CD163+ macrophages restrain vascular calcification, promoting the development of high-risk plaque

Abstract

Vascular calcification (VC) is concomitant with atherosclerosis, yet it remains uncertain why rupture-prone high-risk plaques do not typically show extensive calcification. Intraplaque hemorrhage (IPH) deposits erythrocyte-derived cholesterol, enlarging the necrotic core and promoting high-risk plaque development. Pro-atherogenic CD163+ alternative macrophages engulf hemoglobin:haptoglobin (HH) complexes at IPH sites. However, their role in VC has never been examined to our knowledge. Here we show, in human arteries, the distribution of CD163+ macrophages correlated inversely with VC. In vitro experiments using vascular smooth muscle cells (VSMCs) cultured with HH-exposed human macrophage - M(Hb) - supernatant reduced calcification, while arteries from ApoE-/- CD163-/- mice showed greater VC. M(Hb) supernatant-exposed VSMCs showed activated NF-κB, while blocking NF-κB attenuated the anticalcific effect of M(Hb) on VSMCs. CD163+ macrophages altered VC through NF-κB-induced transcription of hyaluronan synthase (HAS), an enzyme that catalyzes the formation of the extracellular matrix glycosaminoglycan, hyaluronan, within VSMCs. M(Hb) supernatants enhanced HAS production in VSMCs, while knocking down HAS attenuated its anticalcific effect. NF-κB blockade in ApoE-/- mice reduced hyaluronan and increased VC. In human arteries, hyaluronan and HAS were increased in areas of CD163+ macrophage presence. Our findings highlight an important mechanism by which CD163+ macrophages inhibit VC through NF-κB-induced HAS augmentation and thus promote the high-risk plaque development.

Keywords: Atherosclerosis; Cell Biology; Macrophages; Plaque formation; Vascular Biology.

Figure 1. CD163⁺ macrophages are associated with reduced vascular calcification in human carotid artery advanced atheromas.

(A–C) FCP sections from 82-year-old male. Low-power H&E image (A). High-power AR (B) and CD163 immunostaining (C) from the corresponding black rectangle in A. No CD163⁺ cells were observed around calcification stained as red in AR. (D–F) PR sections from 46-year-old male. Low-power H&E image (D). High-power images of AR (E) and CD163 immunostaining (F) from the corresponding black rectangle in D. Multiple CD163⁺ cells were observed around hemorrhagic area without any AR-positive lesions. (G and H) Seventy atheroma sections from 32 patients were classified into FCP or FA-Ca (n = 36, blue), FA or TCFA (n = 13, green), and PR or HR (n = 21, red), with the corresponding percentage of calcification (G) and CD163⁺ macrophages (H) per plaque area. (I–R) FA sections from 62-year-old male. Low-power Movat image (I). Mid-power image of H&E (J) of the corresponding black rectangle in I. High-power images of VK (K and O), AR (L and P), as well as immunostaining for CD68 (a general macrophage marker) (M and Q) and CD163 (N and R) of the corresponding red (K–N) and blue rectangles (O–R) in J. Note the absence of calcification in areas of CD163⁺ macrophages but its presence in CD163^– areas. (S and T) CD163⁺CD68⁺ and CD163^–CD68⁺ areas were determined in 70 sections (the method of outlining was in Supplemental Figure 1). Each CD163⁺CD68⁺ and CD163^–CD68⁺ area was put in adjacent AR sections. Then AR-positive microcalcification areas in each field were digitally analyzed. The total area of CD163⁺CD68⁺ and CD163^–CD68⁺ was similar (S); however, % microcalcification in CD163^–CD68⁺ was greater than in CD163⁺CD68⁺ (T). Results are presented as median and interquartile range (G, H, S, and T). Kruskal-Wallis test followed by post hoc Dunn’s test (G and H) or Wilcoxon’s matched-pair signed-rank test (S and T) was applied. Data normality was tested by Shapiro-Wilk test. Scale bars: 2 mm (A and D), 0.2 mm (B, C, E, and F) 1 mm (I), 0.2 mm (J), 0.1 mm (K–R). AR, Alizarin Red; Ca, calcification; FA, fibroatheroma; FA-Ca, fibroatheroma with calcification; FCP, fibrocalcific plaque; HR, healed plaque rupture; PR, plaque rupture; TCFA, thin-cap fibroatheroma; VK, von Kossa.

Figure 2. The distribution pattern of calcification and CD163⁺ macrophages in the underlying plaque and thrombus occluded lumen of human coronary CTOs.

(A–D) Section with thrombotic total occlusion of RCA from 39-year-old man who died of recent inferior myocardial infarction. Low-power Movat image showing fibrin-rich organizing thrombus (A). High-power images of H&E (B) and immunostaining for CD163 (C) and α-SMA (D) (corresponding black rectangle in A). Occluded lumen with fibrin-rich thrombus (yellow-dot outline in A) was observed. CD163⁺ cells and small amount of α-SMA⁺ cells were found without visible calcification at occluded lumen. (E–H) CTO section from 92-year-old man who died of a subdural hemorrhage. Low-power Movat image showing PG-rich organized thrombus (E). High-power images of H&E (F) and immunostaining for CD163 (G) and α-SMA (H) (corresponding black rectangle in E). Sheet Ca was observed in the original plaque. Multiple CD163⁺ cells and α-SMA⁺ cells were found in PG-rich thrombus at occluded lumen (yellow-dot outline in E) without visible calcification. (I–L) CTO section from 56-year-old man with triple-vessel disease. Low-power Movat image with type I collagen–rich organized thrombus (I). High-power images of H&E (J) and immunostaining for CD163 (K) and α-SMA (L) (corresponding black rectangle in I). Occluded thrombus was replaced by type I collagen over time. No CD163⁺ cells or calcification was observed at occluded lumen (yellow-dot outline in I). (M–R) We obtained 145 CTO sections from 21 vessels (19 patients). Percentage area calcification (M), % CD163⁺ area (N), and % α-SMA⁺ area (O) were compared between original plaque and occluded lumen. Sections were classified into fibrin-rich organizing thrombus (n = 23), PG-rich organized thrombus (n = 31), and type I collagen–rich organized thrombus (n = 91). Percentage area calcification (P), % CD163⁺ area (Q), and % α-SMA⁺ area (R) in occluded lumen were compared. Results are presented as the median and interquartile range (M–R). Wilcoxon’s matched-pairs signed-rank test (M–O) or Kruskal-Wallis test followed by post hoc Dunn’s test (P–R) was applied. Data normality was tested by Shapiro-Wilk test. Scale bars: 1 mm (A, E, and I), 50 μm (B–D, F–H, and J–L). RCA, right coronary artery; α-SMA, α–smooth muscle actin; col, collagen; PG, proteoglycan; Th, thrombus.

Figure 3. BCA and aortic calcification in aged ApoE^–/– versus ApoE^–/– CD163^–/– mice.

(A–D) Representative BCA sections with H&E and AR staining obtained from ApoE^–/– (A and B) and ApoE^–/– CD163^–/– (C and D) mice (1.5 years old). (E–I) Vessel size (EEL) (E), plaque area (F), % area stenosis (G), calcification area assessed by AR (H), and % calcification area (I) at the most stenotic BCA lesion were compared between ApoE^–/– and ApoE^–/– CD163^–/– mice (n = 11 per group). (J) Schematic diagram of a mouse aorta from which total calcium contents were extracted for colorimetric analysis. Aortic tissues from ascending (cut just above aortic valve) to descending (cut at the level of diaphragm) were obtained for further calcium extraction process. BCA (main branch only), left common carotid, and subclavian arteries with the equivalent length with BCA were also included in the specimens. (K) Total aortic calcium contents assessed by colorimetric assay with the comparison of ApoE^–/– and ApoE^–/– CD163^–/– mice (n = 9 per group). Results are presented as the mean ± standard deviation (E–G and K) or median and interquartile range (H and I). T test (E–G and K) or Mann-Whitney test (H and I) was applied. Data normality was tested by Shapiro-Wilk test. EEL, external elastic lamina.

Figure 4. Effect of M(Hb) supernatant on HASMC calcification in vitro.

(A–C) HASMCs were exposed for 2 days to control basic culture media with OS (containing CaCl₂, β-glycerophosphate, l-ascorbic acid, insulin, and dexamethasone) with or without HH, or with M(con)sup or M(Hb)sup. Representative macro- (inset) and microscopic AR findings (4× original magnification) in each condition (A). Summary of % AR-positive area in each condition (B) (n = 3 per group). Amount of calcium examined by colorimetric assay adjusted by protein amount (C) (n = 4 per group). (D and E) HASMCs were cultured for 24 hours in M(con) or M(Hb) with or without OS. Representative immunoblotting image of RUNX2 and β-actin (D). Summary of densitometry analysis (E) (n = 3 per group). (F–K) HASMCs were cultured for 6 hours in M(con) or M(Hb) with OS. Extracted RNA samples were analyzed by Affymetrix Clariom S microarray (n = 4 per group). We demonstrated 548 upregulated and 249 downregulated DEGs in M(Hb)+OS versus M(con)+OS condition (threshold: FC ≥ 2.0 or ≤ –2.0 with P ≤ 0.05). Heatmap of 797 DEGs visualized by row z score scaling (F). Volcano plots detailing the magnitude of expression difference (G). GO enrichment analysis of unbiased DEGs was conducted by DAVID v6.8 bioinformatics resources. Top 25 GO BP terms with highest statistical significance in up- and downregulated DEGs (H and I). Selected genes related to inflammation and NF-κB signaling from upregulated genes (J) and these related to calcification/osteogenic differentiation-related genes from downregulated DEGs (K) were visualized by heatmap with row z score scaling of log₂ FC. *P < 0.05, **P < 0.01. Results are presented as the mean ± standard error and ANOVA followed by post hoc Tukey’s test was applied (B, C, and E). Data normality was tested by Shapiro-Wilk test. The experiments of A–E were performed at least 3 times to confirm their reproducibility. AR, Alizarin Red; BP, biological process; CTL, control; DEGs, differentially expressed genes; FC, fold change; GO, gene ontology; HH, hemoglobin-haptoglobin complex; M(con)sup, control macrophage supernatant; M(Hb)sup, HH-differentiated macrophage supernatant; Neg-reg, negative regulation; OS, osteogenic components supplementation; Pos-reg, positive regulation; TC, transcription; TF, transcription factor.

Figure 5. Stimulation of NF-κB signaling by M(Hb) supernatants directs its anticalcific effect on HASMCs.

(A and B) Representative immunoblotting and summary of densitometry analysis for HASMCs after exposure to M(con) or M(Hb) supernatants for 24 hours without OS: p-p65/t-p65 (whole-cell lysate, n = 4 per group) (A) and t-p65/TBP (nuclear fraction, n = 3 per group) (B). (C–E) Immunoblotting image of t-p65/β-actin for HASMC whole-cell lysate transfected with the scrambled control siRNA or p65-siRNA (C). M(con)sup or M(Hb)sup with OS exposure (48 hours) was performed. Representative AR staining images (D) and summary of % AR-positive area (n = 4 per group) (E). (F) HASMC were exposed to M(con)sup or M(Hb)sup with CTLpep or NBDpep (both 25 μM) without OS for 24-hours. Representative WB images and summary of densitometry analysis for p-p65/t-p65 (whole-cell lysate, n = 3 per group). (G and H) HASMCs were exposed to M(con)sup or M(Hb)sup with OS and CTLpep or NBDpep (6.25, 25, and 100 μM) for 48 hours. Representative AR staining images (G) and summary of % AR-positive area (n = 3 per group) (H). *P < 0.05, **P < 0.01, ^†P < 0.05 vs. M(con) CTLpep 6.25 μM, ^‡P < 0.05 vs. M(con) CTLpep 25 μM, ^§P 0.05 vs. CTLpep 100 μM, ^||P < 0.05 vs. M(Hb) CTLpep 25 μM, ^¶P < 0.05 vs. M(Hb) CTLpep 100 μM. Results are presented as the mean ± standard error (A, B, E, F, and H). T test was applied to A and B. ANOVA followed by post hoc Tukey’s test was applied to E, F, and H. Data normality was tested by Shapiro-Wilk test. All experiments were performed at least 3 times to confirm the reproducibility. CTLpep, IKK-NBD control peptide; NBDpep, NF-κB inhibitor NBD peptide; OS, osteogenic medium; p-p65, phosphorylated p65 (Ser536); t-p65, total-p65; Scr, scrambled control siRNA; si p65, p65-siRNA; TBP, TATA-binding protein; WB, Western blotting.

Figure 6. Augmented HA synthesis via NF-κB signaling by M(Hb) supernatants is responsible for its anticalcific effect on HASMCs.

(A–D) Representative WB images and summary of densitometry analysis for HASMCs after exposing to M(con)sup or M(Hb)sup without OS for 24 hours (whole-cell lysate); HAS1/β-actin (A), HAS2/β-actin (B), HAS3/β-actin (C), and CD44/β-actin (D) (n = 4 in each). (E) HASMCs were exposed to M(con)sup or M(Hb)sup with CTLpep or NBDpep (both 25 μM) without OS for 24 hours. Representative WB images and summary of densitometry analysis for HAS1/β-actin (whole-cell lysate, n = 4 per group). (F) Summary of HA concentration in M(con) and M(Hb) supernatant pre- and post-exposing (24 hours) to HASMCs (ELISA, n = 3 per group). (G) Summary of HA concentration in M(con) and M(Hb) supernatant post-exposing (24 hours) to HASMCs with CTLpep or NBDpep (both 100 μM) without OS for 24 hours (n = 3 per group). (H) Calcium content in HASMCs after 48 hours’ exposure in OS to HA supplementation (0, 50, and 200 μg/mL) (colorimetric assay, n = 3 per group, adjusted by protein amount). (I) Calcium content in HASMCs after 24 hours’ exposure to basal growth medium with OS and 4-MU, an HAS inhibitor, (0, 2.5, and 25 μM) (colorimetric assay, n = 4 per group, adjusted by protein amount). (J–L) HASMCs were transfected with Scr or siHAS1. The effect of siHAS1 was confirmed by WB (J). M(con) or M(Hb) supernatant with OS was exposed for 48 hours. Representative AR images (K) and summary of % AR-positive area (n = 4 per group) (L). *P < 0.05, **P < 0.01. Results are presented as the mean ± standard error (A–H and K). T test was applied to A–D. ANOVA followed by post hoc Tukey’s test was applied to E–I and L. Data normality was tested by Shapiro-Wilk test. All experiments were performed at least 3 times to confirm their reproducibility. HAS, hyaluronan synthase; siHAS1, HAS1 siRNA; 4-MU, 4-methylumbelliferone.

Figure 7. NF-κB signaling, HA, and RUNX2 expression in plaques of aged ApoE^–/– versus ApoE^–/– CD163^–/– mice.

(A–D) Representative low- and high-power images of BCA sections with total p-65 immunofluorescence (red) and DAPI counterstaining obtained from aged ApoE^–/– (A and B) and ApoE^–/– CD163^–/– (C and D) mice (1.5-year-old). Nuclei in the plaque showing t-p65 colocalization are indicated by yellow arrowheads. (E–G) Summary of plaque area (E), number of nuclei/plaque area (F), and % p65 nuclear colocalization (G) in the most stenotic BCA lesions were compared between ApoE^–/– and ApoE^–/– CD163^–/– mice (n = 6 per group). (H) Schematic diagram of the methodology for plaque extraction from mouse aortas. Whole visible plaques from ascending (including aortic root) to descending (cut at the level of diaphragm) aorta were peeled out under dissecting microscope. Plaques in BCA, left common carotid, and subclavian arteries with the equivalent length with BCA were also included in the samples. Plaques were mechanically ground in equivalent volumes of PBS, and further ELISA and WB analyses were performed. (I) HA content in aortic plaque assessed by ELISA (n = 9 per group). (J–L) Representative WB images for mouse plaque samples including p-p65, t-p65, RUNX2, and β-actin (J). Summary of densitometry analysis is also shown in K (p-p65/t-p65) and L (RUNX2/β-actin) (n = 6 per group). (M–P) Representative low- and high-power images of BCA sections with RUNX2 staining (purple) obtained from aged ApoE^–/– (M and N) and ApoE^–/– CD163^–/– (O and P) mice (1.5-year-old). (Q–S) Summary of % RUNX2⁺ cells in the plaque (Q), % RUNX2⁺ cells in the medial layer (R), and % RUNX2⁺ cells in the vessel (plaque+medial layer) (S) in the most stenotic BCA lesion were compared between ApoE^–/– and ApoE^–/– CD163^–/– mice (n = 6 per group). *P < 0.05, **P < 0.01. Results are presented as the mean ± standard deviation (F, G, L, and Q–S) or median and interquartile range (E, I, and K). T test (F, G, L, and Q–S) or Mann-Whitney test (E, I, and K) was conducted for statistical analysis. Data normality was tested by Shapiro-Wilk test. Scale bars: 100 μm (A and C), 20 μm (B and D), and 100 μm (M–P).

Figure 8. Effect of NF-κB inhibition by NBDpep on development of atheroma and VC in ApoE^–/– mice with HFD feeding.

(A) Schematic diagram of study design. HFD was started in ApoE^–/– mice at the age of 8 weeks and continued to the end of the experiment. Treatment by continuous subcutaneous injection of CTLpep or NBDpep by Alzet osmotic pump (100 μg/kg/d) was started at the age of 14 weeks. Since the pump needed to be replaced every 6 weeks, pump replacing surgeries were performed, and Tx was continued for 18 weeks (by the age of 32 weeks). (B–E) Representative BCA sections with H&E and AR staining obtained from HFD feeding ApoE^–/– mice with CTLpep (B; H&E, C; AR) or NBDpep Tx (D; H&E, E; AR). (F–J) Vessel size (EEL) (F), plaque area (G), % area stenosis (H), calcification area assessed by AR (I), and % area calcification (J) at the most stenotic BCA lesion were compared between CTLpep- and NBDpep-treated ApoE^–/– mice (n = 9 per group). (K and L) Total aortic calcium contents assessed by colorimetric assay (n = 4 per group) (K) and HA content in aortic plaque assessed by ELISA (n = 5 per group) (L) with the comparison of CTLpep- and NBDpep-treated ApoE^–/– mice. (M–O) Representative WB images of mouse plaque samples including p-p65 (Ser536), t-p65, RUNX2, and β-actin (M). Summary of densitometry analysis is shown in N (p-p65/t-p65) and O (RUNX2/β-actin) (n = 5 in each). Results are presented as the mean ± standard deviation (F, N, and O) or median and interquartile range (G–L). T test (F, N, and O) or Mann-Whitney test (G–L) was conducted for statistical analysis. Data normality was tested by Shapiro-Wilk test. Scale bars: 200 μm (B–E). Tx, treatment.

Figure 9. CD163 activity and its association with NF-κB signaling, HA synthesis, and calcification in human atherosclerotic arteries.

(A–F) CTO section of left circumflex artery (LCX) from 37-year-old man who died from an acute coronary syndrome (culprit plaque rupture was found in Left anterior descending artery, LAD). Low-power Movat (A) and high-power H&E (B) image of the corresponding black rectangle in A. Yellow-dot border of occluded lumen with thrombus. Fibrin deposition and neo-angiogenesis (*) in the occluded lumen were observed. CD163 (red) (C and D) and HA (red)/HAS1 (green) (E and F) immunofluorescence images of the adjacent sections of A. (G–K) Representative immunoblotting (G) and summary of densitometry analysis (H–K) for CD163, p-p65 (Ser536), t-p65, HAS1, and β-actin of protein extracted from human carotid atheroma expressing high or low levels of CD163 (n = 6 per group). *P < 0.05, **P < 0.01. (L) X-ray image of postmortem heart from 41-year-old African American with WT allele of the rs7136716 (AA genotype). White arrowheads indicate severe calcification in the coronary tree. (M) X-ray image of postmortem heart from 44-year-old African American with 2 copies of the minor allele for rs7136716 (GG genotype) without visible coronary calcification. Pathology of coronary arteries revealed triple-vessel disease. (N) Summary of ex vivo x-ray–based calcification score in age-matched AA versus GG genotype carriers (n = 15 per group). (O–S) Representative histopathology images of coronary artery sections (H&E) from AA (O) and GG (P) genotype carriers. Percentage area stenosis (Q), total calcification area (R), and % calcification/plaque area (S) between 2 genotype carriers. Results are presented as the mean ± standard deviation (H, I, K, and Q) or median and interquartile range (J, N, R, and S). T test (H, I, K, and Q) or Mann-Whitney test (J, N, R, and S) was conducted for statistical analysis. Data normality was tested by Shapiro-Wilk test. Scale bars: 0.5 mm (A, C, and E), 0.1 mm (B, D, and F), 0.5 mm (O and P).

Figure 10. scRNA-Seq data analysis for human coronary artery samples.

(A) UMAP embeddings of human coronary artery (HCA) lesion scRNA-Seq data (26) integration into HCA snATAC-Seq data of samples with varying clinical presentations of CAD (41 individuals, ~30,000 cells) (27). scRNA-Seq cell type labels and expression profiles were transferred to snATAC-Seq clusters using the ArchR implementation (29) of canonical correlation analysis (28) to maximize the number of cells available for gene expression statistical analyses. The UMAP plot shows 8 identified major cell types including macrophages, contractile/modulated SMCs, endothelial cells, fibroblasts, B cells, pericytes, and a mixed population of T/NK cells. (B) Box plot depicting stratification of individual patients based on the CD163 mean expression level of their corresponding macrophages. Individuals below the CD163 mean macrophage expression distribution 25% quartile were denoted as low CD163 (n = 11, norm expression < 3.87), whereas individuals above the distribution 75% quartile were denoted as high CD163 (n = 11, norm expression > 5.33). Box plots show the median and interquartile range with upper (75%) and lower (25%) quartiles shown. The difference in CD163 mean expression between low and high CD163 patients was statistically significant (P < 0.001) as calculated by an unpaired Student’s t test. (C and D) Bar plots depicting enrichment of TF in gene sets from a transcriptome-wide differential expression analysis of SMCs from high versus low CD163 patients. Genes were ranked based on the mean expression difference between the 2 conditions, and the top 100 genes upregulated in CD163^hi SMCs (C) as well as top 100 genes upregulated in CD163^lo groups (D) were used for TF target overrepresentation analyses using ChEA3 (30). The x axis of the plot denotes the negative log₁₀ of the calculated Fisher Exact Test (FET) P value. Metadata for each patient within high and low CD163 groups can be found in Supplemental Table 5.