The study protocol received approval from the Animal Care and Use Committee at Zhongshan Hospital of Fudan University. All experimental procedures followed the guidelines for the Care and Use of Laboratory Animals. Human aortic arch specimens were collected from organ donors or patients who underwent cardiovascular surgery at our institute between 2014 and 2019, with prior consent obtained from the deceased person’s family.

Mice expressing Cre recombinase under the control of the lysosome promoter (pLys-Cre; #004781), histone deacetylase 3 (HDAC3) floxed mice (HDAC3floxed; #024119), and ApoE-KO mice (#002052) were all acquired from the Jackson Laboratory (Bar Harbor, ME, USA) and housed under sterile conditions. In pLys-Cre mice, a nuclear-localized Cre recombinase was inserted at the first coding ATG of the lysozyme 2 gene, thereby abolishing endogenous lysozyme 2 gene function and allowing Cre expression to be controlled by the endogenous lysozyme 2 promoter/enhancer elements (17). In HDAC3floxed mice, loxP sites were positioned flanking exons 4-7 of the HDAC3 gene. Mice homozygous for this allele are viable and fertile. When bred with mice expressing tissue-specific Cre recombinase, the resulting offspring exhibit deletion of exons 4-7 in the Cre-expressing cells (18). The pLys-Cre and HDAC3floxed mice were first crossbred to generate macrophage-specific HDAC3 knockout mice, and then crossbred with ApoE-KO mice to produce ApoE-KO/pLys-Cre/HDAC3floxed mice. ApoE-KO mice (ApoE-KO) (19) were used as controls without macrophage depletion of HDAC3 in ApoE-KO/pLys-Cre/HDAC3floxed mice. Both female and male mice were randomly and equally assigned to each experimental group. AS was induced in ApoE-KO mice by feeding them at age of 8 weeks a high-fat diet (HFD) for 12 weeks to accelerate plaque formation in the arterial wall. The recipe for the HFD contained 21% butter-derived fat by weight and 0.15% cholesterol along with standard rodent chow ingredients. In contrast, control mice are fed a normal diet (ND) that contains 4.5% plant-based fat by weight along with standard rodent chow ingredients. Mice at 20 weeks of age were analyzed for AS development. Four groups of mice were included in the study: ApoE-KO mice with ND, ApoE-KO mice with HFD, ApoE-KO/pLys-Cre/HDAC3floxed mice with ND and ApoE-KO/pLys-Cre/HDAC3floxed mice with HFD.

LCM was used to isolate CD31+ endothelial cells and CD68+ macrophages from human aortic arch specimens. Briefly, after collection and processing of the aortic arch tissue, which involved fixation, embedding, and sectioning to generate thin tissue slices, the sections were then mounted on the specialized glass slides for LCM and stained with either rabbit anti-CD31 antibody (Ab28364, Abcam, Shanghai, China) for endothelial cells or rabbit anti-CD68 antibody (Ab213363, Abcam) for macrophages, to visualize and identify the target cell populations. Fibronectin staining (Ab2413, Abcam) was performed to help visualization of the tissue. Before performing LCM, the stained tissue sections were dehydrated through a series of alcohol and xylene washes to remove any residual water and to ensure compatibility with the LCM process. The tissue sections were then placed under the LCM microscope, where the regions containing CD31+ endothelial cells or CD68+ macrophages were carefully identified based on fluorescence. Using an ultraviolet laser, the selected regions are microdissected and captured onto a thermoplastic cap to be transferred to a collection tube for protein extraction and ELISA analysis.

The mouse aortic arch dissection and subsequent histological analysis involved several executed steps. First, the aortic arch was carefully removed, ensuring that the tissue remained intact during the process. Following dissection, the aortic arch was fixed in a 4% paraformaldehyde solution for 4 hours to preserve the tissue structure and morphology. After fixation, the tissue was dehydrated through a series of graded alcohol solutions and then embedded in paraffin wax to provide support and protection. Once the tissue was thoroughly infiltrated with paraffin, it was positioned in a mold and allowed to solidify. The paraffin-embedded aortic arch was then sectioned at a thickness of 5 micrometers using a microtome, and the resulting tissue sections were mounted on glass slides for further processing and analysis. To evaluate the presence and extent of atherosclerotic lesions, a Hematoxylin and eosin (H&E) staining was employed. The stained sections were examined under a light microscope. The atherosclerotic lesions were visible as areas of cellular accumulation and extracellular matrix deposition within the intimal layer of the artery wall. The high-resolution images of the stained sections were captured using a microscope equipped with a camera attachment. The atherosclerotic lesion area in each section was outlined with ImageJ analysis software, and the total lesion area was calculated. The lesion area between different treatment groups or experimental conditions was compared to assess the extent of AS.

To measure the lipid content of an atherosclerotic plaque, we used an Oil Red O staining. First, the aortic arch segments containing atherosclerotic plaques were dissected and then fixed in 4% paraformaldehyde. The fixed tissue was embedded and sectioned into slices of 5μm-thickness, using a cryostat. The sections were fixed in cold acetone for 10 minutes. After that, the slides were stained with freshly prepared Oil Red O solution for 15 minutes. The stained slides were then rinsed with distilled water and counterstained with hematoxylin to visualize nuclei. This staining method selectively stained neutral lipids, such as triglycerides and cholesterol esters, enabling the visualization and quantification of lipid-rich areas within the aortic arch tissue sections. Upon completion of the staining procedures, the slides were examined under a light microscope to assess the presence, distribution, and severity of atherosclerotic lesions in the mouse aortic arch samples.

To isolate, culture, and differentiate mouse bone marrow-derived macrophages (BMDMs) into M1 or M2 macrophages, we followed the protocol outlined below. Begin by euthanizing the mice according to institutional guidelines. Spray the hind limbs with 70% ethanol and dissect the femurs and tibias. Remove any attached muscle tissue and transfer the bones to a sterile Petri dish containing phosphate-buffered saline (PBS, Sigma-Aldrich). Cut the ends of the bones using sterile scissors, and flush the bone marrow out with a syringe filled with cold PBS supplemented with 2% fetal bovine serum (FBS). Collect the flushed bone marrow cells in a sterile 50 mL tube. Gently pipette the bone marrow suspension up and down to break up any clumps. Pass the cell suspension through a 70μm cell strainer to remove any debris and obtain a single-cell suspension. Centrifuge the cell suspension at 300 x g for 5 minutes and discard the supernatant. Resuspend the cell pellet in Dulbecco’s Modified Eagle Medium (DMEM, Sigma-Aldrich, Shanghai, China) supplemented with 10% FBS, 1% penicillin-streptomycin, and 25 ng/mL macrophage colony-stimulating factor (M-CSF, Sigma-Aldrich). Count the cells using a hemocytometer and seed them at a density of 10⁶ cells per 10 cm dish or appropriate cell culture plate. Incubate the cells in a humidified 37°C, 5% CO₂ incubator for 6 days to allow the differentiation of bone marrow cells into macrophages. Replace the culture medium containing M-CSF every 3 days. After the differentiation period, harvest the macrophages using a cell scraper and replate them in new cell culture plates at the desired density. To polarize the macrophages into M1 or M2 phenotypes, treat them with specific stimuli. For M1 polarization, add 100 ng/mL lipopolysaccharide (LPS, Sigma-Aldrich) and 20 ng/mL interferon-gamma (IFN-γ, Sigma-Aldrich) to the culture medium and incubate the cells for 48 hours. For M2 polarization, add 20 ng/mL interleukin-4 (IL-4, Sigma-Aldrich) to the culture medium and incubate the cells for 48 hours.

The measurement of macrophage metabolism, including glucose utilization, lactate production, succinate levels, and oxygen consumption rate, was done as following. The undifferentiated or differentiated macrophages were treated with or without experimental stimuli for a specific duration, after which the conditioned media were collected. For glucose utilization, glucose levels were measured with a glucose assay kit according to the manufacturer’s instructions (Ab102517, Abcam). The glucose levels in the media before and after the incubation period were compared to determine glucose utilization by the macrophages. For lactate production, the conditioned media were used to measure lactate levels with a lactate assay kit following the manufacturer’s protocol (Ab65331, Abcam). The amount of lactate produced by the macrophages was used as an indicator of glycolytic activity. For succinate levels, macrophage cell pellets were collected and extracted metabolites using cold methanol as an extraction buffer. The samples were then centrifuged, and the supernatants were collected for quantification of succinate levels using a succinate assay kit, following the manufacturer’s instructions (Ab204718, Abcam). For measuring oxygen consumption rate (OCR) to evaluate the mitochondrial function and oxidative phosphorylation in macrophages, a Seahorse XF Analyzer was used (Aligent, Beijing, China). Seed macrophages in an appropriate cell culture microplate and equilibrate them in a CO₂-free incubator for 1 hour before analysis, following the manufacturer’s protocol.

RIPA buffer was used to lyse cells to collect proteins. ELISA for human HDAC3 (LS-F55924, LSBio, Seattle, WA, USA), human enhancer of zeste homolog 2 (EZH2; LS-F7278, LSBio), human DNA Methyltransferase 1 (DNMT1; LS-F7340, LSBio), human sirtuin 1 (SIRT1; LS-F12606, LSBio), human jumonji domain containing 3 (JMJD3; LS-F74347, LSBio), mouse IL-6 (Ab100713, Abcam), mouse IL-1β (Ab197742, Abcam), mouse tumor necrosis factor alpha (TNFα; Ab208348, Abcam), mouse interferon gamma (IFNγ; Ab282874, Abcam), mouse arginase 1 (ARG1; Ab269541, Abcam), mouse CD163 (Ab272204, Abcam), mouse α-smooth muscle actin (α-SMA; NBP2-66429, Novus Biologicals, Shanghai, China), Vimentin (LS-F7624, LSBio, Seattle, WA, USA) and mouse Collagen IV (LS-F20750, LSBio) were utilized as per the manufacturer’s instructions.

In this study, the data were presented as individual values along with the mean and standard deviation (SD). The statistical analysis was performed using one-way ANOVA, followed by a Bonferroni post-hoc correction to adjust for multiple comparisons (GraphPad Prism, GraphPad Software, Inc., La Jolla, CA, USA). A p-value less than 0.05 was considered statistically significant. The term “NS” was used to denote instances where there was no significant difference between the groups. Additionally, correlation analyses were conducted using Pearson’s correlation coefficient to assess the strength and direction of the relationships between the variables under investigation.

In this study, we utilized the LCM technique to assess the expression levels of key epigenetic regulators in both CD31+ endothelial cells and CD68+ macrophages within atherosclerotic lesions of human patients ( Figure 1A ; Supplementary Table ). Patient specimens were categorized into non-diabetic and diabetic groups based on their serum HbA1C percentages ( Figure 1B ). We evaluated the expression of five major epigenetic regulators, namely HDAC3, EZH2, DNMT1, SIRT1, and JMJD3, in CD31+ endothelial cells and CD68+ macrophages from the plaques of both non-diabetic and diabetic cases using ELISA ( Figures 1C–G ). While no significant differences were observed in the expression levels of all five epigenetic regulators in CD31+ endothelial cells or in the expression of four of these regulators (EZH2, DNMT1, SIRT1, and JMJD3) in CD68+ macrophages between non-diabetic and diabetic cases ( Figures 1C–G ), we detected a notable increase in HDAC3 levels in macrophages from diabetic patients compared to those from non-diabetic individuals ( Figure 1C ). Thus, plaque macrophages in diabetic patients exhibit significantly elevated HDAC3 expression.

We subsequently evaluated the correlation between macrophage HDAC3 expression levels and lipid metabolic parameters, such as serum Low-density Lipoprotein (LDL), High-density Lipoprotein (HDL), Triglycerides (TRIG), as well as glucose levels (HbA1C%) in patients ( Supplementary Table ). Our findings revealed a negative correlation between macrophage HDAC3 levels and serum HDL ( Figure 2A ), while a positive correlation was observed with serum LDL ( Figure 2B ) and TRIG ( Figure 2C ). Additionally, a positive correlation was found between macrophage HDAC3 levels and patients’ blood glucose levels ( Figure 2D ). These results, demonstrating a correlation between macrophage HDAC3 expression and lipid metabolism as well as glucose levels in patients, suggest that HDAC3-mediated epigenetic alterations in macrophages may contribute to the metabolic changes associated with AS in diabetes.

In vitro, we investigated the impact of high glucose on glucose utilization, lactate production, succinate levels, oxygen consumption, and polarization in both undifferentiated and differentiated mouse BMDMs. Our findings demonstrated that high glucose significantly enhanced glucose utilization ( Figure 3A ), lactate production ( Figure 3B ), succinate levels ( Figure 3C ), and oxygen consumption ( Figure 3D ) in undifferentiated BMDMs as well as in differentiated BMDMs (M1 by LPS and IFNγ; M2 by IL-4). These results indicate that elevated glucose levels boost macrophage metabolism, which may contribute to their activation and involvement in plaque formation.

We proceeded to examine the effects of high glucose on macrophage polarization by evaluating the expression levels of several critical M1/M2 macrophage markers. Our findings revealed that high glucose significantly upregulated M1 markers, such as IL-6 ( Figure 4A ), IL-1β ( Figure 4B ), TNFα ( Figure 4C ), and IFNγ ( Figure 4D ), while it notably downregulated M2 markers, including Arg1 ( Figure 4E ) and CD163 ( Figure 4F ), in undifferentiated BMDMs. These observations suggest that elevated glucose levels promote proinflammatory polarization of macrophages, which may contribute to their transformation into foam cells and subsequent plaque formation.

To investigate whether increased HDAC3 expression in macrophages might contribute to the development of AS in vivo, we generated macrophage HDAC3 knockout mice in an ApoE-KO background. In pLys-Cre mice, the promoter of the macrophage-specific endogenous lysozyme 2 gene was utilized to drive Cre recombinase expression (17). In HDAC3floxed mice, loxP sites were placed flanking exons 4-7 of the HDAC3 gene to facilitate Cre-mediated HDAC3 depletion (18). We first crossbred pLys-Cre and HDAC3floxed mice to generate macrophage-specific HDAC3 knockout mice and then crossed them with ApoE-KO mice to produce ApoE-KO/pLys-Cre/HDAC3floxed mice. ApoE-KO mice were used as controls without macrophage depletion of HDAC3 in ApoE-KO/pLys-Cre/HDAC3floxed mice. Both strains of mice were fed either a high-fat diet (HFD) or a normal diet (ND), resulting in four experimental groups: ApoE-KO mice with ND, ApoE-KO mice with HFD, ApoE-KO/pLys-Cre/HDAC3floxed mice with ND, and ApoE-KO/pLys-Cre/HDAC3floxed mice with HFD ( Figure 5A ). After 12 weeks of ND/HFD treatment, we observed that ND did not lead to changes in body weights between ApoE-KO and ApoE-KO/pLys-Cre/HDAC3floxed mice, while HFD resulted in significantly higher body weights in ApoE-KO/pLys-Cre/HDAC3floxed mice compared to ApoE-KO mice ( Figure 5B ). Furthermore, no differences were found in fasting blood sugar ( Figure 5C ) or glucose responses ( Figure 5D ) between the two strains when treated with either ND or HFD. Subsequently, these mice were analyzed for the development and severity of AS.

We assessed the impact of HDAC3 depletion in macrophages on the development of AS in these diabetes-prone, AS-susceptible mice. We observed a significant increase in aortic lesion size in the aortic sinus of HFD-treated mice compared to ND-treated mice in both strains ( Figure 6A ). However, macrophage-specific HDAC3 depletion significantly reduced the aortic lesion size in HFD-treated mice ( Figure 6A ). Additionally, the aortic sinus exhibited a significant increase in plaque lipid content in HFD-treated mice compared to ND-treated mice in both strains ( Figures 6B, C ). Nevertheless, macrophage-specific HDAC3 depletion substantially reduced plaque lipid content in HFD-treated mice ( Figures 6B, C ). We then isolated the aortic arch to measure the levels of mesenchymal markers α-SMA, Vimentin, and Collagen IV using ELISA. We detected significantly elevated levels of α-SMA, Vimentin, and Collagen IV in HFD-treated mice in both strains. However, these increases in all mesenchymal markers were significantly attenuated in HFD-treated ApoE-KO/pLys-Cre/HDAC3floxed mice compared to ApoE-KO mice ( Figure 6D ). Collectively, these findings suggest that macrophage-specific HDAC3 depletion significantly ameliorates AS severity in diabetes-prone, AS-susceptible mice.