High-mobility group N proteins are required for regulation of antigen-presentation genes in macrophages

Abstract

The expression, localization, and secretion of certain chromatin proteins are tightly regulated in response to inflammatory challenges or tissue damage. High mobility group (HMG) nucleosome-binding proteins can translocate out of the nucleus to act as alarmins in immune cell communications. The consequences of nuclear depletion of HMGs in genome organization and gene regulation are not understood in the context of innate immunity. Here, we report alterations of tissue-specific macrophage transcriptome using a genetic knockout (KO) of HMGN1 and HMGN2, an HMGN double knockout mouse line. Lack of both HMGN proteins disrupted the expression of key macrophage genes, including those encoding MHCs, M-CSF, and ApoE, as well as acute responses to LPS, in a sex-dependent manner. These transcriptional effects were associated with concomitant changes in chromatin accessibility, as profiled by an assay for transposase accessible chromatin by sequencing, in regulatory sites distal and proximal to the corresponding genes. These findings highlight the importance of macrophage HMGNs in maintaining tissue-specific gene expression and supporting inflammatory responses.

Introduction

High mobility group (HMG) proteins are abundant chromatin-binding proteins that bind nucleosomes in a DNA sequence–nonspecific manner.^1,2 Previous studies have documented their roles in maintaining chromatin landscapes that support cell identity or regulate cellular differentiation.^3–6 Despite their constitutive presence in the nucleus, the HMG proteins can also translocate to the cytoplasm and be released into the extracellular space. Because this has often been observed in inflammatory settings, the secreted HMGs are classified as alarmins in cell-cell communications.⁷ This remarkable property has been documented for an increasing list of HMG proteins, including HMGB1, HMGB2, and HMGN1.^8–10 The release of HMGs induces their transient depletion from the epigenome in the HMG-secreting cells. Consequently, this phenomenon has dual effects: (i) an immediate impact on the epigenome of the alarmin-sending cells and (ii) the signaling response of the alarmin-receiving cells. The latter aspect has gained much attention in recent years. For example, HMGBs mediate neuroinflammatory signaling in the microglia of experimental autoimmune encephalomyelitis mice.¹¹

However, the chromatin impact from the reduction of HMGs in the alarmin-sending cells is still not understood. Relevant cell types for addressing this issue include macrophages, which are often implicated as the source immune cells secreting HMGs.^12,13 Some data from a few nonimmune cell types indicate that the reduction of HMG proteins in the nucleus during acute inflammatory responses leads to decreased core histone content and a consequent global effect on transcription.^14,15 Moreover, the depletion of HMGs has been associated with inflammatory signaling and cellular senescence.¹⁶ In particular, HMGB2 depletion may occur early in cellular senescence, and it disrupts genome organization.¹⁷ Such altered states of chromatin landscape can occur in otherwise healthy cells where HMGs translocate from the nucleus to the cytoplasm,¹⁴ because they do not require rare genetic events such as oncogenic mutations affecting chromatin regulators. Therefore, it is critically important to understand the effects of altered nuclear abundance of HMG in acute or chronic inflammatory conditions.

Here, we examined the macrophages from HMGN1 and HMGN2 double knockout (HMGN DKO) mice and determined the altered transcriptomic and open chromatin profiles in the absence of both HMGNs. The gene expression profiles of 3 macrophage populations revealed a striking elevation of MHC class I and other antigen-presentation genes. Differentially expressed genes (DEGs) showed sex-specific patterns. In bone marrow–derived macrophages (BMDMs) from HMGN DKO mice, the DEGs were preferentially located near putative enhancers that were concomitantly altered in their chromatin accessibility in HMGN DKO. These results describe a key gene expression program of tissue-resident macrophages that requires HMGNs, suggesting that HMGN-secreting macrophages may alter their phenotype through the reprogramming genes involved in antigen presentation.

Materials and methods

Mice

All animals were maintained under specific pathogen–free conditions at the animal facility of the National Institute on Aging (NIA) of the National Institutes of Health (NIH). All the animal care and procedures in this study were carried out in accordance with the guidelines of NIH and approved by the NIA Animal Care and Use Committee.

Tissue and cell extraction

Bone marrow–derived macrophage

The femur and tibia were removed from mice. Bone marrow was flushed from each bone using a 27-gauge needle and collected. The cell suspension was passed through a 70-μm cell strainer and collected via centrifugation. Cells were plated in DMEM supplemented with 10% FBS and M-CSF (60 ng/mL) for 6 days. In RNA sequencing (RNA-seq) experiments, on day 6, BMDMs were plated at 1 million cells/well in a 6-well plate. On the following day, cells were treated with 10 ng/mL LPS and/or harvested as indicated in each experiment.

Peritoneal cavity–resident macrophage isolation

Peritoneal lavage was performed with 5 mL of freshly prepared, room temperature, 0.2-µm sterile-filtered PBS containing 2 mM EDTA and 1% FBS, per animal. All cells for each sex and genotype group were collected, pooled, washed, and resuspended in cold DMEM supplemented with 10% FBS and 2 mM EDTA. Blocking was performed by adding 20 µg/mL anti-CD16/32 (Biolegend, 101301) and incubated for 15 min on ice in the dark. Subsequently, 1.24 µg/mL anti-F4/80 BV421 (Biolegend, 123137) and anti-CD11b PE (Biolegend, 101208) were added, followed by an incubation in the dark for 30 min. Cells were washed and resuspended in 500 µL of cold DMEM supplemented with 10% FBS and 2 mM EDTA on ice. ToPro3 (ThermoFisher Scientific, T3605) was added at a dilution of 1:1,000 prior to sorting to gate out dead cells. CD11b^hi F4/80^hi cells were sorted into 1 mL of DMEM supplemented with 20% FBS. Approximately 0.5 × 10⁶ cells were subsequently cultured in DMEM supplemented with 10% FBS in 6-well tissue culture dishes overnight before proceeding to LPS stimulation.

Microglia isolation

Whole brains were isolated from adult mice after animal euthanization according to institutional regulations. Tissue dissociation was performed using Miltenyi Biotec’s mouse adult brain dissociation kit (catalog no. 130-107-677) according to the manufacturer’s protocol. Microglia were purified from total collected cells using CD11b magnetic beads (Mitenyi Biotec, 130-049-601) on columns (Miltenyi Biotec, 130-042-201). The percentage of microglia among cells was determined by CD45 (Miltenyi Biotec, 130-110-632) and CX3CR1 (Biolegend, 149039) double-positive cells via flow cytometry. Purity was ascertained to be greater than 90%. Total cells from 3 mice were pooled and used per group, and cells were split between the indicated conditions. Equal numbers of male and female mice were used, and 2 biological replicates were performed for each sex. Seeding density was ∼0.45 × 10⁶ cells/1.9 cm² (24-well tissue culture–treated dish) with DMEM-F12 in the presence of M-CSF (10 ng/mL) and TGF-β (50 ng/mL). Cells were cultured for 72 h before subsequent processing. Culture medium was supplemented with 10% FBS and penicillin/streptomycin (1,000 units/mL and 1,000 µg/mL, respectively).

RNA-seq analysis

Adherent cells were frozen in TRIzol and subjected to a freeze-thaw cycle prior to RNA isolation. Total RNA was isolated from TRIzol/cell lysate mixture by using an RNeasy Mini Kit (Qiagen, 74104). Total RNA was sequenced on an Illumina NextSeq using Illumina TruSeq Stranded Total RNA Library Prep and paired-end sequencing at the National Cancer Institute Center for Cancer Research Sequencing Facility.

The stranded paired-end 75 bp reads were mapped to a mouse reference genome (GRCm38, mm10)¹⁸ with the STAR, version 2.5.3a, mapper¹⁹ using the Gencode M14 primary assembly annotation.²⁰ The reads were counted per gene using RSEM, version 1.3.0.²¹ Before further analysis, we filtered out the following genes: (i) short genes (< 200 bp); (ii) ribosomal, transporter RNA genes and other small RNA genes; and (iii) genes from unplaced chromosomes. We also required a gene to be expressed at 1 fragment per kilobase of transcript per million mapped reads (FPKM) level in at least 1 sample. Calculated FPKM values were used to confirm reproducibility between replicates.

The count data were then used for differential expression analysis with edgeR, version 3.18.1,²² in R, version 3.4.2 (R Core Team, 2017). Read counts were normalized using the negative binomial generalized linear model distributions for each sample. A detectable batch effect was caused by using multiple flow cells in the sequencing; therefore, this factor was included in our gene-wise negative binomial generalized linear model fit (∼ flow cell + group) before testing for differential expression. Differential expression was determined pairwise for each comparison by likelihood ratio tests using contrasts. A gene was considered differentially expressed if the log2 fold change (log2FC) was ≥ 0.8 and the false discovery rate (FDR) was < 0.05. The partitioning around medoids algorithm was used to draw heat maps of gene clusters with the ComplexHeatmaps package in R.²³

Gene set enrichment analysis was performed in R (version 4.4.3) using clusterProfiler (version 4.14.6).²⁴ Gene Ontology (GO) biological process over-representation analysis used GO annotations from org.Mm.eg.db (version 3.20.0). Enriched GO terms were identified using a significance threshold of P < 0.01 and FDR < 0.05. Redundant GO terms were removed using the simplify function in clusterProfiler with a semantic similarity cutoff of 0.7. The top 5 or more most significant GO terms (lowest adjusted P values) from each group were visualized using the dot plot function from enrichplot (version 1.26.6).

In total, 1,693 genes were identified as upregulated DEGs and 2,029 genes as downregulated DEGs. After removing duplicate gene identifiers, 1,956 unique DEGs were identified. Sex-specific DEGs were defined as genes showing differential expression exclusively in 1 sex, with an absolute log2FC > 1. Based on this criterion, 1,310 genes were classified as sex-specific DEGs, and the remaining 646 genes were classified as shared DEGs. Heat maps of these 2 gene sets were generated using the ComplexHeatmap package (version 2.22.0). For each heat map, genes were grouped into 9 clusters using the partitioning around medoids clustering algorithm. The 3 genes with the highest absolute z scores within each cluster were annotated as representative genes.

An UpSet plot was generated using the UpSetR package (version 1.4.0) to illustrate the overlap of DEGs between DKO and wild type (WT) among different cell types and sex (Fig. S1). The left horizontal bars indicate the total number of DEGs in each group (cell type and sex). The height of each vertical bar represents the number of genes shared among the indicated comparisons (filled connected dots below the vertical bars). Only 1 filled dot represents the DEGs that are exclusive to the corresponding group.

Assay for transposase accessible chromatin by sequencing analysis

The assay for transposase accessible chromatin by sequencing (ATAC-seq) was performed following methods previously described.²⁵ After harvesting, 50,000 cells were pelleted at 500 g for 5 min, washed with 50 μL of cold 1× PBS, and centrifuged at 500 g for 5 min. Cells were lysed using cold lysis buffer (10 mM Tris-Cl, pH 7.4, 10 mM NaCl, 3 mM MgCl₂, and 0.1% IGEPAL CA-630). Immediately after lysis, nuclei were centrifuged at 500 g for 20 min at 4 °C. The nuclei were then resuspended in 50 µL of transposase reaction mix containing 2.5 µL of Tn5 transposase, 25 µL of TD buffer, and 22.5 µL of nuclease-free water (Nextera Sample preparation kit from Illumina). The transposition reaction was incubated for 30 min at 37 °C. Tagmented DNA was purified using a Qiagen MinElute kit, followed by PCR library amplification. Libraries were amplified for a total of 7 to 10 PCR cycles. Amplified libraries were again purified to remove excess primers and nucleotides. Sequencing of libraries was carried out using 76-bp paired-end sequencing on an Illumina NextSeq at the NCI Center for Cancer Research Sequencing Facility. All ATAC-seq experiments were performed with 2 biological replicates, and each biological replicate included 2 technical replicates.

The ATAC-seq paired-end 75 bp reads were first trimmed with cutadapt, version 1.18, with the following options: -j 6 -b file: adapters.fa -B adapters.fa—nextseq-trim = 2—trim-n -n 5 -O 5 -q 10,10 -m 35:35, and then mapped to mouse reference genome (GRCm38, mm10)¹⁸ with the bowtie2, version 2.26, mapper.²⁶ The mapped reads were filtered with samtools, version 1.9,²⁷ with -q 10 –F 2828 and -f 3 options to get uniquely mapped reads.

ATAC-seq peak regions of each sample were called using MACS2 with parameters: nomodel—shift -75—extsize 150 (https://github.com/ENCODE-DCC/atac-seq-pipeline).²⁸ Blacklisted regions were excluded from the peaks called. We generated a consensus set of unique peaks by merging ATAC-seq peaks. In total, we identified 84,118 peaks. We annotated the peaks using Homer.²⁹ Counts for peaks were generated using featureCounts from the Subread package.³⁰

We assessed the significant change of chromatin accessibility between different groups using edgeR.²² The significantly changed peak was identified if the peak showed log2FC > 0.5 and FDR < 0.05. Gene set enrichment analysis was performed using clusterProfiler, version 3.6.0.²⁴ MEME-ChIP (http://meme.suite.org/tools/meme-chip) was used for motif analysis with default setting.

More and less accessible ATAC-seq peaks were stratified into 4 groups according to their FC values (<2×, 2× to 5×, and >5×). Genomic regions corresponding to the peaks in each group were annotated, and their distributions across genomic features were visualized as histograms using ggplot2 (version 4.0.2). The distributions of these 4 peak groups were further examined and plotted within each cell type, sex, and LPS treatment group.

Results

Macrophage gene expression programs are perturbed in multiple tissues of HMGN DKOs

Because macrophages reside in all major tissues in vivo and maintain the host tissues against infectious agents or damage, we isolated macrophages from multiple sources to compare HMGN DKO and control mice. We cultured BMDMs, peritoneal macrophages, and microglia from the bone marrow, peritoneal cavity, and brain, respectively. RNA-seq analysis was applied to assess the gene expression profiles of the HMGN DKO macrophages in the basal state and in response to the inflammatory stimulus, LPS (Fig. 1A, Data Files S1–S3). Three hours and 8 hours after LPS treatment were selected for early and late response time points, respectively. Principal component analysis indicated the gene expression profiles were most affected by the resident tissue, followed by LPS stimulation status (Fig. 1B). This is because the effect size of genotype-specific expression changes is modest and thus dominated by the much larger size of cell type–specific effects and LPS-responsive changes. Despite the relatively modest magnitude of genotype effects, numerous genes were differentially expressed in the basal state in a tissue- and sex-specific manner (Fig. 1C). DEGs in HMGN DKO versus WT in basal conditions depended on sex and cell type (Fig. S1A, B). In particular, 79% of DEGs in basal female peritoneal macrophages, 59% of DEGs in basal female BMDMs, and 57% of DEGs in basal male BMDMs were unique to each group. Only 21 DEGs were common in all 6 groups, including the upregulated H2-M2, Pttg1, Rprl3, H2-T22, H2-T23, and the downregulated H2-T10 and Lama3. However, the direction of changes in DKO versus WT was largely maintained, with only a small fraction of genes showing changes between groups (Fig. S1C). Female peritoneal macrophages also differed from the other cell types: Hmgn1 and Hmgn2 expression was lower than male cells in WT. On the other hand, the proportion of up- and downregulated genes in response to LPS stimulation remained comparable in HMGN DKO mice (Fig. S2, Data Files S4–S7).

Figure 1.

Transcriptome-wide changes in macrophage populations in the absence of HMGN1 and HMGN2. (A) RNA-seq analysis of 3 macrophage types along the time course of responses to LPS. (B) Principal component (PC) analysis of the RNA-seq data based on log2 (FPKM + 0.1). (C) MA plots showing the DEGs in HMGN DKO versus WT. Red, upregulated DEGs; blue, downregulated DEGs; CPM, count per million. For peritoneal macrophages and microglia, 8-h LPS samples were not prepared, due to insufficient cell yields.

The combined set of DEGs in HMGN DKO versus WT from each cell type was subjected to cluster analysis (Fig. 2, Fig. S3). The heat map confirms principal component analysis in the sense that gene clusters primarily reflect the tissue of residence (in Fig. 2, “High in BMDM,” “High in Microglia,” and “High in Peritoneal”). Other genes had expression patterns mostly dominated by LPS regulation (in Fig. 2, “LPS induced,” and “LPS repressed”). These cell type–specific or LPS-regulated genes had relatively subtle alterations in the HMGN DKO, which are not easily appreciable in the heat map visualization. Only a small cluster of genes were strongly downregulated across all 3 cell types (“DKO repressed” in Fig. 2); these included Apoc2, Apoe, H2-T10, Dmpk, C5ar1, Stard13, Wdfy1, and Rab4a. Differential gene expression of representative DEGs was confirmed by quantitative RT-PCR (Fig. S4).

Figure 2.

DEGs in HMGN DKO versus WT show gene clusters based on cell type and LPS responses. A total of 1,693 upregulated DEGs and 2,029 downregulated DEGs were obtained by comparing HMGN DKO and WT samples in each condition or cell type. The expression data of the DEGs were subject to cluster analysis (see Materials and methods). Each of the 8 clusters is annotated based on the predominant pattern of gene expression exhibited by the constituent genes. Highlighted genes were selected manually for their important immune function.

Antigen-presentation and cytokine-related genes are differentially expressed in the absence of HMGN1 and HMGN2

To examine how HMGN1/2 deficiency affects different cell types in the basal state, we performed GO analysis to assess the extent of shared or distinct gene alterations at a higher level. Despite the predominantly cell type–specific disruption of individual genes (Fig. 2), numerous shared GO)classes were commonly upregulated in all 3 macrophage types (Fig. 3A, Figs. S5 and S6). Notably, shared ontology classes were related to antigen processing and presentation or T-cell immunity. Although these ontology classes were over-represented, the identity of individual genes that showed higher expression in the DKO versus WT varied between cell types and sexes (Fig. S6). For example, Il1b and Cd74 were upregulated only in BMDMs and male peritoneal macrophages, respectively, but both are associated with the adaptive immune response class.

Figure 3.

The antigen-presentation pathway is significantly altered in the basal state of macrophages from HMGN DKO. DEGs from the unstimulated samples (LPS 0 h) were subject to a GO analysis for upregulated genes (A) and downregulated genes (B). Only the groups showing significant enrichment are included in the dot plots. p.adjust, adjusted P value.

In contrast, genes downregulated in HMGN DKO shared few common GO classes, and the GO classes were often not shared even between the sexes within the same cell type (Fig. 3B, Fig. S5 and S7). The lack of HMGN1/2 resulted in reduced expression of many cytokines and their receptors in BMDMs in a sex-dependent manner, such as Il12b, Il10, Ccl22, Cxcl12, Cxcl14, Ccr2, Ccr7, and Cx3cr1 (Fig. S7). Apoe, important for macrophage lipid metabolism and phagocytosis, was a notable exception in that it was downregulated in BMDMs and female peritoneal macrophages.

Given the reported changes of HMGs and chromatin proteins in aging,^31–35 we next asked whether some of the HMGN-dependent genes may also show age-associated patterns of expression. To address this, we compared the DEGs from HMGN DKO against our previous transcriptomic dataset for BMDMs, peritoneal macrophages, and microglia from WT mice at defined ages.³⁶ On a global level, there was minimal overlap in the DEG lists between the old versus young comparison and the HMGN DKO versus WT comparison. However, we noted some genes had concordant patterns of expression changes. For example, Wdfy1 and Rab4a, which are downregulated in aging, were also lower in the DKO versus WT. Meanwhile, several immunoglobulin-encoding genes, especially in chromosome 12, such as Ighm, had elevated expression in BMDMs with aging as well as in male DKO. However, there were also genes that had nonconcordant patterns. Examples include Cxcl5, Cxcl12, Csf1, Hist2h3c2, which exhibited elevated expression in BMDMs from old animals but showed reduced expression in BMDMs from HMGN DKO, especially in female mice. On the other hand, Marco expression was lower in BMDMs from old mice but higher in BMDMs from DKO.

Overall, a predominant effect in HMGN DKO was the upregulation of the antigen-presentation pathway, such as MHC class I genes, which encode cell surface proteins responsible for presenting peptide antigens to cytotoxic T lymphocytes.

DEGs have altered chromatin accessibility of proximal and distal enhancers

To gain further insight about the impact of HMGN DKO, we performed ATAC-seq analysis on the macrophage populations. Among the 3 cell types, BMDMs yielded a high-quality dataset for computational analysis and integration with the RNA-seq dataset. In DKO BMDMs, roughly equal numbers of sites showed increased and decreased accessibility in all conditions, except for the 8-h LPS treatment (Fig. 4A). Genomics sites with large FCs in HMDN DKO were preferentially located in intergenic or intronic regions (Fig. 4B, C, Data Files S8 and S9). The ATAC-seq sites with basally elevated chromatin accessibility in HMGN DKO were located near genes whose ontology is enriched for antigen-processing and -presentation pathways (eg MHC loci) (Fig. S8), which was the most predominantly enriched GO found in our RNA-seq analysis. Additional enriched categories included T cell–mediated immunity and related terms, with representative genes such as Il12b (Fig. S9). In contrast, sites with decreased accessibility in HMGN DKO did not show a noticeable concordance of enriched ontology classes with those identified in the RNA-seq analysis (Fig. 5B). In addition, sites with reduced accessibility in HMGN DKO showed more sex-dependent enrichment patterns than those with increased accessibility (Fig. 5).

Figure 4.

Chromatin accessibility is increased in HMGN DKO near genes related to antigen presentation and regulation of T-cell immunity. (A) Differentially accessible ATAC-seq sites are shown for those more accessible (red) and less accessible (blue) in BMDMs of HMGN DKO versus WT. (B, C) FC profiles of ATAC-seq intensity at differentially accessible peaks and their source samples for sites higher (B) or lower (C) accessibility in HMGN DKO.

Figure 5.

Chromatin accessibility is increased in HMGN DKO near genes related to antigen presentation and regulation of T-cell immunity. Ontology analysis of genes near differentially accessible ATAC-seq sites using data from unstimulated BMDMs. Enriched ontologies for sites with higher (A) and lower (B) accessibility in HMGN DKO.

We then integrated the ATAC-seq and RNA-seq datasets to examine their relative genomic distribution patterns more closely. ATAC-seq peaks represent cell-specific open chromatin sites associated with putative regulatory elements such as enhancers, whose relevant gene targets are often unclear. Consistent with this, simply assigning ATAC-seq peaks to the closest transcription start site yielded no relationship between the effects in RNA-seq versus ATAC-seq profiles. Therefore, we examined the distribution of putative enhancers across a broader range of distances from each DEG.

The genes whose expression were elevated in HMGN DKO were strikingly enriched with ATAC-seq peaks that showed increased accessibility in HMGN DKO, with preferential locations ranging from promoter-proximal sites to far distal (50 kb or more) sites (in Fig. 6A, the “Up” genes showing mostly red bars). On the other hand, the genes repressed in HMGN DKO were predominantly colocalized with ATAC-seq peaks showing decreased accessibility (in Fig. 6A, the “Down” genes showing mostly blue bars). This dichotomy was observed in both male and female mice across all time points. To complement this result, we performed the reciprocal enhancer-centric analysis by examining the distribution of genes near differentially accessible ATAC-seq peaks. We confirmed a similar concordance between ATAC-seq and RNA-seq datasets regarding the directionality of regulation in HMGN DKO (Fig. S10).

Figure 6.

Concordantly modified chromatin accessibility in genomic sites from DEGs in HMGN DKO BMDMs. The distribution of ATAC-seq peaks relative to DEGs. Histograms show the occurrence patterns of more (red) or less (blue) accessible sites along the distance (in base-pairs) from the transcription start sites of DEGs (“Up” or “Down”). RNA-seq and ATAC-seq data from matching sex and LPS stimulation conditions were integrated for the location analysis. (B, C) Transcription factor–binding motifs enriched at ATAC-seq peaks with increased (B) or decreased (C) accessibility in HMGN DKO versus WT.

Differentially accessible chromatin sites in HMGN DKO versus WT revealed distinct sets of enriched motifs within the up- and downregulated genomic sites (Fig. 6B, C). Consistent with previously documented roles of HMGNs in maintaining tissue-specific epigenomes, the sites with reduced accessibility in HMGN DKO were enriched for motifs of transcription factors important for macrophage cell identity and function, eg PU.1 (Spi1) and Ikaros (Ikzf1).^37,38 In contrast, regions that gained accessibility in HMGN DKO were enriched for motifs of transcription factors not typically active in macrophages, eg Oct2 (POU2F2) and KLF15, suggesting potential aberrant regulatory activity.

Discussion

The comprehensive changes to the epigenome and consequent transcriptome that we document here are unexpectedly specific, despite the global role of HMGN proteins in maintaining tissue-specific chromatin architecture. Rather than broadly affecting gene expression, only a relatively select set of genes and distinct functional categories were perturbed in HMGN DKO mice. Moreover, the most significantly affected GO was the antigen-presentation pathway, which is a key function of macrophages linking innate and adaptive immune cells. Although the comprehensive phenotyping of HMGN DKO mice has been conducted,³⁹ the genome-wide impact in macrophages, an alarmin-relevant cell type, has not been previously characterized, to our knowledge.

Although the genes downregulated in HMGN DKO macrophages shared few ontology categories across sex and cell types, they included important macrophage genes, such as Apoe, which was also strongly repressed in HMGN DKO embryonic fibroblasts.³⁹ Another gene, Il12b, showed an opposite pattern between RNA-seq and ATAC-seq datasets: its expression was repressed (Fig. S7), whereas nearby chromatin sites showed greater accessibility (Fig. S9). Further investigation of these individual genes and chromatin sites may clarify their regulation by HMGN proteins.

We identified specific transcriptional reprogramming resulting from the loss of HMGN1 and HMGN2 in alarmin-sending macrophages, which suggests an enhancement of a key immune function, namely, antigen processing and presentation. However, it remains to be examined whether the transcriptional and epigenomic changes lead to increased MHC class I protein expression and enhanced functional antigen presentation in HMGN DKO macrophages and in alarming-secreting macrophages. Alarmin-sending cells activate a transcriptional program that supports chemotaxis,¹⁴ another key function of activated macrophages. It is noteworthy that the alarmin-secreting macrophages may be hyperactive in multiple functional aspects.

Regarding the chromatin impact of lacking HMGNs, we found the specific changes in chromatin accessibility complemented previously reported findings that the alarmin-secreting cells also lose their histones, resulting in reduced nucleosome occupancy genome-wide.^{14,15  40} For example, intact histones can be released into the circulation in either soluble or extracellular vesicle-associated forms, thereby stimulating inflammatory transcriptional responses in recipient cells via TLR4.⁴⁰ This global reduction in nucleosome occupancy may facilitate a poised transcriptional landscape across numerous genes. Interestingly, long-lived histones colocalize with heterochromatin of nondividing cells.⁴¹

The acute effects of HMGN depletion in the nuclei of alarmin-secreting cells may not be the same as our observations in genetic KO macrophages reported in this study. This is because the effects in the KO animals represent not only the direct effects from the absence of HMGN1 and HMGN2 but also long-term organismal adaptations involving multiple indirect molecular and cellular mechanisms. Nevertheless, our dataset is a valuable reference resource for future investigations aiming to isolate the acute impact of HMGN loss on the epigenome and transcriptional program of macrophages actively secreting HMGN alarmins. Our results provide clues to the phenotype of the macrophages following alarmin secretion.

Author contributions

Kyu-Seon Oh (Investigation [Equal], Writing—original draft [Supporting]), Mahamat Babagana (Investigation [Supporting]), Sayantan Chakraborty (Investigation [Supporting]), Mohammad Aqdas (Investigation [Supporting]), Alicja Pacholewska (Data curation [Lead], Formal analysis [Lead]), Daisy Dahiya (Data curation [Equal], Data analysis [Supporting], Writing—review & editing [Supporting]), Michael Bustin (Conceptualization [Equal], Resources [Equal], Supervision [Supporting]), and Myong-Hee Sung (Conceptualization [Equal], Funding acquisition [Lead], Supervision [Lead], Writing—original draft [Equal], Writing—review & editing [Lead])

Supplementary material

Supplementary material is available at The Journal of Immunology online.

Funding

This research was supported by the Intramural Research Program of the National Institutes of Health (NIH) at the National Institute on Aging (grant ZIA AG000390 to M.-H.S.). The contributions of the NIH author are considered Works of the U.S. Government. The findings and conclusions presented in this article are those of the authors and do not necessarily reflect the views of the NIH or the U.S. Department of Health and Human Services.

Conflicts of interest

None declared.

Data availability

The RNA-seq and ATAC-seq datasets have been deposited to the Gene Expression Omnibus database under accession numbers GSE307848 and GSE309182, respectively. The HMGN DKO mice have been deposited to and were cryopreserved at The Jackson Laboratory under the name B6.Cg-Hmgn2^tm1.1Mbus Hmgn1^tm1Mbus/Mmjax.