Increased MAPK Activation and Impaired Insulin Signaling in Subcutaneous Microvascular Endothelial Cells in Type 2 Diabetes: The Role of Endothelin-1

Human needle biopsies of the subcutaneous adipose tissue were obtained from the abdominal region of nondiabetic (n = 12) and type 2 diabetic (n = 12) subjects under local anesthesia (Xylocain 10 mg/ml) (Astra Zeneca AB, Södertälje, Sweden), as previously described (12). All subjects were recruited via an advertisement in a local newspaper, and their clinical characteristics are shown in Table A1 (available in the online appendix at http://diabetes.diabetesjournals.org/content/early/2009/06/23/db08-0961/suppl/DC1).

About 3 g of tissue was incubated in sterile Medium199 (Invitrogen, Paisley, U.K.) containing 25 mmol/l HEPES, 4% BSA, 5.5 mmol/l glucose, and 0.8 mg/ml collagenase in a shaking water bath at 37°C. After 30–50 min, the digest was filtered through a sterile nylon mesh (pore size 250 μm) and washed four times with fresh medium as described before (20). Two fractions were obtained: the upper fraction containing the floating isolated adipose cells and the lower fraction containing the stromavascular cells.

The lower fraction was collected into 50-ml Falcon tubes, and, after centrifugation (1,300 rpm for 5 min), the pellet was resuspended in endothelial cell growth medium Bulletkit (EGM-2-MV) consisting of endothelial cell basal medium-2 (EBM-2) supplemented with EGM-2-MV SingleQuots containing hEGF, hydrocortisone, FBS, vascular endothelial growth factor, R3-IGF-1, ascorbic acid, and gentamicin sulfate/amphotericin-B (GA-1000) (Clonetics, BioWhittaker, Verviers, Belgium). FGFb was not added to the medium.

The cells obtained were seeded into a 75 cm² culture flask and allowed to grow at 37°C and 5% CO₂ in the same media. Before the cells reached confluence, after approximately 1 week, the heterogeneous pool of cells was exposed to a CD31-positive selection as described below. The isolated adipocytes in the floating fraction were washed four times in fresh Medium 199, resuspended at 2% cytocrit, and used as described below. All subjects gave their written informed consent, and the study was approved by the Ethics Committee of the University of Gothenburg.

The human MVEC selection was performed with the Dynabeads magnetic CD31 MicroBeads cell sorting system (Invitrogen).

The heterogeneous cell population, containing the adipose tissue–derived stromavascular cells, was washed twice with PBS and incubated on a rocking platform (25 revolution/min) during 15 min with 4 ml of PBS/0.2% BSA containing 100 μl of Dynabeads (2.8 × 10⁵ beads/ml) coated with CD31 antibody (PECAM-1). The beads rapidly target and partially coat the endothelial cells expressing the CD31 receptor.

After the incubation, the culture was washed twice with PBS and trypsinized. The cells were collected into a tube and centrifuged for 3 min at 1,300 rpm. The pellet was resuspended in PBS/0.2% BSA, collected into a 2 ml microcentrifuge tube, and placed in a magnet (Dynal MPC-S) (Invitrogen), following the manufacturer's recommended protocol for washings and final extraction. The CD31-negative cells (with no beads attached) were removed during the successive washings.

The positive selected endothelial cells were resuspended in endothelial cell growth medium Bulletkit (EGM-MV2) and maintained at 37°C and 5% CO₂. The cells obtained with this procedure were 99% endothelial cells with typical cobblestone morphology. The endothelial cells were characterized by indirect immunofluorescense and acetylated LDL uptake. The cells were used before passage seven.

For immunofluorescence, the cells were incubated in eight-well plastic chamber slides (Nunc, A/S, Roskilde, DK) for 48 h. The cells were washed twice with PBS and fixed for 10 min in methanol. After three rinses in PBS, the cells were incubated with the human primary antibody against von Willebrand Factor (vWF) (DAKO A082) (1:50 dilution in PBS/3% BSA) for 1 h. The culture was washed three times with PBS and incubated with the fluorescein isothiocyanate–labeled secondary antibody (DAKO F0205) (1:50 dilution in PBS/3% BSA) for 1 h and washed again.

The cells were incubated in eight-well plastic chamber slides for 48 h. Acetylated LDL labeled with 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindo-carbocyanine perchlorate (Dil-Ac-LDL) (Biomedical Technologies) was added to the culture medium at a final concentration of 10 μg/ml, and the cells were incubated at 37°C for 4 h. The cells were fixed for 30 min in 3% formaldehyde solution, mounted, and observed by fluorescence microscopy.

Rat-tail collagen type I, 10 mg/ml (BD Biosciences, Stockholm, Sweden), was neutralized by the addition of 1N NaOH, H₂O, and PBS following the manufacturers recommendations. The mix was poured into 12-well culture plates (200 μl/well) and allowed to solidify for 1 h in a humidified atmosphere of 5% CO₂ at 37°C. After solidification, the gels were preincubated with growth medium 30 min before the cells were seeded. The tube formation was monitored by phase-contrast microscopy.

The cultured cells were thoroughly washed with PBS, deprived of serum and EGM-2-MV SingleQuots additions for 4 h, before adding insulin for 20 min. ET-1, BQ123, BQ788 (Sigma-Aldrich, Stockholm, Sweden), or PD98059 (Calbiochem-Merck Chemicals, Nottingham, U.K.) were added to the culture media for 24 h at the indicated concentrations. When used in combination with other factors, PD98059 was added 1 h before.

The cells were washed with PBS and then lysed in lysis buffer on ice (21). The lysate was vortexed and rocked for 2 h at 4°C and the proteins quantified by the bicinchonic acid method (Pierce, Rockford, IL).

The isolated human adipocytes, as described above, were distributed into plastic vials (12–15% cell suspension) and incubated with 100 nmol/l insulin for 15 min. The cells were rapidly separated by centrifugation through silicone oil and suspended in lysis buffer. The lysate was vortexed and rocked for 2 h at 4°C as previously reported (21). The proteins were quantified by the bicinchonic acid method. RNA was extracted according to Chirgwin et al. (22).

The cultured cells were deprived of serum and EGM-2 MV SingleQuots additions and insulin was added, when specified, for 24 h. Total cellular RNA was extracted with RNeasy (Quiagen Gmbh, Hilden, Germany). The mRNA expression was analyzed with the ABI PRISM 7900 sequence detection system (TaqMan, Applied Biosystems, Foster City, CA) using18s rRNA as endogenous control subjects (23). The gene expression predesigned primer and probe set #Hs00167166_ml was used for eNOS and #Hs00174961_m1 for ET-1 gene analysis (Applied Biosystems).

Human umbilical vein endothelial cells (HUVECs) (ATCC-CRL-1730) were grown in Kaighn's modified F12K media (American Type Culture Collection [ATCC]) from LGC Promochem (Boras, Sweden). The media was supplemented with 0.1 mg/ml heparin and 0.03 mg/ml endothelial cell growth supplement, both from Sigma-Aldrich, and 10% FBS (GIBCO, Invitrogen).

Crude cell extracts were boiled in Laemmli buffer containing 150 mmol/l dithiothreitol for 5 min. The samples were analyzed by SDS-PAGE (7.5 or 10%). Proteins were transferred from the gel to nitrocellulose sheets, blocked in 5% fat-free milk, and probed with the different primary antibodies according to the manufacturer's recommendations. IRS-1 and MAPK antibodies were from Upstate Biotechnology (Millipore, Solna, Sweden), p-PKB, PKB, and p-MAPK were from Cell Signaling (New England Biolabs, Hertfordshire, U.K.), insulin receptor and eNOS were from Transduction (BD Biosciences, Erembodegem, Belgium), and antiphosphotyrosine (pY99) was from Santa Cruz Biotechnology (SDS, Falkenberg, Sweden). The proteins were detected by enhanced chemiluminescence using horseradish peroxidase–labeled secondary antibodies (Amersham Biosciences, Buckinghamshire, U.K.), and the intensity of the bands was quantitated with a laser densitometer (Molecular Dynamics, Sunnyvale, CA).

Cell lysates (60 μg) were immunoprecipitated with anti-insulin receptor antibody according to the recommendation of the manufacturers. Subsequently, the immune complexes were precipitated with Protein A/G Plus-Agarose (Santa Cruz Biotechnology) for 90 min at 4°C. The immunoprecipitates were washed three times with lysis buffer. The samples were analyzed by SDS-PAGE (7.5%) and immunoblotted as described above.

Cellular cGMP was measured by the cGMP Enzyme-immunoassay Biotrak System (Amersham Biosciences) following the manufacturer's recommendations.

Data analyses are presented as means ± SE. Student's t test was used to test for statistical significance.

At least 99% of the primary MVEC generated by the CD31 extraction showed a positive staining for vWF and LDL uptake. However, there was a certain variation between the subjects regarding the number of nonendothelial cells (CD31 negative) that were recovered with the positive cells. This variation was not related to the subject phenotype (control or type 2 subjects) nor to the defined experimental conditions used. The nonendothelial cells were easy to distinguish from the MVEC in culture and were easily removed by performing a new CD31 extraction in a subsequent passage.

MVEC proliferation was slow and sensitive to seeding densities. A suitable density of cells was around 3 × 10³ cells/cm². The adherent cultured human MVEC monolayer assumed the typical cobblestone morphology, which was maintained throughout the 10–13 passages studied (Fig. 1 A). Later passages were associated with reduced growth and distorted cell borders. The cells never formed thick aggregates and maintained their contact inhibition.

The MVEC were positive for the endothelial-specific vWF immunostaining (Fig. 1,B). In addition, the cells incorporated acetylated LDL, another marker characterizing endothelial cells (Fig. 1 C).

The ability to form a capillary/tube-like network is a specialized function of vascular endothelial cells. We characterized the subcutaneous MVEC by their ability to form capillary-like structures. Figure 1 D shows the morphogenic response of MVEC in collagen and EGM-MV2 media. Morphologic changes and reorganization of the cells were apparent as early as 12 h after initiation of the culture.

We investigated the responsiveness to insulin of the adipose tissue–derived MVEC from healthy and type 2 diabetes individuals. For comparison, we performed parallel experiments with the HUVEC line, a well-documented endothelial cell line.

Expression of the insulin receptor was low in MVEC compared with HUVEC, but there was a clear insulin-stimulated tyrosine phosphorylation of the receptor in both type 2 and control subjects in MVEC and in HUVECs (Fig. 2).

Because the presence of insulin receptor in MVEC is a controversial issue, we wanted to further validate the presence of insulin receptors by immunoprecipitating the insulin receptors and immunoblotting for both phosphotyrosine and receptor protein. As shown (Fig. 2, right panel), insulin receptors are indeed present in MVEC albeit at a low level.

Both main docking proteins for insulin, IRS-1 and IRS-2, were expressed as well as the downstream signaling proteins PKB/Akt and the MAPKs extracellular signal–related kinase (ERK)1/2. IRS-1 was clearly the predominant protein while IRS-2 was only expressed at a low level (data not shown).

Similar to our previous observations in human adipocytes (12), IRS-1 protein expression was markedly decreased in MVEC from type 2 diabetic subjects (Fig. 3).

This reduction was also reflected by the impaired activation and serine phosphorylation of the downstream protein, protein kinase B (PKB)/Akt, in MVEC from type 2 diabetic subjects, whereas PKB/Akt protein expression was unchanged (Fig. 4, upper panel). This observation was confirmed in four separate immunoblots quantified by scanning and shown in Fig. 4 (bottom panel). A pronounced insulin-stimulated activation of PKB was also seen in the HUVEC cells (Fig. 4, upper panel).

The MAPKs ERK1/2 had an increased phosphorylation in the basal state in MVEC from type 2 diabetic subjects (Fig. 5, upper panel). Insulin further increased the MAPK phosphorylation in control subject MVECs (P = 0.057) while there was no effect of insulin (P = 0.18) in the type 2 diabetic subject cells, probably masked by the high basal MAPK/ERK activation. This observation was confirmed in five separate experiments measuring the p-MAPK/MAPK volume ratios (Fig. 5, bottom panel), supporting that the major change was because of an increased phosphorylation of the MAPK in the basal state.

Surprisingly, there was also a very low protein expression of eNOS in the MVEC and, as expected, in the isolated adipose cells from the same individuals compared with HUVEC (Fig. 6, top). The gene expression of eNOS was also almost undetectable in the MVEC, from both control and type 2 diabetic subjects, while there was a clear expression of both the gene and protein in HUVECs (Fig. 6, bottom panel). Insulin had no effect on the expression of eNOS mRNA in MVEC while the levels were significantly increased in HUVEC. To further pursue the ability of insulin to activate eNOS, we incubated the cells with or without the hormone for 1, 3, 5, 15, or 20 min and measured phosphorylation of Ser¹¹⁷⁷ and Thr⁴⁹⁵. Because no activation was found, we investigated the effect of insulin on cyclic GMP formation. However, insulin did not increase cyclic GMP levels in human cultured MVEC from subcutaneous adipose tissue (data not shown). Thus, we conclude that eNOS is not an important target for insulin in these cells, at least under the culture conditions used.

As shown in Fig. 7, ET-1 mRNA levels were significantly higher in cells from type 2 diabetes individuals, both when compared with cells from the control group as well as with HUVECs, while there was no significant difference between the control group and HUVECs. Adding insulin (100 nmol/l) for 24 h did not increase ET-1 expression in any of the three groups studied (Fig. 7).

To further explore its role, we added ET-1 (100 nmol/l) to the culture medium and found that it increased ERK1/2 phosphorylation in control subject MVEC (Fig. 8, upper panel).

The addition of the ET_A (BQ123) and ET_B (BQ788) receptor antagonists (20 μmol/l) as well as the MEK-1 inhibitor PD98059 (50 μmol/l) prevented this effect of ET-1 (Fig. 8, upper panel). It should be noted that both ET-1 receptor antagonists were required because both A and B receptors were present in MVEC (data not shown).

To examine if the increased MAPK activation in the nonstimulated state in MVEC from type 2 diabetes individuals was because of the increased endothelin, we incubated the cells with both ET_A and ET_B receptor antagonists. As shown in Fig. 8 (middle and bottom panels), the increased basal MAPK activity in MVEC from type 2 diabetic subjects was normalized by the ET-1 receptor antagonists. This was also seen with the MEK-1 inhibitor PD98059. Taken together, our results show that ET-1 is increased in MVEC from type 2 diabetes individuals and that this activation can account for the increased MAPK activation seen in the basal state.

This study describes an improved method for the primary culture of an apparently pure population of MVEC obtained from a subcutaneous needle biopsy (around 3 g), using CD31 extraction. In general, the purification steps had to be performed at least twice to obtain a pure cell population. The cells displayed the typical cobblestone morphology, positive immunostaining for the vWF, uptake of acetylated LDL, as well as capacity to form a capillary/tube-like network on a collagen gel. The usefulness of our procedure is illustrated by the fact that the MVEC can be expanded and used to characterize gene expression and hormone signaling. This opens up new possibilities to study molecular mechanisms in MVEC in relation to the clinical phenotype including endothelial dysfunction.

We focused on characterizing insulin signaling and action in the MVEC because these cells are considered to be important target cells for insulin. We have previously shown that IRS-1 protein expression, like GLUT4, is reduced in the adipose cells in insulin resistance and is, indeed, a biomarker of the IRS (12). Moreover, a group of subjects exhibiting low adipocyte IRS-1 expression displayed early signs of atherosclerosis, including an increased intima-media thickness (13) and an increased vascular stiffness as measured with pulse tonometry (24) compared with subjects with “normal” adipocyte IRS-1 expression. Animal studies have also shown that in the vasculature of obese Zucker rats, insulin resistance affects the PI3Kinase pathway but not other pathways of insulin signaling including the MAPK pathway (25). This “selective” insulin resistance would provide insights into the mechanisms for the increased shedding of cellular adhesion molecules in insulin-resistant subjects (26), the enhanced vasoconstrictory effect of insulin (27), and the pathophysiology of microvascular complications in diabetes (28).

The MVEC showed a low, but similar, expression of the insulin receptor and its tyrosine phosphorylated form in type 2 diabetic patients and control subjects, whereas IRS-1 protein expression and phosphorylation by insulin were markedly attenuated in type 2 diabetic patients. Consequently, serine phosphorylation of PKB/Akt was impaired in agreement with an impaired signaling through the PI3Kinase pathway. However, we were unable to observe any difference in eNOS expression and phosphorylation in response to insulin between the groups because of a very low expression of eNOS in the MVEC. This was unexpected but low eNOS expression in MVEC has been reported before by some (29) but not all studies with MVEC (11). However, our findings were corroborated by measuring cGMP production that also was extremely low.

An intriguing finding was the increased nonstimulated (basal) as well as insulin-stimulated phosphorylation of ERK1/2 in cells from diabetic subjects. This observation fits with the known imbalance of insulin action in the microvasculature often referred to as pathway-selective insulin resistance in obesity and diabetes (25,26,30).

The increased phosphorylation of ERK1/2 in diabetes may be an important reason for the decreased IRS-1 protein expression since MAPK phosphorylate IRS-1 on serine residues leading to an increased degradation of the protein (31). Interestingly, basal ERK1/2 phosphorylation is also increased in adipose cells in type 2 diabetes (32), and IRS-1 protein is also decreased (33). The finding of an increased ERK1/2 phsophorylation together with decreased IRS-1 in MVEC cultured for up to seven passages in vitro suggests a genetic cause and/or that the MVEC release factors that activate the cells.

DNA methylation and posttranslational histone modifications are epigenetic mechanisms that are known to influence mammalian gene expression. Type 2 diabetes shows a progressive development with increasing age and environmental sensitivity. These characteristics make this disease suitable for epigenetic changes, but it is only recently that we have started to understand the role of these epigenetic mechanisms. Therefore, it is possible that epigenetic mechanisms, because of high glucose concentrations or other factors (34,35), may lead to a long-term activation of genes in the type 2 diabetes MVEC. However, further studies are needed to address this point.

Interestingly, ET-1 expression was significantly increased in the MVEC from diabetic subjects when compared with both nondiabetic subjects and HUVECs. In fact, there was no difference between the nondiabetic and HUVECs, further corroborating the increase in the diabetic cells. Increased circulating ET-1 levels have also been reported in type 2 diabetes (36,37), and this finding is consistent with the increased ERK1/2 activation seen since these kinases have been associated with an increased endothelin expression (30).

In our study, the increased ET-1 expression was not altered by the presence of insulin and, thus, is more likely a consequence of other activators. The nature of these upstream factors is a current focus of research.

Addition of ET-1 to the culture media increased MAPK phosphorylation in control subject MVEC. This effect was prevented by PD98059 showing that ERK1/2 is a target for ET-1 in MVEC. Importantly, the increased nonstimulated (basal) MAPK activation in MVEC from type 2 diabetic subjects was completely antagonized by the addition of the ET-1 receptor antagonists BQ123 and BQ788. The use of both antagonists was required because MVEC were found to express both the ET_A and ET_B receptors. These results support the concept that the increased ET-1 expression plays a major role for the increased basal MAPK activation seen in MVEC of type 2 diabetic subjects.

ET-1 was recently shown to impair insulin signaling in adipose cells by decreasing the tyrosine phosphorylation and expression of IRS-1 (37,38). Thus, it is also possible that the reduced IRS-1 protein in adipose cells from type 2 diabetic subjects is because of the increased ET-1 and MAPK activation as discussed above. If this is indeed the case, it would imply that the endothelial cells can directly cross talk with the adipose cells and modulate their insulin sensitivity and action. It should also be added that only ERK1/2 were increased in the MVEC since p38 phosphorylation was not increased (data not shown).

There has been much discussion recently about the putative presence of insulin and/or IGF-I receptors, or chimeric receptors, in endothelial cells (39,40). We have also found that the MVEC express IGF-1 receptors (data not shown). Thus, we cannot exclude that signaling downstream the IGF-1 receptor is more important than that downstream the insulin receptor. In fact, previous studies have suggested that IGF-1 is more important than insulin for the generation of nitric oxide (NO) via the cognizant receptor or via hybrid receptors in endothelial cells (39,40). Nevertheless, our data show that the insulin receptor was expressed and activated in the MVEC, and a discussion about the relative quantity of each receptor and/or the presence of hybrid receptors is beyond the scope of this study.

It could be argued that subcutaneous MVEC are not representative cells for studies of NO signaling. This may to some extent be true because endothelial cells derived from the microvasculature can in many ways differ from macrovascular endothelial cells (e.g., Bovine Aorta and Human Umbilical Vein). Indeed, tissue-specific expression patterns in different MVEC suggest that they are distinct cell types involved in the local regulation of their respective organs (41). Our observation that ERK1/2 is activated in type 2 diabetes MVEC, like in the adipose cells (32), highlights the potential of these cells to be involved in both adipose tissue expansion through angiogenesis (42) as well as adipose cell insulin sensitivity and action.

In conclusion, we have developed a procedure whereby the subcutaneous MVEC can be isolated and cultured from small tissue specimens in human subjects. Using this technique, we find that MVEC from type 2 diabetic patients have a reduced IRS-1 protein expression and an impaired insulin-stimulated PKB/Akt activation, whereas activation of ERK1/2 was increased in the basal state. This activation of ERK1/2 can be accounted for by the increased ET-1 levels in MVEC from type 2 diabetic subjects. These observations support the concept of “selective” insulin resistance as an underlying factor for a dysregulated microvascular endothelium in type 2 diabetes. Furthermore, our results are consistent with the possibility that the endothelial cells can cross talk with the parenchymal cells and, thereby, be important modulators of insulin sensitivity and action.

This study was supported by grants from the Swedish Medical Research Council (K2005-72X-15389-01A), (K2008-55X-15358-04-3), and (K2007-54X-03506-36-3), the Novo Nordisk Foundation, the Swedish Diabetes Association, the Torsten and Ragnar Söderbergs Foundations, the Inga-Britt and Arne Lundberg Foundation, the Foundation for Strategic Research and the European Community as part of the project EUGENE2, which is supported by the European Commission (Contract LSH MCT 2004-512013).

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