Hyperlipidaemia is a major risk factor of atherosclerotic cardiovascular disease (ASCVD). Risk of cardiovascular events depends on cumulative lifetime exposure to low-density lipoprotein cholesterol (LDL-C) and, independently, on the time course of exposure to LDL-C, with early exposure being associated with a higher risk1. Furthermore, LDL-C fluctuations are associated with ASCVD outcomes2–4. However, the precise mechanisms behind this increased ASCVD risk are not understood. Here we find that early intermittent feeding of mice on a high-cholesterol Western-type diet (WD) accelerates atherosclerosis compared with late continuous exposure to the WD, despite similar cumulative circulating LDL-C levels. We find that early intermittent hyperlipidaemia alters the number and homeostatic phenotype of resident-like arterial macrophages. Macrophage genes with altered expression are enriched for genes linked to human ASCVD in genome-wide association studies. We show that LYVE1+ resident macrophages are atheroprotective, and identify biological pathways related to actin filament organization, of which alteration accelerates atherosclerosis. Using the Young Finns Study, we show that exposure to cholesterol early in life is significantly associated with the incidence and size of carotid atherosclerotic plaques in mid-adulthood. In summary, our results identify early intermittent exposure to cholesterol as a strong determinant of accelerated atherosclerosis, highlighting the importance of optimal control of hyperlipidaemia early in life, and providing insights into the underlying biological mechanisms. This knowledge will be essential to designing effective therapeutic strategies to combat ASCVD.
Early intermittent feeding of mice with a high-cholesterol Western-type diet accelerates atherosclerosis compared with late continuous exposure to the Western-type diet, despite similar cumulative circulating LDL-C levels.
publisher-imprint-nameNature Portfoliovolume-issue-count5issue-article-count62issue-toc-levels0issue-pricelist-year2024issue-copyright-holderSpringer Nature Limitedissue-copyright-year2024article-contains-esmYesarticle-numbering-styleUnnumberedarticle-registration-date-year2024article-registration-date-month8article-registration-date-day27article-toc-levels0toc-levels0volume-typeRegularjournal-productNonStandardArchiveJournalnumbering-styleUnnumberedarticle-grants-typeOpenChoicemetadata-grantOpenAccessabstract-grantOpenAccessbodypdf-grantOpenAccessbodyhtml-grantOpenAccessbibliography-grantOpenAccessesm-grantOpenAccessonline-firstfalsepdf-file-referenceBodyRef/PDF/41586_2024_Article_7993.pdfpdf-typeTypesettarget-typeOnlinePDFissue-online-date-year2024issue-online-date-month10issue-online-date-day9issue-typeRegulararticle-typeOriginalPaperjournal-subject-primaryScience, Humanities and Social Sciences, multidisciplinaryjournal-subject-secondaryScience, Humanities and Social Sciences, multidisciplinaryjournal-subject-secondaryScience, multidisciplinaryjournal-subject-collectionScience (multidisciplinary)open-accesstrue
These authors contributed equally: Xiaohui Zhao, Hwee Ying Lim, Costan G. Magnussen, Olli T. Raitakari, Véronique Angeli
Main
A large body of evidence supports a causal role of cholesterol in the development of atherosclerosis, in part through activation of the NLRP3–IL-1β pathway5,6. Despite these advances, the precise mechanisms that are responsible for the development of inflammatory plaques in response to cholesterol overload are not well understood.
Besides the circulating concentration of LDL-C, exposure duration and therefore the area under the LDL-C versus age curve is an independent predictor of incident ASCVD events1. Notably, the time course of area of cholesterol accumulation is a strong independent determinant of ASCVD events, with early age accumulation being associated with a greater cardiovascular disease risk1. Furthermore, periodic variations in circulating cholesterol levels are associated with incident ASCVD, independently of mean LDL-C concentration2,3, even under statin treatment4. The questions therefore regard (1) how early and intermittent variations in serum cholesterol increase ASCVD risk; and (2) how this can be modelled to understand the underlying mechanisms. Here we have begun to answer these important questions.
Early intermittent WD and atherosclerosis
Animal models of atherosclerosis rely on the induction of hyperlipidaemia7,8, the extent and size of atherosclerotic plaques being proportional to the achieved levels of circulating cholesterol7–10. Since Ignatowski developed the first animal model of diet-induced atherosclerosis in 1908, these models have mostly relied on inducing sustained elevations in circulating cholesterol levels for a defined period of time until the experiment is terminated for analysis of plaque size and characteristics. As such, these models address the effect of ‘late’ hyperlipidaemia (that is, late during lifetime) on the development of atherosclerosis, but ignore the potential effect of lifelong variable exposure to circulating cholesterol on disease development and progression, as is the case in humans.
To address these deficiencies, we designed an experimental protocol of diet-induced atherosclerosis that accounts for these parameters. In this model of early intermittent hyperlipidaemia, the area under the cholesterol–time curve (the cumulative cholesterol exposure) is unchanged when compared to the current models of late sustained hyperlipidaemia, but is spread over the lifetime of the animal (Fig. 1a and Extended Data Fig. 1a).
Early intermittent hyperlipidaemia accelerates atherosclerosis in mice.
a, The experimental set-up. LDL receptor-deficient (Ldlr−/−) male mice on a normal diet were fed a WD for 6 weeks either continuously (cWD) or intermittently (iWD). b, Calculation of the estimated area under the curve of plasma cholesterol levels over the whole period of the experiments. n = 6 per group. c, Representative photomicrographs of Oil Red O staining of the aortic sinus in the two groups of mice at the end of the experiment. Scale bars, 500 μm. d, The mean atherosclerotic plaque size in the aortic sinus of the two groups of mice. n = 16 (cWD) and n = 15 (iWD). e,f. Ldlr−/− male mice on a normal diet were put under a cWD or iWD diet for 12 weeks. e, Representative photomicrographs of Oil Red O staining of the aortic sinus in the two groups of mice et the end of the experiment. Scale bars, 500 μm. f, The mean atherosclerotic plaque size in the aortic sinus of the two groups of mice. n = 5 (cWD) and n = 6 (iWD). g, Ldlr−/− male mice on a normal diet were put on a cWD versus iWD diet for 6 weeks. To deplete gut microbiota, the two groups of mice were treated with oral antibiotics (Methods) starting from week 9 until the end of the experiment. The mean atherosclerotic plaque size in the aortic sinus of the two groups of mice at the end of the experiment is shown. n = 12 (cWD) and n = 14 iWD. h, Ldlr−/−Rag2−/− male mice on a normal diet were put on a cWD versus iWD diet for 6 weeks. The mean atherosclerotic plaque size in the aortic sinus of the two groups of mice at the end of the experiment is shown. n = 8 (cWD) and n = 7 (iWD). Data are mean ± s.e.m. Statistical analysis was performed using two-tailed unpaired t-tests (d and f–h). P values are indicated on the graphs.
Source Data
In a first series of experiments, we subjected LDL-receptor-deficient (Ldlr−/−) male mice to 6 weeks of late continuous Western-type cholesterol-rich diet (cWD) versus 6 weeks of early intermittent WD (iWD) (Fig. 1a and Extended Data Fig. 1a). We measured the plasma cholesterol levels repetitively during the experiment (Extended Data Fig. 1b–d) and calculated the averaged cumulated plasma cholesterol levels over the duration of the experiment (Fig. 1b). The overall cholesterol load (cholesterol × time) was similar between the two groups of mice. The weight, heart rate, blood pressure and plasma corticosterone levels were also similar between the two groups (Extended Data Fig. 1e–h). On the basis of the paradigm that cumulative exposure to cholesterol is the major determinant of atherosclerosis, we should have detected atherosclerotic lesions of similar sizes in the two groups of mice. However, we were surprised by the substantial increase of plaque size in mice subjected to the iWD compared with the cWD (Fig. 1c,d). The results were reproducible in female mice (Extended Data Fig. 2a,b), although the fold increase (1.3 to 2.5) in lesion size was more variable than in male mice. Analysis of the lesion composition (Extended Data Fig. 2c–g) revealed a pro-inflammatory phenotype with increased accumulation of MOMA2+ macrophages and CD3+ T cells, and large necrotic cores. Despite an increase in the αSMA+ area, suggesting higher smooth muscle cell plaque content in the iWD group, collagen accumulation as measured by Sirius Red staining was not different between the two groups, suggesting altered plaque healing in iWD mice (Extended Data Fig. 2f,g). We tested several periods of iWD and cWD and always obtained an acceleration of atherosclerosis in iWD mice (Extended Data Fig. 3a,b). Ldlr−/− mice subjected to 12 weeks of iWD still showed substantially larger atherosclerotic plaques compared with mice that were subjected to 12 weeks of late cWD (Fig. 1e,f), with very large necrotic cores associated with defective efferocytosis and increased accumulation of apoptotic debris (Extended Data Fig. 3c–j). The lesion size tended to be smaller (although not statistically different) when iWD was started at 22 weeks of age compared with at 6 weeks of age (Extended Data Fig. 3k–l). In summary, early exposure to WD is substantially more pro-atherogenic than late exposure.
WD feeding affects the composition of the gut microbiota, which in turn may affect the development of atherosclerosis11. We assessed the gut microbiota composition using 16S rRNA sequencing of the faeces after both 3 and 6 weeks of iWD and cWD (Extended Data Fig. 4a). The gut microbiota composition was similar between the iWD and cWD groups after 3 weeks of WD, but differed slightly between the two groups after 6 weeks (Extended Data Fig. 4a–c). We therefore designed an experiment to deplete the gut microbiota with oral antibiotics during the last 3 weeks of iWD (Extended Data Fig. 4d). Antibiotic treatment significantly depleted the gut microbiota (Extended Data Fig. 4e–g). We found that atherosclerosis was still accelerated in iWD versus cWD under antibiotics (Fig. 1g), although to a lesser degree than in the absence of antibiotics, which may in part be due to lower levels of plasma cholesterol in the iWD group compared with the cWD group under antibiotics (Extended Data Fig. 4h). Together, our results suggest a limited effect of gut microbiota on iWD-induced acceleration of atherosclerosis.
We also examined the role of adaptive immunity and found that iWD-induced acceleration of atherosclerosis still occurred in Ldlr−/−Rag2−/− mice (deficient in T and B cells) (Fig. 1h), despite similar circulating cholesterol levels (Extended Data Fig. 4i). Although the roles of specific T and B cell subsets require additional studies, our results suggest a critical role of non-adaptive immune responses in this process.
iWD reshapes arterial macrophages
We conducted a series of additional experiments and found that the atherosclerotic lesion size was not significantly different between the iWD and cWD groups after 3 weeks of WD (Extended Data Fig. 5a,b). We therefore selected this timepoint to perform RNA-sequencing (RNA-seq) analysis of aortae from both groups of mice with the aim of identifying causal processes involved in the subsequent acceleration of atherosclerosis in iWD mice. We focussed on macrophages, given the prominent role of these innate immune cells in atherosclerosis12. We conducted two separate experiments with nine mice per group and analysed the data together with a total of six separate pools (three mice each) per group. In total, 746 genes were differentially expressed (DEGs) in aortic macrophages between iWD- and cWD-fed mice (Fig. 2a,b and Supplementary Table 1). Gene Ontology (GO) term enrichment identified reduced (macro)autophagy as a potential major causal process (Fig. 2a and Supplementary Table 2), which fits with the known role of autophagy in limiting plaque inflammation, necrosis and progression in mice13. Notably, this defective autophagy pathway includes several autophagy-related genes (for example, Atg12 and Wdfy4, which operate in the Wdfy3 pathway14, and Vps33a; Fig. 2b and Supplementary Table 1) recently identified as top-ranked regulators of macrophage efferocytosis14. This is consistent with the increased accumulation of apoptotic debris and the larger necrotic cores in plaques of iWD-fed mice (Extended Data Fig. 3), and the known role of impaired macrophage efferocytosis in promoting the formation of large necrotic and inflammatory atherosclerotic lesions15.
Early intermittent hyperlipidaemia alters resident-like arterial macrophages.
Ldlr−/− male mice were put on 3 weeks of cWD versus 3 weeks of iWD (Extended Data Fig. 5a,b). The aortas (n = 6 separated pools of 3 mice each per group) were collected and digested for flow cytometry cell sorting of aortic macrophages and then analysed using bulk RNA-seq (Methods). a, Ten selected GO pathways (biological process) corresponding to the 746 significant DEGs between the iWD and cWD groups (the full list is shown in Supplementary Table 2). Blue and orange colour represents overall gene direction. Grey shading shows the adjusted P value (Padj), calculated using one-sided Fisher exact tests followed by adjustment using the Benjamini–Hochberg procedure. b, Volcano plot for DEGs; yellow/cyan colour highlights upregulated genes in iWD/cWD (Padj < 0.05; two-sided Wald test with Benjamini–Hochberg adjustment) and |log2[fold change]| ≥ 1, and grey colour displays non-significant DEGs. Some significant DEGs related to resident/TLF-like (red) or MacAIR-like (violet) macrophages are highlighted (a full list is provided in Supplementary Table 1). c–e, Apoe−/− mice at 6 weeks of age were put on a cWD diet for 8, 10 or 18 weeks to develop early-stage, intermediate or advanced atherosclerosis. Atherosclerotic lesions of the innominate and brachiocephalic arteries were stained with DAPI (nuclei), and antibodies against LYVE1, CD68 and smooth muscle alpha-actin (SMA). c, Representative photomicrographs are shown. Scale bars, 100 μm. d,e, Quantification (mean ± s.e.m.) of adventitial (adv.) LYVE1+CD68+ macrophages (n = 4 per group; two-tailed Mann–Whitney U-test) (d) and correlation with the atherosclerotic plaque size (n = 13) (e). R is the Spearman correlation coefficient (two-tailed). f,g. Tissue sections of healthy (n = 7 individuals) and atherosclerotic (athero.; n = 9 individuals) (Methods) arteries were stained with DAPI (nuclei) and antibodies against LYVE1 and CD68. f, Representative photomicrographs. Scale bars, 30 μm. g, The percentages of adventitial LYVE1+CD68+ macrophages among adventitial CD68+ macrophages. Statistical analysis was performed using two-tailed unpaired t-tests.
Source Data
iWD alters resident-like macrophages
Notably, we found that most of the highly significant macrophage DEGs (Fig. 2b and Supplementary Table 1) were enriched in resident-like macrophages, for example, aortic-intima-resident-like (MacAIR) macrophages, adventitial and intimal TLF-like (expressing TIMD4 and/or LYVE1 and/or FOLR2) macrophages, and were downregulated under a iWD versus a cWD (Fig. 2b and Supplementary Table 3). This is typically the case of Cadm1, which is enriched in MacAIR-like macrophages16, and Phactr1, which is enriched in resident-like TLF+ aortic macrophages17.
To examine whether downregulation of the resident-macrophage-associated genes was due to a loss or a phenotypic alteration of resident arterial macrophages, we fate-mapped resident arterial macrophages using Cx3cr1creERT2/+Rosa26LSL-tdTomatoLdlr−/− mice16. After treatment with tamoxifen to induce tdTomato expression (in resident macrophages and blood monocytes), the mice were rested for 2 weeks on a normal diet to allow for clearance of labelled blood monocytes. The mice were then placed on 3 weeks of cWD versus 3 weeks of iWD (Extended Data Fig. 5c). The total numbers of intimal macrophages (CD45+CD68+) were similar between the cWD and iWD groups (Extended Data Fig. 5d,e), consistent with the comparable plaque size between the two groups at this timepoint (Extended Data Fig. 5a,b), but the percentage and the number of tdTomato+ intimal MacAIR macrophages was significantly reduced in iWD mice (Extended Data Fig. 5f,g). Thus, there is an accelerated loss of MacAIR macrophages under an iWD, which is compensated by an increased accumulation of unlabelled monocyte-derived macrophages. These data also indicate that the accelerated depletion of MacAIR macrophages may in part be responsible for the reduced expression of MacAIR-associated genes in the arterial macrophage dataset. By contrast, the number of adventitial tdTomato+ macrophages was similar between the two groups (Extended Data Fig. 5h), suggesting a predominantly phenotypic alteration of these macrophages at this early timepoint.
To gain further insights into the regulatory mechanisms that may drive the change in macrophage gene expression in the aortas of iWD versus cWD mice, we performed transcription factor (TF) enrichment analysis and found significant enrichment of binding motifs for 27 TFs in aortic macrophage DEGs compared with non-DEGs (Extended Data Fig. 6a and Supplementary Table 4). Most of these TFs are members of the ETS TF family and have recently been shown to affect gene expression within expression quantitative trait loci in humans, and to drive the regulatory effects of inherited human genetic variation18. The most enriched ETS TF member was SPIB (Extended Data Fig. 6a), previously identified as a multitissue key driver of a network of genes with causal relationship to coronary heart disease19. SPIB is not expressed in myeloid cells but its highly related TF, SPIC, which shares the same DNA-binding specificity, is expressed in macrophages and was one of the most enriched TFs in the aortic macrophage DEGs (Extended Data Fig. 6a). Notably, SPIC was recently identified in a single-cell-based integrated analysis of mononuclear phagocytes of mouse and human atherosclerosis as one of top-3-ranked TFs for mouse TLF+ and MacAIR macrophages, with notable activity in human LYVE1+ macrophages20. Pathway analysis of our data showed that aortic macrophage DEGs enriched for SPIC binding motifs mostly associate with resident-like macrophages and are involved in biological processes related to autophagy, ribosome biogenesis, RNA processing and actin cytoskeleton reorganization (Extended Data Fig. 6b and Supplementary Table 5), recapitulating the GO term enrichment of the whole set of aortic macrophage DEGs (Fig. 2a and Supplementary Table 2). We therefore generated Lyz2cre+/−Spicflox/flox mice (and Lyz2cre+/−Spicflox/+ controls) to assess the role of myeloid-specific expression of SPIC on atherogenesis. We found that reconstitution of Ldlr−/− mice with Lyz2cre+/−Spicflox/flox bone marrow substantially accelerated atherosclerosis in response to cWD compared with reconstitution with Lyz2cre+/−Spicflox/+ control bone marrow (Extended Data Fig. 6c,d), despite similar circulating cholesterol levels (Extended Data Fig. 6e). However, iWD-induced acceleration of atherosclerosis was limited in mice with myeloid-specific deletion of Spic (Extended Data Fig. 6f). Thus, macrophage expression of SPIC, a TF that was previously identified as a potential regulator of resident-like macrophages20 and of which the binding motifs are enriched in the arterial macrophage DEGs that are downregulated in iWD mice (Extended Data Fig. 6a and Supplementary Table 4), protects against the development of atherosclerosis. Taken together, our data suggest that alteration of resident-like arterial macrophages is involved in the acceleration of atherosclerosis in response to early intermittent hyperlipidaemia.
Effect of LYVE1<sup>+</sup> macrophage deletion
To examine the specific role of resident arterial macrophages in atherosclerosis, we focussed on Lyve1-expressing macrophages, which reside in the adventitia of healthy arteries21 and are part of intimal resident-like TLF macrophages in atherosclerotic lesions22, although intimal TLF macrophages often do not express LYVE1 protein. Three weeks of iWD did not affect the accumulation of resident adventitial macrophages compared with the cWD (Extended Data Fig. 5h), but it profoundly reduced the expression of a substantial number of their prototypical genes (Fig. 2b and Supplementary Table 3). We reasoned that such a profound alteration of gene expression may potentially compromise macrophage survival in the longer term. We therefore assessed the accumulation of LYVE1+ macrophages at various stages of atherosclerosis in mice (Fig. 2c). We detected a significant progressive reduction in aortic LYVE1+ macrophages over time, which was strongly and inversely correlated with the extent of atherosclerotic plaques (Fig. 2d,e). This is consistent with the results of our reanalysis of the single-cell RNA-seq (scRNA-seq) data of published mouse atherosclerosis studies16,22–24, which shows that prototypical genes associated with arterial resident-like macrophages, including Lyve1 and Nrp1, are progressively downregulated with increasing durations on a high-fat diet (Extended Data Fig. 7a,b). We also analysed human healthy and atherosclerotic coronary arteries and found a significant reduction in the accumulation of resident-like LYVE1+ macrophages in atherosclerotic compared with healthy human arteries (Fig. 2f,g), supporting the clinical relevance of our findings.
To directly address the role of Lyve1-expressing macrophages in the development of atherosclerosis, we generated Lyve1cre+/WTCsf1rflox/flox mice (and Csf1rflox/flox controls) on an apolipoprotein-E-deficient (Apoe−/−) background to selectively ablate LYVE1+ macrophages21 and assessed the effect of their genetic deletion on atherosclerosis. LYVE1+ macrophages were substantially reduced in Apoe−/−Lyve1cre+/WTCsf1rflox/flox mice compared with in Apoe−/−Csf1rflox/flox mice (Extended Data Fig. 8a,b), and this was associated with a significant increase in atherosclerotic plaque size (Fig. 3a,b), despite similar plasma cholesterol levels (Extended Data Fig. 8c). Analysis of plaque composition revealed comparable smooth muscle cell and collagen content (Extended Data Fig. 8d,e). However, we found increased accumulation of intimal CD68+ macrophages (Fig. 3c), and increased necrotic core area (Fig. 3d), which was consistent with impaired efferocytosis within the plaques of Apoe−/−Lyve1cre+/WTCsf1rflox/flox mice (Extended Data Fig. 8f,g). We also used another model of hyperlipidaemia. Csf1rflox/flox and Lyve1cre+/WTCsf1rflox/flox mice were injected with AAV8-D377Y-mPCSK9 and subjected to 6 weeks of cWD or 6 weeks of iWD (Extended Data Fig. 8h,i). We found that, in control Csf1rflox/flox mice, the iWD led to a significant increase in plaque size and necrotic core area compared with the cWD (Fig. 3e–h). However, these iWD-dependent effects were blunted in mice with deletion of LYVE1+ macrophages (Fig. 3i–l). Our data indicate that resident-like arterial macrophages are altered during the progression of atherosclerosis, and their alteration further exacerbates disease progression.
Deletion of LYVE1<sup>+</sup> resident macrophages accelerates atherosclerosis.
a–d, Apoe−/−Lyve1cre+/−Csf1rflox/flox mice (n = 7) and Apoe−/−Csf1rflox/flox littermate controls (n = 7) were put on a cWD diet for 20 weeks. a, Representative photomicrographs of aortic sinus atherosclerotic plaques from the two groups of mice are shown, stained with DAPI and antibodies against LYVE1, CD68 and SMA. b, The mean atherosclerotic plaque size in the aortic sinus of the two groups of mice. c, Quantification of the CD68+ area in aortic sinus lesions of the two groups of mice. d, The percentage of necrotic area (Methods) in the aortic sinus lesions of the two groups of mice. e–l, Csf1rflox/flox (e–h) and Lyve1cre+/−Csf1rflox/flox (i–l) female mice were injected with AAV8-D377Y-mPCSK9 to induce hyperlipidaemia. The mice were then subjected to 6 weeks of cWD or 6 weeks of iWD. Csf1rflox/flox: n = 8 (cWD) and n = 9 (iWD) mice; Lyve1cre+/−Csf1rflox/flox: n = 11 (cWD) and n = 11 (iWD) mice (Methods and Extended Data Fig. 8h). f,g,j,k, Quantification of the atherosclerotic lesion size (f and j) and necrotic core area (g and k) in the aortic root. h,l, Plasma cholesterol levels at the end of the experiment. Data are mean ± s.e.m. Statistical analysis was performed using two-tailed Mann–Whitney U-tests (b–d,g,h and j–l). P values are indicated on the graphs. For a,e,i, scale bars, 200 μm.
Source Data
iWD alters human ASCVD pathways
To further understand the mechanisms that are responsible for iWD-dependent acceleration of atherosclerosis, we assessed the relevance of our findings to the mechanisms of human ASCVD. We found that macrophage DEGs (Fig. 2b and Supplementary Table 1) were enriched in genes shown (or predicted) to be causally involved in human ASCVD in genome-wide association studies (GWAS) (Fig. 4a and Supplementary Table 6). Among the DEGs that could be assigned to a specific macrophage subset, a large proportion belonged to MACAIR/TREM2-like or resident-like TLF+ macrophages (Fig. 4b and Supplementary Table 7). This is the case for Phactr1 (Fig. 2b), the deficiency of which in macrophages accelerates atherosclerotic plaque necrosis in mice17, and the expression of which in human arteries is controlled by a genetic variant that is causally linked to human ASCVD25–27 and angiographically determined burden of coronary atherosclerosis28.
Alteration of ASCVD pathways in resident arterial macrophages accelerates atherosclerosis.
a, The percentage of genes with link to coronary artery disease (CAD) in GWAS for significant (Padj ≤ 0.05 and |log2[fold change]| ≥ 1) DEGs (209 out of 746) and non-significant (Padj > 0.1 and |log2[fold change]| < 1) DEGs (831 out of 3,461); Padj values were calculated using two-sided two-proportions z-tests. In total, 64 out of 209 DEGs linked to CAD-GWAS could be assigned to three types of arterial macrophages (the gene list is provided in Supplementary Table 7). b, The distribution of 64 CAD-GWAS-related DEGs among macrophage subtypes. c, Volcano plot for CAD-GWAS-associated macrophage DEGs that are differentially expressed in symptomatic (sym.) versus asymptomatic (asym.) human carotid atherosclerotic plaques, obtained after reanalysis of available scRNA-seq data29. Yellow/cyan colour highlights upregulated genes in symptomatic/asymptomatic plaques. Significant DEGs related to resident/TLF-like (red), MacAIR/TREM2-like (violet) and IFN/inflammatory-like (IFN/inflam.; grey) macrophages are highlighted. P values were calculated using two-sided Wald tests followed by adjustment using Benjamini–Hochberg correction. d,e Dot plots of macrophage markers from the meta-analysis of scRNA-seq data of mouse (d) and human (e) atherosclerotic plaques. NRP1 is shown for comparison. Res, resident. f,h, Selected GO (biological process) pathways for significant DEGs enriched in human NRP1high versus NRP1low (f) and mouse Nrp1+/+ versus Nrp1−/− (h) macrophages. Gradient red colour represents Padj values calculated using one-side Fisher’s exact tests followed by adjustment with Benjamini–Hochberg correction. The bar size represents the gene count in each pathway. g, Ldlr−/− male mice were lethally irradiated and reconstituted with bone marrow (BM) from Lyz2cre+/−Nrp1flox/flox mice and Lyz2cre+/−Nrp1+/+ controls. Mice were put on a cWD (n = 14 (Lyz2cre+/−Nrp1flox/flox) and n = 14 (Lyz2cre+/−Nrp1+/+)) or iWD (n = 9 (Lyz2cre+/−Nrp1flox/flox) and n = 7 (Lyz2cre+/−Nrp1+/+)) for 6 weeks (Methods). The mean atherosclerotic plaque size in the aortic sinus is shown. Data are mean ± s.e.m. Statistical analysis was performed using two-tailed unpaired t-tests. P values are indicated on the graphs.
Source Data
To identify new targets involved in iWD-dependent acceleration of atherosclerosis of relevance to human ASCVD, we analysed publicly available scRNA-seq data29 and found that 22 out of the GWAS-ASCVD-associated DEGs showed significant differential expression in macrophages of symptomatic compared with asymptomatic carotid atherosclerotic plaques (Fig. 4c and Supplementary Table 8). Here we focussed on Nrp1 for several additional reasons. We were interested in genes that operate in the same biological pathway as Phactr1, and our data indicate that the latter is part of an actin filament organization pathway that shows differential enrichment in arterial macrophages of cWD compared with iWD mice (Extended Data Fig. 6b and Supplementary Table 5). Notably, this pathway includes other DEGs that are involved in actin polymerization (such as Actr2 and Cyfip2) and were recently shown, like PHACTR1, to regulate macrophage efferocytosis14 and to be linked to ASCVD in GWAS28. Notably, Nrp1 is one of the actin filament organization DEGs30,31 (Supplementary Table 5), and previous work suggested that Phactr1 may be induced downstream of NRP1 signalling32. Furthermore, our meta-analysis of the available scRNA-seq data on arterial macrophages shows that NRP1 is preferentially expressed in resident-like TLF+ macrophages both in mice (Fig. 4d) and in humans (Fig. 4e), with NRP1high macrophages being enriched in biological pathways related to cellular migration and actin cytoskeleton reorganization (Fig. 4f and Supplementary Tables 9 and 10). Other biological pathways of importance to atherosclerosis, for example, receptor-mediated endocytosis, regulation of leukocyte chemotaxis, apoptotic processes and complement activation, were also enriched in NRP1high macrophages (Fig. 4f and Supplementary Tables 9 and 10).
To further assess the clinical relevance of the actin filament organization pathway to the causal pathways leading to ASCVD in humans, we focussed on the results of a recent study in which the investigators systematically characterized risk variants and genes for coronary artery disease in over a million participants33. Combining several complementary approaches, 220 candidate causal genes were prioritized. We applied GO analysis to those 220 genes and found that the most represented biological pathways were pathways related to lipid metabolism, cell migration and vasculature development (Supplementary Table 11 and Extended Data Fig. 9). Genes related to actin filament organization, including NRP1, were over-represented in the pathways related to cell migration, further highlighting the relevance of our findings obtained in mice to the pathophysiology of ASCVD in humans. NRP1 was also represented in pathways related to extrinsic apoptotic signalling and apoptotic cell clearance (Supplementary Table 11).
Together, our data point to actin filament organization and NRP1 expression in macrophages as potentially important targets, the alteration of which may be causally linked to iWD-dependent acceleration of atherosclerosis. However, the role of myeloid NRP1 in atherosclerosis is unclear.
Effect of <italic>Nrp1</italic> deletion in macrophages
A subset of lymphatic and blood endothelial cells express both LYVE1 and NRP1, precluding the use of Lyve1cre+/− mice to address the role of NRP1 in resident-like macrophages. Lyz2cre+/− mice have previously been used successfully to delete Nrp1 in myeloid cells and tissue resident-like macrophages34. We therefore generated Lyz2cre+/−Nrp1flox/flox mice (and Lyz2cre+/−Nrp1+/+ controls) and performed a bone marrow transplant experiment into Ldlr−/− mice to assess the role of myeloid-specific expression of NRP1 on atherogenesis (Extended Data Fig. 10a). We successfully reconstituted Ldlr−/− mice with NRP1-deficient and NRP1-sufficient macrophages (Extended Data Fig. 10b). iWD accelerated the development of atherosclerosis compared with cWD in Ldlr−/− mice with Lyz2cre+/−Nrp1+/+ bone marrow (Fig. 4g). Reconstitution of Ldlr−/− mice with Lyz2cre+/−Nrp1flox/flox bone marrow increased the development of atherosclerosis under the cWD compared with reconstitution with Lyz2cre+/−Nrp1+/+ bone marrow (Fig. 4g), independently of changes in circulating cholesterol levels (Extended Data Fig. 10c). By contrast, deletion of Nrp1 in myeloid cells had no effect on the development of atherosclerosis under iWD (Fig. 4g). Our findings suggest that the protective effect of NRP1-expressing macrophages in cWD is not maintained under iWD, consistent with the substantial downregulation of Nrp1 expression in aortic macrophages under iWD (Fig. 2b and Supplementary Table 1). Bulk RNA-seq analysis of NRP1-deficient and NRP1-sufficient peritoneal macrophages from reconstituted Ldlr−/− mice showed that Nrp1 deletion altered macrophage phenotype and affected biological pathways related to actin-filament-based processes, cytoskeleton organization, cell adhesion and apoptosis (Fig. 4h and Supplementary Tables 12 and 13). Consistently, we show that Nrp1 deletion in macrophages reduces efferocytosis (Extended Data Fig. 10d). Taken together, our data implicate NRP1 as part of a biological pathway that affects the arterial macrophage cytoskeleton and becomes altered in response to early intermittent hyperlipidaemia, leading to acceleration of atherosclerosis.
Lifetime cholesterol and carotid plaque
Our results led us to hypothesize that exposure to cholesterol early in life (childhood, adolescence and young adulthood) may be a strong determinant of atherosclerosis at mid-adulthood, independently of cumulative lifetime exposure. We therefore analysed data from the Cardiovascular Risk in Young Finns Study (YFS)35, which is a prospective cohort study that began in 1980 when the participants were aged 3–18 years. Follow-up clinics were conducted regularly thereafter. We performed analyses on two subsets of participants in the YFS when carotid ultrasound studies were performed: 2,653 participants who attended either the 2001 or 2007 follow-ups when aged 24–45 years, and 2,062 participants who attended in 2018 when aged 41–56 years. The participant characteristics are shown in Supplementary Tables 14 and 15. Carotid plaques (n = 88 at the 2001/2007 visits, and n = 817 at the 2018 visit) were defined as protruding focal >50% arterial wall thickenings36. The associations of non-high-density lipoprotein cholesterol (non-HDL-C) from three early life periods (childhood, adolescence, young adulthood) with the relative risk of carotid plaque (dichotomous outcome, no/yes) and carotid plaque area (continuous outcome) in mid-adulthood are shown in Supplementary Tables 16 and 17 for the 2001/2007 follow-ups and in Table 1 and Supplementary Table 18 for the 2018 follow-up. Data stratification by sex for carotid plaque outcomes is shown in Supplementary Tables 16 and 17 for the 2001/2007 follow-ups and in Supplementary Tables 19–22 for the 2018 follow-up. The lifetime effect indicated that higher non-HDL-C accumulated across the observed life stages was associated with higher risk of plaque presence (Table 1 and Supplementary Table 16), and higher plaque area (Table 1 and Supplementary Table 17). These findings were consistent in minimally and fully adjusted analyses, including sex, year of birth and potential confounding variables. The results also remained consistent after imputation for missing values in the adjusted model (Supplementary Table 18). Data stratification by sex suggested that the associations were slightly stronger for male than for female participants (Supplementary Tables 16, 17 and 19–22). For each model, the Euclidian distance generally indicated that the accumulation life course model best represented the association with mid-adulthood plaque presence and plaque area (Supplementary Tables 16, 17 and 19–22, Supplementary Methods and Supplementary Data). This was despite some variation in the relative contributions of non-HDL-C levels at each life stage, as shown in Table 1, Supplementary Tables 16–22 and accompanying ternary plots (Supplementary Methods and Supplementary Data), noting that the credible intervals for each life stage were wide and overlapped. Although there was some suggestion that exposure to non-HDL-C at adolescence and young adulthood was contributing most to the observed lifetime effects, exposure to non-HDL-C levels before adulthood (that is, childhood and adolescence) consistently contributed more than half of the lifetime effect to plaque outcomes in mid-adulthood. This suggests that exposure to cholesterol before adulthood contributes to the development of atherosclerosis in mid-adulthood.
Association of non-HDL-C with plaque presence and plaque area
Carotid plaque
log-transformed carotid plaque area (mm2)
Model 1 (n = 2,062)
Model 2 (n = 1,761)
Model 1 (n = 817)
Model 2 (n = 689)
RR
95% CrI
Relative weight (%)
95% CrI
RR
95% CrI
Relative weight (%)
95% CrI
β
95% CrI
Relative weight (%)
95% CrI
β
95% CrI
Relative weight (%)
95% CrI
Lifetime effect
1.22
1.14–1.30
1.23
1.14–1.33
0.27
0.22, 0.33
0.27
0.21, 0.33
Life-period, age
Childhood, 6 to 12 years
1.04
1.001–1.14
21.4
0.7–64.1
1.06
1.002–1.17
27.0
1.0–73.8
0.06
0.002–0.16
20.6
0.8–57.6
0.06
0.002–0.17
21.7
0.8–60.3
Adolescence, 15 to 18 years
1.07
1.004–1.19
35.5
1.9–82.4
1.07
1.003–1.19
34.2
1.7–81.4
0.16
0.04–0.25
57.4
17.1–90.4
0.14
0.03–0.25
51.9
10.7–88.0
Young adulthood, 21 to 24 years
1.09
1.01–1.20
43.1
3.4–87.1
1.09
1.005–1.20
38.8
2.4–84.3
0.06
0.003–0.15
21.9
1.2–55.1
0.07
0.004–0.17
26.5
1.6–62.2
Life-course modela
Accumulation
Accumulation
Accumulation
Accumulation
Association of non-HDL-C in three life-periods (childhood, adolescence, young adulthood) with plaque (dichotomous, no/yes) and plaque area (continuous measure) among those with plaque in mid-adulthood (in 2018) aged 41 to 56 years. β, relative risk (RR) and relative weights are shown for a 1 s.d. increase in non-HDL-C. Model 1 was adjusted for sex and year of birth in all participants. Model 2 adjusted for sex, year of birth, area under the curve between ages 6 and 24 years for body mass index, HDL-C and systolic blood pressure, area under the curve between ages 9 and 24 years for blood glucose and physical activity index, education (years studied), ever smoked daily before age 24 years, family history of cardiovascular disease in all participants. CrI, credible interval.
aIndicated by Euclidian distance.
Discussion
Hyperlipidaemia is a major causal risk factor for atherosclerosis. We show that early intermittent hyperlipidaemia is a major determinant of atherosclerotic plaque size compared with late hyperlipidaemia, even when adjusted for cumulative plasma cholesterol levels over time. Atherosclerosis is known as an age-associated arterial disease. Our results show that events (here, high circulating lipid levels) occurring early in life substantially affect the subsequent development and progression of atherosclerosis. Our observation may explain, at least in part, the association between early (and periodic) time course of cholesterol elevation and subsequent cardiovascular events in adulthood1–3. Our results may also be relevant to the increased risk of cardiovascular events after statin discontinuation and re-exposure to high cholesterol levels37.
Understanding why atherosclerosis is accelerated in response to early and intermittent hyperlipidaemia may lead to the identification of biological mechanisms that are involved in disease development and progression. As a first step, we show that early intermittent hyperlipidaemia impairs critical autophagy- and efferocytosis-related pathways in arterial macrophages with causal links to atherosclerosis in mice and GWAS ASCVD in humans. We also show that most of these pathways are enriched in resident-like arterial macrophages, and that early intermittent hyperlipidaemia accelerates atherosclerosis, at least in part through the loss of the atheroprotective properties of TLF+ macrophages. We identify biological pathways in arterial macrophages, in particular genes that are involved in actin cytoskeleton organization such as NRP1, that contribute to atheroprotection, in part through the promotion of efferocytosis. Other relevant genes that we identified in these pathways include PHACTR1, which has causal links to ASCVD in humans25–27 and has previously been shown to limit atherosclerosis through improved efferocytosis17.
Our results indicate that resident arterial macrophages become altered very early during the development of atherosclerosis. Intermittent hyperlipidaemia prevents the newly recruited monocyte-derived macrophages from acquiring atheroprotective tissue-residency characteristics, leading to substantial disease acceleration. Additional work will be required to assess the differential roles of monocyte-, yolk-sac- or erythromyeloid-progenitor-derived resident macrophages in this process. Nevertheless, our results are in agreement with the concept that the balance of homeostatic versus disease-associated signals is a major determinant of the local diversity of macrophage cell states and, therefore, tissue homeostasis38.
We found that intermittent hyperlipidaemia can still accelerate atherosclerosis after extended durations of a high-fat diet, long after the early alteration of resident macrophages. This suggests that alteration of resident macrophages sets the stage for further acceleration of atherosclerosis by additional mechanisms, which become revealed by the absence of resident macrophages. This hypothesis merits further investigation.
Our study has limitations. Cx3cr1creERT2/+ mice could not be used to assess the role of resident macrophages in atherosclerosis because Cx3cr1 heterozygosity affects the development of atherosclerosis39. Mice were lethally irradiated before reconstitution with the bone marrow of interest to address the role of macrophage SPIC and NRP1 expression. Resident arterial macrophages are ablated by lethal irradiation but most (approximately 80%) of them reconstitute from the donor bone marrow40. Besides macrophages, the use of Lyz2cre+/− also affects gene expression in neutrophils. However, compared to macrophages, Nrp1 and Spic show very low or no expression, respectively, in blood neutrophils. Thus, the results using Lyz2cre+/−Nrp1flox/flox and Lyz2cre+/−Spicflox/flox mice can mostly be attributed to the roles of NRP1 and SPIC in macrophages. Note also that efferocytosis was not directly assessed in plaque-derived macrophages. Finally, although the human data show that hyperlipidaemia early in life determines future atherosclerotic plaque development, these data are observational and are subject to bias and confounding, and additional studies will be required to address the effect of cholesterol variability in childhood or adolescence on plaque development and progression.
In summary, we provide evidence that exposure to high circulating lipid levels in early life is a major determinant of atherosclerosis development and progression in adulthood. This finding has the potential to radically change our understanding of the mechanisms of accelerated atherosclerosis. We unravel major homeostatic roles of resident arterial macrophages in limiting the development of atherosclerosis, and show that early intermittent hyperlipidaemia alters these homeostatic mechanisms to accelerate atherosclerosis. Our results have implications regarding the optimal timing of initiation of lipid-lowering therapies41, the earlier being the better42, and the need for sustained lifelong control of circulating lipid levels43,44.
MethodsMice
All in vivo experiments using mice were approved by local Institutional Review Boards: the Home Office, UK, under PPL PA4BDF775; the Ethical Committee of INSERM (APAFIS, 29371) and the French Ministry of Agriculture (MESRI 674 29371); the Institutional Animal Care and Use Committee (IACUC) at National University of Singapore (protocol number R21-1562); and under animal protocol 2111–39587A for experiments performed in the United States. All mice were on the C57BL/6 background. Both males and females were used. Atherosclerosis experiments were started at the age of 6 weeks unless otherwise stated. Lyz2cre+/−, Ldlr−/− and Rag2−/− mice were initially from Charles River. Spicflox/flox mice were from T. Kaisho and W. Ise45. Lyz2cre+/−Nrp1flox/flox and Lyz2cre+/−Nrp1+/+ mice were generated by the group of Christiana Ruhrberg, University College London, UK. Cx3cr1creERT2/+Rosa26LSL-TomatoLdlr−/− mice were generated by J. Williams. Apoe−/− mice were purchased from Comparative Medicine (National University of Singapore). Lyve1cre/cre and Csf1rflox/flox mice were purchased from Jackson Laboratory. Apoe−/−Lyve1cre/WTCsf1rflox/flox mice were generated by crossbreeding Lyve1cre/cre with Apoe−/−Csf1rflox/flox mice. In some experiments, 8–10-week-old Lyve1cre+/WTCsf1rflox/flox and Csf1rflox/flox female mice were injected intraperitoneally with 0.5 × 1012 viral genome copies of AAV8-D377Y-mPCSK9 virus per mouse to induce hypercholesterolemia. AAV8-D377Y-mPCSK9 viruses were purchased from and generated by Phoenix Laboratory of Gene Therapy and Cell Therapy (National University of Singapore). Mice were randomly assigned to their respective experimental groups. Sample size was determined to detect a significant difference (α = 0.05 and 80% power) of at least 30% in lesion size between groups. Analyses were performed in a blinded manner. Main atherosclerosis experiments were repeated several times and showed consistent results (see the main text). All animals were maintained in pathogen-free animal facilities and were handled in accordance with protocols approved by local institutional animal care and use committees. Mice were housed under a 12 h–12 h dark–light cycle, at ambient temperature of 22 °C and 45–65% humidity. No wild animals and no field-collected samples were used in the study.
Bone marrow transplants
Ldlr−/− mice were irradiated with two doses of 5.5 Gy (separated by 4 h) followed by reconstitution with 0.5–1 × 107 sex-matched donor bone marrow cells. Animals were randomly assigned to receive the appropriate bone marrow. Mice were then maintained for a 4-week recovery period before being continued on normal or WD (21% fat, 0.15% cholesterol).
Measurement of plasma cholesterol levels
Blood was obtained from through the saphenous vein (during the experimental period) or the vena cava (at euthanasia) on EDTA. Plasma was separated by centrifugation at 12,000 rpm and kept at −80 °C. Lipids (and corticosterone) were measured by the Core Biochemical Assay laboratory at the University of Cambridge and, in some experiments, using a commercially available cholesterol colorimetric assay kit (BioVision).
Histology, immunohistochemistry and morphometry of mouse atherosclerotic lesions
Tissues were collected into 4% PFA overnight before washing into PBS. Aortic root and innominate artery cryosections were stained with Oil Red O (Sigma-Aldrich) for atherosclerotic lesion size, Picrosirius Red (Sigma-Aldrich) for collagen content and haematoxylin and eosin and DAPI for acellular debris (necrosis). For immunofluorescence staining, cryosections (kept at −80 °C) were dried for 1 h followed by rehydration in PBS for 10 min before the staining. The samples were then permeabilized in 0.1% Triton X-100, 0.1% citrate in PBS for 30 min. The samples were then washed in PBS and incubated with the blocking solution (staining solution (PBS, 1% BSA, 2 mM EDTA, 0.01 % NaN3) + 5% serum of secondary antibody species) for 30 min followed by incubation with primary antibodies diluted in the blocking solution overnight at 4 °C. The samples were washed with PBS and incubated with secondary antibodies diluted in the staining solution for 3 h. The samples were washed in PBS and nuclei was counterstained with DAPI and mounted with CC mount (Sigma-Aldrich). Quantification of lesion size and composition was performed using ImageJ (National Institutes of Health) according to accepted guidelines46. The LYVE1+ macrophage number in aortic sections was quantified using colocalization of pan-macrophage marker CD68 or the tissue-resident M2 macrophage markers CD206 with LYVE1. We have previously shown that the majority of adventitial resident LYVE1+ macrophages also co-express CD20621. The adventitial area of the innominate artery section was demarcated as the area between GFP autofluorescence of the external elastic laminae and the border of the periadventitial fat. The extent of plaque spread was determined by the measurement of external elastic laminae length beneath the plaque. Necrotic area was determined by demarcating DAPI-negative regions within plaques of the aortic sinus sections. Data analysis was conducted in a blind manner where appropriate. Primary antibodies used for immunofluorescence analysis included anti-MOMA2 (rabbit monoclonal, Bio-Rad, MCA519G; 1:200), anti-CD68 (rat monoclonal, Bio-Rad, MCA1957, 1:300), anti-CD206 (rat monoclonal, Bio-Rad, MCA2235, 1:200), anti-CD3 (rabbit polyclonal, DAKO/Agilent, A045229-2, 1:200), anti-LYVE1 (rabbit monoclonal, Abcam, ab218535, 1:600), anti-CD31 (Armenian-hamster monoclonal, Millipore, MAB1398Z, 1:200), anti-collagen Type I (rabbit polyclonal, Millipore, ABT257, 1:300) and Cy3-conjugated anti-smooth muscle actin (1A4, Sigma-Aldrich, C6198; 1:1,000). Rat IgG (eBioscience) and rabbit IgG (Jackson ImmunoResearch) were used for isotype controls. AF488- (1:300), AF555- (1:300), Cy3- (1:500) and AF647 (1:300)-conjugated antibodies (Jackson ImmunoResearch) were used for fluorescence detection. The apoptosis assay was performed using the Roche In Situ Cell Death Detection Kit, TMR Red (Roche). The number of extracellular (not internalized) apoptotic cells was determined on three ×40 magnification fields of aortic sinus lesions as TUNEL+DAPI+ double-positive cells within the necrotic core of atherosclerotic plaques.
To examine whether NRP1 expression in macrophages affects efferocytosis, bone-marrow-derived macrophages from Lyz2cre+/−Nrp1flox/flox and Lyz2cre+/−Nrp1+/+ mice were incubated with fluorescent dye PKH26-labelled apoptotic Jurkat cells (E6-1; ATCC) induced by ultraviolet irradiation. After 45 min of incubation, macrophages were washed with PBS and then fixed with 4% paraformaldehyde, rinsed with PBS and imaged. The Jurkat cell line used in this experiment tested negative for mycoplasma at first provision. However, it has not been retested later on and has not been authenticated.
Microbiota analysis using 16S amplicon sequencing
DNA was isolated from faecal material of the experimental mice using the DNeasy PowerSoil Pro Kit (Qiagen) according to the manufacturer’s protocol. Extracted DNA was eluted from the spin-filter silica membrane with 100 µl of elution buffer and stored at −80 °C. 16S profiling and MiSeq sequencing was performed as described earlier47,48 with the following modifications: the V3–V4 region of the 16S gene was amplified using the dual barcoded primers 341F (GTGCCAGCMGCCGCGGTAA) and 806R (GGACTACHVGGGTWTCTAAT). Each primer contained additional sequences for a 12 base Golay barcode, Illumina adaptor and a linker sequence49.
PCR was performed using the Phusion Hot Start Flex 2× Master Mix (NEB) in the GeneAmp PCR system 9700 (Applied Biosystems) and the following program: 98 °C for 3 min; 25× 98 °C for 20 s, 55 °C for 30 s and 72 °C for 45 s; then 72 °C for 10 min and hold at 4 °C. The performance of the PCR reactions was checked using agarose gel electrophoresis. Normalization was performed using the SequalPrep Normalization Plate Kit (Thermo Fisher Scientific) according to the manufacturer’s instructions. Equal volumes of SequalPrep-normalized amplicons were pooled and sequenced on the Illumina MiSeq (2 × 300 nucleotides) system. MiSeq sequencing data were analysed using MacQIIME v.1.9.1 (http://www.wernerlab.org/software/macqiime). In brief, all sequencing reads were trimmed, retaining only nucleotides with a Phred quality score of at least 20, then paired-end assembled and mapped onto the different samples using the barcode information. Rarefaction was performed at 4,062 reads per sample to normalize all samples to the minimum shared read count and to account for differential sequencing depth. Sequences were assigned to operational taxonomic units (OTUs) using uclust and the greengenes reference database (gg_13_8 release) with 97% identity. Representative OTUs were picked and taxonomy assigned using uclust and the greengenes database. Quality filtering was performed by removing chimeric sequences using ChimeraSlayer and by removing singletons and sequences that did not align with PyNAST. The reference phylogenetic tree was constructed using FastTree 2. Unequal read counts were normalized by rarefaction to 5,257 and 9,200 (ABx experiment) reads per sample. The relative abundance was calculated by dividing the number of reads for an OTU by the total number of sequences in the sample. β-Diversity was calculated using Bray–Curtis dissimilarity and visualized by principal coordinate plots.
Gut microbiota was depleted using a combination of oral antibiotics (metronidazole 1 g l−1, amoxicillin 0.5 g l−1, vancomycin 0.5 g l−1, neomycin 1 g l−1) dissolved in the drinking water with sucralose (4 g l−1). The success of bacterial depletion was determined by 16S qPCR on the isolated faecal DNAs as described previously50. 16S copy numbers were normalized per g of faecal material and to the untreated control.
Fate mapping of and quantification of resident arterial macrophages
Cx3cr1creER/+R26lsl-tdTomatoLdlr−/− reporter mice were orally gavaged with 200 µl (20 mg ml−1) tamoxifen in corn oil to induce tdTomato-expression in CX3CR1-expressing MacAIR and adventitia-resident macrophages. Mice were then rested on a normal diet for 2 weeks to allow for monocyte turnover in the blood before initiating a 7 week experimental feeding period. The control mice were rested for four additional weeks on a normal diet followed by three weeks on a WD (TD.88137; adjusted calories diet, 42% from fat, Envigo). Mice in the intermittent feeding group underwent three cycles of iWD; this consisted of 1 week of WD followed by 2 weeks on a chow diet, with euthanasia after the third week of WD feeding. Aortae were collected, fixed in a 4% paraformaldehyde with 30% sucrose solution and then stained for macrophages using CD45 (30-F11, BioLegend, 0.5 μg ml−1) and CD68 (FA-11, BioLegend, 1 μg ml−1) antibodies overnight. Adventitia and aortic intima were whole-mount imaged using the Leica Sp8 laser-scanning confocal microscope using LAS X software. Images were processed using Imaris software (Bitplane), and quantification of cells was performed using the spot function. Adventitia samples were collected at three different regions of interest (approximately 1 mm × 1 mm each), and quantification of resident adventitia cell density was averaged for each mouse. Macrophages in the aortic intima were quantified by CD68 expression, and MacAIR cells were identified for co-expression of tdTomato. Sample imaging and quantification were performed with sample identification blinded to the researcher.
Total RNA extraction, cDNA synthesis and qPCR
Total RNA was extracted from cells using the RNeasy mini kit (Qiagen) and cDNA was synthesized using the QuantiTect Reverse Transcription Kit. Quantitative PCR (qPCR) was performed using the SYBR Green qPCR mix (Eurogentec) on the Roche Lightcycler. Rplp0 (also known as 36B4) was used as a reference gene. Data were analysed using the ΔΔCt method.
The primer list was as follows: mouse Nrp1 forward, 5′-CATCTCCCGGTTACCCTCATTCTT-3′; mouse Nrp1 reverse, 5′-GCGGCCGCCTTCATTCTC-3′; Rplp0 forward, 5′-TCTCCAGAGGCACCATTGAAA-3′; Rplp0 reverse, 5′-CTCGCTGGCTCCCACCTT-3′.
Artery tissue digestion
Animals were perfused with cold PBS to remove any trace of blood. Aortas were collected and cleaned of contaminated tissues under a dissection microscope. Aortas were minced with scissors and subjected to two different enzymatic digestion methods. For batch 1, collagenase D (0.2 mg ml−1, Sigma-Aldrich), dispase (1 U ml−1, Stem Cell Technologies), DNase I grade II (0.2 mg ml−1, Sigma-Aldrich) and elastase (1 mg ml−1, Worthington) were used. For batch 2, collagenase type I (450 U ml−1, Sigma-Aldrich), collagenase type XI (125 U ml−1, Sigma-Aldrich), hyaluronidase type 1 (60 U ml−1, Sigma-Aldrich), DNase I (60 U ml−1, Sigma-Aldrich) were used, then incubated for 45 min at 37 °C. Cell suspensions were pipetted up and down until disappearance of all residual piece of tissue and filtered through a 70 µm cell strainer.
RNA-seq and analysesRNA-seq
For aortic macrophages, single-cell suspensions of aortic cells were prepared. Cell suspensions were incubated with Zombie Aqua Fixable Viability Kit (BioLegend) for 10 min at room temperature and the cells were then stained with fluorescently labelled anti-mouse antibodies against CD45, CD11b and CD64 at 4 °C. Aortic macrophages were identified as CD45+CD11b+CD64+. Peritoneal macrophages were obtained by intraperitoneal injection of cold PBS from Ldlr−/− mice reconstituted with Lyz2cre+/−Nrp1flox/flox or Lyz2cre+/−Nrp1+/+ bone marrow. Peritoneal exudates were allowed to adhere to the six-well plates by culturing them for 2 h at 37 °C. Nonadherent cells were removed by gently washing with warm PBS and the remaining peritoneal macrophages were used for bulk RNA-seq.
RNA extraction and residual DNA removal after DNase digestion was performed using the RNeasy mini kit (Qiagen) according to the manufacturer’s instructions. After quality checking using the Agilent Bioanalyser 2100 system, total RNA was made into sequencing libraries. For aortic macrophages, we used the Takara SMART-Seq v4 Ultra Low Input RNA Kit. In brief, total RNA was fragmented before reverse transcription. Second-strand cDNA was PCR synthesized with the incorporation of SMART technology. cDNA originally from rRNA was removed selectively before Illumina-compatible barcoded libraries were generated by PCR amplification. Indexed libraries were normalized, pooled and were sequenced on the Illumina HiSeq 4000 platform, single-end reads (SE50). For peritoneal macrophages, total RNA was purified using poly-T oligo beads and fragmented before reverse transcription. After adenylating 3′ ends and ligating the adapter, cDNA fragments were amplified by PCR. Indexed libraries were normalized, pooled and were sequenced on the Illumina NovaSeq 6000 platform, paired-end reads (PE50). All sequencing was performed at the Genomics Core Facility, Cancer Research UK Cambridge Institute.
RNA-seq analyses
For iWD and cWD Ldlr−/− mice, single-end, unstranded sequencing was performed using SMART-Seq v4 plus Nextera (Ultra Low Input RNA Kit (Takara)) with read lengths of 50 bp. Raw fastq data have been deposited in ArrayExpress under accession number E-MTAB-12759. There are two separate experiments for this study with iWD and cWD groups, each group has three independent pools per experiment, and each pool comprises three different mice. All raw data were merged and analysed together, with a total of six different pools for each group. The sample summary table is provided in Supplementary Table 23a. The quality control, alignment and gene quantification analysis were run using the nextflow (v.21.05.0.edge)51 RNA-seq pipeline with nf-core/rnaseq (v.3.2, https://nf-co.re/rnaseq)52 with the option ‘--aligner star_salmon’ and Ensembl reference genome with the annotation for mouse GRCm39. The alignment results are with a mean of 72% reads uniquely mapping and mean of 34 million reads per sample (Supplementary Table 23a). Corresponding software/packages and versions from nextflow pipeline are listed in Supplementary Table 24. All scripts, with details of software versions, expression raw count files and results are available at GitHub (https://github.com/CAD-ZM-BFX/Takaoka_Mallat) and Zenodo (https://doi.org/10.5281/zenodo.13137366).
Counts extracted with htseq-counts were used to perform the DEG analysis using DESeq2 (v.1.36.0; R package, v.4.2.1; https://www.r-project.org/)53. In total, 746 genes were identified as significant DEGs when comparing the iWD group to the cWD group using the cut-off threshold Padj < 0.05 and |log2[fold change]| ≥ 1 (an absolute fold change of 2). The list of significant DEGs is provided in Supplementary Table 1 and a volcano plot is presented in Fig. 2b. The two-sided Wald-test P value was then adjusted using the Benjamini–Hochberg false-discovery rate multiple-test correction method. GO and KEGG pathway analysis were performed using R package clusterProfiler (v.4.4.4)54,55 on the significant DEGs (n = 746); all identified pathways are listed in Supplementary Table 2, with ten selected pathways plotted in Fig. 2a. The enrichment P value was calculated using the hypergeometric distribution with the one-side Fisher’s exact test.PX≥x=1−PX≤x−1=1−∑i=0x−1MiN−Mn−iNn\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$P\left(X\ge x\right)=1-P\left(X\le x-1\right)=1-\,\mathop{\sum }\limits_{i=0}^{x-1}\frac{\left(\genfrac{}{}{0ex}{}{M}{i}\right)\left(\genfrac{}{}{0ex}{}{N-M}{n-i}\right)}{\left(\genfrac{}{}{0ex}{}{N}{n}\right)}$$\end{document}
In this equation, N is the total number of genes in the background distribution, M is the number of genes within that distribution that are annotated (either directly or indirectly) to the gene set of interest, n is the size of the list of genes of interest and x is the number of genes within that list which are annotated to the gene set. The background distribution by default is all of the genes that have annotation. P values are adjusted for multiple comparison using Benjamini–Hochberg correction.
TF-binding motif enrichment analysis was performed using Rcistarget package (v.1.16.0). Both significant DEGs and redefined (Padj > 0.1 and |log2[fold change]| < 1) non-significant DEG after TF analysis were performed. We identified 73 common enriched binding motifs corresponding to 27 TFs (Supplementary Table 4) and we performed two-sample proportion z-tests (two-tailed) and adjusted the P values using Benjamini–Hochberg correction. We generated a dot plot by plotting the difference in normalized enrichment score (NES) between significant and non-significant DEGs (Extended Data Fig. 6a). Moreover, SPIC was selected and GO analysis was performed on significant DEGs (n = 199) showing enrichment for SPIC binding motifs (Extended Data Fig. 6b and Supplementary Table 5).
To assign significant DEGs to specific mouse macrophage cell types, we performed integrated analysis of the published scRNA-seq data on mouse atherosclerosis. We selected four main mouse studies from a previous article20 (Supplementary Table 25). Both individual and integrated study analyses were applied for the data using R package Seurat (v.5.0.0)56 with integration method scVI57. We applied FeaturePlot to a set of key mouse macrophage genes (Csf1r, Sepp1, Pf4, F13a1, Folr2, Nrp1, Lyve1, Cd163, Ednrb, Timd4, Phactr1, Egr1, Lgals3, Ctsd, Cxcl2, Ccl4, Trem2, Spp1, Cd9, Gpnmb, Il1b, Cd74, Stat1, Ifit3, Mnda, Irf7, Isg15, Ifit2 and Fcgr4) to identify the macrophage cluster. We then reclustered the macrophage cell types into six main subtypes of macrophages identified as resident-like, IFN-, inflammation-, MacAIR- and TREM2-like and CCR2-MHCII. In early atherosclerosis, MacAIR macrophages are the first macrophages that become foam cells, and MacAIR- and TREM2-like macrophages share common markers and gene programs16. Therefore, they were grouped together for some analyses. For mouse macrophages, we identified NRP1high and NRP1low groups, then performed GO pathway analysis using enrichR (R package, v.3.1)58–60 (Supplementary Table 10).
Further differential expression analysis between chow diet and high-fat diet of mouse scRNA-seq data across different time courses of high-fat diet was performed for three main subsets of macrophages using the FindMarkers function with the MAST method in Seurat (v.5.0.0, R package)56. A log2[fold change] heat map was generated with key genes in each annotated subtype of macrophage across time (Extended Data Fig. 7a). A proportion bar plot presents the general information for three main macrophage subtypes and the change in their proportions with increased duration of high-fat diet (Extended Data Fig. 7b).
GWAS analysis was performed using the extracted list related to atherosclerosis/CAD (coronary artery disease) from the GWAS Catalog (https://www.ebi.ac.uk/gwas/) (Supplementary Table 26) and from supplementary table 13 of ref. 28. The GWAS list human symbols were converted to mouse symbols using g:Profiler2 orthology (v.0.2.1)61 in R (Supplementary Table 27). We then performed the overlap analysis between the GWAS list and our DEGs list (significant DEGs and non-significant DEGs; Fig. 4a,b and Supplementary Tables 6 and 7). To assign the 210 significant DEGs that are linked by coronary artery GWAS to specific macrophage cell types, we performed integrated analysis of the published scRNA-seq data on human atherosclerosis. Six human studies from cardiovascular PlaqView were selected (https://www.plaqview.com/; Supplementary Table 28). Integrated analysis was performed first using R package Seurat (v.5.0.0)56, and the macrophage cluster was identified by applying the FeaturePlot for a set of key human macrophage genes (CSF1R, PF4, FOLR2, NRP1, LYVE1, LGALS3, CXCL2, SPP1, GPNMB, IL1B, APOC1, APOE, FCGR4, IFIT2 and FABP4). We then reclustered the macrophage cell types into five main subtypes of macrophages identified as resident-like, IFN-, inflammation-, foamy-like and mixed subtypes. Next, we assigned our GWAS-related 210 significant DEGs (Supplementary Table 6) to each macrophage cell type key marker list (Fig. 4b and Supplementary Table 7). For both human macrophages, we identified NRP1high and NRP1low groups, then performed GO pathway analysis using enrichR (R package, v.3.1)58–60 (Supplementary Table 9).
We analysed data from a recent GWAS study in which the investigators systematically characterized risk variants and genes for CAD in over a million participants33. Two hundred and twenty candidate causal genes at 279 CAD associations were prioritized, combining eight complementary approaches. We used clusterProfiler (v.4.4.4, R package)54,55 and performed GO enrichment analysis of biological process orthology. In total, 178 out of 220 causal genes contributed to the GO enrichment analysis, and 867 enriched biological processes were identified (Supplementary Table 11). Selected pathways and corresponding causal gene network plots are shown in Extended Data Fig. 9 using the cnetplot function in R (enrichplot v.1.22.0; https://bioconductor.org/packages/release/bioc/html/enrichplot.html). Significant differentially expressed genes in our RNA-seq data of aortic macrophages from iWD versus cWD (Supplementary Table 1) are shown (Extended Data Fig. 9).
RNA-seq of peritoneal macrophages
Paired-end (PE50), reverse-stranded sequencing for Nrp1 WT versus Nrp1-KO peritoneal macrophages in reconstituted Ldlr−/− male mice was performed using the Illumina NovaSeq6000 platform. Raw fastq data have been deposited in ArrayExpress under accession number E-MTAB-12761. DEG analysis was performed using the same workflow as mentioned for iWD versus cWD. The sample summary and mapping statistics table is provided in Supplementary Table 23b. There were 113 significant DEGs identified (Padj < 0.05 and |log2[fold change]| ≥ 0.6; Supplementary Table 12). GO was performed using clusterProfiler (R package, v.4.4.4)54,55 with the 113 DEGs, and the list of pathways is given in Supplementary Table 13.
Immunohistochemistry of human coronary healthy and atherosclerotic arteries
Samples of human aortic tissue were obtained from the INSERM cardiovascular tissue Biobank (member of European BBMRI-RIC organization). The study was performed according to the Guidelines of the World Medical Association Declaration of Helsinki. The local ethics committee of the INSERM Paris (01-024) approved the study, and written informed consent for permission was given by all patients. Human coronary arteries were obtained from patients with coronary artery disease from explanted hearts of transplant recipients (average age, 59 years) INSERM. Coronary artery disease classification was based on coronary angiography before transplant and macroscopic aspect at dissection to confirm the presence of atherosclerotic plaque. Furthermore, control healthy human coronary arteries were obtained from explanted hearts of organ donors (average age, 56 years). Paraffin sections were stained with H&E to assess atherosclerosis. For immunostaining, antisera were first optimized using different dilutions to determine the best staining results with minimal background. Immunostainings were performed as previously described62 using the following marker antibodies: goat anti-human LYVE1 (R&D, AF2089, 1:100) and mouse anti-human CD68 (DAKO, M0876, 1:50). After antigen retrieval and incubation with primary antibodies, visualization was performed using secondary antibodies (donkey anti-goat Cy3, Jackson ImmunoResearch, 705-165-147, 1:300 for LYVE1 and donkey anti-mouse Alexa 647, Jackson ImmunoResearch, 705-605-150, 1:300 for CD68) as previously described62,63. DAPI was used to stain the cell nuclei. For negative controls, staining was performed without primary antibodies. Stained slides were imaged using the Leica TCS SP8 3X or by DM6 B Thunder Imager (Leica Microsystems). Images were acquired under identical microscope settings using sequential acquisition of different channels to avoid cross-talk between fluorophores, and non-overlapping fluorophores were used for colocalization analysis. All images were prepared as TIF files and quantified using ImageJ software. For quantification of LYVE1+ macrophages in the adventitia, LYVE1+CD68+ macrophages among total adventitial CD68+ cells were quantified in 5–8 images per section in three sections per sample and expressed as percentage of LYVE1+ macrophages from adventitial CD68+ cells.
Participants of the YFS
The Cardiovascular Risk in Young Finns Study is a prospective cohort study that began in 1980 when the participants were aged 3–18 years. Follow-up clinics were conducted in 1983, 1986, 1989 (sub-sample), 1992 (sub-sample), 2001, 2007, 2011 and 2018. The primary sample for this analysis included up to 2,062 participants who attended the most recent 2018 follow-up when aged 41–56 years and who had carotid ultrasound studies performed. Moreover, a secondary sample included 2,643 participants who attended either the 2001 or 2007 follow-ups then aged 24–45 years when carotid ultrasound studies were also performed. Ethics approval for the baseline study was granted by the University of Turku Faculty of Medicine Ethical Committee (letter dated 14/11/1978, minutes 3/1978) and for the most recent follow-up, granted by the Hospital District of Southwest Finland (Ethics approval number ETMK:68/1801/2017).
Measures performed in the YFS
At each timepoint, venous blood samples were taken after an overnight fast (12 h) using standard methods to measure serum total cholesterol and HDL-C that were calibrated to shifts in assay kits over time. Non-HDL-C was calculated as total cholesterol minus HDL-C. In 2001 and 2007, B-mode ultrasound with 13.0 MHz linear array transducers (Sequoia 512, Acuson) was performed on the left common carotid artery and the beginning of the bifurcation; and, in 2018, using the GE Logiq S8 system (GE Vingmend Ultrasound A/S) on both the left and right common carotid artery, bifurcation and internal carotid artery. Carotid plaques were defined as focal structures protruding into the arterial lumen of at least 0.5 mm or 50% of the surrounding intima–media thickness value or demonstrating a thickness of 1.5 mm as measured from the media–adventitia interface to the intima–lumen interface64. Carotid artery plaque measures were performed off-line by a single reader blinded to the participant’s details. Where ultrasound data were available on participants at both 2001 and 2007 follow-ups, data from 2007 was used. The prevalence of plaque was 39.6% in 2018 (817 out of 2,062) and 3.3% (88 out of 2,643) in 2001/07.
Statistics and reproducibility
Statistical analyses were performed using Graph Pad Prism 7 (Graph Pad Software). Experimental groups were compared using two-tailed Student’s unpaired t-tests or Mann–Whitney tests as appropriate. Data are presented as mean ± s.e.m. P < 0.05 was considered to be significant.
Statistical analyses in the YFS
The BRLM. The Bayesian relevant life-course exposure model (BRLM) was used to identify relative contribution of non-HDL-C measured in childhood (aged 6, 9 and 12 years), adolescence (aged 15 and 18 years), and young adulthood (aged 21 and 24 years) on the atherosclerosis outcome variables in mid-adulthood (carotid plaque presence with Poisson regression and plaque area with linear regression). Detailed methods on the BRLM have been provided elsewhere65,66. The BRLM estimates an overall or total lifetime effect, representing the maximum accumulated effect of non-HDL-C measured at different stages across the observed life-course (in this case childhood, adolescence and young adulthood) on each outcome variable in mid-adulthood (β for continuous outcomes, and relative risk for categorical outcomes). The BRLM also allows estimation of the relative weights and their posterior distributions for non-HDL-C measured in childhood, adolescence and young adulthood that represent the direct or relative contribution of non-HDL-C at each life stage to each outcome. These weight parameters help to determine the life-course model that is best supported by the data. The life-course association of non-HDL-C and each outcome forms one of three prevailing hypotheses: (1) an accumulation life-course model (non-HDL-C at each life stage is having an equal contribution); or subclasses of an accumulation model; (2) a sensitive life-course model (non-HDL-C levels at each life stage has different importance); or (3) a critical life-course model (non-HDL-C at only one life stage is of importance). Using the accumulated effect and relative weight, the BRLM also estimates the life-stage-specific effect. Here we fit the BRLM using a Cauchy (0, 2.5) prior for the lifetime effect and a non-informative Dirichlet (1, 1, 1) prior for weights.
The IGC model. The individual growth curve (IGC) model67 was used to interpolate over ages with missing non-HDL-C values68. The IGC model is an advanced multilevel mixed-effect regression model that enables the simultaneous modelling of the interindividual differences in intraindividual systematic changes over time (repeated measurements at the individual level). The non-HDL-C trajectory in our study was best described by an IGC model with a cubic age polynomial (age67) random intercept, quadratic age random slope and inclusion of sex as a modifier. If observed non-HDL-C data at the ages used in the BRLM were missing, we used individual-level data derived from the IGC model to fill these gaps. These non-HDL-C levels at each timepoint were transformed into age- and sex-specific z scores. Life-stage averaged non-HDL-C z scores were calculated as the mean values between ages 6 to 12 years for childhood ((z6 years old + z9 years old + z12 years old)/3), ages 15 and 18 years for adolescence ((z15years old + z18 years old)/2), and ages 21 and 24 years for young adulthood ((z21years old + z24 years old)/2). In our primary analyses, we show the association for a 1 s.d. increase in non-HDL-C z-score accounts for variations in non-HDL-C that naturally occur with age and sex. We have done this to ensure a uniform metric for comparison across each life-stage. However, we also provide data for a per 1 mmol l−1 increase in non-HDL-C in the supplement, which might be more clinically relevant. We routinely adjusted for potential confounders in a minimally adjusted model (model 1) that included sex and year of birth, and a fully adjusted model (model 2) that included sex, year of birth, education (years studied), ever smoked daily before age 24 years, family history of cardiovascular disease and areas under the curve of measurements available up to age 24 years for body mass index, HDL-C, systolic blood pressure, fasting blood glucose and physical activity index. All models were additionally performed stratified by sex. Owing to missing covariates, there was a reduction in sample size between models 1 and 2. Methods for imputing missing values have been described (https://github.com/MadathilSA/BayesianRelevantLifeCourseExposure/blob/master/Manuscript III/HeNCeIndiaSmkModelImput.jags). Multiple imputation was undertaken for the covariates of youth age, ever smoked, family history of cardiovascular disease, education (number of years studied), and cumulative body mass index, systolic blood pressure and HDL-C, using predictive mean matching via the R library mice. Three imputations were made, and the means and credible intervals were calculated using the draws across all imputations.
The rstan package of R studio (v.3.5.3, R Foundation for Statistical Computing) was used to fit the BRLM in the probabilistic programming language, Stan69. The Lme4 package of R studio (v.3.5.3, R Foundation for Statistical Computing) was used to perform the IGC modelling.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Online content
Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/s41586-024-07993-x.
Acknowledgements
Z.M. is supported by a BHF chair grant (CH/10/001/27642), a BHF Special Project Grant (SP/F/20/150010), the BHF Centre of Research Excellence (RE/18/1/34212), the Leducq Foundation (20CVD03) and the NIHR Cambridge Biomedical Research Centre (NIHR203312). The views expressed are those of the authors and not necessarily those of the NIHR or the Department of Health and Social Care; M.N. by a BHF Intermediate Fellowship (G103314). S.A.N. is supported by a BHF Project Grant (PG/21/10816); and T.X.Z. through the BHF (CH/10/001/27642) and by the BMA Foundation, and the Academy of Medical Sciences (SGL025/1071). This work is also supported by a NMRC-funded grant to H.Y.L. (OFYIRG20nov-0013) and to V.A. (OF-IRG19Nov-0112). J.W.W. and P.R.S. were supported by the American Heart Association (CDA855022) and the US National Institutes of Health (R01 AI165553); D.T. by the European Research Council (ERC Starting grant), the Austrian Science Fund (I4963, P35233) and the European Research Area Network on Cardiovascular Diseases (I4647); F.S. by the German Research Foundation (DFG) through the grant SO1141/10-1 and the Research Unit FOR5042 “miTarget—The Microbiome as a Target in Inflammatory Bowel Diseases” (project P5); S.K.M. through a Corona Foundation grant (S199/10087/2022), and a DFG grant (SFB 1123/Z1); and C.R. by a Wellcome Investigator Award (095623/Z/11/Z). This work is also supported by the Academy of Finland (grant numbers 286284, 134309 (Eye), 126925, 121584, 124282, 129378 (Salve), 117787 (Gendi) and 41071 (Skidi)); the Social Insurance Institution of Finland; Competitive State Research Financing of the Expert Responsibility area of Kuopio, Tampere and Turku University Hospitals (grant number X51001); the Turku University Foundation; Tampere University Hospital Supporting Foundation; the Juho Vainio Foundation; Paavo Nurmi Foundation; the Finnish Foundation of Cardiovascular Research; Orion-Farmos Research Foundation; Sigrid Juselius Foundation; Emil Aaltonen Foundation; Yrjö Jahnsson Foundation; Signe and Ane Gyllenberg Foundation; Diabetes Research Foundation of Finnish Diabetes Association; Tampere Tuberculosis Foundation; the Finnish Cultural Foundation; and EU Horizon 2020 (grant number 755320 for TAXINOMISIS). J.S.K. is supported by the European Research Council (grant number 742927 for MULTIEPIGEN project), Maud Kuistila Memorial Foundation, Päivikki and Sakari Sohlberg Foundation, Finnish Medical Foundation, Paulon Foundation and Licentiate of Medicine Paavo Ilmari Ahvenainen Foundation; and C.G.M. by a National Health and Medical Research Council (NHMRC) Investigator Grant (APP1176494). The contents of the published material are solely the responsibility of the individual authors and do not reflect the views of the NHMRC. We thank W. Ise for his contribution to generating the Spicflox/flox mice; J.-B. Michel for providing tissue sections of healthy and atherosclerotic human coronary arteries; D. Cuchet-Lourenco and E. F. Warner for providing Jurkat cells and assistance with BMDM experiments, respectively; the staff at the Phenotyping Hub and Biochemical Assay Laboratory of Cambridge University Hospitals, the University of Cambridge Biomedical Services at the Anne McLaren Building, the Cancer Research UK Cambridge Institute Genomics Core facility and the Genomics & Bioinformatics Core, Wellcome-MRC Institute of Metabolic Science-Metabolic Research Laboratories and Medical Research Council Metabolic Diseases Unit, University of Cambridge; the members of the Advanced Imaging and Histology cores of Life Sciences Institute, NUS; and S. U. Gan and S. K. Chuan for providing AAV8-D377Y-mPCSK9.
Author contributions
M.T. led and performed most of the experiments on atherosclerosis and contributed to the RNA-seq data. X.Z. analysed the RNA-seq data. V.A. and H.Y.L. designed the experiments related to LYVE1+ expressing adventitial macrophages, and H.Y.L., O.A. and W.S.O. performed the corresponding experiments and analysed the data. C.G.M. and O.T.R. organized data collection, designed the analysis plan of the YFS and interpreted the data, and Y.M. and F.W. performed those analyses. NS performed and analysed atherosclerosis experiments. P.R.S. and J.W.W. performed fate-mapping experiments and analysed data. D.T. performed atherosclerosis and RNA-seq experiments. F.S. performed 16S amplicon sequencing and analysed/interpreted corresponding data. M.N. contributed to analysis of the microbiota depletion experiments. S.K.M., M.H. and A.J.R.H. performed and analysed experiments related to adventitia LYVE1+ macrophages in human coronary arteries, and T.X.Z. provided artery samples. J.H., Y.L., L.M. and S.A.N. contributed to the atherosclerosis and RNA-seq experiments. L.L., M. Clement, M. Chajadine and S.T. maintained genetically modified mice and helped in atherosclerosis experiments. L.D. and C.R. maintained, mated and genotyped genetically modified mice for bone marrow donation. M.J. and J.S.K. did the plaque measurements in the YFS. M.K. contributed to the set-up of the YFS. K.P., S.P.R. and J.M. collected data from the 2018 visit. R.T. conducted statistical analyses on the YFS. T.K. generated and provided genetically modified mice. A.T. and H.A.-O. contributed to data interpretation. Z.M. conceived the idea, designed, initiated and supervised the project, established the required collaborations and wrote the manuscript. All of the authors reviewed and edited the manuscript.
Peer reviewPeer review information
Nature thanks Sreenath Madathil and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.
Data availability
All the associated raw data presented in this paper are available from the corresponding author on reasonable request. Raw RNA-seq data are accessible through EMBL-EBI ArrayExpress under accession numbers E-MTAB-12759 and E-MTAB-12761. All of the scRNA-seq datasets used in this study are publicly available. For mouse datasets, see Supplementary Table 25. For human datasets, see Supplementary Table 28. Source data are provided with this paper.
Code availability
Custom codes for RNA-seq analysis used in this paper are available at GitHub and Zenodo70 (https://github.com/CAD-ZM-BFX/Takaoka_Mallat, https://doi.org/10.5281/zenodo.13137366).
Competing interests
The authors declare no competing interests.
ReferencesDomanskiMJTime course of LDL cholesterol exposure and cardiovascular disease event riskJ. Am. Coll. Cardiol.202076150715161:CAS:528:DC%2BB3cXhvFyisbnN32972526KimMKCholesterol variability and the risk of mortality, myocardial infarction, and stroke: a nationwide population-based studyEur. Heart J.201738356035661:CAS:528:DC%2BC1cXisV2murvP290694586251576KimMKAssociations of variability in blood pressure, glucose and cholesterol concentrations, and body mass index with mortality and cardiovascular outcomes in the general populationCirculation2018138262726371:CAS:528:DC%2BC1cXitlOksbzE30571256BangaloreSVisit-to-visit low-density lipoprotein cholesterol variability and risk of cardiovascular outcomes: insights from the TNT trialJ. Am. Coll. Cardiol.201565153915481:CAS:528:DC%2BC2MXmsFSntbs%3D25881936DuewellPNLRP3 inflammasomes are required for atherogenesis and activated by cholesterol crystalsNature2010464135713612010Natur.464.1357D1:CAS:528:DC%2BC3cXltl2mu7w%3D204281722946640RidkerPMAntiinflammatory therapy with canakinumab for atherosclerotic diseaseN. Engl. J. Med.2017377111911311:CAS:528:DC%2BC2sXhsFOqurnE28845751IshibashiSGoldsteinJLBrownMSHerzJBurnsDKMassive xanthomatosis and atherosclerosis in cholesterol-fed low density lipoprotein receptor-negative miceJ. Clin. Invest.199493188518931:CAS:528:DyaK2cXivFOgtbw%3D8182121294295GetzGSReardonCAAnimal models of atherosclerosisArterioscler. Thromb. Vasc. Biol.201232110411151:CAS:528:DC%2BC38XlslWqur8%3D223837003331926SkalenKSubendothelial retention of atherogenic lipoproteins in early atherosclerosisNature20024177507542002Natur.417..750S1:CAS:528:DC%2BD38XksVGjuro%3D12066187TeupserDPerskyADBreslowJLInduction of atherosclerosis by low-fat, semisynthetic diets in LDL receptor-deficient C57BL/6 J and FVB/NJ mice: comparison of lesions of the aortic root, brachiocephalic artery, and whole aorta (en face measurement)Arterioscler. Thromb. Vasc. Biol.200323190719131:CAS:528:DC%2BD3sXnslWisbs%3D12907460Lindskog JonssonAImpact of gut microbiota and diet on the development of atherosclerosis in Apoe−/− MiceArterioscler. Thromb. Vasc. Biol.201838231823261:CAS:528:DC%2BC1cXhslOru7nE299037356166703SmithJDDecreased atherosclerosis in mice deficient in both macrophage colony-stimulating factor (op) and apolipoprotein EProc. Natl Acad. Sci. USA199592826482681995PNAS...92.8264S1:CAS:528:DyaK2MXnvVymt7c%3D766727941137LiaoXMacrophage autophagy plays a protective role in advanced atherosclerosisCell Metab.2012155455531:CAS:528:DC%2BC38XksFegt7o%3D224456003322248ShiJA genome-wide CRISPR screen identifies WDFY3 as a regulator of macrophage efferocytosisNat. Commun.2022132022NatCo..13.7929S1:CAS:528:DC%2BB38XjtF2gtLzN365662599789999YurdagulAJrDoranACCaiBFredmanGTabasIAMechanisms and consequences of defective efferocytosis in atherosclerosisFront. Cardiovasc. Med.201748629379788WilliamsJWLimited proliferation capacity of aortic intima resident macrophages requires monocyte recruitment for atherosclerotic plaque progressionNat. Immunol.202021119412041:CAS:528:DC%2BB3cXhslKju77I328955397502558KasikaraCDeficiency of macrophage PHACTR1 impairs efferocytosis and promotes atherosclerotic plaque necrosisJ. Clin. Invest.2021131e145275336307588262505AbellNSMultiple causal variants underlie genetic associations in humansScience2022375124712542022Sci...375.1247A1:CAS:528:DC%2BB38XnsF2rsL4%3D352982439725108HuanTA systems biology framework identifies molecular underpinnings of coronary heart diseaseArter. Thromb. Vasc. Biol.201333142714341:CAS:528:DC%2BC3sXps1yksL4%3DZernecke, A. et al. Integrated single-cell analysis based classification of vascular mononuclear phagocytes in mouse and human atherosclerosis. Cardiovasc. Res.119, 1676–1689 (2022).LimHYHyaluronan receptor LYVE-1-expressing macrophages maintain arterial tone through hyaluronan-mediated regulation of smooth muscle cell collagenImmunity2018493263411:CAS:528:DC%2BC1cXhtlOrtbnI30054204CochainCSingle-cell RNA-seq reveals the transcriptional landscape and heterogeneity of aortic macrophages in murine atherosclerosisCirc. Res.2018122166116741:CAS:528:DC%2BC1cXhtVyltLzN29545365KimKTranscriptome analysis reveals nonfoamy rather than foamy plaque macrophages are proinflammatory in atherosclerotic murine modelsCirc. Res.2018123112711421:CAS:528:DC%2BC1cXhvFyktrrO303592006945121VafadarnejadEDynamics of cardiac neutrophil diversity in murine myocardial infarctionCirc. Res.2020127e232e2491:CAS:528:DC%2BB3cXitVSqtb7N32811295Myocardial Infarction Genetics Consortium. Genome-wide association of early-onset myocardial infarction with single nucleotide polymorphisms and copy number variants. Nat. Genet.41, 334–341 (2009).ConsortiumCADLarge-scale association analysis identifies new risk loci for coronary artery diseaseNat. Genet.2013452533BeaudoinMMyocardial infarction-associated SNP at 6p24 interferes with MEF2 binding and associates with PHACTR1 expression levels in human coronary arteriesArterioscler. Thromb. Vasc. Biol.201535147214791:CAS:528:DC%2BC2MXos1OjsLc%3D258384254441556TcheandjieuCLarge-scale genome-wide association study of coronary artery disease in genetically diverse populationsNat. Med.202228167916921:CAS:528:DC%2BB38XitVehs7jO359151569419655FernandezDMSingle-cell immune landscape of human atherosclerotic plaquesNat. Med.201925157615881:CAS:528:DC%2BC1MXhvFaksbnI315916037318784RaimondiCImatinib inhibits VEGF-independent angiogenesis by targeting neuropilin 1-dependent ABL1 activation in endothelial cellsJ. Exp. Med.2014211116711831:CAS:528:DC%2BC2cXhtVSgurbP248630634042645FantinANRP1 regulates CDC42 activation to promote filopodia formation in endothelial tip cellsCell Rep.201511157715901:CAS:528:DC%2BC2MXhtVejtLbE260519424528263AllainBNeuropilin-1 regulates a new VEGF-induced gene, Phactr-1, which controls tubulogenesis and modulates lamellipodial dynamics in human endothelial cellsCell Signal.2012242142231:CAS:528:DC%2BC3MXhsVSksLvO21939755AragamKGDiscovery and systematic characterization of risk variants and genes for coronary artery disease in over a million participantsNat. Genet.202254180318151:CAS:528:DC%2BB38XjtVGru7%2FJ364740459729111WilsonAMNeuropilin-1 expression in adipose tissue macrophages protects against obesity and metabolic syndromeSci. Immunol.20183eaan462629549139RaitakariOTCardiovascular risk factors in childhood and carotid artery intima-media thickness in adulthood: the Cardiovascular Risk in Young Finns StudyJAMA2003290227722831:CAS:528:DC%2BD3sXovVCjsro%3D14600186KoskinenJSChildhood risk factors and carotid atherosclerotic plaque in adulthood: the Cardiovascular Risk in Young Finns StudyAtherosclerosis202029318251:CAS:528:DC%2BC1MXitlOmur%2FF31830725GiralPNeumannAWeillACosteJCardiovascular effect of discontinuing statins for primary prevention at the age of 75 years: a nationwide population-based cohort study in FranceEur. Heart J.20194035163525313623076855142ParkMDSilvinAGinhouxFMeradMMacrophages in health and diseaseCell2022185425942791:CAS:528:DC%2BB38XivVKms7jE363683059908006CombadiereCDecreased atherosclerotic lesion formation in CX3CR1/apolipoprotein E double knockout miceCirculation2003107100910161:CAS:528:DC%2BD3sXjt1yjs78%3D12600915EnsanSSelf-renewing resident arterial macrophages arise from embryonic CX3CR1+ precursors and circulating monocytes immediately after birthNat. Immunol.2016171591681:CAS:528:DC%2BC2MXhvFKgsb%2FJ26642357LuirinkIK20-year follow-up of statins in children with familial hypercholesterolemiaN. Engl. J. Med.2019381154715561:CAS:528:DC%2BC1MXitVShs7jE31618540WilliamsKJEradicating atherosclerotic events by targeting early subclinical disease: it is time to retire the therapeutic paradigm of too much, too lateArterioscler. Thromb. Vasc. Biol.20244448641:CAS:528:DC%2BB3sXislWisLfE37970716RayKKTwo phase 3 trials of inclisiran in patients with elevated LDL cholesterolN. Engl. J. Med.2020382150715191:CAS:528:DC%2BB3cXnvV2iurg%3D32187462MusunuruKIn vivo CRISPR base editing of PCSK9 durably lowers cholesterol in primatesNature20215934294342021Natur.593..429M1:CAS:528:DC%2BB3MXhtFanur3I34012082KayamaHHeme ameliorates dextran sodium sulfate-induced colitis through providing intestinal macrophages with noninflammatory profilesProc. Natl Acad. Sci. USA2018115841884232018PNAS..115.8418K1:CAS:528:DC%2BC1cXitVCiu7%2FJ300614156099887DaughertyARecommendation on design, execution, and reporting of animal atherosclerosis studies: a scientific statement from the american heart associationArterioscler. Thromb. Vasc. Biol.201737e131e1571:CAS:528:DC%2BC2sXhtlOhsb%2FO28729366SommerFAltered mucus glycosylation in core 1 O-glycan-deficient mice affects microbiota composition and intestinal architecturePLoS ONE20149e852542014PLoSO...985254S244163703887012SommerFThe gut microbiota modulates energy metabolism in the hibernating brown bear Ursus arctosCell Rep.201614165516611:CAS:528:DC%2BC28Xit1WmurY%3D26854221CaporasoJGUltra-high-throughput microbial community analysis on the Illumina HiSeq and MiSeq platformsISME J20126162116241:CAS:528:DC%2BC38XhtVOjtL3N224024013400413Moltzau AndersonJNOD2 influences trajectories of intestinal microbiota recovery after antibiotic perturbationCell Mol. Gastroenterol. Hepatol.202010365389322894997327897Di TommasoPNextflow enables reproducible computational workflowsNat. Biotechnol.20173531631928398311EwelsPAThe nf-core framework for community-curated bioinformatics pipelinesNat. Biotechnol.2020382762781:CAS:528:DC%2BB3cXjtFelsbo%3D32055031LoveMIHuberWAndersSModerated estimation of fold change and dispersion for RNA-seq data with DESeq2Genome Biol.201415255162814302049YuGWangLGHanYHeQY. clusterProfiler: an R package for comparing biological themes among gene clustersOMICS2012162842871:CAS:528:DC%2BC38XmsFarsLw%3D224554633339379WuTclusterProfiler 4.0: a universal enrichment tool for interpreting omics dataInnovation202121001411:CAS:528:DC%2BB38XitVemt7jI345577788454663StuartTComprehensive Integration of Single-Cell DataCell2019177188819021:CAS:528:DC%2BC1MXhtFens77L311781186687398GayosoAA Python library for probabilistic analysis of single-cell omics dataNat. Biotechnol.2022401631661:CAS:528:DC%2BB38XjtFSit78%3D35132262ChenEYEnrichr: interactive and collaborative HTML5 gene list enrichment analysis toolBMC Bioinform.201314128KuleshovMVEnrichr: a comprehensive gene set enrichment analysis web server 2016 updateNucleic Acids Res.201644W90W971:CAS:528:DC%2BC2sXhtV2itrfF271419614987924XieZGene set knowledge discovery with EnrichrCurr. Protoc.20211e901:CAS:528:DC%2BB3MXntl2gtb0%3D337801708152575RaudvereUg:Profiler: a web server for functional enrichment analysis and conversions of gene lists (2019 update)Nucleic Acids Res.201947W191W1981:CAS:528:DC%2BB3cXktVyit70%3D310664536602461MohantaSKNeuroimmune cardiovascular interfaces control atherosclerosisNature20226051521592022Natur.605..152M1:CAS:528:DC%2BB38XhtFersLjK35477759YinCApoE attenuates unresolvable inflammation by complex formation with activated C1qNat. Med.2019254965061:CAS:528:DC%2BC1MXmtFKjsLc%3D306926996420126JeejeebhoyKNHepatic manifestations of total parenteral nutrition: need for prospective investigationHepatology198884284291:STN:280:DyaL1c7os1SitQ%3D%3D3128470MadathilSJosephLHardyRRousseauMCNicolauBA Bayesian approach to investigate life course hypotheses involving continuous exposuresInt. J. Epidemiol.20184716231635299123846208282MadathilSBlaserCNicolauBRichardHParentMEDisadvantageous socioeconomic position at specific life periods may contribute to prostate cancer risk and aggressivenessFront. Oncol.20188515304986926249302BuscotMJThe combined effect of common genetic risk variants on circulating lipoproteins is evident in childhood: a longitudinal analysis of the Cardiovascular Risk in Young Finns StudyPLoS ONE201611e0146081267312814701181ArmstrongMKAssociation of non-high-density lipoprotein cholesterol measured in adolescence, young adulthood, and mid-adulthood with coronary artery calcification measured in mid-adulthoodJAMA Cardiol.20216661668335024547841578CarpenterBStan: a probabilistic programming languageJ. Stat. Softw.201776132365683349788645Takaoka, M et al. Code for ‘Early intermittent hyperlipidaemia alters tissue macrophages to fuel atherosclerosis’. Zenodohttps://doi.org/10.5281/zenodo.13137366 (2024).Extended data figures and tables
Induction of early intermittent hyperlipidaemia in mice.
a, Experimental set up. LDL receptor-deficient (Ldlr−/−) male mice on normal diet were fed a Western-type diet (WD) for 6 weeks (w) either continuously (cWD) or intermittently (iWD). b, c, Calculation of plasma cholesterol levels in mice subjected to iWD over a period of 6 weeks. Here, examples are given for measurements that started at the end of week 9 and the beginning of week 10 and lasted until end of week 15. b, shows cholesterol measurement over 14 consecutive days (n = 4 per time point). c, shows weekly cholesterol measurements in Ldlr−/− mice under cWD (during the last 6 weeks, n = 6) and iWD (n = 6). d, weekly LDL cholesterol measurements in Ldlr−/− mice under cWD (during the last 6 weeks, n = 6) and iWD (n = 6). e-h, Body weight (n = 6 per group) (e), systolic blood pressure (f), heart rate (g), and plasma corticosterone levels (h) in randomly selected cWD (n = 5) or iWD (n = 6) fed Ldlr−/− mice at the end of the experiment. Mean ± s.e.m.; two-tailed unpaired t-test (e-h). P values are indicated on the graphs.
Source Data
Early intermittent hyperlipidaemia accelerates atherosclerosis in mice and increases plaque inflammation and necrosis.
a, LDL receptor-deficient (Ldlr−/−) female mice were fed a Western-type diet (WD) for 6 weeks (w) either continuously (cWD) or intermittently (iWD). Representative photomicrographs (left panels) and mean atherosclerotic plaque size (right panel) in the aortic sinus of the 2 groups of mice are shown (n = 10 cWD and n = 11 iWD). b, Ldlr−/− female mice irradiated and reconstituted with WT bone marrow and subjected to 6 weeks of cWD (n = 10) or iWD (n = 8). Representative photomicrographs (left panels) and mean atherosclerotic plaque size (right panel) in the aortic sinus of the 2 groups of mice are shown. c-g, Ldlr−/− male mice were put on 6 weeks cWD (n = 6) versus 6 weeks iWD (n = 6). Mean plaque area staining positively for MOMA2 (c), CD3 (d), acellular debris (necrosis) (e), αSMA (f), and Sirius red (collagen) (g) in one set of a series of experiments. Mean ± s.e.m.; two-tailed unpaired t-test. P values are indicated on the graphs.
Source Data
Various regimens of early intermittent hyperlipidaemia accelerate atherosclerosis in mice.
a, Experimental set up. In group 1, LDL receptor-deficient (Ldlr−/−) male mice first received 3 intermittent weeks of Western diet (WD) at the beginning of the experiment, then 3 weeks of continuous WD just before the end of the experiment (n = 6). Group 2 mice received 6 weeks of intermittent WD (n = 6). b, Mean atherosclerotic plaque size in the aortic sinus of the 2 groups of mice (n = 6 per group). c-j, Experimental set up (c). Ldlr−/− male mice were put on 12 weeks of cWD (n = 5) versus 12 weeks of iWD (n = 6). d, Mean atherosclerotic plaque size on en face aorta in the 2 groups of mice (n = 5 to 6 per group). Mean plaque area staining positively for MOMA2 (e), CD3 (f), αSMA (g), Sirius red (collagen) (h), and acellular debris (necrosis) (i), (n = 5 per group). j, Representative photomicrographs of TUNEL staining (red) and CD68 staining (green) (scale bar = 10 μm) and quantification of the proportion of internalized apoptotic cells in plaques from the 2 groups of mice (n = 5 to 6 per group). k, Experimental set up. Ldlr−/− male mice were put on 6 weeks iWD starting at 6 or 22 weeks of age (n = 11 and n = 9 mice, respectively). l, Mean atherosclerotic plaque size in the 2 groups of mice. Mean ± s.e.m.; two-tailed unpaired t-test (b, d-j, l). P values are indicated on the graphs.
Source Data
Early intermittent hyperlipidaemia, gut microbiota, and accelerated atherosclerosis in mice.
a, Experimental set up. LDL receptor-deficient (Ldlr−/−) male mice were fed a normal (ND, control, n = 4) diet or a Western-type diet (WD) for 3 or 6 weeks (w) either continuously (cWD) or intermittently (iWD) (n = 8 cWD 3w, n = 10 iWD 3w, n = 8 cWD 6w, n = 6 iWD 6w). Faecal samples were collected during and at the end of the experiment for analysis of gut microbiota using 16S rRNA sequencing (b-c). b, Principal coordinate analysis reveals distinct separation in microbiota composition due to diet. c, WD-induced changes in microbiome composition on the phylum level (mean relative abundance). d, Experimental set up. Ldlr−/− male mice were put on 6 weeks cWD versus 6 weeks iWD. Treatment with oral antibiotics (see Methods) to deplete the gut microbiota was started at the beginning of week 9. Faecal samples collected at the end of week 8 (before treatment with antibiotics), and at the end of the experiment (after treatment with antibiotics) were analysed using 16S rRNA sequencing (n = 10 per group) (e-g). Bacterial quantification via 16S qPCR (e), principal coordinate analysis (f), and proportional abundance of bacteria (g) demonstrate a great impact of the antibiotic treatment (ABx) on microbiome abundance and composition. h, Plasma levels of total cholesterol at the end of the experiment (n = 10 per group). i, Ldlr−/−/Rag2−/− male mice were put on 6 weeks of cWD (n = 8) versus iWD (n = 7). Calculation of plasma cholesterol levels over the whole period of the experiment. Mean ± s.e.m.; two-tailed unpaired t-test (h, i). P values are indicated on the graphs.
Source Data
Impact of early intermittent hyperlipidaemia on atherosclerosis and resident arterial macrophages in mice.
a, LDL receptor-deficient (Ldlr−/−) female mice were fed a Western-type diet (WD) for 3 weeks (w) either continuously (cWD) or intermittently (iWD). b, Mean atherosclerotic plaque size in the aortic sinus of the 2 groups of mice is shown (n = 5 cWD and n = 6 iWD). c, Experimental set up. Cx3cr1creERT2/+Rosa26LSL-TomatoLdlr−/− mice on normal diet (ND) received tamoxifen treatment (TAM) to induce Tomato expression. The mice were rested for 2 weeks on normal diet to allow for clearance of labelled blood monocytes. The mice were then placed on 3 weeks of cWD versus 3 weeks of iWD. Mice were killed at the end of the experiment for analysis of fate-mapped Tomato+ resident arterial macrophages (n = 15 cWD; n = 21 iWD). d, Representative photomicrographs showing staining for macrophages (CD45+ CD68+) to quantify the numbers of intimal MacAIR macrophages. e, Total CD45+ CD68+ macrophage count in the aortic arch of the 2 groups of mice at the end of the experiment. f, Percentage of Tomato+ CD45+ CD68+ macrophages in the intima (MacAIR) (among intimal CD45+ CD68+ macrophages) of the 2 groups of mice. g, Number of Tomato+ CD45+ CD68+ intimal (MacAIR) macrophages. h, Number of Tomato+ adventitial (adv.) macrophages. n = 15 cWD and n = 21 iWD (e-h). Mean ± s.e.m.; two-tailed unpaired t-test (b, e-h). P values are indicated on the graphs.
Source Data
Impact of macrophage expression of SPIC on the development of atherosclerosis in mice.
a, Dotplot (n = 1 to 8) with mean (red bar) for common transcription factor (TF) binding motifs enriched in significant DEGs and their corresponding normalized enrichment score (NES) difference with nonsignificant DEGs. The order of the TFs corresponds to the mean of Padj value. Two-tailed proportion z-test performed then adjusted by Benjamini-Hochberg (BH) correction. SPIC is highlighted. Detailed information is provided in Supplementary Table 4. b, Barplot of selected 6 Gene Ontology (Biological Process) pathways for significant DEGs enriched in SPIC binding motifs. Gradient blue colour represents the padj values (one-sided Fisher exact test then adjusted by BH correction), and the size of the bar represents the number of genes that contribute to the corresponding pathways. Detailed information is provided in Supplementary Table 5. c, Ldlr−/− female mice were lethally irradiated and reconstituted with bone marrow (BM) from Lyz2Cre+/−Spicflox/flox mice (n = 27) and Lyz2Cre+/−Spicflox/+ controls (n = 22) (see Methods). After 5 weeks of recovery mice were put on cWD for 8 weeks to assess the role of myeloid-specific expression of SPIC on atherogenesis. Mean atherosclerotic plaque size in the aortic sinus of the 2 groups of mice is shown. d, Mean atherosclerotic plaque size in the en face aorta of the 2 groups of mice (n = 12 Lyz2Cre+/−Spicflox/flox and n = 10 Lyz2Cre+/−Spicflox/+). e, Mean plasma total cholesterol levels at the end of the experiment (n = 24 Lyz2Cre+/−Spicflox/flox and n = 25 Lyz2Cre+/−Spicflox/+). f, Ldlr−/− female mice were lethally irradiated and reconstituted with Lyz2Cre+/−Spicflox/flox bone marrow. After recovery, mice were put on 6 weeks of cWD (n = 9) or 6 weeks iWD (n = 7). Mean atherosclerotic plaque size in the aortic sinus of the 2 groups of mice. Mean ± s.e.m.; two-tailed unpaired t-test (c-f). P values are indicated on the graphs.
Source Data
Time-dependent changes in the expression of prototypical genes associated with distinct arterial macrophage subsets in <italic>Ldlr</italic><sup><italic>−/−</italic></sup> mice subjected to continuous western diet.
a, A heatmap of selected markers for each macrophage subtype (Res = Resident; MacAIR/Trem2; IFN/Inflam = interferon/inflammatory) derived from analysis of the mouse public data set on scRNASeq of atherosclerotic lesions in Ldlr−/− mice subjected to various durations of continuous high fat diet (HFD). Hierarchical clustering is shown based on log2FoldChange. b, Barplot for three macrophage subtype proportions for each duration of continuous HFD.
Deletion of LYVE1<sup>+</sup> resident macrophages in atherosclerotic mice.
a, Representative examples of staining for LYVE1, CD68, SMA and DAPI in arteries of Apoe−/−Csf1rflox/flox and Apoe−/−Lyve1Cre+/−Csf1rflox/flox mice. b, Quantification of adventitial LYVE1+ macrophages in the 2 groups of mice (n = 7 Csf1rflox/flox and n = 4 Lyve1Cre+/−Csf1rflox/flox). c, Plasma levels of total cholesterol at the end of the experiment in the different groups of mice fed on cWD for 18 weeks (n = 15 Apoe−/−Csf1rflox/flox and n = 15 Apoe−/−Lyve1Cre+/−Csf1rflox/flox mice. d, Quantification of SMA+ area in lesion cap of the aortic sinus (n = 7 per group). e, Quantification of collagen I area in the plaques of the aortic sinus (n = 7 per group). Mean ± s.e.m.; two-tailed unpaired t-test (b-e). P values are indicated on the graphs. f, Representative photomicrographs (scale bar = 50 µm) of plaques stained with anti-CD68, DAPI, and TUNEL. g, Number of free apoptotic (TUNEL+DAPI+) cells in aortic lesions of Apoe−/−Csf1rflox/flox (n = 4) and Apoe−/−Lyve1Cre+wtCsf1rflox/flox (n = 5) were quantified (see Methods). Mean ± s.e.m.; two-tailed unpaired t-test. P values are indicated on the graphs. h, Experimental design. Csf1rflox/flox and Lyve1Cre+/−Csf1rflox/flox female mice were injected with AAV8-D377Y-mPCSK9 to induce hyperlipidaemia. The mice were then subjected to 6 weeks of cWD or 6 weeks of iWD. Csf1rflox/flox mice (n = 8 cWD and n = 9 iWD), and Lyve1Cre+/−Csf1rflox/flox mice (n = 11 cWD and n = 11 iWD) (see Methods). i, Representative values of plasma cholesterol levels at the indicated time points. Mean ± s.e.m. (n = 5 per group).
Source Data
Gene ontology enrichment analysis on 220 human coronary artery disease causal genes.
Gene ontology enrichment analysis of biological processes (BP) was conducted on 220 coronary artery disease causal genes prioritized in Aragam et al.33. A total of 178 out of 220 causal genes contributed to the GO enrichment analysis, and 867 enriched BP were identified (see Supplementary Table 11). The most enriched pathways and their corresponding causal gene network are plotted here using cnetplot function in R. Significant differentially expressed genes in our RNASeq data of aortic macrophages from iWD vs cWD (see Supplementary Table 1) are shown here (red: upregulated in iWD, blue: downregulated in iWD). NRP1 is represented in the most enriched biological pathways, which are related to cell migration, actin filament organization, and vascular development.
Source Data
Deletion of NRP1 expression in myeloid cells of atherosclerotic mice.
a, Experimental set up. LDL receptor-deficient (Ldlr−/−) male mice were lethally irradiated and reconstituted with Lyz2Cre+/−Nrp1flox/flox (Nrp1−/−) or Lyz2Cre+/−Nrp1+/+ (WT) bone marrow. After recovery, mice were put on a Western-type diet (WD) for 6 weeks (w) either continuously (cWD) or intermittently (iWD). b, Nrp1 gene expression by qPCR in peritoneal macrophages at the end of the experiment (n = 12 per group). Fold change is relative to Rplp0 (36B4) expression. c, Mean plasma total cholesterol levels at the end of the experiment in the 4 groups of mice (n = 8 cWD WT and Nrp−/−, n = 7 iWD WT, and n = 9 iWD Nrp−/−). d, Bone-marrow-derived macrophages were generated from Lyz2Cre+/−Nrp1+/+ (WT) and Lyz2Cre+/−Nrp1flox/flox mice and cultured with PKH26-labelled apoptotic Jurkat cells (see Methods). e, f, Quantification of the percentage of macrophages (M0 in e and M1 in f) with internalized apoptotic cells; two separate experiments with n = 3 biological replicates/group. Mean ± s.e.m.; two-tailed unpaired t-test. P values are indicated on the graphs.
Source Data
Supplementary information
Supplementary Methods and Supplementary Data. Description of the methods for diagnostics assessments of the Bayesian relevant life-course exposure models, as well as Supplementary Figs. 1–10, showing the corresponding results.
Reporting Summary
Supplementary Tables 1–28.
Peer Review File
Source data
Source Data Figs. 1–4 and Source Data Extended Data Figs. 1–6 and 8–10
Extended data
is available for this paper at https://doi.org/10.1038/s41586-024-07993-x.
Supplementary information
The online version contains supplementary material available at https://doi.org/10.1038/s41586-024-07993-x.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.