Single-cell transcriptomics reveals subtype-specific molecular profiles in Nrf2-deficient macrophages from murine atherosclerotic aortas

Abstract

Nuclear factor erythroid 2-related factor 2 (Nrf2) is a transcriptional regulator of antioxidant and anti-inflammatory response in all cell types. It also activates the transcription of genes important for macrophage function. Nrf2 activity declines with age and has been closely linked to atherosclerosis, but its specific role in this vascular pathology is not clear. Atherosclerotic plaques contain several macrophage subsets with distinct, yet not completely understood, functions in the lesion development. The aim of this study was to analyze the transcriptome of diverse Nrf2-deficient macrophage subpopulations from murine atherosclerotic aortas. Mice with transcriptionally inactive Nrf2 in Cdh5-expressing cells (Nrf2 ^Cdh5tKO) were used in the experiments. These mice lack transcriptional Nrf2 activity in endothelial cells, but also in a proportion of leukocytes. We confirmed that the bone marrow-derived and tissue-resident macrophages isolated from Nrf2 ^Cdh5tKO mice exhibit a significant decline in Nrf2 activity. Atherosclerosis was induced in Nrf2 ^Cdh5tKO and appropriate control mice via adeno-associated viral vector (AAV)-mediated overexpression of murine proprotein convertase subtilisin/kexin type 9 (Pcsk9) in the liver and high-fat diet feeding. After 21 weeks, live aortic cells were sorted on FACS and single-cell RNA sequencing (scRNA-seq) was performed. Unsupervised clustering singled out 13 distinct aortic cell types. Among macrophages, 9 subclusters were identified. Differential gene expression analysis revealed cell subtype-specific expression patterns. A subset of inflammatory macrophages from atherosclerotic Nrf2 ^Cdh5tKO mice demonstrated downregulation of DNA replication genes (e.g. Mcm7, Lig1, Pola1) concomitant with upregulation of DNA damage sensor Atr gene. Atherosclerotic Nrf2 ^Cdh5tKO Lyve1+ resident macrophages showed strong upregulation of IFN-stimulated genes, as well as changes in the expression of death pathways-associated genes (Slc40a1, Bcl2a1). Furthermore, we observed subtype-specific expression of core ferroptosis genes (e.g. Cp, Hells, Slc40a1) in inflammatory versus tissue resident macrophages. This observation suggested a link between ferroptosis and inflammatory microenvironment appearing at a very early stage of atherogenesis. Our findings indicate that Nrf2 deficiency in aortic macrophages leads to subtype-specific transcriptomic changes associated with inflammation, iron homeostasis, cell injury or death pathways. This may help understanding the role of aging-associated decline of Nrf2 activity and the function of specific macrophage subtypes in atherosclerotic lesion development.

Introduction

Atherosclerosis is an inflammatory disease of the wall of large- and medium-sized arteries. It can begin in childhood with the development of fatty streaks due to an accumulation of lipids in the intimal layer of the artery (1). With time, the fatty streak evolves into established plaque with the thick fibrous cap produced by vascular smooth muscle cells. Then, increasing lipid content and numbers of inflammatory macrophages, enlarged necrotic core and thinner fibrous cap lead to the formation of unstable plaque and sometimes its rupture. That end stage disease is typical for humans, but practically does not occur in mouse models of atherosclerosis (2).

Inflammation and oxidative stress are crucial elements in progression of atherosclerosis. In fact, these two processes are interrelated and form a vicious circle during atherogenesis. Reactive oxygen species induce the expression of inflammatory cytokines, chemokines and soluble mediators of inflammation via activation of various transcription factors (3). In turn, cytokines and chemokines produced by inflammatory cells recruit additional inflammatory cells to the sites of inflammation, what increases oxidative stress and exacerbates this adverse cycle.

Macrophages are the most numerous immune cells in the pathogenesis of atherosclerosis, present through all stage of the disease from lesion initiation to plaque rupture (4). Lipoproteins sequestered by macrophages in the arterial wall undergo various modifications including oxidation and aggregation. These pro-inflammatory particles lead to activation of the overlying endothelial cells and recruitment of monocytes differentiating to macrophages inside the vessel wall (5). However, lesional macrophages accumulate not only through recruitment and differentiation of circulating monocytes, but also via local proliferation, (trans)-differentiation of vascular smooth muscle cells or local progenitors (4). During atherosclerosis, macrophages are exposed to many environmental signals, which modulate their functional phenotypes. For e.g. cholesterol crystals, found not only in advanced plaques, but also at early stages of atherosclerotic lesions, were shown to activate the caspase-1-activating-NLRP3 inflammasome and thus were shown to act as a proinflammatory stimulus in LPS-primed human peripheral blood mononuclear cells in culture and in vivo in mice (6).

Recent research involving the single-cell transcriptomics identified several distinct macrophage clusters within human and mouse atherosclerotic plaques (7–9). These populations may play either beneficial or harmful functions in atherosclerosis, which may depend on the stage of the disease (10, 11). Undeniably, the beneficial role of plaque macrophages, especially at the early stage of the lesion development, results from their ability to scavenge cytotoxic lipoproteins and remove dead cells. But, similarly to other cells in the plaque, macrophages also undergo different forms of cell death. It is currently proposed that targeting the diverse types of macrophage death may affect different stages of atherosclerosis development (12).

Nuclear factor erythroid 2-related factor 2 (Nrf2), a transcription factor encoded in humans by the NFE2L2 gene, is regarded as a master transcriptional regulator of cellular redox homeostasis. Nrf2 induces the expression of a battery of genes involved in defence against oxidative stress (13). This transcription factor is ubiquitously expressed in vascular cells, where it regulates expression of various atheroprotective enzymes, such as heme oxygenase-1 (14) or peroxiredoxin-1 (15). Moreover, Nrf2 is involved in regulation of iron metabolism. It may affect the labile iron pool via regulation of the expression of ferroportin, an iron transporter (16, 17) and ferritin, an iron storage protein (18). Nrf2 also plays an important role in biosynthesis and degradation of iron-containing heme via regulation of the expression of genes such as ATP binding cassette subfamily B member 6 or ferrochelatase (19), as well as the above-mentioned heme degrading enzyme - heme oxygenase-1 (14). Noteworthy, Nrf2 activity declines with age what may lead to disturbed redox homeostasis, cell senescence and even cell death (18, 20, 21).

The data on the role of Nrf2 in atherosclerosis are contradictory. Nrf2 was shown to confer protection against foam cells formation by regulating the expression of antioxidant proteins and scavenger receptors in bone marrow transplantation model (22). However, a decreased susceptibility to apolipoprotein E (ApoE)-mediated atherosclerotic plaque formation was shown in Nrf2^−/−ApoE^−/− mice (23, 24). On the other hand, increased atherosclerosis in low-density lipoprotein receptor–deficient (Ldlr^–/–) mice was observed following Nrf2^−/− bone marrow transplantation (25). Another study demonstrated that the ablation of Nrf2 in the bone marrow-derived cells suppresses atherosclerotic lesion area (26). More recently, global Nrf2 deficiency was shown to promote the signs of plaque instability in Ldlr^−/−ApoB^100/100 mice via increased inflammation and oxidative stress (27).

The contradictory data on the role of Nrf2 in atherosclerosis may result from different approaches in atherosclerosis induction, the genetic background of mice and/or a combination of systemic and local effects in conventional Nrf2 knockout mice. Thus, it is crucial to investigate the association between Nrf2 signaling and fate of specific vascular cell types in atherosclerosis. In this study we aimed to analyze at the single-cell level the transcriptome of diverse macrophage subpopulations from murine atherosclerotic aortas in the early stages of the disease development. Our data indicate that decreased Nrf2 transcriptional activity affects expression of genes involved in inflammatory pathways, cell proliferation and programmed cell death in a subtype-specific manner.

Materials and methods

Animals

All mouse experiments were carried out in accordance with Directive 2010/63/UE of the European Parliament on the protection of animals used for scientific purposes and approved by the 2nd Institutional Animal Care and Use Committee (IACUC) in Kraków, Poland (approval number 142/2020). The animals were housed in specific pathogen-free (SPF) conditions with water and food available ad libitum. To generate mice with Nrf2 transcriptional knockout in Cdh5-expressing cells (Nrf2^Cdh5tKO) and control mice (Nrf2^flox/flox), C57BL/6-Nfe2l2^tm1.1Sred/SbisJ mice (The Jackson Laboratory, Strain #:025433) were crossed with B6;129-Tg(Cdh5-cre)1Spe/J mice (The Jackson Laboratory, Strain #: 017968). Genotyping of animals was performed by PCR on the DNA isolated from the tail tips.

AAV8-Pcsk9 vector production

Production of the adeno-associated viral vector serotype 8 coding for proprotein convertase subtilisin/kexin type 9 (AAV8-Pcsk9) was performed using CellRoll Roller Bottle system (Integra) in three-plasmid Helper-free system as previously described (25). Briefly, AAV293 cells seeded on collagen-precoated Roller Bottles (Corning) were cultured until 60% confluence and transfected with 130 µg of pHelper (Stratagene), 100 µg of pAAV2/8 (kindly provided by Prof. James Wilson, University of Pennsylvania), and 90 µg of pAAV/D377Y-mPCSK9. The plasmid encoding murine Pcsk9 under the control of liver-dependent promoter (pAAV/D377Y-mPCSK9) was a gift from Jacob Bentzon (Addgene plasmid # 58376; http://n2t.net/addgene:58376; RRID: Addgene_58376). According to Addgene specification, the plasmid possessed an alternate 5’ITR sequence which, in our hands, prevented AAV production. Thus, to replace the alternate 5’ITR with the functional 5’ITR, the pAAV-MCS (Stratagene) and pAAV/D377Y-mPCSK9 were digested with Not1 (New England Biolabs) according to the vendor’s protocol. The fragment encoding functional 5’ITR from pAAV-MCS (backbone) and the fragment with Pcsk9 coding gene from pAAV/D377Y-mPCSK9 (GOI) were excised from 1% agarose gel and purified using Zymoclean Gel DNA Recovery Kit (ZymoResearch). Then the backbone was dephosphorylated and ligated with GOI. The proper modified plasmid sequence was confirmed by restriction analysis and functional tests. AAV293 cell transfection was performed using PEI MAX (2.58 mg/ml, 1 µl per 1 µg of DNA). Then, 72 hours after transfection, the cells were detached and washed with PBS. The cells were suspended in a small volume (2-3 ml) of PBS containing calcium and magnesium and subjected to 3 freeze-thaw cycles by placing them in liquid nitrogen and thawing in a 37°C water bath with vigorous mixing after each cycle. The lysate was incubated with HS nuclease (50 U/ml; MoBiTec) for 1 hour at 37°C. Then lysate was centrifuged (4000 × g, 30 min, 4°C) twice, the supernatant was collected and stored at -20°C. Next, AAV8-Pcsk9 vector was purified by Iodixanol (OptiPrep™, Sigma-Aldrich) gradient ultracentrifugation as previously described (28).

The AAV8-Pcsk9 titer was determined using quantitative PCR (qPCR). Briefly, the DNA was isolated from 5 µl of stock using the phenol-chloroform extraction method. Quantification of genome copies in the sample was performed using TaqMan-quantitative real-time PCR. The reaction was performed using primers recognizing the sequence of promoter (HCR_F: 5’-TGGAGTGCAGTGACACAATC-3’, HCR_R: 5’-AGGCCTGTAATCCCAGTTAC-3’), the custom Taqman probe (HCR_probe: 5’-6-FAM-GTAGCTGGGATTACAAGCATGTGC-BHQ-1-3’) and TaqMan Gene Expression Master Mix (Applied Biosystems). The serial dilutions of linearized modified pAAV/D377Y-mPCSK9 were used to generate a standard curve. The reaction conditions were applied according to the manufacturer’s protocol. The AAV titer was calculated as the number of vector genomes per µl of purified stock vector preparation.

Atherosclerosis model

The experiment started when the mice were 3-month-old. To induce hyperlipidemia and atherosclerosis, male Nrf2^Cdh5tKO and Nrf2^flox/flox mice received a single tail vein injection of AAV8-Pcsk9 at a dose of 1×10¹¹ vector genomes (vg) per mouse and were subsequently placed on a high-fat diet (35% fat; 60% kcal from fat; ZooLab, DP-1E-60S) for 21 weeks. Normocholesterolemic 3-month-old male Nrf2^Cdh5tKO and Nrf2^flox/flox control mice received a single tail vein injection of saline and were fed a standard diet. At the end of the experiment, the mice were euthanized by inhalation of an overdose of isoflurane. Next, the blood was collected by cardiac puncture using a syringe containing heparin (10 U/ml) and used for blood cell count analysis (ABC Vet, Horiba ABX) and preparation of plasma. Plasma was obtained by centrifugation (2000 × g, 10 min, 4°C) and stored at -80°C. After blood collection, the mice were perfused with PBS containing 0.5 U/ml heparin and the aortic arch, heart, and liver samples were collected for further analysis. In the western blot analysis of liver samples of both AAV8-Pcsk9-treated atherogenic groups we observed highly efficient depletion of the LDL receptor compared with control animals that did not receive AAV8-Pcsk9 (not shown).

Plasma cholesterol and triglyceride analysis

The concentration of cholesterol of plasma LDL/VLDL lipoproteins was evaluated with HDL and LDL/VLDL Quantitation Kit (Sigma-Aldrich) according to vendor’s protocol. The level of triglycerides was determined with Liquick-Cor TG kit (Cormay, Poland) according vendor’s instruction.

Histological examination

The hearts and brachiocephalic arteries (BCAs) collected from the atherosclerotic Nrf2^Cdh5tKO and Nrf2^flox/flox mice were embedded in OCT compound. Sequential cross sections (8 µm thick) were fixed in formalin, immersed in 60% isopropyl alcohol, and stained with freshly prepared Oil Red O working solution (0.5% Oil Red O in isopropanol was mixed with distilled water in a ratio 3:2) for 5 min. Then the sections were briefly immersed in 60% isopropyl alcohol solution and subsequently examined under a light microscope.

Isolation and culture of bone marrow-derived macrophages

The healthy 5-month-old Nrf2^Cdh5tKO and Nrf2^flox/flox mice were euthanized by inhalation of an overdose of isoflurane. The BMDM were obtained according to a protocol described previously (29), with modifications. Briefly, bone marrow cells were collected by flushing the femurs and tibias with sterile physiological saline. The cells were washed with PBS, red blood cells were lysed and the remaining cells were cultured in RPMI medium supplemented with 10% fetal bovine serum (FBS, Biowest), penicillin (100 U/ml) and streptomycin (100 µg/ml; pen-strep, Gibco), and 10 ng/ml of purified human recombinant M-CSF (R&D Systems) for 5 days.

Isolation of Kupffer cells

The Kupffer cells were isolated from the liver of the same animals, which were used for BMDM isolation. Kupffer cells isolation was performed as previously described (28).

RT-qPCR

The cells were lysed in Fenozol (A&A Biotechnology). Total RNA was isolated using Total RNA Mini kit (A&A Biotechnology) and then reverse transcribed into cDNA using a High-Capacity cDNA Reverse Transcription Kit (Life Technologies) according to the manufacturer’s instruction. The qPCR reaction was performed with primers targeting Nrf2 exon 3, Nrf2 exon 5 sequence, Nqo1 gene, Gclm gene, Eef2 reference gene (Nrf2_ex3_F: 5’-CAGAGACATTCCCATTTGTAG-3’, Nrf2_ex3_R: 5’-ATTCGGGAATGGAAAATAGC-3’, Nrf2_ex5_F: 5’-CATTCCCGAATTACAGTGTC-3’, Nrf2_ex5_R: 5’-GGAGATCGATGAGTAAAAATGG-3’, Nqo1_F: 5’-GGTTTGGAGTCCCTGCCATT-3’ Nqo1_R: 5’-GTGGATCCCTTGCAGAGAGT-3’, Gclm_F: 5’-TGGAGTTCCCAAATCAGCCC-3’, Gclm_R: 5’-CAACTCCAAGGACGGAGCAT-3’, Eef2_F: 5’-GACATCACCAAGGGTGTGCAG-3’, Eef2_R: 5’-TCAGCACACTGGCATAGAGGC-3’) and AceQ qPCR SYBR Green Master Mix (Vazyme) in StepOne Plus Real-Time PCR system (Applied Biosystems). All the procedures were performed according to the manufacturer’s instructions. The level of Nrf2 transcriptional activity (exon 5 deletion) was calculated as a relative expression of Nrf2 exon 5 in comparison to Nrf2 exon 3 expression.

Preparation of single cells suspension from mouse aortas

The aortic arches with main branches including the perivascular tissue were isolated from atherosclerotic Nrf2^Cdh5tKO and Nrf2^flox/flox mice (n=3 per group) and control normocholesterolemic (no AAV transduction and standard diet) Nrf2^Cdh5tKO and Nrf2^flox/flox mice (n=4 per group). The aortas of each group of mice were pooled, cut into small pieces, and digested in 2.5 ml enzyme mix from Multi Tissue Dissociation Kit 2 (Miltenyi Biotec) prepared as described in the protocol provided by the vendor. The samples were placed at 37°C for 30 min with shaking (250 RPM). The suspension was gently pipetted every 15 min. Then the cells were filtered through a 70 μm strainer, centrifuged at 500 × g for 5 min, and resuspended in red blood cells lysis buffer (BioLegend). After 5 min, the cells were centrifuged at 500 × g for 5 min, washed with PBS, and resuspended in 300 µl of PBS containing 2% FBS and 200 ng/ml DAPI. To exclude dead cells and collect only the viable ones, the DAPI-negative cells were sorted with MoFlo XPD (Beckman Coulter) cell sorter.

Library preparation/data analysis/subclustering/DEG analysis

A total of ~12.000 DAPI-negative cells per sample were loaded on a chromium single cell controller to generate single cell GEMs and then cDNA libraries were prepared using Chromium™ Next GEM Single Cell 3’ GEM, Library & Gel Bead Kit v3.1 (10x Genomics, PN-1000121) following manufacturers protocol. Following QC analysis, cDNA libraries were sequenced on Illumina NovaSeq instrument (paired end, single indexing, 28-8-0-91 cycles) at desired sequencing depth of 50 000 reads/cell. Following sequencing output fastq files were mapped to the mm10 mouse reference genome using 10x Genomics Cell Ranger (cellranger-6.1.2) software. Subsequent downstream analysis was performed in R and R studio softwares (packages: Seurat, Patchwork, Dplyr, Harmony and ggplot2). Any possible cell doublets (nFeature>6000) as well as cells expressing less than 200 genes and more than 15% of mitochondrial genes were excluded from analysis. Cells were clustered using Seurat package (v4.3.0). Results were displayed by UMAP and Heatmaps. Main cell populations were annotated based on the expression of 30 marker genes with highest expression within the cluster. Differentially expressed genes between comparable groups were found using FindMarkers function with the parameters test.use=wilcox, min.pct=0.2, tresh.use=0.25, the genes were filtered based on the p value <0.05.

Kegg pathways analysis

The differentially expressed genes (DEGs) were uploaded to the KEGG Pathways (DAVID; version DAVID 2021, Dec. 2021) (30) to identify biological pathways enriched in specific mononuclear phagocyte subtypes of 1) atherosclerotic Nrf2^flox/flox mice in comparison to normocholesterolemic Nrf2^flox/flox mice and 2) atherosclerotic Nrf2^Cdh5tKO mice in comparison to atherosclerotic Nrf2^flox/flox mice. The threshold for the statistical significance was p<0.05. The plots were drawn using the “ggplot2” packages in the R software (version 4.2.1).

Statistical analysis

Results are presented as mean ± standard deviation (SD) unless stated otherwise. The Student’s t-test was used to compare the means of two groups and a value of p<0.05 was considered statistically significant. Statistical analyses were performed with GraphPad Prism (GraphPadSoftware Inc., San Diego, CA).

Results

Expression of Nrf2 is significantly decreased in macrophages from Nrf2^Cdh5tKO mice

Previous studies investigating the effect of macrophage-specific loss of Nrf2 on atherosclerosis were based on bone marrow transplantation of Nrf2-deficient bone marrow to lethally irradiated Ldlr^−/− (25, 31) or ApoE^−/− mice (26). Unfortunately, these results did not provide a clear answer on the role of macrophage-specific Nrf2 in atherosclerosis. Here we used animals with transcriptionally inactive Nrf2 in cadherin 5 (Cdh5)-expressing cells (Nrf2^Cdh5tKO) and the control Nrf2^flox/flox mice. In Nrf2^Cdh5tKO mice, Cre recombinase excises exon 5 of Nrf2 gene, which is responsible for DNA binding and activation of transcription, making Nrf2 transcriptionally inactive. As a major component of the endothelial cell adherens junction, Cdh5 (VE-cadherin) plays an important role in vascular permeability and angiogenesis (32). Thus, Cdh5-Cre is frequently used for endothelial gene deletion. However, Cdh5-dependent constitutive Cre recombinase expression is present also in a proportion of hematopoietic cells (33–35). To verify whether Nrf2 transcriptional activity is affected in macrophages in our mouse model, we isolated macrophages of two different origins: bone marrow-derived macrophages (BMDM) and Kupffer cells from the liver. For both cell types, we confirmed a decreased level of Nrf2 exon 5 ( Figures 1A, B ) in comparison to Nrf2 exon 3 (the one not affected by Cre recombinase). In support of the decreased Nrf2 activity, we observed lower expression of the two selected Nrf2 target genes: NAD(P)H quinone dehydrogenase 1 (Nqo1) and glutamate-cysteine ligase modifier subunit (Gclm) in BMDM ( Figures 1C, E , respectively), as well as in Kupffer cells ( Figures 1D, F , respectively). In addition, the complete blood count analysis showed no difference in white blood cells (WBC, Figure 1G ) and monocytes ( Figure 1H ) between analyzed healthy mice. It indicates that in steady state conditions decreased transcriptional activity of Nrf2 in hematopoietic and endothelial lineages does not affect formation of WBC and monocytes.

Figure 1.

Evaluation of Nrf2 expression in macrophages from Nrf2^Cdh5tKO mice. (A) The qPCR confirming excision of exon 5 in Nrf2^Cdh5tKO BMDM presented as a ratio of exon 5 to exon 3. (B) The qPCR confirming the excision of exon 5 Nrf2^Cdh5tKO Kupffer cells presented as a ratio of exon 5 to exon 3. (C) The qPCR analysis of Nqo1 in BMDM. (D) The qPCR analysis of Nqo1 in Kupffer cells. (E) The qPCR analysis of Gclm in BMDM. (F) The qPCR analysis of Gclm in Kupffer cells. (G) The number of WBC in control Nrf2^flox/flox and Nrf2^Cdh5tKO mice. (H) The percentage of blood monocytes in control Nrf2^flox/flox and Nrf2^Cdh5tKO mice. Results are presented as mean ± SD (*p<0.05, ** p<0.01).

Pcsk9 overexpression and high fat feeding changes the blood lipids level and induces early stage plaque development in the aortic arch branches of Nrf2^flox/flox and Nrf2^Cdh5tKO mice

Our atherosclerosis model was based on intravenous delivery of the AAV8 vector encoding Pcsk9, a protein controlling the expression of LDL receptor in the liver what, in combination with high-fat feeding, induces hypercholesterolemia in mice (36). Two groups of mice were subjected to this treatment: Nrf2^Cdh5tKO and Nrf2^flox/flox. The mouse plasma samples were collected 21 weeks after introducing proatherogenic factors. The analysis revealed increased level of plasma cholesterol carried by LDL/VLDL lipoproteins ( Figure 2A ) and increased level of triglycerides ( Figure 2B ) in atherosclerotic mice when compared to control littermates, with no significant differences between both genotypes. The Oil Red O (ORO) staining of aortic roots cross-sections confirmed the presence of plaques and atherosclerotic phenotype in both groups of mice ( Figure 2C ). The ORO staining of brachiocephalic artery cross-sections showed small lipid deposits within the aortic wall corresponding to an early stage of plaque development in the aortic arch branches of both genotypes ( Figure 2D ). No lipid deposits were detected within the aortic walls of normocholesterolemic Nrf2^flox/flox and Nrf2^Cdh5tKO mice (data not shown). After 21 weeks of atherosclerosis induction, the complete blood count analysis showed comparable numbers of WBC ( Figure 2E ) and monocytes ( Figure 2F ) in Nrf2^flox/flox and Nrf2^Cdh5tKO mice.

Figure 2.

Characterization of atherosclerotic phenotype of Nrf2^flox/flox and Nrf2^Cdh5tKO mice. (A) The level of plasma cholesterol carried by LDL/VLDL lipoproteins in control healthy and atherosclerotic Nrf2^flox/flox and Nrf2^Cdh5tKO mice. (B) The level of plasma triglycerides in control healthy and atherosclerotic Nrf2^flox/flox and Nrf2^Cdh5tKO mice. (C) The staining of lipid deposits within aortic root of Nrf2^flox/flox and Nrf2^Cdh5tKO mice 21 weeks after AAV8-Pcsk9 injection and HFD. (D) The staining of lipid deposits in the brachiocephalic artery of Nrf2^flox/flox and Nrf2^Cdh5tKO mice 21 weeks after AAV8-Pcsk9 injection and HFD. Lumen of the blood vessel is marked with an ‘L’. The arrows indicate lipid deposits. (E) The number of WBC in the blood of Nrf2^flox/flox and Nrf2^Cdh5tKO mice 21 weeks after AAV8-Pcsk9 injection and HFD. (F) The percentage of monocytes in WBC population in the blood of atherosclerotic Nrf2^flox/flox and Nrf2^Cdh5tKO mice 21 weeks after AAV8-Pcsk9 injection and HFD. Results are presented as mean ± SD.

Single-cell RNA sequencing identifies several types of mononuclear phagocytes in the aortas of Nrf2^flox/flox and Nrf2^Cdh5tKO mice

Previous studies using single-cell RNA sequencing (scRNA-seq) highlighted the complexity of macrophage biology and their heterogeneity in human and murine aortas (7, 9). In our study we applied this technology to analyze the transcriptomic changes in different macrophage subpopulations deficient in Nrf2 transcriptional activity in the context of their possible contribution to atherosclerosis development. We collected control and atherosclerotic aortic arches with branches from Nrf2^flox/flox and Nrf2^Cdh5tKO mice. Then, the pooled fragments of aortas of each group were subjected to scRNA-seq ( Figure 3A ). The unsupervised clustering of 14 125 cells identified, among others, two clusters of macrophages. They were annotated as proinflammatory and tissue-resident based on the expression of Pf4, Mrc1, C1qa, C1qb, C1qc genes, and Lyz1, Rentla, Lyz2, Cd74 genes, respectively. Within the merged (proinflammatory and tissue-resident) macrophage population of 1821 cells, 92.3% were Cd45(Ptprc)-positive, 92.5% were Cd68-positive, 82.3% were F4/80(Adgre1)-positive, and 73.1% were Cd11b(Itgam)-positive ( Figure 3B ). The variation in the content of these immune cell markers suggested that the population is heterogenous and, besides macrophages, consists of other cell types. The subclustering analysis of this heterogenous population, here called mononuclear phagocytes, revealed 9 subpopulations ( Figure 3C ) with a high abundance of lymphatic vessel endothelial receptor-1 (Lyve1+) resident macrophages, proinflammatory macrophages, cavity macrophages and monocytes ( Figure 3D ). The other subtypes - the population of stem, endothelial and mesenchymal (SEM) cells, hematopoietic progenitor cells (HPC)/fibrocytes, aortic intima resident (AIR) macrophages, conventional type 1 dendritic cells (cDC1) and conventional type 2 dendritic cells (cDC2) were less abundant ( Figure 3D ). Lyve1+ resident macrophages were characterized according to previous reports (8, 9, 37) by the expression of Lyve1, Folr2, Hmox1, Maf, and Cd209f, whereas cavity macrophages (8) by Lyz1, Fn1, Ear2, Rentla, and Lpl ( Figure 3E ). The expression of Cxcl1, Nlrp3, Sdc4, Cxcl16, and Ccl3 genes characterized both proinflammatory and AIR macrophages (8), but the presence of typical AIR macrophage markers Cd72, Ccrl2, Ctss, Acp5, Mpeg1 (8) distinguished this population from the proinflammatory subset ( Figure 3E ). Monocyte cluster (8) showed high expression of Plac8, Chil3, Msrb1, Clec4e, and Slpi ( Figure 3E ); HPC/fibrocytes (38) were enriched in Dcn, Sparc, Gsn, Col3a1, Mgp ( Figure 3E ), while SEM cluster (39) highly expressed Lum, Runx1, Apoe, Acta2, Ctsb ( Figure 3E ). For cDC1 (8) and cDC2 (8) we found overlapping expression of classical marker genes with higher expression of Naaa, Cst3, Ifi205, Cd24a, H2afz for cDC1 ( Figure 3E ) and Napsa, H2-ab1, H2-eb1, Traf1, Wnt11 for cDC2 ( Figure 3E ).

Figure 3.

Identification of distinct mononuclear phagocytes in the analyzed fragments of aortas of Nrf2^flox/flox and Nrf2^Cdh5tKO mice. (A) Schematic depiction of the scRNA-seq procedure. (B) The percentage of cells expressing classical macrophage markers: Cd45, Cd68, F4/80, Cd11b within the whole macrophage population before subclustering. (C) UMAP representation of integrated scRNA-seq gene expression data from normocholesterolemic Nrf2^flox/flox (n=4) and Nrf2^Cdh5tKO (n=4), as well as atherosclerotic Nrf2^flox/flox (n=3) and Nrf2^Cdh5tKO (n=3) murine aortic arches and branches with identification of the major cell types. (D) Proportions of distinct cell types singled out among whole macrophage population. (E) Dot plot of average gene expression of the indicated marker transcripts for each cell cluster.

The analysis of classical macrophage markers - Cd45, Cd68, F4/80, Cd11b ( Figures 4A–D , respectively) revealed their lowest expression in HPC/fibrocytes and SEM cells. Moreover, the level of Cd68, F4/80, and Cd11b ( Figures 4B–D , respectively) was moderately or strongly decreased in cDC1 and cDC2 cell clusters. Analysis of the abundance of various populations of mononuclear phagocytes in the aortas of healthy and atherosclerotic mice of both genotypes revealed a higher percentage of monocytes (9.18% and 5.25% respectively for normocholesterolemic Nrf2^flox/flox and Nrf2^Cdh5tKO vs. 27.06% and 10.17% respectively for atherosclerotic Nrf2^flox/flox and Nrf2^Cdh5tKO) and cDC2 cells (2.05% and 2.62% respectively for normocholesterolemic Nrf2^flox/flox and Nrf2^Cdh5tKO vs. 11.76% and 6.10% respectively for atherosclerotic Nrf2^flox/flox and Nrf2^Cdh5tKO) in atherogenic conditions ( Figure 4E ). Interestingly, in atherosclerotic Nrf2^Cdh5tKO mice a lower proportion of monocytes was associated with a higher proportion of proinflammatory macrophages (17.44% for atherosclerotic Nrf2^Cdh5tKO vs. 7.65% for atherosclerotic Nrf2^flox/flox).

Figure 4.

Characteristics of the identified mononuclear phagocyte clusters. The expression of typical macrophage genes (A) Cd45, (B) F4/80, (C) Cd68, (D) Cd11b within particular mononuclear phagocyte population shown in the violin plot. (E) Proportions of identified cell subsets in the analyzed fragments of aortas of control normocholesterolemic and atherosclerotic Nrf2^flox/flox and Nrf2^Cdh5tKO mice.

The transcriptome of distinct aortic mononuclear phagocytes is differently affected by decreased Nrf2 transcriptional activity under atherogenic conditions

Next, we analyzed macrophage subtype-specific gene expression using scRNA-seq technology in the aortic arches and branches from control normocholesterolemic and atherosclerotic mice of two genotypes - Nrf2^flox/flox and Nrf2^Cdh5tKO. We focused mostly on three highly abundant clusters that play an important role in plaque development: Lyve1+ resident macrophages, proinflammatory macrophages and monocytes ( Figure 3C ).

Lyve1+ resident macrophages

Lyve1+ resident macrophages are considered to originate from embryonic erythro-myeloid progenitors (40) but, under certain circumstances, circulating monocytes can differentiate into self-maintaining tissue-resident macrophages that resemble their embryonic counterparts. In our analysis, this macrophage subset is characterized by high expression of Lyve1, Folr2, Hmox1, Maf, and Cd209f ( Figures 3E , 5A ), and was the most numerous subpopulation in all samples subjected to scRNA-seq ( Figure 4E ). First, we analyzed how atherosclerosis itself affects gene expression in this macrophage subset. In Lyve1+ resident macrophages of atherosclerotic Nrf2^flox/flox mice, we identified 16 downregulated and 38 upregulated genes when compared to normocholesterolemic Nrf2^flox/flox controls ( Table 1 ). The KEGG pathway analysis ( Figure 5B ) showed that many differentially expressed genes (DEGs) were significantly enriched in cytokine-cytokine receptor interaction (Ccl9, Bmp2, Gdf15, Il1b, Ccl4, Cxcl13, Ccr2), C-type lectin receptor signaling pathway (Cd209d, Clec4b1, Cd209b, Il1b, Lsp1), chemokine signaling pathway (Lyn, Ccl9, Ccl4, Cxcl13, Ccr2), NF-kappa B signaling pathway (Lyn, Il1b, Ccl4). The majority of the identified DEGs were associated with a pro-inflammatory cell phenotype.

Figure 5.

The analysis of transcriptomic changes in Lyve1+ resident macrophages. (A) Expression of marker genes identified for Lyve1+ resident macrophages projected onto the UMAP plot. (B) KEGG pathway enrichment analysis of upregulated and downregulated genes for atherosclerotic Nrf2^flox/flox vs. control Nrf2^flox/flox and (C) atherosclerotic Nrf2^Cdh5tKO vs. atherosclerotic Nrf2^flox/flox Lyve1+ resident macrophages. The x-axis represents the strength (Log10(observed/expected)) of enriched pathways, whereas the color denotes –log10(pValue) and dot size corresponds to gene count in each pathway. (D) STRING interaction network - graphical representation of Interferon-simulated genes (ISG) cluster. (E) Volcano plot of the top differentially expressed ISG in atherosclerotic Nrf2^Cdh5tKO vs. atherosclerotic Nrf2^flox/flox Lyve1+ resident macrophages.

Table 1.

DEGs in Lyve1+ resident macrophages (atherosclerotic Nrf2^flox/flox vs. normocholesterolemic Nrf2^flox/flox).

Upregulated
Gene	p-value	avg_log2FC
Cxcl13	2,39E-08	0,861436
Ccr2	1,47E-07	0,701267
Arhgdib	0,032756	0,641391
Gfra2	4,55E-07	0,488265
Btg1	0,04049	0,472143
Furin	0,008615	0,468665
Cbr2	0,042708	0,448178
Tyrobp	0,021066	0,442919
Cd209d	0,002953	0,436895
Gm42418	0,039606	0,396878
Ugdh	8,43E-11	0,389207
Pde8a	0,00899	0,379347
Mtus2	1,19E-13	0,377547
Cd68	0,023115	0,376554
F830016B08Rik	1,32E-13	0,354216
Fcrls	0,023415	0,347959
S100a8	0,000941	0,336188
Bmp2	0,000161	0,330332
Fstl1	1,66E-17	0,321947
Gpr183	0,008656	0,316971
Cited2	0,014093	0,31122
Ppp1r12a	0,025934	0,310632
Smc4	1,67E-05	0,30858
Nfkbid	0,034186	0,307896
Ccnd1	1,63E-06	0,301505
Iigp1	5,58E-12	0,295462
Anp32e	0,001897	0,293936
Ifi213	5,80E-12	0,285891
Apobec1	5,18E-05	0,285051
Cd209b	0,022909	0,28255
Odc1	0,002432	0,282423
Igfbp4	0,011171	0,277896
Tsc22d1	5,64E-05	0,277236
Clec5a	6,09E-10	0,276247
Mfap4	1,17E-09	0,274002
Ramp1	2,78E-11	0,271583
Myl9	3,51E-05	0,264222
Nr4a3	0,000564	0,251389
Downregulated
Gene	p-value	avg_log2FC
S100a4	0,027411	-0,77728
Ccl4	0,002927	-0,74716
Il1b	2,66E-14	-0,49265
Gdf15	0,047466	-0,45099
Lyve1	0,007756	-0,36497
Ccl9	0,036021	-0,36179
Adgre5	0,046157	-0,33137
Lyn	0,015787	-0,30115
Cd300c2	0,044796	-0,29036
Lilra5	0,008648	-0,2749
Clec4b1	3,51E-11	-0,27359
Marcksl1	0,036684	-0,2674
Tuba4a	1,44E-05	-0,26151
Fxyd5	0,042591	-0,25779
Spic	4,08E-09	-0,25625
Lsp1	0,009876	-0,25385

Next, we investigated the effect of Nrf2 deficiency on the transcriptome of Lyve1+ resident macrophages under atherogenic conditions. The scRNAseq analysis revealed that in this macrophage subset, 17 genes were downregulated and 20 were upregulated in atherosclerotic Nrf2^Cdh5tKO vs. atherosclerotic Nrf2^flox/flox mice ( Table 2 ). The DEGs were subjected to KEGG functional enrichment analysis ( Figure 5C ). Among the significantly enriched KEGG pathways, we found retinoic acid-inducible gene I (RIG-I)-like receptor signaling pathway (Ifih1, Cxcl10, Irf7, Isg15), IL-17 signaling pathway (Cxcl10, Ccl7, S100a8), viral protein interaction with cytokine and cytokine receptor (Cxcl10, Ccl7, Ccr2).

Table 2.

DEGs in atherosclerotic Nrf2^Cdh5tKO Lyve1+ resident macrophages (atherosclerotic Nrf2^Cdh5tKO vs. atherosclerotic Nrf2^flox/flox).

Upregulated
Gene	p-value	avg_log2FC
Cxcl10	0,001899	1,233588
Rsad2	0,022622	0,94986
Cmpk2	0,00588	0,574843
Ifit3	0,000666	0,569893
Ifit2	4,45E-07	0,568756
Isg15	0,001899	0,540737
Oasl1	0,000293	0,482284
Ifih1	0,048154	0,383909
Pbxip1	2,95E-09	0,344674
Ifit3b	0,012109	0,338988
Ptgs1	0,004127	0,320827
Adgre5	0,000181	0,318622
Bcl2a1b	0,003372	0,317612
Cd93	0,001162	0,30791
Rasgef1b	0,000174	0,302515
S100a4	0,022434	0,299475
Rgs1	1,99E-07	0,296326
Irf7	1,51E-05	0,290236
Ifi211	0,009514	0,288928
Oasl2	0,003925	0,258254
Downregulated
Gene	p-value	avg_log2FC
Ccl7	0,026876	-0,63902
Gm12840	4,76E-07	-0,61068
Fcrls	0,003512	-0,52932
Arhgdib	0,019785	-0,50975
Gfra2	0,000528	-0,45824
Ugdh	3,77E-05	-0,43686
Slc40a1	3,25E-07	-0,4262
S100a8	1,13E-05	-0,42252
Mtus2	5,65E-09	-0,3913
Tnfaip2	5,40E-06	-0,35575
Ccr2	3,42E-05	-0,31412
Cd209d	0,021881	-0,30924
F830016B08Rik	0,000383	-0,28949
Mia	1,11E-07	-0,28862
Ramp1	0,02397	-0,27211
Gadd45g	0,003512	-0,27173
Cd209b	0,005131	-0,27114

RIG-I not only participates in antiviral signaling pathways, but may also affect non-viral diseases such as atherosclerosis (41, 42). Once activated, RIG-I initiates a signaling cascade that leads to the activation of transcription factors, including interferon regulatory factors (IRFs) and nuclear factor kappa B (NF-κB), which induce the expression of interferon-related genes and pro-inflammatory cytokines (43). In fact, our results confirmed that observation revealing upregulation of interferon-stimulated genes (ISGs, Figures 5D, E ), Rsad2, Cxcl10, Irf7, Ifit2, Ifit3, Ifit3b, Ifih1, Isg15, Oasl1, Oasl2, Cmpk2, in Nrf2^Cdh5tKO Lyve1+ resident macrophages highlighting that over 50% of upregulated genes were associated with downstream IFN-response. Among them, the key transcription factor interferon regulatory factor 7 (Irf7) was upregulated. Interestingly, comparison of normocholesterolemic Nrf2^Cdh5tKO vs. normocholesterolemic Nrf2^flox/flox mice revealed increased, among others, expression of Rsad2 and Oasl1 ( Table 3 ) suggesting that Nrf2-defficient Lyve1+ resident macrophages can be predisposed to develop an interferon-stimulated response under favorable proatherogenic conditions. In terms of atherosclerosis, we also found that the top upregulated in the atherosclerotic Nrf2^Cdh5tKO mice Cxcl10 gene, was involved in all the most enriched KEGG pathways. Its altered expression, together with other chemokine and chemokine receptors such as Ccl7, Ccr2, Cd97, Rgs1 and S100a8, may play a role in leukocyte chemotaxis and migration.

Table 3.

DEGs in normocholesterolemic Nrf2^Cdh5tKO Lyve1+ resident macrophages (normocholesterolemic Nrf2^Cdh5tKO vs. normocholesterolemic Nrf2^flox/flox).

Upregulated
Gene	p-value	avg_log2FC
Mucl1	4,80E-52	0,456824
Rsad2	1,47E-09	0,452826
Myl4	1,77E-07	0,444174
Oasl1	5,31E-06	0,41565
H2-Eb1	0,010924	0,387727
Cd36	0,00958	0,38566
Sdc3	0,019887	0,343128
Ccr2	7,64E-18	0,341965
Cd209f	4,75E-07	0,336083
Cd209g	7,93E-14	0,319793
H2-DMb1	0,001735	0,312869
Cryab	1,02E-07	0,312213
Arhgdib	0,029654	0,281278
Slc15a3	0,038431	0,278945
Gm21188	7,36E-07	0,260851
Downregulated
Gene	p-value	avg_log2FC
S100a4	0,000526	-0,63943
Retnla	0,000711	-0,3919
Id2	0,037052	-0,38605
Atf3	0,014253	-0,37579
Apoe	0,00021	-0,30321
Ctla2a	2,93E-35	-0,29131
Il1b	0,013429	-0,2505

Among DEGs in atherosclerotic Nrf2^Cdh5tKO Lyve1+ resident aortic macrophages ( Table 2 ), we also identified genes involved in different aspects of cell death regulation. Slc40a1, whose expression may be indirectly regulated by Nrf2 (18), encodes ferroportin and has been implicated in the regulation of iron homeostasis and ferroptotic cell death (44). In addition, we detected the upregulation of Pbxip1 gene which has been shown to promote apoptosis by inhibiting anti-apoptotic Bcl-2 family members, while on the other hand, the anti-apoptotic Bcl2a1b gene was upregulated. We also observed changes in Gadd45g, involved in promoting cell death in response to DNA damage and other stress signals (45); Ifi211 that may inhibit cell growth via p53/TP53 and RB1-dependent and independent pathways (46); Tnfaip2 that may regulate cell death in the context- and cell-dependent manner (47, 48).

Proinflammatory macrophages

Proinflammatory macrophages are the subtype of immune cells mostly derived from monocytes infiltrating the tissues in response to different environmental stimuli. In our scRNAseq analysis, proinflammatory macrophages were the second most abundant cell population within aortic wall, characterized by Cxcl1, Nlrp3, Sdc4, Cxcl16, Ccl3 marker genes ( Figures 3E , 6A ). The atherogenic conditions led to the downregulation of 29 and upregulation of 38 genes ( Table 4 ) in proinflammatory macrophages of Nrf2^flox/flox mice when compared to the normocholesterolemic Nrf2^flox/flox controls. Among the KEGG enriched pathways ( Figure 6B ), we identified cytokine-cytokine receptor interaction (Cxcl10, Cxcl9, Cd4, Cxcl12, Ccl6, Il1b, Il21r, Il2rg, Tnfsf13b, Ccr2), chemokine signaling pathway (Lyn, Cxcl10, Hck, Itk, Cxcl9, Cxcl12, Ccl6, Ccr2), viral protein interaction with cytokine and cytokine receptor (Cxcl10, Cxcl9, Cxcl12, Ccl6, Il2rg, Ccr2), NF-kappa B signaling pathway (Lyn, Cxcl12, Lck, Il1b, Tnfsf13b). Here, similarly to Lyve1+ resident macrophages, the proatherogenic conditions altered mainly the expression of genes involved in immune responses and proinflammatory cell phenotype.

Figure 6.

The analysis of transcriptomic changes in proinflammatory macrophages. (A) Expression of marker genes identified for proinflammatory macrophages projected onto the UMAP plot. (B) KEGG pathway enrichment analysis of upregulated and downregulated genes for atherosclerotic Nrf2^flox/flox vs. control Nrf2^flox/flox and (C) atherosclerotic Nrf2^Cdh5tKO vs. atherosclerotic Nrf2^flox/flox proinflammatory macrophages. The x-axis represents the strength (Log10(observed/expected)) of enriched pathways, whereas the color denotes –log10(pValue) and dot size corresponds to gene count in each pathway. (D) STRING interaction network - graphical representation of DNA replication and repair genes cluster. The upregulated genes are presented in red color, the downregulated in blue. (E) Volcano plot of the top differentially expressed DNA replication and repair genes in atherosclerotic Nrf2^Cdh5tKO vs. atherosclerotic Nrf2^flox/flox proinflammatory macrophages.

Table 4.

DEGs in proinflammatory macrophages (atherosclerotic Nrf2^flox/flox vs. normocholesterolemic Nrf2^flox/flox).

Upregulated
Gene	p-value	avg_log2FC
Trbc1	2,15E-17	1,504752
H2-M2	7,87E-14	1,376942
Il1b	0,000461	1,193289
Glipr1	0,002198	1,0943
Ms4a7	0,002158	0,758854
Tmsb4x	0,001944	0,746055
Tyrobp	3,16E-05	0,663754
Dusp2	0,004814	0,653287
Mylk	2,71E-05	0,625095
Hist1h1c	3,07E-06	0,529187
Cd24a	4,03E-32	0,463632
Dpep2	0,000217	0,428303
Tcf7	1,22E-09	0,39592
Dusp10	7,23E-08	0,392107
Zcwpw1	0,003704	0,385932
Clec4d	7,19E-05	0,381738
Tnfsf13b	0,000751	0,368559
Plk4	1,29E-16	0,366538
Rag1	3,48E-12	0,362176
Hist1h4d	0,000195	0,355199
Jph2	0,001703	0,348627
Klra2	1,12E-05	0,338862
Brip1	2,60E-12	0,335571
Trbc2	3,43E-05	0,31769
Cd4	1,36E-09	0,316175
Lck	2,24E-07	0,308246
Areg	4,52E-14	0,290733
Cxcl10	8,60E-17	0,289706
Rcan1	4,46E-07	0,283454
Itk	9,96E-18	0,278292
Clspn	3,65E-11	0,277299
Ezh2	2,23E-09	0,272555
Slc7a11	8,49E-06	0,272049
Dbf4	1,84E-12	0,262024
3830403N18Rik	0,003334	0,256676
Gimap6	0,000261	0,250604
Smc2	1,58E-11	0,250261
Pclaf	5,46E-20	0,250237
Downregulated
Gene	p-value	avg_log2FC
Ccl6	0,003904	-1,37766
Lyz1	0,040356	-1,12272
Mgl2	0,005566	-1,0127
Pltp	0,007962	-0,9206
Frmd4b	0,001727	-0,79907
Cbr2	0,019255	-0,76134
Clec10a	2,96E-06	-0,73313
Mt1	0,028923	-0,71533
Lyn	0,028313	-0,59088
Cxcl9	7,68E-13	-0,56984
Ciita	1,60E-05	-0,51057
Tppp3	0,008265	-0,49995
Slamf8	0,025101	-0,48998
Cd53	0,038255	-0,38001
Hmgn5	0,003166	-0,37605
Dcn	0,037169	-0,35097
Cxcl12	0,03321	-0,34359
Apex1	0,02609	-0,33454
Ptgis	0,000182	-0,32926
Lrmp	0,011707	-0,32879
Ifi209	0,003534	-0,32394
Pla2g2d	0,000148	-0,30379
Cav2	0,00013	-0,29992
Dqx1	7,97E-35	-0,29513
Rnasel	0,012607	-0,29187
Tsc22d1	0,032091	-0,28398
Gpr34	0,024385	-0,27839
Myh11	0,04207	-0,27628
Gpr183	0,013806	-0,27444

Nrf2 deficiency additionally modified the transcriptome of these cells. The scRNAseq analysis revealed that 38 genes were downregulated and 15 genes were upregulated in proinflammatory macrophages of atherosclerotic Nrf2^Cdh5tKO in comparison to atherosclerotic Nrf2^flox/flox ( Table 5 ). The KEGG pathway analysis ( Figure 6C ) revealed that the majority of DEGs were associated with DNA replication (Pola1, Lig1, Mcm7), hematopoietic cell lineage (Cd14, Il7r, Cd34), NF-kappa B signaling pathway (Lyn, Lck, Cd14), C-type lectin receptor signaling pathway (Clec4d, Cd209f, Clec4e) and cell cycle (Dbf4, Mcm7, Atr). Interestingly, comparison of normocholesterolemic Nrf2^Cdh5tKO vs. normocholesterolemic Nrf2^flox/flox mice revealed that the top upregulated gene in proinflammatory macrophages was Il1b ( Table 6 ). This, together with other DEGs involved in cytokine-cytokine receptor interaction, antigen processing and presentation, cell adhesion molecules, or toll-like receptor pathway, may suggest a more inflammatory phenotype of this Nrf2-deficient macrophage subtype already in steady state conditions.

Table 5.

DEGs in atherosclerotic Nrf2^Cdh5tKO proinflammatory macrophages (atherosclerotic Nrf2^Cdh5tKO vs. atherosclerotic Nrf2^flox/flox).

Upregulated
Gene	p-value	avg_log2FC
Cd209f	9,05E-05	1,042335
Frmd4b	0,016363	0,828339
Lyn	0,02315	0,670389
Slc6a6	0,042782	0,476034
Atr	0,000761	0,43571
Mctp1	0,0184	0,399953
Arrdc3	1,52E-06	0,366464
Apex1	0,00551	0,350366
Ifi209	0,006861	0,342648
Gm21188	0,001551	0,313762
C3	0,035912	0,297601
Uap1	0,016363	0,297434
Fbxo30	0,038538	0,276554
Hmgn5	0,011867	0,272808
Il7r	0,014231	0,256876
Downregulated
Gene	p-value	avg_log2FC
Fabp5	0,010931	-1,9532
Trbc1	0,024954	-1,47603
Mmp12	0,009443	-1,34337
Glipr1	0,005759	-1,07391
Lig1	7,56E-05	-0,92399
S100a4	0,00527	-0,84746
Ccr2	0,032259	-0,70196
Satb1	0,002522	-0,67599
Cd14	0,02315	-0,64989
Gpnmb	0,012874	-0,6213
Clec4e	0,002522	-0,61655
Ccr9	0,000102	-0,60723
Ms4a7	0,016363	-0,60247
Clec4d	0,045826	-0,53573
Tyrobp	0,012874	-0,44672
Il1rn	0,022291	-0,42433
Dusp10	0,042782	-0,40252
Ptprcap	0,000154	-0,38418
Cd34	0,027888	-0,36934
Cp	0,042782	-0,35777
AA467197	0,003045	-0,353
Jph2	0,026877	-0,33817
Areg	0,002404	-0,33282
Pola1	0,047415	-0,31824
Brca1	0,002522	-0,31297
Rrm1	0,002773	-0,30989
Cd247	0,000194	-0,29907
Smc2	0,011867	-0,29682
Dbf4	0,000446	-0,2788
Prrx1	0,010931	-0,27569
Lck	0,004801	-0,27034
Rmnd5a	0,041325	-0,26977
Ctla2a	0,000145	-0,26973
Adgre4	2,33E-06	-0,26094
Gem	0,002644	-0,26062
Mcm7	0,001888	-0,26053
Cenph	0,009222	-0,25088
Hells	0,000243	-0,25018

Table 6.

DEGs in normocholesterolemic Nrf2^Cdh5tKO proinflammatory macrophages (normocholesterolemic Nrf2^Cdh5tKO vs. normocholesterolemic Nrf2^flox/flox).

Upregulated
Gene	p-value	avg_log2FC
Il1b	0,027338	0,912593
Pou2f2	0,000477	0,515043
Errfi1	0,012599	0,397658
Entpd1	0,00028	0,394454
Cd300a	0,011064	0,360637
Tmsb4x	6,67E-07	0,349386
Cd209f	7,32E-10	0,347549
Bmp2	0,026157	0,341992
Lst1	0,007295	0,338405
Basp1	0,011608	0,331997
Tyrobp	3,07E-05	0,329781
Cotl1	0,036986	0,326228
Pde4b	0,016117	0,308501
Zmynd15	5,06E-06	0,289084
Runx1	0,012652	0,282018
Mctp1	0,005341	0,279978
Tgfbi	0,02097	0,277903
Myl6	0,036142	0,272899
Ccr1	1,50E-06	0,272622
Ncf4	0,002907	0,265326
Actc1	1,06E-07	0,264654
Ranbp1	0,042688	0,256888
Downregulated
Gene	p-value	avg_log2FC
Retnla	0,048006	-2,55668
Gdf15	0,001614	-0,78194
S100a4	0,035294	-0,74965
Lyz1	1,64E-05	-0,68443
Mgl2	2,43E-05	-0,58338
Cxcl9	4,65E-42	-0,54531
Vcam1	0,000307	-0,53983
Ccl3	0,044231	-0,52939
Pepd	2,60E-05	-0,46906
H2-Ab1	0,007047	-0,4344
H2-Aa	0,04455	-0,36261
H2-Eb1	0,027323	-0,32848
Cd74	0,00508	-0,31822
Clec10a	0,006858	-0,31712
Dqx1	1,71E-39	-0,30632
Plbd1	0,021344	-0,28927
Lcp2	0,00315	-0,25308
Pcna	0,000764	-0,25137

Moreover, in atherogenic conditions, several DEGs in proinflammatory macrophages of Nrf2^Cdh5tKO vs. Nrf2^flox/flox mice ( Table 5 ) were identified as hub genes associated with DNA replication and maintenance (downregulation of Brca1, Smc2, Cenph, Lig1, Mcm7, Rrm1, Dbf4, Pola1, Hells and upregulation of Atr, Apex1, Figures 6D, E ). As the Atr and Apex1 genes are the main sensors of oxidative stress-induced DNA damage, we checked for any other genes associated with increased oxidative stress in Nrf2^Cdh5tKO proinflammatory macrophages. Among these, we observed upregulation of Lyn, which is involved in response to DNA damage/redox imbalance (49); Fbxo3, which is a part of SCF complex and may be involved in DNA damage response (50); Hmgn5, which may regulate histone modification, DNA replication, repair, and gene transcription through binding to chromatin regulators (51). Among the downregulated DEGs, we found Dbf4 gene which decreases S-phase checkpoint signaling and maintains DNA replication and cell cycle progression induced by DNA damage (52), DNA damage-induced Areg (53), Satb1 that is involved in DNA repair (54), and Gem which regulates chromatin remodeling and DNA replication (51).

Some DEGs were associated with the C-type lectin receptor (CLR) signaling pathway. CLRs are widely expressed by myeloid cells. The expression of macrophage-specific members of CLR family were decreased in Nrf2^Cdh5tKO proinflammatory macrophages. Among them, we identified Clec4e (macrophage inducible C-type lectin; Mincle) and Clec4d (macrophage C-type lectin; MCL). The first one is activated by dead cells/necrosis and both of them are able to sense damage-associated molecular patterns (55).

Within the Src family kinases, Lck was downregulated, the level of Ptprcap, that stabilizes the association of CD45 with Lck (56), was decreased, similarly to Tyrobp (Dap12) that may be phosphorylated by Src kinases. On the other hand, the other Src family proteins, Cd14 and Lyn were upregulated.

Monocytes

Monocytes possess the ability to infiltrate the tissues and mature into macrophages. In our scRNAseq analysis they were characterized by high expression of Plac8, Chil3, Msrb1, Clec4e, Slpi ( Figures 3E , 7A ) and were one of the most abundant mononuclear phagocyte subset. The monocytes of Nrf2^flox/flox mice were strongly affected by the atherogenic conditions what was reflected by the number of DEGs – 146 genes (57 down- and 89 upregulated vs. normocholesterolemic Nrf2^flox/flox mice; Table 7 ). The KEGG pathway analysis ( Figure 7B ) revealed that many DEGs were related to inflammatory responses: viral protein interaction with cytokine and cytokine receptor (Ccr1, Il10, Il6, Ccl7, Ccl5, Ccl2, Il2rg, Cxcl3, Pf4), cytokine-cytokine receptor interaction (Ccr1, Il10, Il1r2, Inhba, Cxcl3, Il2rg, Cxcl16, Il6, Ccl7, Ccl5, Ccl2, Il7r, Pf4), chemokine signaling pathway (Ccr1, Grk3, Gngt2, Ccl7, Ccl5, Ccl2, Cxcl3, Cxcl16, Pf4), IL-17 signaling pathway (Il6, Ccl7, Ccl2, Cxcl3, S100a9, S100a8), complement and coagulation cascades (C1qb, C4b, C1qa, Serpinb2, C3ar1, C1qc), genes associated with TNF signaling pathway (Il6, Ccl5, Ccl2, Cxcl3, Creb5).

Figure 7.

The analysis of transcriptomic changes in monocytes. (A) Expression of marker genes identified for monocytes projected onto the UMAP plot. (B) KEGG pathway enrichment analysis of upregulated and downregulated genes for atherosclerotic Nrf2^flox/flox vs. control Nrf2^flox/flox and (C) atherosclerotic Nrf2^Cdh5tKO vs. atherosclerotic Nrf2^flox/flox monocytes. The x-axis represents the strength (Log10(observed/expected)) of enriched pathways, whereas the color denotes –log10(pValue) and dot size corresponds to gene count in each pathway.

Table 7.

DEGs in monocytes (atherosclerotic Nrf2^flox/flox vs. normocholesterolemic Nrf2^flox/flox).

Upregulated
Gene	p-value	avg_log2FC
S100a9	4,22E-05	2,244782
Basp1	2,38E-10	1,60785
Retnlg	9,22E-21	1,577205
S100a8	0,004065	1,510919
Ccl2	0,000946	1,498106
Ccl7	2,41E-09	1,288358
Spp1	0,036512	1,215382
Il1r2	4,72E-11	1,140975
Inhba	1,46E-25	1,117781
Vegfa	0,00013	1,082856
Cxcl3	3,60E-20	1,080253
Prg4	2,20E-25	1,078775
Rgs1	0,000105	1,061712
C5ar1	0,015234	1,044396
Cstb	0,035832	1,023867
Lilr4b	0,000666	0,877581
Sdc4	3,79E-21	0,873767
Pf4	1,58E-07	0,86228
Lgmn	0,046372	0,836777
Mxd1	1,34E-07	0,829167
Id2	0,000335	0,779114
Dab2	1,11E-17	0,728129
Il1b	0,018683	0,709909
Clec4d	0,044953	0,702688
Ptafr	0,006671	0,679941
C3ar1	7,69E-20	0,654895
Entpd1	0,001438	0,645294
C1qb	6,42E-10	0,625359
Marcksl1	0,042589	0,599681
Ecm1	2,83E-05	0,565733
Runx3	0,005149	0,558581
Ms4a7	1,75E-10	0,555418
Eps8	6,21E-05	0,555199
Nlrp3	0,040919	0,552225
Fabp4	0,047085	0,532537
Rab11fip1	0,00121	0,516488
Rab20	0,000329	0,45967
Capg	0,010068	0,4592
Plin2	0,030932	0,443532
Cxcl16	0,000537	0,442237
Slc7a11	0,048725	0,4358
Ptgs2	0,047623	0,435474
Olr1	0,002084	0,432372
Clec4n	2,21E-06	0,428953
Dok2	3,26E-13	0,425217
Fcgr3	0,021193	0,422065
C1qc	2,77E-16	0,41703
Adam8	0,02293	0,404946
Cd93	0,024288	0,398992
Furin	0,000996	0,395016
Il6	4,87E-06	0,391441
Gpr65	0,005499	0,38869
C1qa	3,74E-08	0,386056
Pdpn	8,19E-27	0,382998
Serpinb2	0,00494	0,367
Acod1	4,72E-16	0,366281
Odc1	0,006713	0,363343
Rab7b	7,32E-13	0,352893
Hspa1a	0,028078	0,352667
Ankrd33b	6,81E-17	0,350538
Trf	0,000773	0,349309
Creb5	1,56E-18	0,343503
Fam46c	0,001612	0,34167
Il7r	1,80E-09	0,340033
Ass1	5,99E-06	0,337812
Efhd2	0,036588	0,33655
Clec5a	6,12E-16	0,336504
Rgcc	0,013707	0,327872
Ccr1	4,96E-05	0,324188
Maf	0,000167	0,321496
Ran	0,015117	0,317006
C4b	6,74E-16	0,314411
Tmem37	1,78E-21	0,314265
Nrp2	2,18E-18	0,311033
Plekho1	0,007307	0,307844
Arg2	2,74E-08	0,303814
Glrx	3,35E-06	0,302841
Bcl2a1b	0,012414	0,299724
Fcgr1	0,000566	0,299692
Lacc1	3,32E-09	0,275824
Cenpa	8,29E-14	0,274966
Cysltr1	2,91E-14	0,273945
Tsc22d1	4,82E-07	0,26936
Ifit3	2,14E-07	0,267745
Bcl2a1d	0,027019	0,26495
Mrc1	1,16E-19	0,264818
Stab1	7,27E-19	0,256998
Cmah	6,67E-09	0,254944
Cd86	0,007618	0,25333
Downregulated
Gene	p-value	avg_log2FC
Hes1	5,73E-08	-1,18421
Gngt2	0,00481	-1,02475
Pou2f2	0,000119	-0,73768
Itgal	0,010299	-0,71632
Fam107b	0,000992	-0,62345
Plac8	0,009667	-0,59459
H2-Eb1	0,010857	-0,58955
Adgre5	0,003784	-0,5846
Spn	0,02998	-0,58211
Samhd1	0,001794	-0,58062
Il2rg	0,002521	-0,57655
Napsa	0,002012	-0,57128
Gm9733	0,00556	-0,56823
Samsn1	0,011576	-0,55471
Ly6i	1,49E-08	-0,53254
Ikzf1	0,000899	-0,50793
Ceacam1	0,00772	-0,50464
Nabp1	0,000779	-0,49656
Gsr	0,00512	-0,49127
Ptprc	0,02186	-0,47549
Coro1a	0,000812	-0,46232
C3	0,01185	-0,43997
Nfam1	0,006526	-0,43844
Stk10	0,013361	-0,43065
Ppp1r12a	0,014837	-0,43023
Ets1	6,73E-06	-0,42168
Grk3	0,000688	-0,40919
Bin2	0,006256	-0,40871
Ifitm6	0,036091	-0,40527
Ear2	0,001697	-0,3956
Fgr	0,036856	-0,38661
Adgre4	0,001166	-0,3842
Cytip	0,009957	-0,38144
Tppp3	0,004165	-0,37381
Il10ra	0,017348	-0,37184
Spi1	0,039732	-0,3638
Saa3	1,04E-17	-0,3629
H2-Aa	0,006533	-0,35594
Lst1	0,032867	-0,35426
Ptpn6	0,038215	-0,35237
Gm19951	3,45E-08	-0,35106
Cbfa2t3	0,000225	-0,34962
Ccl5	1,84E-13	-0,32136
H2-Ab1	0,025519	-0,32122
Limd2	0,025979	-0,31346
Il10	1,04E-17	-0,30918
Msrb1	0,032869	-0,30777
Itgb7	0,0182	-0,30705
Adssl1	0,033508	-0,30264
Mfap5	0,013216	-0,29002
Icam2	1,22E-05	-0,28042
Trbc2	2,74E-11	-0,27078
Igfbp5	2,61E-06	-0,2638
Tyrobp	0,001737	-0,26266
Ptpn22	1,85E-05	-0,26183
Cd300e	4,41E-09	-0,25778
Slpi	0,002152	-0,25694

In monocytes of atherosclerotic Nrf2^Cdh5tKO mice, we found 19 downregulated and 24 upregulated genes when compared to atherosclerotic Nrf2^flox/flox ( Table 8 ). The KEGG pathway analysis ( Figure 7C ) uncovered 5 enriched pathways. Among these, we found cytokine-cytokine receptor interaction (Ccr1, Il6, Il1rn, Ccl7, Gdf15, Il1r2, Ccl3, Cxcl13), viral protein interaction with cytokine and cytokine receptor (Ccr1, Il6, Ccl7, Ccl3, Cxcl13), chemokine signaling pathway (Ccr1, Ccl7, Ccl3, Cxcl13), IL-17 signaling pathway (Il6, Ccl7, S100a9), complement and coagulation cascades (C4b, Serpinb2, Vsig4). The last pathway was the only one to be enriched when normocholesterolemic Nrf2^Cdh5tKO and Nrf2^flox/flox mice were compared ( Table 9 ).

Table 8.

DEGs in atherosclerotic Nrf2^Cdh5tKO monocytes (atherosclerotic Nrf2^Cdh5tKO vs. atherosclerotic Nrf2^flox/flox).

Upregulated
Gene	p-value	avg_log2FC
Mgl2	0,000151	1,111916
Cxcl13	2,33E-09	1,084349
Slpi	0,022409	1,072164
Mmp12	0,015239	0,59813
Ifitm6	0,044762	0,558191
AA467197	0,000115	0,49493
Hbb-bt	0,000493	0,443647
Gm12840	0,011678	0,438294
Icam2	0,001169	0,425176
Ccl7	0,003695	0,411656
Il1rn	0,048982	0,400177
Ednrb	2,25E-08	0,392623
Timp3	0,00014	0,368488
Saa3	0,000654	0,363723
Ccr1	0,039922	0,33807
Vsig4	6,93E-06	0,33687
Gm19951	2,56E-07	0,328188
Padi4	0,001794	0,311341
Fmo2	0,000111	0,299015
Lum	4,69E-07	0,291819
Socs2	0,000256	0,269916
C4b	7,76E-05	0,261613
Serpinb2	0,003269	0,251719
Trem3	0,00595	0,250948
Downregulated
Gene	p-value	avg_log2FC
S100a9	9,80E-07	-2,71359
Il1r2	0,000157	-1,04943
Gdf15	0,00317	-1,04697
Ccl3	0,002041	-0,85663
Il6	0,00317	-0,55114
Lair1	0,019751	-0,46441
Sdc4	0,000229	-0,43374
Olr1	0,000427	-0,42507
2010015M23Rik	7,77E-08	-0,36718
Fbxo32	0,00668	-0,35142
Maf	0,003927	-0,33295
Gm14548	0,00263	-0,32188
Dab2	0,003695	-0,26731
H2-DMa	0,031563	-0,26226
Mrc1	2,10E-06	-0,26101
Cenpa	0,018293	-0,25761
Gm26870	7,77E-08	-0,25656
P2ry6	0,02798	-0,25297
Basp1	0,022976	-0,2505

Table 9.

DEGs in normocholesterolemic Nrf2^Cdh5tKO monocytes (normocholesterolemic Nrf2^Cdh5tKO vs. normocholesterolemic Nrf2^flox/flox).

Upregulated
Gene	p-value	avg_log2FC
Thbs1	0,000882	0,698712
F13a1	0,008909	0,637967
Lbr	0,01143	0,459449
Clec4d	0,047093	0,429932
Slfn2	0,032847	0,421368
Evi2a	0,018258	0,41203
Gm26778	0,030337	0,407959
Mxd1	0,007663	0,406864
Gadd45a	0,001254	0,394997
Naaa	0,015887	0,393913
Sell	5,54E-05	0,372804
Dab2	0,018307	0,353464
St8sia4	0,000414	0,347379
Ifi211	0,004227	0,344962
Mx1	6,52E-06	0,327849
Nfkbie	0,014707	0,315773
Ifit3	0,048771	0,313544
Clec4a2	0,038668	0,30933
Eps8	0,006948	0,308254
Ciita	1,21E-06	0,303733
C5ar1	2,86E-05	0,297725
Rnasel	2,27E-05	0,294714
Pcdh7	1,70E-13	0,283886
Lif	1,66E-16	0,263507
Tnip3	0,034321	0,263488
Sla	0,04905	0,262487
Downregulated
Gene	p-value	avg_log2FC
Hes1	0,004695	-1,2028
C1qa	0,000305	-0,82092
S100a8	0,001518	-0,80354
C1qc	1,44E-05	-0,78584
Prg4	6,03E-06	-0,64874
S100a9	0,003547	-0,61631
Serpinb2	0,033818	-0,60127
Rgs1	9,27E-06	-0,59562
Icam2	5,10E-09	-0,5471
Cd36	0,041232	-0,54095
Hmgn5	2,92E-06	-0,38335
Ly6i	0,025161	-0,38235
Il10	1,34E-10	-0,36946
Unc119	0,049872	-0,35086
Hilpda	7,20E-07	-0,3472
Ccl5	5,46E-19	-0,33329
Ranbp1	0,029044	-0,33061
Tuba4a	0,023782	-0,30078
Acp5	0,023156	-0,29388
Saa3	4,83E-13	-0,29215
Trbc2	1,87E-10	-0,28844
Gm20186	1,42E-06	-0,26269

The transcriptome of monocytes from atherosclerotic Nrf2^Cdh5tKO mice ( Table 8 ) was strongly changed in terms of the expression of genes encoding chemokines, cytokines, and their receptors. Besides the genes associated with cytokine-cytokine receptor interaction (downregulated Il6, Gdf15, Il1r2, Ccl3 and upregulated Ccr1, Il1rn, Ccl7, Cxcl13), in atherosclerotic Nrf2^Cdh5tKO monocytes we also identified upregulation of Ednrb, which is a receptor for endothelin and angiotensin (57). Interestingly, in monocyte DEGs between atherosclerotic Nrf2^Cdh5tKO and Nrf2^flox/flox mice, we also found several protease inhibitors, such as Timp3, Serpinb2, and Slpi, that may change the activity of peptidases and contribute to vascular remodeling.

Discussion

Macrophages drive atherosclerosis at all stages of plaque development. It was previously shown that local tissue environment shapes the transcriptome and identity of macrophages and various macrophage types may differentially respond to the same stimuli (58). One of the stimuli acting in atherosclerotic plaques is oxidative stress, which has been implicated in the disease progression from the early fatty streak lesions to advanced atherosclerotic plaques (22). The accumulation of reactive oxygen species within the cells can affect different signaling pathways regulating cell cycle, migration and survival leading to cellular dysfunction and even cell death (59).

In this study, we analyzed the transcriptomic changes caused by atherogenic conditions in different populations of murine aortic mononuclear phagocytes deficient in Nrf2 transcriptional activity. In physiological conditions this redox-activated transcription factor provides protection via activation of expression of a number of genes mediating antioxidant response (22). However, Nrf2 effects in atherosclerosis seem to be more complex and currently not completely understood. Here, we analyzed the data obtained from mice with transcriptionally inactive Nrf2 in Cdh5-expressing cells and their progeny. The prevailing scientific consensus is that hematopoiesis, the process of blood cell formation, primarily occurs through two distinct mechanisms during development: yolk-sac hematopoiesis and hemogenic endothelium-derived hematopoiesis. The exact contribution of these sources to adult hematopoiesis is still a subject of active research and ongoing debate (40, 60). Nevertheless, according to the literature data even 50-96% of CD45+ adult bone marrow–derived cells may be affected by Cdh5-dependent Cre recombinase active during embryonic development (34, 35). Therefore, the Cre-driven mutation is usually present not only in endothelial cells, but also in a number of leukocytes. In our mouse model, macrophages of two different origins – BMDM and Kupffer cells - were characterized by significantly decreased level of Nrf2 exon 5 responsible for the transcriptional Nrf2 activity.

Atherosclerosis was induced via AAV8-mediated Pcsk9 overexpression and a high-fat diet (36). This approach successfully elevated the cholesterol level in the LDL/VLDL plasma lipoproteins, as well as the level of triglycerides in the mouse plasma in the range previously described for similar proatherogenic approaches (61–63). In the fragments of the aorta (aortic arch and branches) used for scRNAseq procedure, we observed lipids accumulation within the first layers of intima and media corresponding to fatty streaks, which represent an early stage of atherosclerotic plaque formation. In such early stage lesions, macrophage infiltration, initiation of inflammation and chemoattraction of other cell types is usually observed.

Nowadays, mainly with the use of scRNA-seq technology, several macrophage subtypes were identified in mouse and human atherosclerotic plaques, each with distinct functions and properties (7–9, 37, 38). It is well-recognized that monocyte-derived inflammatory macrophages are not the only players in atherosclerotic plaque development. Thus, we performed subclustering analysis to visualize the macrophage heterogeneity and macrophage subtype-specific effects of decreased Nrf2 transcriptional activity. That strategy uncovered 9 subpopulations of mononuclear phagocytes, in majority comprising macrophages of different origin, associated with different layers of aortic wall or peritoneal cavity. Each subpopulation was characterized by the typical markers described in previous reports (8, 9, 37, 38).

We selected three most abundant subtypes, namely Lyve1+ resident macrophages, proinflammatory macrophages and monocytes, for in-depth transcriptomic analysis in proatherogenic conditions. Tissue-resident macrophages maintain tissue homeostasis, remove pathogens and abnormal (apoptotic or senescent) cells, participate in tissue repair and regeneration (60). In the aortic wall, the tissue-resident macrophages are present mainly in adventitia in steady-state conditions, whereas in inflammation they can migrate towards the media and intima layer (64). The proinflammatory macrophages release several proinflammatory cytokines and ingest modified lipids what results in foam cells formation (64). Monocytes are blood-circulating cells that play an important role in atherosclerosis as protagonists of plaque development. When the lining of the artery is damaged, monocytes are intensively recruited to the site of injury.

The comparison between atherosclerotic Nrf2^flox/flox and control normocholesterolemic Nrf2^flox/flox mice revealed that each analyzed mononuclear phagocytes subset contained DEGs encoding cytokines, chemokines and their receptors involved in immune responses. This confirms that inflammation is an important component in the pathogenesis of atherosclerosis already in the early stage lesions (4, 5). In our analysis, the highest number of DEGs induced by atherogenic conditions was identified in the monocyte subset. Interestingly, among monocyte DEGs between atherosclerotic Nrf2^flox/flox and control normocholesterolemic Nrf2^flox/flox mice we found upregulation of NOD-like receptor pyrin domain-containing protein 3 (Nlrp3) and interleukin 1β (Il1b), two components of Nlrp3 inflammasome activation, which normally leads to Il-1β release. Inhibition of this cytokine production is currently of particular interest for the secondary prevention of atherosclerotic events (65).

Next, we investigated the effect of decreased Nrf2 transcriptional activity on the transcriptomic changes in monocytes and selected macrophage subsets in atherosclerotic mice. Previous in vitro studies indicated that Nrf2 activity may promote the anti-inflammatory phenotype of macrophages (66, 67), but the effect of Nrf2 on tissue-resident macrophages is still poorly investigated. It was shown that in alveolar macrophages Nrf2 increases phagocytic ability (68), macrophage-driven efferocytosis and apoptotic neutrophil clearance (69), and may protect from ferroptosis in sepsis-induced acute lung injury (20). Other studies performed on dermal resident macrophages showed the involvement of Nrf2 in IL-23–IL-17A–TRPV1 axis and pain perception (70) or wound repair (71). In our study, the RIG-I like pathway was the top altered pathway in atherosclerotic Nrf2^Cdh5tKO Lyve1+ resident macrophages. RIG-I may activate IRFs and NF-κB, and then induce the expression of antiviral genes, such as type I interferons and pro-inflammatory cytokines (43). The latest research shows that RIG-I-like pathway is not only triggered by viral nucleic acids, but also by damage-associated molecular patterns (72). It was suggested that Nrf2 can interact with RIG-I-like pathway and modulate its function. Specifically, Nrf2 has been shown to regulate the expression of genes involved in the antiviral response, including interferon-stimulated genes (ISGs) that are downstream effectors of RIG-I-like signaling (41). Interestingly, in Lyve1+ resident macrophages of atherosclerotic Nrf2^Cdh5tKO vs. atherosclerotic Nrf2^flox/flox, we also observed upregulation of Bcl2a1b, an IFN-dependent master switch for the function of Cd11b+ cells exerting anti-apoptotic and pro-survival effects (42). In addition, in Lyve1+ resident macrophages from atherosclerotic Nrf2^Cdh5tKO mice we observed downregulation of ferroportin gene Slc40a1. This is in accordance with previous reports describing involvement of Nrf2 in regulation of this gene expression, as well as in macrophage resistance towards ferroptosis (16, 17). In this context, decreased Nrf2 activity might promote ferroptotic death of Lyve1+ resident macrophages at early stages of plaque development and in this way contribute to disease progression. This, however, would require further investigation.

In terms of bone marrow-derived macrophages, the role of Nrf2 is better recognized. Previous in vivo research demonstrated strong evidence that Nrf2 works as a negative upstream regulator of macrophage proinflammatory phenotype, while the majority of in vitro data indicates that Nrf2 deficiency promotes macrophage polarization towards proinflammatory M1 phenotype (25, 26, 31, 66). Macrophages are generally considered as non-dividing cells with limited capacity for self-renewal. However, our current data demonstrate that in Nrf2^Cdh5tKO proinflammatory macrophages the pathway associated with DNA replication was significantly enriched. The precise role of DNA replication in macrophage proliferation and activation (especially in atherosclerosis) is still not fully understood. There are indications that macrophages may undergo DNA replication and proliferation upon certain, usually considered as harmful or danger, stimuli (73, 74). The upregulation of Atr and Apex1 in Nrf2^Cdh5tKO proinflammatory macrophages may suggest increased oxidative stress and DNA damage that often accompany Nrf2 deficiency (74). Some data indicate that high level of DNA synthesis aiming at DNA repair in macrophages is a response to oxidative DNA damage and is highly associated with apoptotic cell death (11). In chronic inflammation, DNA synthesis may raise generation of polyploid macrophages (73). In addition, we observed that Nrf2 deficiency in proinflammatory macrophages was associated with upregulation of genes from the CLR family, including Mincle and MCL, involved in lipids recognition (including cholesterol) and their endocytosis (75). Our transcriptomic analysis of proinflammatory macrophages from atherosclerotic Nrf2^Cdh5tKO mice also revealed decreased expression of genes connected with autophagy (Mcm7, S100a4 and Gpnmb; Table 5 ) and genes described as ferroptosis-related (Hells and Cp; Table 5 ). Observations of the last years indicate that the process of autophagy and ferroptosis are intertwined (76). Autophagy promotes intracellular lipid hydrolysis and cholesterol efflux thereby inhibiting development of macrophage-derived foam cells. Inhibition of autophagy in macrophages leads to foam cells formation, macrophage death, pro-inflammatory factor release and contributes to atherogenesis (77). Autophagy plays a key role in inhibiting ferroptosis of macrophages via maintaining cellular iron homeostasis and cellular reactive oxygen species generation, and alleviates atherosclerosis (78, 79). Overall, our data indicate that Nrf2 deficiency in atherosclerotic proinflammatory macrophages affects the expression of genes involved in DNA replication and repair mechanisms, as well as in the recognition of lipids and cell death pathways, what may impact these cells activation, phagocytic ability, and survival.

In atherosclerotic conditions, Nrf2-deficiency caused transcriptomic changes also in the monocyte subset. Some proinflammatory genes were downregulated and anti-inflammatory were upregulated, whereas on the other hand, the expression of some chemoattractants was increased when compared to monocytes from the aortas of atherosclerotic Nrf2^flox/flox mice. The observed changes suggest modulation of the immune phenotype of monocytes, but not necessarily towards the proinflammatory one. In addition, in atherosclerotic Nrf2^Cdh5tKO vs. atherosclerotic Nrf2^flox/flox monocytes, increased level of Mmp12 and protease inhibitors may suggest their involvement in ECM remodeling and monocyte infiltration.

In conclusion, this exploratory study identified several subtype-specific differences in monocytes and two selected, the most numerous in the aortic wall, types of macrophages from atherosclerotic Nrf2^Cdh5tKO mice. The introduced proatherogenic factors caused significant changes particularly in the expression of genes encoding inflammatory cytokines and chemokines. Comparison of atherosclerotic Nrf2^flox/flox and Nrf2^Cdh5tKO mice enabled identification of Nrf2-dependent macrophage subtype-specific transcriptomic changes associated with inflammation, iron homeostasis, DNA repair, cell injury and death pathways. Our data demonstrate a possible link between ferroptosis and inflammatory microenvironment appearing at a very early stage of atherogenesis. A limitation of this study can be our mouse model based on Cdh5-dependent Cre activity. First, because Nrf2-deficient endothelial cells may affect the microenvironment and their interaction with other cell types (e.g. monocytes) may differ. Second, this mouse model is characterized by a significant downregulation, but not a complete knockout of Nrf2 in mononuclear phagocytes. Although, we think that such model better reflects the observed in humans aging-associated decline of Nrf2 activity, in the future it may be worth to reproduce some of the findings in mice with the lysozyme 2 (LysM) Cre-driven recombination.

Data availability statement

The data presented in the study are deposited in the Gene Expression Omnibus (GEO) repository, accession number GSE245820.

Ethics statement

The animal study was approved by the 2nd Institutional Animal Care and Use Committee (IACUC) in Kraków, Poland. The study was conducted in accordance with the local legislation and institutional requirements.

Author contributions

KSa and AJ-K designed the study. KSa, MS, IK, and KSz performed the experiments. MS and KSa performed the bioinformatic analyses and data visualization. KSa and AJ-K were involved in data interpretation and wrote the manuscript. JCS, PB, and JD provided the intellectual contribution and revised the manuscript. AJ-K secured funding and coordinated the project. All authors provided constructive feedback, helped shape the manuscript and approved the submitted version.

Funding Statement

The author(s) declare financial support was received for the research, authorship, and/or publication of this article. This work was supported by grant ERA-CVD/MEND-AGE/6/2019 from the National Centre for Research and Development and project No. 2021/43/B/NZ5/03336 financed from the funds of the National Science Center, Poland.

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.