Abstract

   Vascular restenosis is a serious clinical issue initiated and
   aggravated by macrophage inflammation, with no effective treatments
   available, in cardiovascular and autoimmune diseases. However, the
   untapped mechanisms and new targets that can regulate macrophage
   polarization and vascular restenosis remain elusive. The research
   identifies interferon regulatory factor 4 (IRF4) expression as crucial
   in macrophage polarization during arterial restenosis. Myeloid‐specific
   Irf4 deficiency and overexpression experiments showed that IRF4
   promoted M2 macrophage polarization, inhibited M1 macrophage
   transitions, and disrupted the interaction between macrophages and
   vascular smooth muscle cells to reduce neointimal hyperplasia by
   directly upregulating krüppel like factor 4 (KLF4) expression.
   Artesunate, an FDA‐approved drug, is screened as a potent activator of
   IRF4 expression in M2 polarization, and its treatment attenuated
   arterial restenosis in rodents and non‐human primates. The findings
   reveal a significant protective role of IRF4 in the development of
   neointimal hyperplasia by regulating macrophage polarization, and
   artesunate may be proposed as a novel therapy for vascular restenosis.

   Keywords: artesunate, IRF4, macrophages, neointimal hyperplasia,
   restenosis
     __________________________________________________________________

   This study shows that macrophage interferon regulatory factor 4 (IRF4)
   improved arterial injury‐induced neointimal hyperplasia by promoting of
   M2 polarization via up‐regulating krüppel‐like factor 4 (KLF4) in
   rodent and nonhuman primate models. Notably, artesunate is identified
   as a potent inducer of IRF4 in macrophages, and proposed as a promising
   novel therapy for vascular restenosis.

   graphic file with name ADVS-12-2408992-g001.jpg

1. Introduction

   Cardiovascular disease (CVD) is the leading cause of death and
   disability worldwide.^[ [54]^1 ^] Atherosclerosis, the primary
   pathological basis of CVD, is a chronic inflammatory disease that can
   be initiated and accelerated by local and systemic inflammation.^[
   [55]^2 ^] For instance, autoimmune rheumatic diseases had an increased
   risk and mortality for cardiovascular events.^[ [56]^3 ^] Vascular
   interventions, including intravascular stenting, endarterectomy, and
   angioplasty, are the mainstays for treating atherosclerotic
   occlusion.^[ [57]^4 ^] However, postoperative arterial restenosis and
   the need for repeated revascularization still occur in many patients.^[
   [58]^5 ^] While numerous drugs and drug‐eluting stents have been
   developed to mitigate this problem, their long‐term efficacy remains
   unsatisfactory, and the precise mechanisms and regulators contributing
   to restenosis are unclear. This ongoing challenge highlights the need
   for further investigation into the underlying mechanisms of restenosis
   to improve clinical outcomes.

   Despite the extensive focus on vascular smooth muscle cells (VSMCs),
   mounting evidences indicate that inflammation response is a critical
   driver of restenosis occurrence and progression.^[ [59]^6 ^] Recent
   clinical trials have suggested that canakinumab^[ [60]^7 ^] and
   colchicine,^[ [61]^8 ^] with its unique anti‐inflammatory efficacy, are
   effective in reducing the risks of cardiovascular events and restenosis
   in patients with post‐intravascular stenting.^[ [62]^9 , [63]^10 ^]
   Macrophages, as primary inflammatory cells, significantly impact the
   inflammation in CVD and regulate aggressive neointimal proliferation,
   thereby contributing to restenosis by expressing specific phenotypes.^[
   [64]^11 , [65]^12 ^] During the formation of vascular restenosis,
   macrophages typically polarize to the classically activated M1
   phenotype, secreting proinflammatory cytokines that promote VSMC
   proliferation and migration.^[ [66]^13 ^] Conversely, alternatively
   activated M2 macrophages exhibit anti‐inflammatory properties, leading
   to a significant reduction in vascular restenosis.^[ [67]^14 ^]
   Consequently, it is crucial to elucidate the regulatory mechanisms of
   macrophage polarization to vascular injury, and further regulating
   macrophage polarization maybe a promising strategy to alleviate
   vascular inflammation and restenosis.

   Within the hierarchical molecular network controlling macrophage
   polarization, several transcription factors have been identified to
   significantly impact the macrophage phenotypes.^[ [68]^15 , [69]^16 ^]
   Interferon regulatory factors (IRFs) are hub transcription factors that
   modulate immune responses in fighting viral infection.^[ [70]^17 ,
   [71]^18 ^] In addition to the N‐terminal DNA‐binding domain that binds
   a consensus recognition sequence, most IRFs contain a protein
   interaction domain that allows for interaction with other IRF family
   members or transcription factors for diverse cellular functions.^[
   [72]^17 , [73]^18 ^] For instance, IRF4 is constitutively expressed in
   macrophages and histone demethylation of the IRF4 promoter contributes
   to the M2 macrophage polarization under helminth parasite conditions.^[
   [74]^19 , [75]^20 ^] However, given that the phenotype and function of
   macrophages differ in specific diseases, the precise role of IRF4 in
   governing macrophage phenotypes and the subsequent inflammatory
   response during vascular restenosis has yet to be clarified. In the
   present study, we explored the regulatory effect of IRF4 on macrophage
   polarization during vascular restenosis. Importantly, through
   high‐throughput screening analysis combined with in vitro and in vivo
   experiments, we investigated the protective effects of selected drugs
   against restenosis in both mouse and nonhuman primate models.

2. Results

2.1. Identification of IRF4 as a Regulator of Neointimal Hyperplasia and
Macrophage Polarization

   To identify the key regulators potentially associated with post‐injury
   remodeling in the vasculature, a composite analysis was performed to
   explore the critical molecular changes by mining the public Gene
   Expression Omnibus (GEO) datasets obtained from mouse aortic tissues in
   response to vascular injuries (Figure [76] 1A). By Estimating the
   Proportion of Immune and Cancer cells (EPIC) algorithm, we performed
   the analysis of three independent public GEO datasets and found that
   macrophages were the only common cell type in injured tissues which are
   significantly higher in proportions than in control tissues
   (Figure [77]1B–D). Interestingly, enrichment analysis also identified
   macrophage‐related biological processes as the main events across these
   GEO datasets (Figure [78]1E). Among them, macrophage polarization is
   closely linked to macrophage activation, migration, and cytokine
   production, and is involved in vascular restenosis.^[ [79]^13 , [80]^21
   ^] To further delineate the key molecules regulating macrophage
   polarization, we analyzed 4 separate public datasets of polarized
   macrophages and identified a total of 23 shared differentially
   expressed genes (DEGs) in M1 and M2 macrophages, as illustrated in the
   Venn diagram (Figure [81]1F–H). These 23 DEGs were then verified with
   our RNA‐seq data of primary mouse bone marrow‐derived macrophages
   (BMDMs), and as compared to M0 (Vehicle) cells, IRF4 was the most
   significantly downregulated gene in lipopolysaccharides (LPS)‐induced
   M1 cells, but its expression was upregulated in interleukin‐4
   (IL‐4)‐induced M2 macrophages (Figure [82]1I). Furthermore, we
   performed iRegulon analysis of the common DEGs to predict the key
   transcription factors (TFs) involved in vascular injury
   (Figure [83]1J). As illustrated in the Venn diagram, 222 DEGs were
   shared among vascular injury‐related GEO datasets (Figure [84]1K).
   Additionally, we identified IRF4 as the top‐ranked TF
   (Figure [85]1L,M), indicating that IRF4 is a potential key regulator
   involved in macrophage polarization in response to vascular injury.

Figure 1.

   Figure 1
   [86]Open in a new tab

   IRF4 is a regulator of neointimal hyperplasia and macrophage
   polarization. A) Flowchart for bioinformatics analysis of publicly
   available data from GEO databases of vascular injury in mice. B) Venn
   diagram showing the number of cells with significantly different
   abundances from the [87]GSE119549, [88]GSE40637, and [89]GSE70410
   datasets by the EPIC algorithm. C) Difference in predicted cell
   proportions between the vascular injury and control groups in GEO
   datasets. *p < 0.05, vascular injury versus control group. D)
   Difference in macrophage proportion between the vascular injury and
   control groups in GEO datasets. E) GO‐biological process enrichment
   analysis of the 222 common DEGs. F) Flowchart for analysis of
   macrophage polarization from GEO databases. G) Volcano plot showing the
   DEGs in 4 GEO datasets of macrophage polarization. Blue dots represent
   the genes with significantly downregulated expression in M1 and
   upregulated expression in M2 macrophages. Red dots represent the genes
   with significantly upregulated expression in M1 and downregulated
   expression in M2 macrophages. H) Venn diagram showing the common DEGs
   shared by the 4 GEO datasets. I) Heatmaps showing the expression of 23
   DEGs identified in (H) in the BMDMs of mice with M1 (LPS) and M2 (IL‐4)
   compared to M0 (Vehicle) macrophages from our RNA‐seq dataset. *p
   < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 versus M0 group. J)
   Flowchart for transcription factor (TF) prediction of the 222 common
   DEGs from vascular injury databases. K) Venn diagram showing the common
   DEGs in GEO databases of vascular injury ([90]GSE119549, [91]GSE40637,
   and [92]GSE70410). L) The TFs of the hub DEGs predicted by iRegulon
   analysis. M) Vascular injury‐related DEGs regulated by IRF4. N)
   Relative IRF4 mRNA expression levels in mouse BMDMs upon LPS and IL‐4
   treatment. n = 5 to 6 per group. O) Western blot analysis of IRF4
   protein levels in mouse BMDMs upon LPS and IL‐4 treatment. P,Q)
   Immunofluorescence staining of Mac3 in sections of mouse wire‐induced
   carotid neointima (P) and human coronary arteries (Q). Arrowheads
   indicate Mac3‐positive macrophages. Scale bar = 25 µm. R,S)
   Immunofluorescence costaining of IRF4 (red) with Mac3 (green) in
   sections of mouse wire‐induced carotid neointima (R) and human coronary
   arteries (S). Arrowheads indicate Mac3‐ and IRF4‐positive macrophages.
   Scale bar = 25 µm. ns, not significant. EPIC, Estimating the Proportion
   of Immune and Cancer cells. DEGs, differentially expressed genes. LPS,
   lipopolysaccharides. Data information: The data with error bars are
   presented as mean ± SD. One‐way ANOVA followed by Bonferroni's test (D,
   I) and Two‐tailed Student's t‐test (D, N) are used.

   Consistent with above‐mentioned results of bioinformatics analysis, the
   expression of IRF4 was significantly decreased in primary
   LPS‐stimulated M1 macrophage, but increased in IL‐4‐stimulated M2
   phenotype (Figure [93]1N,O; Figure [94]S1A, Supporting Information).
   Furthermore, abundant IRF4‐positive macrophage infiltration was
   observed in the neointima of mouse carotid arteries (Figure [95]1P,R)
   and human coronary arteries (Figure [96]1Q,S) by immunofluorescence
   microscopy. In addition, the levels of infiltration M1 macrophages
   (marked with CD80) was higher than that of M2 macrophages (marked with
   CD206) in neointima of mouse arteries post injured, and IRF4 expression
   was significantly increased in M2 macrophages (Figure [97]S1B,
   Supporting Information). All these data implicate IRF4 as a potential
   key regulatory factor in macrophage polarization and neointima
   formation post‐injury.

2.2. IRF4 Regulates Macrophage Polarization and its Crosstalk with VSMCs

   To evaluate IRF4 function in macrophage polarization in vivo, we
   generated myeloid‐specific Irf4 knockout (Lyz2‐Irf4‐KO) mice using
   Lyz2‐Cre‐mediated deletion of the Irf4 gene (Figure [98]S2A, Supporting
   Information). Western blotting showed that IRF4 expression was more
   significantly suppressed in BMDMs prepared from Lyz2‐Irf4‐KO mice
   compared to that in control Lyz2‐Cre mice (Figure [99]S2B, Supporting
   Information). The LPS‐stimulated M1 macrophage‐associated genes,
   including iNOS, Cox‐2, Il‐6, and Tnf‐α, were further induced at the
   mRNA level in the Lyz2‐Irf4‐KO BMDMs compared to that in the control
   BMDMs (Figure [100] 2A). In contrast, the IL‐4‐induced M2
   macrophage‐associated genes Arg‐1, Chi3l3, Il‐10, and Mrc‐1 were
   significantly attenuated at the mRNA level in the Lyz2‐Irf4‐KO BMDMs
   (Figure [101]2B). Consistently, Western blotting showed a similar
   impact of Irf4 KO on M1/M2 signature genes at the protein level in
   BMDMs (Figure [102]2C,D; Figure [103]S2C,D, Supporting Information).

Figure 2.

   Figure 2
   [104]Open in a new tab

   IRF4 regulates macrophage polarization and smooth muscle cell activity.
   A) Relative mRNA expression levels of M1 macrophage markers in BMDMs
   from Lyz2‐Cre and Lyz2‐Irf4‐KO mice upon LPS treatment. n = 5 per
   group. B) Relative mRNA expression levels of M2 macrophage markers in
   BMDMs from Lyz2‐Cre and Lyz2‐Irf4‐KO mice upon IL‐4 treatment. n = 5
   per group. C,D) Western blot analysis of M1 macrophage (C) or M2
   macrophage (D) marker expression from mouse BMDMs upon LPS and IL‐4
   treatment. E) Schematic diagram of the collection of conditioned
   macrophage medium and smooth muscle cell activity analysis. F,G) EdU
   incorporation was performed to assess VSMC proliferation in conditioned
   medium collected from cultured BMDMs of Lyz2‐Cre and Lyz2‐Irf4‐KO upon
   DMSO (Vehicle), LPS (F) or IL‐4 (G) treatment. Scale bar = 200 µm.
   n = 5 per group. H,I) The number of migrated VSMCs detected using the
   Transwell migration assay induced by coculturing with conditioned
   medium of Lyz2‐Cre and Lyz2‐Irf4‐KO BMDMs treated with DMSO (Vehicle),
   LPS (H) or IL‐4 (I) in the bottom. Scale bar = 100 µm. n = 5 per group.
   J,K) Relative mRNA expression levels of smooth muscle cell activation
   markers in VSMCs cocultured with conditioned medium from Lyz2‐Cre and
   Lyz2‐Irf4‐KO BMDMs upon DMSO (Vehicle), LPS (J) or IL‐4 (K) treatment.
   n = 5 per group. Data information: The data with error bars are
   presented as mean ± SD. One‐way ANOVA followed by Bonferroni's test (A,
   B, F, K, G, J) and Tamhane's T2 test (A, B, H, I, J) are used.

   Next, we evaluated whether IRF4‐mediated macrophage polarization could
   affect VSMC proliferation, migration, and inflammation. Supernatants
   from Lyz2‐Cre and Lyz2‐Irf4‐KO BMDMs stimulated by LPS or IL‐4 were
   collected and used as conditioned media to treat VSMCs, followed by EdU
   assay and transwell chamber assay to measure proliferation and
   migration (Figure [105]2E). The LPS‐conditioned media from Lyz2‐Irf4‐KO
   BMDMs further exacerbated VSMC proliferation and migration compared to
   the media from control BMDMs (Figure [106]2F,H). And the conditioned
   medium from IL‐4‐induced M2 macrophages promoted VSMC proliferation and
   migration, which is further enhanced in Lyz2‐Irf4‐KO BMDMs
   (Figure [107]2G,I). Additionally, LPS‐conditioned medium increased the
   mRNA levels of proinflammatory cytokines and chemokines (Il‐6, Il‐1β,
   Tnf‐α, Ccl2, and Cxcl2) in VSMCs, with this increase further amplified
   in Lyz2‐Irf4‐KO BMDMs (Figure [108]2J). The expression of
   proinflammatory cytokines and chemokines were significantly decreased
   in VSMCs treated with IL‐4‐conditioned medium from BMDMs, but markedly
   increased by IL‐4‐conditioned medium from Lyz2‐Irf4‐KO BMDMs
   (Figure [109]2K). Further experiments showed that IRF4‐mediated
   polarization of M1 and M2 macrophages primarily regulate VSMC
   proliferation and migration via IL‐6 and IL‐10 secretion, respectively
   (Figure [110]S2E‐I, Supporting Information). These results revealed the
   impact of IRF4 expression on macrophage polarization and the role of
   crosstalk signaling from macrophages to VSMCs in determining their
   proliferation, migration, and inflammatory profiles.

2.3. IRF4 Regulates Macrophage Polarization In Vivo

   The function of macrophage IRF4 in vascular injury was evaluated in the
   mouse carotid artery following wire injury. Based on immunofluorescence
   staining, IL‐6 and iNOS expression in neointimal macrophages was
   significantly elevated in the Lyz2‐Irf4‐KO mice at day 14 after wire
   injury (Figure [111] 3A,B). In contrast, the expression of the
   immunosuppressive genes Arg‐1 and PPARγ was significantly suppressed in
   neointima macrophages in Lyz2‐Irf4‐KO mice compared to that in control
   mice (Figure [112]3C,D). To further demonstrate the regulatory role of
   IRF4 in macrophage polarization, we generated myeloid‐specific
   Irf4‐overexpressing transgenic mice (Lyz2‐Irf4‐TG) by using a
   transgenic construct containing mouse Irf4 cDNA under the control of a
   myeloid‐specific promoter of the mouse Lyz2 gene (Figure [113]S2J,
   Supporting Information). IRF4 expression in BMDMs prepared from
   Lyz2‐Irf4‐TG mice was significantly upregulated in comparison to that
   in BMDMs prepared from wild‐type (WT) controls (Figure [114]S2K,
   Supporting Information). A significant decrease in the expression of
   IL‐6 and iNOS (Figure [115]3E,F) and a significant increase in Arg‐1
   and PPARγ (Figure [116]3G,H) were observed in neointimal macrophages in
   Lyz2‐Irf4‐TG mice at day 14 post‐injury. Therefore, the in vivo loss‐
   and gain‐of‐function analyses both support the notion that IRF4
   expression promotes macrophage polarization to the M2 phenotype in the
   mouse carotid artery following wire injury.

Figure 3.

   Figure 3
   [117]Open in a new tab

   IRF4 regulates macrophage polarization in vivo. A–D) Immunofluorescence
   costaining of IL‐6 (A), iNOS (B), Arg‐1 (C), or PPARγ (D) (red) with
   Mac3 (green) in sections of wire‐induced carotid neointima from
   Lyz2‐Cre and Lyz2‐Irf4‐KO mice. n = 5 per group. E–H)
   Immunofluorescence costaining of IL‐6 (E), iNOS (F), Arg‐1 (G), or
   PPARγ (H) (red) with Mac3 (green) in sections of wire‐induced carotid
   neointima from WT and Lyz2‐Irf4‐TG mice. n = 5 to 6 per group.
   Immunofluorescence data are graphically represented as the relative
   positive area of IL‐6, iNOS, Arg‐1, or PPARγ immunostaining normalized
   to Mac3 and the relative expression level of Mac3 in the neointima. All
   scale bar = 25 µm. IOD, Integrated Optical Density. Data information:
   The data with error bars are presented as mean ± SD. Two‐tailed
   Student's t‐test is used.

2.4. IRF4 Inhibits Arterial Neointima Formation in Mice and Nonhuman Primates

   To identify the pathophysiological role of macrophage IRF4 in neointima
   formation, wire injury‐induced carotid arterial restenosis was
   evaluated in Lyz2‐Irf4‐KO, Lyz2‐Irf4‐TG, and their corresponding
   control mice. The severity of restenosis was determined by the
   neointimal area and the intima/media (I/M) ratio. The Lyz2‐Irf4‐KO mice
   showed exacerbated restenosis compared with the Lyz2‐Cre mice at day 14
   after wire injury (Figure [118] 4A). As shown by Western blotting, the
   expression of proliferation markers (PCNA and CyclinD1) and matrix
   metalloproteinase in remodeling of the vasculature (MMP2) were further
   upregulated in injured vessels in Lyz2‐Irf4‐KO mice compared to the
   control mice (Figure [119]4B; Figure [120]S2L, Supporting Information).
   In addition, molecular markers for cell proliferation (Ki67) and
   remodeling (MMP2) were further increased in neointima smooth muscle
   cells in the Lyz2‐Irf4‐KO mice compared with that in the controls
   (Figure [121]4C,D). In contrast, neointima formation in the carotid
   arteries was significantly attenuated in Lyz2‐Irf4‐TG mice compared
   with that in WT controls (Figure [122]4E). Western blotting showed that
   the expression of PCNA, CyclinD1 and MMP2 was also downregulated in the
   Lyz2‐Irf4‐TG mice compared to the WT mice after injury (Figure [123]4F;
   Figure [124]S2M, Supporting Information). Consistently, Ki67 and MMP2
   signals were suppressed in the neointima smooth muscle cells in the
   Lyz2‐Irf4‐TG mice compared to the WT mice following injury
   (Figure [125]4G,H). All these data suggest that myeloid‐specific Irf4
   exerts a protective effect on neointima formation in mice. Furthermore,
   to explore the clinical relevance of macrophage IRF4 in the treatment
   of vascular restenosis, we constructed a myeloid‐specific adenovirus
   vector carrying the full‐length IRF4 coding sequence driven by the CD68
   promoter (Ad‐PCD68‐IRF4). The IRF4 adenovirus was injected
   intravascularly following the femoral artery injury in monkeys (Figure
   [126]S3A, Supporting Information). Immunofluorescence staining revealed
   that IRF4 was specifically expressed in CD68‐positive macrophages in
   the transfected femoral artery (Figure [127]S3B, Supporting
   Information). Importantly, the neointimal area and the I/M ratio in the
   femoral arteries were significantly reduced in
   Ad‐PCD68‐IRF4‐transfected monkeys compared to the control monkeys
   transfected with Ad‐PCD68‐GFP (Figure [128]4I). Immunofluorescence
   staining showed a lower IL‐6 (M1 phenotype marker) expression
   (Figure [129]4J) and a higher Arg‐1 (M2 phenotype marker) expression
   (Figure [130]4K) in the injured vessels from Ad‐PCD68‐IRF4‐treated
   monkeys than in those from Ad‐PCD68‐GFP‐treated monkeys at day 28
   post‐injury. Furthermore, Ad‐PCD68‐IRF4 treatment reduced α‐smooth
   muscle actin (α‐SMA), Ki67 and MMP2 expression levels in the injured
   vessels (Figure [131]4L–N). These data indicate that IRF4 expression in
   macrophages regulates macrophage polarization and inhibits neointima
   formation in nonhuman primates.

Figure 4.

   Figure 4
   [132]Open in a new tab

   IRF4 inhibits arterial neointima formation in mice and nonhuman
   primates. A) Elastica van Gieson (EVG)‐stained sections showing the
   intimal areas in Lyz2‐Cre and Lyz2‐Irf4‐KO mice after wire‐induced
   carotid injury. Quantified intimal area and the intima‐to‐media (I/M)
   ratio. n = 5 to 6 per group. Scale bar = 25 µm. B) Western blot
   analysis of PCNA, CyclinD1 and MMP2 proteins in the arteries of
   Lyz2‐Cre and Lyz2‐Irf4‐KO mice after sham or at 14 days post‐injury.
   C,D) Immunofluorescence costaining of Ki67 (C) or MMP2 (D) (red) with
   α‐SMA (green) in arteries from Lyz2‐Cre and Lyz2‐Irf4‐KO mice. n = 5
   per group. Scale bar = 25 µm. E) EVG‐stained sections demonstrating the
   intimal areas in WT and Lyz2‐Irf4‐TG mice after wire‐induced carotid
   injury. n = 5 to 6 per group. Scale bar = 25 µm. F) Western blot
   analysis of PCNA, CyclinD1 and MMP2 proteins in the arteries of WT and
   Lyz2‐Irf4‐TG mice after sham or at 14 days post‐injury. G,H)
   Immunofluorescence costaining of Ki67 (G) or MMP2 (H) (red) with α‐SMA
   (green) in arteries from WT and Lyz2‐Irf4‐TG mice. n = 5 per group.
   Scale bar = 25 µm. I) EVG‐stained sections demonstrating the intimal
   areas in sham monkeys and balloon‐induced femoral artery injury from
   Ad‐PCD68‐GFP and Ad‐PCD68‐IRF4 monkeys. Quantified intimal area and the
   I/M ratio. n = 4 per group. Scale bar = 100 µm. J–N) Immunofluorescence
   staining of IL‐6 (J), Arg‐1 (K), α‐SMA (L), Ki67 (M), and MMP2 (N) in
   arteries from Ad‐PCD68‐GFP and Ad‐PCD68‐IRF4 monkeys. n = 4 per group.
   Scale bar = 20 µm. The mouse immunofluorescence data is graphically
   represented as the relative positive area of Ki67 or MMP2
   immunostaining normalized to α‐SMA and the relative expression level of
   α‐SMA in the neointima. The monkey immunofluorescence data is
   graphically represented as the relative expression level of these
   markers in the neointima. IOD, Integrated Optical Density. Data
   information: The data with error bars are presented as mean ± SD.
   Two‐tailed Student's t‐test is used.

2.5. KLF4 Mediates IRF4‐Regulated Macrophage Polarization and Neointimal
Formation

   To further elucidate the mechanisms underlying the anti‐inflammatory
   effects of IRF4 in macrophage regulation and vascular restenosis, IRF4
   chromatin immunoprecipitation (ChIP)‐seq data obtained from 11 public
   GEO datasets were integrated and dissected (Figure [133] 5A). A total
   of 18 core binding targets of IRF4 were identified in heatmap
   (Figure [134]5A,B). Among them, krüppel like factor 4 (KLF4) is a
   well‐established regulator of macrophage polarization and vascular
   neointima formation.^[ [135]^22 , [136]^23 ^] Western blot analysis
   showed a significant increase in KLF4 protein in the carotid arteries
   of Lyz2‐Cre mice 14 days after injury, but its level was significantly
   suppressed in Lyz2‐Irf4‐KO mice (Figure [137]5C; Figure [138]S4A,
   Supporting Information). Further, the mRNA and protein levels of KLF4
   were markedly decreased by LPS treatment in BMDMs from Lyz2‐Cre mice
   but were further reduced in BMDMs from Lyz2‐Irf4‐KO mice
   (Figure [139]5D,F; Figure [140]S4B, Supporting Information). In
   contrast, KLF4 level was markedly elevated by IL‐4 treatment in BMDMs
   from Lyz2‐Cre mice, and this elevation was attenuated in BMDMs from
   Lyz2‐Irf4‐KO mice (Figure [141]5E,G; Figure [142]S4C, Supporting
   Information). These data showed that inactivation of IRF4 suppressed
   KLF4 expression in macrophages. To determine the transcriptional
   regulation of Klf4 by IRF4, we investigated IRF4 interaction with the
   Klf4 promoter in BMDMs. ChIP assays demonstrated that a fragment −1543
   to ≈−1433 bp (P3) from the transcription initiation site of the Klf4
   gene showed significant binding activity to IRF4 and contained a
   putative interferon‐stimulated response element (ISRE)
   (Figure [143]5H). A luciferase assay showed that Klf4 promoter activity
   was induced by IRF4 overexpression in BMDMs (p‐KLF4), and mutation of
   the putative IRF4 binding sites in the Klf4 promoter (m‐pKLF4)
   attenuated IRF4‐induced Klf4 transcription (Figure [144]5I). Treatment
   of BMDMs with APTO‐253, an inducer of KLF4,^[ [145]^24 ^] significantly
   reversed the LPS‐induced Klf4 suppression and further potentiated
   IL‐4‐induced Klf4 expression (Figure [146]S4D, E, Supporting
   Information). In addition, APTO‐253 treatment attenuated LPS‐induced
   upregulation of proinflammatory cytokine expression (Figure [147]5J)
   and potentiated IL‐4‐induced immunosuppressive gene expression
   (Figure [148]5K) in BMDMs from Lyz2‐Irf4‐KO mice, further supporting
   the role of KLF4 in macrophage phenotype transition.

Figure 5.

   Figure 5
   [149]Open in a new tab

   KLF4 mediates IRF4 regulation of macrophage polarization and neointimal
   formation. A) Venn diagram showing the core binding targets of IRF4
   conserved across GEO ChIP‐Seq databases. B) Heatmaps showing the peak
   scores of 18 core binding targets identified in (A). C) Western blot
   analysis of KLF4 protein in the arteries of Lyz2‐Cre and Lyz2‐Irf4‐KO
   mice after sham or at 14 days post‐injury. D,E) Relative mRNA
   expression KLF4 level in BMDMs from Lyz2‐Cre and Lyz2‐Irf4‐KO mice upon
   LPS (D) or IL‐4 (E) treatment. n = 6 per group. F,G) Western blot
   analysis of KLF4 protein in BMDMs from Lyz2‐Cre and Lyz2‐Irf4‐KO mice
   upon LPS (F) or IL‐4 (G) treatment. H) Relative enrichment of IRF4
   binding at the indicated putative interferon‐stimulating response
   elements as determined by ChIP assays. mIL12‐pro84 served as a positive
   control. n = 5 per group. I) Relative KLF4 promoter activity of BMDMs
   transfected with AdIRF4, pKLF4‐luciferase, and m‐pKLF4‐luciferase (IRF4
   binding site pro110 deletion). n = 6 per group. J,K) Relative mRNA
   expression levels of M1 macrophage (J) or M2 macrophage (K) markers in
   BMDMs from Lyz2‐Cre and Lyz2‐Irf4‐KO mice upon DMSO or APTO‐253 (APTO)
   stimulation under LPS or IL‐4 treatment. n = 5 per group. L,M)
   EVG‐stained sections demonstrating the intimal areas in Lyz2‐Cre,
   Lyz2‐Irf4‐KO, Lyz2‐Klf4‐TG, and Lyz2‐Irf4‐KO/Klf4‐TG mice (L) and WT,
   Lyz2‐Irf4‐TG, Lyz2‐Klf4‐KO, and Lyz2‐Irf4‐TG/Klf4‐KO mice (M) after
   wire‐induced carotid injury. n = 5 to 6 per group. Scale bar = 25 µm.
   TSS, transcription start site. Data information: The data with error
   bars are presented as mean ± SD. One‐way ANOVA followed by Bonferroni's
   test (D, I, J, K, L, M) or Tamhane's T2 test (E, J, K), and Two‐tailed
   Student's t‐test (H) are used.

   Next, to investigate whether KLF4 in macrophages affects vascular
   restenosis in vivo, we generated a myeloid‐specific Klf4 knockout mouse
   line (Lyz2‐Klf4‐KO) and a Klf4 overexpressing transgenic mouse line
   (Lyz2‐Klf4‐TG) (Figure [150]S5A‐D, Supporting Information). Then,
   myeloid‐specific Lyz2‐Irf4‐KO/Klf4‐TG mice and Lyz2‐Irf4‐TG/Klf4‐KO
   mice were created and tested (Figure [151]S5E‐H, Supporting
   Information). Compared with the Lyz2‐Cre control group, neointima
   formation following wire injury was significantly inhibited in the
   Lyz2‐Klf4‐TG mice (Figure [152]5L). In addition, KLF4 expression also
   reversed the neointimal area and the I/M ratio in Lyz2‐Irf4‐KO/Klf4‐TG
   mice compared to Lyz2‐Irf4‐KO mice (Figure [153]5L). However, in the
   Lyz2‐Irf4‐TG/Klf4‐KO mice, the protective effects of macrophage IRF4
   overexpression were eliminated by simultaneous KLF4 deletion
   (Figure [154]5M). Therefore, KLF4 is an essential downstream attributor
   for IRF4‐mediated protection against neointima formation.

2.6. Screening of Drugs Targeting IRF4 for Macrophage Polarization

   Considering the protective effect of IRF4 on intimal hyperplasia,
   targeting upregulation of IRF4 is a potential therapeutic approach for
   preventing restenosis. Therefore, we screened a small molecule library
   containing 3008 FDA‐approved drugs using a highly sensitive
   dual‐luciferase‐based reporter assay, among which 9 compounds
   significantly improved IRF4 luciferase activity (Figure [155] 6A,B).
   The effects of screened compounds on IRF4 expression and macrophage
   polarization were further evaluated in BMDMs. Results showed that 9
   compounds were able to upregulate Irf4 levels in BMDMs (Figure
   [156]S6A, Supporting Information). Concurrently, daidzein, artesunate,
   mycophenolic acid, and montelukast significantly inhibited M1 and
   promoted M2 polarization, as evidenced by their effects on reducing
   Il‐6 and Tnf‐α (Figure [157]6C; Figure [158]S6B, C, Supporting
   Information) while increasing Arg‐1 and Il‐10 mRNA levels
   (Figure [159]6C; Figure [160]S6D, E, Supporting Information). Moreover,
   the supernatants from BMDMs treated with the drug candidates and LPS
   were collected as conditioned media to evaluate their effects on VSMC
   proliferation and migration. Consistent with their effect on macrophage
   polarization, daidzein‐, mycophenolic acid‐, and artesunate‐treated
   conditioned medium significantly reduced the proliferation rate
   (Figure [161]6D) and migration rate (Figure [162]6E) of VSMCs. Notably,
   artesunate demonstrated the highest potency in the aforementioned assay
   and was subsequently selected for further evaluation.

Figure 6.

   Figure 6
   [163]Open in a new tab

   Screening drugs as potent regulators of macrophage polarization. A)
   Schematic diagram illustrating the experimental workflow of the
   dual‐luciferase‐based reporter assay kinome screen. B) The relative
   luciferase activity of Irf4 in HEK293T cells treated with 9 screened
   compounds. n = 5 per group. p value versus dimethyl sulfoxide (DMSO)
   group. C) The table shows the comprehensive effects of 9 compounds on
   regulating M1 (Il‐6 and Tnf‐α) and M2 (Arg‐1 and Il‐10) macrophage
   polarization. Blue dots represent a significant decrease compared with
   the control. Red dots represent a significant increase compared with
   the control. D) EdU incorporation was performed to assess VSMC
   proliferation in conditioned medium collected from cultured LPS‐treated
   BMDMs upon stimulation of daidzein, mycophenolic acid, montelukast and
   artesunate. n = 5 per group. p value versus DMSO group. Scale bar =
   200 µm. E) The number of migrated VSMCs detected using the Transwell
   migration assay induced by coculturing with conditioned medium of
   LPS‐treated BMDMs upon compound stimulation in the bottom. n = 5 per
   group. p value versus DMSO group. Scale bar = 100 µm. F) Immunoblotting
   analysis of IRF4 protein levels in LPS‐treated BMDMs upon DMSO or
   artesunate stimulation. G,H) In LPS‐treated BMDMs stimulated with DMSO
   and artesunate, heatmaps show the expression profile of genes related
   to macrophage activation and the inflammatory response (G), and gene
   set enrichment analysis (GSEA) plots show the enrichment score of
   macrophage activation and inflammatory response gene sets based on
   RNA‐seq data (H). n = 4 per group. I,J) In VSMCs cocultured with
   conditioned medium of LPS‐treated BMDMs upon DMSO or artesunate
   stimulation, heatmaps showing the expression profile of genes related
   to cell proliferation and migration (I), GSEA plots showing the
   enrichment score of cell proliferation and migration gene sets based on
   RNA‐seq data (J). n = 4 per group. Data information: The data with
   error bars are presented as median ± interquartile range (B) or mean ±
   SD (D, E). Two‐tailed Student's t‐test (B, D, E) and Mann‐Whitney U
   test (B) are used.

   Western blotting demonstrated that the optimal concentration of
   artesunate to upregulate IRF4 expression was 10 µM (Figure [164]S6F,
   Supporting Information), and at this concentration, endogenous IRF4
   expression in BMDMs was significantly upregulated by artesunate
   treatment both with and without LPS (Figure [165]6F; Figure [166]S6G,
   Supporting Information). Further, Irf4 and Klf4 mRNA levels in LPS‐ or
   IL‐4‐ treated Lyz2‐Cre BMDMs, but not in Lyz2‐Irf4‐KO and Lyz2‐Klf4‐KO
   BMDMs, were significantly increased when treated with artesunate
   (Figure [167]S7A, B, Supporting Information). In conjunction, the mRNA
   levels of Cox‐2, Il‐6, Tnf‐α, and Il‐β in LPS‐treated Lyz2‐Cre BMDMs,
   but not in Lyz2‐Irf4‐KO and Lyz2‐Klf4‐KO BMDMs, were decreased when
   treated with artesunate (Figure [168]S7C, Supporting Information), and
   similar phenomena could be observed in IL‐4‐treated Lyz2‐Irf4‐KO and
   Lyz2‐Klf4‐KO BMDMs (Figure [169]S7D, Supporting Information).
   Additionally, these results were further confirmed by Western blotting
   assay (Figure [170]S8A‐D, Supporting Information). Meanwhile, when
   conditioned medium from artesunate‐treated BMDMs was applied to VSMCs,
   the proliferation and remodeling markers in VSMCs, including Ki‐67,
   Pcna, Mmp2, and Mmp9, were significantly decreased (Figure [171]S8E,
   Supporting Information). RNA‐seq analyses also showed significantly
   suppressed macrophage activation and inflammation‐associated genes in
   the artesunate‐treated BMDMs (Figure [172]6G,H; Figure [173]S9A,
   Supporting Information). Furthermore, artesunate treatment attenuated
   the activation of a broad spectrum of genes related to the
   proliferation and migration of conditioned medium‐treated VSMCs
   (Figure [174]6I,J). Finally, further mechanism data showed that
   artesunate dramatically increased the phosphorylation of signal
   transducer and activator of transcription 6 (pSTAT6) (Figure [175]S9B,
   Supporting Information), and artesunate‐biotin was able to pull down
   the STAT6 protein (Figure [176]S9C, Supporting Information). In
   addition, the expression of pSTAT6, IRF4, KLF4 and Arg‐1 activated by
   IL‐4 and artesunate were abrogated by pSTAT6 inhibitor AS1517499
   (Figure [177]S9D, Supporting Information).

2.7. Artesunate Attenuates Arterial Neointima Formation in Mice

   To examine the potential efficacy of artesunate in vivo, we evaluated
   the effect of artesunate on vascular restenosis in mice. Mice with
   carotid arterial injury were randomized into the vehicle and artesunate
   treatment groups. There was no obvious neointima formation in the
   carotid artery in sham controls (Figure [178] 7A). The neointimal area
   and the I/M ratio were significantly reduced in artesunate‐treated mice
   compared with vehicle‐treated mice 14 days after injury
   (Figure [179]7A). Immunofluorescence staining showed that, compared to
   the vehicle group, the expression of IRF4 in neointimal Mac3‐positive
   macrophages was upregulated in the artesunate‐treated mice
   (Figure [180]7B). Likewise, the expression of IL‐6 and iNOS (M1
   phenotype) was downregulated, but the levels of Arg‐1 and PPARγ (M2
   phenotype) were increased in the artesunate‐treated mice compared with
   the vehicle controls 14 days after injury (Figure [181]7C–F). Moreover,
   immunofluorescence staining for Ki67 and MMP2 in SMCs demonstrated that
   artesunate treatment inhibited VSMC proliferation and remodeling
   (Figure [182]7G,H). These data suggest that artesunate has a protective
   effect against arterial neointima formation, partly by upregulating
   IRF4 expression and promoting M2 macrophage polarization in mice.

Figure 7.

   Figure 7
   [183]Open in a new tab

   Artesunate attenuates arterial neointima formation in mice. A)
   EVG‐stained sections show the intimal areas in the vehicle‐ and
   artesunate‐treated mice after sham or at 14 days post wire‐induced
   carotid injury. Quantified intimal area and the intima‐to‐media (I/M)
   ratio. n = 6 to 7 per group. Scale bar = 25 µm. B–F) Immunofluorescence
   costaining of IRF4 (B), IL‐6 (C), iNOS (D), Arg‐1 (E), or PPARγ (F)
   (red) with Mac3 (green) in arteries from vehicle‐ and
   artesunate‐treated mice. Immunofluorescence data are graphically
   presented as the relative positive area of IRF4, IL‐6, iNOS, Arg‐1, or
   PPARγ immunostaining normalized to Mac3 and the relative expression
   level of Mac3 in the neointima. n = 5 per group. Scale bar = 25 µm.
   G,H) Immunofluorescence costaining of Ki67 (G) or MMP2 (H) (red) with
   α‐SMA (green) in arteries from vehicle‐ and artesunate‐treated mice.
   n = 5 per group. Scale bar = 25 µm. Immunofluorescence data are
   graphically presented as the relative positive area of Ki67 or MMP2
   immunostaining normalized to α‐SMA and the relative expression level of
   α‐SMA in the neointima. IOD, Integrated Optical Density. Data
   information: The data with error bars are presented as mean ± SD.
   Two‐tailed Student's t‐test is used.

2.8. Artesunate Attenuates Arterial Neointima Formation in Nonhuman Primates

   To further explore the translational implications of artesunate
   treatment in arterial stenosis, we evaluated the effect of artesunate
   on balloon injury‐induced femoral artery neointima formation in rhesus
   monkeys. Balloon injury‐induced arterial stenosis 28 days after injury
   was measured by MRI, and the results showed that the percent lumen
   stenosis and percent change in arterial wall thickness in the
   artesunate treatment group tended to decrease (Figure [184] 8A). This
   protective effect on vascular restenosis was substantiated by
   histological analysis based on the reduced intima area and I/M ratio
   (Figure [185]8B). In line with the observations in mouse models,
   artesunate treatment significantly augmented IRF4 expression in the
   injured arteries compared with vehicle controls (Figure [186]8C), along
   with the decreased expression of IL‐6 (Figure [187]8D) and the
   increased expression of Arg‐1 (Figure [188]8E) in neointimal
   Mac3‐positive macrophages. Expression of Ki67 and MMP2 in neointimal
   α‐SMA‐positive SMCs were also significantly downregulated in the
   injured arteries from the artesunate‐treated monkeys, compared to the
   vehicle controls (Figure [189]8F,G). Therefore, these data support that
   artesunate ameliorates arterial stenosis in nonhuman primates.

Figure 8.

   Figure 8
   [190]Open in a new tab

   Artesunate attenuates arterial neointima formation in nonhuman
   primates. A) In rhesus monkey treated with vehicle and artesunate pre‐
   and post‐balloon‐induced femoral artery injury, 3D volume rendering
   reconstruction of enhanced magnetic resonance angiography (MRA) images
   (upper row) showing the luminal blood flow and 3D turbo spin echo
   T2‐weighed (black blood) MR images (lower row) showing the vascular
   wall. The percent lumen stenosis and percent change in vessel wall
   thickness were calculated to assess neointima formation. Arrows
   indicate the area of arterial injury. S, superior. I, inferior. n = 4
   per group. B) EVG‐stained sections demonstrating the intimal areas in
   the vehicle‐ and artesunate‐treated monkeys at 28 days post
   balloon‐induced femoral artery injury. Quantified intimal area and the
   I/M ratio. n = 4 per group. Scale bar = 100 µm. C–E) Immunofluorescence
   costaining of IRF4 (C), IL‐6 (D), or Arg‐1 (E) (red) with Mac3 (green)
   in arteries from vehicle‐ and artesunate‐treated monkeys.
   Immunofluorescence data are graphically presented as the relative
   positive area of IRF4, IL‐6, or Arg‐1 immunostaining normalized to Mac3
   and the relative expression level of Mac3 in the neointima. n = 4 per
   group. Scale bar = 20 µm. F,G) Immunofluorescence costaining of Ki67
   (F) or MMP2 (G) (red) with α‐SMA (green) in arteries from vehicle‐ and
   artesunate‐treated monkeys. Immunofluorescence data are graphically
   presented as the relative positive area of Ki67 or MMP2 immunostaining
   normalized to α‐SMA and the relative expression level of α‐SMA in the
   neointima. n = 4 per group. Scale bar = 20 µm. IOD, Integrated Optical
   Density. Data information: The data with error bars are presented as
   mean ± SD. Two‐tailed Student's t‐test is used.

3. Discussion

   Vascular restenosis and the need for repeated revascularization remain
   a prevalent clinical challenge globally.^[ [191]^5 , [192]^25 ^]
   Through an integrated analysis of post‐injury vasculature in mice and
   nonhuman primates, we have discovered that macrophage IRF4 serves as a
   protective effector against arterial restenosis. This effect is
   mediated, in part, by IRF4‐induced KLF4 expression and the subsequent
   modulation of M2 macrophage polarization. Furthermore, a high
   throughput drug screening based on IRF4 transcriptional induction
   identified an FDA‐approved drug, artesunate, as a potent activator of
   IRF4. Its therapeutic effect against vascular injury‐induced restenosis
   was demonstrated in vivo across different species. Thus, our study
   uncovers a previously unappreciated role of IRF4 in vascular protection
   by modulating inflammatory responses and highlights a potential
   therapeutic approach against injury‐induced remodeling.

   Arterial injuries resulting from intravascular stenting,
   endarterectomy, and angioplasty trigger complex and dynamic
   inflammatory responses,^[ [193]^26 ^] where infiltrating monocytes and
   monocyte‐derived macrophages have pivotal impact on the entire process
   from inflammation to repair in post‐injury remodeling.^[ [194]^27 ^] In
   the first week after stenting, nearly 40% of neointimal cells are
   infiltrated macrophages.^[ [195]^28 ^] As established in the
   literature, macrophages are highly heterogeneous at the functional and
   molecular levels in the post‐injury vasculature, and are generally
   characterized by their M1 versus M2 polarization status.^[ [196]^13 ^]
   M1 macrophages secrete proinflammatory factors and promote VSMC
   proliferation and migration, while M2 macrophages exhibit
   anti‐inflammatory properties and reduce the incidence of in‐stent
   restenosis.^[ [197]^13 , [198]^14 ^] However, the molecular mechanisms
   for macrophage polarization regulation, particularly in vascular
   restenosis, remain to be fully elucidated. The therapeutic potential of
   targeting macrophage polarization to treat this major disease has not
   yet been clearly demonstrated.

   Among the myriads of molecular pathways controlling macrophage
   polarization, a number of transcription factors have been uncovered
   that play pivotal roles in modulating macrophage phenotypes.^[ [199]^15
   , [200]^16 ^] IRFs are hub transcription factors that were initially
   discovered to regulate immune responses during viral infections.^[
   [201]^17 , [202]^18 ^] Earlier studies from our group also found that
   IRF4 is expressed and exerts its effects in nonimmune cells, which is
   further confirmed in this study, where several other cell types, in
   addition to macrophages, in the vascular neointima also express IRF4
   (Mac3^−IRF4^+ cells). As previously reported, IRF4 promotes cardiac
   hypertrophy in cardiomyocytes^[ [203]^29 ^] and has neuroprotective
   effects during stroke.^[ [204]^30 ^] Through VSMC‐specific knockout and
   transgenic over‐expression in mice, our group has revealed that IRF4
   inhibits VSMC proliferation and migration, thus having a protective
   function against restenosis in post‐injury arteries.^[ [205]^23 ^] In
   addition, IRF4 has been implicated as an anti‐inflammatory mediator in
   dendritic cells^[ [206]^31 ^] and macrophages.^[ [207]^32 , [208]^33 ^]
   However, the effects of IRF4 on macrophage polarization and subsequent
   neointima hyperplasia, particularly in drug screening targeting IRF4
   and efficacy evaluation, still require extensive study. Consistent with
   previous reports,^[ [209]^19 , [210]^20 ^] we found that macrophage
   IRF4 skewed the M1/M2 polarization balance toward the anti‐inflammatory
   M2 type in vitro and in vivo during post‐injury vascular remodeling.
   Further molecular studies uncovered KLF4 as a key downstream factor in
   IRF4‐mediated regulation of macrophage polarization. This observation
   aligns with the previously reported role of KLF4 in the monocyte
   differentiation^[ [211]^34 , [212]^35 ^] and vascular pathology.^[
   [213]^36 ^] Since macrophage infiltration is an early event in the
   inflammatory cascade that leads to VSMC proliferation/migration and
   neointima formation after endothelial injury,^[ [214]^37 , [215]^38 ^]
   IRF4‐mediated macrophage polarization should provide early protection
   against neointimal formation. In our current work, this concept was
   validated in vivo through macrophage‐specific manipulation of IRF4
   expression both in mice and nonhuman primates, further supporting its
   potential for clinical translation.

   Given the pivotal role of macrophages in neointima formation, systemic
   inactivation or depletion of macrophages has been proposed as a
   therapeutic strategy.^[ [216]^39 ^] However, such blunt treatment may
   have unintended adverse consequences, such as fatal infection due to
   non‐specific immunosuppression.^[ [217]^40 ^] Furthermore, due to the
   complexity of macrophage‐mediated signaling pathways in CVD
   pathogenesis, the mere applications of cell mediators (such as IL‐1,
   IL‐10, and MMP‐9) often fail to yield expected benefits, imposing the
   challenges encountered in most cardiovascular clinical trials.^[
   [218]^41 , [219]^42 ^] Therefore, it is imperative to investigate
   mechanism‐based macrophage phenotype regulatory molecules that are
   integral to the pathological vascular remodeling, and to ensure that
   these targets do not trigger serious adverse effects. In this study, we
   discovered an unexpected effect of artesunate as a potent activator of
   IRF4 expression in macrophages through unbiased high‐throughput
   screening using a library of FDA‐approved compounds. Artesunate, a
   semi‐synthetic derivative of artemisinin, is an effective first‐line
   antimalarial agent recommended by the World Health Organization (WHO)
   for severe malaria, and has a worldwide safety record for adults,
   children and pregnant women.^[ [220]^43 , [221]^44 ^] Additionally,
   artesunate has been reported to have therapeutic properties against
   infections, cancers, autoimmune diseases, and atherosclerosis.^[
   [222]^45 , [223]^46 , [224]^47 ^] Early reports have shown that
   artemisinin ameliorates inflammatory bowel disease by promoting M2
   polarization^[ [225]^48 ^] and alleviates atherosclerotic lesions by
   promoting macrophage autophagy.^[ [226]^48 ^] Our screening results
   indicate that artesunate enhances IRF4 expression by modulating the
   IRF4 promoter activity. To further explore the interaction between
   artesunate and IRF4, as well as its impact on M2 polarization, we
   initially analyzed the potential signals linked to artesunate and
   macrophage polarization. Previous studies have reported that the STAT6
   signaling pathway is crucial for IRF4 expression and participates in M2
   activation.^[ [227]^50 ^] M2 polarization occurs via the STAT6‐IRF4
   signaling pathway.^[ [228]^51 , [229]^52 ^] Moreover, artesunate has
   been shown to enhance pSTAT6 expression in macrophages, suggesting that
   STAT6 is a key regulator in balancing M1/M2 polarization and mediating
   the anti‐colitis effects of artesunate.^[ [230]^53 ^] Similarly, our
   results demonstrate that IRF4 and pSTAT6 induction is pivotal to
   artesunate‐mediated anti‐inflammatory function and artesunate is
   proposed as a new anti‐restenosis therapy in both rodent and nonhuman
   primates. Further, since macrophage‐rich areas are prevalent in the
   lesion of inflammatory vascular diseases, such as atherosclerosis,^[
   [231]^54 ^] graft vasculopathy,^[ [232]^55 ^] and autoimmune diseases
   associated vasculitis,^[ [233]^56 ^] it is worthwhile to investigate
   the therapeutic potential of artesunate in these vascular disorders.

   The recent clinical trials showed that anti‐inflammatory strategies
   have successfully reduced cardiovascular events, which give rise
   inflammation as a rational target for treating CVD. Canakinumab, a
   neutralizing antibody against IL‐1β (mainly produced by macrophages),
   is the first drug shown in clinical trials to specifically constrain
   inflammation and the recurrence rate of cardiovascular events.^[
   [234]^7 ^] Colchicine, a broad anti‐inflammatory drug, significantly
   reduces the risk of cardiovascular event in patients with myocardial
   infarction and chronic coronary artery disease.^[ [235]^8 , [236]^10 ^]
    Furthermore, the most common autoimmune rheumatic diseases, such as
   rheumatoid arthritis and systemic lupus erythematosus, are associated
   with increased adverse cardiovascular risk, which further proves that
   inflammatory process per se is the risk factor for CVD and related
   complications.^[ [237]^3 , [238]^57 , [239]^58 ^] Given the conclusive
   relationship between autoimmune diseases and cardiovascular events,
   more anti‐inflammatory drugs for rheumatism are being tested in
   clinical trials for CVD. For instance, hydroxychloroquine and
   methotrexate have both been found to reduce CVD risk in rheumatoid
   arthritis.^[ [240]^59 , [241]^60 ^] Sarilumab and ziltivekimab,
   monoclonal antibodies targeting the IL‐6 receptor and IL‐6, are being
   evaluated for their effects on cardiovascular risk in patients with
   rheumatoid arthritis ([242]NCT04350216) and chronic kidney disease
   ([243]NCT05021835) with high CRP levels, respectively. Etanercept, a
   TNF‐α blocker, is also undergoing clinical trials in patients with
   acute myocardial infarction ([244]NCT01372930). In our work, due to its
   anti‐inflammatory activities, artesunate represents a promising
   cardiovascular medicine, and its demonstrated efficacy further supports
   the potent role of inflammation in restenosis. Over time, we anticipate
   that, in addition to lipid‐lowering therapy, combination therapies with
   anti‐inflammatory agents will become the standard of care for patients
   with CVD, especially restenosis.

   There are still some limitations in the current study. While cell
   culture, mouse and monkey experiments provide essential and highly
   valuable approaches to exploring mechanisms and therapeutic strategies,
   a comprehensive analysis of the IRF4‐KLF4 effects on macrophage
   polarization and vascular restenosis in clinical cases is needed in the
   future. Second, only male animal models were performed in our study to
   avoid the unknown impact of periodic changes in the abundance of
   estrogen in female animals throughout the menstrual cycle. This
   limitation leads to insufficient exploration of the benefits and
   potential side effects of artesunate for clinical translation in
   females. Further, the therapeutic efficacy of artesunate in reversing
   disease progression, off‐target effects, timing and dosage of
   administration, and potential side effects from long‐term treatment
   remains to be studied. However, drug delivery using the drug‐eluting
   stent allows local enrichment of artesunate to a vascular segment to
   reduce neointimal hyperplasia without causing serious systemic side
   effects is one of the possible strategies.

4. Conclusion

   Collectively, we revealed the novel role of IRF4 in macrophage
   polarization with a potent protective effect against arterial
   injury‐induced restenosis. This beneficial effect is mediated, in part,
   by IRF4‐dependent induction of KLF4 expression and subsequent M2
   macrophage polarization. In particular, artesunate was screened and
   identified to possess unexpected anti‐inflammatory activity, and
   exhibited conserved therapeutic efficacy for vascular restenosis in
   rodents and primates. Overall, this study reveals a new mechanistic
   basis and a novel therapeutic approach for vascular restenosis and
   related inflammatory cardiovascular diseases.

5. Experimental Section

Study Design

   The primary goal of this study was to explore the key determinant of
   inflammation‐derived vascular restenosis. To this end,
   transcriptome‐wide cross‐species gene expression analysis was performed
   and identified IRF4 as a promising driver of macrophage polarization
   and vascular restenosis. Then, myeloid‐specific and targeted gain‐ and
   loss‐of‐function of IRF4 models were established and characterized on
   injury‐induced artery neointima formation and macrophage polarization
   in mice or nonhuman primates. The molecular mechanisms underlying the
   anti‐inflammatory effects of IRF4 were determined by bioinformatics
   analysis and ChIP assay, followed by biological exploration and
   validation. Finally, small molecule screening for IRF4 transcriptional
   activators was implemented based on a reporter assay and the top
   candidate was validated in mouse and nonhuman primate models of
   arterial restenosis. The sample size of animals was determined based on
   past experience and published literature and was specified in the
   figure legends. Animals with identical background were randomly
   assigned to each group. Histopathological evaluation was performed in a
   double‐blind manner. In vitro, experiments were carried out at least
   three replicates unless otherwise specified.

Experimental Animals

   All animal protocols were approved by the Animal Care and Use Committee
   of the National Translational Science Center for Molecular Medicine
   (approval number: 2020‐NTSCMM‐ID005) and Zhongnan Hospital of Wuhan
   University (approval number: ZN2022116). Animal experiments were
   conducted according to institutional ethical guidelines. The animals
   received humane care according to the Guide for the Care and Use of
   Laboratory Animals published by the National Academy of Sciences and
   the National Institutes of Health. Male mice aged 10 to 12 weeks and
   weighing 20 to 30 g were included in the experiments. Mice were housed
   strictly inbred in pathogen‐free individually ventilated cages (3‐5
   mice per cage) with free access to clean tap water and rodent chow. For
   monkey experiments, male rhesus monkeys aged 5 to 7 years (5 to 12 kg)
   were purchased from Topgene Biotechnology (Guangzhou, China). The
   monkeys included in this study were healthy, passed the physical
   examination, and met the standards of local quarantine inspection. The
   monkdeys were individually housed and acclimated for at least 1 week
   before experiments. All monkeys were fed a regular chow diet twice
   daily, seasonal fruits once daily, and clean water freely. The mice and
   monkeys were housed at a controlled temperature (23 ± 2 °C), humidity
   (50 to 65%) and photoperiod [12:12 hour light/dark cycle].

Vascular Injury Models in Mice

   Wire injuries induced neointima formation in the mouse carotid artery,
   as previously described.^[ [245]^61 ^] Briefly, after anesthetization
   (sodium pentobarbital, 80 mg kg^−1, intraperitoneally), the left
   carotid artery was carefully exposed via a midline neck incision. The
   left external carotid artery was ligated proximally to the bifurcation
   point with 8‐0 silk suture (Ethicon). Additional vascular clamps
   temporarily interrupted the internal and common carotid arterial blood
   flow. A transverse arteriotomy was made in the external carotid artery,
   and a flexible angioplasty guidewire was introduced into and withdrawn
   from the arterial lumen five times with a rotating motion. Then, the
   wire was removed, and vascular clamps were released to resume blood
   flow. The skin incision was closed with a single suture. For treatment
   experiments, the mice received either a vehicle and artesunate
   following sham or vascular injury for 14 days. Since the maximum oral
   dose and rectal dose of artesunate for clinical application in malaria
   treatment is 8 mg kg^−1[ [246]^62 ^] and 10 mg kg^−1,^[ [247]^63 ^]
   respectively, according to the animal equivalent dose (AED)
   calculation,^[ [248]^64 ^] the AED for mice is approximately 98.4‐123
   mg kg^−1. Additionally, considering the reported doses from previous
   studies on artesunate administration in mice, the dosage of 100
   mg kg^−1 (gavage, qd, Sigma‐Aldrich) was ultimately used for mice in
   this study. At 14 days post‐injury, mice were euthanized via an
   overdose of sodium pentobarbital (150 mg kg^−1). Both injured left and
   uninjured right carotid arteries were excised.

Vascular Injury Models in Rhesus Monkeys

   For the monkey model of intimal hyperplasia, injury to the femoral
   artery was achieved as described below. Monkeys were first concurrently
   anesthetized with atropine (0.04 mg kg^−1), xylazine (0.1 mL kg^−1),
   and Zoletil (15 mg kg^−1) by intramuscular injection. Once adequately
   anesthetized, the monkey was fixed, and their heart rate, blood
   pressure, and oxygen saturation were measured throughout the process.
   The left femoral artery and its branches were surgically exposed and
   carefully dissected. Then, the branches of the femoral artery were
   ligated from the bifurcation with 8‐0 silk sutures and the proximal and
   distal blood vessels were clamped. After the femoral artery puncture,
   the angioplasty balloon was inflated to a pressure of 12 atmospheres.
   The injury paradigm used three balloon inflations for 3 min each, with
   1‐minute reperfusion between each inflation. After removing the
   catheter, the injured vessel was examined visually to ensure that blood
   flow was restored. The femoral artery was then surgically stapled, and
   the incision was sutured. After the surgery, monkeys received aspirin
   (25 mg kg^−1, qd) and clopidogrel (75 mg kg^−1, qd) in the diet for 7
   days to decrease platelet activation and to prevent possible
   thrombosis. The monkeys were allowed to recover in their original cage.
   For treatment experiments, the monkeys were given vehicle or artesunate
   with diet following vascular injury for 28 days. The dosage of
   artesunate for monkeys was converted from that of the mice based on
   body surface area (Meeh‐Rubner formula). Considering that the average
   weight of mice is 25 g and that of monkeys is 10 kg, their respective
   body surface areas were calculated, and the dosage of artesunate for
   the monkeys was determined to be 17.6 mg kg^−1 (qd, Sigma‐Aldrich). At
   28 days post‐injury, monkeys were anesthetized, and the injured femoral
   arteries were excised.

Genetic Modification of Mice

   Myeloid‐specific genetic modification mice were generated by
   constructing modification elements downstream of the Lyz2 promoter.^[
   [249]^65 ^] Lyz2‐Irf4‐KO (Irf4^flox/flox /Lyz2‐Cre) mice were generated
   by mating Irf4^flox/flox (stock number: 0 09380, Jackson Laboratory)
   mice with Lyz2‐Cre mice (stock number: 0 04781, Jackson Laboratory). To
   obtain myeloid‐specific IRF4 transgenic mice, full‐length mouse IRF4
   cDNA was subcloned downstream of the mouse Lyz2 promoter (Lyz2‐Irf4‐TG
   mice). Klf4‐floxed mice were generated using the CRISPR/Cas9 system.^[
   [250]^23 ^] Lyz2‐Klf4‐KO mice were generated by mating Klf4^flox/flox
   mice with Lyz2‐Cre mice. Lyz2‐Klf4‐TG mice were generated by mating
   CAG‐loxP‐CAT‐loxP‐Klf4 mice with Lyz2‐Cre mice. Lyz2‐Irf4‐KO/Klf4‐TG
   mice were generated by mating Irf4^flox/flox /Lyz2‐Cre mice with Klf4
   conditional transgenic mice. Lyz2‐Klf4‐KO/Irf4‐TG mice were generated
   by mating Lyz2‐Klf4‐KO mice with Irf4‐TG mice. The constructed
   expression vectors were microinjected into fertilized C57BL/6J embryos.

Human Samples

   Human samples of coronary arteries with in‐stent restenosis were
   obtained from coronary heart disease (CHD) patients who previously had
   percutaneous coronary intervention (PCI) and were undergoing heart
   transplants. Coronary arteries from matched donors who wanted heart
   transplantation but were unsuitable for transplantation were used as
   control samples. The demographic and clinical characteristics of human
   donors are summarized in Table [251]S1 (Supporting Information). All
   procedures involving human samples adhered to the principles of the
   Declaration of Helsinki and were approved by the ethics committees of
   the Renmin Hospital of Wuhan University (approval number: 2014‐K‐016).
   Written informed consent was obtained from each participating subjects.

Cell Culture and Preparation of Conditioned Medium

   BMDMs were isolated and differentiated from mice as previously
   described.^[ [252]^66 ^] Briefly, bone marrow was harvested from the
   femur and tibia of the indicated mice and filtered through a 70 µm
   strainer. After treatment with Red Cell Lysis Buffer (Sigma), the
   isolated cells were cultured in RPMI medium (Gibco) containing 10% FBS
   (Gibco) and 1% penicillin‐streptomycin (Gibco), and stimulated with
   50 ng mL^−1 recombinant mouse M‐CSF (Biolegend) for 5 days. BMDM purity
   was assessed by flow cytometry following immunostaining with anti‐CD11b
   and anti‐F4/80 (both from Biolegend), resulting in > 98% CD11b^+
   F4/80^+ cells, indicating a highly pure population of BMDMs. To induce
   an M1‐like phenotype, BMDMs were treated with LPS (100 ng mL^−1, Sigma)
   for 24 h. To obtain an M2‐like phenotype, BMDMs were stimulated with
   IL‐4 (10 ng mL^−1, Biolegend) for 24 h. In some experiments, as
   indicated, BMDMs were treated with vehicle (DMSO), chemotherapy
   compounds (10 µM, Selleck Chemicals), or APTO‐253 (5 µM,
   MedChemExpress). To obtain a conditioned medium, BMDMs were initially
   stimulated with either LPS or IL‐4 for 24 h. After this stimulation
   period, the supernatant was removed and the cells were washed. Then,
   BMDMs were treated with fresh medium and incubated for an additional 24
   h. Following this incubation, the supernatant was collected as
   conditioned medium.

   The VSMC cell line generated from mouse aortic vascular smooth muscle
   cells (MOVAS) was obtained from the American Type Culture Collection
   (ATCC). The HEK 293T cell line was purchased from the Cell Bank of the
   Type Culture Collection of the Chinese Academy of Sciences (Shanghai,
   China). These cells were cultured in DMEM (Gibco) containing 10% FBS
   (Gibco) and 1% penicillin‐streptomycin (Gibco) at 37 °C in a 5%
   CO[2]‐humidified atmosphere.

Adenovirus Construction and Injection

   To generate myeloid‐specific gene expression vectors in rhesus monkeys,
   a CD68 gene promoter was used to construct adenovirus. The adenovirus
   plasmid carrying the full‐length IRF4 coding sequence (Ad‐P
   [CD68]‐IRF4) and the adenovirus GFP control plasmids (Ad‐P [CD68]‐GFP)
   were constructed by Hanbio (Shanghai, China). As described
   previously,^[ [253]^67 ^] Ad‐P [CD68]‐GFP and Ad‐P [CD68]‐IRF4 were
   injected intravascularly (1 × 10^9 PFU per monkey) at 1 week after
   femoral artery injury. At 28 days post‐injury, monkeys were
   anesthetized and the femoral arteries were excised.

Elastica Van Gieson and Immunofluorescence Staining

   Carotid and femoral arteries were harvested at the indicated times
   post‐injury. The arterial segments were excised after circulation
   perfusion and fixed with 4% paraformaldehyde in PBS before dehydration
   in ethanol and xylene, and embedding in paraffin for further
   preparation for histological analysis and immunofluorescence. Each
   paraffin section was 5 µm in thickness.

   To evaluate vascular tissue structures and the extent of carotid artery
   luminal occlusion, consecutive sections from the bifurcation site of
   the carotid artery were stained with Elastica van Gieson (EVG).
   Measurements included luminal area, internal elastic lamina and
   external elastic lamina. The intimal area and intima‐to‐media (I/M)
   ratio were calculated to assess neointima formation. All sections were
   analyzed by a single observer blinded to the study design using
   Image‐Pro Plus 6.0.

   For immunofluorescence staining, the arterial sections were collected,
   placed onto gelatin‐coated glass slides, and blocked with 20% donkey
   serum for 1 h. Tissue sections were then incubated with corresponding
   primary antibodies (Table [254]S2, Supporting Information) overnight at
   4 °C, followed by incubation for 1 h with appropriate secondary
   antibodies. The secondary antibodies used: included Alexa Fluor
   488‐conjugated donkey anti‐rat IgG (A21208; 1:200; Invitrogen,
   Carlsbad, CA, USA), Alexa Fluor 488‐conjugated donkey anti‐mouse IgG
   (A21202; 1:200; Invitrogen), Alexa Fluor 568‐conjugated donkey
   anti‐rabbit IgG (A11042; 1:200; Invitrogen) or Alexa Fluor
   568‐conjugated donkey anti‐mouse IgG (A10037; 1:200; Invitrogen). After
   nuclear staining for 30 s with 4′,6‐diamidino‐2‐phenylindole (DAPI),
   the slides were imaged using an inverted confocal microscope (TCS SP8,
   Leica, Germany).

Multiplex Immunohistochemistry Assay

   Briefly, the tissue sections were dewaxed, followed by antigen
   retrieval in TRIS‐EDTA (pH = 9) buffer. After blocked with 5% goat
   serum in PBS, the tissue sections were sequentially incubated with
   primary antibodies. After being washed, the bound antibodies were
   reacted with goat anti‐rabbit IgG (Panovue, Beijing, China) and stained
   using the tyramine signal amplification (TSA)‐Indirect kit
   (PerkinElmer, USA), according to the manufacturer's manual. The primary
   and secondary antibodies were removed by microwaving, and then blocked
   again by blocking solution. The sections were stained with another
   primary antibody, and the above processes were repeated. Finally, DAPI
   was used to stain nuclei. Images were analyzed with HALO Image analysis
   software and Inform software (version 3.0).

Magnetic Resonance Imaging (MRI) and Analysis

   The vehicle‐ and artesunate‐treated monkeys were imaged by
   high‐resolution MRI with a 3.0‐Tesla (Magnetom Prisma, Siemens
   Healthcare GmbH, Germany) before and 28 days after balloon‐induced
   femoral artery injury. The MRI protocol consisted of the following 4
   steps: (1) gradient echo series to localize anatomic structures; (2)
   time‐of‐flight (TOF) magnetic resonance angiography (MRA) to localize
   the vascular stenosis and provide anatomic landmarks to define the
   precise location of the stenosis; (3) unenhanced T2‐weighted 3D
   isotropic TSE sequences to show the vascular wall; and (4) pre‐ and
   post‐contrast fast low‐angle shot 3D (fl3d) MRA sequences to show blood
   flow in the femoral artery. Contrast‐enhanced MRA was performed with
   2.5 mL s^−1 of 0.1 mL kg^−1 gadobutrol (Gadovist; Bayer Vital GmBH,
   Leverkusen, Germany). The distance between the common femoral artery
   bifurcation and maximum stenosis was used to predefine anatomic
   landmarks and obtain reproducible image positions. Subsequent
   angiography sequences were performed using this superficial
   receive‐only coil centered at the stenosis level. A blinded operator
   performed computer‐assisted morphometric analysis of MRI images,
   including measuring vascular lumen diameter and wall thickness by
   manually tracing vascular borders (RadiAnt DICOM Viewer, Medixant,
   Poznan, Poland). Percent lumen stenosis was calculated according to the
   following formula: 100 × [(diameter of stenosis distal lumen) −
   (diameter of the lumen at stenosis) / (diameter of stenosis distal
   lumen)]. The percent change in vessel wall thickness was calculated as
   follows: 100 × [(diameter of the vessel wall at stenosis) − (diameter
   of stenosis distal vessel wall) / (diameter of the vessel wall at
   stenosis)]. Vascular landmarks were used to match the MR images.

EdU Incorporation Assay

   VSMCs were cultured in a 96‐well microplate overnight, followed by
   serum starvation for 24 hours. Then starved VSMCs were treated with
   conditioned medium. After 24 h, VSMC proliferation was analyzed via a
   BeyoClick EdU Cell Proliferation Kit with Alexa Fluor 555 (Beyotime,
   China) according to the manufacturer's instructions. Briefly, cells
   were incubated with EdU (10 µM) for 2 h at 37 °C and subsequently fixed
   with 4% paraformaldehyde. Then, the cells were incubated with the
   permeable solution and the Click Reaction Solution containing Azide 555
   at room temperature. Cell nuclei were stained with Hoechst 33 342 dye.
   Images of the stained cells were taken and recorded by an Operetta CLS
   High Content Analysis System (Perkin Elmer).

BrdU Incorporation Assay

   VSMCs were cultured overnight in 96‐well microplate and subsequently
   serum‐starved for 24 hours. The starved VSMCs were treated with
   conditioned medium from BMDMs. After 24 h, VSMCs proliferation was
   assessed using a BrdU cell proliferation kit (Roche, Basel,
   Switzerland) following the manufacturer's instructions. Briefly, the
   cells were incubated with BrdU for 2 h, then fixed, and their DNA
   denatured and labeled with a peroxidase‐conjugated BrdU antibody. After
   incubation with the peroxidase substrate and acid stop solution,
   absorbance values representing cell proliferation were measured by a
   microplate reader at 450 nm (OD450) with reference measurement at
   490 nm (OD490).

Migration Assay

   The migration of VSMCs was evaluated via the Transwell chamber assay.
   After serum starvation for 24 h, VSMCs (1 × 10^5 per well) were added
   to the upper chamber (8 µm pore size, Corning) in 200 µL of serum‐free
   medium. The lower chamber was filled with cultured BMDMs stimulated
   with LPS (100 ng mL^−1) or IL‐4 (10 ng mL^−1). After 10 h of
   incubation, cells were fixed with 4% paraformaldehyde for 20 min and
   stained with 0.1% crystal violet for 10 min. The non‐migrating VSMCs on
   the upper chamber of the filter were removed with a cotton bud. The
   cells on the underside were imaged and counted under a phase contrast
   microscope for quantification.

Quantitative Real‐Time PCR

   Total RNA was extracted from cells and tissues via TRIzol reagent
   (T9424, Sigma‐Aldrich) and reverse‐transcribed into complementary DNA
   by Transcriptor HiScript III RT SuperMix for qPCR (+gDNA wiper)
   (Vazyme). Then, SYBR Green PCR Master Mix (0 488 735 2001, Roche) and a
   LightCycler 480 Real‐time PCR System (Roche) were used to quantify the
   mRNA levels of the target genes in accordance with the manufacturer's
   protocol. The relative expression levels of the target genes were
   calculated and normalized to GAPDH mRNA expression. The PCR primer
   pairs used in this study are listed in Table [255]S3 (Supporting
   Information).

Western Blotting

   Total proteins from tissue or cell samples were lysed using RIPA assay
   buffer (65 mM Tris‐HCl pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% NP‐40, 0.5%
   sodium deoxycholate, 0.1% SDS, 13 protease inhibitor cocktail and 13
   phosphatase inhibitor PhosStop) and quantified with a BCA Protein Assay
   Kit (Thermo Fisher Scientific). Protein samples were separated using
   10% SDS–PAGE gels and then transferred to polyvinylidene difluoride
   (PVDF) membranes (Millipore). After blocking with 5% skim milk for 1 h
   at room temperature, the membranes were incubated with the appropriate
   primary antibodies (Table [256]S2, Supporting Information) overnight at
   4 °C. Then, the membrane was incubated with secondary horseradish
   peroxidase (HRP)‐conjugated antibodies for 1 h at room temperature and
   visualized via a FluorChem E imager (ProteinSimple, FluorChem E). GAPDH
   was used as the loading control.

Pull‐Down Assay

   BMDMs from mice were incubated with biotin, artesunate, and
   artesunate‐biotin probes for 2 h, respectively. The cells were lysed
   using RIPA lysis buffer. After centrifugation at 12 000 rpm for 15
   mins, the supernatant was incubated with BeaverBeads Streptavidin
   (Beaver, China) on a rotary mixer at room temperature for 1 h. The
   mixture was then separated using a magnetic separator, and the
   supernatant was transferred to a new centrifuge tube. Subsequently,
   1 mL of PBS (containing 0.05% Tween‐20) was added, followed by magnetic
   separation and removal of the supernatant. The washing process was
   repeated five times, and 100 µL of 1 × SDS‐PAGE Loading Buffer was
   added to elute the sample proteins. The expression levels of the
   collected proteins were detected by Western blotting assay.

Chromatin Immunoprecipitation (ChIP)

   ChIP was performed via a ChIP Assay Kit (P2078, Beyotime) according to
   the manufacturer's instructions. Briefly, macrophages were stimulated
   with LPS for 4 h, followed by fixation with 1% formaldehyde. After
   washing with ice‐cold PBS (1 mM PMSF), cells were resuspended in SDS
   lysis buffer. Then cell lysates were sonicated at low power (28 cycles)
   by a Bioruptor Sonication Systemat. The samples were precleared with
   Protein A/G Agarose and then incubated with goat‐anti‐IRF4 (sc‐6059,
   1 mg mL^−1, goat, Santa Cruz Biotechnology) or normal goat IgG (sc2028,
   0.2 mg mL^−1, goat, Santa Cruz Biotechnology) overnight. After
   incubation with Protein A/G Agarose for 2 h, the antibody/protein/DNA
   complexes were washed with a series of wash buffers. The DNA‐protein
   complexes and input control were eluted with elution buffer (1% SDS and
   0.1 M NaHCO[3]) and de‐crosslinked by 0.2 M NaCl incubation. Then, the
   samples were treated with proteinase K, and the DNA fragments were
   purified using a FastPure Gel DNA Extraction Mini Kit (Vazyme).
   mIL12‐pro84 was used as a positive control, and P6 was used as a
   negative control. The primers for ChIP‐PCR were as follows: P1,
   5‐CTTTCTTTCTTGACACCACAATGGG‐3 (forward) and
   5‐AGGTGTTCTTTTTGTTGTTCCTTTG‐3 (reverse); P2, 5‐CTTTGACAGACTGACTGGGC‐3
   (forward) and 5‐ATAGGTCCCAAGGATTCCG‐3 (reverse); P3,
   5‐TCTGTGGTCGGCCAGAAGT‐3 (forward) and 5‐GTGAAACGACAGAGGACGTGG‐3
   (reverse); P4, 5‐TCACCTCCCAGTGAAGTCCC‐3 (forward) and
   5‐GACCTTTACCTCGCCGCTTC‐3 (reverse); P5, 5‐ATGACTACGTGCCCAATCCC‐3
   (forward) and 5‐CGGAACCCCTTCTCCTAGCTT‐3 (reverse); P6,
   5‐TGGCGGACATCAATGACGTG‐3 (forward) and 5‐CTGAAGGCGTCTCTGACCAC‐3
   (reverse).

Dual‐Luciferase Reporter Assay

   Screening of the potential compounds that could affect IRF4 expression
   was performed by dual‐luciferase reporter assay. A total of 3000+
   compounds from the FDA‐approved Drug Library (L1300, Selleck Chemicals)
   were used for high‐throughput screening. HEK 293T cells were seeded in
   a 10‐cm Petri dish (5 × 10^6 per dish) and co‐transfected with a
   firefly luciferase reporter plasmid encoding the IRF4 promoter (4 µg)
   and a Renilla luciferase reporter plasmid (4 µg). After 24 h of
   transfection, the cells were seeded in 96‐well plates (8 × 10^4 per
   well) and an individual library compound (10 µM) was added to each
   well. Then the cells were lysed with Passive Lysis Buffer (Promega)
   after 12 h of incubation. The supernatant was then collected from the
   cellar lysate for the luciferase activity assay. The firefly and
   Renilla luciferase activities were measured by a Dual‐Luciferase
   Reporter Assay System (Promega) according to the manufacturer's
   instructions. Firefly luciferase activity was normalized to the
   activity of Renilla for each sample. The inhibitory ability of the
   screening compounds was presented and ranked by log2‐fold change
   normalized to the control. All experiments were performed in
   triplicate. Finally, the corresponding compounds that inhibited IRF4
   luciferase reporter activity more than twofold are shown in
   Figure [257]6A.

Luciferase Reporter Assays

   By amplifying the IRF4 gene‐coding region with the primers
   5‐CGGAATTCATGAACTTGGAGACGGGCAGC‐3 (forward) and
   5‐CCGCTCGAGTCACTCTTGGATGGAAGAAT‐3 (reverse) from the cDNA library, the
   plasmid psi‐flag‐IRF4 was generated after cloning the amplicon into the
   psi‐flag plasmid between the EcoRI and XhoI sites. The luciferase
   reporter plasmid pGL3‐pmKLF4 was generated by amplification of the KLF4
   promoter from the C57BL/6 genome ([258]NC_000076.6) with the primers
   5‐CTAGTCTAGATATGCTGGAGCATCGACCC‐3 (forward) and
   5‐GGAAGATCTCGCGCTACGATCACTCCG‐3 (reverse) and ligation into the
   pGL3‐basic vector (Promega, Madison) between the NheI and BglII sites.
   The potential IRF4 binding sites in the KLF4 promoter were predicted
   using Genomatix ([259]http://www.genomatix.de/) online software. The
   ISRE core sequence deletion plasmid pGL3‐pmKFL4‐del was generated using
   the fusion‐PCR approach described earlier with the primers
   5‐CCTCTTCCCAGCGCAGGAACCTCTTTATGACTCACTTCTC‐3 and
   5‐GAGAAGTGAGTCATAAAGAGGTTCCTGCGCTGGGAAGAGG‐3. Primary macrophages were
   co‐transfected with 2 µg of psi‐flag‐IRF4, 100 ng of pGL3‐pmKFL4 or
   pGL3‐pmKFL4‐del and 1 ng of pRL‐TK‐Luc for 24 h. The cells were treated
   with LPS for 4 h and then harvested and lysed with 100 µL of passive
   lysis buffer (Promega). After removing the cell debris by
   centrifugation, the supernatant was analyzed using a Single‐Mode
   SpectraMax Microplate Reader in accordance with the manufacturer's
   instructions.

RNA‐Seq Data Analysis

   For the RNA‐Seq assay, total RNA was isolated by TRI Reagent (T9424,
   Sigma‐Aldrich). The quality of the extracted total RNA was assessed
   using an RNA 6000 Nano kit (5067‐1511, Agilent, CA, USA). RNA‐seq
   libraries were prepared by an MGIEasy RNA Library Prep Kit
   (1 000 006 384, MGI Tech Co., Ltd, Shenzhen, China) according to the
   manufacturer's instructions. For gene expression analysis, samples were
   sequenced on a BGISEQ‐500 (MGI Tech Co.) with a single‐end 50 bp
   module. The clean reads were aligned using HISAT2 software (version
   2.21) with Ensembl human (hg38/GRCh38) reference genomes. The mapped
   fragments were stored and converted to the BAM format by SAMtools
   (version 1.4). The read counts of each identified gene were quantified
   by StringTie (version 1.3.3b) and were then used as input to calculate
   differential gene expression and statistical significance in DESeq2
   (version 3.5). In pairwise comparisons, DEGs were defined as genes with
   fold changes of more than 1.5 and an adjusted p‐value of less than
   0.05.

Enrichment Analysis

   Based on the DEGs, Gene Ontology (GO) biological process term
   enrichment analysis was performed using the Metascape online tool and
   clusterProfiler R package (version 3.12.0). Kyoto Encyclopedia of Genes
   and Genomes (KEGG) pathway enrichment analysis was performed by
   hypergeometric testing with the in‐house Perl script. Gene set
   enrichment analysis (GSEA) was performed on the Java GSEA platform
   (version 3.0) based on gene sets defined by the GO and KEGG results.
   The “Signal2Noise” metric was used to generate a ranked list and
   permutation type of the gene set. Gene sets with a p value of less than
   0.05 and a false discovery rate (FDR) value of less than 0.25 were
   considered statistically significant.

Public Data Analysis

   The bulk RNA‐seq and ChIP‐Seq datasets used in this study were
   collected and downloaded from GEO database. In the analysis of
   potential key regulators in vascular injury, three bulk RNA‐seq
   datasets in mice numbered [260]GSE40637 (control versus injured),
   [261]GSE119549 (control versus injured_6 h and control versus
   injured_24 h), and [262]GSE70410 (non‐ligated versus ligated) were
   collected. The RNA‐seq datasets related to macrophage polarization
   numbered [263]GSE86298, [264]GSE99296, [265]GSE106700, and
   [266]GSE158094 were analyzed. In addition, eleven ChIP‐Seq datasets
   were used for analysis. The Estimate the Proportion of Immune and
   Cancer Cells (EPIC) algorithm was applied to estimate the proportion of
   infiltrated cell types in each dataset. The DEGs were identified by the
   R software package “Limma”. Transcription factors were predicted by the
   iRegulon plugin of Cytoscape software (version 3.7.2).

Statistical Analysis

   All data were analyzed by SPSS software (version 19.0) and are
   presented either as the mean ± standard deviation (SD) or median ±
   interquartile range. For data with a normal distribution, a two‐tailed
   unpaired Student's t‐test was employed to compare differences between
   two groups, and one‐way ANOVA was applied for multiple comparisons,
   followed by Bonferroni's post hoc test (for data meeting homogeneity of
   variance) or Tamhane's T2 post hoc test (for data showing
   heteroscedasticity). For data showing a skewed distribution, the
   Mann‐Whitney U test was used to analyze differences between the two
   groups, and the Kruskal‐Wallis H test was applied for multiple
   comparisons. For each experiment, the sample size is indicated in the
   figure legends. Age‐ and weight‐matched male animals were randomly
   assigned to groups. Data from animal studies were collected in a
   blinded fashion. p values less than 0.05 were considered significant.

Conflict of Interest

   The authors declare no conflict of interest.

Author Contributions

   J.M., Y.Y., Z.Z., and K.Z. contributed equally to this work. P.Z. and
   J.C. conceptualized the study. J.M., Y.Y., Z.Z., K.Z., W.L., J.L., and
   S.Z. designed and performed experiments. W.L. and J.L. developed
   methodology. J.M., Y.Y., J.‐J.Q., H.S., and Y.W. analyzed and organized
   the data. X.F. and X.L. validated the data. S.C. and Z.‐G.S. provided
   resources. J.M., Y.Y., and K.Z. wrote the original draft. P.Z., J.C.,
   and Z.Z. reviewed and edited the manuscript.

Supporting information

   Supporting Information
   [267]ADVS-12-2408992-s001.docx^ (4.1MB, docx)

Acknowledegments

   The authors thank Drs Xiaojing Zhang and Xu Cheng for the valuable
   technological assistance. This work was supported by grants from the
   National Natural Science Foundation of China (NO. 92169211, NO.
   82000299 and NO. 81770053).

   Miao J., Yong Y., Zheng Z., Zhang K., Li W., Liu J., Zhou S., Qin J.,
   Sun H., Wang Y., Fu X., Luo X., Chen S., She Z.‐G., Cai J., Zhu P.,
   Artesunate Inhibits Neointimal Hyperplasia by Promoting IRF4 Associated
   Macrophage Polarization. Adv. Sci. 2025, 12, 2408992.
   10.1002/advs.202408992

Contributor Information

   Jingjing Cai, Email: caijingjing83@hotmail.com.

   Ping Zhu, Email: zhuping@fmmu.edu.cn.

Data Availability Statement

   The data that support the findings of this study are available from the
   corresponding author upon reasonable request.

References