasoreactivity in muscle, shifting the balance of insulin’s vasoactive effects toward insulin-induced vasodilation. Our finding that PVAT controls vasoreactivity is in line with previous observations in the aorta and adipose tissue microvessels that PVAT controls vascular tone by antagonizing vasoconstriction in response to norepinephrine, endothelin-1, and angiotensin II (11,28,29). We did not observe an effect of PVAT on basal smooth muscle tone, indicating that the interaction of PVAT with insulin-induced vasoregulation is specific. As direct effects of PVAT on smooth muscle tone have been observed in adipose tissue and the aorta, the effects of PVAT on vasoregulation vary among different vascular beds.

We have shown that release of adiponectin and activation of AMPK by PVAT are critical to its control of insulin-induced vasoregulation in muscle. Previous studies have indeed shown that adiponectin can activate AMPK in endothelial cells (30), and previous data from our group have shown that adiponectin uncovers insulin-induced vasodilation in muscle resistance arteries through AMPK (31). Our data are in good agreement with the earlier data of Greenstein et al. (11) who showed that PVAT effects on norepinephrine-induced vascular tone in subcutaneous adipose tissue are mediated by adiponectin. Although PVAT releases adiponectin, and a soluble fragment of the adiponectin receptor 1 (AdipoR1) inhibited PVAT control of insulin-induced vasoregulation, a role of other adipokines in PVAT-endothelial communication cannot be completely excluded. The effect of PVAT on insulin-induced vasodilation was partially mimicked by recombinant globular adiponectin, either in the absence or presence of db/db PVAT as well as by dual incubation of db/db and C57BL/6 PVAT. This suggests that other adipokines than adiponectin play a supporting role or stimulate the insulin-induced vasoconstrictor pathway. Indeed, obesity seems to promote secretion of adipokines by PVAT that inhibit insulin-induced vasodilation and others that enhance insulin-induced vasoconstriction, because the vasodilator effect of C57BL/6 PVAT was partially mimicked by adiponectin, and db/db PVAT decreases the vasodilator effect of C57BL/6 PVAT. Moreover, in AMPKα2^+/+, no insulin-induced vasodilation was observed with PVAT; however, knocking out AMPKα2 leads to insulin-induced vasoconstriction. This is further supported by equal amounts of AMPKα phosphorylation by C57BL/6 and db/db PVAT.

Although AMPK is an important effector of PVAT in vascular endothelium, we have observed AMPK-independent effects of PVAT. In AMPKα2-deficient resistance arteries, PVAT uncovers insulin-induced vasoconstriction. Furthermore, PVAT from db/db mice enhances AMPKα phosphorylation in muscle resistance arteries but inhibits insulin-stimulated phosphorylation of Akt. These results suggest that AMPKα2 activation is a necessary, but not sufficient, step to uncover insulin-induced vasodilation. Moreover, db/db PVAT must regulate insulin’s vasoactive properties through a second mechanism, counteracting the insulin-induced vasodilation-stimulating effects of PVAT. The difference in net effects of PVAT between C57BL/6 and AMPKα2^+/+ mice strongly suggests that genetic factors determine adipokine secretion by PVAT.

Indeed, inhibition of inflammation with L-JNKi fully restored the vasodilator capacity of db/db PVAT (Fig. 6B), whereas blocking adiponectin with the soluble adiponectin receptor fragment completely abrogated this effect (Fig. 6C). Inflammation therefore plays an important role in the defective PVAT properties of db/db mice. Interestingly, L-JNKi did not restore insulin-induced Akt phosphorylation during exposure to db/db PVAT (Fig. 6D). This finding may be explained by increased adiponectin secretion by db/db PVAT during JNK inhibition, or inhibition of insulin’s vasoconstrictor effects, which are independent from Akt (27). Indeed, inflammation of PVAT leads to a loss of its anticontractile properties (11,32). Nevertheless, adipokines other than adiponectin likely contribute to vasodilator effects of PVAT, as supplementation of adiponectin to db/db PVAT could not completely restore insulin-induced vasodilation (Fig. 4D). Recently, a new class of adipokines, C1q/tumor necrosis factor (TNF)–related proteins, has been shown to activate AMPK in endothelial cells through the adiponectin receptor AdipoR1 (33). These C1q/TNF-related proteins could explain the AMPK phosphorylation and the vasodilator response to anti-inflammatory treatment with L-JNKi in the absence of increased Akt phosphorylation. The PVAT-derived adipokine that activates JNK in the muscle microcirculation remains to be identified in future studies. We have assessed secretion of the JNK agonists TNF-α and free fatty acids by muscle-derived PVAT but found only minor secretion of these adipokines (R.I.M., unpublished data). This finding does not exclude an autocrine role of TNF-α in regulating PVAT function.

Our results yield two novel insights into the pathogenesis of insulin resistance. First, we have shown that local depots of PVAT, which surround the resistance arteries in the muscle microcirculation, are capable of controlling insulin-dependent microvascular function. In humans, local accumulation of adipose tissue in the extremities is related to flow-mediated vasodilation and insulin sensitivity (21,34). These observations and the present dataset support our previous proposition (9) that adipokine-induced microvascular dysfunction in obesity is a paracrine, rather than an endocrine, process (9,35). In contrast to our initial hypothesis (9), we have found that altered adipokine secretion by intramuscular PVAT, rather than the formation of previously absent PVAT in muscle, impairs insulin-induced vasoregulation in obesity. Also, anti-inflammatory treatment restores insulin-induced vasodilation, at least partly, through adiponectin.

Second, our results provide a novel mechanism by which AMPK regulates insulin sensitivity. AMPK is well known as a “metabolic gauge” that controls both basal and insulin-stimulated glucose uptake (36). Paradoxically, AMPKα2 controls muscle insulin sensitivity in vivo but not in isolated muscles (26). Our observation that AMPKα2^−/− mice show impaired insulin-induced vasoregulation in the presence of PVAT, in combination with recent evidence that insulin-induced vasoregulation controls insulin sensitivity (2), suggests that AMPKα2 controls muscle insulin sensitivity by interacting with insulin-induced microvascular recruitment. Further support for this notion comes from our recent observation that AMPK activation by AICA-riboside enhances muscle microvascular blood volume (15).

In conclusion, in this study, we show that PVAT unveils vasodilator responses to insulin through the secretion of adiponectin, which is antagonized in db/db PVAT by the inflammatory kinase JNK. Adiponectin activates AMPKα2, which subsequently shifts the balance between vasodilation and vasoconstriction toward vasodilation. In obesity, the quantity of PVAT increases and adiponectin secretion diminishes. Due to the diminished adiponectin secretion and inflammatory burden, insulin no longer induces vasodilation. However, the insulin-stimulated vasodilation can be restored by inhibition of JNK.

This research is sponsored by the Netherlands Heart Foundation (Grant 2009B098), the Netherlands Organization for Scientific Research (Grant 916.76.179), and the Danish Medical Research Council.

No potential conflicts of interests relevant to this article were reported.

R.I.M. wrote the manuscript and researched and analyzed data. W.B. and C.-L.A.F.A. researched data and edited the manuscript. P.S. and E.A.R. supervised and edited the manuscript. J.S.Y., Y.M.S., V.W.M.v.H., and E.H.S. supervised, contributed to discussions, and edited the manuscript. B.V. created the AMPKα2^+/+ and AMPKα2^−/− mice and edited the manuscript. E.C.E. researched data, supervised, contributed to discussions, 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.

Parts of this study were presented in poster form at the 70th Scientific Sessions of the American Diabetes Association, Orlando, Florida, 25–29 June 2010.