All rabbit experiments conform to the European Union guidelines for the care and use of laboratory animals and were approved by the Animal Ethical Committee of the University of Liège (file number 1951). Seventy-three eight-week-old male New Zealand White rabbits (1.83 ± 0.14 Kg) were purchased from the Charles River, France. The animals were kept in a temperature and humidity-controlled environment in an air-conditioned room with a standard rabbit chow and tap water ad libitum. After 2 weeks of acclimation, the rabbits were randomly assigned to seven different groups and diet habituation was conducted for 2 weeks. Protocols started when the rabbits were 12 weeks old and it lasted 16 weeks. Control group (Ctrl, n = 13) received a standard chow diet. Vitamin D2 group (Vit.D2, n = 7) received a chow diet with 15 days of Vitamin D2 supplementation (ergocalciferol, Sigma-Aldrich) in drinking water at the beginning of the protocol (25,000 U/day/2.5 Kg). The cholesterol-enriched diet group (CHT, n = 17) was treated with 0.3% of the cholesterol-enriched diet and received vitamin D2 (Vit.D2) supplementation as the Vit.D2 group. The lard-enriched diet group (Lard, n =13) received 5% of the lard-enriched diet (17) with Vit.D2 supplementation. Irradiation group (IR, n = 7) received a chow diet and underwent X-ray cardiac irradiation focused on the aortic valve and the thoracic aorta during the first week of the protocol (24Gy dose divided into 3 fractions of 8Gy delivered at 48-h intervals, chosen to mimic clinically relevant doses used for anticancer radiotherapy) using image-guided radiation on an irradiator specifically made for small animals (X-Rad 225Cx, Precision X-Ray, Inc.). Irradiation combined with cholesterol-enriched diet group (IR/CHT, n = 15) received 0.3% of cholesterol-enriched diet plus Vit.D2 supplementation and underwent cardiac irradiation at mid protocol (e.g., after 8 weeks of cholesterol-rich diet). Cholesterol- and lard-enriched diets were produced by Safe (France). Bodyweight was measured on a weekly basis. Blood draws and cardiac CT (X-Rad 225Cx, Precision X-Ray, Inc.) were performed at baseline and at the end of every four weeks (W4, W8, W12, and W16), and aortic wall calcification was visually assessed as absent or present in each of the following segments: ascending aorta, aortic cross, and the descending aorta. An extra blood draw was performed at W2. Anesthesia was induced prior to any procedure (except for body weight measurements) by intramuscular injections of droperidol (0.625 mg/kg), xylazine (5 mg/kg), and ketamine (35 mg/kg) in the hind legs. Euthanasia was conducted in the anesthetized animals by intracardiac injection of pentobarbital (200 mg/kg).

Hearts were harvested, weighed, and fixed in 4% of paraformaldehyde–phosphate-buffered saline (PBS) solution for 24 h before dehydration into 70% of ethanol solution overnight at room temperature (Ctrl: n = 6, Vit.D2: n = 7, CHT: n = 11, Lard: n = 6, IR: n = 7, IR/CHT: n = 7). They were then embedded in paraffin wax and sectioned at 7 μm. Before histological and immunofluorescence analyses, the sections were incubated overnight at 58°C and rehydrated through xylene baths followed by a series of graded ethanol baths. Serial sections were used for the histological assessment of tissue structure on hematoxylin-eosin-stained sections, tissue calcification on alizarin red stained section, and extracellular matrix (ECM) composition on sirius red and Movat's pentachrome-strained sections. The aortic valve area was evaluated by isolating the aortic valve leaflets and measuring their area on multiple serial hematoxylin-eosin-stained sections (4 to 9 per rabbit) by using Adobe Photoshop software. For the immunofluorescence study, the sections were incubated overnight at 58°C prior to chemical rehydration. All sections were treated for 30 min at 96°C with Dako Target Retrieval solution, pH9 (Agilent, S2367) for antigen retrieval, then incubated for 30 min at room temperature with normal goat serum (Abcam, ab7481). Co-staining was conducted by sequential use of monoclonal mouse anti-rabbit RAM-11 primary antibody (Agilent, M0633, 1:50 dilution), TRITC-conjugated goat anti-mouse secondary antibody (Abcam, ab5885, 1:100 dilution), and AlexaFluor 488-conjugated monoclonal mouse anti-alpha smooth muscle actin (αSMA) antibody (ThermoFisher Scientific, #53-9760-82, 1:100 dilution). Nuclei were stained with 4′, 6-diamidino-2-phenylindole (DAPI) (ThermoFisher Scientific, D1306, 1:500,000 dilution). Negative controls corresponding to TRITC-conjugated goat anti-mouse secondary antibody without anti-RAM-11 primary antibody and AF488-conjugated mouse IgG2a kappa isotype control isotype were performed (Supplementary Figure S1). QuPath 0.2.3 software was used to analyze fluorescence staining in the delimitated aortic valve, the mitral valve, and the aorta area. Pixel classifiers were first defined to set a threshold value for each fluorochrome. A script was then applied in batch to automatically count the cell number, mean fluorescence intensities as well as the total tissue area, and the percentages of positive area for each immunostaining.

Aorta ring and heart valves were dissected (Ctrl: n = 7, CHT: n = 6, Lard: n = 7, IR/CHT: n = 7) at the end of the 16-week protocols, snap frozen in liquid nitrogen, and stored at −80°C until RNA extraction. Total RNA was extracted by the use of miRNeasy Mini Kit (Qiagen) according to the manufacturer's instructions. The integrity of the RNA sample was assessed on a Bioanalyser 2,100 with RNA 6,000 Nano and RNA 6,000 Pico chips (Agilent Technologies). The RNA libraries were prepared with 10 ng of RNA using the SMARTer® Stranded Total RNA-Seq Kit v2 - Pico Input Mammalian (Takara Bio USA, Inc.) according to the manufacturer's instructions. The RNA quantification and normalization were performed with the KAPA Library Quantification kit (comprising P5-P7 primers) on the ABI 7900HT system (ThermoFisher Scientific). Sequencing of the libraries was performed on a NovaSeq6000 System (Illumina) with an S4 reagent kit (300 cycles) and 1.2 nM library loading.

Paired FastQ files for all the samples were assessed for Quality Control using FastQC and then they were processed through nf-core rnaseq pipeline v1.4.2 (https://nf-co.re/rnaseq/1.4.2), with default parameters except for -clip_R2 = 3 and -nextseq-trim = 10, in order to perform mapping, quantification, and QC, and to produce the gene count matrix. The reference genome used was Oryctolagus cuniculus OryCun2.0 with annotation from Ensembl release 99 (http://jan2020.archive.ensembl.org/index.html). Gene counts were normalized by the default method (median of ratios) in the DESeq2 package (22). Gene expressions in each diet or treatment group were compared to the control group by DESeq2; a total of 29,587 tests were performed for each comparison and were followed by Benjamini and Hochberg multiple test correction. An adjusted p-value (FDR) less than 0.05 and a fold-change greater than 1.5 were used to highlight a significant difference in gene expression between the two groups. Human orthologs of differentially expressed genes (DEGs) were identified by the online tool, Better Bunny (23) with an identity of 50% as a threshold (24), and were ranked by their π-value calculated with a log-fold change (LFC) and the adjusted p-value (25). TopGO package was used to perform gene ontology (GO) analysis with a ranked gene list. The whole RNA-seq analysis was performed in R (version 4.0.4; R Foundation for Statistical Computing, Vienna, Austria).

At each blood draw, differential blood cell counts (white blood cells, red blood cell indices, and platelets) were measured in 3.8% of citrated whole blood on a Cell-Dyn 3700 hematocytometer equipped with veterinary software (Abbott Laboratories). Serum aliquots were prepared after 30-min resting to allow for blood clotting followed by centrifugation at 1,700 g and room temperature. EDTA and 3.8% citrated plasma aliquots were prepared by two successive centrifugations at 1,700 g and room temperature maximum of 1 h after blood draw. Serum and plasma samples were stored at −80°C until further analysis.

Measurements of the levels of total cholesterol, triglycerides, LDL-c, HDL-c, calcium, phosphorous, iron, albumin, AST, ALT, creatinine, and urea were performed in serum samples on an AU480 Chemistry Analyzer (Beckman Coulter). Serum levels of IL-6 and plasma [ethylenediamine tetraacetic acid (EDTA)-anticoagulated plasma] levels of DKK1 and PF4 were measured by enzyme-linked immunosorbent assays according to the manufacturer's recommendations (Cusabio). The ent-8-iso-15(S)-PGF2α levels were measured in EDTA-plasma by the ultra-high performance liquid chromatography coupled to tandem mass spectrometry detection. Levels of circulating H3.1-nucleosomes and citrullinated histone H3R8 nucleosomes (H3R8cit-nucleosome), an epigenetic modification, were assessed by blinded Nu.Q® chemiluminescence immunoassays (Volition SRL, Belgium) in EDTA-plasma (Ctrl: n = 7–13, Vit.D2: n = 7, CHT: n = 10–15, Lard: n = 7–12, IR: n = 7, IR/CHT: n = 15).

Between-treatment comparisons of all variables were performed by ANOVA and post-hoc Dunnett test on log transformed data, except for leaflet area quantification, nested GLM model, and post-hoc SIDAK tests were performed. Over time changes between two time points in the blood cells and circulating biomarkers were assessed by Wilcoxon signed-rank test. Data are presented as median [interquartile range (p25–p75)]. All tests were performed 2-sided and P-value ≤ 0.05 was considered significant. All statistical analyses were performed using SAS 9.4 (SAS Institute, Cary NC).

Rabbits fed for 16 weeks with lard- or cholesterol-enriched diets supplemented with Vit.D2 displayed aortic wall structural alterations that were not observed in rabbits fed with chow diet, or in those receiving the chow diet supplemented with Vit.D2 (Figure 1A). Alizarin red staining of heart sections from rabbits treated with a lard-enriched diet revealed massive media calcification with collagen fiber accumulation and disorganization around the calcific lesions as shown by Sirius Red and Movat's Pentachrome staining. The CT imaging revealed aortic wall calcifications in 70 to 80% of rabbits fed with lard depending on considered aorta segments (Figure 1B). Such macrocalcifications were not observed in Ctrl or Vit.D2 groups, except for one rabbit treated with Vit.D2 which developed calcification in the aortic cross.

In agreement with previous studies (26), rabbits fed for 16 weeks with a cholesterol-enriched diet developed atherosclerotic lesions. Hematoxylin-eosin staining of aortic sections showed intima thickening with cell and lipid infiltration, whereas Sirius Red and Movat's Pentachrome showed disorganized ECM component deposition (collagen fibers and glycosaminoglycans, GAGs). Alizarin red staining revealed calcification nodules at the intima-media junction and diffuse microcalcifications in the thickened intima. Approximately 35% of rabbits fed with a cholesterol-enriched diet displayed calcifications detected by a CT (Figure 1B). We also assessed the infiltration of the aortic wall by macrophages (or macrophage-like cells) by immunodetection of the rabbit RAM-11 marker (27). As described (28), a cholesterol-enriched diet-induced significant infiltration of RAM-11 positive cells in atheroma as compared to Vit.D2 alone (Figure 4C).

Alterations of the aortic valve structure were observed in the two diet groups (Figure 2A). In the cholesterol group, the aortic valve leaflets were significantly thickened with sparse calcification spots, cell infiltration and accumulation of collagen fibers, and GAGs (Figures 2A,B). Immunostaining of RAM-11 and of the muscle-specific αSMA marker revealed significant macrophage infiltration and few αSMA positive cells, both located in valvular ventricularis (Figure 3). In contrast, rabbits fed with a lard-enriched diet displayed calcific nodules in the aortic valve with only slight leaflet thickening compared to cholesterol (Figure 2) and no detectable RAM-11 or αSMA positive cells (Figure 4). No aortic valve alteration was observed in the chow diet and Vit.D2 groups.

The mitral valve was also affected in the cholesterol-enriched diet group (Figure 3), with cellular infiltrates and ECM remodeling in spongiosa, without leaflet thickening. The signal for RAM-11 positive cells was significantly increased as compared to the control and Vit.D2 groups (Figure 4C). As for the aortic valve, only few αSMA positive cells were observed in mitral valves. Rabbits from the chow diet, Vit.D2, and lard groups did not show any visible modification of the mitral valve structure.

None of the diets caused histological modifications of the tricuspid and pulmonary valves (data not shown).

Our histological analyses did not reveal any aortic histological alterations when cardiac irradiation was applied in rabbits fed with a chow diet kept for 8 weeks post-irradiation (Figure 1A). However, rabbits fed for 16 weeks with a cholesterol-enriched diet and receiving cardiac irradiation at week 9 displayed more macrocalcification in the aorta than rabbits that were not irradiated. Calcification was detected by cardiac CT in all aortic segments in 40% of rabbits (Figure 1B). Also, the atherosclerotic plaque core of rabbits that received cardiac irradiation combined with cholesterol diet significantly increased in the numbers of RAM-11 positive cells as compared to rabbits fed with a chow diet, a cholesterol diet without irradiation, and in rabbits that received irradiation whilst on chow diet (Figures 4A,C).

The addition of irradiation to the cholesterol-enriched diet induced the aortic valve structure alterations that were not observed in the irradiation-only group. Thickened aortic valve leaflets displayed more cellular infiltrates and RAM-11 positive cells than the control groups (Figures 4A,C). Aortic valves were more calcified than those of rabbits from the cholesterol group (Figures 2A,B).

As for mitral valves, the combination of irradiation with the cholesterol diet led to increased infiltration of RAM-11 positive cells as compared to the irradiation alone (Figure 4C).

To get insight into the biological mechanisms underlying vascular and valvular modifications, we then performed RNAseq experiments on RNA extracted from the dissected proximal aorta and all cardiac valves. The RNAseq identified 29.587 genes annotated in the rabbit genome, among which 15.135 genes could be assigned to human genes. The cholesterol-rich diet modified the expression of 1.102 (898 unique human orthologs) genes in the aorta, 612 (492) genes in the aortic valve, 281 (229) genes in the mitral valve, and only 4 (3) and 6 (6) genes in the tricuspid and pulmonary valve, respectively. The lard-rich diet yielded much less DEGs than the cholesterol-rich diet, comprising 76 (66) genes in the aorta and 40 (36) genes in the aortic valve. No DEGs were found in the three other cardiac valves in response to the lard diet.

We then assessed the impact of these changes in RNA expression on biological processes in each tissue. To detect the differential effects of each diet in every tissue, both tissue-specific and common DEGs were included in these analyses.

We first studied the effect of a cholesterol-rich diet on the aorta and aortic valve. Among the 612 DEGs identified in the aortic valve of rabbits from the cholesterol group, 443 DEGs were commonly affected in the aorta (Figure 5A, Supplementary Tables S1–S3). These common DEGs were involved in inflammatory and cellular defense responses, including the production of major inflammatory cytokines and neutrophil degranulation. Mitotic spindle organization was also affected, pointing to an effect on cell division. The 659 aorta-specific DEGs were enriched in genes contributing to additional effects on immune response and signaling pathways, positive regulation of cell migration, angiogenesis, and platelet degranulation, while the 169 aortic valve-specific DEGs were involved in cardiac muscle contraction, tissue morphogenesis, and myofibril assembly. These data indicate that inflammation likely contributes to aortic valve alterations induced by a cholesterol-rich diet similarly as in the aorta, but additional muscle-related processes could also be affected.

We next focused on the comparison between the aortic valve and mitral valve in the cholesterol-fed group. Whereas the DEGs commonly found in the mitral and aortic valves included genes involved in inflammation and cellular defense response, we surprisingly observed that the cholesterol-rich diet mainly affected lipid biosynthetic and metabolic processes as well as folic acid transport in the mitral valve (Figure 5B, Supplementary Tables S4–S6). This finding may thus reveal that not only inflammation but also key metabolic cellular processes underlie the mitral valve alterations in hypercholesterolemia.

We did not identify the genes that were commonly affected in the aorta and aortic valve by the lard-rich diet. The few DEGs that were found to be specific to the aortic valve were all involved in muscle-related processes, including contraction, filament sliding, and sarcomere organization (Figure 5D, Supplementary Tables S7, S8). Intriguingly, these processes were similarly found to be affected by the cholesterol-rich diet, suggesting that dietary cholesterol and palmitate could have a similar biological impact on the aortic valve. In contrast, in the aorta, ECM organization, osteoblast differentiation, and ossification processes were more prevalent in response to lard than cholesterol. Of interest, inflammation would not be the main process induced by the lard diet, which is in sharp contrast to the cholesterol diet.

The addition of irradiation to a cholesterol-rich diet significantly increased the numbers of DEGs both in the aorta and the aortic valve. This was particularly true for the aortic valve. Indeed, 2,097 (1714) and 3,474 (2715) DEGs were identified in the aorta and aortic valve, respectively, which indicated that the aortic valve might be more impacted than the aorta following irradiation. The addition of irradiation to the cholesterol-rich diet modified the expression of 15 (12) additional genes in the aorta and 602 (512) genes in the aortic valve as compared to cholesterol without irradiation. Interestingly, we observed that the common DEGs identified both in the aorta and aortic valve contributed to the same biological processes as a cholesterol-rich diet alone, namely neutrophil degranulation, inflammatory, and immune responses (Figure 5C, Supplementary Tables S9–S11). However, the gene frequency in these processes was doubled by the addition of irradiation, indicating that irradiation could promote inflammation in a neutrophil-dependent manner (Figures 5A,C). Processes, such as neutrophil chemotaxis and leukocyte migration indeed became predominant. In the aortic valve, specific processes were nevertheless affected, such as ephrin receptor and Wnt signaling pathways, heart development, and glucocorticoid metabolism, suggesting that irradiation could disturb aortic valve tissue homeostasis and metabolism. In the aorta, effects on collagen fibril and ECM organization predominated in addition to inflammation and platelet degranulation (Figures 5A,C).

We wanted to determine whether diet- and irradiation-induced structural and RNA changes could be reflected systemically. For this purpose, we measured the plasma levels of biomarkers related to some of the biological processes identified in our RNAseq dataset. These measurements were performed at baseline and after 16 weeks. We first analyzed differential blood cell counts. Neutrophil count was the only parameter to be increased in the cholesterol-treated rabbits, with or without cardiac irradiation, as compared to Ctrl rabbits (Supplementary Table S12). We then studied the effects of lard-enriched diet and cholesterol-rich diet, combined either with irradiation or without irradiation on the circulating levels of conventional lipid biomarkers (Table 1). As expected, feeding rabbits for 16 weeks with a cholesterol-enriched diet increased their circulating levels of triglycerides, total cholesterol, LDL, and HDL. Surprisingly, rabbits from the irradiation-only group had increased levels of triglycerides. In agreement with our recent study (17), the lard-enriched diet did not induce changes in conventional lipid biomarkers. The hepatic marker, asparate transaminase, increased upon the intake of a cholesterol-enriched diet, whereas the albumin levels were elevated in rabbits receiving cholesterol combined either with or without irradiation (Supplementary Table S13).

We then analyzed the levels of biomarkers involved in calcification, inflammation, oxidative stress, and platelet activation (Supplementary Figure S2). Rabbits fed for 16 weeks with a lard-rich diet had reduced circulating levels of DKK1, an inhibitor of the Wnt signaling pathway which is involved in the calcification processes (Figure 6A). In contrast, rabbits that received cardiac irradiation combined with a cholesterol diet had elevated levels of DKK1 compared to baseline. Changes in DKK1 levels were neither observed in response to cholesterol diet alone nor in the irradiation alone.

Cholesterol-enriched diet increased the plasma levels of the specific platelet activation marker, platelet factor-4 (PF4), as well as the neutrophil count and the oxidative stress marker, ent-8-iso-15(S)-prostaglandin F2α (ent-8-iso-15(S)-PGF2α) (Figures 6C,E,F). Combined irradiation and a cholesterol-rich diet induced a significant increase in the plasma levels of the major inflammatory cytokine IL-6. as well as the levels of H3.1-nucleosome and citrullinated H3R8-nucleosome (Figure 6A) are consistent with our data of RAM-11 staining and our RNAseq findings on a predominant role of inflammation and neutrophil degranulation in this rabbit group.

Lard-enriched diet did not increase inflammation- or oxidative stress-related biomarkers, in accordance with large, calcified area observed in the aorta and aortic valve without the major implication of inflammatory processes as found in the RNAseq analyses.

To identify possible early biomarkers of lesion development, we performed serial measurements of the neutrophil count, ent-8-iso-15(S)-PGF2α, H3.1-nucleosome, and citrullinated H3R8-nucleosome levels in cholesterol-enriched diet and irradiation combined with cholesterol-enriched diet groups. The neutrophil count increased as early as after 15 days following the dietary cholesterol intake (Figure 6B). Levels of the oxidative stress marker, ent-8-iso-15(S)-PGF2α, were also elevated after 15 days of cholesterol diet (Figure 6C). In the cholesterol diet group, circulating H3.1-nucleosome levels quickly rose after 15 days and 4 weeks and dropped thereafter, while it went on increasing when irradiation was applied at mid-protocol (Figure 6D). Only rabbits that received irradiation combined with a cholesterol diet had increased the circulating levels of citrullinated H3R8-nucleosome after 16 weeks (Figure 6E).