ILF3 Mediated BMP2 and STAT1 Transcription Is Responsible for Hyperlipidemia-Induced Arteriosclerotic Calcication

Calcication is common in atherosclerotic plaque and can induce vulnerability, which further leads to myocardial infarction, plaque rupture and stroke. The mechanisms of atherosclerotic calcication are poorly characterized. Interleukin enhancer binding factor 3 (ILF3) has been identied as a novel factor affecting dyslipidemia and stroke subtypes. However, the precise role of ILF3 in atherosclerotic calcication remains unclear. Here we showed that ILF3 expression is increased in calcied human aortic vascular smooth muscle cells (HAVSMCs) and calcied atherosclerotic plaque in humans and mice. We then found that hyperlipidemia increases ILF3 expression and exacerbates calcication of VSMCs and macrophages by regulating bone morphogenetic protein 2 (BMP2) and signal transducer and activator of transcription 1 (STAT1) transcription. We further explored the molecular mechanisms of ILF3 in atherosclerotic calcication and revealed that ILF3 acts on the promoter regions of BMP2 and STAT1 and mediates BMP2 upregulation and STAT1 downregulation, which promotes atherosclerotic calcication. Our results demonstrate the effect of ILF3 in atherosclerotic calcication. Inhibition of ILF3 may be a useful therapy for preventing and even reversing atherosclerotic calcication. 3.0 false discovery rate (n=3 per group) represent signicantly different genes. and c Gene Ontology analysis of biological processes, molecular functions and cellular components for DEGs related to vascular disease and KEGG pathway analysis of calcication-related classical pathway analysis. The x-axis represents the signicance of different terms with negative log10 (P-values). The y-axis shows the name of related terms. ILF3-related diseases and function analysis and ILF3-related classical pathway analysis were transfected with si-ILF3 versus negative Heatmap shows detailed information of upregulated and downregulated genes for the top induced genes involved in osteoblastic differentiation. normalized transfected with negative control siNC or si-ILF3. RT-PCR mRNA


Introduction
Vascular calci cation is a common phenomenon in many physiological and pathological diseases including aging, end-stage renal disease, diabetes mellitus and cardiovascular diseases 1,2 . Vascular calci cation can occur in different locations of the vessel wall including intima and media but exists mainly in intimal layers in atherosclerosis and can induce atherosclerotic plaque susceptibility and further lead to myocardial infarction, plaque rupture and stroke 3 .
Recent studies suggested that vascular calci cation is an active cell regulatory process characterized by the involvement of various cells such as vascular smooth muscle cells (VSMCs), pericytes, myo broblasts, macrophages, vascular mesenchymal progenitors and endothelial cells 4,5,6,7 . Under multiplepro-calci cstimuli, VSMCs can undergo a phenotype switch from a contractile to osteoblastic phenotype accompanied by loss of contractile markers (smooth muscle 22  Msx and become the main source of osteoblastic cells, which leads to vascular calci cation. In addition, macrophages can undergo a phenotype shift and participate in atherosclerotic calci cation 5,8 . Because of the diversity and complexity of calci cation mechanisms, ideal drugs preventing or reversing atherosclerotic calci cation are unavailable. The underlying molecular mechanisms of atherosclerotic calci cation still need further study. Interleukin enhancer-binding factor 3 (ILF3), as a double-stranded RNA (dsRNA)-binding protein, combines with other proteins, mRNAs, small noncoding RNAs, and dsRNAs to regulate transcription, translation, mRNA stability and noncoding RNA biogenesis 9 . In the cardiovascular system, ILF3 can inhibit myocardial hypertrophy 10 . Also, the association between ILF3 and myocardial infarction is affected by low-density lipoprotein (LDL) and high-density lipoprotein (HDL) cholesterol metabolism, which indicates interactions between genes 11 . Recent studies have reported insights into the possible physiological roles of ILF3 in stroke, in ammation, and dyslipidaemia, but its role in vascular calci cation has not been reported.
In this study, we used human samples and murine models with conditional ILF3 knockout and overexpression in VSMCs and macrophages to explore the roles of ILF3 in atherosclerotic calci cation.

Materials And Methods
Human coronary artery samples Atherosclerotic and control epicardial coronary artery segments were from human specimens with extensive atherosclerotic disease and healthy controls. The specimens were donated by the Shandong Red Cross Society.The experiment protocols were examined and approved by the review committee of Animals ILF3 conditional transgenic and knockout mice (ILF3 f/f mice) were generated with use of CRISPR-Cas9.
Mice were weighed and organs were harvested and xed in 4% paraformaldehyde.

Serum index levels
We collected 0.5 to 1.0 ml of blood from the left ventricle of mice at the time of tissue harvesting after 16 weeks feeding with the HFD. Serum levels of triglycerides (TG), total cholesterol (TC), blood glucose (BG), low density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), calcium and phosphorus were measured by standard enzymatic methods with commercial kits (Roche Diagnostics, Indianapolis, IN).
Oil-red O staining andSirius red staining Brie y, frozen sections were hydrated and washed, then immersed in 0.3% Oil-red O or Sirius red stain. Then sections were counterstained with haematoxylin and visualized under a Nikon Eclipse 80 imicroscope with a camera (DS-Ri1; Nikon). Oil-red O stain was used to detect lipids and Sirius red to examine collagen in aortic root plaques. Atherosclerotic plaque instability index was computed by using the formula (Oil-red O area + MOMA-2 area)/(α-SMA area + collagen I area).

Cell culture
Human aortic vascular smooth muscle cells (HAVSMCs) were obtained from the ScienCell (Carlsbad, CA, US) and cultured in SMC medium (SMCM) including 2% fetal bovine serum (FBS), 1% SMC growth supplement (SMCGS) and 1% penicillin/streptomycin solution. Peritoneal macrophages were collected from ApoE −/− , ILF3 −/− /Lyz Cre and ILF3 over /Lyz Cre mice as described 8 . After being injected intraperitoneally with 6% starch for 3 days, mice were euthanized by spinal dislocation. Macrophages were collected from the peritoneal cavity by perfusion with cold phosphate-buffered saline (PBS, GibcoBRL Life Technologies). Then macrophages were grown in Dulbecco's modi ed Eagle's medium (DMEM, Gibco) supplemented with 10% FBS (Gibco). All cells were incubated at 37 °C in a humidi ed atmosphere and used after passages 3-5. Cells were starved for 24 h before each experiment. For in vitro experiments, we used osteogenic medium containing 10 mM β-glycerophosphate (β-GP, Sigma) to stimulate cell calci cation for 3-14 days. Cells were treated with oxidized LDL (oxLDL; 50 µg/ml) to simulate a high lipid condition.
Lentivirus (Lv) and siRNA transfection Lentiviruses and siRNA duplex were supplied by GenePharma (Suzhou, China). Lentiviruses encoding overexpressed ILF3 (Lv-ILF3) were incubated with VSMCs or macrophages at multiplicity of infection (MOI) of 10. After 24 h, the medium containing lentiviruses was removed and replaced with fresh medium. The siRNA duplex encoding ILF3 knockout cDNA (si-ILF3) was transfected into VSMCs or macrophages by using Lipofectamine 3000 Reagent Protocol (Invitrogen, NY, USA) in Opti-medium (Gibco) for 6 h and then replaced with fresh medium. The Homo sapiens siRNA sequence was 5'-CCUGUGUGAGAAAUCCAUU-3'. The lentivirus NCBI reference sequence is NO. NM_001137673.

RNA-sequencing (RNA-seq)
After HAVSMCs were transfected with ILF3 siRNA, RNAs were extracted with TRIzol reagent. We used the

Western blot analysis
Western blot analysis was conducted as previously described 8

RT-PCR analysis
Total RNA was extracted from cells with TRIzol reagent and reverse-transcribed to cDNA by using HiScriptIIIRT SuperMix for qPCR (Vazyme, Nanjing, China). Quantitative RT-PCR involved using the SYBR Green Master mix kit (Roche, USA). The average cycle threshold (Ct) method was used to determine mRNA expression. The 2-ΔΔCT method was used to calculate the relative change of mRNA. The primer sequences for ILF3, BMP2, STAT1 and β-actin are in supplemental Table S1.

Alizarin-red and von Kossa staining
For Alizarin-red staining, after xing in 4% paraformaldehyde, cells were immersed in 1% Alizarin-red solution (Solarbio, China) for 15 min. The sections of aortic root were depara nized and dyed with Alizarin-red for 5 min. Von Kossa staining involved using a kit (Solarbio). The cells were exposed to 5% silver nitrate solution and exposed to ultravioletray for 1 h.
Alkaline phosphatase (ALP) activity and calcium content detection ALP activity was detected by using an ALP assay kit (Beyotime, China) and normalized to total protein concentration by the BCA Protein Assay Kit (Beyotime, China). Calcium content was determined with the Calcium Assay Kit (Nanjing Jiancheng Bioengineering Institute, China) and normalized to protein content.

Chromatin immunoprecipitation assay (ChIP)
ChIP assay involved using a ChIP Assay kit (CST, USA). Nucleoprotein complexes were extracted from HAVSMCs. For immunoprecipitation, anti-ILF3 antibody (Abcam, Cat# ab131004), normal IgG antibody (CST, USA) and Histone H3 antibody (CST, USA) were used. Speci c primers targeting different DNA sites in the − 160 to + 40 bp fragment of BMP2 and the − 140 to + 50 bp fragment of STAT1 are in supplemental Table 1.

Statistical analysis
Data are presented as mean ± SEM from three replicate experiments. Student's t test and one-way ANOVA were used to evaluate differences between 2 groups and multiple groups, respectively, by using SPSS 18.0. Differences were considered signi cant at P < 0.05.

ILF3 is upregulated in calci ed atherosclerotic plaque
Mice body weight (BW) and serum levels of TC, TG, BG, HDL-C, LDL-C, calcium and phosphorus are in Supplemental Tables 2 and 3. Both VSMC-speci c and macrophage-speci c ILF3-knockout ApoE −/− ILF3 −/ − mice showed lower BW and TC, TG and LDL-C levels but higher HDL-C level than ApoE −/− mice (all P < 0.05). For all ILF3-overexpressed mice, BW and TC, TG and LDL-C levels were higher but HDL-C level was lower relative to ApoE −/− mice (*P < 0.05). BG, calcium and phosphorus levels did not differ between groups.
Some studies have reported that ILF3 is involved in dyslipidemia and participates in the occurrence of acute myocardial infarction 11 . To investigate the role of ILF3 in calci cation of atherosclerotic plaque, we assessed whether ILF3 is increased in calci ed HAVSMCs exposed to ox-LDL and in calci ed atherosclerotic plaque. Calci cation was induced in cultured VSMCs by treatment with osteogenic medium with or without ox-LDL for 14 days. Alizarin-red and von Kossa staining were used to nd signi cantly increased calcium nodules in VSMCs exposed to ox-LDL (Fig. 1a). Ox-LDL-stimulated calci ed VSMCs showed increased protein and mRNA levels of ILF3 as compared with controls ( Fig. 1b and c). In human coronary atherosclerotic plaque, calcium nodules and ILF3 were increased in calci ed plaque as compared with healthy controls (Fig. 1d). In addition, ILF3 expression was signi cantly increased in atherosclerotic plaque calci cation inApoE −/− mice fed an HFD for 16 weeks (Fig. 1e). These results suggest that ILF3 plays an important role in atherosclerotic calci cation.

ILF3 participates in regulating calci cation gene transcription in HAVSMCs
To investigate the mechanisms of ILF3 in atherosclerosis calci cation, we analyzed transcriptomic pro les in human wild-type and ILF3-de cient HAVSMCs. HAVSMCs transfected with siRNA deleting ILF3 (si-ILF3) and normal control siRNA (NC) were subjected to RNA-sEq. On differential expression analysis, inhibition of ILF3 resulted in 684 genes signi cantly downregulated and 940 genes signi cantly upregulated in HAVSMCs as compared with normal controls (Fig. 2a). High enrichment of these differentially expressed genes was associated with GO terms for processes involved in cardiovascular systemic diseases, such as movement of endothelial cells, arterial aneurysm, aortic dilatation, apoptosis and death of cardiomyocytes and vascular lesion (Fig. 2b). KEGG pathway analysis revealed that the signi cantly differentially expressed genes may affect the activation of calci cation-related signaling pathways including transforming growth factor-beta (TGF-β) signaling, STAT3 pathway, BMP2 signaling pathway, role of osteoblasts, osteoclasts and chondrocytes in rheumatoid arthritis and the Janus kinase (JAK)/STAT signaling pathway (Fig. 2c).
From the RNA-seq analyses and in vivo ndings, we focused on the targets of calci cation. The genes involved in calci cation were selected to validate the dysregulated expression of mRNAs with Reads Per Kilobase of transcript, per Million mapped reads (RPKM) values > 50 identi ed by RNA-sEq. To further investigate the roles of ILF3 in calci cation-related genes in HAVSMCs, we used hierarchical clustering based on ILF3 silence-induced change in levels of calci cation-related genes. Calci cation-associated genes, such as BMP2, STAT1, secreted phosphoprotein 1 (SPP1), interleukin-1 beta (IL-1β), Tissue inhibitor of metalloproteinases 2 (TIMP2) and LDL-receptor-related protein 5 (LRP5) were dysregulated (Fig. 2d). We used quantitative real-time PCR (qRT-PCR) and immunoblotting to con rm the target genes regulating calci cation in ILF3-silenced HAVSMCs. Silencing ILF3 in HAVSMCs led to markedly decreased BMP2 protein level but increased STAT1 level, with the reverse for mRNA levels ( Fig. 2e and f).

ILF3 promotes atherosclerotic calci cation by regulating BMP2 and STAT1 gene transcription in VSMCs
To test the roles of ILF3 in atherosclerotic calci cation, we rst investigated the effects of ILF3 knockdown and overexpression on atherosclerotic calci cation in VSMCs. ILF3 overexpression and knockdown were achieved by transduction of lentivirus (Lv-ILF3) and si-ILF3 duplex in VSMCs, respectively. Calcium deposition was estimated incultured VSMCs by using Alizarin-red and von Kossa staining. Ox-LDL stimulation increased calcium deposition in VSMCs as compared with controls, and ILF3 knockout abolished the ox-LDL effect. In addition, Lv-ILF3 overexpression induced calci cation under hyperlipemia (Fig. 3a). Detection of calcium content and ALP activity showed similar trends in cultured VSMCs (Fig. 3b and 3c). Western blot analysis was used to con rm the levels of osteogenic markers including BMP2, Runx2 and STAT1 in calci ed VSMCs. ILF3 knockdown reversed the ox-LDLincreased BMP2 and Runx2 levels and ox-LDL-decreased STAT1 level. In contrast, ILF3 overexpression increased BMP2 and Runx2 levels and reduced STAT1 level relative to ox-LDL alone (Fig. 3d). A BMP2induced Smads complex is transported from cytoplasm to nucleus to increase the expression of osteogenic gene Runx2 12,13 . Hence, we examined the change in p-smad1/5 level. ILF3 inhibition abolished the ox-LDL-induced Smad1/5 phosphorylation (Fig. 3e). In contrast, enhanced ILF3 expression increased Smad1/5 phosphorylation as compared with ox-LDL alone (Fig. 3e). In addition, Runx2 nuclear translocation is crucial for calci cation of VSMCs 14,15 . Immuno uorescence staining revealed that ox-LDL enhanced Runx2 nuclear localization but also inhibited STAT1 expression. ILF3 silencing reduced the Runx2 nuclear translocation response to ox-LDL and reversed STAT1 activity, and ILF3 overexpression had the opposite result (Fig. 3f).
ILF3 promotes atherosclerotic calci cation via a VSMC phenotypic switch A VSMC phenotypic switch from a contractile to synthetic phenotype is crucial for VSMC calci cation 6 . To verify whether ILF3 enhances atherosclerotic calci cation via promoting a VSMC phenotypic switch, primary HAVSMCs were cultured and treated with ox-LDL. Immunoblotting assay showed that ox-LDL induced a VSMC phenotypic switch from a contractile to synthetic phenotype by increasing levels of OPN and Vimentin (markers of VSMC synthetic phenotype) but decreasing α-SMA level (marker of VSMC contractile phenotype). ILF3 silencing signi cantly alleviated the VSMC phenotypic switch induced by ox-LDL (Fig. 4a). Conversely, the role of ox-LDL in the VSMC phenotypetransition was further boosted by ILF3 overexpression. Furthermore, immuno uorescence and immunohistochemical staining revealed that ApoE −/− ILF3 −/− /SMA Cre mice showed lower OPN and Vimentin levels but higher α-SMA level as compared with ApoE −/− mice. In ApoE −/− ILF3 over /SMA Cre mice, OPN and Vimentin levels were increased, but a-SMA level was downregulated (Fig. 4b and c). These results suggest that ILF3 promotes atherosclerotic calci cation via a VSMC phenotypic switch.

ILF3 promotes atherosclerotic calci cation by enhancing macrophage calci cation and macrophage polarization
Previous study reported the role of macrophages in atherosclerotic calci cation and osteogenic differentiation 8 . Next, we investigated whether ILF3 accelerates macrophage calci cation in atherosclerosis. Peritoneal macrophages isolated from wild-type (WT), ILF3 −/− /Lyz Cre and ILF3 over /Lyz Cre mice were treated with or without ox-LDL. Alizarin-red and von Kossa staining showed that the knockdown of ILF3 signi cantly reduced calcium nodule formation in macrophages treated with ox-LDL (Fig. 5a), which was accompanied by reduced ALP activity and calcium content (Fig. 5b and c). As compared with WT macrophages incubated with ox-LDL, in ILF3-overexpressed macrophages, calcium deposition was more severe and ALP activity and calcium content were higher (Fig. 5a-c). Also, in ILF3 −/− macrophages, the ox-LDL-induced levels of osteogenic markers BMP2 and Runx2 was markedly suppressed, but STAT1 level was concomitantly increased; ILF3 overexpression increased BMP2 and Runx2 levels and decreased STAT1 level as compared with ox-LDL alone (Fig. 5d). We also detected the effect of ILF3 on p-smad1/5 and Runx2 nuclear translocation in macrophages under hyperlipidemia. As in VSMCs, ILF3 deletion attenuated ox-LDL-induced phosphorylation of Smad1/5, but ILF3 overexpression induced Smad1/5 phosphorylation versus ox-LDL alone (Fig. 5e). Macrophagic ILF3 knockout led to decreased Runx2 nuclear translocation with ox-LDL induction and increased STAT1 level (Fig. 5f). Overexpressed ILF3 increased Runx2 nuclear translocation as compared with ox-LDL alone, with a signi cant decrease in STAT1 level.
To better discern whether ILF3 expedites macrophage calci cation in atherosclerotic plaque, ApoE −/− , ApoE −/− ILF3 −/− /Lyz Cre and ApoE −/− ILF3 over /Lyz Cre mice were fed an HFD for 16 weeks. In atherosclerotic lesions, ILF3 deletion in macrophages resulted in fewer calcium nodules together with lower expression of osteogenic markers BMP2 and Runx2 and higher STAT1 level relative to ApoE −/− mice (Fig. 5g and h). ILF3 overexpression in macrophages resulted in more serious calci cation, higher BMP2 and Runx2 expression and reduced STAT1 level relative to ApoE-/-mice ( Fig. 5g and h). Additionally, we tested the effect of macrophages on VSMC calci cation. Cultured medium from ILF3-de cient macrophages was used to incubate VSMCs. VSMCs showed weaker calci cation and less Runx2 level as compared with WT macrophages under ox-LDL treatment (Fig. 5i and j). Thus, ILF3 promoted macrophage calci cation and participated in atherosclerotic calci cation.
We next assessed whether ILF3 is involved in macrophage polarization in atherosclerotic calci cation. In vitro, ox-LDL treatment increased the activity of iNOS (an M1 macrophage marker) but decreased the expression of ARG1 (an M2 macrophage marker) in WT peritoneal macrophages (Fig. 6a). However, ILF3 deletion impeded the ox-LDL-induced expression of iNOS and reversed ARG1 level. Also, ILF3 overexpression accelerated ox-LDL-induced expression of iNOS but reduced ARG1 level.
To better reveal the roles of ILF3 in macrophage polarization of atherosclerotic calci cation, we evaluated the expression of iNOS and ARG1 in atherosclerotic lesions by immuno uorescence and immunohistochemistry analysis in ApoE −/− , ApoE −/− ILF3 −/− /Lyz Cre and ApoE −/− ILF3 over /Lyz Cre mice. iNOS level was decreased, and ARG1 level was markedly increased in ApoE −/− ILF3 −/− /Lyz Cre mice relative to ApoE −/− mice ( Fig. 6b and c). Conversely, the expression of iNOS was enhanced and that of ARG1 was reduced in ApoE −/− ILF3 over /Lyz Cre versus ApoE −/− mice ( Fig. 6b and c). These results indicate that ILF3 promotes macrophages to the in ammatory M1 macrophagic phenotype in atherosclerotic calci cation.
ILF3 directly binds to the BMP2 and STAT1 promoters to accelerate arteriosclerotic calci cation Studies have demonstrated that ILF3 predominantly locates in the nucleus and plays an important role in regulating transcription in mammalian cells 16,17 either negatively or positively 18 . From the mRNA-seq results, ILF3 may promote arteriosclerotic calci cation by regulating the transcription of BMP2and STAT1. To verify this hypothesis, we constructed pGL3-BMP2-Luc reporter plasmids with progressively deleted 5'-ankingregions from the − 1360-to + 597-bp region and pGL3-STAT1-Luc reporter plasmids containing progressively deleted 5'-anking regions from the − 526-to + 246-bp region.
To con rm that ILF3 is the transcription factor of BMP2, pGL3-BMP2-Luc reporter plasmids with progressive deletions were transfected into HEK293T cells along with normal control siRNA (si-NC) or si-ILF3. The luciferase activities of the si-ILF3 group were weakened as compared with the si-NC group, which suggests that ILF3 binds to the BMP2 promoter region (Fig. 7a). In addition, luciferase activity was signi cantly reduced in the upstream region + 41 to + 597 bp of the BMP2 promoter, which suggests an ILF3 binding site at -160 to + 40 bp (Fig. 7a). To further verify whether ILF3 binds to the − 160-to + 40-bp region of BMP2, ChIP assay was used with speci c primers covering the promoter region − 160 to + 40 bp of BMP2. qRT-PCR results with speci c primers revealed signi cantly elevated DNA levels that speci cally bound to ILF3 as compared with negative IgG control but less to the positive H3 control (Fig. 7b and c). To further discover the binding sequences of BMP2 promoters with ILF3, we used HADDOCK software to analyze the spatial structure binding domains of the BMP2 promoter with ILF3. Figure 7d revealed that ILF3 might bind to the AGGGAG site (-53-to -48-bp fragment) of the BMP2 promoter. With the same methods, ILF3 could simultaneously bind to the GCGCCC site (-28-to -23-bp fragment) of the STAT1 promoter and regulate STAT1 transcription (Fig. 7e, f, g and h). These ndings suggest that ILF3 can increase BMP2 level but decrease STAT1 level by binding to BMP2 and STAT1 promoter regions and promote arteriosclerotic calci cation.
In sum, our ndings illustrate the working schematic that hyperlipidemia-induced ILF3 activation mediates acceleration of atherosclerotic calci cation (Fig. 8). Hyperlipidemia-increased ILF3 expression mediates BMP2 and STAT1 transcription by directly binding to their promoter regions. ILF3 upregulates BMP2 level and activates Smads signaling to elevate Runx2 transcription. Meanwhile, ILF3 suppresses STAT1 transcription, which promotes Runx2 nuclear translocation and regulates osteogenic differentiation. In addition, ILF3-mediated hyperlipidemia induces a phenotypic switch of VSMCs from contractile to a dedifferentiated synthetic phenotype and macrophages to a pro-in ammatory M1 phenotype, which in turn aggravates VSMC calci cation to promote atherosclerotic calci cation.

Discussion
ILF3 has been veri ed to play important roles in dyslipidemia and the cardiovascular system 10,11 . However, whether ILF3 is linked to dyslipidemia-induced atherosclerotic calci cation has not been reported. In the present study, we used ILF3 conditional genetic deletion and transgenic mouse models to investigate the role of ILF3 in atherosclerotic calci cation. Hyperlipidemia could augment ILF3 expression in calci ed VSMCs and macrophages and in atherosclerotic calci cation in humans and mice. Inhibition of ILF3 blocked osteogenic differentiation in both VSMCs and macrophages under dyslipidemia. The underlying mechanisms may involve ILF3 regulating the transcription of osteogenic markers BMP2 and STAT1.
Genetic lineage-tracing studies revealed that most of the chondrocyte-like cells (98%) originate from VSMCs in atherosclerotic lesion calci cation 19 . We found that knockout of ILF3 blocked osteogenic differentiation and calcium deposition in VSMCs under HFD feeding. Conversely, overexpressed ILF3 promoted atherosclerotic calci cation by inducing VSMC calci cation. Besides VSMCs, macrophages are involved in atherosclerotic calci cation. Previous studies found that in ltrated macrophages during atherogenesis induce VSMC transdifferentiation into an osteochondrogenic phenotype by producing pro-in ammatory cytokines and regulatory molecules, which contribute to mineral deposition in plaques 20,21 . Co-culture studies of macrophages and VSMCs in vitro suggested that VSMC-induced Runx2 expression promotes macrophage migration and formation of osteoclast-like cells 22 . In this study, ILF3 promoted macrophage calci cation in atherosclerotic calci cation and isolated macrophages under dyslipidemia. Also, medium from cultured ILF3 −/− macrophages incubated with VSMCs eliminated calci ed nodules and a decrease in Runx2 expression. Our results suggest that ILF3 plays an important role in driving arteriosclerosis calci cation by accelerating the osteogenic differentiation of VSMCs and macrophages.
Increasing evidence suggests that vascular calci cation is an active cell-regulatory process and Runx2 is indispensable for calci cation occurrence and development 4,23,24 . Here, we provide evidence that hyperlipidemia promotes Runx2 expression in calci ed VSMCs and macrophages and atherosclerotic calci cation. ILF3 deletion decreased hyperlipidemia-induced Runx2 protein content, but ILF3 overexpression increased Runx2 level concomitant with increased ALP activity.
VSMCs are non-terminally differentiated and show phenotypic plasticity, which is considered a major contribution to vascular calci cation 6 . Also, a VSMC osteogenic phenotype switch facilitates intimal calci cation in atherosclerotic plaque 19 . In this study, ILF3 induced a VSMC phenotype switch from a contractile to synthetic phenotype with an increase in levels of synthetic markers OPN and Vimentin and decrease in level of the contractile marker α-SMA.
The role and mechanism of macrophages in atherosclerotic calci cation are still unclear. Studies reported that the pro-in ammatory M1 phenotype of macrophages is activated during atherosclerotic calcium deposition, and the pro-in ammatory atherogenic phenotype of macrophages was signi cantly correlated with atherosclerotic calci cation 5,25 . In our studies, hyperlipidemia activated macrophagic ILF3 and increased the level of the M1 marker iNOS and weakened that of the M2 marker ARG-1 in macrophages, which indicates that ILF3 aggravated a shift to a pro-in ammatory phenotype.
Hyperlipidemia upregulated ILF3, increased Runx2 activity and promoted atherosclerotic calci cation in the M1 macrophage phenotype. Moreover, medium from cultured ILF3 −/− macrophages incubated with VSMCs eliminated calci ed nodules and decreased Runx2 expression, so cytokines from M1 macrophages signi cantly promoted the calci cation of VSMCs.
Studies indicated that BMP2 and STAT1 play important roles in vascular calci cation by regulating Runx2 expression 15,26 . BMP2 induces Smad proteins phosphorylation, and activated Smads complex are transported from cytoplasm to the nucleus to increase the expression of the osteogenic gene Runx2 12,13 .
Our data showed that ILF3 increased BMP2 level and Smad1/5 phosphorylation and further upregulated Runx2 expression, so ILF3 is involved in arteriosclerotic calci cation by modulating BMP2/Smads signaling. STAT1 has a negative regulatory role in Runx2-mediated osteoblast differentiation 27 . STAT1 suppresses osteoblast differentiation by interfering with Runx2 nuclear localization and transcriptional activity 15,28,29 . In our studies, ILF3 regulated Runx2 nuclear translocation by suppressing STAT1 activity in VSMCs and macrophages under hyperlipidemia. In the hyperlipidemic condition, nuclear Runx2 level was increased but that of STAT1 was decreased. Inhibition of ILF3 induced STAT1 expression and Runx2 cytoplasmic translocation as compared with the hyperlipidemic condition. Furthermore, ILF3 overexpression enhanced Runx2 level in the nucleus but reduced STAT1 level under the hyperlipidemic condition. These ndings demonstrate that one of the targets for ILF3 to promote atherosclerotic calci cation is mediated by modulation of STAT1 transcription and thusRunx2 nuclear translocation.
Some studies have found that ILF3 can regulate transcription as a transcriptional activator 30 . In addition, DNA a nity chromatography and subsequent ChIP assays con rmed that NF90/NF110 can associate with gene regulation by binding DNA 31,32,33,34,35 . In the current study, mRNA-seq demonstrated that ILF3 genetic de ciency in VSMCs decreased BMP2 mRNA level but enhanced STAT1 level. Furthermore, we veri ed that ILF3 could enhance BMP2 and weaken STAT1 transcription by directly binding to their promoter regions.
In conclusion, we report the roles of ILF3 in the osteogenic switch of VSMCs and macrophages by regulating BMP2 and STAT1 transcription. ILF3 could mediate upregulation of BMP2 and suppression of STAT1 expression under hyperlipidemia to promote atherosclerotic calci cation, which augmented the risk of atherosclerotic lesion instability. Inhibition of ILF3 may be a potential therapeutic target for preventing atherosclerotic calci cation and lesion rupture.

Con ict of Interest
The authors declare that they have no con ict of interest. Figure 1 ILF3 is upregulated in hyperlipidemia-induced calci ed vascular smooth muscle cells (VSMCs) and atherosclerotic plaque. a HAVSMCs were incubated in osteogenic medium with or without ox-LDL (50μg/ml) for 21 days. Calci cation was detected by Alizarin-red and Von Kossa staining (n =5 per group).*P<0.05, vs. control in osteogenic medium without ox-LDL. b, c Immunoblots and qRT-PCR analyses of the effect of ox-LDL on expression in HAVSMCs (n=5 per group). *P<0.05, vs. control. D Representative immunohistochemical staining of ILF3 and Alizarin-red staining in arteries of healthy control and atherosclerosis mice (n=6 per group, Scale bar: 100μm).*P<0.05, vs. healthy control. e Representative immunohistochemical staining of ILF3 and Alizarin-red staining in atherosclerotic lesions of ApoE-/-mice fed a normal diet (ND) or high fat diet (HFD) for 16 weeks (n=10 per group, Scale bar: 50μm). *P<0.05, vs. ApoE-/-mice fed a normal diet. Data are mean ± SEM (n=5 per group).

Figure 2
ILF3 participates in regulating transcription of osteogenic-related genes. a Volcano plot of genome-wide transcriptomic analysis shows the differentially expressed genes (DEGs) between normal control (NC) and siRNA-deleted ILF3 (si-ILF3)HAVSMCs. Red points were screened according to the standard of |fold change|≥ 3.0 and false discovery rate <0.05 (n=3 per group) and represent signi cantly different genes. b and c Gene Ontology analysis of biological processes, molecular functions and cellular components for DEGs related to vascular disease and KEGG pathway analysis of calci cation-related classical pathway analysis. The x-axis represents the signi cance of different terms with negative log10 (P-values). The yaxis shows the name of related terms. ILF3-related diseases and function analysis (b) and ILF3-related classical pathway analysis (c). d HAVSMCs were transfected with si-ILF3 versus negative control (NC).
Heatmap shows detailed information of upregulated and downregulated genes for the top induced genes involved in osteoblastic differentiation. Color scale is based on normalized read counts. e HAVSMCs were transfected with negative control siNC or si-ILF3. RT-PCR shows BMP2 and STAT1 mRNA levels in HAVSMCs (n=3 per group). f Representative immunoblot assay of ILF3, BMP2 and STAT1 protein levels in HAVSMCs (n=3 per group) transfected with siNC or si-ILF3. Data are mean ± SEM. *P<0.05 vs NC.  Western blot analysis of BMP2, STAT1 and Runx2 protein levels (d) and p-smad1/5 and Smad1 levels (e) and quanti cation. f Immuno uorescence double staining for Runx2 (green) and STAT1(red) in ILF3 accelerates atherosclerotic calci cation by inducing macrophage calci cation. a-f Primary macrophages from wild type (WT), ILF3-/-/LyzCre and ILF3over/LyzCre mice were incubated with osteogenic medium and with or without ox-LDL (n=3 per group). a Calci cation were evaluated by Alizarin-red or von Kossa staining after treatment with ox-LDL for 14 days. b Quanti cation of calcium content. c Measurement of ALP activity. Western blot analysis of BMP2, STAT1 and Runx2 protein levels (d) in macrophages and p-smad1/5 and Smad1 levels (e) and quanti cation. f Immuno uorescence ILF3over/LyzCre mice fed a HFD for 16 weeks (n=10 per group, Scale bar: 50μm). c Representative immunohistochemical staining of iNOS and ARG1 (n=10 per group, Scale bar: 50μm). *P<0.05, vs. ApoE-/-. Data are mean ± SEM. Figure 7 ILF3 binds to the BMP2 and STAT1 promoter regions and contributes to arteriosclerotic calci cation. a Relative luciferase activity assay of HEK293T cells after transfection with pGL3-promoter constructs containing DNA fragments serially deleted from -1160 to +597 bp of the promoter BMP2 (n=5 per group). *P<0.05, vs. transfection with negative control siRNA (siNC) or pGL-160. b ChIP assay to verify binding of ILF3 to the BMP2 promoter (n=3 per group). c qPCR analysis of DNA level of BMP2 promoter containing -160 to +40 bp (n=3 per group). *P<0.05, vs. IgG and #P<0.05, vs. Histone H3. d HADDOCK software was used to evaluate the spatial structure of ILF3-BMP2 promoter regions, and ILF3 binding sites were identi ed as the AGGGAG site (-53 to -48 bp) in the BMP2 promoter. e Luciferase activity assay after transfection with the STAT1 promoter serially deleted from -526 to +246 bp in HEK293T cells (n=5 per group). *P<0.05, vs. transfection with siNC or pGL-140. f ChIP assay to con rm binding of ILF3 to the STAT1 promoter (n=3 per group). g qPCR of DNA level of STAT1 promoter containing -140 to +50 bp (n=3 per group). *P<0.05, vs. IgG and #P<0.05, vs. Histone H3. h HADDOCK software analysis to evaluate spatial structure of ILF3-STAT1 promoter regions and ILF3 binding sites identi ed as the GCGCCC site (-28 to -23 bp) of STAT1 promoter. Data are mean ± SEM.

Figure 8
Schematic diagram of ILF3 promoting atherosclerotic calci cation ILF3 functions as an indirect regulator of Runx2 expression by targeting BMP2 and STAT1 transcription. The increase in BMP2 level and decrease in STAT1 level induces Runx2 expression and function, which results in an osteogenic switch of both VSMCs and macrophages. VSMCs undergo a phenotype alteration from a contractile to dedifferentiated synthetic phenotype, and macrophages transform from an anti-in ammatory M2 to proin ammatory M1 phenotype. So, ILF3 accelerates the osteogenic differentiation of both VSMCs and macrophages and further promotes atherosclerotic calci cation by acting on the regulation of BMP2 and STAT1 transcription levels.