C-type lectin receptor CLEC4A2 promotes tissue adaptation of macrophages and protects against atherosclerosis

Macrophages are integral to the pathogenesis of atherosclerosis, but the contribution of distinct macrophage subsets to disease remains poorly defined. Using single cell technologies and conditional ablation via a LysMCre+ Clec4a2flox/DTR mouse strain, we demonstrate that the expression of the C-type lectin receptor CLEC4A2 is a distinguishing feature of vascular resident macrophages endowed with athero-protective properties. Through genetic deletion and competitive bone marrow chimera experiments, we identify CLEC4A2 as an intrinsic regulator of macrophage tissue adaptation by promoting a bias in monocyte-to-macrophage in situ differentiation towards colony stimulating factor 1 (CSF1) in vascular health and disease. During atherogenesis, CLEC4A2 deficiency results in loss of resident vascular macrophages and their homeostatic properties causing dysfunctional cholesterol metabolism and enhanced toll-like receptor triggering, exacerbating disease. Our study demonstrates that CLEC4A2 licenses monocytes to join the vascular resident macrophage pool, and that CLEC4A2-mediated macrophage homeostasis is critical to combat cardiovascular disease.

M acrophages support organ function and immune responses to tissue damage and disease 1 . Ontogeny and organ-specific tissue signals drive the phenotype and function of resident macrophages through activation of specific transcription factors 2,3 . In some organs, monocytes replenish the resident macrophage pool, adapting their transcriptional signatures to the tissue microenvironment 4,5 . The arterial wall is structurally divided into three layers: the intima, the media, and the adventitia. In the steady state, two distinct macrophage populations reside in the intima 6 and adventitial layers 7 . Both require colony-stimulating factor 1 (CSF1) for their survival and originate from embryonic precursors 6,8 . The majority of lymphatic-vessel endothelial hyaluronan receptor-1 (Lyve1) + adventitial macrophages are replenished by bone marrow (BM)derived monocytes after birth 8 , while arterial intimal resident macrophages are replaced by monocytes during early atherosclerosis 6 . The former regulate collagen production in smooth-muscle cells and prevent unfavourable arterial remodelling 7 while the latter are responsible for cholesterol uptake and initiation of vascular inflammation by producing interleukin-1β (IL-1β) 6 .
Atherosclerosis is the most prevalent chronic inflammatory disease in man and the main driver of cardiovascular disease 9 . During atherogenesis, monocytes are recruited into the subendothelium of the artery, where they give rise to lesional macrophages 10 . Single cell technologies have revealed three broad macrophage states in murine atherosclerosis, comprising resident, inflammatory, and lipid-driven subsets [11][12][13][14] . Genetic lineage tracing studies showed that monocytes have the capacity to differentiate into most macrophage subpopulations during atherosclerosis 15 . The fate of monocyte-derived macrophages is influenced by cues from surrounding tissue, as indicated by the conversion of monocytes into resident-like macrophages during disease regression 16 . The mechanisms of macrophage tissue adaptation in vascular health and disease are currently unknown.
CLEC4A, also known as dendritic cell immunoreceptor (DCIR), is one of the few C-type lectin receptors (CLR) that contain an immunoreceptor tyrosine-based inhibition motif (ITIM) 17 . The ITIM domain recruits Src homology region 2 domain-containing protein tyrosine phosphatase 1 (SHP1) and SHP2, that downmodulate tyrosine phosphorylation signalling 18 . Clec4a polymorphisms are positively associated with human autoimmune diseases 19,20 . CLEC4A2, the murine homologue of CLEC4A, protects from experimental arthritis and autoimmune encephalomyelitis by negatively regulating activation of dendritic cells (DC) 21,22 . Mice lacking Clec4a2 are resistant to excessive inflammation and tissue damage during hepatitis, cerebral malaria, and tuberculosis [23][24][25] . Collectively, this evidence implies an immunomodulatory role for CLEC4A2, however, its functions in macrophage biology and atherosclerosis are unknown.
In this paper, we reveal that the expression of CLEC4A2 identifies a subset of vascular resident macrophages with atheroprotective functions. CLEC4A2 promotes appropriate tissue adaptation of vascular macrophages ensuring homeostasis of the vessel wall and thus combating atherosclerosis.

Results
The expression of Clec4a2 is associated with a homeostatic signature in vascular resident macrophages. We previously identified two broad subsets of macrophages in the atherosclerotic aorta, comprising the CD206 + CD169 + and CD11c + CD44 + subsets by mass cytometry 12 . Here, we compared the composition of vascular-macrophage subsets in the steady state and atherosclerosis using mass cytometry ( Supplementary Fig. 1a). In the aorta of C57BL/6 J wild type (WT) mice, CD206 + CD169 + macrophages represented 96.3 ± 0.4% of total macrophages ( Supplementary Fig. 1b). Atherogenesis promoted a reduction in the representation of this aortic macrophage population to 73.7 ± 2.0% in high fat diet (HFD)-fed apolipoprotein E-deficient (ApoE) −/− mice. Conversely, CD11c + CD44 + macrophages significantly increased with atherosclerosis from 3.8 ± 0.4% in WT to 26.3 ± 2.1% in ApoE −/− (HFD) mice.
The spatial distribution of these macrophage populations was assessed by microscopy and mass cytometry. CD206 + macrophages were mainly detected in the adventitia in healthy and atherosclerotic aortas ( Supplementary Fig. 1c,d), whereas macrophages co-expressing CD11c and CD64 were predominantly located in the intima of the atherosclerotic aorta ( Supplementary  Fig. 1d). To clarify the macrophage heterogeneity across arterial compartments, we performed mass cytometry on single cells from separated intima-media and adventitial layers of aortas of ApoE −/− mice ( Supplementary Fig. 1e). Unbiased tSNE clustering demonstrated that adventitial macrophage clusters expressed CD206, CD169, and Lyve1, while macrophage clusters in the intima-media compartment expressed CD11c and CD44 (Supplementary Fig. 1f,g,h,i).
Next, we examined the cluster-specific gene lists to identify a regulator of a homeostatic gene signature within the resident macrophage clusters. Among the differentially expressed genes (P < 0.05), Clec4a2 was one of the top genes that distinguished two Lyve1 + Mrc1 + Pf4 + resident macrophage subsets (C1 and C5) from the activated resident subset C0 (Fig. 1c). CLEC4A2 expression in C1 and C5 was negatively associated with the average expression of 170 macrophage-activation genes, while a weaker correlation was noted in C3 that expressed lower levels of Clec4a2 (Supplementary Fig. 2c

Cd209a
Myeloid cell defining markers Mac cluster genes cellular response to lipopolysaccharide inflammatory response neutrophil chemotaxis chemokine-mediated signaling pathway cellular response to tumour necrosis factor cellular response to interleukin-1 lymphocyte chemotaxis cellular response to interferon-gamma monocyte chemotaxis response to wounding innate immune response inflammatory response heme catabolic process receptor-mediated endocytosis heme oxidation antigen processing and presentation of antigen positive regulation of angiogenesis cellular response to iron ion phagocytosis, engulfment phagocytosis, recognition clathrin-dependent endocytosis positive regulation of phagocytosis cellular response to lipopolysaccharide complement activation response to wounding regulation of Cdc42 protein signal transduction positive regulation of myoblast differentiation receptor-mediated endocytosis positive regulation of smooth muscle cell apoptotic process regulation of blood vessel endothelial cell migration positive regulation of macrophage derived foam cell differentiation blood vessel remodeling phagocytosis, engulfment positive regulation of cholesterol efflux antigen processing and presentation of antigen complement activation, classical pathway response to interferon-gamma regulation of macrophage differentiation neutrophil chemotaxis positive regulation of T cell differentiation innate immune response macrophage migration inhibitory factor signaling pathway phagocytosis, engulfment -Log 10 (P value) -Log 10 (P value) - CLEC4A2 is highly expressed by resident macrophages in the aorta, heart, and lung. The expression of CLEC4A2 and human CLEC4A was reported in neutrophils, monocytes, and DCs in lymphatic organs and blood in WT mice 29 and in human blood 17 . We profiled the cell surface expression of CLEC4A2 in cells from the aorta, heart, and lungs of ApoE −/− and WT mice using mass cytometry. CLEC4A2 was not expressed by nonimmune CD45 − cells ( Supplementary Fig. 3a).  Fig. 3b,c,d). It was detectable at very low levels in other vascular myeloid cells including CD11c + macrophages, monocytes, and neutrophils. CLEC4A2 was also expressed by cardiac resident and alveolar macrophages ( Supplementary Fig. 3d). In keeping with these findings, purified aortic and alveolar macrophages expressed higher Clec4a2 gene expression compared with DCs (Supplementary Fig. 3e). Next, we assessed the regulation of Clec4a2 expression in macrophages by tissue-derived factors and inflammatory stimuli. Ex vivo conditioned media from murine aortas and hearts significantly enhanced the expression of Clec4a2 in CSF1-cultured bone marrow-derived macrophages (BMDMs), as well as the expression of Mrc1 (CD206) ( Supplementary Fig. 3f). Conversely, stimulation with lipopolysaccharide (LPS) and IFNγ significantly inhibited Clec4a2 expression in CSF1-BMDMs, while IL-4 and IL-13 did not alter its expression ( Supplementary  Fig. 3g).
To understand the relationship between the human homologue of CLEC4A2, CLEC4A, and markers of resident macrophages in human atherosclerotic plaques, we interrogated transcriptomics data from human carotid arterial tissues in the Tampere Vascular Study biobank. We performed Spearman correlation between Clec4a and macrophage genes and found that Clec4a gene expression was strongly correlated with markers of resident and alternatively activated macrophages (Cd163, Csf1r, Cd200r1, Clec7a, Tlr7, Scarb1, Cd68, Msr1, Mrc1, and Ms4a4a) (Supplementary Fig. 4). This dataset shows that a close relationship between Clec4a expression and macrophages is also conserved within human vascular tissues.
Diphtheria toxin-mediated cell ablation in LysM Cre+ Clec4a2flox/DTR mice targets CLEC4A2 + tissue macrophages. In order to study the role of CLEC4A2 + macrophages in vascular disease, we generated a mouse strain Clec4a2 flox/DTR carrying a construct containing neomycin, diphtheria-toxin receptor (DTR), and eGFP upstream of exon 8 of the Clec4a2 gene ( Supplementary  Fig. 5a). This mouse line was further crossed with LysM Cre knock-in mice so that diphtheria toxin (DT)-mediated ablation in cells expressing both genes could be achieved while preserving CLEC4A2 + cDC2, which expressed CLEC4A2 in aortic tissues. The mouse line was viable and displayed no differences in body weight (Clec4a2 flox/DTR : 28.6 ± 0.9 g (n = 28) and LysM Cre+ Cle-c4a2 flox/DTR : 29.4 ± 0.4 g (n = 33)). The expression of eGFP was detected in neutrophils and to a lesser extent in Ly6C + monocytes in the BM and blood ( Supplementary Fig. 5b), but not in other myeloid cells. In keeping with the data in Supplementary  Fig. 3c, gene expression of Clec4a2, egfp, and Hbegf (DTR) was higher in aortic and alveolar macrophages than DCs and Ly6C + monocytes from the same tissues ( Supplementary Fig. 5c).
Aorta sections immunostained with an antibody against CD64 confirmed macrophage ablation in the adventitia of LysM Cre+ Clec4a2 flox/DTR mice 24 h after administration of one dose of DT (Fig. 2a). Flow cytometry analysis showed that CLEC4A2 + aortic macrophages were ablated at day 1 (65% reduction), day 2 (80%), and day 3 (75%) post DT injection (Fig. 2b). The aortic macrophage pool was mostly reconstituted at day 5 and fully recovered at day 7, day 14, and day 70 post ablation. An influx of Ly6C + monocytes was observed post DT administration, peaking at day 3 and waning at day 5, in line with the kinetics of macrophage replenishment (Fig. 2c). No significant DT-mediated depletion was observed in CD11c + macrophages and neutrophils in the artery at all time points (Fig. 2d,e). CLEC4A2 expression was gradually restored in vascular macrophages from day 7 post ablation ( Fig. 2f).
We assessed the effects of DT administration on circulating cells in the BM, blood, and spleen. DT caused a temporary ablation of neutrophils in the BM at 24 and 30 h but the population was restored by 48 h (Supplementary Fig. 5d). In the blood, neutrophils were not significantly reduced by DT treatment nor in the spleen ( Supplementary Fig. 5e,f). No depletion was observed in Ly6C + monocytes in the BM, blood, or spleen ( Supplementary Fig. 5g-i).
We studied the feasibility and selectivity of repeated DT treatment by flow and mass cytometry on vascular leucocyte populations from saline or DT-treated LysM Cre+ Clec4a2 flox/DTR mice. DT-treated mice showed selective vascular macrophage ablation (92.7 ± 1.6%), without depleting aortic monocytes, cDC1, cDC2, neutrophils, eosinophils, B cells, or T cells ( Fig. 2g- To trace whether Ly6C + monocytes repopulate the vascular macrophage pool post depletion, we utilised monocyte-adoptive transfer. Monocytes isolated from the BM of Cx3cr1 eGFP+ CD45.1 + CD45.2 + mice were transferred into LysM Cre+ Clec4a2 flox/DTR mice (steady state) and hypercholesterolemic low-density lipoprotein-receptor (Ldlr) −/− mice (CD45.1) after DT treatment ( Supplementary Fig. 6a-c). The transferred monocytes were traceable in the blood ( Supplementary Fig. 6d,e), and the aortas in the steady state and atherosclerotic conditions ( Supplementary Fig. 6f,g). The reconstituted vascular macrophages from both donor and host origins expressed CLEC4A2 (Supplementary Fig. 6f-g). We detected the presence of the donor-origin GFP + macrophages in the adventitia of LysM Cre+ Clec4a2 flox/DTR and Ldlr −/− chimeric mice ( Supplementary  Fig. 6h,i), in keeping with previous parabiosis 8 and fatemapping 15 studies. In addition, we also confirmed that ablation and replenishment of macrophages occurred in the adventitia of healthy and atherosclerotic mice, leaving lesional macrophages intact ( Supplementary Fig. 6j,k).
To investigate the contribution of blood Ly6C + monocytes and local proliferation to replenishment of vascular macrophages, we performed bromodeoxyuridine (BrdU)-proliferation assays on BM chimeras where the host-derived cells and donor cells were traceable by CD45.1 and CD45.2 expression. CD45.1 + hosts were reconstituted with the BM from LysM Cre+ Clec4a2 flox/DTR (CD45.2) mice and received DT or saline followed by BrdU injections ( Supplementary  Fig. 6l). The majority of vascular macrophages in the saline group were donor (CD45.2 + )-derived but a small population of host (CD45.1 + ) cells (radio-resistant) were detectable, which represented approximately 10% of aortic macrophages ( Supplementary Fig. 6m). In the DT group, proliferation rates increased from 9.0 ± 1.2% to 43.9 ± 1.9% in the donor origin and 1.2 ± 0.3% to 8.3 ± 0.8% in the host macrophages. Collectively, our findings indicate that the replenished resident macrophage pool has a predominant contribution of donor cells.
CLEC4A2 exerts a protective function in atherosclerosis by maintaining myeloid homeostasis. After demonstrating the protective role of CLEC4A2 + resident macrophages in vascular disease, we sought to assess how this CLR protects from atherosclerosis. Clec4a2 −/− mice displayed no discrepancy in numbers and frequencies of cells in lymphatic tissues and blood 21 . We found that WT and Clec4a2 −/− littermate mice showed no changes in immune cell populations in aortas, lungs and BM in the steady state (Supplementary Table 3). To assess the role of CLEC4A2 in hyperlipidemic conditions, we generated ApoE −/− and ApoE −/− Clec4a2 −/− littermate mice and fed them a chow diet or a HFD for 12 weeks to allow for the development of atherosclerosis ( Fig. 4a; Supplementary Fig. 7a). No significant differences in body weight or serum cholesterol levels were detected between the two groups (Supplementary Table 4). In the chow-fed mice, we detected no significant differences between the two genotypes in lesion size, arterial size, or necrotic core content in the aortic root ( Supplementary Fig. 7a-d). Expression of the mononuclear phagocyte marker CD68 in aortic root lesions was also not altered by CLEC4A2 deficiency in the chow-diet group ( Supplementary Fig. 7e). However, we found increased lesion formation in the aortic arch and abdominal aorta of ApoE −/− Clec4a2 −/− mice fed a HFD compared to their ApoE −/− counterparts (Fig. 4b). In addition, Clec4a2 deficiency significantly aggravated lesion development in the aortic root (Fig. 4c). ApoE −/− Clec4a2 −/− mice exhibited an enlarged arterial diameter, indicating outward arterial remodelling (Fig. 4d). Furthermore, ApoE −/− Clec4a2 −/− mice showed a more vulnerable plaque phenotype compared with ApoE −/− mice, with a threefold increase in the necrotic core area (Fig. 4e). ApoE −/− Clec4a2 −/− mice showed increased immuno-positive area for the mononuclear phagocyte markers CD68, CD11c, and CD64 within atherosclerotic lesions ( Fig. 4f-h), indicating a higher macrophage content of the lesions. Next, we investigated whether CLEC4A2 deficiency altered the immune cell composition in the aortas of ApoE −/− and ApoE −/− Clec4a2 −/− mice using mass cytometry ( Fig. 4i; Supplementary  Fig. 7f). Chow-fed ApoE −/− Clec4a2 −/− mice showed a significant reduction in vascular resident macrophages while no changes were detected in other leucocyte populations (Supplementary Fig. 7g,h). When mice developed advanced disease following feeding of a HFD, we consistently found a significant decrease in resident macrophages (Fig. 4j,k). In contrast to the earlier stages of atherogenesis, an increase of CD11c + macrophages, Ly6C + monocytes, and neutrophils was detected in the aorta of HFD-fed ApoE −/− Clec4a2 −/− mice (Fig. 4j,k). No differences in other vascular leucocyte populations were found between the two genotypes. Furthermore, the frequency of immune cells in the blood, BM, spleen and para-aortic lymph nodes was not altered in HFD-fed ApoE −/− and ApoE −/− Clec4a2 −/− mice, except for a small reduction in Ly6C + monocytes in the blood of ApoE −/− Clec4a2 −/− mice (Supplementary Table 5). Taken together, CLEC4A2 deficiency results in excessive loss of the vascular resident macrophage pool followed by an increase in lesion size, necrotic core formation, and accumulation of inflammatory myeloid cells.
A similar reconstitution of myeloid cells by WT and Clec4a2 −/− cells was observed in the atherosclerotic aorta (Fig. 6g). Cells deficient in CLEC4A2 displayed a significantly reduced capacity to replenish the vascular resident macrophage pool compared with WT cells (Fig. 6h), but preferentially accumulated as Ly6C + monocytes, cDC2, and CD11c + macrophages (Fig. 6i-k), the latter predominantly emerging during atherogenesis (Supplementary Fig. 1b). In summary, CLEC4A2 promotes the maintenance of the CSF1-dependent resident macrophage pool in the healthy and atherosclerotic artery, limiting the accumulation of pro-inflammatory myeloid subsets in atherogenesis.  We then sought to investigate whether the proliferation capacity of vascular macrophages was affected by CLEC4A2 deficiency by injecting BrdU in the steady state and in atherosclerotic mice. We detected that there were no apparent differences in BrdU incorporation in the resident macrophages from WT and Clec4a2 −/− mice in the steady state (Supplementary Fig. 8a-c). However, differential effects of genetic deletion were observed in the atherosclerotic mice. ApoE −/− Clec4a2 −/− mice compared with their ApoE −/− littermate mice showed a reduction in proliferation in resident macrophages and an increase in CD11c + macrophages ( Supplementary Fig. 8d-h). These findings support our claim that CLEC4A2 drives loss of the resident macrophage pool while promoting lesional macrophage accumulation during atherosclerosis. ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-27862-9 CLEC4A2 protects from atherosclerosis by controlling homeostatic and anti-atherogenic functions in vascular resident macrophages. To understand how CLEC4A2 deficiency affects the vascular myeloid landscape during atherogenesis at the single-cell level, we performed scRNA-seq on live CD45 + CD11b + myeloid cells isolated from aortas of ApoE −/− and ApoE −/− Clec4a2 −/− mice fed a HFD for 12 weeks. Clustering was performed using Seurat and similar myeloid cell populations were identified as described in Fig. 1 (Fig. 7a; Supplementary Data 4). scRNA-seq in Fig. 7b revealed a decrease in the resident macrophage cluster (C1) in ApoE −/− Clec4a2 −/− mice, in line with the finding by mass cytometry (Fig. 4j). Clustering did not identify CD209b + Lyve1 + macrophages as a separate cluster in this dataset (compared with Fig. 1), possibly due to the considerable decrease of the number of resident macrophages in ApoE −/− Clec4a2 −/− mice. Compared with ApoE −/− mice, an increase in Itgax (CD11c) expressing macrophage clusters (C2 and C6), proliferating macrophages (C8), Ly6C + monocytes (C5), and cDC2 (C7) was observed in the aorta of ApoE −/− Clec4a2 −/− mice.
To verify the functional significance of these transcriptomic changes in CLEC4A2-deficient macrophages, we tested the ability of Clec4a2 −/− CSF1-BMDMs to respond to toll-like receptor (TLR) ligands. Clec4a2 −/− macrophages produced higher levels of the pro-inflammatory cytokine IL-6 when stimulated by FSL-1 (TLR2 ligand) or LPS (TLR4 ligand) compared with the WT cells (Fig. 7d,e), demonstrating that CLEC4A2 inhibits TLR signalling in macrophages.
Second, we tested how CLEC4A2 modulates the ability of macrophages to uptake and process oxidised LDL (oxLDL). CSF1-BMDMs from WT and Clec4a2 −/− mice were treated with oxLDL for 24 h. Greater lipid levels accumulated in the cytoplasm of Clec4a2 −/− BMDMs compared with the WT counterpart (Fig. 7f,g). We then examined the reverse cholesterol-transport capacity, and found that upon oxLDL loading Clec4a2 −/− macrophages were less capable of processing and secreting cholesterol than WT cells (Fig. 7h). Consistently, oxLDL-laden Clec4a2 −/− macrophages showed lower expression of cholesterolefflux genes (Abca1, Abcg1) relative to the WT cells (Fig. 7i). Since CLEC4A2 signals through its ITIM with phosphatase SHP1, we tested whether SHP1 inhibition can phenocopy the effect of CLEC4A2 deficiency on oxLDL uptake. Suppression of SHP1 activity increased lipid accumulation in WT BMDMs but not in Clec4a2 −/− macrophages (Fig. 7j) and reduced cholesterol efflux genes in WT BMDMs (Fig. 7k). Our results demonstrate that CLEC4A2 prevents pro-atherogenic macrophage behaviour by promoting cholesterol handling in a SHP1-modulated manner.

Discussion
In our study, we show that the C-type lectin receptor CLEC4A2 identifies a subpopulation of Lyve1 + vascular resident macrophages endowed with athero-protective functions (Figs. 1, 2, 3). CLEC4A2 licences monocyte differentiation to vascular resident macrophages and their acquisition of homeostatic properties by promoting CSF1-driven over CSF2-driven monocyte-fate decisions and proliferation, avoiding losses of the vascular resident macrophage pool (Figs. 5, 6, Supplementary Fig. 8). During atherogenesis, CLEC4A2 limits TLR signalling and promotes cholesterol efflux in vascular macrophages (Fig. 7), which in turn protects from atherogenesis (Fig. 4). Our findings advance the understanding of the mechanisms of macrophage tissue adaptation to the vascular niche through CLEC4A2 and highlight the role of CLEC4A2 + macrophages in atheroprotection.
Vascular macrophages are heterogeneous and the contribution of each subset to atherogenesis remains elusive. By using mass cytometry, we show that the intima and adventitia of the atherosclerotic aorta have a different immune landscape with two distinct macrophage compartments (Supplementary Fig. 1). Our scRNA-seq data on myeloid cells in murine atherosclerotic aortas provide a highresolution landscape of the myeloid lineage, identifying 12 cell populations including 8 macrophage clusters (Fig. 1). These macrophage subsets include previously identified clusters such as TREM2 high 11 , inflammatory 13 , IFN-inducible 15 , and proliferating 15 . Our analysis uncovers significant heterogeneity in Lyve1 + Mrc1 + Pf4 + vascular resident macrophages, segregating them into 3 subsets. These comprise two subsets that share a homeostatic gene signature (cholesterol handling, vascular cell maintenance, phagocytosis, and wound healing) and one with an activated resident signature (responses to wounding, chemotaxis, responses to oxidised phospholipids, NLRP3 inflammasome, and NFκB pathways). Importantly, Clec4a2 is one of the top identifier genes discriminating homeostatic from activated Lyve1 + Mrc1 + Pf4 + resident macrophages in the atherosclerotic aorta. Ablation of CLEC4A2 + -resident macrophages using LysM Cre+ Clec4a2 flox/DTR mice increases lesion development during atherogenesis (Figs. 2, 3), indicating that CLEC4A2 + macrophages are a subpopulation of Lyve1 + vascular resident macrophages endowed with protective functions as predicted by their gene expression profile in scRNA-seq (Fig. 1). . d-f Representative mass cytometry plots and percentages of resident macrophages, Ly6C + monocytes, and cDC2 from each origin in the aorta of the WT host. Two-tailed Student's t-test. Resident Mac: ****P < 0.0001, Ly6C + Mon: ***P = 0.0003, cDC2: ****P < 0.0001. g Representative density plots of myeloid cells in the aorta of HFD-fed Ldlr −/− mice by WT and Clec4a2 −/− donor origins (n = 9, pooled from three independent experiments). h-k Representative mass cytometry plots and percentages of resident macrophages, CD11c + macrophages, Ly6C + monocytes and cDC2 from each origin in the aorta of the Ldlr −/− host. Two tailed Student's t-test. Resident Mac, CD11c + Mac and Ly6C + Mon: ****P < 0.0001, cDC2: **P = 0.0011. All data are presented as mean ± SEM. Source data are provided as a Source Data file. Although CLEC4A2 expression identifies macrophages with specific properties, its expression is not exclusive to vascular macrophages 29 . CLEC4A2 is highly expressed by alveolar and cardiac macrophages (Supplementary Fig. 3), and they are also ablated by DT in LysM Cre+ Clec4a2 flox/DTR mice ( Supplementary  Fig. 5). Nonetheless, there is no evidence to date that these macrophages have a direct impact on atherogenesis. In addition, neutrophils express CLEC4A2 in the BM and blood, but their CLEC4A2 expression is low in the artery. We observed a transient depletion of neutrophils in the BM after DT administration to LysM Cre+ Clec4a2 flox/DTR mice, with full recovery at 48 h (Supplementary Fig. 5d). DT-mediated depletion did not affect neutrophils in peripheral tissues such as spleen and aorta ( Fig. 2;  Supplementary Fig. 5). Ablation of neutrophils has been shown to reduce the progression of atherogenesis 30 . Therefore, the effect of short-term neutrophil depletion in the BM may contribute to underestimate the pro-atherogenic effect of ablation of CLEC4A2 + macrophages. Finally, DT-mediated cell ablation did not induce systemic inflammation or increased serum cholesterol levels (Supplementary Table 2). Thus, the pro-atherogenic effects of DT-mediated cell ablation are on the whole driven by ablation of CLEC4A2 + macrophages.
Our findings show that CLEC4A2 has an essential role in deciding the fate of monocytes as they differentiate in situ (Figs. 5,  6). CSF1 induces housekeeping features that characterise tissue resident macrophages, while CSF2-dependent macrophages acquire pro-inflammatory functions 31 . The fine balance between CSF1 and CSF2 signalling is important in tissue homeostasis, for instance, by preserving the protective function of the microglia 32 . Vascular resident macrophages in the steady state are CSF1dependent 6,7 while CSF2 underpins the survival and activation of intimal CD11c + myeloid cells in atherosclerosis 33,34 . We provide in vitro (Fig. 5) and in vivo ( Fig. 6; Supplementary Fig. 8) evidence that CLEC4A2 biases the monocytes to macrophage transition in response to CSF1 by monitoring proliferation and differentiation in WT and Clec4a2 −/− cells. As a result of this bias, CLEC4A2-competent monocytic cells are more apt to differentiate into CSF1-dependent resident macrophages than CLEC4A2-deficient ones, and Clec4a2 −/− macrophages remain monocytes or differentiate into CSF2-dependent cell states such as cDC2 and CD11c + macrophages (Figs. 5 and 6). Our in vivo data recapitulate in vitro cellular phenotypes as shown by the shared genes regulated by CLEC4A2 in macrophages collected from the aorta and in culture as assessed by scRNA-seq and bulk RNA-seq (Supplementary Table 6). Our RNA-seq data demonstrate that CLEC4A2 deletion induces a differentiation bias by causing profound transcriptional changes, with loss of core macrophage genes (e.g., Mrc1, Siglec1, Slco2b1, Folr2, Adgre1, Fcgrt, and Lyve1 in BMDMs; Fcgr1, Cd209g and Timd4 in vascular resident macrophages) and regulators of macrophage maturation (Notch1, Cebpa, and Irf4 in BMDMs; Notch2, Zeb2os, and Irf4 in vascular resident macrophages) (Figs. 5, 7). Notch signalling is required for appropriate differentiation of macrophages that perform tissue repair 35 and contributes to maturation of Kupffer cells 36,37 . Interferon regulatory factor 4 (IRF4) is a key transcription factor in polarisation of alternatively activated macrophages 38 . ZEB2os is a long noncoding RNA that regulates the translation of ZEB2 39 . Loss of macrophage identity by ZEB2 deletion results in diminished tissue-macrophage pools in the liver and lung 40 . Downregulation of these genes in CLEC4A2deficient macrophages coincides with the enhanced expression of CSF2-dependent transcription factors, Id2 41 and Bhlhe40 42 (Fig. 5). These genes are consistently identifiable in CD11c + macrophage subsets in the aorta (Id2 in inflammatory macrophages (C2) and TREM2 high macrophages (C6), and Bhlhe40 in C2; Fig. 7). Both subsets are increased by CLEC4A2 deficiency during atherosclerosis (Fig. 7). Together, functional and transcriptional findings indicate that CLEC4A2-dependent changes in macrophages are ascribable to a single overarching mechanism of bias in monocyte-fate decision from CSF2 to CSF1 within the arterial wall. Future studies will help mine further the mechanisms underlying CSF1/CSF2-regulated events in arterial macrophages.
CLEC4A2 has an atheroprotective role and its genetic deletion causes an increase in lesion size, accumulation of inflammatory myeloid cells, and necrotic core formation (Fig. 4). We show that CLEC4A2 has an important role in minimising losses of the vascular resident macrophage pool during atherogenesis. A concomitant dilatation of the vessel wall, consistent with the finding by Lim et al. 7 , also further supports this conclusion. The lack of systemic increase in myeloid subsets in the BM, blood, spleen and para-aortic lymph nodes in Clec4a2-deficient atherosclerotic mice ( Supplementary  Fig. 7), indicates that CLEC4A2 protects from atherogenesis through its effects within the arterial wall. Our findings in a competitive BM chimera setting indicate that the cell-intrinsic effect of CLEC4A2 on differentiation is driving the loss of the vascular resident macrophage pool (Fig. 6). The rise in CD11c + macrophages by CLEC4A2 deficiency is observed when the lesional macrophage niche is expanded in atherosclerosis (Fig. 4 and Supplementary Fig. 7). Notably, the loss of the vascular resident macrophage pool in the absence of CLEC4A2 precedes the changes in lesion formation and myeloid landscape (Fig. 4), demonstrating that the primary function of CLEC4A2 is to preserve the size of the vascular resident macrophage pool during atherogenesis. These findings, together with the evidence of a proatherogenic effect of CLEC4A2 + macrophage depletion (Fig. 3), reveal that CLEC4A2 is essential in preventing loss of the vascular resident macrophage pool by promoting monocyte differentiation into homeostatic vascular resident macrophages and thus protecting the aorta from atherosclerosis.
Macrophages are integral to inflammation and cholesterol homeostasis, key biological processes in atherogenesis. CLEC4A2 not only controls the vascular resident macrophage pool size, but also ensures their appropriate functional tissue adaptation. Using a combination of scRNA-seq and functional assays, we reveal that CLEC4A2-dependent macrophage programming has two key functional consequences that protect from atherogenesis. The first is its ability to dramatically reduce TLR2 and TLR4 signalling in macrophages (Fig. 7). Resident macrophages restrain their TLR responses to nucleic acids when clearing cellular debris 43 . Human CLEC4A and murine CLEC4A4, an isoform of CLEC4A2 bearing the ITIM, hamper TLR8 44 or TLR2/3/4/9 responses in DCs 45 . SHP1 can act as a negative regulator of TLR signalling 46,47 and the deletion of the TLR/IL-1R adaptor MyD88 in motheaten mice rescues its hyperinflammatory phenotype 48 . We also find that in the absence of CLEC4A2, vascular resident macrophages upregulate inflammatory gene expression (Il1b, Tlr2, and Stat5a) and BMDMs display enhanced responses when exposed to TLR2 and TLR4 ligands. In atherosclerosis, lipoproteins trigger TLR2 and TLR4 signalling pathways 49,50 . TLR2 and TLR4 are proatherogenic and enhance plaque inflammation via the production of cytokines and activation of the inflammasome 51,52 . TLR activation is also key in driving changes in macrophages in the vessel wall. For instance, TLR4-mediated IRF5 activation increases CD11c expression in vascular macrophages 53 . Our functional and transcriptional data indicate that CLEC4A2 desensitises macrophages from continuous TLR triggering during atherogenesis, protecting from excessive inflammation. It is notable that Clec4a2 expression in macrophages is downregulated upon inflammatory stimuli ( Supplementary Fig. 3), suggesting that CLEC4A2 may act as a switch to allow sensitisation of macrophages to TLR signalling when inflammation is detected in the vessel wall.
Last, our scRNA-seq data show that vascular resident macrophages have a transcriptional signature of genes involved in cholesterol efflux (Fig. 1), which is essential for appropriate lipid metabolism within the vessel wall 54 . Loss of CLEC4A2 interferes with the expression and function of reverse cholesterol-transport pathways. In the absence of CLEC4A2, macrophages express lower levels of essential cholesterol-efflux genes (Abca1 and Abcg1) (Figs. 5, 7), and are consequently less capable of performing the reverse cholesterol transport in response to oxLDL loading, resulting in accumulation of lipids and foam-cell formation (Fig. 7). CLEC4A2 is one of the few CLRs bearing a cytoplasmic ITIM that recruits SHP1 and SHP2 18,55 . Both SHP1-deficient (motheaten) and CLEC4A2-deficient mice are characterised by DC expansion, enhanced TLR and CSF2 signalling, and susceptibility to autoimmune diseases 21,56,57 . While the role of the CLEC4A2 and SHP1 in TLR and CSF2 signalling is well documented 21,56-58 , whether these pathways modulate cholesterol handling is unknown. We demonstrate that intracellular cholesterol accumulation and Abca1 and Abcg1 expression is modulated by SHP1, directly linking activation of the CLEC4A2 signalling pathway and atheroprotection. Our study indicates that CLEC4A2 is a crucial regulator of vascularmacrophage functions at the crossroads of immune and cholesterol homeostasis in atherosclerosis.
The findings of our work challenge the widely held belief that macrophages are solely detrimental in cardiovascular disease with the discovery of the protective subset of CLEC4A2 + vascular macrophages. Our data identify CLEC4A2 as a crucial receptor for tissue adaptation of macrophages, which has a protective role in vascular health and disease. Guided by single-cell biology, our data pave the way to discoveries on the fine functional heterogeneity of vascular macrophages and how we modulate specific subsets to alter the course of atherogenesis. Fostering the homeostatic functions of macrophages through the CLEC4A pathway might represent a transformative therapeutic strategy for inflammatory and cardiovascular diseases. Bone marrow chimeras. In total, 8-10-week-old CD45.1 or Ldlr −/− CD45.1 mice were sub-lethally irradiated twice with a 4-hour interval using an X-ray irradiator (Gulmay; 1.8 Gy/min, 210 KV, 13 mA for 3 min). The irradiated mice received 3-5 × 10 6 bone marrow (BM) cells by tail vein injection 1 h after the last irradiation. 1 % antibiotic (Baytril, Bayer) in autoclaved drinking water was given for 2 weeks.
Tissue preparation. All enzymes were purchased from Sigma-Aldrich, unless stated otherwise. Tissues were isolated in 5% FBS (Gibco) in RPMI-1640 (Gibco). Tissues were cut into small pieces and digested in the appropriate enzyme mixture at 37°C in a water-bath shaker: Aorta: 59 450 U/ml collagenase I, 125 U/ml collagenase XI 60 U/ml hyaluronidase and 60 U/ml DNase I (ThermoFisher Scientific) for 50 min; Lung: 0.4 mg/ml collagenase IV and 0.15 mg/ml DNase I for 40 min; Heart: 1 mg/ml collagenase II and 0.15 mg/ml DNase I for 30 min. The cells were retrieved by passing tissue pieces through a 70 μm cell strainer (Greiner Bio-One). Spleen and lymph nodes were mechanically dissociated by passing cells through a 70 μm cell strainer. For separating aortic layers into the intima/media and the adventitia, the whole aorta was pre-incubated with 125 U/ml collagenase II and 3.75 U/ml elastase at 37°C for 10 min. The adventitia layer was separated from the inner tube (the intima and media) by pulling with two pairs of fine forceps. The tissues were then digested as performed in the aorta sample. For lungs, hearts, and spleens, erythrocytes were lysed using Red Cell Lysis Buffer (Sigma-Aldrich) at room temperature for 2 min.
Flow cytometry. Cells were stained with Live/Dead fixable aqua dye (Thermo-Fisher Scientific) for 10 min at room temperature and were then washed with 1% FBS in PBS (Lonza). Nonspecific binding of antibodies was blocked by incubating cells with rat anti-mouse CD16/32 (BD Biosciences) at 4°C for 10 min, followed by incubating with fluorochrome-conjugated antibodies recognising cell-surface markers for 30 min at 4°C. Following by two wash steps, cells were fixed using either BD fix (BD Biosciences) for 10 min at 4°C or Foxp3/Transcription Factor Staining Buffer Set (ThermoFisher Scientific) for 60 min at 4°C for intracellular staining. For intracellular staining, fixed and permeabilised cells were washed and stained with fluorochrome-conjugated antibodies against intracellular markers. Cell acquisition was performed using an LSRII (BD Biosciences) or LSRFortessa (BD Biosciences) flow cytometer. The data were analysed using FlowJo 10.5.3.
Mass cytometry. Some metal-conjugated antibodies were purchased from Fluidigm. Some purified unlabelled antibodies were conjugated in-house using the Maxpar X8 Metal Labelling Kit (Fluidigm) according to the manufacturer's instructions. All antibodies were titrated to determine the optimal staining concentration using murine splenocytes, aortic cells, and BM cells. The antibodies used are listed in Supplementary Table 7. Single cells from murine tissues were first stained with a Rhodium DNA intercalator (Fluidigm) to distinguish live or dead cells followed by barcoding using the palladium-based 20-Plex Pd Barcoding Kit (Fluidigm) according to the manufacturer's instructions. Barcoded cells were then combined into a single tube prior to addition of rat anti-mouse CD16/32 (BD Biosciences) and staining with metal-conjugated antibodies for 30 min at 4°C. The cells were then washed and were permeabilised using the Foxp3/Transcription Factor Staining Buffer Set (ThermoFisher Scientific) for 60 min at 4°C prior to staining with antibodies against intracellular markers for 30 min at 4°C. Finally, the stained cells were incubated with an Iridium DNA intercalator (Fluidigm) in Maxpar fix and perm buffer (Fluidigm) overnight at 4°C. Prior to acquisition, cells were washed twice with cell staining buffer (Fluidigm) followed by two washes with water (Fluidigm) and filtered through 40 μm cell strainer (Corning) before introducing into a Helios mass cytometer (Fluidigm). Data were generated in an fcs file format. All fcs files in the experiment were normalised and debarcoded using the normalisation and debarcoding tools within the Helios software. These fcs files were then uploaded to Cytobank (www.cytobank.org) for all gating and further analysis using the automated dimensionality-reduction algorithm tSNE 12 .
Tissue and cell coculture. Heart, lungs and aorta were dissected and were rinsed in PBS (Lonza) to remove the excessive blood. Finely cut tissues were placed in RPMI containing 1% FBS (Bio-Sera), 10 units/ml penicillin (Gibco), and 0.1 mg/ml streptomycin (Gibco), and were then transferred to an insert containing 0.5 μmsized pores (Greiner Bio-One). This insert allows migration of small molecules such as cytokines and growth factors while blocking cell migration. The inserts were placed in the 6 well plate where cells were seeded a day before, and tissues and cells were cocultured at 37°C for 24 h. The insert containing the tissues was removed and the cells were washed in PBS. Cells were lysed in 700 μl of RLT buffer (Qiagen) for gene expression assay.
Enzyme-linked immunosorbent assay (ELISA). IL-6 in the supernatant from macrophage culture was measured using Mouse IL-6 DuoSet ELISA (R&D systems), according to the manufacturer's instructions.
Gene expression assay. Total RNA was extracted from cell lysates in RLT buffer using a RNeasy Mini kit (Qiagen) according to the manufacturer's instructions. About 50 ng/ml of RNA was synthesised into cDNA using a High Capacity cDNA Reverse Transcription kit (ThermoFisher Scientific). A pre-amplification step was performed according to the manufacturer's instructions using appropriate TaqMan assays (ThermoFisher Scientific) and TaqMan Pre-Amp Master Mix (Thermo-Fisher Scientific). The pre-amplified cDNA, TaqMan Assays, and TaqMan Gene Expression Master Mix (ThermoFisher Scientific) were used for qPCR in a 384-well plate and the reaction was run by Viia7 Real Time PCR system (ThermoFisher Scientific), according to the manufacturer's instructions. TaqMan probes used are 18 S rRNA (Hs99999901_s1), egfp (Mr04097229_mr), Hbegf (Hs00181813_m1), Clec4a2 (Mm04210363_mH), Mrc1 (Mm00485148_m1), Actb (Mm02619580_g1), Abca1 (Mm00442646_m1), and Abcg1 (Mm00437390_m1).
RNA sequencing. Single-cell RNA sequencing was performed on live CD45 + CD11b + aortic cells from 9 ApoE −/− and ApoE −/− Clec4a2 −/− female mice fed a HFD for 12 weeks. Briefly, cells were sorted using FACSAria III (BD Biosciences), washed in PBS with 0.04% BSA and resuspended before capturing single cells in droplets on a Chromium 10x Genomics platform. Library generation for 10x Genomics v2 chemistry was performed following the Chromium Single Cell 3ʹ Reagents Kits (User Guide: CG00052). Quantification of cDNA and library construction was performed using a Qubit dsDNA HS Assay Kit (ThermoFisher Scientific) and high-sensitivity DNA tapestation (Agilent). Libraries were sequenced on an Illumina HiSeq4000 platform to achieve an average of 50,000 or more reads per cell. The raw sequence data were generated in a FASTQ file format and the alignment data were generated in a BAM file format.
Poly(A) RNA sequencing was carried out on BM cells obtained from CD45.1 and Clec4a2 −/− mice and cultured with 100 ng/ml CSF1 for 10 days. Macrophages were collected and sorted into CD11b + F4/80 + CD206 + populations that then were separated by the expression of CD45.1 (WT) or CD45.2 (Clec4a2 −/− ). Total RNA was isolated by using the RNeasy micro kit (Qiagen). Total RNA quantity and integrity were assessed using Quant-IT RiboGreen RNA Assay Kit (ThermoFisher Scientific) and high-sensitivity DNA tapestation (Agilent). Purification of mRNA, generation of double-stranded cDNA, and library construction were performed using NEBNext Poly(A) mRNA Magnetic Isolation Module (New England Biolabs) and NEBNext Ultra II Directional RNA Library Prep Kit (New England Biolabs) for Illumina with adaptors and barcode tags (dual indexing) 61 . The final size distribution of the pool was determined using high-sensitivity DNA tapestation (Agilent), and quantification was determined by Quant-IT RiboGreen RNA Assay Kit (ThermoFisher Scientific) before sequencing on an Illumina HiSeq 4000 as 75 bp paired end.
Raw gene counts across multiple biological samples were pooled using the CellRanger command line interface. Reads were subsampled from higher-depth libraries until they all had an equal number of confidently mapped reads per cell. Integrated raw gene matrix was passed to Seurat R package (v 3.0) 62 for preprocessing and quality control. Quality control pipeline removed n = 1,467,497 cells that have fewer than 200 expressed genes and n = 14,872 genes that were detected in fewer than 3 cells. Cells with greater than 5% mitochondrial gene expression were also excluded from the analysis. The count matrix was lognormalised and scaled to unit variance, while mitochondrial and cell count effects were removed using a linear regression. Normalised count matrix was reduced to 50 principal components (PCs) and significant PCs [Jackstraw test 62 ] were used as the input for graph-based clustering (as implemented in Seurat). Data visualisation and dimensionality reduction was achieved using three algorithms including tSNE 63 , LargeVis 64 , and ivis 65 .
Gene expression in human atherosclerosis. Human-plaque RNA expression data were obtained from the Tampere Vascular Study 66 . The study was approved by the Ethics Committee of Pirkanmaa Hospital District (Finland) and informed consent was previously acquired from patients enroled in the study. Vascular samples were collected from patients subjected to carotid endarterectomy. Clinical characteristics of the study population are available in Supplementary Table 8.
Total RNA was extracted using a RNeasy Mini kit (Qiagen) according to the manufacturer's instructions. About 200 ng of RNA was synthesised into cDNA using the Ambion's Illumina RNA Amplification kit according to the instructions (Ambion). Microarray assays were performed using Illumina HumanHT-12 version 3 Expression BeadChip (Illumina), according to the manufacturer's instructions (Illumina). The gene-association data from the carotid samples were analysed using a Spearman's rank-correlation coefficient test.
Murine atherosclerotic lesion assays. All materials were purchased from Sigma-Aldrich unless stated otherwise. Frozen sections of the aortic root (5 μm each) were stained in Oil Red O solution (3 parts 1% w/v in isopropanol and 2 parts 1% w/v dextrin in distilled water) and counterstained with haematoxylin for measuring lesion formation. The whole aortas were dissected and fixed in 4% PFA. The fixed tissue was cut open longitudinally from the iliac arteries to the aortic root and pinned flat. The aortas were stained in the Sudan IV solution (0.5% w/v in equal parts 50% ethanol and acetone). Sudan IV-stained plaque area was measured. Hematoxylin and eosin (H&E) staining was used to determine necrotic areas in the lesion within the aortic root. The areas within the lesion that were free from H&E stain and bigger than 3000 μm 2 were designated to be necrotic core.
Avidin-biotin complex (ABC) staining. Immunohistochemistry of tissue sections was performed using the Avidin-Biotin Complex (ABC) kits (Vector Laboratories) based on the optimised protocol 67 . Acetone-fixed slides were washed in PBS and blocked with 10% serum (rabbit or goat serum from Abcam) for 1 h. After washing in PBS, Avidin and Biotin blocking solutions (Vector Laboratories) were used to block endogenous avidin and biotin for 15 min each. The slides were incubated with a primary antibody for 1 h and washed in PBS. The slides were then incubated with a biotinylated secondary antibody (Vector Laboratories) diluted by 1:400 in PBS for 1 h. After two washing steps, the slides were then treated with 0.3% hydrogen peroxide (Sigma-Aldrich) for 15 min to block endogenous peroxidase activity. The slides were then washed and incubated with an ABC-Elite-Standard Kit for Peroxidase (Vector Laboratories) for 40 min. After two washing steps, the slides were treated with DAB (3,3′-diaminobenzidine) HRP substrate solution (Vector Laboratories) for 1-5 min depending on the intensity of the signal. The slides were then washed in PBS, counter-stained in Haematoxylin solution (Sigma-Aldrich) for 15-30 s, and washed in deionised water. The slides were dehydrated in 70% ethanol for 1 min and 100% ethanol for 1 min. Finally, the slides were airdried and were mounted using Histomount (National Diagnostics) and cover glasses (VWR).
Immunofluorescence. Acetone or PFA-fixed slides were washed in PBS and were blocked with 20% normal serum (rabbit or goat serum from Abcam) for 1 h. After washing steps in PBS, the slides were treated with primary antibodies for 1 h and were washed in PBS twice. The slides were then incubated with secondary antibody conjugated with AF488, AF568 or AF647 (10 μg/ml, ThermoFisher Scientific) for 1 h in the dark. The slides were then washed in PBS followed by dipping in deionised water briefly. The slides were then mounted with Fluoromount-G with DAPI (ThermoFisher Scientific).
Imaging and analysis. All stained sections were imaged using an Olympus BX51 osteometric bright-field and fluorescence microscope, and en face aorta images were taken using the Leica M60 dissection microscope. All images were randomised for blind analysis. Quantification of stained images were analysed using Clemex Vision Lite software. Some images were analysed using ImageJ 2.1.0/1.53c (NIH, USA).
Serum cholesterol measure. Levels of cholesterol in the murine blood were measured by using cholesterol-detection reagent (Randox) alongside the cholesterol standard (Randox). The enzymatic endpoint method was adopted to calculate the concentration of cholesterol based on the standard. Statistical analysis. All statistical analyses were performed using GraphPad Prism. All data are presented as Mean ± SEM. Two tailed Student's t-test, two-way ANOVA or ordinary one-way ANOVA with Holm-Sidak's multiple-comparison test was performed for statistical significance. For the scRNA-seq analyses, differentially expressed genes were identified using the two-sided Wilcoxon Rank Sum test with Bonferroni's correction for multiple-hypothesis testing. In the bulk RNAseq analyses, differential expression was assessed using the two-sided Robinson and Smyth exact test with Bonferroni's correction for multiple-hypothesis testing. P < 0.05 was considered statistically significant: *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001.
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
All data generated in this study are provided in the Supplementary Information files and the Source Data file. The scRNA-seq data of myeloid cells from murine atherosclerotic aortas are publicly available at ArrayExpress using the following accession codes (Fig. 1: E-MTAB-10743; Fig. 7: E-MTAB-10746). The bulk RNA-seq data of murine CSF1cultured BM-derived macrophages in Fig. 5 are also freely accessible (accession code: E-MTAB-10734). The data included in Supplementary Fig. 4 are obtained from the Tampere Vascular Study (TVS) 66 , which comprises health related participant data. The use of data is restricted by regulations on professional secrecy (Act on the Openness of Government Activities, 612/1999) and sensitive personal data (Personal Data Act, 523/ 1999, implemented in the EU data-protection directive 95/46/EC). Due to these restrictions, the data cannot be stored in public repositories or otherwise made publicly available. Data access may be permitted on a case-by-case basis upon request. Data sharing outside the group is done in collaboration with the TVS group and requires a data-sharing agreement. Investigators can submit an expression of interest to the Chairman of the publication committee (Prof Terho Lehtimäki, Department of Clinical Chemistry, Fimlab Laboratories and Finnish Cardiovascular Research Center-Tampere, Faculty of Medicine and Health Technology, Tampere University, Arvo building room D338, P.O Box 100, FI-33014 Tampere, Finland; email: terho.lehtimaki@tuni.fi; phone: +358 50 4336285). The timeframe for response to request is within two weeks. The right to use the data is granted with the understanding that the collaborator will protect the data and not share it with any other parties, academic or commercial, other than those mentioned in the agreement. The data can be used by the collaborator only for scientific purposes and as specified in the appendix of the agreement in the form of a short scientific proposal. After the contract has ended, the collaborator will upon directions of TVS, return or destroy copies of the data. In case the work has involved transfer of biospecimens, the collaborator will return the remaining biospecimens and related data to the TVS group. Source data are provided with this paper.