Introduction

Atherosclerosis and its complications (stroke and myocardial infarction) are the worldwide leading cause of death [1]. It is now considered an inflammatory disorder induced by low-density lipoprotein (LDL) particles, implicating both innate and adaptive immunity [1]. This new paradigm led to the assessment of different immunosuppressive or immunomodulatory treatments in human atherosclerosis targeting neutrophils, lymphocytes, monocytes and/or their leading cytokines with colchicine [2, 3], methotrexate [4] or canakinumab [5]. Colchicine and canakinumab improved clinical outcomes in selected patients, decreasing the risk of cardiovascular events in secondary prevention by 15 to 30% [2, 3, 5]. However, as reported with interleukin-1 inhibitors and methotrexate, their use may be hampered by significant weight gain, potentially accelerating the course of atherosclerosis, or by cytolytic hepatitis [6, 7]. Hence, an optimal and refined understanding of the pathophysiology of atherogenesis and the consequent development of new therapeutic targets remain necessary.

In ApoE−/− mice, the absence of apolipoprotein E (ApoE) leads to hypercholesterolemia and to the development of atherosclerotic lesions from 10 weeks of age [8, 9]. Positron emission tomography (PET) is a noninvasive method for the detection of aortic inflammation that has been used for assessing atherogenesis in vivo in ApoE−/− mice [10,11,12,13,14]. Hybrid PET-magnetic resonance (MR) imaging allows for quantitative or semiquantitative analysis of aortic glucose uptake within vascular structures by noncontrast MR angiography. Hence, in removing the life-threatening injected volumes of CT contrast, PET-MR is theoretically most fitted to the physiology of small animals.

Among the proposed methods to accelerate atherosclerosis in ApoE−/− mice, a high-fat high-cholesterol Western diet (WD) multiplies the burden of atherosclerotic lesions by approximately 3-fold [15] and increases the number of proinflammatory cells [16]. The reduction of the nephron mass of at least 50% also potentiates the development of atherosclerosis [17, 18], and subtotal 5/6 nephrectomy-induced uremia increases the area of atherosclerotic lesions 10-fold after 12 weeks of uremia [19]. Despite their high relevance for translational research, no study has compared the performances of these two methods in accelerating atherogenesis in ApoE−/− mice, especially by means of hybrid PET-MR imaging.

We therefore aimed to compare 18F-fluorodeoxyglucose (18F-FDG) PET-MR imaging and immunohistological findings in ApoE−/− mice, with or without WD or uremia, and in wild-type mice.

Materials and methods

Mice

Male C57BL/6 ApoE−/− (B6.129P2-Apoetm1Unc/J) and control C57BL/6J wild-type mice were purchased from The Jackson Laboratory (Sacramento, CA, USA) and Janvier Labs (Le Genest St. Isle, France), respectively, and housed in a temperature-controlled specific pathogenic free environment with ad libitum access to food and water. Five groups of mice were studied, including a control group of 6 standard diet-fed C57BL/6 mice. Twenty-seven ApoE−/− mice were randomly allocated to one of 4 groups: standard diet-fed ApoE−/− (SD, n = 7), SD and uremic ApoE−/− (SD-U, n = 7), WD ApoE−/− (WD, n = 7), WD and uremic ApoE−/− (WD-U, n = 6). Chronic renal failure was induced by electrocoagulation of the right kidney at 8 weeks old, followed by a contralateral nephrectomy two weeks later, as previously described [13]. Analgesia was obtained by preoperative subcutaneous injection of buprenorphine (0.05 mg/kg) and then every 12 h as long as needed. Mice were fed standard chow up to 10 weeks old, and then for the next 8 weeks, 14 mice continued on standard chow and 13 mice switched to WD (SAFE Western 1635 v35, SAFE Diet, Augy, France, made of 21.2% fat, 17% proteins, 48.5% carbohydrates, 0.2% total cholesterol) according to the experimental groups. The mice were separated according to the defined diet, but uremic mice were cohoused with nonuremic mice to decrease the cage effect. An additional analysis was performed to compare all mice that were fed an SD versus a WD and to compare all nonuremic mice to uremic mice.

All animal procedures were approved by the regional animal ethics committee (Comité d’Éthique NOrmand en Matière d’EXpérimentation Animale, CENOMEXA 054, n°29982) and were performed in accordance with the European directive 2010/63/EU on the protection of animals used for scientific purposes and specific French laws (decree n°2013–118). The study is reported in accordance with ARRIVE guidelines.

PET-MR imaging

Images were acquired in 18-week-old mice, i.e., 8 weeks after the introduction of uremia and/or WD, using a dedicated hybrid PET-MR 7T system (BioSpec 70/18, Bruker, Germany). The MR acquisition sequences consisted of a localizer, an axial 2D TOF FLASH angiography (TE: 1.657 ms, TR: 12 ms, 125 ​× ​125 matrix, bandwidth: 700 Hz/pixel, flip angle: 80°, averages: 2, slice thickness: 0.6 mm, slices: 60), an axial T2 TurboRARE (TE: 24 ms, TR: 1720 ms, image size: 150 ​× ​150, RARE factor: 5, averages: 8, slice thickness: 0.5 mm, slices: 30) and a whole-body FISP (TE: 2.6 ms, TR: 5.5 ms, flip angle: 10°, image size: 80 ​× 80, averages: 3, slice thickness: 0.5 mm, slices: 60). The 2D TOF imaging was gated to the ECG to avoid cardiac motion artifacts, and T2 TurboRARE was gated to the respiratory signal. MR images were used as landmarks to quantify 18F-FDG uptake in vessels of interest.

The mice were fed a ketogenic diet for 2 days and then fasted 18 h before 18F-FDG (Curium, Glisy, France) was intraperitoneally injected into conscious mice to decrease myocardial uptake [20,21,22]. Then, anesthesia was induced with 5% isoflurane gas and maintained with isoflurane 1–2% in an N2O/O2 mixture (2:1), and the mice were placed in a prone position headfirst in the PET-MR system. A static whole-body PET acquisition (15 min) was acquired 90 ± 1 min after the 18F-FDG injection, with energy windows and coincidence windows of 357–664 keV and 5 ns, respectively. The axial scan length was 117 mm. The image data were corrected for dead time, radioactive decay, attenuation, scatter and randoms, and sinograms were reconstructed using a 3-dimensional OSEM algorithm (number of iterations: 12, number of subsets: 16) into a 128 × 128 matrix (slice thickness 0.75 mm).

3D Slicer 4.11 (http://www.slicer.org/) with the PET-IndiC extension was used to analyze PET-MR images [23,24,25,26]. Alignment of the PET data to MR images was refined using a rigid transformation (6 degrees of freedom). The TOF sequences were used to manually draw volumes of interest (VOIs) encompassing the entire available portion of the ascending thoracic aorta, descending thoracic aorta, abdominal aorta and inferior vena cava. We divided the aorta into these three sections because atherosclerotic lesions in ApoE−/− mice do not develop at the same speed in each arterial segment, with the first lesions appearing in the thoracic aorta, particularly the aortic root, and because the division between ascending and descending thoracic aorta can lessen the impact of a possible spillover effect [9, 27]. Then, the VOIs of the aorta were dilated by the neighborhood method (8 neighbors) to include the vessel wall. The target-to-background ratio (TBR) was defined as the ratio of the maximum uptake within each VOI of the aorta and of the maximum uptake of vascular background measured in the inferior vena cava.

Blood analyses and ex vivo radioactivity measurement

Immediately after PET-MR imaging, blood samples were obtained by intracardiac puncture and collected in dry tubes while mice were under deep anesthesia. Analyses were performed by an external laboratory according to their internal protocols (LABÉO Frank Duncombe, Saint-Contest, France). Then, the mice were euthanized by decapitation under deep anesthesia. All precautions were taken to minimize suffering. After the aorta was removed, periaortic brown adipose tissue was removed and a piece at the thoracoabdominal junction was harvested, regardless of the PET-MR results, weighed and gamma counted with an ad hoc18F protocol for 60 s (Wizard 2470, PerkinElmer, Boston, MA, USA), 163 ± 8 min after the 18F-FDG injection. The results were expressed as percentage of injected dose per gram of tissue (%ID/g).

Histological staining

Atherosclerotic lesions were revealed with Oil red O staining. Immunostaining was performed on a BenchMark XT (Ventana Medical Systems Inc., Tucson, AZ, USA). The following rabbit antibodies were used according to the manufacturers’ protocols: anti-CD4 (1:50, Cell Signaling, catalog number: #25229), anti-CD8α (1:200, Cell Signaling, catalog number: #98941), anti-CD20 (1:50, Abcam, catalog number: #ab64088), anti-CD31 (1:100, Cell Signaling, catalog number: #77699), anti-CD11c (1:100, Cell Signaling, catalog number: #97585), anti-CD68 (1:600, Cell Signaling, catalog number: #97778), anti-CD163 (1:200, Abcam, catalog number: #ab182422), anti-interleukin (IL)-1β (1:150, Abcam, catalog number: #ab205924), anti-IL-1α (1:100, Abcam, catalog number: #ab7632), anti-IL-6 (1:100, ThermoFisher Scientific, catalog number: #BS-0782R), anti-IL-17 A (1:100, ThermoFisher Scientific, catalog number: #PA5-79470), and anti-interferon-γ (1:100, Abcam, catalog number: #ab216642). Bound antibodies were detected using avidin-biotin-peroxidase complex (ChromoMap DAB and OmniMap anti-Rabbit HRP, Roche Diagnostics, Meylan, France). Images were acquired using an Olympus VS120 slide scanner (Olympus, Tokyo, Japan) at 20× magnification. Tissue segmentation was manually performed by the same investigator (SD) using the QuPath software package (v0.2.3) [28], and data were expressed as the mean percentage of positive cells using the “Positive cell detection” option (optical density sum, Score compartment: DAB OD max with single threshold at 0.25) or as the mean percentage of positive area for the Oil red O staining (threshold set to 0.5 and smoothing sigma set to 1 with a full resolution of 0.35 μm/px).

Statistical analyses

Quantitative data were expressed as the mean ± standard error of the mean and analyzed using the nonparametric Mann–Whitney test, or, when more than two groups were compared, by the nonparametric Kruskal–Wallis test, followed if significant by Dunn’s multiple comparison test. GraphPad Prism 7 (GraphPad Software Inc., San Diego, CA, USA) was used for statistical analyses. A p value < 0.05 was considered significant.

Results

Five groups of male mice were analyzed: SD ApoE−/− (n = 7), SD-U ApoE−/− (n = 7), WD ApoE−/− (n = 7), WD-U ApoE−/− (n = 6) and control C57BL/6 mice (n = 6). All experiments were performed in 18-week-old mice, with a mean weight of 29.85 ± 0.42 g.

Biological analyses

Blood total cholesterol levels were significantly higher in the WD ApoE−/− and WD-U ApoE−/− groups than in the control C57BL/6 mice (p < 0.05, Table 1). Blood triglyceride levels were significantly higher in SD-U ApoE−/− mice than in control C57BL/6 mice (p = 0.042).

Table 1 Results of biological analyses
Full size table

PET-MR imaging and ex vivo radioactivity measurement

The mice received a mean of 22.15 ± 0.74 MBq of 18F-FDG, without a significant difference among the five groups (p = 0.54). An accidental intracolic injection of 18F-FDG prevented the analysis of the images for one mouse in the SD ApoE−/− group. No significant difference was found between groups regarding the TBR, regardless of the aortic segment (Figs. 1 and 2). The results were similar with the mean uptake within each VOI (see Online Resource 1). The heart uptake was similar between groups (p = 0.13, see Online Resource 2).

Fig. 1
figure 1

Results of 18F-FDG PET quantification in 18-week-old standard diet-fed ApoE−/− (n = 6), standard diet-fed and uremic ApoE−/− (n = 7), Western diet-fed ApoE−/− (n = 7), Western diet-fed and uremic ApoE−/− (n = 6) or wild-type C57BL/6J mice (n = 6)

Full size image
Fig. 2
figure 2

Representative 18F-FDG PET-MR imaging (sagittal sections of 2D TOF FLASH MR angiography sequences). a: standard diet ApoE−/− mice; b: standard diet and uremic ApoE−/− mice; c: Western diet ApoE−/− mice; d: Western diet and uremic ApoE−/− mice; e: wild-type C57BL/6J mice; H: heart; Ab: abdominal aorta; Des: descending thoracic aorta; Asc: ascending thoracic aorta; BAT: paravertebral brown adipose tissue

Full size image

In addition, no significant differences were found regarding the radioactivity measured by gamma counting: 1.7 ± 0.49%ID/g in the SD ApoE−/− group (n = 2), 5.2 ± 0.83%ID/g in the SD-U ApoE−/− group (n = 5), 4.7 ± 0.81%ID/g in the WD ApoE−/− group (n = 7), 4.3 ± 0.62%ID/g in the WD-U ApoE−/− group (n = 5) and 2.8 ± 0.62%ID/g in the control C57BL/6 mice (n = 6), p = 0.07.

Histological staining

Because the results for thoracic or abdominal segments did not add specific relevance, the results for these segments were combined (Table 2). There were significantly more positive cells marked by anti-CD8 and anti-CD11c in WD-U ApoE−/− mice than in control C57BL/6 mice (p = 0.042 and 0.018, respectively) (see Online Resource 3). In contrast, there was a decreased proportion of positive cells marked by CD31 in the SD ApoE−/− group and in the SD-U ApoE−/− group compared with the control C57BL/6 mice (p = 0.007 and p = 0.002, respectively). There was also a decreased percentage of IL-17 A-positive cells in the SD ApoE−/− group and in the SD-U ApoE−/− group compared with the control C57BL/6 mice (p = 0.0004 and p = 0.003, respectively) and in the SD ApoE−/− group compared with the WD ApoE−/− group (p = 0.032). Regarding anti-CD68 staining, there were significantly more positive cells in the WD-U ApoE−/− group than in the SD ApoE−/− group (p = 0.026) or the control C57BL/6 group (p = 0.002). For IL-6 staining, the WD ApoE−/− group and the WD-U ApoE−/− group had significantly more positive cells than the SD-U ApoE−/− group (p = 0.022 and 0.006, respectively). There was also an increased proportion of interferon-γ-positive cells in the WD ApoE−/− group and in the WD-U ApoE−/− group compared with the SD ApoE−/− group (p = 0.002 and 0.005, respectively) and in the SD-U ApoE−/− group (p = 0.013 and 0.029, respectively). The results with the anti-IL-1α antibody showed an increased proportion of positive cells in the WD-U ApoE−/− group compared with the SD-U ApoE−/− group (p = 0.047). The results regarding the anti-IL-1β and the Oil red O staining were different between groups (p = 0.035 and 0.029, respectively), but the p values with the post hoc Dunn’s multiple comparison test were not significant.

Table 2 Results of histological staining of mouse whole aortas in each group
Full size table

To compare the respective contribution of each intervention, an analysis compared all ApoE−/− mice that were fed with an SD versus a WD, regardless of their renal status, and compared all nonuremic mice to uremic mice, regardless of their diet. The TBR, regardless of the aortic segment, was comparable between SD and WD mice and between nonuremic mice and uremic mice (Fig. 3). Similarly, no significant results were found by gamma counting (p = 0.17 and 0.11, respectively). Regarding histological staining (Table 3), no significant differences were observed between nonuremic and uremic mice. In contrast, when combining the results for the thoracic and abdominal aortas, each staining was significantly higher in WD mice than in SD mice, except for the CD4 antibody.

Fig. 3
figure 3

Results of 18F-FDG PET quantification in 18-week-old ApoE−/− mice according to their diet or renal status. Upper line: standard diet (n = 13) versus Western diet (n = 13)-fed ApoE−/− mice. Bottom line: nonuremic (n = 13) versus uremic (n = 13) ApoE−/− mice

Full size image
Table 3 Histological staining results of mouse whole aortas according to diet or renal status
Full size table

Discussion

To our knowledge, this is the first study assessing two frequently used methods of accelerated atherosclerosis in ApoE−/− mice. We found that with its current sensitivity, our 18F-FDG PET protocol at the 8-week follow-up did not detect metabolic remodeling within the aortic wall, whereas inflammation was revealed by immunostaining. In addition, this vessel wall inflammation was found only in WD mice.

The similarity of aortic 18F-FDG uptake assessed by 18F-FDG uptake or gamma counting between all groups, including the wild-type mice, was previously reported. Indeed, in the study by Ahmed et al. [11]., the aortic 18F-FDG uptake was similar in WD ApoE−/− mice or in WD wild-type mice of 16 and of 20 weeks of age. In addition, in the study by Hag et al. [10]., aortic 18F-FDG uptake, assessed by SUVmean and gamma counting, was significantly higher in high-fat diet-fed ApoE−/− mice than in SD ApoE−/− mice after 16 weeks of diet but not after only 8 weeks. Similarly, in the study by Zhao et al. [12]., gamma counting of aortic 18F-FDG uptake was similar in high fat, high cholesterol diet-fed ApoE−/− mice and in high fat, high cholesterol diet-fed wild-type mice compared with SD wild-type mice after 5 weeks of diet but was higher after 13 weeks of diet. The study by Laurberg et al. [14]. even suggests that aortic 18F-FDG uptake is lower in nonatherosclerotic segments of the aorta than in advanced atherosclerotic lesions. Taken together, these results emphasized the lack of sensitivity of 18F-FDG PET in assessing aortic inflammation in ApoE−/− mice, even when using a hybrid PET-MR approach. We have chosen a study endpoint of 8 weeks after the accelerated atherosclerosis method because intralesional macrophages at the early stage of foam cell formation are responsible at least in part for 18F-FDG uptake [29] and because atherosclerotic lesions can be histologically detected as early as 3 weeks of WD [9] and as 6 weeks of uremia [30]. But, based on the results, it might be possible that our study endpoint was too early for demonstrating differences between groups. Although PET acquisition 3 h after 18F-FDG injection has been reported to optimize the quantitation of atherosclerotic plaque inflammation in humans by decreasing blood pool activity [31], our PET acquisition 90 min after 18F-FDG injection is in line with previous studies, with a delay ranging from 60 to 120 min after injection [12, 32]. In addition, to our knowledge, no study comparing different 18F-FDG PET acquisition times has been performed in the ApoE−/− mouse model.

In contrast to metabolic imaging, histological staining clearly highlighted differences between ApoE−/− versus wild-type mice. As expected, because of the involvement of CD8+ cells and macrophages in the pathophysiology of atherosclerosis [33], more CD8- and CD68-positive cells were found in ApoE−/− mice than in wild-type mice, but the difference was significant only in the WD-U group. This group also exhibited more CD11c-positive cells, a phenotypic marker that is expressed by dendritic cells that exert both pro- and antiatherosclerotic functions but also by other cells, especially the monocyte-macrophage lineages [34]. Of note, the staining of CD163, a marker of anti-inflammatory and atheroprotective macrophages [35], is similar between groups, contrary to the results with the anti-CD68 antibody that is expressed on both pro- and anti-inflammatory macrophages.

In addition, immunostaining also revealed some differences between ApoE−/− mouse groups. Indeed, higher proportions of CD68-, interferon-γ-, IL-1α- and IL-6-positive cells were found in WD ApoE−/− mice and/or in WD-U ApoE−/− mice than in SD ApoE−/− and/or in SD-U ApoE−/− mice. Although some atheroprotective functions have been described for interferon-γ and IL-6, these cytokines and IL-1α are mainly considered atherogenic [36,37,38].

Two other features seem paradoxical. CD31, an adhesion molecule that is expressed in particular by endothelial cells, participates in the development of atherosclerotic lesions by sensing shear stress and through the transmigration of leukocytes. Therefore, double knockout ApoE−/−/CD31−/− mice exhibit a significant reduction in atherosclerotic lesions [39]. We found that SD ApoE−/− and SD-U ApoE−/− mice have fewer positive CD31 cells within the aortic wall than control C57BL/6 mice. This may be secondary to the decreased expression of CD31 within the endothelial layer from 6 weeks to 16–20 weeks in this model, in parallel with the evolution of atherosclerotic lesions from the fatty to fibrous stage [40]. In addition, some proinflammatory cytokines can be responsible for both CD31 internalization and decreased synthesis, as shown with tumor necrosis factor-α and interferon-γ [41, 42]. The role of IL-17 A in atherogenesis is controversial, with both atherogenic and atheroprotective functions depending on the studies [43]. Indeed, IL-17 A may be involved in the early stage of atherosclerosis development and may explain a different percentage of positive cells among ApoE−/− mouse groups. Of note, the staining by anti-IL-17 A antibody was mainly localized in endothelial cells, which is unusual and may suggest a lack of specificity of the antibody.

In this study, the WD groups of mice exhibited a higher percentage of inflammatory cells than the SD groups. In contrast, the induced uremia did not seem to increase inflammatory cells. In line with these findings, no infiltration by B or T cells has been observed in previous studies with uremic mice [17, 44]. Previous results are controversial regarding the involvement of macrophages, demonstrating [45, 46] or not [18, 30] an increased macrophage content in atherosclerotic lesions. Uremia, although contradicted by some studies [17, 44], contributes to accelerating the calcification processes as previously described [30, 47], including by our team [13]. Therefore, it may be possible that uremia leads to arterial mineralization, but perhaps with a lesser inflammatory component. Of note, the mice were separated according to the defined diet, but uremic mice were cohoused with nonuremic mice. The gut microbiota is involved in atherosclerosis [48], and this cohousing may have led a partial normalization of fecal microbiota in these two latter groups [49], which may explain some of our results.

Our study has several limitations. Besides the small number of mice within each group resulting in a lack of statistical power, the PET acquisition time point at 90 min and the endpoint 8 weeks after the introduction of the method to accelerate atherosclerosis, previously discussed, we only studied male mice, whereas atherosclerotic lesions are larger and more advanced in females in 14- to 16-week-old uremic and nonuremic ApoE−/− mice [30, 50]. We did not assess other radionuclides that could be of interest in atherosclerosis, such as 18F-sodium fluoride, which targets active vascular microcalcifications and is not affected by myocardial uptake [13, 51, 52]. We have chosen ApoE−/− mice over other models, notably rats, because of the extensive study of this model in atherosclerosis, even if a partial volume effect is probable in our study, as in previous vascular PET studies in mice [10, 11, 14]. We did not characterize the composition of the atherosclerotic lesions, and we were unable to analyze the colocalization of immunostaining. Also, the ketogenic diet could not induce complete suppression of myocardial uptake and potential spillover into the ascending thoracic aorta cannot be excluded, although the myocardial uptake was similar between groups. In addition, the gamma counting of aortic 18F-FDG uptake was similar between groups. We used the intraperitoneal route instead of the more commonly used intravenous route. However, the pharmacokinetics of the intraperitoneal route is similar to the intravenous route, provided that the PET acquisition is performed more than 60 min after 18F-FDG administration [22]. In addition, the intraperitoneal route allows injecting conscious mice, a condition that further decrease 18F-FDG myocardial uptake [20]. We were also unable to obtain enough blood from each mouse to carry out all the necessary analyses, affecting statistical power. However, biological results in these mice have previously been well studied [8,9,10,11,12,13, 15, 17,18,19, 30, 44, 45, 47, 53,54,55,56].

Conclusions

This first comparative study, with a dual staining and metabolic imaging assessment, of two frequently used methods of accelerated atherosclerosis in predisposed ApoE−/− mice provided two major results with their own consequences. First, 18F-FDG PET-MR imaging at the 8-week follow-up failed to reveal an early atherogenic remodeling of the aorta that was nevertheless demonstrated using immunostaining. Second, only WD, and not uremia, significantly led to accelerated atherosclerosis, at least in its inflammatory component. Therefore, future studies with new anti-inflammatory and targeted therapies in ApoE−/− mice should use a WD to accelerate (inflammatory) atherosclerosis and should not rely solely on 18F-FDG PET to assess the efficacy. Other possible discriminating imaging approaches could be proposed, such as other radiotracers than 18F-FDG or micron-sized particles of iron oxide conjugated with antibodies targeting early endothelial activation and/or mineralization processes involved in atherosclerosis.