Macrophage Activation in the Lung during the Progression of Nitrogen Mustard Induced Injury is Associated with Histone Modifications and Altered miRNA Expression

Abstract

Activated macrophages have been implicated in lung injury and fibrosis induced by the cytotoxic alkylating agent, nitrogen mustard (NM). Herein, we determined if macrophage activation is associated with histone modifications and altered miRNA expression. Treatment of rats with NM (0.125 mg/kg, i.t.) resulted in increases in phosphorylation of H2A.X in lung macrophages at 1 d and 3 d post-exposure. This DNA damage response was accompanied by methylation of histone (H) 3 lysine (K) 4 and acetylation of H3K9, marks of transcriptional activation, and methylation of H3K36 and H3K9, marks associated with transcriptional repression. Increases in histone acetyl transferase and histone deacetylase were also observed in macrophages 1 d and 28 d post-NM exposure. PCR array analysis of miRNAs (miR)s involved in inflammation and fibrosis revealed unique and overlapping expression profiles in macrophages isolated 1, 3, 7, and 28 d post-NM. An IPA Core Analysis of predicted mRNA targets of differentially expressed miRNAs identified significant enrichment of Diseases and Functions related to cell cycle arrest, apoptosis, cell movement, cell adhesion, lipid metabolism, and inflammation 1 d and 28 d post NM. miRNA-mRNA interaction network analysis revealed highly connected miRNAs representing key upstream regulators of mRNAs involved in significantly enriched pathways including miR-34c-5p and miR-27a-3p at 1 d post NM and miR-125b-5p, miR-16-5p, miR-30c-5p, miR-19b-3p and miR-148b-3p at 28 d post NM. Collectively, these data show that NM promotes histone remodeling and alterations in miRNA expression linked to lung macrophage responses during inflammatory injury and fibrosis.

INTRODUCTION

Nitrogen mustard (NM) is a bifunctional alkylating agent and cytotoxic vesicant known to cause acute injury to the respiratory tract that progresses to fibrosis. This is associated with an accumulation of activated macrophages in the lung (Malaviya et al., 2012). We previously showed that these cells consist of phenotypically distinct subpopulations with characteristics of classically activated M1 macrophages, which promote inflammation and cytotoxicity, and alternatively activated anti-inflammatory/wound repair M2 macrophages which contribute to fibrosis (Sunil et al., 2011; Venosa et al., 2016).

Macrophage activation and polarization are controlled, in part, by alterations at the transcriptional (e.g., histones) and post-transcriptional (e.g., miRNAs) levels (Shanmugam and Sethi, 2013; Van den Bossche et al., 2014; Chen et al., 2020). Changes in histones and miRNAs in response to environmental cues provide macrophages with the capacity to rapidly between cellular phenotypes. Histones are nuclear scaffolding proteins involved in DNA packaging into nucleosomes. Four pairs of core histones have been identified: H2A, H2B, H3, and H4. Modifications of the amino acid tails of these histones have been shown to govern chromatin accessibility to transcription factors, promoting expression of genes best suited to respond to inflammatory signals in the surrounding environment (Lawrence and Natoli, 2011; Ivashkiv, 2013). Of particular note in terms of inflammation, are modifications of histone H3 in macrophages. Thus, while an active transcriptional state is characterized by positive marks such as methylation on lysine-4 and acetylation of lysine-9 on histone H3, a repressed transcriptional state is manifest by trimethylation of H3 lysine-9 and lysine-27, and demethylation of lysine-36 (Kondo et al., 2008; Kapellos and Iqbal, 2016; Wiles and Selker, 2017). Evidence suggests that the addition and removal of methyl and acetyl groups from histone H3 at different locations and times during the inflammatory response is important in regulating macrophage phenotype (Van den Bossche et al., 2014).

A second level of control of macrophage activation is via miRNAs (Curtale et al., 2019). Mature miRNAs are localized within a multiprotein miRNA-induced silencing complex (miRISC). Mechanistically, the miRISC directs mRNA degradation or translational gene silencing by sequestering its targets based on miRNA sequence homology with the 3′ untranslated region (Gebert and MacRae, 2019). This creates a complex system where several miRNAs are capable of targeting a single mRNA and individual miRNAs can target multiple mRNAs (Krek et al., 2005). Accumulating evidence suggests that phenotypic activation of macrophages during the initiation, progression, and resolution of inflammatory responses is regulated by groups of miRNAs and other noncoding RNAs that act in concert to coordinate cellular activity (Piccolo et al., 2017; Xu et al., 2019).

The present studies were aimed at determining if changes in the phenotype of macrophages accumulating in the lung after NM exposure are associated with alterations in histones and miRNAs linked to phenotypic activation of these cells. We found that NM caused modifications of macrophage histones that reflected a poised/open state and increases in miRNAs known to regulate inflammatory and fibrotic pathways. These findings suggest that histone alterations and miRNA expression may be valuable to exploit as pharmacological targets to modulate macrophage responses in the lung during the pathogenic response to NM.

MATERIALS AND METHODS

Animals and treatments

Male Wistar rats (8 wk, 225–250 g) were used in these studies in order to correlate our findings with previous RNA-sequencing analysis of macrophages isolated from male mice treated with NM (Venosa et al., 2019). Rats were purchased from Harlan Laboratories (Indianapolis, IN) and maintained in an AALAC approved animal care facility. Animals were housed in filter top microisolation cages and provided food and water ad libitum. Animals received humane care in compliance with the guidelines outlined in the Guide for the Care and Use of Laboratory Animals, published by the National Institutes of Health. Animals were anesthetized with 2.5% isoflurane, and then administered PBS or NM (0.125 mg/kg, mechlorethamine hydrochloride, Sigma-Aldrich, St. Louis, MO) intratracheally as previously described (Sunil et al., 2011). This dose was used in the current studies as we previously demonstrated that it produces progressive pathologic changes in the lung similar to those observed in humans after exposure to sulfur mustard, a bifunctional alkylating agent and cytotoxic vesicant related to NM (Balali-Mood and Hefazi, 2005; Malaviya et al., 2012; 2015; Sunil et al., 2011). All instillations were performed by the same individual from Rutgers Comparative Medicine Resources. NM was prepared immediately before administration in a designated room under a chemical hood following Rutgers University Environmental Health and Safety guidelines.

Lung cell collection and preparation of nuclear extracts

Animals were euthanized by intraperitoneal injection of pentobarbital (Sleepaway, 50 mg/kg, Fort Dodge Animal Health, Fort Dodge, IA) 1 d, 3 d, 7 d and 28 d after administration of PBS or NM. The lung was removed and 10 mL of ice-cold PBS slowly instilled and withdrawn through a cannula in the trachea while gently massaging the tissue; this procedure was repeated 4 times. The cells were then centrifuged (300 x g, 8 min), resuspended in 10 mL PBS and viable cells enumerated using a hemocytometer with trypan blue. Differential analysis of BAL + massage cells revealed that they consisted of >96% macrophages; the remaining cells were neutrophils and mononuclear cells. All of the cells recovered by BAL + massage were found to stain positively for CD45. For preparation of nuclear extracts, cells (8–10 × 10⁶) were processed using an EpiQuik Nuclear Extraction kit according to the manufacturer’s protocol (EpiGentek Group Inc, Farmingdale, NY). Nuclear proteins were quantified using a BCA protein assay kit (Pierce Biotechnologies Inc., Rockford, IL) with bovine serum albumin as the standard. All samples were assayed in triplicate.

HDAC activity assay

Macrophage histone deacetylase (HDAC) activity was analyzed in nuclear protein extracts (4 μg) using an EpiQuick HDAC kit (Epigentek). All assays were performed in triplicate. A standard curve was generated to convert fluorescence to HDAC activity; data are presented as nanograms of deacetylated substrate/μg protein. As a control, in separate studies the assays were run in the presence of the non-specific HDAC inhibitor trichostatin-A (TSA). TSA was found to block HDAC activity (Supplementary Fig. 1).

Histology and immunohistochemistry

Prior to preparation of tissue sections, the lungs were gently lavaged by slowly instilling and withdrawing 10 mL of ice-cold PBS into the lungs through a cannula inserted into the trachea. In previous studies we demonstrated that this procedure had no effect on lung histology (Venosa et al., 2016). Ten mL of 2% paraformaldehyde in PBS was then instilled into the lungs through the cannula, the lungs removed and fixed in 2% paraformaldehyde for 48 h at 4º C. After washing twice with PBS + 2% sucrose, the lungs were transferred to 70% ethanol, and embedded in paraffin. For immunostaining, random tissue sections (5–6 μm) were deparaffinized with xylene followed by decreasing concentrations of ethanol (100% - 50%) and then water. After antigen retrieval using citrate buffer (10.2 mM sodium citrate, 0.05% Tween 20, pH 6.0, 10 min) and quenching of endogenous peroxidase with 3% hydrogen peroxide in methanol (30 min), sections were incubated for 2 h at room temperature with 10% serum to block nonspecific binding. This was followed by overnight incubation at 4°C in a humidified chamber with primary antibodies (Supplementary Table 1) or the appropriate serum/IgG controls diluted in blocking buffer. Sections were then washed and incubated at room temperature for 30 min with biotinylated secondary antibody (Vectastain Elite ABC kit, Vector Labs, Burlingame, CA). Binding was visualized using a DAB Peroxidase Substrate Kit (Vectastain). Random sections from at least 3 rats were analyzed for each treatment group. Representative sections stained with IgG control antibody are shown in Supplementary Fig. 2.

miRNA isolation, microarray analysis and RT-qPCR

Total RNA was collected from lung macrophages by phenol-chloroform extraction. Samples were enriched for miRNAs using an RNeasy MinElute Cleanup kit (QIAGEN Inc., Valencia, CA). Purified miRNA was reverse transcribed using a QIAGEN QuantiTect Reverse Transcription kit and analyzed using a QIAGEN miScript Rat Inflammatory Response and Autoimmunity miRNA PCR Array (MIRN-105Z) or by RT-qPCR using a QIAGEN miScript SYBR Green kit according to the manufacturer’s protocol. Amplification was performed using a 7300HT Real Time PCR system (Applied Biosystems, Grand Island, NY). Fold changes were calculated using the ∆∆Ct method. miRNA expression was normalized to Snord72 and Rnu6–6p for the PCR array and RT-qPCR assays, respectively. Expression of these miRNAs was not affected by NM administration (Supplementary Fig. 3). Results of the miRNA PCR Array were deposited in NCBI’s Gene Expression Omnibus (Edgar et al., 2002) and are accessible through GEO Series accession number GSE172290 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE172290).

miRNA target profiling

MicroRNA Target Filter in Ingenuity Pathway Analysis (IPA) software version 51963813 was used to predict biological consequences of differential miRNA expression after exposure to NM relative to control (https://www.qiagenbioinformatics.com/products/ ingenuity pathway-analysis) (Kramer et al., 2014). Significantly altered miRNAs were then linked to previously published mRNA-sequencing data deposited in NCBI’s Gene Expression Omnibus and accessible through GEO Series accession number GSE125619 (Venosa et al., 2019). A core analysis was performed on predicted mRNA targets of differentially expressed miRNAs (Log₂(Fold Change) > |1| and p-value < 0.05). Predicted mRNA targets were limited to those identified as: (1) significantly different from control (Log₂(Fold Change) > |1| and FDR-corrected p-value < 0.05) at 1 d and 28 d after NM exposure (Venosa et al., 2019); (2) exhibiting inverse expression profiles (increased miRNA expression coordinate with reduced mRNA expression and vice versa); and (3) involved in Diseases and Functions including ‘Inflammatory Disease’, ‘Inflammatory Response’, and ‘Respiratory Disease’. The Path Explorer function in IPA was used to construct miRNA-mRNA interaction networks to identify highly connected miRNAs representing key upstream regulatory miRNAs. This analysis included both direct and indirect relationships between upstream miRNAs and downstream mRNA targets.

Statistical analysis

HDAC activity and RT-qPCR data were analyzed using one-way ANOVA with Tukey’s post-hoc test. A p-value of < 0.05 was considered statistically significant. miRNA array data were analyzed using the QIAGEN GeneGlobe Data Analysis Center. miRNAs were considered differentially expressed if Log₂(Fold Change) > |1| and p < 0.05. Significantly enriched Diseases and Functions among mRNA targets were determined in Ingenuity Pathway Analysis using the Benjamini-Hochberg approach to correct for multiple comparisons testing.

RESULTS

Effects of NM on histone methylation and acetylation

NM exposure resulted in rapid (within 1 d) and transient evidence of double strand DNA breaks in lung macrophages and some neutrophils, as assessed by expression of the phosphorylated histone variant γH2A.X (Fig. 1). In previous studies, we analyzed lung macrophages isolated from rats treated with NM by RNA sequencing (Venosa et al., 2019). IPA analysis of these data revealed that NM-induced double strand DNA breaks in macrophages were associated with activation of DNA damage response pathways 1 d post-NM; this was evidenced by significant enrichment of p53 signaling [-log(p-value)=2.12], apoptosis signaling [-log(p-value)=3.97], and G2/M DNA damage checkpoint regulation [-log(p-value)=2.31], as well as genes involved in signaling pathways linked to proliferation of mononuclear leukocytes [-log(p-value)=20.84].

Fig. 1. Effects of NM on phosphorylated histone H2A.X (ɣH2A.X).

Tissue sections prepared 1 – 28 d after exposure of rats to NM or PBS control (CTL) were immunostained with antibody to ɣH2A.X. Binding was visualized using a Vectastain kit. Original magnification, 200x.

Representative sections from 3 rats/treatment group are shown.

We next determined if NM exposure resulted in chromatin modifications that regulate gene transcription. In these studies, we analyzed methylation and acetylation of histone H3. Following NM administration, increased numbers of macrophages expressing monomethylated H3K4 (H3K4MM) and dimethylated H3K36 (H3K36DM) were observed at 1 d and 7 d post-exposure (Fig. 2 and Supplementary Fig. 4). Increases in epithelial cells expressing these histone marks were also observed at 7 d and 28 d post-NM exposure. Some lung epithelial cells in control animals also expressed H3K4MM and H3K36DM. We also detected transient upregulation of trimethylated H3K4 (H3K4TM) in both macrophages and epithelial cells 3 d following NM exposure, as well as trimethylated H3K9 in macrophages (Figs. 2 and 3). H3K9 acetylation (H3K9Ac) was also evident in some lung macrophages beginning at 1 d post-NM and persisting for at least 7 d (Fig. 3). Whereas at 28 days post-NM, expression of H3K9Ac was at control levels in macrophages, it was upregulated in some epithelial cells. In contrast to these findings, NM had no effect on total H3 protein (not shown).

Fig. 2. Effects of NM on histone H3 lysine-4.

Tissue sections prepared 1 – 28 d after exposure of rats to NM or PBS control (CTL) were immunostained with antibody to H3 monomethylated lysine-4 (H3K4MM) or H3 trimethylated lysine-4 (H3K4TM). Binding was visualized using a Vectastain kit. Asterisks, macrophages; Arrows, epithelial cells. Original magnification, 200x.

Representative sections from 3 rats/treatment group are shown.

Fig. 3. Effects of NM on histone H3 lysine-9.

Tissue sections prepared 1 – 28 d after exposure of rats to NM or PBS control (CTL) were immunostained with antibody to trimethylated (H3K9TM) or acetylated H3 lysine-9 (H3K9Ac). Binding was visualized using a Vectastain kit. Asterisks, macrophages; Arrows, epithelial cells. Original magnification, 200x. Representative sections from 3 rats/treatment group are shown.

In further studies we assessed whether changes in histone acetylation were associated with alterations in enzymes that acetylate and deacetylate histones. Expression of the histone acetyltransferase (HAT) p300 was upregulated in alveolar macrophages at 1 d, 3 d and 7 d post-NM; this was most pronounced at 3 d and 7 d; subsequently p300 returned to control levels (Fig. 4). Epithelial cells also expressed p300 at 7 d and to a lesser extent 28 d post-NM. The histone deacetylase, HDAC2, was also rapidly upregulated in both macrophages and epithelial cells following NM exposure, most notably at 3 d (Fig. 4). In both cell types, HDAC2 expression returned to baseline levels at 7 d; this was followed by a secondary increase at 28 d. In line with these findings, a biphasic increase in HDAC activity was observed in isolated lung macrophages at 1 d, 3 d, and 28 d after NM exposure (Fig. 5). In contrast, NM had no effect on expression of the NAD-dependent HDAC, sirtuin-1 (data not shown).

Fig. 4. Effects of NM on histone acetylase and histone deacetylase expression.

Tissue sections prepared 1 – 28 d after exposure of rats to NM or PBS control (CTL) were immunostained with antibody to the histone acetylase p300 and histone deacetylase 2 (HDAC2). Binding was visualized using a Vectastain kit. Asterisks, macrophages; Arrows, epithelial cells. Original magnification, 200x. Representative sections from 3 rats/treatment group are shown.

Fig. 5. Effects of NM on macrophages HDAC activity.

HDAC activity was measured spectrophotometrically in nuclear extracts from lung macrophages collected 1 d, 3 d, and 28 d after exposure of rats to NM or to PBS control (CTL). Activity was calculated as ng of substrate deacetylated per minute. Bars, mean ± SE (n = 6–10 rats/treatment group). *Significantly different (p ≤ 0.05) from CTL.

Effects of NM exposure on lung macrophage miRNA expression

We next analyzed NM-induced alterations in macrophage expression of miRNAs associated with acute and chronic inflammation and fibrosis (Supplementary Table 2) (Sonkoly and Pivarcsi, 2009; Vettori et al., 2012). The majority of the 84 miRNAs analyzed in the array were upregulated or unchanged after NM administration at all post-exposure times; the only notable exceptions were miR-20a-5p and miR-20b-5p, which were downregulated at 28 d, however, this response was not statistically significant (Fig. 6 and Supplementary Table 2). Seventeen miRNAs were identified as significantly upregulated relative to PBS controls at all time-points (Log₂(Fold Change) > |1| and p-value < 0.05); these included miR-322–5p, miR-351–5p, miR-125b-5p, miR-222–3p, and miR-182, which are predominantly involved in macrophage anti-inflammatory signaling (Li et al., 2015; Lu et al., 2016; Luo et al., 2017; Zhang et al., 2017; Yang et al., 2020), miR-221 which is pro-inflammatory (Zhao et al., 2016), and miR-125a-5p which has both pro- and anti-inflammatory activity (Graff et al., 2012; Banerjee et al., 2013) (Fig. 6 and Supplementary Table 2). At 1 d post-NM, pro-inflammatory miR-183–5p was identified as significantly upregulated relative to PBS control, while at 3 d post-NM, anti-inflammatory miR325–3p was upregulated (Kim et al., 2020; Sun et al., 2020) (Fig. 6 and Supplementary Table 2). At 7 d post-NM, two miRNAs were upregulated relative to PBS control, miR-743b and miR-878, which have both been implicated in suppressing fibrosis or are negatively correlated with fibrosis progression (Gao et al., 2018; Li et al., 2011). At 28 d post-NM, six miRNA were upregulated in lung macrophages: pro-inflammatory miR-664, anti-inflammatory miR23b-3p, miR-26a and miR-181d, pro-fibrotic miR-181b, and pro-wound repair miR-320 (Li et al., 2020; Lopetuso et al., 2018; Nejad et al., 2018; Zhang et al., 2020; Zheng et al., 2015).

Fig. 6. Effects of NM on macrophage miRNA expression.

Lung macrophages were collected 1 d, 3 d, 7 d, and 28 d after exposure of rats (n = 3 rats/treatment group) to NM or PBS control. miRNA expression was analyzed using a QIAGEN miScript Rat Inflammatory Response and Autoimmunity miRNA PCR Array. Fold changes were calculated relative to PBS controls by the ∆∆Ct method and normalized to SNORD72 endogenous control. Left panel: Heat map of miRNAs differentially expressed (Log₂(Fold Change > |1| and p-value < 0.05) in at least one time-point post-NM relative to CTL (PBS). Red, upregulation; Green, downregulation; Black, no change. Right panel: Venn diagram of differentially expressed miRNAs highlighting unique and overlapping expression profiles. Numbers indicate miRNAs differentially expressed within each time-point.

Select miRNAs identified in the array as upregulated after NM were assessed by RT-qPCR, along with 3 miRNAs (miR-146b, miR-127 and miR-15a) not included in the array but relevant to inflammatory responses (Ying et al., 2015; He et al., 2019; Shang et al., 2019). RT-qPCR expression data were generally consistent with the array data, however, differences in the timing of expression were noted for some of the miRNAs. Thus, as observed in the miRNA array, miR-125b and miR-9 were significantly increased after NM exposure when analyzed by RT-qPCR, while miR-29c was not significantly altered (Fig. 7). Conversely, while expression of let-7a was significantly reduced when analyzed by RT-qPCR, no changes were observed in the array data. Similarly, increased expression of miR-21 was observed in lung macrophages analyzed by RT-qPCR but not by miRNA array. We speculate this is a result of differences in the miRNA isotype analyzed (Choo et al., 2014; Báez-Vega et al., 2016). RT-qPCR analysis also showed that pro-inflammatory miR-15a and miR-127 were upregulated at 1 d and 3 d post NM; miR-15a was also increased at 28 d. In contrast, anti-inflammatory miR-146b was down regulated at 7 d and 28 d.

Fig. 7. RT-qPCR analysis of miRNA expression in macrophages following NM administration.

Lung macrophages were collected 1 d, 3 d, 7 d, and 28 d after exposure of rats to NM or PBS control. miRNA expression was analyzed by RT-qPCR using a QIAGEN miScript SYBR Green kit. Data are presented as fold change relative to CTL. Bars, mean + SE (n = 3–5 rats/treatment group). *Significantly different (p ≤ 0.05) from CTL.

In further studies, we used Ingenuity Pathway Analysis (IPA) software to infer biological consequences of changes in miRNA expression profiles in macrophages by identifying significant enrichment of Diseases and Functions among their downstream mRNA targets. In these studies, we cross-referenced previously published RNA-seq data on lung macrophages collected from rats 1 d and 28 d post-NM exposure (Venosa et al., 2019) with our miRNA array data using MicroRNA Target Filter analysis to connect differentially expressed miRNAs and mRNAs. Subsequently, IPA Core Analysis was performed to elucidate pathways regulated by predicted mRNA targets of miRNAs identified as upregulated in the present analysis (Log₂(Fold Change) > |1| and p-value < 0.05). We found that predicted mRNA targets were associated with Diseases and Functions related to cellular movement, inflammation, and lipid metabolism (Fig. 8). We then performed analyses to identify miRNAs highly connected to mRNAs regulating the enriched Diseases and Functions pathways. At 1 d post-NM, miR-27a and miR-34c shared the greatest number of relationships with differentially expressed mRNAs suggesting that they are key regulators; this was followed by miR221-3p, miR9a-5p and let7b-3p (Fig. 9). miR-125b, miR-30c, miR-19b, miR-148b, and miR-16 were predicted as major regulators of mRNAs involved in the enriched Diseases and Functions at 28 d post-NM, a time coordinate with increases in their expression (Fig. 10).

Fig. 8. Ingenuity Pathway Analysis (IPA) of miRNAs.

An IPA Core Analysis was performed on predicted mRNA targets of differentially expressed miRNAs at 1 d and 28 d post-NM. The mRNA targets were identified as differentially expressed (Log₂(Fold Change) > |1| and FDR-corrected p-value < 0.05) in a previous RNA-seq analysis [20]. mRNAs involved in ‘Inflammatory Disease’, ‘Inflammatory Response’, and ‘Respiratory Disease’ pathways, that exhibited inverse expression compared to upstream miRNAs were then selected for further analysis. Significantly enriched Diseases and Functions (p < 0.05) among predicted mRNA targets of differentially expressed miRNAs. Bars, enrichment [-log(p-value)] of Diseases and Functions 1 d (top) and 28 d (bottom) post-NM exposure.

Fig. 9. Association between miRNAs and mRNAs identified as significantly enriched in Diseases and Functions pathways 1 d post-NM.

IPA was used to identify miRNAs that are central regulators of differentially expressed mRNAs in the significantly enriched Diseases and Functions pathways. Differentially expressed mRNAs (Log₂(Fold Change) > |1| and FDR-corrected p-value < 0.05)) in lung macrophages 1 d post-NM identified in a previous RNA-seq analysis were connected to differentially expressed miRNAs (Log₂(Fold Change) > |1| and p-value < 0.05)) in their respective time-points using the Path Explorer function in IPA. Direct (solid lines) and indirect (dashed lines) relationships were included. Red, upregulation; green, downregulation.

Fig. 10. Association between miRNAs and mRNAs identified as significantly enriched in Diseases and Functions pathways 28 d post-NM.

IPA was used to identify miRNAs that are central regulators of differentially expressed mRNAs in the significantly enriched Diseases and Functions pathways. Differentially expressed mRNAs (fold change > |2| and FDR-correct p-value < 0.05) in lung macrophages 28 d post-NM identified in a previous RNA-seq analysis were connected to differentially expressed miRNAs (Log₂(Fold Change) > |1| and p-value < 0.05)) in their respective time-points using the Path Explorer function in IPA. Direct (solid lines) and indirect (dashed lines) relationships were included. Red, upregulation; green, downregulation.

DISCUSSION

It is now well established that changes in DNA methylation, histone marks and expression of non-coding RNA are hallmarks of inflammation driven diseases (Stylianou, 2018). Each of these regulatory pathways play a role in fine tuning macrophage gene expression and function (Davis and Gallagher, 2019). Previous studies have reported increases in serum levels of a functionally broad group of miRNAs in individuals exposed to sulfur mustard, a related mustard vesicant (Gharbi et al., 2018; Salimi et al., 2019). Alterations in histone status and DNA methylation have also been described in rodent skin and in cultured endothelial cells treated with sulfur mustard, (Steinritz et al., 2016; Simons et al., 2018). The present studies demonstrate that pulmonary exposure of rats to NM causes dynamic changes in lung macrophage histones and in expression of miRNAs during acute injury, tissue remodeling and fibrogenesis. These findings provide clues on pathways regulating macrophage activation and may help identify novel targets for therapeutic intervention. This is supported by our earlier findings that the histone deacetylase inhibitor, valproic acid, reduced lung injury and oxidative stress induced by NM in rats, a response associated with decreases in inflammatory macrophage activation (Venosa et al., 2017).

NM is known to cause DNA alkylation leading to single and double strand DNA breaks (Boldogh et al., 2003). Double strand breaks have also been observed in activated proinflammatory macrophages as a consequence of reactive oxygen species production (Pereira-Lopes et al., 2015). Between 2.5–25% of total histone H2A in the mammalian genome is present as the H2A.X variant (Rogakou et al., 1998). Phosphorylation of serine 139 in H2A.X (γH2A.X) is a marker of double strand DNA breaks (Kuo and Yang, 2008; Redon et al., 2012). Increases in γH2A.X is associated with activation of the DNA damage response pathway, characterized by rises in cell cycle checkpoints and DNA repair pathways, and alterations in signaling pathways controlling proliferation (Giglia-Mari et al., 2011; Mazouzi et al., 2016). Our findings of increases in γH2A.X and signaling pathways associated with p53, apoptosis and proliferation in macrophages 1 d post-NM indicate that these cells are directly targeted by NM and that this is associated with activation of the DNA damage response (Venosa et al., 2019; Smith et al., 2020). This may be an initiating factor driving their pro-inflammatory activity (Rodier et al., 2009; Rodier et al., 2011; Sharma et al., 2012). The macrophage DNA damage response may also be important in countering the toxic effects of mustards by reducing intracellular oxidative stress and promoting survival (Polo and Jackson, 2011; Van Houten et al., 2018). Interestingly, γH2A.X was not detected in epithelial cells despite evidence of NM-induced injury to these cells. It is possible that DNA damage occurs more rapidly in epithelial cells following NM exposure. In this regard, previous studies have demonstrated that NM upregulates γH2A.X in cultured epithelial cells within 2 h of exposure (Jan et al., 2019).

Post-translational modifications of histone tails, including methylation and acetylation, coordinate chromatin accessibility for transcription and consequent gene expression (Jenuwein and Allis, 2001). In macrophages, methylation and acetylation of H3 are associated with M1 macrophage activation and increases in IL-1β, TNFα and iNOS (Daskalaki et al., 2018; Davis and Gallagher, 2019). H3K4 trimethylation and H3K9 acetylation have also been linked to activation of NF-κB (Lecoeur et al., 2020). Following NM exposure, rapid (1 d - 3 d) increases in numbers of macrophages expressing H3K4MM, H3K4TM and H3K36DM, as well as H3K9Ac were observed. Expression of these histone marks correlated with pro-inflammatory macrophage activation, as evidenced by increased NF-κB signaling and upregulation of Tnfa, Nos2, Ptgs2, and Il1b (Venosa et al., 2019; Smith et al., 2020). Our findings of increases in H3 methylation and acetylation in macrophages responding to NM are consistent with reports linking these histone marks to pro-inflammatory and antioxidant responses in the lung and/or peripheral blood monocytes following exposure of humans or rodents to traffic related air pollution (Ding et al., 2017; Zheng et al., 2017). At 28 d post-NM, expression of H3K4TM, H3K36DM, and H3K9Ac were largely at control levels, consistent with a decline in macrophage M1 pro-inflammatory activation and a shift to an M2 phenotype (Venosa et al., 2016). It remains to be determined if changes in histone marks observed in the present studies are directly linked to macrophage phenotype and function following NM exposure.

Interestingly, some alveolar epithelial cells were found to constitutively express H3K4MM and H3K36DM, a response that increased at 7 d and 28 d following NM exposure. Additionally, increases in H3K4TM and H3K9Ac were observed in epithelial cells at 3 d and at 28 d, respectively. Epithelial cells are known to contribute to inflammatory and fibrotic responses of the lung to pulmonary toxicants, including ozone (Yang et al., 2013; Cooper and Loxham, 2019). In line with this, previous studies have demonstrated that basal expression levels of methylated H3 in lung epithelial cells are correlated with individual susceptibility to ozone-induced increases in inflammatory mediator production by these cells (McCullough et al., 2016). Further studies are required to determine if epithelial cells are also involved in the pathophysiological responses to NM, and if this is due to histone modifications.

HATs catalyze the acetylation of histones, while HDACs are important in histone deacetylation (Peserico and Simone, 2011; Hoeksema and de Winther, 2016). While a balance between HAT/HDAC function maintains homeostasis, increases in HAT activity and hyperacetylation have been reported to enhance chromatin accessibility in macrophages resulting in excessive pro-inflammatory activation, aberrant wound repair, and fibrosis (Ishii et al., 2009; Ogiwara et al., 2011; Peserico and Simone, 2011). Treatment of rats with NM resulted in increases in HAT p300 in macrophages at 3 d post-exposure; this correlated with upregulation in H3K9Ac. Earlier studies have shown that p300 can promote transcriptional activation of M1 genes (Tnfa, Il6, Ifna) in macrophages (Hu et al., 2017; Tikhanovich et al., 2017; Daskalaki et al., 2018). We speculate that p300 plays a role in pro-inflammatory macrophage activation in the lung early after NM exposure. We also found that p300 was upregulated in lung epithelial cells following NM exposure. p300 has been implicated in NF-κB activation and IL-8/CXCL8 expression in human lung epithelial cells (Osterlund et al., 2005; Lin et al., 2020). Studies are in progress to determine if p300 activates these signaling pathways in lung epithelial cells following NM exposure and if this contributes to tissue injury.

NM exposure was also associated with increases in HDAC2 expression and activity in lung macrophages at 1 d, 3 d and 28 d. HDACs have been shown to be essential for optimal regulation of a number of macrophage genes involved in host defense (Roger et al., 2011). Increases in HDAC2 in macrophages may represent a compensatory attempt to limit excessive inflammatory responses due to hyperacetylation of macrophage histones at 3 d and 28 d post-NM exposure. HDAC2 has been positively correlated with both inflammation/alveolitis and epithelial bronchiolization/fibrosis in experimental models of bleomycin-induced fibrosis and in patients with idiopathic pulmonary fibrosis (Korfei et al., 2015; Li et al., 2017). It may be that HDAC2 plays a role in both of these pathogenic responses to NM in the lung.

Histone modifications and miRNAs work in a coordinated manner to regulate macrophage gene expression and inflammatory responses, with active promoter regions (e.g., enriched in H3K4DM and H3K4TM) exhibiting higher degrees of miRNA targeting (Cao et al., 2016; Tao et al., 2017; Curtale et al., 2019). In line with this, we found that NM exposure was associated with differential upregulation of miRNAs over time, consistent with the idea that various stages of the inflammatory response are associated with distinct miRNA expression profiles (Nejad et al., 2018). Of note, over the time-course of our studies, we observed upregulation of pro- (e.g., miR-221, miR-183-5p, miR-664) and anti-inflammatory (e.g., miR-322-5p, miR-351-5p, miR-125b-5p, miR-222-3p, miR-182, miR-325-3p, miR-26a-5p, and miR-181d-5p), and pro- (e.g., miR-181b-5p) and anti-fibrotic (e.g., miR-743b-3p and miR-878) miRNAs, as well as miRNAs (e.g., miR-125a-5p) that act on signaling pathways involved in both pro- and anti-fibrotic activity (Li et al., 2011; Su et al., 2015; Chen et al., 2017; Huang et al., 2018; Lin et al., 2019; Fu et al., 2020; Yang et al., 2017; Guiot et al., 2020). Moreover, while some of these miRNAs were upregulated at only at one post-NM exposure time point (e.g., miR-183-5p at 1 d, miR-325-3p at 3 d, miR-743b-3p and miR-878 7 d, and miR-181b-5p, miR-181d-5p, miR-26a-5p, miR-321-3p, and miR-664-3p at 28 d), the majority were upregulated at multiple time points. These findings highlight the importance of miRNAs in balancing inflammatory responses and the complexity of their interactions (Nejad et al., 2018; Locati et al., 2020).

To assess the potential function of miRNAs upregulated in lung macrophages following NM exposure, we used IPA MicroRNA Target Filter analysis to connect them to differentially expressed mRNAs previously identified by RNA-seq (Venosa et al., 2019). We then performed an IPA Core Analysis to identify biological pathways significantly enriched among the predicted mRNA targets. This analysis showed that pathways involved in cell viability, necrosis, apoptosis and cell cycle progression were enriched at both 1 d and 28 d post-NM, consistent with a prolonged DNA damage response. Our findings also suggest that multiple miRNA-mRNA pathways are involved in regulating the DNA damage response in macrophages following NM exposure (Jowsey et al., 2012; Jowsey and Blain, 2015). Diseases and Functions related to inflammation (Chronic inflammatory disorder, Inflammation of organ) were significantly enriched at both 1 d and 28 d post-NM, along with pathways related to adhesion and migration of cells. These data are in accord with increases in inflammatory cells in the lung at these NM postexposure times (Venosa et al., 2015; Venosa et al., 2016; Sunil et al., 2020). Our miRNA-mRNA interaction network analysis also identified miRNAs highly connected to mRNAs involved in both promoting and resolving inflammation. These included pro-inflammatory miR-221-3p and let-7b-5p and anti-inflammatory miR-125b, miR-16, and miR-34c at 1 d post-NM, and anti-inflammatory miR-125b, miR-30c, and miR-16, as well as miR-19b which possesses both pro- and anti-inflammatory activity, at 28 d post NM (Ceolotto et al., 2017; Lv et al., 2020; Quero et al., 2020; Teng et al., 2013; Xu et al., 2016; Yamada et al., 2020; Ye et al., 2017; Zhu et al., 2019). Anti-inflammatory miR-16, miR-125b and miR-34c (Xu et al., 2016; Yamada et al., 2020; Zhu et al., 2019) were identified as highly connected at both 1 d and 28 d post NM. These findings support the idea that a regulatory network of miRNAs controls the outcome of the inflammatory response to tissue injury (Tahamtan et al., 2018).

Our pathway analysis also showed that miRNAs predicted to target mRNAs involved in lipid metabolism were also upregulated at 1 d (e.g., miR-27a and miR-29a) and 28 d (e.g., miR-27a, miR-30c, miR-148b, and miR-19b) after NM exposure (Irani et al., 2015; Cruz-Gil et al., 2018; Zhou et al., 2020; Gil-Zamorano et al., 2014; Massart et al., 2017; Miranda et al., 2018; Yao et al., 2017). We previously demonstrated that the pathogenesis of NM-induced lung fibrosis is associated with altered lipid signaling and the appearance of foamy macrophages in the lung in areas adjacent to fibrotic foci (Venosa et al., 2019). We speculate that these miRNAs are involved in dyslipidemia and foam cell formation during the progression of fibrosis; however, this remains to be determined.

Regulation of macrophage gene expression represents an attractive therapeutic target for treatment of inflammatory diseases (Nebbioso et al., 2012; Nebbioso et al., 2018; Jones et al., 2019). The present studies identify temporal changes in chromatin accessibility in macrophages following NM exposure. This was correlated with high pro-inflammatory activity during early injury, followed by a less transcriptionally favorable state during tissue remodeling/fibrogenesis. In addition, our macrophage miRNA assessment and mRNA-miRNA interaction network analysis identified specific signaling pathways associated with macrophage recruitment and activation at 1 d, fibrosis at 28 d, and lipid metabolism at both 1 d and 28 d. Taken together, these data provide important new insights into the complexity of the transcriptional and post transcriptional machinery involved in regulating macrophage responses to vesicant-induced lung injury, which may be useful in the development of countermeasures aimed at mitigating toxicity and disease pathogenesis.

Supplementary Material

1

Supplementary material 1

2

Supplementary material 2

Highlights.

• Nitrogen mustard (NM) exposure activates the DNA damage response in

• macrophages

• NM causes early histone modifications linked to transcriptional activation

• Distinct macrophage miRNAs are upregulated during acute injury and fibrosis after NM

• Predicted mRNA targets of miRNAs are involved in inflammation and lipid metabolism