All animal studies were performed with the approval of the Institutional Animal Care and Use Committee of Columbia University. Male C57BL/6 mice and ApoE^−/− mice with the genetic background of C57BL/6 mice were purchased from The Jackson Laboratories (Bar Harbor, ME, USA).

At 6 weeks of age, mice were made diabetic via 5 consecutive daily doses of streptozotocin (Sigma, 50 mg/kg in citrate buffer, pH 4.5; St. Louis, MO, USA). Blood glucose levels were monitored weekly via tail vein sampling using blood glucose monitor (FreeStyle Lite, Abbott, North Chicago, IL, USA). The mean blood glucose level for the duration for diabetic mice was 309 ± 39 mg/dL. Animals were studied at 26 to 28 weeks of age. On arrival, all of the apoE−/−mice had similar weights, but by the time of study, the diabetic mice were smaller than the non-diabetic mice with average weights of 23.1 ± 1.2 and 30.7 ± 1.1 g (P = 0.0001).

scV/Tc is based on scVEGF protein which combines two 3 to 112 fragments of human VEGF fused head to tail and is expressed with an N-terminal 15-aa cysteine-containing tag (Cys-tag) for site-specific conjugation 19 . scVEGF-PEG-DOTA and its inactivated analog that lacks VEGFR-binding ability were prepared as described 19 . Briefly, scVEGF was site-specifically derivatized on C4 in Cys-tag with radionuclide chelator 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA) via 3.4 kDa PEGylated linker. The inactivated VEGF-PEG-DOTA was prepared by biotinylation of 10 to 12 lysine ε amino groups in scVEGF moiety. For ^99mTc labeling, lyophilized 0.15 mg scVEGF-PEG-DOTA (approximately 5 nmol) was reconstituted with 75 μl of 0.1 M NaOAc buffer (pH 5.5) and added to 30 to 40 mCi (1,110 to 1,480 MBq) of ^99mTc. The mixture was purged with N₂ for 2 min followed by addition of 50 μg of tin-tricine in 450 μl of deionized water, briefly purged with N₂ and incubated for 20 min at 55°C. Free ^99mTc was separated on PD-10 column (Pharmacia) equilibrated with PBS/0.1% BSA. Fractions (0.5 ml) containing the tracer in the void volume were pooled. The radiochemical purity of scV/Tc was 97% ± 1.6% with the specific activity of 109 ± 11 μCi/μg. The inactivated scVEGF-PEG-DOTA was similarly labeled with ^99mTc and the resulting tracer was named in-scV/Tc.

Blood pool clearance was measured in 3 C57BL/6 mice and 3 ApoE^−/−diabetic mice. Blood samples were taken by tail vein knicking at 5, 10, 30, 60, 90, 200, 250, and 350 min. Biodistribution of scV/Tc in non-target organs was analyzed in diabetic and non-diabetic mice after imaging. At necropsy, organs were removed, weighed, and counted in the well counter (Wallac Wizard 1470, PerkinElmer, Waltham, MA, USA).

Non-diabetic ApoE^−/− mice (n = 8) and diabetic ApoE^−/− mice (n = 10) mice were anesthetized with isofluorane (4% to induce, 1% to maintain) and injected with scV/Tc 0.38 ± 0.08 mCi (14.06 ± 2.92 MBq) via femoral vein catheter. To assess nonspecific (non VEGFR-mediated) tracer uptake, a separate group of non-diabetic ApoE^−/− mice (n = 5) were injected with in-scV/Tc (tracer control). To assess uptake of tracer in non-diseased mice, 2 C57BL/6 mice were injected with scV/Tc (disease control). Three to 4 h after tracer injection (blood pool clearance), all mice were re-anesthetized and injected with 150 to 200 μl (160 mg/mL) of eXIA contrast agent (Binitio Biomedical, Ottawa, Canada) and immediately underwent SPECT/CT imaging on nanoSPECT/CT system (Bioscan, Washington DC, USA).

CT images were acquired with an integrated CT scanner using an X-ray tube at 45 kVp and an exposure time of 1,000 ms per view. The helical SPECT scans were acquired using dual-headed detectors each outfitted with collimators with nine pinholes. Each pinhole had a diameter of 1.4 mm with each collimator providing a transaxial field-of-view (FOV) of 30 mm and an axial FOV of 16 mm, extendable through helical scanning to 270 mm. SPECT data were acquired with the following parameters: step and shoot rotation, 30° step in 360° rotation using 24 projections, 60 s per projection, 256 × 256 frame size with 1.0 mm pixels, and 140 keV with 10% energy window. The obtained projection data were reconstructed by ordered subset expectation-maximization algorithm with subset and iteration number set to 16 and 8, respectively, and a voxel size of 300 μm and SPECT and CT datasets fused. At the end of imaging, mice were euthanized by an intraperitoneal injection of pentobarbital (100 mg/kg).

The scans were reconstructed and processed using InVivoScope software (Invicro, Boston, MA, USA). Tracer uptake identified as focal areas of activity in the ascending aorta, arch, and brachiocephalic trunk were identified on transverse sections by triangulating from coronal and/or sagittal images and using contrast in the vessels on the CTA for anatomical localization. Regions of interest (ROIs) were drawn around all these areas of focal activity and counts in the ROIs converted to mCi using a calibration algorithm and activity for all regions summed for the total vascular tracer uptake (Figure 1). The summed activity from the ascending aorta and arch (excluding the brachiocephalic vessels) was correlated with ex vivo gamma counting of the same tissue. To address whether there were differences in blood pool activity between the diabetic and non-diabetic mice at the time of imaging, ROIs were drawn in the center of the mid-LV cavity on the transverse slice to measure activity (in mCi/mm³).

The chest was opened and the proximal aorta and arch and brachiocephalic vessels photographed in situ. Due to time-consuming and technical challenges in dissecting the brachiocephalic vessels, the proximal aorta and arch showing plaque was dissected for correlation with activity measured on scans for the same anatomical segments. The fragment were excised and washed in PBS and weighed, and the radioactivity was counted in a gamma well counter (Wallac Wizard 1470, PerkinElmer, Waltham, MA, USA) and expressed as the percentage of injected dose per gram (%ID/g) of tissue. Radiotracer accumulation in non-target organs was determined similarly.

The proximal aorta was harvested and fixed for 24 h in formalin (10%) and paraffin-embedded. Tissue blocks were sectioned (5-μm-thick) and stained with hematoxylin-eosin (H&E) for morphological evaluation. Serial sections (5-μm-thick) were deparaffinized in xylene, treated with 0.3% hydrogen peroxide for 20 min, followed by incubation in protein-free block (Dako, Carpinteria, CA, USA) for 10 min to inhibit the nonspecific binding of the primary antibody. Sections were stained for macrophage marker Mac-3 (RM0029-11H3, rat monoclonal antibody, 1:50 dilution, Santa Cruz Biotechnology, Dallas, TX, USA), VEGFR-1 (rabbit monoclonal antibody, 1:100 dilution, Abcam, Cambridge, MA, USA), VEGFR-2 (rabbit monoclonal antibody, 1:1,500 dilution, Cell Signaling Technology, Beverly, MA, USA), and pan-endothelial cell marker FVIII (rabbit polyclonal antibody, 1:250 dilution, Dako, Carpinteria, CA, USA). Detection was performed with HRP-conjugated respective secondary system followed by diaminobenzidine (DAB substrate kit for peroxidase, Vector Laboratories) and counterstaining with Gill's hematoxylin solution.

Morphometric and immunohistochemical analyses of the arterial segments were performed using a Nikon microscope and Image-Pro Plus software (Media Cybernetics Inc., Silver Spring, MD, USA). The plaque area in the proximal aorta was measured as percent lesion area per total area of the aorta. The positively stained area in the lesion for each marker was calculated as percentage of immunostained area per total area of the vessel.

To determine the cell types expressing VEGFR-1 and VEGFR-2 in aortic lesions, dual fluorescent confocal microscopy studies were undertaken. Briefly, aortic sections (5-μm-thick) were stained for VEGFR-1 and VEGFR with the respective antibodies as described above. The slides were then incubated with fluorescent tagged secondary antibodies (Texas Red) and co-stained with fluorescent antibodies for endothelial cells (anti-FVIII, 1:200; fluorescein isothiocyanate) or macrophages (anti-Mac-3, 1:50; fluorescein isothiocyanate). The images were examined using confocal fluorescence microscope (Nikon, Melville, NY, USA) and SPOT imaging software (Diagnostic Instruments, Inc., Sterling Heights, MI, USA).

Data are reported as means ± standard deviation. Statistical comparisons between the diabetic and non-diabetic groups and between the inactivated and active tracers were made using Welch's t test for groups of unequal size and variance. Estimated power to detect an observed twofold difference between groups of non-diabetic mice imaged with inactivated tracer (n = 5) vs. active tracer (n = 8) is better than 0.8; and between groups of non-diabetic (n = 8) vs. diabetic mice (n = 10) imaged with active tracer is better than 0.95. Differences between the groups were considered significant at a value of P < 0.05. Correlation for individual values for tracer uptake between scan and well counting was assessed using the Pearson product-moment correlation coefficient.

The clearance of scV/Tc in diabetic ApoE^−/− and C57BL/6 mice was fitted to similar biexonential curves with t_1/2 for the first component of the curve 10 min and for the second 150 min (Figure 2A for diabetic ApoE^−/− mice). The biodistribution of scV/Tc in non-target organs of non-diabetic and diabetic ApoE^−/− mice was found to be similar, with highest uptake in the kidneys and liver (Figure 2B).

Scans from diabetic and non-diabetic ApoE^−/− mice showed unequivocal uptake of scV/Tc in the proximal aorta and arch on the co-registered CT scan (Figure 3A,B,C,D as representative scans) with tracer uptake extending into the brachiocephalic vessels in the diabetic mice (compare representative scan Figure 3A,B for diabetic vs. Figure 3C,D for non-diabetic). ApoE^−/− mice injected with in-scV/Tc and C57BL/6 mice injected with scV/Tc showed little or no tracer uptake in areas corresponding to the location of the proximal aorta and arch on the co-registered SPECT/CT scan (Figure 3E,F,G,H).

The blood pool activity levels determined from the scans for the non-diabetic and diabetic mice were not significantly different: 18.6 × 10^–3 vs. 19.4 × 10⁻³ mCi/mm³ (P = 0.86).

The mean cross-sectional area of the proximal aortic lesions, expressed as percent lesion area over total aortic area, in the diabetic group (32.3% ± 6.05%) was significantly larger than in the non-diabetic group receiving scV/Tc (15.3 ± 4.3%; P < 0.01) The lesion size in the non-diabetic ApoE^−/− mice receiving the in-scV/Tc (12.0% ± 2.9%) was not statistically different from the non-diabetic mice receiving scV/Tc.

The average uptake of scV/Tc or in-scV/Tc in thorax areas corresponding to the proximal aorta, arch, and brachiocepahalic vessels was calculated from scans as %ID, as described in Methods (Figure 4A). The average uptake of scV/Tc in diabetic ApoE^−/− mice was significantly higher than in non-diabetic ApoE^−/− mice (0.036 ± 0.017 vs. 0.017 ± 0.005%ID; P < 0.01) (Figure 4A). The non VEGFR-mediated uptake of in-scV/Tc in ApoE^−/− mice, was significantly lower than that of functionally active scV/Tc in age-matched ApoE^−/− mice with similar size lesions (0.008 ± 0.005 vs. 0.017 ± 0.005%ID; P < 0.01).

These findings were confirmed by ex-vivo gamma well counting of harvested fragments of aorta with the most conspicuous plaques (Figure 4B). Uptake of scV/Tc calculated for these fragments as %ID/g in diabetic ApoE^−/− mice (1.81 ± 0.50%ID/g) was significantly higher than that in the non-diabetic ApoE^−/− group (0.9 ± 0.25%ID/g, P < 0.01). In fragments harvested from non-diabetic ApoE^−/− mice with similar extent of atherosclerosis the nonspecific (non VEGFR-mediated) uptake of in-scV/Tc (0.44 ± 0.12%ID/g) was significantly lower than scV/Tc uptake (P < 0.01). Tracer uptake in the proximal aorta and arch of each mouse, determined as %ID from scans, plotted vs. %ID/g (well counter) determined for the same anatomical aortic section in the same mouse correlated significantly, (r = 0.75, P = 0.0017) (Figure 4C).

Lesions from the diabetic ApoE^−/− mice showed complex plaque anatomy with cholesterol clefts whereas lesions from age-matched non-diabetic ApoE^−/− mice were at an earlier (fatty streak) stage. Immunostaining of serial sections through the plaque in the proximal aorta revealed greater staining for VEGFR-1, VEGFR-2 in the diabetic group than in the non-diabetic group, and this staining was observed on the luminal aorta surface as well as in the inner plaque area enriched in cells stained for endothelial marker FVIII and macrophage marker Mac-3 (Figure 5). Quantitative immunohistochemistry of the aortic tissue revealed greater expression of VEGFR-1 in the diabetic mice (9.17% ± 4.46%) compared with the non-diabetic (1.88% ± 1.48%; P < 0.01). Similarly, VEGFR-2 expression was significantly higher in the diabetic mice (3.25% ± 1.27%) compared with the non-diabetic (1.31% ± 0.86%; P < 0.01). Total macrophage burden in the atherosclerotic lesions of diabetic mice (5.85% ± 2.86%) was also significantly higher compared with the non-diabetic mice (0.99% ± 0.38%; P < 0.01). Immunofluorescence studies showed co-localization of VEGFR-1 primarily with macrophage marker Mac3 and VEGFR-2 with endothelial cell marker FVIII (Figure 6).