All animal experiments were performed in accordance with the approval of the Institutional Animal Care and Use Committee of Columbia University. Homozygous male RAGE null (RAGE^-/-) mice (backcrossed >10 generations into C57BL/6) were generated as described previously 9 . Male wild-type (WT) C57BL/6 mice were obtained (Jackson Laboratories). At age 6 weeks, half of the WT and half of the RAGE^-/- mice were treated with streptozotocin (STZ; Sigma). Two months later, all mice underwent femoral artery (FA) ligation.

Mice were treated with five consecutive daily doses of STZ dissolved in citrate buffer (55 mg/kg, pH 4.5) via the intraperitoneal route. One week after the first dose, glucose levels were assessed by glucometer. The criteria of two consecutive glucose levels >250 mg/dL was used to indicate diabetes. If glucose levels were <250 mg/dL, then the mice received two additional doses of STZ (55 mg/kg).

Under isoflurane anesthesia, the hair on the abdominal wall and pelvis and both upper legs was shaved and the skin prepped with iodine and alcohol. An incision was made on the upper thigh of both the left and right legs of each mouse. The inguinal ligament and the upper half of the femoral artery were exposed. On the left side, the vascular bundle was isolated from below the inguinal ligament proximally to just above the bifurcation into the superficial and deep femoral arteries distally. The femoral artery was dissected free, and two ligatures were placed around it with 8/0 non-absorbable sutures and tied. Both skin incisions were closed with sterile 5/0 nylon suture.

Aliquots of 5 μg of HYNIC-RGD were incubated with 0.5 ml of tricine solution (70 mg/ml in distilled water) and approximately 0.5 ml of ^99mTcO₄ ^- solution (50 mCi = 1,850 MBq) and 20 μl of tin(II) solution (10 mg of SnCl₂·2H₂O in 10 ml of nitrogen-purged 0.1 N HCl for 20 min) at room temperature. To test the specificity of the HYNIC-RGD, cyclo [Arg-Ala-Asp-D-Phe-Lys (HYNIC)] (Peptides International, Louisville, KY, USA) was similarly radiolabeled and used as control peptide. Radiochemical purity was >94% by Tec-control chromatography (Biodex, Shirley, NY, USA).

Under isoflurane anesthesia (1.5% isoflurane at a flow rate of 0.5% L/min oxygen per mouse), a cutdown was made over the jugular vein and a specially designed vascular catheter was placed (Braintree Scientific, Braintree, MA, USA). Mice in each of four groups were injected with ^99mTc-HYNIC-RGD and imaged 3 or 7 days after FA ligation: WT without diabetes (n = 14), WT with diabetes (n = 14), RAGE^-/- without diabetes (n = 16), RAGE^-/- with diabetes (n = 14), and five WT without diabetes were injected with control peptide. All mice were injected through the jugular vein catheter with 1 ± 0.2 mCi of ^99mTc-HYNIC-RGD in 0.05 to 0.1 ml (corresponding to 1 μg of peptide) or control peptide. Blood pool clearance was measured in five mice injected with ^99mTc-HYNIC-RGD. By 60 to 75 min after injection, residual blood pool activity was below 10% of peak. Whole-body planar gamma images in the anteroposterior view were acquired on a high-resolution high-sensitivity dedicated small animal camera with parallel hole collimator (provided by Jefferson Lab, Newport News, VA, USA). The camera is based on a 5-in. Hamamatsu position sensitive photomultiplier type R3292 with an active field of view of about 95 mm diameter. The scintillator sensor is 1.6-mm-step 6-mm-thick pixelated NaI(Tl) scintillator array. The photo peak was set at 140 keV with a 15% energy window.

At completion of the imaging session, each animal was euthanized by an intraperitoneal injection of pentobarbital (100 mg/kg). The anterior tibialis muscles were dissected, weighed, and counted in a gamma counter (Wallac Wizard 1470, PerkinElmer, Waltham, MA, USA) for determination of the percent injected dose of radiotracer per gram (%ID/g) tissue. The radiotracer activity in the samples was corrected for background, decay time, and tissue weight. Limb counting was performed in 28 animals. The remaining animals were used for immunohistochemistry.

For immunohistochemical analyses, tibialis anterior muscles were harvested and fixed in 10% formalin for 48 h. Specimens were embedded in paraffin, and tissue slices (5 μm in thickness) were prepared. Serial sections were stained with hematoxylin and eosin (H&E) for morphology. Immunostaining was performed for capillary sprouting using biotinylated Griffonia Bandeiraea Simplicifolia Isolectin I (Vector Laboratories, Burlingame, CA, USA) for β₃ (1:50; Abcam, Cambridge, MA, USA.) and for α_ν (1:100; Millipore, Temecula, CA, USA). Serial sections were also stained for RAGE using a monoclonal antibody against RAGE (50 μg/ml). Secondary stains were performed using avidin-biotin visualization systems (Vectastain ABC Kit, Vector Laboratories). All brown staining capillaries were counted for each of 5 to 6 sections for both the left and right anterior tibialis muscles for each experiment and then were averaged. The average number of capillaries for the left anterior tibialis muscle was divided by the average number for the right (control) anterior tibialis muscle. RAGE staining was quantified as area staining positive for the brown chromagen per 100× field.

Dual immunofluorescent studies were undertaken to determine the cell types expressing α_ν integrin. Serial sections (5 μm in thickness) obtained from the ischemic hindlimb were deparaffinized in xylene and incubated with α_ν (rat anti-mouse integrin α_ν, 1:100) and co-stained with endothelial cell marker (FVIII, 1:200) or macrophage marker (Mac-3, 1:50). Secondary fluorescent antibodies were Texas Red anti-rabbit and FITC anti-mouse. The images were captured and processed using confocal fluorescence microscope (Nikon, Tokyo, Japan) and SPOT imaging software (Diagnostic Instruments, Inc., Sterling Heights, MI, USA).

Radiotracer counts in the ischemic hindlimb were determined from the in vivo scans by using the region of interest (ROI) method in the mini gamma camera image using public domain Image J software (NIH, Bethesda, MD, USA). A region was drawn around the focal uptake, and the mean counts were determined. Radioactivity in the contralateral control limb was similarly determined using a comparable ROI (same anatomic location and the number of pixels). The counts from each of these areas were used to determine the ischemic to non-ischemic ratios.

Continuous variables were expressed as mean ± standard deviation. Normality was assessed using the Shapiro-Wilk test. Comparisons between two groups were made using the Student's t test. Correlation was assessed using the Pearson product-moment correlation coefficient. All statistical tests were two-tailed, with P < 0.05 denoting significance. All statistical analyses were performed using STATA 10.1 (StataCorp, College Station, TX, USA).

Mean uptake ratios of counts between the left and right limbs were not different between days 3 and 7 for any of the four groups: WT non-diabetic (P = 0.52), WT diabetic (P = 0.39), RAGE^-/- non-diabetic (P = 0.41), and RAGE^-/- diabetic (P = 0.39). Therefore, days 3 and 7 data were combined as the early time period.

Representative scans from the four groups and the control peptide are shown in Figure 1. All scans in the non-diabetic WT group were positive visually, while three of the left limbs in the diabetic group were negative, one was equivocal, and one weakly positive. Scans of the WT mice injected with control peptide showed no tracer uptake in either limb.

Data from scans and ex vivo well counting for both hindlimbs are shown in Figure 2. For the WT non-diabetic group, the mean scan count ratio for L/R hindlimbs was 1.91 ± 0.34 (range, 1.46 to 2.79), and for the WT diabetic group, it was 1.38 ± 0.26 (range, 1.05 to 1.74) (P < 0.001) (Figure 2A). The mean value for the RAGE^-/- non-diabetic group was 2.02 ± 0.29 (range, 1.54 to 2.62) not statistically significantly different from the WT non-diabetic group. The mean value for the RAGE^-/- diabetic group was 1.75 ± 0.22 (range, 1.53 to 2.35) which was significantly lower than the RAGE^-/- non-diabetic group (P < 0.001) and was significantly higher than the WT diabetic group (P < 0.001).

Figure 2B shows values as %ID/g for the four groups for the left and right hindlimbs. The counts in the left (ischemic) hindlimbs showed the same pattern of differences among the four groups as shown for the scan ratios except for values for the WT diabetic and RAGE^-/- diabetic (1.42 and 1.43). However, the ratios of L/R hindlimb %ID/g for RAGE^-/- diabetic was higher than for WT diabetic (2.85 ± 0.40 vs. 2.13 ± 0.67, P = 0.03) (Figure 2C). This difference is due to lower mean %ID/g in the right limb for RAGE^-/- diabetic group. For the remaining limb ratios for %ID/g values, WT non-diabetic was significantly higher than WT diabetic (3.04 ± 0.95 vs. 2.13 ± 0.67, P = 0.03) and RAGE^-/- non-diabetic was higher than RAGE^-/- diabetic (4.08 ± 1.00 vs. 2.85 ± 0.40, P = 0.02). All of these significance levels for intergroup differences were lower than for the scan data (Figure 2A) possibly due to the technical challenge to cleanly dissect the anterior tibialis muscles in the mouse. This limitation may have weakened the correlation for the plot of the ratios of L/R limbs against %ID/g (R = 0.059), although the correlation is highly significant (P = 0.001) (Figure 3).

Examples of tissue sections stained for H&E, α_ν, β₃, and lectin are shown in Figure 4A. Quantitative lectin staining for capillaries from anterior tibialis muscle sections (n = 20 per group) for both the left (ischemic) and right (sham operated) hindlimbs of WT non-diabetic, WT diabetic, RAGE^-/- non-diabetic, and RAGE^-/- diabetic are shown in Figure 4B. The average capillary staining for the WT non-diabetic left limbs was significantly lower than the RAGE^-/- non-diabetic left limbs (P = 0.05) and significantly higher than the WT diabetic left limbs (P < 0.001). The capillary staining for the WT diabetic left limbs was borderline significantly lower than for the RAGE^-/- diabetic left limbs (P = 0.06). These histological results support the scan findings. Co-staining of sections for both endothelial cells and macrophages showed colocalization with α_ν (Figure 5).

RAGE staining also supported the scan findings. There was positive staining for RAGE in the ischemic sections of hindlimbs from both diabetic and non-diabetic mice, and no staining in the contralateral control limbs (Figure 6). The RAGE^-/- mice both non-diabetic and diabetic showed no RAGE staining.