Insulin Suppresses Plasma Concentration of Vascular Endothelial Growth Factor and Matrix Metalloproteinase-9

Ten obese nondiabetic subjects (age range 29–64 years; mean 48.3 ± 10.9 years), all with BMI >37 kg/m², (mean BMI 42.6 ± 9.1 kg/m²) were included in this study. All patients had a fasting venous plasma glucose <100 mg/dl. All had impaired glucose tolerance with a 2-h postchallenge glucose ranging between 145 and 180 mg/dl. None of the obese subjects were on vitamin E or C or any other antioxidant therapy. There were eight female and two male subjects. Insulin was infused into subjects (2–2.5 IU/h), along with 5% dextrose (100 ml/h) and 10 mmol KCl for 4 h. The rate of insulin infusion was titrated such that glucose concentrations were maintained as close to the basal levels as possible. Blood glucose concentrations were measured every 15 min from superpricks (Hemocue) to ensure that they did not deviate significantly. The rate of insulin infusion was altered accordingly as necessary. Minimal alterations were necessary at the low rates of infusion of both glucose and insulin. Blood samples for detailed analysis were collected at baseline, 2, 4, and 6 h and were collected in Na-EDTA as the anti-coagulant. Subjects returned after 2–3 weeks on two occasions and were infused with 5% dextrose (100 ml/h), 10 mmol KCl, and normal physiological saline (100 ml/h and 10 mmol KCl/h) for control data. The Institutional Review Board of State University of New York at Buffalo based at Millard Fillmore Hospital approved the study. Written informed consent was obtained from all subjects.

Insulin was measured from fasting plasma samples using an enzyme-linked immunosorbent assay kit from Diagnostic Systems Laboratories (Webster, TX). Glucose was measured in whole blood by a Hemocue glucose analyzer (Hemocue, Mission Viejo, CA).

Plasma VEGF and MMP-9 were assayed with enzyme-linked immunosorbent assay kits from R&D Systems (Minneapolis, MN). The minimum detectable dose of VEGF assay was 5 pg/ml, with an intra-assay variation of 4.5% and inter-assay variation of 7%. The minimum detectable dose of MMP-9 assay was 0.156 ng/ml, with an intra-assay variation of 2% and interassay variation of 8%. The variability in fasting MMP-9 concentrations is in the range of 10%.

Statistical analysis was performed using SigmaStat software (Jandel Scientific, San Rafael, CA). All data on VEGF and MMP-9 were normalized to a baseline of 100% in view of the interindividual variability and are expressed accordingly as the percentage of the basal level. One-way ANOVA for repeated measures was used to compare all of the indexes measured in this study. The results are expressed as mean ± SE.

Plasma glucose concentration remained steady during insulin infusion. Glucose levels were 73 ± 2 mg/dl at baseline, 76 ± 5 mg/dl at 2 h, and 73 ± 2 mg/dl at 4 h. Plasma glucose concentration also remained steady following dextrose or saline infusion. The concentrations of glucose were similar to those after glucose-insulin infusion. Plasma insulin concentrations increased from a basal level of 13 ± 2 to 28 ± 3 μU/ml at 2 h, and 24 ± 4 μU/ml at 4 h after insulin infusion (Table 1). Insulin levels decreased after the cessation of insulin infusion and returned to the basal level (11 ± 3 μU/ml) at 6 h. Dextrose infusion caused a slight increase in glucose levels at 2 h. There was no significant increase in insulin in the glucose-infused group.

VEGF concentrations decreased significantly following insulin infusion (P < 0.001) (Fig. 1) and continued to be inhibited at 6 h, even after the cessation of insulin infusion. Basal plasma VEGF concentration was 307.2 ± 163.8 pg/ml. Levels decreased to 73.5 ± 20.9% of the basal level at 2 h, 67.1 ± 23.2% at 4 h, and 82.0 ± 18.5% at 6 h. Dextrose infusion alone did not change plasma VEGF concentrations significantly. Plasma MMP-9 concentrations decreased from 375.1 ± 196.3 ng/ml (100%) to 83 ± 7.9% of the basal level at 2 h, 82 ± 7.4% of the basal level at 4 h, and reverted to 99 ± 9.0% at 6 h (P < 0.05). Plasma MMP-9 concentrations increased significantly following dextrose infusion alone (P < 0.05), as shown in Fig. 2. Thus, the fall in MMP-9 concentration with insulin was even more impressive (by 40%) when compared with glucose-infused control subjects. Saline infusion caused no change in glucose, insulin, VEGF, or MMP-9 concentrations.

Our data demonstrate clearly that insulin acutely causes a lowering of plasma VEGF and MMP-9 concentrations. The magnitude of this fall was 35% for VEGF and 40% for MMP-9 when compared with glucose-infused control subjects. The fall was also significant when compared with the baseline of saline-infused control subjects. These data are consistent with our previous observations (17,18) that insulin inhibits the expression of NF-κB and early growth response-1, the two proinflammatory transcription factors that modulate the inflammation-inducing genes such as ICAM-1, MCP-1, C-reactive protein, p47phox, plasminogen activator inhibitor 1, and tissue factor. This inhibitory effect lasted for 4 h, the duration of insulin infusion. Because NF-κB is a modulator of VEGF expression (19), it is possible that NF-κB suppression by insulin causes the suppression of VEGF concentrations. It is of interest that VEGF also induces NF-κB activity (20).

This observation is relevant to the pathogenesis of retinopathy in that the absence of insulin and of adequate insulin action, as in type 2 diabetes, may result in an increase in VEGF expression/secretion. This will promote neovascularization at the microvasculature level and from adventitia into the media and intima to supply blood to the growing atherosclerotic plaque (8). Hyperglycemia also increases the expression and secretion of VEGF at the cellular level (21) and may thus contribute to the endothelial abnormalities leading to local inflammatory changes, which trigger leukocytic adhesion to endothelium. Leukocytic adhesion to retinal capillaries mediated by the adhesion molecule, ICAM-1 (22,23), can cause plugging of the capillaries, capillary nonperfusion, retinal ischemia, and the changes that trigger increased VEGF expression locally. We have previously demonstrated that insulin suppresses ICAM-1 expression by endothelial cells and therefore lowers the concentration of ICAM-1 in plasma. Our current observations indicate that insulin lowers VEGF concentration rapidly, suggesting that it may have a direct effect on the expression of VEGF at the cellular level. MMP-9 implicated in the lysis of the matrix proteins, especially collagen, was also suppressed by insulin. MMP action on the extracellular matrix allows the endothelial tubes to be formed from the proliferation of endothelial cells under the action of VEGF. In addition, MMPs are responsible for the rupture of the atherosclerotic plaque through the lysis of the fibrous cap of the plaque. The increase in MMP-9 concentration by glucose is intriguing but is consistent with a proinflammatory effect of glucose. Saline-infused control subjects showed no changes in MMP-9.

Although VEGF is usually secreted under hypoxic conditions, inflammatory stimuli may also cause it to be expressed and secreted. Thus, hypoxia-inducing factor, the transcription factor that regulates VEGF gene expression, may be induced through NF-κB activation by interleukin-1 and tumor necrosis factor-α (24).

There is one observation showing an increase in the expression of VEGF retinal pigment endothelial cells following incubation with insulin in vitro. However, these increases were observed with pharmacological concentrations of insulin (5 nmol/l ≅ 800 μU/ml) (25). The same authors showed that a similarly high concentration of insulin injected into the eye also induced an increased expression of VEGF in the retina.

It is also noteworthy that there is a lack of correlation between plasma concentrations of VEGF and the prevalence or severity of retinopathy. This is probably due to the fact that the local concentrations of VEGF in any tissue determine the action of VEGF and thus the neovascularization in that tissue. Therefore, local ischemia in the retina determines the expression of VEGF in the retina; this then triggers MMPs and neovascularization. Although insulin may have a general tonic inhibitory effect on the expression of VEGF, ischemia may turn out to be the major determinant locally. On the other hand, the lack of adequate insulin bioavailability and action in diabetes may trigger VEGF release and neovascularization. Indeed, intensive insulin treatment in patients with diabetes over prolonged periods has been shown to markedly reduce the incidence of background and proliferative retinopathy in many studies: the Diabetes Control and Complications Trial (26), the Kumamoto Study (27), and the Oslo study (28). Although a reduction in glucose concentrations may have largely induced this improved outcome, an increase in the bioavailablity of insulin at appropriate times during intensive insulin therapy may have also contributed to this effect. It should also be appreciated that there may be factors other than VEGF that determine endothelial proliferation and neovascularization, e.g., fibroblast growth factor and angiopoietin. This requires further investigation.

In conclusion, insulin acutely suppresses the concentration of VEGF and MMP-9 in plasma. This has implications for the pathogenesis of proliferative retinopathy of diabetes and for atherogenesis and plaque rupture in patients with diabetes and insulin resistance.