Open/close ratio reveals cell differentiation state as the primary determinant of chromatin balance in immune memory

Open/close ratio reveals cell differentiation state as the primary determinant of chromatin balance in immune memory | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Open/close ratio reveals cell differentiation state as the primary determinant of chromatin balance in immune memory Jeongsoon Yong This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9364995/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 8 You are reading this latest preprint version Abstract Trained immunity involves epigenetic reprogramming of innate immune cells, yet the genome-wide balance between chromatin opening and closing during this process has not been systematically quantified. Here, we introduce the open/close ratio, a single metric that captures the global chromatin accessibility balance by dividing the number of regions gaining accessibility by those losing it. Applying this metric to 10 independent ATAC-seq datasets encompassing 20 comparisons across human and mouse immune cells, we find that the ratio is primarily determined by cell differentiation state rather than the training stimulus itself. Directly stimulated monocytes and progenitor cells consistently show closing-dominant ratios (0.03–0.86), while bone marrow-derived macrophages that underwent in vitro differentiation show opening-dominant ratios (2.6–160). Methodological validation using MACS3 peak calling, sensitivity analysis across window sizes (500 bp–10 kb), and fold-change thresholds (1.2x–3.0x) confirms that the closing-dominant pattern is robust and method-independent. Time-resolved analysis reveals that chromatin closing begins within one hour of β-glucan exposure, with selective opening of a small gene set that shifts from cytokine receptors (day 1) to response machinery with built-in brakes such as IL1RN (day 6). Cross-dataset corroboration using ATAC-seq, H3K4me3, H3K27ac, and RNA-seq identifies IRF3 as being in a poised state: chromatin open, H3K4me3 elevated, but not yet transcriptionally active. These findings highlight the need to delineate stimulus-driven chromatin changes from differentiation-associated remodeling, and establish the open/close ratio as a standardized metric for comparing chromatin states across immune memory paradigms. trained immunity chromatin accessibility ATAC-seq open/close ratio epigenetic memory poised state meta-analysis Figures Figure 1 Figure 2 Figure 3 Figure 4 Introduction Trained immunity, the enhanced responsiveness of innate immune cells upon restimulation, is mediated by long-lasting epigenetic and metabolic reprogramming [ 1 – 3 ]. Since the landmark studies by Saeed et al. [ 4 ] and Novakovic et al. [ 5 ], chromatin accessibility changes have been considered central to this phenomenon. The prevailing model describes trained immunity as driven by chromatin opening and the deposition of activating histone marks such as H3K4me1 and H3K27ac at inflammatory gene loci [ 4 – 6 ]. However, the field has predominantly focused on regions that gain accessibility, while regions that lose accessibility have been systematically underreported. This bias is embedded in standard analytical pipelines: ATAC-seq studies typically identify gained peaks, perform GO enrichment, and report activated pathways. Lost peaks are rarely quantified. For example, Hargreaves and colleagues reported 15,160 gained and 4,661 lost chromatin sites upon LPS stimulation [ 7 ], but did not compute a ratio between these numbers as a global balance metric. Recent studies have established that trained immunity operates not only at the level of mature immune cells but also through reprogramming of hematopoietic stem and progenitor cells in the bone marrow [ 8 , 9 ]. Kaufmann et al. demonstrated that BCG vaccination induces sustained epigenetic changes in hematopoietic stem cells [ 8 ], while Mitroulis et al. showed that β-glucan modulates myelopoiesis through metabolic reprogramming of bone marrow progenitors [ 9 ]. These findings indicate that differentiation is an integral component of trained immunity in vivo, not merely an experimental artifact. However, this raises an important analytical question: when ATAC-seq is performed on cells that have undergone both training and differentiation, how much of the observed chromatin change reflects the stimulus versus the differentiation process? We reasoned that a simple ratio—the number of regions gaining accessibility divided by the number losing it—could provide an unbiased snapshot of the genome-wide chromatin balance. Here, we apply this open/close ratio across 10 independent ATAC-seq datasets, spanning multiple stimuli, cell types, species, and experimental protocols. Our systematic comparison reveals that the ratio segregates primarily by differentiation state, underscoring the need to delineate stimulus-specific chromatin changes from differentiation-associated remodeling. Results The open/close ratio: a global metric for chromatin balance We developed two complementary approaches to calculate the open/close ratio from public ATAC-seq data. For BigWig signal files, we scanned the genome in non-overlapping windows and classified each region as opened (treatment signal >1.5-fold above control), closed (control signal >1.5-fold above treatment), or unchanged. For narrowPeak files generated by peak callers such as MACS2/MACS3 [10], we computed the set difference between treatment and control peak sets. In both cases, the ratio = opened/closed, where values 1 indicate opening-dominant chromatin remodeling. Chromatin closing begins within one hour of β-glucan stimulation Using the Novakovic et al. time-course dataset (GSE87218) [5], we calculated open/close ratios at four time points following β-glucan or LPS treatment of primary human monocytes (Figure 1). β-glucan induced closing at all time points: ratio = 0.141 (1h), 0.155 (4h), 0.025 (day 1), and 0.032 (day 6). Closing was already dominant at one hour, the earliest available time point, and intensified progressively through day 6. LPS showed a different temporal pattern. While most time points were closing-dominant (0.119 at 1h, 0.061 at day 1, 0.239 at day 6), LPS at 4 hours produced the only opening-dominant ratio in the dataset (6.730). This transient burst of opening, followed by a return to closing, distinguishes the acute inflammatory response from trained immunity at the chromatin level. Methodological validation: the ratio is robust across analytical approaches To address whether the observed closing dominance depends on the analytical method, we performed three independent validations (Table 1). First, we varied the window size from 500 bp to 10 kb; the ratio remained consistently <1 across all resolutions (0.159 at 500 bp, 0.025 at 2 kb, 0.007 at 10 kb). Second, we varied the fold-change threshold from 1.2x to 3.0x; the ratio remained <1 at all thresholds (range: 0.025–0.160). Third, we performed MACS3 peak calling on the BigWig data and computed the ratio from consensus peaks with signal comparison, obtaining a ratio of 0.126—consistent with closing dominance. Additionally, for datasets where MACS3-called narrowPeak files were available from the original authors (GSE183485, GSE230337), the peak-based ratios matched the BigWig-based ratios exactly. Table 1. Sensitivity analysis of open/close ratio for GSE87218 BG d1 vs RPMI d1 Method Opened Closed Ratio BigWig 500 bp window 427,918 2,686,105 0.159 BigWig 1 kb window 97,268 1,361,591 0.071 BigWig 2 kb window 17,017 686,444 0.025 BigWig 5 kb window 3,483 274,773 0.013 BigWig 10 kb window 970 135,652 0.007 Fold 1.2x (2 kb) 51,069 1,022,139 0.050 Fold 2.0x (2 kb) 6,651 257,593 0.026 Fold 3.0x (2 kb) 4,206 26,294 0.160 MACS3 consensus + signal 211,075 1,672,084 0.126 All methods yield ratio <1, confirming closing dominance independent of analytical parameters. Cell differentiation state shapes the open/close ratio across 10 datasets To test whether closing dominance generalizes beyond a single dataset, we extended our analysis to 10 independent datasets (Figure 2). The results revealed a clear pattern: directly stimulated cells—human monocytes treated in vitro, mouse bone marrow progenitors analyzed directly, and iPSC-derived macrophages—consistently showed closing-dominant ratios (range: 0.025–0.86). In contrast, protocols involving bone marrow isolation followed by in vitro differentiation to BMDM consistently produced opening-dominant ratios (range: 2.6–160). Importantly, this distinction should not be interpreted as evidence that differentiation-associated chromatin opening is artifactual. Recent work has established that reprogramming of hematopoietic progenitors and their subsequent differentiation are integral to trained immunity in vivo [8,9]. Rather, our finding highlights the need to delineate the contributions of stimulus-driven epigenetic changes from those associated with differentiation when interpreting ATAC-seq data. The open/close ratio provides a tool for making this distinction explicit. Selective gene opening shifts from receptors to response machinery with built-in brakes Despite the overwhelming closing dominance in β-glucan-trained monocytes (ratio = 0.025 at day 1), 17,017 genomic regions gained accessibility. Gene-level annotation revealed that none of the canonical trained immunity effectors (TNF, IL6, IL1B, MTOR, TLR4, HK2, PKM) were among the opened genes. Instead, the selectively opened genes were predominantly cytokine receptors (IL6R, IL4R, IL7R, IL15) and signaling intermediates (PIK3CA, IRF8, KAT2A). By day 6, this open gene set had largely turned over. Of the 11 immune-related genes open at day 1, 9 closed while new genes appeared, including IRF3, NFKB2, and IL1RN (Figure 3). IL1RN encodes the interleukin-1 receptor antagonist, a natural negative regulator of IL-1 signaling [15]. Its opening at day 6 indicates that trained cells prepare both accelerators (IRF3, NFKB2) and brakes (IL1RN) simultaneously—a built-in safeguard that may prevent excessive inflammation upon restimulation. Training and tolerance share IL1RN as a common brake Comparison of β-glucan training (ratio = 0.032) and LPS tolerance (ratio = 0.239) at day 6 revealed 9 shared open immune genes including IL1RN, IRF3, NFKB2, and IL6R. Training opened 14 immune genes with one negative regulator (IL1RN), while tolerance opened 199 immune genes with multiple negative regulators (IL1RN, SOCS1, DUSP1). This quantitative difference—minimal opening with minimal braking versus extensive opening with redundant braking—distinguishes training from tolerance at the chromatin level. IRF3 exhibits hallmarks of a poised state: cross-dataset corroboration To assess the functional significance of selective opening, we integrated ATAC-seq data (GSE87218 [5]) with H3K27ac ChIP-seq, H3K4me3 ChIP-seq, and RNA-seq data (GSE86940 [16]) from the same research group using matched experimental conditions. We note that these are separate experiments, not simultaneous measurements from the same cells, which limits the strength of the corroboration. For genes classified as closed by ATAC-seq, 10/10 showed either unchanged or decreased RNA expression, confirming that chromatin closing corresponds to transcriptional silencing. For opened genes, IRF3 exhibited hallmarks of a poised state: ATAC open, H3K4me3 increased (+44%), H3K27ac increased (+24%), but RNA unchanged (Figure 4). This combination is consistent with established models of epigenetic poising, where chromatin is prepared for rapid activation but transcription awaits an external signal [4,17]. Like a designate who has been appointed but not yet assumed office, the gene holds a confirmed position—chromatin open, histone marks deposited—but awaits the activation signal to begin transcription. IL32 showed a similar pattern (H3K4me3 +20%). Of the 10 opened genes examined, only these 2 showed consistent histone mark elevation, indicating that the 2 kb window-based ATAC analysis overestimates the number of functionally poised genes. Discussion The open/close ratio provides a simple, reproducible metric for quantifying the global chromatin balance in immune cells. By applying this ratio across 10 datasets and 20 comparisons, with methodological validation using MACS3 peak calling and sensitivity analysis, we arrive at three principal findings. First, directly stimulated immune cells predominantly close chromatin. In every dataset where cells were stimulated without an intervening differentiation step, the ratio was closing-dominant (range: 0.025–0.86). This holds across species (human and mouse), stimuli (β-glucan, LPS, BCG, hemin, DSS), and cell types (monocytes, GMPs, iPSC-derived macrophages). The sole exception is the acute inflammatory peak at LPS 4 hours (ratio = 6.73), which resolves to closing by day 1. Second, stimulus-driven and differentiation-associated chromatin changes require delineation. Datasets in which cells underwent differentiation consistently showed opening-dominant ratios. This does not mean that differentiation-associated opening is artifactual—indeed, reprogramming of hematopoietic progenitors is now recognized as a central mechanism of trained immunity [8,9]. Rather, it means that ATAC-seq studies comparing trained versus untrained cells that have undergone different degrees of differentiation may conflate two distinct biological processes. The open/close ratio makes this conflation visible and quantifiable. Third, trained cells are poised, not active. Cross-dataset corroboration identified IRF3 as a poised gene: chromatin open, H3K4me3 elevated, but transcriptionally silent. This is consistent with the established concept of poised enhancers and bivalent chromatin [17,18]. The co-opening of IL1RN alongside inflammatory genes at day 6 indicates that trained cells prepare both accelerator and brake simultaneously. We speculate that failure to establish this dual poised state—for example through genetic variants in IL1RN [19] or through incomplete chromatin remodeling—could contribute to dysregulated inflammation such as cytokine storm, though this hypothesis requires direct experimental testing. Limitations Several limitations should be noted. First, our initial BigWig-based analysis uses a sliding window rather than formal peak calling. We addressed this through MACS3 validation and sensitivity analysis across window sizes and thresholds, which confirmed that the closing-dominant pattern is method-independent. However, the absolute ratio values differ between methods, and formal peak calling from raw sequencing data (FASTQ → BAM → MACS3) would provide the most rigorous quantification. Second, most comparisons involve n=1 per condition, precluding formal statistical testing. Third, the cross-dataset corroboration (ATAC from GSE87218, ChIP/RNA from GSE86940) involves different experiments, limiting its interpretive strength; only 2 of 10 opened genes showed consistent histone mark changes. Fourth, the distinction between direct stimulation and differentiation is not binary, as even in vitro monocyte culture involves some differentiation. Fifth, the gene-level analysis using TSS±5 kb annotation likely overestimates the number of functionally opened genes, as validated by the low corroboration rate with histone marks. Methods Data sources All data were obtained from GEO. Ten ATAC-seq datasets: GSE87218 [5], GSE280935 [11], GSE324548, GSE141968, GSE230337, GSE172116 [12], GSE131446 [13], GSE183485, GSE266967, GSE190004 [14]. Validation: GSE86940 [16] (H3K27ac, H3K4me3 ChIP-seq, RNA-seq). Open/close ratio calculation For BigWig files, the genome was scanned in non-overlapping windows (default 2 kb). At each window, mean signal was computed for treatment and control. Windows where both signals were below 0.1 were excluded. A window was classified as opened if treatment > control × 1.5, closed if control > treatment × 1.5. Only autosomal chromosomes were included. For narrowPeak files, peaks were binned to 500 bp resolution and the set difference computed. Ratio = opened / closed. Sensitivity analysis To assess robustness, the ratio was recalculated with window sizes of 500 bp, 1 kb, 2 kb, 5 kb, and 10 kb, and with fold-change thresholds of 1.2x, 1.5x, 2.0x, and 3.0x. Additionally, MACS3 (v3.0.4) [10] bdgpeakcall was used to call peaks from BigWig-derived bedGraph files, and the ratio was computed from consensus peaks with signal comparison. Gene-level annotation Opened regions were mapped to genes using UCSC refGene annotations (hg38), assigning regions within ±5 kb of the transcription start site. Cross-dataset corroboration H3K27ac and H3K4me3 ChIP-seq BigWig files from GSE86940 [16] were queried at promoter regions (TSS ±2 kb). Fold change was computed as BG signal / RPMI signal. RNA-seq expression data (mmseq format) were converted from Ensembl gene IDs to gene symbols using BioMart. Expression change was computed as the difference in log_mu values. Declarations Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Ethics declaration Not applicable. This study exclusively analyzed publicly available, de-identified datasets from the Gene Expression Omnibus. Consent to publish Not applicable. Competing interests The author declares no competing interests. Author contributions J.Y. conceived the study, performed all analyses, and wrote the manuscript. Data availability All ATAC-seq, ChIP-seq, and RNA-seq datasets analyzed in this study are publicly available from NCBI GEO under accession numbers GSE87218, GSE280935, GSE324548, GSE141968, GSE230337, GSE172116, GSE131446, GSE183485, GSE266967, GSE190004, and GSE86940. Acknowledgements We thank Novakovic et al. for making their comprehensive ATAC-seq time-course data publicly available. References Netea MG, Joosten LAB, Latz E, et al. Trained immunity: a program of innate immune memory in health and disease. Science. 2016;352(6284):aaf1098. Netea MG, Domínguez-Andrés J, Barreiro LB, et al. Defining trained immunity and its role in health and disease. Nat Rev Immunol. 2020;20(6):375–388. Quintin J, Saeed S, Martens JHA, et al. Candida albicans infection affords protection against reinfection via functional reprogramming of monocytes. Cell Host Microbe. 2012;12(2):223–232. Saeed S, Quintin J, Kerstens HHD, et al. Epigenetic programming of monocyte-to-macrophage differentiation and trained innate immunity. Science. 2014;345(6204):1251086. Novakovic B, Habibi E, Wang SY, et al. β-glucan reverses the epigenetic state of LPS-induced immunological tolerance. Cell. 2016;167(5):1354–1368. Cheng SC, Quintin J, Cramer RA, et al. mTOR- and HIF-1α-mediated aerobic glycolysis as metabolic basis for trained immunity. Science. 2014;345(6204):1250684. Liao J, Ho J, Burns M, Dykhuizen EC, Hargreaves DC. Collaboration between distinct SWI/SNF chromatin remodeling complexes directs enhancer selection and activation of macrophage inflammatory genes. Immunity. 2024;57(8):1780–1795. Kaufmann E, Sanz J, Dunn JL, et al. BCG educates hematopoietic stem cells to generate protective innate immunity against tuberculosis. Cell. 2018;172(1–2):176–190. Mitroulis I, Ruppova K, Wang B, et al. Modulation of myelopoiesis progenitors is an integral component of trained immunity. Cell. 2018;172(1–2):147–161. Zhang Y, Liu T, Meyer CA, et al. Model-based analysis of ChIP-Seq (MACS). Genome Biol. 2008;9(9):R137. [Aurora kinase A promotes trained immunity via regulation of endogenous S-adenosylmethionine metabolism. eLife. 2025 (reviewed preprint)]. Dossou AS, Gardeux V, Holtz A, et al. Epigenomic analysis reveals a dynamic and context-specific macrophage enhancer landscape associated with innate immune activation and tolerance. Genome Biol. 2022;23:136. Ciernia AV, Link VM, Engeln M, et al. ATAC-sequencing of BTBR and C57BL6/J BMDM with repeated LPS treatment. GEO GSE131446. Cheong JG, Ravishankar A, Sharma S, et al. Epigenomic landscape of COVID-19: durable alterations of hematopoiesis and chromatin. Cell. 2023;186:1–17. Arend WP, Malyak M, Guthridge CJ, Gabay C. Interleukin-1 receptor antagonist: role in biology. Annu Rev Immunol. 1998;16:27–55. Arts RJW, Novakovic B, ter Horst R, et al. Glutaminolysis and fumarate accumulation integrate immunometabolic and epigenetic programs in trained immunity. Cell Metab. 2016;24(6):807–819. Bernstein BE, Mikkelsen TS, Xie X, et al. A bivalent chromatin structure marks key developmental genes in embryonic stem cells. Cell. 2006;125(2):315–326. Rada-Iglesias A, Bajpai R, Swigut T, et al. A unique chromatin signature uncovers early developmental enhancers in humans. Nature. 2011;470(7333):279–283. Aksentijevich I, Masters SL, Ferguson PJ, et al. An autoinflammatory disease with deficiency of the interleukin-1-receptor antagonist. N Engl J Med. 2009;360(23):2426–2437. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Reviews received at journal 08 May, 2026 Reviews received at journal 28 Apr, 2026 Reviewers agreed at journal 17 Apr, 2026 Reviewers agreed at journal 16 Apr, 2026 Reviewers invited by journal 16 Apr, 2026 Editor assigned by journal 11 Apr, 2026 Submission checks completed at journal 10 Apr, 2026 First submitted to journal 09 Apr, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9364995","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":627368098,"identity":"41f02cd9-b4e9-4cab-bb4c-77b2bd1a19fc","order_by":0,"name":"Jeongsoon Yong","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA20lEQVRIie3OMQrCMBTG8VcKcYl2TSh4hlcKxUHqVSKBTi5uDg4RQZeiq+DgFTxCpNCp6Cp09AIFQVwsOgkupm4O+Y0P/rwPwLL+kLO7aKwmfXxfmClxVSLGmyL5IWmpEV7bi+yHpK10sKfkFHp+dqhgGgPf6u8JnymJjJYRXyWSQS7B74jvSTCHHJGVfSwoMiAautQwbJA7y7vA4yvxrneoGyRO6gJqoaPXF8KchQbfmGwIBErLkKck6g1XkvLUmHhV+KjjYE3dy7m6xV1WGJJPAsA0y7Isy2riCc+FPZw1EZl4AAAAAElFTkSuQmCC","orcid":"","institution":"Korea University","correspondingAuthor":true,"prefix":"","firstName":"Jeongsoon","middleName":"","lastName":"Yong","suffix":""}],"badges":[],"createdAt":"2026-04-09 07:54:04","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9364995/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9364995/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":107793886,"identity":"bc808da2-a34c-4d8d-8864-93da0f9f081b","added_by":"auto","created_at":"2026-04-25 14:27:23","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":101582,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eTemporal dynamics of the open/close ratio in human monocytes. β-glucan (blue) shows sustained closing. LPS (red) shows a transient opening peak at 4 hours. Dashed line = ratio of 1.0. Data from GSE87218 [5].\u003c/em\u003e\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-9364995/v1/51287d0dc058c099b3b5d640.png"},{"id":107869574,"identity":"2a56f147-2796-433b-852d-373da65ddb6f","added_by":"auto","created_at":"2026-04-27 07:37:27","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":373334,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eMeta-analysis of open/close ratios across 10 ATAC-seq datasets. Blue = directly stimulated cells; red = cells that underwent differentiation; orange = acute transient; green = in vivo. Data from GSE87218 [5], GSE280935 [11], GSE324548, GSE141968, GSE230337, GSE172116 [12], GSE131446 [13], GSE183485, GSE266967, GSE190004 [14].\u003c/em\u003e\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-9364995/v1/403cd6cbe430b0f4f28ad01c.png"},{"id":107793889,"identity":"0b5de794-1231-4d1f-bc5c-71a774f02c84","added_by":"auto","created_at":"2026-04-25 14:27:23","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":218884,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eSelective gene opening shifts from day 1 to day 6. Left: day 1 opened genes (receptors). Center: 9/11 genes close by day 6. Right: day 6 genes include response machinery and IL1RN (brake). GSE87218 [5].\u003c/em\u003e\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-9364995/v1/d15bcffb1d9c44a1fa582c64.png"},{"id":107869566,"identity":"873ae4fa-1e97-44e0-b546-766915c8fd6f","added_by":"auto","created_at":"2026-04-27 07:37:26","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":173202,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eCross-dataset corroboration. (A) H3K4me3 at ATAC-opened genes; IRF3 and IL32 show elevation consistent with poised state. (B) RNA at ATAC-closed genes: 10/10 unchanged or decreased. GSE87218 [5] + GSE86940 [16]. Note: different experiments, not simultaneous measurements.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-9364995/v1/c504c8bc0aa330eeebb02e1b.png"},{"id":107871716,"identity":"839d74ad-da5f-4e4a-95fb-33ac1854f398","added_by":"auto","created_at":"2026-04-27 07:53:49","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":900476,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9364995/v1/6277a603-b432-42d3-97bb-c8c2382e09cb.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Open/close ratio reveals cell differentiation state as the primary determinant of chromatin balance in immune memory","fulltext":[{"header":"Introduction","content":"\u003cp\u003eTrained immunity, the enhanced responsiveness of innate immune cells upon restimulation, is mediated by long-lasting epigenetic and metabolic reprogramming [\u003cspan additionalcitationids=\"CR2\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. Since the landmark studies by Saeed et al. [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e] and Novakovic et al. [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e], chromatin accessibility changes have been considered central to this phenomenon. The prevailing model describes trained immunity as driven by chromatin opening and the deposition of activating histone marks such as H3K4me1 and H3K27ac at inflammatory gene loci [\u003cspan additionalcitationids=\"CR5\" citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eHowever, the field has predominantly focused on regions that gain accessibility, while regions that lose accessibility have been systematically underreported. This bias is embedded in standard analytical pipelines: ATAC-seq studies typically identify gained peaks, perform GO enrichment, and report activated pathways. Lost peaks are rarely quantified. For example, Hargreaves and colleagues reported 15,160 gained and 4,661 lost chromatin sites upon LPS stimulation [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e], but did not compute a ratio between these numbers as a global balance metric.\u003c/p\u003e \u003cp\u003eRecent studies have established that trained immunity operates not only at the level of mature immune cells but also through reprogramming of hematopoietic stem and progenitor cells in the bone marrow [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. Kaufmann et al. demonstrated that BCG vaccination induces sustained epigenetic changes in hematopoietic stem cells [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e], while Mitroulis et al. showed that β-glucan modulates myelopoiesis through metabolic reprogramming of bone marrow progenitors [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. These findings indicate that differentiation is an integral component of trained immunity in vivo, not merely an experimental artifact. However, this raises an important analytical question: when ATAC-seq is performed on cells that have undergone both training and differentiation, how much of the observed chromatin change reflects the stimulus versus the differentiation process?\u003c/p\u003e \u003cp\u003eWe reasoned that a simple ratio\u0026mdash;the number of regions gaining accessibility divided by the number losing it\u0026mdash;could provide an unbiased snapshot of the genome-wide chromatin balance. Here, we apply this open/close ratio across 10 independent ATAC-seq datasets, spanning multiple stimuli, cell types, species, and experimental protocols. Our systematic comparison reveals that the ratio segregates primarily by differentiation state, underscoring the need to delineate stimulus-specific chromatin changes from differentiation-associated remodeling.\u003c/p\u003e"},{"header":"Results","content":"\u003ch2\u003eThe open/close ratio: a global metric for chromatin balance\u003c/h2\u003e\n\u003cp\u003eWe developed two complementary approaches to calculate the open/close ratio from public ATAC-seq data. For BigWig signal files, we scanned the genome in non-overlapping windows and classified each region as opened (treatment signal \u0026gt;1.5-fold above control), closed (control signal \u0026gt;1.5-fold above treatment), or unchanged. For narrowPeak files generated by peak callers such as MACS2/MACS3 [10], we computed the set difference between treatment and control peak sets. In both cases, the ratio = opened/closed, where values \u0026lt;1 indicate closing-dominant and \u0026gt;1 indicate opening-dominant chromatin remodeling.\u003c/p\u003e\n\u003ch2\u003eChromatin closing begins within one hour of \u0026beta;-glucan stimulation\u003c/h2\u003e\n\u003cp\u003eUsing the Novakovic et al. time-course dataset (GSE87218) [5], we calculated open/close ratios at four time points following \u0026beta;-glucan or LPS treatment of primary human monocytes (Figure 1). \u0026beta;-glucan induced closing at all time points: ratio = 0.141 (1h), 0.155 (4h), 0.025 (day 1), and 0.032 (day 6). Closing was already dominant at one hour, the earliest available time point, and intensified progressively through day 6.\u003c/p\u003e\n\u003cp\u003eLPS showed a different temporal pattern. While most time points were closing-dominant (0.119 at 1h, 0.061 at day 1, 0.239 at day 6), LPS at 4 hours produced the only opening-dominant ratio in the dataset (6.730). This transient burst of opening, followed by a return to closing, distinguishes the acute inflammatory response from trained immunity at the chromatin level.\u003c/p\u003e\n\u003ch2\u003eMethodological validation: the ratio is robust across analytical approaches\u003c/h2\u003e\n\u003cp\u003eTo address whether the observed closing dominance depends on the analytical method, we performed three independent validations (Table 1). First, we varied the window size from 500 bp to 10 kb; the ratio remained consistently \u0026lt;1 across all resolutions (0.159 at 500 bp, 0.025 at 2 kb, 0.007 at 10 kb). Second, we varied the fold-change threshold from 1.2x to 3.0x; the ratio remained \u0026lt;1 at all thresholds (range: 0.025\u0026ndash;0.160). Third, we performed MACS3 peak calling on the BigWig data and computed the ratio from consensus peaks with signal comparison, obtaining a ratio of 0.126\u0026mdash;consistent with closing dominance. Additionally, for datasets where MACS3-called narrowPeak files were available from the original authors (GSE183485, GSE230337), the peak-based ratios matched the BigWig-based ratios exactly.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 1.\u0026nbsp;\u003c/strong\u003eSensitivity analysis of open/close ratio for GSE87218 BG d1 vs RPMI d1\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"624\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.28%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMethod\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.24%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eOpened\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.24%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eClosed\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.24%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eRatio\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.28%;\"\u003e\n \u003cp\u003eBigWig 500 bp window\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.24%;\"\u003e\n \u003cp\u003e427,918\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.24%;\"\u003e\n \u003cp\u003e2,686,105\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.24%;\"\u003e\n \u003cp\u003e0.159\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.28%;\"\u003e\n \u003cp\u003eBigWig 1 kb window\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.24%;\"\u003e\n \u003cp\u003e97,268\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.24%;\"\u003e\n \u003cp\u003e1,361,591\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.24%;\"\u003e\n \u003cp\u003e0.071\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.28%;\"\u003e\n \u003cp\u003eBigWig 2 kb window\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.24%;\"\u003e\n \u003cp\u003e17,017\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.24%;\"\u003e\n \u003cp\u003e686,444\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.24%;\"\u003e\n \u003cp\u003e0.025\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.28%;\"\u003e\n \u003cp\u003eBigWig 5 kb window\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.24%;\"\u003e\n \u003cp\u003e3,483\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.24%;\"\u003e\n \u003cp\u003e274,773\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.24%;\"\u003e\n \u003cp\u003e0.013\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.28%;\"\u003e\n \u003cp\u003eBigWig 10 kb window\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.24%;\"\u003e\n \u003cp\u003e970\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.24%;\"\u003e\n \u003cp\u003e135,652\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.24%;\"\u003e\n \u003cp\u003e0.007\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.28%;\"\u003e\n \u003cp\u003eFold 1.2x (2 kb)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.24%;\"\u003e\n \u003cp\u003e51,069\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.24%;\"\u003e\n \u003cp\u003e1,022,139\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.24%;\"\u003e\n \u003cp\u003e0.050\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.28%;\"\u003e\n \u003cp\u003eFold 2.0x (2 kb)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.24%;\"\u003e\n \u003cp\u003e6,651\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.24%;\"\u003e\n \u003cp\u003e257,593\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.24%;\"\u003e\n \u003cp\u003e0.026\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.28%;\"\u003e\n \u003cp\u003eFold 3.0x (2 kb)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.24%;\"\u003e\n \u003cp\u003e4,206\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.24%;\"\u003e\n \u003cp\u003e26,294\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.24%;\"\u003e\n \u003cp\u003e0.160\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.28%;\"\u003e\n \u003cp\u003eMACS3 consensus + signal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.24%;\"\u003e\n \u003cp\u003e211,075\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.24%;\"\u003e\n \u003cp\u003e1,672,084\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.24%;\"\u003e\n \u003cp\u003e0.126\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cem\u003eAll methods yield ratio \u0026lt;1, confirming closing dominance independent of analytical parameters.\u003c/em\u003e\u003c/p\u003e\n\u003ch2\u003eCell differentiation state shapes the open/close ratio across 10 datasets\u003c/h2\u003e\n\u003cp\u003eTo test whether closing dominance generalizes beyond a single dataset, we extended our analysis to 10 independent datasets (Figure 2). The results revealed a clear pattern: directly stimulated cells\u0026mdash;human monocytes treated in vitro, mouse bone marrow progenitors analyzed directly, and iPSC-derived macrophages\u0026mdash;consistently showed closing-dominant ratios (range: 0.025\u0026ndash;0.86). In contrast, protocols involving bone marrow isolation followed by in vitro differentiation to BMDM consistently produced opening-dominant ratios (range: 2.6\u0026ndash;160).\u003c/p\u003e\n\u003cp\u003eImportantly, this distinction should not be interpreted as evidence that differentiation-associated chromatin opening is artifactual. Recent work has established that reprogramming of hematopoietic progenitors and their subsequent differentiation are integral to trained immunity in vivo [8,9]. Rather, our finding highlights the need to delineate the contributions of stimulus-driven epigenetic changes from those associated with differentiation when interpreting ATAC-seq data. The open/close ratio provides a tool for making this distinction explicit.\u003c/p\u003e\n\u003ch2\u003eSelective gene opening shifts from receptors to response machinery with built-in brakes\u003c/h2\u003e\n\u003cp\u003eDespite the overwhelming closing dominance in \u0026beta;-glucan-trained monocytes (ratio = 0.025 at day 1), 17,017 genomic regions gained accessibility. Gene-level annotation revealed that none of the canonical trained immunity effectors (TNF, IL6, IL1B, MTOR, TLR4, HK2, PKM) were among the opened genes. Instead, the selectively opened genes were predominantly cytokine receptors (IL6R, IL4R, IL7R, IL15) and signaling intermediates (PIK3CA, IRF8, KAT2A).\u003c/p\u003e\n\u003cp\u003eBy day 6, this open gene set had largely turned over. Of the 11 immune-related genes open at day 1, 9 closed while new genes appeared, including IRF3, NFKB2, and IL1RN (Figure 3). IL1RN encodes the interleukin-1 receptor antagonist, a natural negative regulator of IL-1 signaling [15]. Its opening at day 6 indicates that trained cells prepare both accelerators (IRF3, NFKB2) and brakes (IL1RN) simultaneously\u0026mdash;a built-in safeguard that may prevent excessive inflammation upon restimulation.\u003c/p\u003e\n\u003ch2\u003eTraining and tolerance share IL1RN as a common brake\u003c/h2\u003e\n\u003cp\u003eComparison of \u0026beta;-glucan training (ratio = 0.032) and LPS tolerance (ratio = 0.239) at day 6 revealed 9 shared open immune genes including IL1RN, IRF3, NFKB2, and IL6R. Training opened 14 immune genes with one negative regulator (IL1RN), while tolerance opened 199 immune genes with multiple negative regulators (IL1RN, SOCS1, DUSP1). This quantitative difference\u0026mdash;minimal opening with minimal braking versus extensive opening with redundant braking\u0026mdash;distinguishes training from tolerance at the chromatin level.\u003c/p\u003e\n\u003ch2\u003eIRF3 exhibits hallmarks of a poised state: cross-dataset corroboration\u003c/h2\u003e\n\u003cp\u003eTo assess the functional significance of selective opening, we integrated ATAC-seq data (GSE87218 [5]) with H3K27ac ChIP-seq, H3K4me3 ChIP-seq, and RNA-seq data (GSE86940 [16]) from the same research group using matched experimental conditions. We note that these are separate experiments, not simultaneous measurements from the same cells, which limits the strength of the corroboration.\u003c/p\u003e\n\u003cp\u003eFor genes classified as closed by ATAC-seq, 10/10 showed either unchanged or decreased RNA expression, confirming that chromatin closing corresponds to transcriptional silencing. For opened genes, IRF3 exhibited hallmarks of a poised state: ATAC open, H3K4me3 increased (+44%), H3K27ac increased (+24%), but RNA unchanged (Figure 4). This combination is consistent with established models of epigenetic poising, where chromatin is prepared for rapid activation but transcription awaits an external signal [4,17]. Like a designate who has been appointed but not yet assumed office, the gene holds a confirmed position\u0026mdash;chromatin open, histone marks deposited\u0026mdash;but awaits the activation signal to begin transcription. IL32 showed a similar pattern (H3K4me3 +20%). Of the 10 opened genes examined, only these 2 showed consistent histone mark elevation, indicating that the 2 kb window-based ATAC analysis overestimates the number of functionally poised genes.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThe open/close ratio provides a simple, reproducible metric for quantifying the global chromatin balance in immune cells. By applying this ratio across 10 datasets and 20 comparisons, with methodological validation using MACS3 peak calling and sensitivity analysis, we arrive at three principal findings.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFirst, directly stimulated immune cells predominantly close chromatin.\u003c/strong\u003e In every dataset where cells were stimulated without an intervening differentiation step, the ratio was closing-dominant (range: 0.025\u0026ndash;0.86). This holds across species (human and mouse), stimuli (\u0026beta;-glucan, LPS, BCG, hemin, DSS), and cell types (monocytes, GMPs, iPSC-derived macrophages). The sole exception is the acute inflammatory peak at LPS 4 hours (ratio = 6.73), which resolves to closing by day 1.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSecond, stimulus-driven and differentiation-associated chromatin changes require delineation.\u003c/strong\u003e Datasets in which cells underwent differentiation consistently showed opening-dominant ratios. This does not mean that differentiation-associated opening is artifactual\u0026mdash;indeed, reprogramming of hematopoietic progenitors is now recognized as a central mechanism of trained immunity [8,9]. Rather, it means that ATAC-seq studies comparing trained versus untrained cells that have undergone different degrees of differentiation may conflate two distinct biological processes. The open/close ratio makes this conflation visible and quantifiable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eThird, trained cells are poised, not active.\u003c/strong\u003e Cross-dataset corroboration identified IRF3 as a poised gene: chromatin open, H3K4me3 elevated, but transcriptionally silent. This is consistent with the established concept of poised enhancers and bivalent chromatin [17,18]. The co-opening of IL1RN alongside inflammatory genes at day 6 indicates that trained cells prepare both accelerator and brake simultaneously. We speculate that failure to establish this dual poised state\u0026mdash;for example through genetic variants in IL1RN [19] or through incomplete chromatin remodeling\u0026mdash;could contribute to dysregulated inflammation such as cytokine storm, though this hypothesis requires direct experimental testing.\u003c/p\u003e\n\u003ch2\u003eLimitations\u003c/h2\u003e\n\u003cp\u003eSeveral limitations should be noted. First, our initial BigWig-based analysis uses a sliding window rather than formal peak calling. We addressed this through MACS3 validation and sensitivity analysis across window sizes and thresholds, which confirmed that the closing-dominant pattern is method-independent. However, the absolute ratio values differ between methods, and formal peak calling from raw sequencing data (FASTQ \u0026rarr; BAM \u0026rarr; MACS3) would provide the most rigorous quantification. Second, most comparisons involve n=1 per condition, precluding formal statistical testing. Third, the cross-dataset corroboration (ATAC from GSE87218, ChIP/RNA from GSE86940) involves different experiments, limiting its interpretive strength; only 2 of 10 opened genes showed consistent histone mark changes. Fourth, the distinction between direct stimulation and differentiation is not binary, as even in vitro monocyte culture involves some differentiation. Fifth, the gene-level analysis using TSS\u0026plusmn;5 kb annotation likely overestimates the number of functionally opened genes, as validated by the low corroboration rate with histone marks.\u003c/p\u003e"},{"header":"Methods","content":"\u003ch2\u003eData sources\u003c/h2\u003e\n\u003cp\u003eAll data were obtained from GEO. Ten ATAC-seq datasets: GSE87218 [5], GSE280935 [11], GSE324548, GSE141968, GSE230337, GSE172116 [12], GSE131446 [13], GSE183485, GSE266967, GSE190004 [14]. Validation: GSE86940 [16] (H3K27ac, H3K4me3 ChIP-seq, RNA-seq).\u003c/p\u003e\n\u003ch2\u003eOpen/close ratio calculation\u003c/h2\u003e\n\u003cp\u003eFor BigWig files, the genome was scanned in non-overlapping windows (default 2 kb). At each window, mean signal was computed for treatment and control. Windows where both signals were below 0.1 were excluded. A window was classified as opened if treatment \u0026gt; control \u0026times; 1.5, closed if control \u0026gt; treatment \u0026times; 1.5. Only autosomal chromosomes were included. For narrowPeak files, peaks were binned to 500 bp resolution and the set difference computed. Ratio = opened / closed.\u003c/p\u003e\n\u003ch2\u003eSensitivity analysis\u003c/h2\u003e\n\u003cp\u003eTo assess robustness, the ratio was recalculated with window sizes of 500 bp, 1 kb, 2 kb, 5 kb, and 10 kb, and with fold-change thresholds of 1.2x, 1.5x, 2.0x, and 3.0x. Additionally, MACS3 (v3.0.4) [10] bdgpeakcall was used to call peaks from BigWig-derived bedGraph files, and the ratio was computed from consensus peaks with signal comparison.\u003c/p\u003e\n\u003ch2\u003eGene-level annotation\u003c/h2\u003e\n\u003cp\u003eOpened regions were mapped to genes using UCSC refGene annotations (hg38), assigning regions within \u0026plusmn;5 kb of the transcription start site.\u003c/p\u003e\n\u003ch2\u003eCross-dataset corroboration\u003c/h2\u003e\n\u003cp\u003eH3K27ac and H3K4me3 ChIP-seq BigWig files from GSE86940 [16] were queried at promoter regions (TSS \u0026plusmn;2 kb). Fold change was computed as BG signal / RPMI signal. RNA-seq expression data (mmseq format) were converted from Ensembl gene IDs to gene symbols using BioMart. Expression change was computed as the difference in log_mu values.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eFunding\u003c/h2\u003e\n\u003cp\u003eThis research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.\u003c/p\u003e\n\u003ch2\u003eEthics declaration\u003c/h2\u003e\n\u003cp\u003eNot applicable. This study exclusively analyzed publicly available, de-identified datasets from the Gene Expression Omnibus.\u003c/p\u003e\n\u003ch2\u003eConsent to publish\u003c/h2\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003ch2\u003eCompeting interests\u003c/h2\u003e\n\u003cp\u003eThe author declares no competing interests.\u003c/p\u003e\n\u003ch2\u003eAuthor contributions\u003c/h2\u003e\n\u003cp\u003eJ.Y. conceived the study, performed all analyses, and wrote the manuscript.\u003c/p\u003e\n\u003ch1\u003eData availability\u003c/h1\u003e\n\u003cp\u003eAll ATAC-seq, ChIP-seq, and RNA-seq datasets analyzed in this study are publicly available from NCBI GEO under accession numbers GSE87218, GSE280935, GSE324548, GSE141968, GSE230337, GSE172116, GSE131446, GSE183485, GSE266967, GSE190004, and GSE86940.\u003c/p\u003e\n\u003ch1\u003eAcknowledgements\u003c/h1\u003e\n\u003cp\u003eWe thank Novakovic et al. for making their comprehensive ATAC-seq time-course data publicly available.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eNetea MG, Joosten LAB, Latz E, et al. Trained immunity: a program of innate immune memory in health and disease. Science. 2016;352(6284):aaf1098.\u003c/li\u003e\n\u003cli\u003eNetea MG, Dom\u0026iacute;nguez-Andr\u0026eacute;s J, Barreiro LB, et al. Defining trained immunity and its role in health and disease. Nat Rev Immunol. 2020;20(6):375\u0026ndash;388.\u003c/li\u003e\n\u003cli\u003eQuintin J, Saeed S, Martens JHA, et al. Candida albicans infection affords protection against reinfection via functional reprogramming of monocytes. Cell Host Microbe. 2012;12(2):223\u0026ndash;232.\u003c/li\u003e\n\u003cli\u003eSaeed S, Quintin J, Kerstens HHD, et al. Epigenetic programming of monocyte-to-macrophage differentiation and trained innate immunity. Science. 2014;345(6204):1251086.\u003c/li\u003e\n\u003cli\u003eNovakovic B, Habibi E, Wang SY, et al. \u0026beta;-glucan reverses the epigenetic state of LPS-induced immunological tolerance. Cell. 2016;167(5):1354\u0026ndash;1368.\u003c/li\u003e\n\u003cli\u003eCheng SC, Quintin J, Cramer RA, et al. mTOR- and HIF-1\u0026alpha;-mediated aerobic glycolysis as metabolic basis for trained immunity. Science. 2014;345(6204):1250684.\u003c/li\u003e\n\u003cli\u003eLiao J, Ho J, Burns M, Dykhuizen EC, Hargreaves DC. Collaboration between distinct SWI/SNF chromatin remodeling complexes directs enhancer selection and activation of macrophage inflammatory genes. Immunity. 2024;57(8):1780\u0026ndash;1795.\u003c/li\u003e\n\u003cli\u003eKaufmann E, Sanz J, Dunn JL, et al. BCG educates hematopoietic stem cells to generate protective innate immunity against tuberculosis. Cell. 2018;172(1\u0026ndash;2):176\u0026ndash;190.\u003c/li\u003e\n\u003cli\u003eMitroulis I, Ruppova K, Wang B, et al. Modulation of myelopoiesis progenitors is an integral component of trained immunity. Cell. 2018;172(1\u0026ndash;2):147\u0026ndash;161.\u003c/li\u003e\n\u003cli\u003eZhang Y, Liu T, Meyer CA, et al. Model-based analysis of ChIP-Seq (MACS). Genome Biol. 2008;9(9):R137.\u003c/li\u003e\n\u003cli\u003e[Aurora kinase A promotes trained immunity via regulation of endogenous S-adenosylmethionine metabolism. eLife. 2025 (reviewed preprint)].\u003c/li\u003e\n\u003cli\u003eDossou AS, Gardeux V, Holtz A, et al. Epigenomic analysis reveals a dynamic and context-specific macrophage enhancer landscape associated with innate immune activation and tolerance. Genome Biol. 2022;23:136.\u003c/li\u003e\n\u003cli\u003eCiernia AV, Link VM, Engeln M, et al. ATAC-sequencing of BTBR and C57BL6/J BMDM with repeated LPS treatment. GEO GSE131446.\u003c/li\u003e\n\u003cli\u003eCheong JG, Ravishankar A, Sharma S, et al. Epigenomic landscape of COVID-19: durable alterations of hematopoiesis and chromatin. Cell. 2023;186:1\u0026ndash;17.\u003c/li\u003e\n\u003cli\u003eArend WP, Malyak M, Guthridge CJ, Gabay C. Interleukin-1 receptor antagonist: role in biology. Annu Rev Immunol. 1998;16:27\u0026ndash;55.\u003c/li\u003e\n\u003cli\u003eArts RJW, Novakovic B, ter Horst R, et al. Glutaminolysis and fumarate accumulation integrate immunometabolic and epigenetic programs in trained immunity. Cell Metab. 2016;24(6):807\u0026ndash;819.\u003c/li\u003e\n\u003cli\u003eBernstein BE, Mikkelsen TS, Xie X, et al. A bivalent chromatin structure marks key developmental genes in embryonic stem cells. Cell. 2006;125(2):315\u0026ndash;326.\u003c/li\u003e\n\u003cli\u003eRada-Iglesias A, Bajpai R, Swigut T, et al. A unique chromatin signature uncovers early developmental enhancers in humans. Nature. 2011;470(7333):279\u0026ndash;283.\u003c/li\u003e\n\u003cli\u003eAksentijevich I, Masters SL, Ferguson PJ, et al. An autoinflammatory disease with deficiency of the interleukin-1-receptor antagonist. N Engl J Med. 2009;360(23):2426\u0026ndash;2437.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"epigenetics-and-chromatin","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"epch","sideBox":"Learn more about [Epigenetics \u0026 Chromatin](http://epigeneticsandchromatin.biomedcentral.com/)","snPcode":"13072","submissionUrl":"https://submission.nature.com/new-submission/13072/3","title":"Epigenetics \u0026 Chromatin","twitterHandle":"@EpigenChromatin","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"trained immunity, chromatin accessibility, ATAC-seq, open/close ratio, epigenetic memory, poised state, meta-analysis","lastPublishedDoi":"10.21203/rs.3.rs-9364995/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9364995/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eTrained immunity involves epigenetic reprogramming of innate immune cells, yet the genome-wide balance between chromatin opening and closing during this process has not been systematically quantified. Here, we introduce the open/close ratio, a single metric that captures the global chromatin accessibility balance by dividing the number of regions gaining accessibility by those losing it. Applying this metric to 10 independent ATAC-seq datasets encompassing 20 comparisons across human and mouse immune cells, we find that the ratio is primarily determined by cell differentiation state rather than the training stimulus itself. Directly stimulated monocytes and progenitor cells consistently show closing-dominant ratios (0.03\u0026ndash;0.86), while bone marrow-derived macrophages that underwent in vitro differentiation show opening-dominant ratios (2.6\u0026ndash;160). Methodological validation using MACS3 peak calling, sensitivity analysis across window sizes (500 bp\u0026ndash;10 kb), and fold-change thresholds (1.2x\u0026ndash;3.0x) confirms that the closing-dominant pattern is robust and method-independent. Time-resolved analysis reveals that chromatin closing begins within one hour of β-glucan exposure, with selective opening of a small gene set that shifts from cytokine receptors (day 1) to response machinery with built-in brakes such as IL1RN (day 6). Cross-dataset corroboration using ATAC-seq, H3K4me3, H3K27ac, and RNA-seq identifies IRF3 as being in a poised state: chromatin open, H3K4me3 elevated, but not yet transcriptionally active. These findings highlight the need to delineate stimulus-driven chromatin changes from differentiation-associated remodeling, and establish the open/close ratio as a standardized metric for comparing chromatin states across immune memory paradigms.\u003c/p\u003e","manuscriptTitle":"Open/close ratio reveals cell differentiation state as the primary determinant of chromatin balance in immune memory","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-25 14:27:19","doi":"10.21203/rs.3.rs-9364995/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"editorInvitedReview","content":"","date":"2026-05-09T00:41:19+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-04-29T01:44:17+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"236267417487996538451167857203459265904","date":"2026-04-17T19:42:40+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"261751833677767519742925168865021230082","date":"2026-04-16T21:42:32+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-04-16T14:37:20+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-04-11T08:06:44+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-04-10T04:53:22+00:00","index":"","fulltext":""},{"type":"submitted","content":"Epigenetics \u0026 Chromatin","date":"2026-04-09T07:44:06+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"epigenetics-and-chromatin","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"epch","sideBox":"Learn more about [Epigenetics \u0026 Chromatin](http://epigeneticsandchromatin.biomedcentral.com/)","snPcode":"13072","submissionUrl":"https://submission.nature.com/new-submission/13072/3","title":"Epigenetics \u0026 Chromatin","twitterHandle":"@EpigenChromatin","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"712945f9-f0bf-4834-83ef-3d20d294e124","owner":[],"postedDate":"April 25th, 2026","published":true,"recentEditorialEvents":[{"type":"editorInvitedReview","content":"","date":"2026-05-09T00:41:19+00:00","index":20,"fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-04-25T14:27:19+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-25 14:27:19","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9364995","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9364995","identity":"rs-9364995","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}