Microvascular Alterations in Diabetic Mice Correlate With Level of Hyperglycemia

Transgenic mice expressing diphteria toxin A chain (DTA) regulated by the uncoupling-protein gene-promotor (UCP promoter) selectively in brown adipose tissue (11) were used. UCP1/DTA mice develop severe brown adipose tissue deficiency leading to decreased energy expenditure. They are characterized by hyperphagia beginning at 8 weeks, which is followed by obesity, hyperglycemia, and hyperinsulinemia (11). An average increase of 54% (male mice)/75% (female mice) in body weight, 30% in blood glucose levels, and an 85- to 125-fold increase in blood insulin levels at the age of 22 weeks (14,15) have been documented in UCP1/DTA mice. To mimic early-stage diabetes, mice at the age of 20 weeks with established obesity and hyperglycemia were used in this study. Nontransgenic littermates were used as controls. Mice were caged individually at 24°C and 50% humidity and fed with standard rodent chow (Chow no. S5714S040; Ssniff, Soest, Germany) and water ad libitum. All animal procedures were performed according to the German animal welfare committee.

Body weight, body temperature, and whole blood glucose levels were measured 2 days before dorsal skin-fold chamber implantation, on the day of the surgical procedures, and 6 and 9 days after implantation. A glucose tolerance test (GTT) was performed on the day of chamber implantation by injecting a 10% glucose solution (1 mg glucose/g body wt) intraperitoneally after an overnight fast (12 h). Blood was collected with a heparinized capillary, and whole blood glucose was measured 0, 15, 30, 60, 90, and 120 min after glucose injection via the glucose oxidase method (One Touch, Life Scan, Milpitas, CA). Serum levels of tumor necrosis factor (TNF)-α were determined using an immunoassay (Mouse TNF-a, Quantikine M; R&D Systems, Minneapolis, MN). Blood was collected by heart puncture with a heparinized syringe.

To noninvasively and continuously measure microcirculatory parameters, we implanted a dorsal skin-fold chamber as previously described (16,17) (Fig. 1A). Before surgical procedures, mice were anesthetized (7.5 mg ketamine hydrochlorid and 2.5 mg xylazine/100 g body wt) and their skin was shaved and depilated. Dorsal skin-fold chambers (weight 3.2 g; Machine shop, Boston, MA) were implanted exposing subcutaneous tissue and striated skin muscle as described previously (16). All surgical procedures were performed at aseptic conditions while maintaining body temperature at physiological levels. The mice were allowed to recover for 6 days.

To obtain microcirculatory parameters, randomly selected areas (four to five locations) were investigated using a fluorescence microscope (Axioplan; Zeiss, Oberkochen, Germany). The microscope is equipped with filter sets for fluorescein isothiocyanate (FITC) and tetramethylrhodamine (TRITC), a 20× objective lens (LD Achroplan 20×/0.40; Zeiss, Germany), an intensified charge coupled device (CCD) video camera (C2400–97; Hamamatsu Photonics, Germany), a Photomultiplier Tube (R4632; Hamamatsu Photonics, Herrschingam Ammersee, Germany), and a computer (Apple Power MacIntosh).

A total of 100 μl FITC labeled Dextran (MW 2,000,000, 50 mg/ml 0.9% saline; Sigma, St. Louis, MO) was injected through the tail vein to visualize blood vessels. During each observation period, FITC images were recorded for 30 s and analyzed off-line (Fig. 1B). The erythrocyte velocity (V_RBC) was measured opto-electronically using a two-slit method and an image processing system (Exbem 3.0; Pixlock e.k., Münster, Germany; National Institutes of Health Image 1.62) as described elsewhere (18). The mean blood flow rates (BFRs) of individual vessels were calculated using diameter (D) and V_mean.

BFR = π/4 × V_mean × D², where V_mean = V_RBC/α (α = 1.3, for blood vessels <10 μm; linear extrapolation 1.3 <α <1.6 for blood vessels between 10 and 15 μm; and α = 1.6 for blood vessels >15 μm) (19). The functional vascular density (VD), defined as the total length of functional vessels per unit area (measured in centimeters per centimeters squared) and the vessel diameter were analyzed using an image processing system (National Institutes of Health Image 1.62), as described elsewhere (16,18).

Effective microvascular permeability (P) was measured as described previously (17,20). After the injection of TRITC-labeled BSA (TRITC-BSA, MW 67,000; Molecular Probes, Eugene, OR) (10 mg/ml, 0.1 ml/25 g body wt), the fluorescence intensity was measured intermittently for 25 min and recorded digitally (PowerLab/200 ADInstruments; Ply, Castle Hill, Australia) (Fig. 1D). The value of P was calculated as P = (1 − HT) V/S [1/(I₀ − I_b) × dI/dt + 1/K], where I is the average fluorescence intensity of the whole image, I₀ is the value of I immediately after the filling of all vessels by TRITC-BSA, I_b is the background fluorescence intensity, and HT is the amount of hematocrit. The average HT of vessels is assumed to be equal to 19% (21). V and S are the total volume and surface area, respectively, of vessels within the tissue volume covered by the surface image, respectively. The time constant of BSA plasma clearance (K) is 9.1 × 10³s (22).

Leukocyte-endothelial interactions (LEIs) in vessels were monitored as described previously (23). Mice were injected intravenously with a bolus (20 μl) of 0.1% rhodamine-6G in 0.9% saline, and leukocytes were visualized via an intensified CCD camera and recorded noncompressed digitally on a computer (Apple Power MacIntosh) (Fig. 1C). The numbers of rolling (Nr) and adhering (Na) leukocytes were counted for 30 s along a 100-μm segment of a vessel. Rolling was defined as a slow movement along the vessel wall (<50% of V_RBC), and adhering was defined as a stable position at the vessel wall for 30 s. The total flux of cells for 30 s was measured (Nt). The total leukocyte flux was normalized by the cross-sectional area of the individual vessel. Leukocyte flux (measured in cells per millimeters squared) = 10^{6 ×} Nt/(π × [D/2]² × 30s). Rolling count, the ratio of rolling leukocytes (Nr) to total cell flux (Nt) was calculated as follows: rolling count (%) = 100 × Nr/Nt. The mean leukocyte adhesion at the vessel wall was calculated by normalizing the vessel surface area. Adhesion density (measured in cells per millimeters squared) = 10⁶ × Na/(π × D × 100 μm). Shear rate of individual blood vessels was calculated according to the following equation: shear rate (s⁻¹) = 8 × V_mean/D (23).

Results are presented as the mean ± SE of the mean. Values of the transgenic versus control group were compared with the Mann-Whitney test using StatView (Abacus, Berkeley, CA). To test correlations between parameters, t tests for Spearman’s rank correlation were performed and groups were compared with the Mann-Whitney test. P values <5% were considered to be significant.

Significantly lower body temperature of UCP1/DTA mice (∼1°C; data not shown) corroborate observations of reduced body temperature from previous studies (15). Overnight fast caused a decrease in body temperature of the controls, whereas no extra decrease in body temperature of UCP1/DTA mice was observed.

Compared with controls, body weight of UCP1/DTA mice increased significantly (52.5%), as described previously (14) (data not shown). The difference in body weight between UCP1/DTA mice and controls decreased from day 0 (52.5%) to day 9 (32.3%) due to a more pronounced loss in body weight in the transgenic group (−18.2%) compared with the control group (–5.7%). The initial loss in body weight (5%) was probably due to the overnight fast, whereas the further loss might be related to the chamber implantation. A 15% loss in body weight has been reported after dorsal skin-fold chamber implantation (16). Analysis revealed a significant correlation of weight loss (day 0 –day 9) with the body weight at day 0 independently of group (data not shown).

We measured blood glucose at day −2, 0, 6, and 9 (Fig. 2A). UCP1/DTA mice showed significant elevated glucose levels throughout the whole observation period. Compared with the control group (107 ± 3 vs. 109 ± 3 mg/dl), an increase to 206 ± 13 mg/dl (+92.1%, day 6) and 195 ± 13 mg/dl (77.9%, day 9) was measured. Additionally, a glucose tolerance test (GTT) was performed (Fig. 2B). The blood glucose after an overnight fast was 151.4 mg/dl (mean ± 9.1) in UCP1/DTA mice, indicating the diabetic state, whereas controls showed physiological fasting glucose (81.4 ± 4.2 mg/dl). GTT was significantly elevated in UCP1/DTA mice compared with controls.

The microvascular permeability (P) was significantly higher (22 × 10⁻⁷ ± 2.3 × 10⁻⁷cm/s) in UCP1/DTA mice than in controls (6.2 × 10⁻⁷ ± 4.2 × 10⁻⁷ cm/s) (Fig. 2C).

Analysis of functional VD revealed a significant decrease in UCP1/DTA mice (day 6, 264 ± 11 cm/cm² and day 9, 276 ± 13 cm/cm²) compared with control mice (day 6, 326 ± 9cm/cm² and day 9, 323 ± 9 cm/cm²) (Fig. 3A). VD showed a significant negative correlation to the blood glucose level (Fig. 4B) and body weight (data not shown). A reduction in VD of 27% was found in the group with blood glucose levels 250–299 mg/dl (246 ± 11 cm/cm²) compared with the group with blood glucose levels 50–99 mg/dl (335 ± 12 cm/cm²). Vessel diameter (D) in UCP1/DTA mice (day 6, 10 ± 0.8 μm and day 9, 10 ± 1.1 μm) was significantly increased compared with controls (day 6, 5.3 ± 0.2 μm and day 9, 5.2 ± 0.2 μm) (Fig. 3B). Analysis of the vessel diameter distribution illustrated that the diameter peak in control mice (∼3 μm) was significantly blunted and shifted to higher values (∼2 μm) in UCP1/DTA mice (Fig. 4A). Therefore, the increased mean diameter in UCP1/DTA mice was mainly due to a reduction in the number of vessels with diameters ∼3 μm. The mean VD was characterized by a positive correlation to blood glucose levels with a plateau phase (9.1 μm) for blood glucose values >160 mg/dl (Fig. 5A). A positive correlation between D and body weight was found (data not shown).

The UCP1/DTA mice showed a significant decrease in blood flow velocity (V_mean) (day 6, 529 ± 44 μm/s and day 9, 535 ± 39 μm/s) compared with controls (day 6, 859 ± 46 μm/s and day 9, 818 ± 42 μm/s) (Fig. 5B). V_mean correlated negatively with blood glucose levels (Fig. 5A) and body weight (data not shown) and showed the maximum velocity (840 ± 31 μm/s) at low blood glucose levels (80–119 mg/dl) and a plateau phase (590.7 ± 38.6 μm/s) at high blood glucose levels with values >160 mg/dl. The BFR was found to be unchanged, indicating that the perfusion of vessels is similar in UCP1/DTA mice and controls (Fig. 5C).

Mean diameter of measured venules in control and transgenic mice was 27.1 ± 0.82 and 25.7 ± 1.3 μm, respectively. The rolling count of the transgenic group was significantly higher with 42.4 ± 2.7% on day 6 and 45.2 ± 2.8% on day 9 compared with the control group with 22.0 ± 3.8% on day 6 and 31.1 ± 3.8% on day 9 (Fig. 6A). Cell adhesion density was at least threefold higher in the transgenic (day 6, 126 ± 17 cells/mm² and day 9, 107 ± 13 cells/mm²) versus control group (day 6, 39 ± 21 cells/mm² and day 9, 13 ± 6 cells/mm²) (Fig. 6B). LEI, leukocyte adhesion density, and the rolling fraction correlated significantly with glucose levels (Fig. 6C) and body weight (data not shown). The cell adhesion density of leukocytes correlated positively with the blood glucose level and mirrored the picture of the leukocyte rolling fraction. The leukocyte cell flux per cross-sectional area was significantly lower in transgenic mice (1,830 ± 176 cells/mm²s) compared with controls at day 6 (2,958 ± 375 cells/mm²s) (Fig. 7A). A significant reduction in shear rate to 43 and 53% could be found in UCP1/DTA mice (day 6, 185 ± 24^s−1 and day 9, 210 ± 23^s−1) compared with control mice (day 6, 432 ± 37^s−1; day 9, 397 ± 40^s−1) due to the decreased V_mean (Fig. 7B). Shear rate correlated significantly with glucose levels and body weight (data not shown) with a strong decrease between 80 and 160 mg/dl blood glucose and a plateau at values >160 mg/dl blood glucose (Fig. 7C).

The UCP1/DTA-group showed a significant increase in mean serum TNF-α concentration (9.28 ± 1.60 pg/ml) compared with the control group (3.11 ± 0.87 pg/ml) (Fig. 8A). Serum TNF-α levels were found to be significantly correlated to blood glucose levels as well as body weight (data not shown) and had an approximately linear (R² = 0.984) increase in dependence of blood glucose (Fig. 8B). An increase from 0.69 ± 0.31 to 13.34 ± 3.41 pg/ml at glucose levels of 55–99 and 200–279 mg/dl, respectively, could be measured.

Microcirculatory properties govern the efficacy of substrate delivery, but they also mediate functional and morphological adaptations to local requirements and metabolic conditions. Our results demonstrate for the first time functional and morphological alterations of the microcirculation in early stage diabetic UCP1/DTA mice. We observed that these parameters correlated with the metabolic state represented by blood glucose levels and body weight. Because of the correlations found between the metabolic state and the microcirculatory parameters independently of the group, it is not likely that the microcirculatory dysfunctions are directly due to the animal model.

Increased permeability, especially to albumin, has been shown to be a very early endothelial dysfunction in diabetes in the retina (26), kidney (27), and rat mesentery (5). We showed that increased permeability throughout the subcutaneous tissue, indicating vascular leakage, is an early sensitive parameter of endothelial dysfunction and is observable in early-stage diabetic disease.

VD was decreased in diabetic mice, whereas the vessel diameter was increased as reported earlier (10,28). The combination of an unchanged BFR in blood vessels and a lower VD leads to decreased tissue perfusion in subcutaneous tissue and may therefore lead to an impaired supply of oxygen and nutrients. Analysis of diameter distribution revealed that the decrease in VD and increase in vessel diameter was mainly due to a reduced number of small vessels with diameters between 3 and 5 μm. The failure of small vessels in UCP1/DTA mice might explain impaired wound healing found in diabetic disease. The reduction in VD and increase of vessel diameter correlated with the metabolic situation of the mice represented by the blood glucose levels and body weight. These correlations support the finding that impairment of metabolic control directly leads to progressive morphologic and functional alterations of the microcirculation in early-stage diabetes. The molecular mechanisms of the impaired vessel formation need to be further elucidated. However, there is some evidence that the disturbance in skin vessel stability might be due to altered expression of angiopoetin-1 and -2 and their receptor Tie-2 (29), which are important determinants in vessel maturation, stabilization, and remodelling (30).

The V_mean showed a decrease at higher blood glucose levels reaching a plateau at blood glucose levels >160 mg/dl. This is consistent with earlier observations measured in rat mesentery (5) and in conjunctival microcirculation of diabetic patients (28). Surprisingly, the BFR, an important determinant for tissue perfusion, was not changed. The unchanged BFR is likely due to the combination of a lower V_mean and the increased diameter.

There is growing evidence that LEI plays an important role in diabetic microangiopathy. Increased LEI have been reported to trigger capillary nonperfusion, blood-retinal barrier breakdown and neovascularization in the eye (24,25), increase ischemia-reperfusion damage (5), and lead to inactivation of nitric oxide (NO) through release of oxigen-derived radicals and therefore to endothelial dysfunction (31). LEIs are mediated by forces of adhesion and blood velocity. Adhesion is determined by the expression of adhesion molecules on the endothelial surface (intracellular adhesion molecule [ICAM]-1, P-selectin, platelet endothelial cell adhesion molecule [PECAM]-1, and vascular cell adhesion molecule [VCAM]-1) (32–35) and the surface of leukocytes (e.g., L-selectin) (33,36). Leukocytes interact with the endothelium by first rolling along the vessel wall, then activating β₂-Integrins expression (32,33), and finally stable binding with possible extravasation thereafter. Counteracting this process is the blood flow (shear rate), which depends on blood flow velocity and vessel diameter. We found significantly increased LEI, leukocyte rolling fraction, and leukocyte density at the vessel wall, which were positively correlated to blood glucose levels and body weight, indicating an inflammatory reaction in the microcirculation induced by impaired metabolic control. These results agree with previous in vivo and in vitro observations of LEI in streptozotocin-induced diabetes in rat mesentery (5), rats given a continuous intravenous glucose infusion (37), and human umbilical vein endothelial cell (HUVEC) in a flow chamber (38). A significantly decreased shear rate indicates that the increased LEI is likely in part due to the reduction of hemodynamic forces in the microcirculation, as described for the mesentery of streptozotocin-induced diabetes of rats (5). But despite the reduction of hemodynamic forces, analysis of correlation of LEI and shear rate with blood glucose levels revealed that shear rate showed a plateau phase at blood glucose levels >160 mg/dl, whereas LEI, rolling count, and leukocyte density, were approximately linearly (R² = 0.96) correlated to blood glucose. Therefore, it is not likely that the increased leukocyte adhesion at higher blood glucose values is only due to the lowered shear rate, as discussed previously (5). This conclusion is additionally supported by in vivo observations showing that local application of glucose to nondiabetic rat mesentery led to an increase of LEI without altering shear rate over 12 h (34). Therefore, other influencing factors, such as glucose-induced expression of cell adhesion molecules, might be an explanation for the increased LEI. TNF-α is known to upregulate adhesion molecules (E-selectin, VCAM-1, and ICAM-1) in a dose-dependent manner in human intestinal microvascular endothelial cells (35) and dermal endothelial cells (39). TNF-α mRNA levels in adipose tissue were reported to be increased in diabetic UCP1/DTA mice (14,40) similar to elevated plasma TNF-α found in obese humans (41). Furthermore, it has been shown that locally administered TNF-α increases LEI in subcutaneous tissue in vivo, without altering hemodynamic parameters (23). Our results show elevated TNF-α levels, which linearly correlate to glycemic levels (R² = 0.98), and can explain this further increase in LEI beyond the plateau phase of shear rate at glucose levels >160 mg/dl.

We have shown for the first time that functional and morpholocical microvascular alterations in subcutaneous tissue are present even at an early stage of diabetes. These microvascular parameters correlated with the metabolic situation represented by blood glucose levels and body weight. From our results, we conclude that even a short time of metabolic dysregulation causes not only functional but also morphological microvascular alterations. Therefore, a precise glycemic and metabolic control in patients can reduce diabetic angiopathies. To which extent patients with pathologic glucose tolerance and increased postprandial glucose levels suffer from microvascular alterations and the impact of other influencing factors (e.g., hyperinsulinemia and advanced glycation end products) on vascular deteriorations has to be further elucidated. To register early microvascular alterations, microvascular monitoring of patients with diabetes might be necessary to recognize early changes in microvascular properties (42). Our results support the growing evidence of the inflammatory component of diabetic disease and indicate that anti-inflammatory strategies might be beneficial to prevent microvascular deterioration. Further investigations must evaluate the benefit of such therapeutic strategies.

This work was supported by a Werner Otto Stiftung research grant to N.H-A. and P.A. P.A. and A.H. were members of the DFG Graduate Kolleg [GRK336]. N.H.-A. and C.S. were members of the DFG Graduate Kolleg [GRK 476].

We thank M. Ammelt for help with the image processing analysis and Tim Padera for fruitful discussions and help with the manuscript.