Molecular imaging of endothelial activation and mineralization in a mouse model of accelerated atherosclerosis

The institutional animal ethics committee (Comité National de Réflexion Ethique sur l’EXpérimentation Animale CENOMEXA) approved the animal experiments (#3979), and all experiments were performed with European and French regulation. A control group consisted of 31 control C57Bl/6 J mice (Ctl), a group of non-uremic 25 C57Bl/6 J ApoE^−/− mice (NUr), and a group of uremic 27 C57Bl/6 J ApoE^−/− mice (Ur) were investigated. All animals were male and maintained on a standard chow diet composed by 8.4% fat, 19.3% protein, 72.4% carbohydrates, 0.55% phosphorus, 0.73% calcium, 0.16% magnesium, and 1000 UI/kg vitamin D₃. We induced chronic renal failure by an electrocoagulation of the right kidney at 8 weeks old, followed 2 weeks later by a contralateral nephrectomy, according to a previously described procedure [11]. Analgesia was induced by injection of buprenorphine (0.05 mg/kg, Buprecase®, Axience, Pantin, France) administered in preoperative and at 9 h interval in postoperative during the first 24 h.

All acquisitions were performed using a dedicated preclinical PET-CT system (Inveon®, Siemens Healthcare, Erlangen, Germany). At 12 and 16 weeks old, anesthesia was induced with 5% isoflurane gas and maintained with isoflurane 2% gas in a mixed of O₂ and N₂O (1:2). Following attenuation CT acquisition, [18F]NaF injection (≈ 18 MBq) was performed intravenously through a tail vein catheter. Fifty minutes after [18F]NaF injection, a 10-min list mode acquisition of emission data was performed (energy and coincidence windows of 350–650 keV and 3.4 ns, respectively). Image series were reconstructed using 3D-ordered subset expectation maximization (3D-OSEM) with 8 iterations and a zoom factor 2 without scatter correction.

Postprocessing was performed using Carimas v2.4 software (Turku PET Centre, Turku, Finland). The [18F]NaF activity was represented with a color scale starting just upon the plasma activity determined drawing a volume of interest over the left ventricle with a maximum set to 200% of the plasma activity measured at the last frame of the data for each examination [12]. An uptake in a region close and above to the heart was considered as a positive aortic [18F]NaF uptake. Then, aortic uptake was semi-quantitatively determined as the ratio of the SUVmax in a VOI drawn over the visible aortic uptake to the SUVmax of vascular background measured in the inferior vena cava. In addition, global cardiac [18F]NaF uptake was also measured in a volume of interest encompassing the whole heart normalized to the same vascular background.

MR experiments were carried out in a second group of animals using a 7-T magnet (Pharmascan® Bruker, Billerica, USA) with dual respiratory and ECG gating. Telediastolic T2*-weighted MR images were acquired using a multi-slice sequence: field of view 1809 × 939 mm², TR/TE 100/4.25 ms, 12 slices, and voxel 0.1 × 0.1 × 0.15 mm³ encompassing the thoracic aorta were performed in 16-week-old mice, with and without the intravenous injection of 200 μL of MPIO-αVCAM-1.

As previously described [9], microparticles of iron oxide (DynaBeads MyOnes Tosyl Activated, ThermoFisher Scientific, Waltham, USA) were conjugated with antibodies anti-αVCAM-1 (clone A(429), BD BioScience, Franklin Lakes, USA) by incubation at 37 °C for 48 h. MRI images were analyzed using Osirix v.6.5.2 software. MPIO-αVCAM-1 binding results in a T2* signal void in the lumen of the vasculature. Intravascular diameters were measured using non-injected MR images on the following segments: (i) the aortic root, (ii) the ascending aorta, (iii) the aorta at the level of the brachiocephalic trunk, (iv) the brachiocephalic trunk, and (v) the aortic arch (Fig. 1). The diameter measurement was indexed to the animal weight measured the day of the MR acquisition.

Mice were sacrificed at 16 weeks old. Blood samples were collected to determine the plasma level of urea and calcium using Beckman-Coulter AU5800 autoanalyzer (Miami, USA). Aortas were harvested from the root to the diaphragm. In 11 Ctl, 9 NUr, and 11 Ur, aortic tissue calcium content was measured by flame atomic absorption spectroscopy (FAAS) (Varian AA240, Varian Inc., Palo Alto, CA). After an overnight bath of aortas in a 0.6 N HCl solution at 4 °C, FAAS was performed in supernatant. Aortas were dried at 37 °C, then weighted, and the vascular calcium concentration was reported on the dry weight of each aorta.

In 6 Ctl, 5 NUr, and 11 Ur, macrocalcifications were assessed using the von Kossa staining. Aortas were dissected under a microscope, perfused with heparinized (50 U/mL) phosphate-buffered saline 5/100 (PBS), and cryomounted in optimal cutting embedding medium (OCT, Thermo Scientific, Waltham, USA). Ten-micrometer-thick slices were collected each 50 μm. Cryosections were placed in 5% silver nitrate solution for 30–60 min then fixed in 5% sodium-thiosulfate solution for 2–3 min. Sections were digitized using ScanScope CS (Leica Byosystems, Wetzlar, Germany,) and tissue segmentation was manually performed using Aperio ImageScope software v12.3 (Leica Biosystem, Wetzlar, Germany). Data were expressed as the relative proportion of stained tissue to total tissue area.

Harvested aortas were homogenized in homogenization buffer (20 mM Tris HCl, 150 mM NaCl, 1 m EDTA, 1% Triton X-100), then sonicated and centrifuged (20,000 G, 15 min, 4 °C). Total protein concentrations were determined using the Bradford assay (Bio-Rad, Hercules, USA), and protein was solubilized in Laemmli 4× buffer (25 mL TrisSDS, 20 mL glycerol, 4 g SDS, 2 mL βmercaptoethanol, 1 mg Bromophenol Blue, 50 mL H₂0). This solution was vortexed, and proteins were denatured using thermal cycler GeneAmp PCR system 2700 (Applied Biosystems, Foster City, USA) at 95 °C (3 min). Proteins (50 μg) were separated on a 12% SDS-PAGE gel (86.4 g glycine, 6 g SDS, 18 g TrisBase) and transferred to a nitrocellulose membrane (BioRad, Hercules, USA). Blots were blocked with a blocking solution (non-fat dry milk). Primary antibodies were incubated 12 h at 4 °C with anti-VCAM-1 from Cell Signaling (1:1000, VCAM-1 (D2T4N) Rabbit mAb (Mouse Specific), Ozyme distributor, Saint Quentin en Yvelines, France), anti-OPN-R (1:2500, OPN-R Antibody (24H5L3) ABfinity Rabbit mAb, ThermoFisher Scientific, Waltham, USA), and anti-GAPDH (1:1000, GAPDH Loading Control Antibody (MA5-15738), ThermoFisher Scientific, Waltham, USA). After incubation with appropriate horseradish peroxidase-conjugated secondary antibodies for 1.5 h at room temperature, proteins were visualized by chemiluminescence. Immunoreactive proteins were scanned using Chemidoc XRS (Bio-Rad, Veenendaal, The Netherlands), and intensities were analyzed with ImageJ NIH image processing software (version 1.52a, Bethesda, MD, USA).

Values were expressed as mean ± SEM. Continuous data were compared using the Wilcoxon signed-rank test or Mann-Whitney U test when appropriate. Multivariate analysis was performed using logistic regression, with post hoc analysis using Tukey’s HSD test or odds ratio when appropriate. The Kruskal-Wallis test was used for global comparison between three independent animal groups. For proportions, the chi-square test was used to compare differences between groups. Statistical analyses were performed using JMP 11 (SAS Institute, Cary, NC, USA). A p value ≤ 0.05 was considered statistically significant.

Urea plasma level was significantly increased in Ur (22.87 ± 1.32 mM, n = 27) compared to NUr and Ctl mice (respectively 8.62 ± 0.25 mM, n = 25, and 9.11 ± 0.15 mM, n = 31; p < 0.0001; Fig. 2a). The concentration of calcium in plasma was significantly higher in Ur (2.54 ± 0.09 mM, n = 17) compared to NUr and Ctl mice (respectively 2.24 ± 0.01 mM, n = 22, and 2.14 ± 0.02 mM, n = 27; p < 0.0001; Fig. 2b). Flame atomic absorption spectroscopy demonstrated a significant increase of vascular calcium in Ur mice (0.51 ± 0.06 μg of Ca²⁺ per milligram of dry weight aorta, n = 11) compared to NUr (0.27 ± 0.05 μgCa²⁺/mg, n = 9, p = 0.013) and Ctl mice (0.28 ± 0.05 μgCa^2+/mg, n = 11 p = 0.014; Fig. 3). The von Kossa staining demonstrated a non-significant increase of vascular macrocalcifications in Ur mice (0.48 % ± 0.06, n = 11) compared to NUr and Ctl mice (respectively 0.41 ± 0.03, n = 5, and 0.43 ± 0.02, n = 6, NS).

At 12 weeks old, [18F]NaF PET demonstrated an aortic uptake in 8/12 (66%) Ur mice (Fig. 4), 5/9 (55%) NUr mice, and 0/3 (0%) in Ctl (Table 1). However, only Ur mice (4/9, 45%) had an aortic uptake at 16 weeks old (0/4 in NUr mice and 0/4 in Ctl; Table 1). There was a significant impact of animal group on aortic [18F]NaF uptake (p = 0.01), and post hoc analysis showed a significant difference between Ur and Ctl mice (p = 0.002). A higher proportion of macrocalcifications on the von Kossa staining was noted in mice without aortic uptake compared to mice with aortic [18F]NaF uptake at 16 weeks old (PET− 0.89% ± 0.09 vs. PET+ 0.42% ± 0.07, p = 0.028). There was a global effect of the genotype and the age of the animals on cardiac [18F]NaF uptake assessment (global p value < 0.05; Table 2). In addition, there was a positive corrrelation between cardiac and aortic [18F]NaF uptake in animals demonstrating positive aortic [18F]NaF uptake (r = 0.69, p < 0.01).

After intravenous MPIO-αVCAM-1 injection, 4/5 (80%) Ur mice (Fig. 5) and 1/4 (25%) NUr mice demonstrated a MR signal void in the aortic wall, whereas images were normal in all control mice (Table 3). There was a significant impact of animal group on MPIO-αVCAM-1 binding (global p value = 0.01), and post hoc analysis showed a significant difference between Ur and Ctl mice (p = 0.003). Signal void was mostly located within the aortic root. The localizations of MPIO-αVCAM-1 binding is summarized in Table 3.

Morphological MR imaging demonstrated a vascular dilation involving the aorta at the level of the brachiocephalic trunk, the brachiocephalic trunk itself, and the aortic arch in Ur compared to NUr and Ctl mice. The indexed vascular diameters were as follows: aorta at the level of the brachiocephalic trunk, 80.23 ± 3.50 μm/g (Ur, n = 4) vs. 67.62 ± 3.62 (NUr, n = 4) and 66.10 ± 2.94 (Ctl, n = 5, global p value < 0.05); brachiocephalic trunk, 36.36 ± 1.39 (Ur, n = 4) vs. 30.80 ± 0.65 (NUr, n = 4) and 31.30 ± 1.37 (Ctl, n = 5, global p value < 0.05); aortic arch, 70.49 ± 2.80 (Ur, n = 4) vs. 57.99 ± 2.92 (NUr, n = 4) and 55.70 ± 1.74 (Ctl, n = 5, global p value < 0.05); aortic root, 76.23 ± 7.62 μm/g (Ur, n = 4) vs. 66.69 ± 3.80 (NUr, n = 4) and 70.86 ± 1.51 (Ctl, n = 5, NS); and ascending aorta, 66.93 ± 3.96 μm/g (Ur, n = 4) vs. 60.30 ± 2.44 (NUr, n = 4) and 59.36 ± 3.58 (Ctl, n = 5, NS).

In addition, we analyzed the results of imaging at 16 weeks (positive vs. negative) with respect to the animal group (Ur, NUr, Ctl) and to the molecular imaging modality (PET and MPIO-αVCAM-1 MRI). The logistic regression demonstrated a significant effect of the animal group on the imaging results (p < 0.001), independent of the imaging modality (p = 0.99, ns). The odds ratios for positive imaging results were significant for Ur vs. NUr (p = 0.01) and Ur vs. Ctl (p < 0.001) but not for NUr vs. Ctl (p = 0.99, ns).

The aortic expression of VCAM-1 protein was significantly different between animal groups (Ur 1.27 ± 0.22 AU, n = 11, NUr 0.94 ± 0.12, and Ctl 0.56 ± 0.09, global p value = 0.02), with a significant difference between uremic ApoE^−/− and control mice (p < 0.01, Fig. 6a). No difference was found between groups for aortic OPN-R protein expression (Fig. 6b).