Abstract

   Inflammation, oxidative stress, and copper (Cu) play an important role
   in cardiovascular disease, including atherosclerosis. We previously
   reported that cytosolic Cu chaperone antioxidant-1 (Atox1) translocates
   to the nucleus in response to inflammatory cytokines or exogenous Cu
   and that Atox1 is localized at the nucleus in the endothelium of
   inflamed atherosclerotic aorta. However, the roles of nuclear Atox1 and
   their function are poorly understood. Here we showed that Atox1
   deficiency in ApoE^−/− mice with a Western diet exhibited a significant
   reduction of atherosclerotic lesion formation. In vitro,
   adenovirus-mediated overexpression of nuclear-targeted Atox1
   (Ad-Atox1-NLS) in cultured human endothelial cells (ECs) increased
   monocyte adhesion and reactive oxygen species (ROS) production compared
   to control cells (Ad-null). To address the underlying mechanisms, we
   performed genome-wide mapping of Atox1-regulated targets in ECs, using
   an unbiased systemic approach integrating sequencing data. Combination
   of ChIP-Seq and RNA-Seq analyses in ECs transfected with Ad-Atox1-NLS
   or Ad-null identified 1387 differentially expressed genes (DEG). Motif
   enrichment assay and KEGG pathway enrichment analysis revealed that 248
   differentially expressed genes, including inflammatory and angiogenic
   genes, were regulated by Atox1-NLS, which was then confirmed by
   real-time qPCR. Among these genes, functional analysis of inflammatory
   responses identified CD137, CSF1, and IL5RA as new nuclear
   Atox1-targeted inflammatory genes, while CD137 is also a key regulator
   of Atox1-NLS-induced ROS production. These findings uncover new nuclear
   Atox1 downstream targets involved in inflammation and ROS production
   and provide insights into the nuclear Atox1 as a potential therapeutic
   target for the treatment of inflammatory diseases such as
   atherosclerosis.

   Keywords: antioxidant-1, copper, transcriptome sequencing, ROS,
   inflammation, endothelial cells, atherosclerosis

1. Introduction

   Inflammation contributes to the initiation and progression of
   atherosclerotic lesion, which is associated with immunity [[32]1],
   clonal hematopoiesis of indeterminate potential (CHIP) [[33]2], and
   oxidative stress [[34]3,[35]4]. The causal role of inflammation in
   atherosclerosis and cardiovascular risk has been supported by recent
   clinical trials, such as the canakinumab Anti-inflammatory Thrombosis
   Outcomes Study (CANTOS) on patients with established atherosclerotic
   disease [[36]5,[37]6]. In addition, excessive reactive oxygen species
   (ROS)/oxidative stress also play an important role in vascular
   inflammatory disease, such as atherosclerosis, by inducing expression
   of redox-sensitive inflammatory genes such as VCAM-1 and ICAM-1 in
   endothelial cells (ECs), oxidative modification of lipoproteins
   [[38]3], and macrophage activation [[39]4]. Furthermore, copper (Cu),
   an essential micronutrient, has also been implicated in inflammation
   and atherosclerosis [[40]7,[41]8]. Implanting a Cu cuff promotes
   vascular neointima thickening [[42]9], whereas Cu chelator treatment
   inhibits inflammation in vascular tissue and development of
   atherosclerotic lesion in ApoE^−/− mice [[43]10], as well as neointimal
   formation in response to vascular injury [[44]11,[45]12]. Cu plays an
   important role in inflammatory responses involved in innate and
   adaptive immunity [[46]13,[47]14]. For example, Cu deficiency or Cu
   chelator decreased leukocyte adhesion to the activated ECs and adhesion
   molecule expression, such as ICAM-1/VCAM-1 [[48]15]. Despite the
   critical role of inflammation, ROS, and Cu in atherosclerosis, the
   mechanistic link between Cu, ROS, and inflammation in atherosclerosis
   remains unknown.

   Antioxidant-1 (Atox1) [[49]8,[50]16,[51]17,[52]18] is known to function
   as a cytosolic Cu chaperone with 68 amino acid residues and is highly
   conserved among mammalian species. Atox1 at the cytosol transfers Cu(I)
   to the Cu-transporting ATPase, ATP7A, at the trans-Golgi network (TGN)
   for delivery of Cu to the secretory pathway or secretory cuproenzymes
   such as extracellular SOD (ecSOD, SOD3) or lysyl oxidase to regulate
   their activity [[53]8,[54]18,[55]19]. Importantly, Hamza et al.
   [[56]20] initially reported that Atox1 is localized not only in the
   cytosol to function as a Cu chaperone for ATP7A, but also localized in
   the nucleus in human cancer cell line HeLa cells. This finding was
   further confirmed by us and others in other cell types
   [[57]16,[58]21,[59]22,[60]23,[61]24,[62]25,[63]26,[64]27,[65]28,[66]29,
   [67]30]. For example, we demonstrated that: Atox1 is predominantly
   localized at the nucleus in the endothelium of inflamed atherosclerotic
   aorta compared to control aorta [[68]17]; Atox1 translocates from the
   cytosol to the nucleus in response to exogenous Cu to function as a
   Cu-dependent transcription factor for cyclin D1 to promote cell
   proliferation [[69]25]; G-protein-coupled receptor agonist angiotensin
   II stimulates Atox1 nuclear translocation to promote transcription of
   ecSOD in vascular smooth muscle cells, which protects against oxidative
   stress in hypertension [[70]29]; inflammatory cytokine TNFα rapidly
   stimulates nuclear translocation of Atox1 in Cu- and TRAF4-dependent
   manner to upregulate cytosolic NADPH oxidase organizer, p47phox in ECs,
   which promotes ROS-dependent inflammatory angiogenesis in vivo
   [[71]16,[72]17,[73]31]; and nuclear translocation of Atox1 potentiates
   activin A-induced cell migration and colony formation in colon cancer
   [[74]27]. Consistent with our findings, others have also reported
   nuclear Atox1 in various cancers [[75]21,[76]22,[77]32,[78]33,[79]34].
   Jin et al. reported that ATOX1 binds to the promoter of mediator of DNA
   damage checkpoint 1, which enhances its transcriptional activity and
   tumor growth [[80]33]. Chen et al. reported that APEX2-based proximity
   labeling of Atox1 identifies cysteine-rich protein 2 (CRIP2) as a
   nuclear Cu-binding protein that regulates autophagy activation in lung
   cancer H1299 cells [[81]32]. Despite these critical advances in our
   knowledge supporting the transcription factor function of Atox1, a
   genome-wide analysis of Atox1 targeted genes in ECs and their function
   has not been described.

   In this study, we performed the genome-wide mapping of Atox1-regulated
   target gene analysis, using unbiased systemic approaches such as
   chromatin immunoprecipitation followed by high-throughput sequencing
   (ChIP-Seq) and RNA-sequencing (RNA-Seq) in human ECs transfected with
   Atox1 with nuclear-targeted sequence (Atox1-NLS). Our broad analysis
   identified a number of differentially expressed genes (846 upregulated
   and 541 downregulated genes) regulated by nuclear Atox1. Moreover,
   motif enrichment assay and a KEGG pathway enrichment analysis indicated
   that Atox1-NLS increases the genes involved in inflammation and ROS
   production in ECs. These findings provide insights into nuclear Atox1
   as a potential therapeutic target for the treatment of inflammatory
   diseases such as atherosclerosis.

2. Methods

2.1. Animals

   Atox1^−/− mice were obtained from Mutant Mouse Regional Resource
   Centers and backcrossed eight times to C57Bl/6J (Jackson laboratory,
   Bar Harbor, ME, USA). ApoE^−/−/Atox1^−/− mice were generated by
   crossing Atox1^−/− and ApoE^−/− mice (Jackson Laboratory, Bar Harbor,
   ME, USA). ApoE^−/−/Atox1^−/− or ApoE^−/− mice were maintained on
   regular chow for 7 weeks and then on a Western diet (TD88137; ENVIGO,
   Indianapolis, IN, USA) for 16 weeks. Diet and water were provided ad
   libitum. The use of mice was in accordance with the National Institutes
   of Health Guide for the Care and Use of Laboratory Animals and relevant
   ethical regulations. Studies were carried out in accordance with
   guidelines approved by the Institutional Animal Care and Use Committee
   of Augusta University.

Atherosclerotic Lesion Analysis

   After being fed a Western diet, mice were anesthetized and perfused
   with phosphate-buffered saline (PBS) and subsequently paraformaldehyde
   (4%) in PBS. Then, periadventitial fat and connective tissue were
   removed from the aortas. Aortas were stained with 2% Oil Red O (Sigma
   Aldrich, St. Louis, MO, USA) and opened longitudinally, and photographs
   were taken. The aortic atherosclerotic lesions were quantified using
   Image-Pro Plus software (Media Cybernetics, Rockville, MD, USA).
   Additionally, the upper one-third of the heart was excised and embedded
   in an optimum cutting temperature compound. Then, the tissue was
   cryosectioned (5 μm thick). Sections were stained with Oil Red O or
   hematoxylin to evaluate the size of necrotic cores or atherosclerotic
   lesions in the aortic sinus. Image-Pro Plus software (Media
   Cybernetics, Rockville, MD, USA) was used for quantification.

2.2. Cell Culture and Reagents

   Human umbilical vein endothelial cells (HUVECs) (Lonza, Portsmouth, NH,
   USA) were cultured in a flask coated with 0.1% gelatin using EndoGRO
   basal medium (Millipore, Burlington, MA, USA) containing 5% fetal
   bovine serum (FBS) and used for experiments until Passage 6. Human
   aortic endothelial cells (HAECs) (Cell Application Inc. San Diego, CA,
   USA) were grown in human endothelial cell growth medium (Cell
   Application Inc.) containing 10% FBS and used for experiments until
   Passage 7. Proinflammatory cytokine cocktail containing TNF-α (10
   ng/mL), IL-1β (10 ng/mL), and IL-6 (10 ng/mL) was used in in vitro
   experiments. siRNAs were transfected using Oligofectamine (Invitrogen,
   Waltham, MA, USA) in HUVECs for 3 h, followed by 48 h growth. The
   sequence for human CD137 siRNA is 5′-AAGCAGTTACTACAAGGATCC-3′; human
   CSF1 siRNA is 5′-GAUCCAGUGUGCUACCUUAAGAAGG-3′; human IL5RA siRNA is
   5′-GCAAAGGAAUGUUAAUCUAGAAUAT-3′. Human Atox1 siRNA was purchased from
   Ambion (Austin, TX, USA). THP-1 monocyte cells were grown in RPMI
   medium with 10% FBS.

2.2.1. siRNA and Adenovirus Transfection

   HUVECs were incubated with either Ad-Null (control) or Ad. Atox1-NLS or
   Ad-catalase (University of Iowa Viral Vector Core, Iowa City, IA, USA)
   in 5% FBS containing culture medium for 24 h, followed by incubation in
   culture medium without virus for 24 h before experiments. For
   nuclear-targeted Atox1 (Flag-Atox1-NLS) constructs, a tripartite NLS
   sequence (PKKKRKVD) derived from the SV40 large T antigen was fused to
   the C terminus of Flag-tagged WT-Atox. For siRNA transfection, HUVECs
   were grown to 50% confluence and transfected with control siRNA, CD137
   siRNA, IL5RA siRNA, CSF1 siRNA, or Atox1 siRNA (30 nM) using
   Oligofectamine. siRNAs were transfected using Oligofectamine
   (Invitrogen, Waltham, MA, USA) in HUVECs for 3 h, followed by 48 h
   growth. For double transfection of siRNA and Ad-virus, cells were
   transfected with siRNAs as described above. After 16 h of siRNA
   transfection, cells were transfected with Ad-null or Ad-Atox1-NLS or
   Ad-Catalase. Cells were used for experiments at 48 h after adenovirus
   transfection.

2.2.2. Immunofluorescence Analysis

   HUVECs were grown on glass coverslips. Cells were washed quickly in
   ice-cold PBS, fixed with paraformaldehyde (4%) in PBS for 10 min at
   room temperature, and permeabilized in Triton X-100 (0.05%) in PBS for
   5 min. Then, cells were rinsed with PBS, quenching aldehydes with 50 μM
   NH[4]Cl, and washed with PBS for 10 min each. After incubation for 1 h
   in blocking buffer (PBS+ 3%BSA), cells were incubated with flag
   antibody (F7425, Sigma Aldrich, St. Louis, MO, USA) or Atox1 antibody
   [[82]25] for 18 h at 4 °C. Then, cells were rinsed with PBS and
   incubated in Alexa Fluor 488-conjugated goat anti-rabbit IgG for 30 min
   at room temperature. Finally, cells on coverslips were mounted onto
   glass slides using Vectashield (Vector Laboratories, Newark, CA, USA)
   and photographed using confocal microscopy.

2.2.3. Nuclear/Cytosolic Fractionation Assay and Immunoblotting

   Nuclear/cytoplasmic fractionation from HUVECs was performed using the
   NEPER Nuclear and Cytoplasmic Extraction Reagents Kit (ThermoFisher,
   Waltham, MA, USA) according to the manufacturer’s protocol. For
   immunoblotting, cell lysates from nuclear/cytosolic fractionation or
   total cell lysate were separated using SDS-polyacrylamide gel
   electrophoresis and transferred to nitrocellulose membranes and blocked
   with blocking buffer (PBS containing 5% nonfat dry milk and 0.1% Tween
   20) for 1 h. Then, the membrane was incubated overnight with primary
   antibodies (anti-Atox1 (this laboratory), anti-flag (Sigma, St. Louis,
   MO, USA), anti-actin (Santa Cruz, Dallas, TX, USA), GADPH, and P84
   antibody (GeneTex, Irvine, CA, USA). After incubation with Goat
   Anti-Rabbit IgG-HRP Conjugate or Goat Anti-Mouse IgG-HRP Conjugate
   (Bio-Rad, Hercules, CA, USA), proteins were detected by ECL
   chemiluminescence.

2.2.4. Monoclonal Atox1 Antibody Production

   Mouse monoclonal Atox1 antibody ([83]Z31112) was developed by using
   baculovirus particles displaying the surface glycoprotein gp64-fusion
   protein as the immunizing agent, as we previously described [[84]16].
   Briefly, the Atox1 cDNA was ligated into the gp64 gene to create a
   fusion protein that is expressed in the viral surface, as previously
   described [[85]16]. Positive clones were selected by Western blotting
   using recombinant Atox1 protein or cells overexpressing Atox1 by
   adenovirus [[86]25,[87]26].

2.2.5. Chromatin Immunoprecipitation (ChIP) and Sequencing

   HUVECs grown to confluence were transfected with Atox1-NLS for 48 h.
   ChIP assays were performed by following the EpiTect ChIP One-Day Kit
   protocol (Qiagen, Germantown, MD, USA), as we described previously
   [[88]16]. Briefly, cells were incubated with 1% formaldehyde and lysed
   with lysis buffer. Then, cells were sonicated and diluted with dilution
   buffer. The lysate was incubated with monoclonal Atox1 antibody
   ([89]Z31112) (see above) or normal mouse IgG as control. ChIP-enriched
   DNA was subjected to library preparation using a ChIP-Seq DNA Sample
   Prep kit (Illumina, San Diego, CA, USA). Sequencing was performed using
   an Illumina HiSeq 2500 platform (GTAC, Washington University, St.
   Louis, MO, USA). For validation, the primer sequence was used as shown
   in [90]Supplementary Table S8.

2.2.6. ChIP Sequencing Analysis

   The FASTQ ChIP-Seq read files were aligned to the reference human
   genome build hg19 using Bowtie 2 (v2.4.2, Johns Hopkins University,
   Baltimore, MD, USA) with default settings, and the generated BAM files
   were sorted and used as input files by MACS (v2.2.7.1, Harvard
   University, Cambridge, MA, USA) with default settings for peaks
   calling, where false discovery rate (FDR) <5% was used to call an Atox1
   binding peak. ChIPseeker (v1.32.0, Southern Medical University,
   Guangzhou, China) was used for the annotations of those Atox1 binding
   peaks, and Integrative Genomics Viewer (v2.10.0, Broad Institute,
   Cambridge, MA, USA) was used for the visualization of the peaks at
   specific genomic regions. HOMER (v4.11, The University of California,
   San Diego, CA, USA) with default settings was used for motif analysis,
   where those significantly enriched motifs were identified using FDR <
   5%.

2.2.7. RNA Sequencing

   Total RNA was extracted using the RNeasy kit (Qiagen, Germantown, MD,
   USA), and RNAs were quantified by spectrophotometry (ND-2000, NanoDrop,
   ThermoFisher, Waltham, MA, USA). The quality of RNAs was determined
   using electrophoresis (2100 Bioanalyzer, Agilent Technologies, Santa
   Clara, CA, USA) prior to next-generation sequencing in the Genome
   Technology Access Center (GTAC) at Washington University, St. Louis,
   MO, USA. Samples with RNA integrity number (RIN) values >8.0 were
   subjected to poly-A selection, chemical fragmentation, random hexamer
   priming, cDNA synthesis, and adapter-ligation using the TruSeq RNA
   Library Prep Kit (Illumina, San Diego, CA, USA), followed by paired-end
   multiplexed sequencing (HiSeq 2500, Illumina) according to the
   manufacturer’s instructions.

2.2.8. RNA-Seq Analysis

   The FASTQ RNA-Seq read files were quantified using Kallisto (v0.46.1,
   California Institute of Technology, Pasadena, CA, USA) with the
   transcriptome indices built on the cDNA FASTA files from human genome
   assembly GRCh38. Genes with lower than 10 total read counts were
   filtered for subsequent differential expression analysis. Differential
   expression analysis was performed using DESeq2 (v1.32.0, The University
   of North Carolina at Chapel Hill, Chapel Hill, NC, USA) by comparing
   Ad.Atox1-NLS and Ad.null samples, and genes with fold changes greater
   than 2.5 (>2.5 as upregulated and <0.4 as downregulated) and adjusted p
   < 0.05 were considered differentially expressed. The Gene Ontology (GO)
   Consortium, Molecular Signatures Database (MSigDB), and Kyoto
   Encyclopedia of Genes and Genomes (KEGG) pathway databases were used to
   perform GO and pathway enrichment analysis using R package
   clusterProfiler (v4.0.2, Southern Medical University, Guangzhou,
   China).

2.2.9. Quantitative Real-Time PCR

   Total RNA of HUVECs was isolated using TRI Reagent (Molecular Research
   Center Inc. Cincinnati, OH, USA). Reverse transcription was carried out
   using the high capacity cDNA reverse transcription kit (Applied
   Biosystems, Waltham, MA, USA) using 2 µg of total RNA. Quantitative PCR
   was performed with the SYBR Green PCR kit (Qiagen, Germantown, MD, USA)
   and ABI Prism 7000. The primer sequence for specific genes
   ([91]Supplementary Table S9). Samples were run in duplicates to reduce
   variability. Expression of genes was normalized with 18S and expressed
   as fold change of relative mRNA expression compared to control siRNA or
   Ad-Null.

2.2.10. Monocyte Adhesion Assay

   HUVECs were transfected with Ad-Null or Ad-Atox1-NLS and grown in
   24-well-plates. THP-1 cells were labeled with Cell Tracker Green CMFDA
   (Invitrogen, Waltham, MA, USA). The labeled THP-1 cells were added to
   the confluent EC monolayer. Cultures were incubated in a CO[2]
   incubator for 1 h. Nonadherent cells were removed from the plate by
   gentle rinsing with PBS, and the number of adherent cells was
   determined by confocal microscopy. Cells were counted in five random
   20× fields per well.

2.2.11. ROS Measurement

   ROS production in HUVEC cells was detected by incubating the cells with
   20 μM of CM-H2DCFDA
   (5-(and-6)-chloromethyl-2′,7′-dichlorodihydrofluorescein diacetate,
   acetyl ester, Invitrogen, Waltham, MA, USA) for 6 min and photographed
   by confocal microscopy using the same exposure condition in each
   condition. Relative DCF-DA fluorescence intensity was analyzed using
   LSM software as we reported previously [[92]16].

2.3. Statistical Analysis

   All experiments were performed at least three times, and all values
   were expressed as mean ± SEM. Comparison between groups was analyzed by
   unpaired Student’s two-tailed t-test for two groups or one-way analysis
   of variance (ANOVA), followed by Bonferroni’s post hoc analysis for
   experiments with more than two subgroups. Statistical tests were
   performed using Prism v9 (GraphPad Software, San Diego, CA, USA).

3. Results

3.1. Role of Atox1 in Inflammatory Disease

   We initially confirmed our previous observation [[93]16,[94]17] that
   Atox1 is mainly localized in the cytosol in unstimulated, basal
   cultured human umbilical vein endothelial cells (HUVECs), or human
   aortic endothelial cells (HAECs) and that Atox1 is translocated to the
   nucleus in response to the proinflammatory cytokines (tumor necrosis
   factor (TNF)-α, interleukin (IL)-1β, and IL-6), which are involved in
   vascular pathology ([95]Figure 1A, [96]Supplementary Figure S1A)
   [[97]35,[98]36]. This is consistent with our previous report that Atox1
   is localized at the nucleus in the endothelium of inflamed
   atherosclerotic aorta [[99]17]. To investigate the functional
   significance of Atox1 in atherosclerosis in vivo, we used either
   Atox1^−/− mice crossed with ApoE^−/− mice (Atox1^−/−/ApoE^−/−) or
   ApoE^−/− (control) mice on a 4-month Western diet. Oil Red O staining
   revealed a significant decrease in atherosclerotic lesions in the
   surface lesion area ([100]Figure 1B) and the aortic sinus ([101]Figure
   1C) in Atox1^−/−/ApoE^−/− mice, as compared to control mice. Masson
   Trichrome staining of collagen in the aortic sinus showed significantly
   higher fibrosis in ApoE^−/−/Atox1^−/− mice as compared to control mice
   ([102]Figure 1C). These results suggest that Atox1 localized in the
   nucleus plays an important role in inflammatory disease, such as
   atherosclerosis.

Figure 1.

   [103]Figure 1
   [104]Open in a new tab

   Role of Atox1 in inflammatory disease. (A). Immunofluorescence staining
   of Atox1. HUVECs stimulated with proinflammatory cytokine cocktail
   (TNF-α (10 ng/mL), IL-1β (10 ng/mL), and IL-6 (10 ng/mL) for 1 h were
   immunostained for Atox1 using anti-Atox1 Ab or nuclear marker DAPI.
   Percentage of Atox1+ cells in nucleus was shown in right (n = 3)
   Scale:50 μm. (B,C). Loss of Atox1 decreases atherosclerotic lesion in
   ApoE^−/− mice. ApoE^−/− or Atox1^−/− /ApoE^−/− were fed a Western diet
   for 16 weeks. (B). Representative images of Oil Red O stained-aortas
   (en face) from ApoE^−/− (n = 6) or Atox1^−/−/ApoE^−/− (n = 9) mice
   after 16 weeks of Western diet. (Lower panel) Quantification data of
   lesion area. (C). Representative images and quantification data of
   cross-sections of aortic sinus of ApoE^−/− or Atox1^−/− /ApoE^−/− mice
   with 16 weeks of Western diet. Sections were stained with Oil Red O
   (lesion), hematoxylin and eosin (plaque), and Masson’s trichrome
   (collagen) with the magnified inset (black box) (n = 8–13). Scale bar:
   200 μm. *** p < 0.001.

3.2. Nuclear Atox1 Increases Inflammatory Responses and ROS Production in ECs

   To determine the roles of nuclear Atox1 in inflammatory responses in
   cultured ECs, we generated adenovirus expressing nuclear-targeted Atox1
   (Ad.Atox1-NLS) in which a tripartite nuclear localization sequence
   (NLS, PKKKRKVD) derived from the SV40 large T antigen was fused to the
   C terminus of Flag-tagged Atox1, as we previously reported [[105]25].
   Immunofluorescence and immunoblotting analyses confirmed that Atox1-NLS
   was predominantly expressed in the nucleus in ECs transduced with
   Ad.Atox1-NLS as compared to Ad.null (control) ([106]Figure 2A,B,
   [107]Supplementary Figure S2A,B). We also found that Atox1 expression
   is increased in total lysate from Ad.Atox1-NLS transfected ECs
   ([108]Supplementary Figure S2A). Functional significance of nuclear
   Atox1 showed that Ad.Atox1-NLS significantly increased monocyte
   adhesion ([109]Figure 2C) and ROS production, as measured by DCF
   fluorescence ([110]Figure 2D), which were inhibited by catalase
   overexpression ([111]Figure 2C,D) in ECs. These results suggest that
   nuclear Atox1 overexpression is sufficient to induce inflammatory
   responses and ROS production in ECs.

Figure 2.

   [112]Figure 2
   [113]Open in a new tab

   Nuclear Atox1 increases inflammatory response and ROS production in
   ECs. (A). Subcellular localization of nuclear-targeted Atox1
   (Flag-Atox1-NLS) fusion proteins. HUVECs were transfected with
   adenovirus expressing Flag-Atox1-NLS (Ad-Atox1-NLS) or adenovirus alone
   (Ad-null), immunolabeled with antibodies for Flag tag (green) and
   nuclear marker, 4′,6-diamidino-2-phenylindole (DAPI) (blue). Scale: 50
   μm. (B). Subcellular localization of Atox1 in Ad-Atox1-NLS or
   Ad-Null-transfected HUVECs. Nuclear or cytosolic fractions were
   subjected to Western blotting with anti-flag, anti-P84 (a marker for
   nuclear fraction), or anti-GAPDH (a marker for cytosolic fraction).
   (C). HUVECs were transfected with Ad-Catalase, Ad-Atox1-NLS, or Ad-Null
   for 48 h. The numbers of bound THP1 monocytes (fluorescently labeled)
   to ECs were measured with a fluorescence microscope. Right panel
   quantification (n = 5). Scale: 100 μm. (D). HUVECs were transfected
   with Ad-Atox1-NLS or Ad-Null in the presence or absence of Ad-Catalase
   for 48 h. Representative images for DCF fluorescence and DAPI staining
   (blue, nucleus marker) and quantification of fluorescence intensity
   (right, n = 3). Scale: 100 μm. *** p < 0.001, ** p < 0.01, * p < 0.05.

3.3. Atox1 Gene Binding and Gene Expression Signatures

   To characterize the genes differentially expressed in response to
   overexpression of nuclear Atox1, we performed genome-wide analysis by
   combining ChIP-Seq and RNA-Seq data sets in HUVECs treated with
   Ad.Atox1-NLS or Ad.null, as outlined in [114]Figure 3A. HUVECs were
   transfected with Ad-Atox1 NLS or Ad.null for 48 h, followed by RNA
   isolation, and subjected to cDNA library preparation and deep
   sequencing. [115]Figure 3A shows the summary of the differentially
   expressed Atox1-bound genes, i.e., genes that are identified as
   differentially expressed in RNA-Seq and have at least one Atox1 binding
   peak identified in ChIP-Seq. We find that Atox1 binds to 541 (85%)
   among 638 downregulated genes ([116]Supplementary Table S1; designated
   as Atox1 downregulated targets) and 846 (85%) among the 997 upregulated
   genes by Atox1 in RNA-Seq ([117]Supplementary Table S2; designated as
   Atox1 upregulated targets). [118]Figure 3B shows the expression levels
   for those 541 downregulated and 846 upregulated genes, and [119]Figure
   3C shows the pattern for all genes analyzed in RNA-Seq with red dots
   representing those differentially expressed genes (DEGs).

Figure 3.

   [120]Figure 3
   [121]Open in a new tab

   Genome-wide Atox1 gene targets identified by combining RNA-Seq and
   ChIP-Seq analysis. (A). Strategy for identifying Atox1 gene targets in
   HUVEC cells transfected with Ad-Atox1-NLS for 48 h. Atox1 gene targets
   represent genes that were bound by Atox1 and showed differential
   expression in response to Atox1-NLS. (B). Heat map of expression
   pattern for differentially expressed genes (DEGs) identified by
   combining ChIP-Seq and RNA-Seq. (C). Volcano plot of the pattern for
   all genes analyzed in RNA-Seq. Red dots represent DEGs with adjusted p
   < 0.05 and fold change >2.5 or <0.4. (D). Heatmap showing signal
   intensity of Atox1 peaks in ChIP-Seq in a 3 kb window centered around
   transcription start sites (TSSs) of genes across the genome. Atox1 was
   specifically precipitated from fragmented DNA-protein complexes by
   Monoclonal Anti-Atox1 antibody but not nonimmune IgG. Samples were used
   for ChIP-Seq. (E). Pie chart showing genomic distribution of
   Atox1-binding peaks. The genomic features (promoters, 5′ UTRs, exons,
   introns, 3′ UTRs, downstream, and intergenic regions) were defined
   based on RefSeq gene annotations. (F). Bar graph showing the
   distribution of Atox1-binding peaks in each genomic feature after
   normalization to the size of each category in 100 kb. (G). Atox1
   binding motif (GAAAGA) in Atox1 ChIP-Seq peaks that correspond to the
   one identified in our previous work [[122]25]. (H). Top 3 enriched
   motifs in Atox1 ChIP-Seq peaks.

   Using MACS, a total of 46,075 Atox1 binding peaks were called at an FDR
   of 5%. [123]Figure 3D shows the intensities of those peaks at the
   transcription start sites (TSSs), and [124]Figure 3E shows the genomic
   distribution of those peaks relative to functional categories across
   the genome, where the majority of Atox1-binding peaks were localized in
   introns (37%) and intergenic regions (54%) ([125]Figure 3E). However,
   after normalization to the size of these functional regions, it became
   clear that Atox1-binding peaks were enriched in gene promoters and
   5′-UTRs ([126]Figure 3E,F). Motif enrichment analysis revealed ten
   highly enriched motifs, including the Atox1 binding motif (Atox1-RE:
   GAAAGA) identified in our previous work [[127]25] and other motif
   families (transcription factor, PRDM14, TEC1, bHLH80) ([128]Figure
   3G,H). These results indicate that nuclear Atox1 regulates gene
   expression via Atox1-RE or other motif families (i.e., coactivator).

3.4. Analysis of Nuclear Atox1-Regulated Genes

   Further genomic annotations of those Atox1 binding peaks identified in
   ChIP-Seq revealed that a total of 20,861 genes contained at least one
   peak within 5 kb upstream or downstream of their genomic loci, and
   among them 541 genes are downregulated and 846 are upregulated by Atox1
   in RNA-Seq ([129]Figure 3A and [130]Figure 4A), which results in 1387
   DEGs. Next, we performed the pathway and Gene Ontology (GO) enrichment
   analysis using those 1387 Atox1-regulated DEGs ([131]Figure 4A).
   Specifically, we conducted enrichment analysis using the following
   databases: GO Consortium (including GO terms in biological process, BP,
   molecular function, MF, and cellular components, CC), Molecular
   Signatures Database (MSigDB), and Kyoto Encyclopedia of Genes and
   Genomes (KEGG) pathways. GO analysis demonstrates that 1101 BP GO
   terms, 90 CC GO terms, and 136 MF GO terms were significantly enriched
   ([132]Supplementary Figure S3A–C and Supplementary Tables S3–S5). In
   KEGG pathway analysis, those 1387 DEGs were classified into 305
   biological pathways, and especially, and 28 pathways were significantly
   enriched ([133]Figure 4A, [134]Supplementary Table S6). Particularly,
   these DEGs were significantly enriched in the inflammatory TNF
   signaling pathways, cell cycle, p53 signaling, and Hippo signaling
   pathways ([135]Figure 4B, [136]Supplementary Figure S4). Based on the
   expression of the DEGs in those four pathways, we identified several
   candidate Atox1 regulated genes involved in those pathways and further
   validated their differential expression levels between Ad.Atox1-NLS and
   Ad.null by qRT-PCR, which are listed as follows: inflammatory TNF
   signaling pathways—TRAF1, CSF1, MAP3K14 ([137]Figure 4C); cell cycle
   pathways—CDC25A, CREEBP, CDC45; P53 signaling pathways—SENS2, MDM2,
   GTSE1; Hippo signaling pathways—TP73, TAZ, FZD5 ([138]Supplementary
   Figure S4). These analyses suggest that Atox1 regulates inflammation
   and other signaling pathways.

Figure 4.

   [139]Figure 4
   [140]Open in a new tab

   Analysis of nuclear Atox1-regulated genes. (A). Venn diagrams showing
   the overlap between differentially expressed genes (DEGs) identified by
   RNA-Seq and Atox1-bound genes identified by ChIP-Seq. (B). KEGG pathway
   enrichment analysis based on those 1387 overlapped DEGs in A. The top
   10 enriched pathways ranked by the number of DEGs in the pathways are
   shown here. (C). Top: Heatmap showing the expression levels of DEGs
   belonging to the TNF signaling pathway in RNA-Seq. Color scale depicts
   range of log2 fold changes in gene expression. Bottom: Three selected
   DEGs in this pathway was validated by quantitative RT-PCR (n = 3) *** p
   < 0.001, ** p < 0.01.

3.5. Genome-Wide Identification of Direct Atox1 Target Genes

   We next analyzed the expression changes of nuclear Atox1 directly
   targeted genes, i.e., genes differentially expressed in response to
   Atox1-NLS in RNA-Seq and having at least one Atox1 binding peak in
   ChIP-Seq and at least one Atox1 binding motif sequence (GAAAGA) within
   3 kb of their promoters, and this analysis identified 248 such genes,
   where 154 genes were upregulated and 94 genes were downregulated in
   HUVEC overexpressing Atox1-NLS compared to control ([141]Figure 5A,
   [142]Supplementary Table S7), and their expression pattern was shown in
   [143]Figure 5B. We further performed pathway enrichment analysis based
   on those genes using the KEGG pathway database, and those significantly
   enriched pathways are shown in [144]Figure 5C. Among them, we validated
   the expression of two groups of genes that are related to specific
   biological processes using qRT-PCR, which are inflammation-related
   genes, including ADRB1, CSF1, ERBB3, DUSP16, TNFSF11, MAFA, CD137, and
   IL5RA ([145]Figure 5D, [146]Supplementary Figure S1B) and
   angiogenic-related genes, including CCN2, LMO2, ADIPOQ and PIM1
   ([147]Supplementary Figure S5A). Furthermore, in both groups, the Atox1
   binding peaks near the transcription start site (TSS) of several
   selected Atox1-regulated targeted genes are shown, which include ADRB1,
   CSF1, IL5RA, and CD137 (inflammation-related genes, [148]Figure 6A) and
   CCN2, PIM1, LMO2 and MDM2 ([149]Supplementary Figure S5B). Furthermore,
   these Atox1-regulated targets were further validated by ChIP-PCR,
   showing the enrichment of Atox1-bound chromatin peaks ([150]Figure 6B
   and [151]Supplementary Figure S5C).

Figure 5.

   [152]Figure 5
   [153]Open in a new tab

   Genome-wide identification of direct Atox1 target genes. (A). Venn
   diagrams showing the overlap of DEGs directly regulated by Atox1, which
   are genes differentially expressed as identified in RNA-Seq and bound
   by Atox1 via Atox1 binding motif (GAAAGA) identified in ChIP-Seq. (B).
   Heatmap of expression pattern for those DEGs directly regulated by
   Atox1 as shown in (A). (C). KEGG pathway enrichment analysis of those
   248 DEGs as shown in A. The 17 enriched pathways with p < 0.05 are
   shown here. (D). Subset of the Atox1 regulated gene targets was
   validated by quantitative RT-PCR (n = 3).

Figure 6.

   [154]Figure 6
   [155]Open in a new tab

   Validation of Atox1 binding peaks. (A). Representative tracks from
   ChIP-Seq results show the locations of Atox1 binding peaks proximal to
   the TSS of Atox1 targeted genes. Those peaks in red box were validated
   in (B). (B). HUVEC cells were transfected with Atox1-NLS for 48 h and
   subject to ChIP using monoclonal Anti-Atox1 Ab and mouse IgG. The
   enrichment of Atox1 binding peaks were determined by qPCR (n = 3). ***
   p < 0.001, ** p < 0.01.

3.6. CD137, IL5RA, and CSF1 as New Atox1 Targets to Regulate Inflammation and
ROS Production in ECs

   Both RNA-Seq and ChIP-Seq analysis identified CD137 (TNFRSF9), IL5RA,
   and CSF1 as new direct targets of Atox1 which regulate inflammatory
   responses. To determine the functional significance of
   Atox1-NLS-induced CD137, IL5RA, and CSF1 expression, we examined their
   roles in monocyte adhesion on ECs in vitro using THP-1 monocytes and
   HUVECs. CD137, IL5RA, and CSF1 siRNA treatment ([156]Figure 7A)
   significantly decreased Atox1-NLS-induced monocyte adhesion in HUVECs
   ([157]Figure 7B). Furthermore, CD137 siRNA, but not either IL5RA or CSF
   siRNA, significantly decreased Atox1-NLS-induced ROS production in ECs,
   as measured by DCF fluorescence ([158]Figure 7C). We next examined the
   roles of endogenous Atox1 in proinflammatory cytokine induced CD137,
   IL5RA, or CSF1 mRNA expression. Atox1 siRNA significantly decreased
   CD137 and IL5RA mRNAs, but not CSF1 mRNA, in response to
   proinflammatory cytokines ([159]Supplementary Figure S6).

Figure 7.

   [160]Figure 7
   [161]Open in a new tab

   CD137, IL15RA, and CSF1 as new Atox1 targets to regulate inflammation
   and ROS production in ECs. ECs were transfected with CD137 or IL5RA or
   CSF1 siRNA. After 16 h, cells were transfected with Ad-Atox1-NLS or
   Ad-null for 48 h. (A). Cells were harvested and CD137 or IL5RA or CSF1
   mRNA expression measured. Panel shows averaged data, expressed as fold
   change over Ad-Null (the ratio in Ad-Null-treated ECs transfected with
   Control siRNA was set to 1). (B). The numbers of bound THP1 monocytes
   (fluorescently labeled) to ECs were measured with a fluorescence
   microscope. Right panel quantification (n = 3). Scale: 100 μm. (C).
   Representative images for DCF fluorescence and DAPI staining (blue,
   nucleus marker) and quantification of fluorescence intensity (right) (n
   = 3). Scale: 100 μm. *** p < 0.001, ** p < 0.01, * p < 0.05. (D).
   Schematic diagram illustrating the mechanisms underlying the effect of
   Atox1-mediated inflammation and atherosclerosis.

4. Discussion

   Antioxidant-1 (Atox1) in the cytosol functions as a Cu chaperone to
   deliver Cu to ATP7A in the secretory pathway and is essential for Cu
   homeostasis. In addition, we and others showed that Atox1 is
   translocated from the cytosol to the nucleus to stimulate transcription
   of several genes regulating cell growth and ROS-dependent inflammatory
   gene expression in ECs in response to Cu or the inflammatory cytokine,
   TNFα [[162]16,[163]17,[164]31]. However, downstream targets and the
   functional significance of nuclear Atox1 are still largely unknown.
   Here we performed whole-transcriptome sequencing analyses of ECs
   transfected with Atox1-NLS to identify genes regulated by nuclear
   Atox1. Using unbiased systemic approaches by combining RNA-Seq with
   ChIP-Seq, 248 DEGs were seen to be directly regulated by nuclear Atox1
   via Atox1-RE, while 1139 DEGs were mediated via DNA binding motifs
   other than Atox1-RE. A KEGG pathway enrichment analysis revealed that
   nuclear Atox1 regulates genes related to inflammatory responses,
   including CD137, IL5RA, and CSF1, confirmed by real-time qRT-PCR and
   functional assays. Functionally, Atox1-deficient mice showed decreased
   atherosclerosis development on a high-fat diet. These findings uncover
   new nuclear Atox1 downstream target genes involved in inflammation and
   highlight the potential for targeting nuclear Atox1 as a treatment for
   inflammatory diseases.

   This study performed genome-wide mapping of Atox1 target genes in ECs
   by combining ChIP-seq with RNA-Seq for the first time and identified
   the 46,075 genome-wide Atox1 binding peaks (annotated to 20,861 genes)
   in ECs overexpressing nuclear-targeted Atox1 compared to control. Among
   the genes around these peaks, 846 genes were upregulated, while 541
   genes were downregulated by Atox1-NLS. The majority of Atox1 binding
   sites were in the intergenic regions and introns, but most of the Atox1
   binding peaks were enriched proximal to the TSS, suggesting
   transcriptional regulatory functions for Atox1. An unbiased genome-wide
   motif search showed Atox1 binding to the DNA consensus sequence
   (Atox1-RE, GAAAGA) in the promoters, consistent with our previous
   reports [[165]25]. Furthermore, PRDM4, TEC1, and bHLH80 motifs were
   enriched at Atox1-bound sites, suggesting that Atox1 binds to DNA via
   these transcription factors. Together with our current findings, which
   show other motifs enriched in Atox1 binding, these findings suggest
   that nuclear Atox1 functions as a transcription factor by direct
   binding to DNA, as well as by binding to DNA via other transcription
   factors (i.e., as a coactivator). In the context of in vitro studies
   using recombinant Atox1 protein and labeled DNA, Kahra et al. reported
   that Atox1 protein alone does not bind to DNA in vitro [[166]24], which
   is in contrast to our findings [[167]25]. Although the reason for this
   discrepancy is not clear, Atox1 may require other transcription factors
   or post-translational modification to stabilize its binding to DNA, as
   shown in transcription coactivator Yap and transcription factor TEAD
   [[168]37]. However, the effect of cooperability of transcription
   factors on their affinity to DNA is complex, as shown in Oct4 and Sox2
   [[169]38,[170]39]. Thus, it is conceivable that Atox1 binding with DNA
   directly or binding with other transcription factors as a cofactor may
   be context or agonist-dependent. For example, Atox1 binding with a
   specific motif or binding with another transcription factor may be
   determined by the physiological or pathological stimulants
   (inflammatory cytokines) in healthy or disease conditions such as
   diabetes and atherosclerosis. The molecular mechanisms underlying the
   transcriptional activity of Atox1 via Atox1-RE or other motifs clearly
   warrant further investigation.

   Gene function annotation and pathway analyses revealed that signaling
   pathways enriched in Atox1-NLS-regulated gene targets are relevant not
   only to inflammation, such as TNFα signaling, but also to cell cycle,
   Hippo signaling, and p53 signaling pathways. We and others previously
   reported that Atox1 functions as a Cu-dependent transcription factor
   for cyclin D1 [[171]25,[172]27,[173]31] or binding with an
   anaphase-promoting complex [[174]40] to regulate cell proliferation. In
   the present study, we found that Atox1-NLS increased gene expression of
   cell cycle pathways and cell division, including cycle 25 A (CDC25A)
   [[175]3] and CREBBP [[176]41], which are known to induce cell
   proliferation, whereas it decreased CDC45, which is known to inhibit
   cell proliferation ([177]Supplementary Figure S4) [[178]42]. These
   results suggest that nuclear-targeted Atox1 increases cell
   proliferation partially through these genes. Moreover, Atox1-NLS also
   increased inflammatory genes, TRAF1, and CSF1 but decreased
   anti-inflammatory gene MAP3K14 in ECs. It has been shown that Cu alters
   the conformation and transcriptional activity of the tumor suppressor
   protein p53 in human Hep G2 cells [[179]26]. Consistent with this
   result, Atox1-NLS changed p53-regulated genes such as SESN2, MDM2, and
   GTSE1, which may reflect the Cu-dependent transcriptional activation of
   p53. Furthermore, Atox-NLS also regulates the Hippo signaling pathway
   (i.e., increase in TP73 gene expression and decrease in TEAD2 and FZD5
   gene expression, see [180]Supplementary Figure S4). These results
   suggest that Atox1 may be important in regulating a transcriptional
   network directed by cell cycle, inflammatory, Hippo, and other
   signaling pathways. The functional significance of cell cycle and
   inflammatory gene regulation by nuclear Atox1 warrants further
   investigation.

   The current study identified 248 genes directly regulated by nuclear
   Atox1 in which Atox1-RE (GAAAGA) is present within the +/− 3 kb region
   of the TSS. Using pathway analysis, we found that many genes regulated
   by Atox1 are involved in the inflammatory response and angiogenesis.
   Atox1 increased inflammatory genes (CREB3L3 [[181]43], CD137 [[182]44],
   CSF1 [[183]45], IL5RA [[184]46]) and decreased the anti-inflammatory
   gene (ADRB1 [[185]47]). We previously reported that Atox1 plays an
   essential role in ROS-NFkB-dependent inflammatory neovascularization in
   response to hindlimb ischemia and proinflammatory cytokine TNFα by
   upregulating the NADPH oxidase organizer, p47phox, in ECs [[186]16].
   However, p47phox was not listed in Atox1-NLS-regulated genes,
   suggesting that Atox1-dependent p47phox gene upregulation may require
   additional factors or post-translational modifications, which are
   stimulated by TNFα. Paradoxically, overexpression of nuclear Atox1
   increased antiangiogenic gene (SLC8A2 [[187]48]) and decreased
   angiogenic gene (LMO2 [[188]49], PIM1 [[189]50]) expression. This
   suggests that persistent Atox1 localization in the nucleus may mimic
   pathological conditions, such as diabetes, in which reparative
   angiogenesis is impaired. It is also possible that too much Atox1
   overexpression may cause irrelevant effects. In the case of
   atherosclerosis, the current study showed decreased atherosclerosis
   development on a high-fat diet in global Atox1-deficient mice,
   suggesting a proatherogenic role of Atox1. Addressing the role of
   endothelial nuclear Atox1 in atherosclerosis by using EC-specific
   Atox1^−/− mice, as well as EC-specific Atox1 overexpressing transgenic
   mice, is the subject of a future study.

   RNA-Seq and ChIP-Seq analysis of Atox1-NLS overexpression, as well as a
   loss of function approach using siRNA, revealed CD137 as a new direct
   target of Atox1 to regulate both inflammatory response (monocyte
   adhesion) and ROS production. CD137 (TNFRSF9) is expressed in a variety
   of immunocytes and nonimmune cells, including ECs and smooth muscle
   cells. CD137 is involved in several known biological functions in ECs,
   including inflammation, angiogenesis, and apoptosis [[190]51]. Previous
   studies have shown that the expression of CD137 is upregulated in
   atheroma-associated vascular cells in both mouse and human
   atherosclerotic plaque lesions in vivo [[191]52,[192]53]. In addition,
   we found that CSF1 and IL5RA are other direct targets of Atox1. CSF1
   has been shown to be involved in atherosclerosis development via
   stimulating differentiation, growth, and survival of
   monocytes/macrophages
   [[193]54,[194]55,[195]56,[196]57,[197]58,[198]59,[199]60]. Thus, the
   effect of CSF1 on the development of atherosclerosis is likely due to
   regulating monocyte/macrophage function. Previous studies showed that
   IL5RA is a hematopoietic receptor that plays a dominant role in
   eosinophil biology and can be influenced by proinflammatory cytokines,
   including IL-3, IL-5, IL-9, and GM-CSF [[200]61,[201]62,[202]63];
   however, the effect of IL5RA in regulating endothelial function has not
   been previously reported. Interestingly, we found that loss of CSF1 or
   IL5RA using siRNA inhibited Atox1-NLS-induced monocyte adhesion in ECs.
   Thus, Atox1-NLS-induced CSF1 and IL5RA may contribute to inflammatory
   responses in ECs. However, Atox1 siRNA significantly decreased CD137
   and IL5RA mRNAs, but not CSF1 mRNA, in response to proinflammatory
   cytokines. These results suggest that proinflammatory cytokine-induced
   CD137 and IL5RA gene expression is Atox1-dependent and that CSF1
   upregulation by cytokines requires additional transcription factors in
   ECs.

   In conclusion, we demonstrated that Atox1-deficient mice showed
   decreased atherosclerosis development and identified downstream target
   genes regulated by nuclear Atox1 on the genome-wide scale in ECs. We
   showed that nuclear-targeted Atox1 overexpression in ECs increases
   inflammatory responses and ROS production via transcriptional-level
   activity in endothelial cells. Atox1 was shown to regulate multiple
   signaling pathways, including those involved in p53 signaling, Hippo
   signaling, and other signaling pathways. These findings provide new
   insights into nuclear Atox1 as a potential therapeutic target and a key
   transcriptional regulator of inflammatory gene expression in ECs.

Supplementary Materials

   The following supporting information can be downloaded at:
   [203]https://www.mdpi.com/article/10.3390/cells11182919/s1, Table S1:
   Down regulated -chipseq_rnaseq_overlap_genes. Table S2: Up regulated
   -chipseq_rnaseq_overlap_genes. Table S3: Atox1_GO_BP pathway. Table S4:
   Atox1_GO_CC pathway. Table S5: Atox1_GO_MF pathway. Table S6:
   Atox1_kegg pathways. Table S7: Atox1 direct target genes. Table S8:
   Primer sequences utilized for ChIP–qPCR experiment. Table S9: Primers
   used for qPCR. Figure S1: Immunofluorescence staining of Atox1and
   inflammatory gene expression in HAECs. Figure S2: Atox1 expression in
   ECs transfected with Ad-Atox1-NLS. Figure S3: Analysis of RNA-Seq and
   ChIP-Seq data. Figure S4: Regulation of genes from KEGG pathways
   enriched by Atox1-NLS. Figure S5. Validation of Atox1 binding peaks.
   Figure S6. Role of Atox1 in CD137 or IL5RA or CSF1 mRNA expression in
   ECs treated with proinflammatory cytokines.
   [204]Click here for additional data file.^ (2.1MB, zip)

Author Contributions

   V.S. designed the study, performed experiments, analyzed data, and
   wrote the article. Y.S. Analyzed and interpret ChiP-Seq and RNAseq
   data. J.H.K. gave suggestions to advance this project and edited
   article. M.U.-F. and T.F. designed the overall study, analyzed data,
   and wrote, reviewed, and edited the article. T.F. is the guarantor of
   this work and, as such, had full access to all the data in the study
   and takes responsibility for the integrity of the data and the accuracy
   of the data analysis. All authors have read and agreed to the published
   version of the manuscript.

Institutional Review Board Statement

   The use of mice was in accordance with the National Institutes of
   Health Guide for the Care and Use of Laboratory Animals and relevant
   ethical regulations. Studies were carried out in accordance with
   guidelines approved by the Institutional Animal Care and Use Committee
   of Augusta University (AUP: 2017-0854).

Informed Consent Statement

   Not applicable.

Data Availability Statement

   The data reported here are available from the corresponding author upon
   reasonable request.

Conflicts of Interest

   The authors declare no conflict of interest.

Funding Statement

   This work was supported by National Institute of Health grants:
   R01HL147550 and HL147500-03S1 (to M.U.-F., T.F.), R01HL133613 (to T.F.,
   M.U.-F.), R01HL116976 (to T.F., M.U.-F.), R01HL070187 (to T.F.),
   R01HL160014 (to M.U.-F.), Veterans Administration Merit Review Award
   101BX001232 (to T.F.). The contents do not represent the views of the
   Department of Veterans Affairs or the United States Government. The
   funders had no role in study design, data collection and analysis,
   decision to publish, or preparation of the manuscript.

Footnotes

   Publisher’s Note: MDPI stays neutral with regard to jurisdictional
   claims in published maps and institutional affiliations.

References