Lorente-Sorolla et al. Genome Medicine (2019) 11:66 https://doi.org/10.1186/s13073-019-0674-2 RESEARCH Open Access Inflammatory cytokines and organ dysfunction associate with the aberrant DNA methylome of monocytes in sepsis Clara Lorente-Sorolla1,2, Antonio Garcia-Gomez1,2, Francesc Català-Moll1,2, Víctor Toledano3,4, Laura Ciudad1,2, José Avendaño-Ortiz3,4, Charbel Maroun-Eid3, Alejandro Martín-Quirós4, Mónica Martínez-Gallo5, Adolfo Ruiz-Sanmartín6, Álvaro García del Campo7, Ricard Ferrer-Roca6, Juan Carlos Ruiz-Rodriguez6, Damiana Álvarez-Errico1, Eduardo López-Collazo3,8,9 and Esteban Ballestar1,2* Abstract Background: Sepsis, a life-threatening organ dysfunction caused by a dysregulated systemic immune response to infection, associates with reduced responsiveness to subsequent infections. How such tolerance is acquired is not well understood but is known to involve epigenetic and transcriptional dysregulation. Methods: Bead arrays were used to compare global DNA methylation changes in patients with sepsis, non- infectious systemic inflammatory response syndrome, and healthy controls. Bioinformatic analyses were performed to dissect functional reprogramming and signaling pathways related to the acquisition of these specific DNA methylation alterations. Finally, in vitro experiments using human monocytes were performed to test the induction of similar DNA methylation reprogramming. Results: Here, we focused on DNA methylation changes associated with sepsis, given their potential role in stabilizing altered phenotypes. Tolerized monocytes from patients with sepsis display changes in their DNA methylomes with respect to those from healthy controls, affecting critical monocyte-related genes. DNA methylation profiles correlate with IL-10 and IL-6 levels, significantly increased in monocytes in sepsis, as well as with the Sequential Organ Failure Assessment score; the observed changes associate with TFs and pathways downstream to toll-like receptors and inflammatory cytokines. In fact, in vitro stimulation of toll-like receptors in monocytes results in similar gains and losses of methylation together with the acquisition of tolerance. Conclusion: We have identified a DNA methylation signature associated with sepsis that is downstream to the response of monocytes to inflammatory signals associated with the acquisition of a tolerized phenotype and organic dysfunction. Keywords: Sepsis, DNA methylation, Cytokines, Endotoxin tolerance, Monocytes Background intensive care units [2, 3]. The immune response during Sepsis is a life-threatening organ dysfunction caused by sepsis is complex and varies over time, with the con- a dysregulated host response to infection [1]. Sepsis can comitant occurrence of both pro-inflammatory and anti- induce acute kidney injury and multiple organ failures inflammatory mechanisms [3]. Despite intense study, the and represents the most common cause of death in cellular and molecular basis of human sepsis remains still unclear and effective therapies are lacking. * Correspondence: eballestar@carrerasresearch.org In many cases, sepsis survivors continue to succumb 1 Epigenetics and Immune Disease Group, Josep Carreras Research Institute (IJC), 08916 Barcelona, Spain to secondary challenges, latent infections, or malignan- 2 Chromatin and Disease Group, Cancer Epigenetics and Biology Programme cies several years after the initial septic episode [4]. It (PEBC), Bellvitge Biomedical Research Institute (IDIBELL), L’Hospitalet de has been recognized that sepsis leads to the acquisition Llobregat, 08908 Barcelona, Spain Full list of author information is available at the end of the article of tolerance, a state of reduced responsiveness to © The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Lorente-Sorolla et al. Genome Medicine (2019) 11:66 Page 2 of 18 subsequent stimulation after a primary bacterial insult several reasons. It has been mainly because DNA methy- that results in reduced cytokine production by mono- lation has a more limited range of effects than histone cytes and macrophages [5]. As a result, most patients modifications [17]. Nevertheless, DNA methylation with sepsis rapidly display signs of profound immuno- changes are generally highly relevant for the biology of suppression, associated with an increase in hypoxia in- myeloid cells [18]. On the one hand, various studies have ducible factor-1α expression that drives functional demonstrated the relevance of DNA (cytosine-5)-meth- reprogramming [6]. Such immune reprogramming is in yltransferase 3A (DNMT3A) and ten-eleven transloca- part due to disruption of homeostasis and defective cel- tion methylcytosine dioxygenase 2 (TET2), both lular energy metabolism which underlies the inability to enzymes respectively essential for the de novo incorpor- respond to secondary or further stimulation [7]. A num- ation and oxidation/removal of methyl groups to cyto- ber of mechanisms are involved in the homeostasis of sines, to the function of monocytes, dendritic cells, and the immune system, where monocytes are pivotal. These macrophages [19, 20]. On the other hand, DNA methy- cells recognize pathogen patterns or intercept and lation is generally associated with the stabilization of a phagocytose antigens, critical steps in eliminating bacter- transcriptional and functional state; thus, it is appealing ial infections, halting the increase in viral load, and to hypothesize that sepsis results in prolonged acquisi- eradicating certain neoplastic growths. Several monocyte tion of DNA methylation changes of the monocytes well subtypes are also crucial to the de-escalation of inflam- beyond the acute phase of sepsis, and perhaps contribut- mation and in wound healing [8]. In vitro experiments ing to stabilize the state of tolerance of monocytes. Most mimicking septic conditions have shown that upon importantly, TET2 has a role in resolution of inflamma- endotoxin re-challenge with Gram-negative bacteria or tion by recruiting HDAC2 to repress inflammatory genes only lipopolysaccharide (LPS), the major component of [21] and to promote sepsis-induced emergency myelo- the outer membrane of Gram-negative bacteria, tolerized poiesis [22]. monocytes/macrophages show a drastic downregulation In this study, we investigated DNA methylation of inflammatory cytokines (e.g., tumor necrosis factor changes in monocytes from individuals who have experi- [TNF] α, interleukin [IL]-6, IL-1β, IL-12) in parallel with enced an episode of sepsis. We performed DNA methy- the upregulation of anti-inflammatory cytokines like IL- lation profiling where we compared sorted monocytes 10, transforming growth factor (TGF) β, and IL-1RA as from patients with sepsis and healthy controls. The ana- compared to non-tolerized cells challenged with the lysis revealed the existence of significant DNA methyla- same stimuli. These tolerant monocytes/macrophages tion differences between the two groups in CpG sites also show an impaired antigen presenting capacity corre- mapping at genes relevant for monocyte-related immune lated with decreased expression of human leukocyte responses. Most importantly, we identified a significant antigen (HLA)-DR and some costimulatory molecules relationship between DNA methylation data and IL-10 [9] and upregulation of the immune checkpoint ligand and IL-6 cytokine levels, which are significantly in- PD-L1 [10]. creased in patients with sepsis, as well as with organ dys- The acquisition of endotoxin tolerance is accompanied function. We have determined that changes in DNA by a remodeling of the epigenomic profiles [11–13]. methylation are determined by TLR stimulation and the Most studies have focused on histone modification altered levels of inflammatory cytokines. Our findings changes. Following LPS stimulation of macrophages, also highlight the implication of TLR stimulation and cy- toll-like receptor (TLR)-induced genes are categorized tokines under sepsis in establishing and perpetuating the into two classes: tolerized and non-tolerized genes. Tol- dysregulated epigenetic signature and phenotype of erized genes, which include inflammatory genes, show monocytes. repressed expression whereas non-tolerized genes in- crease their levels. Transcriptional activation of non- Methods tolerized genes is associated with high levels of histone Human samples acetylation and H3K4me3 at their promoters [11]. In We selected and diagnosed patients with sepsis based on addition, monocytes exposed to LPS showed changes in the Third International Consensus Definitions for Sepsis H3K27ac, H3K4me1, and H3K4me3 [12]. It has also and Septic Shock (Sepsis-3) [1]. For each patient, we cal- been shown that during endotoxin tolerance, leukocytes culated the Sequential [Sepsis-related] Organ Failure As- display increased levels of repressive H3K9me2 mark at sessment (SOFA) score. The study included 14 patients the promoter regions of the IL1B and TNF genes [14, with bacterial infections with SOFA ranging from 2 to 8. 15]. Specifically, the H3K9 histone methyltransferase, Patients were obtained from La Paz University Hospital G9a, is essential for silencing the TNFA gene [16]. and Vall d’Hebron University Hospital. Blood samples In this context, DNA methylation changes have re- were collected at the first 12 h of sepsis diagnosis, which ceived less attention than histone modifications for was confirmed using clinical and analytical data. The Lorente-Sorolla et al. Genome Medicine (2019) 11:66 Page 3 of 18 clinical data of the patients included in the study are For in vitro experiments, we obtained buffy coats from summarized in Table 1 and Additional file 1: Table S1. anonymous donors through the Catalan Blood and Tis- We also studied a group of individuals with non- sue Bank (CBTB). The CBTB follows the principles of infectious systemic inflammatory response syndrome the World Medical Association (WMA) Declaration of (SIRS), formed by 4 patients in the immediate postop- Helsinki. Before providing blood sample, all donors re- erative period of cardiac surgery. In this group, the ceived detailed oral and written information and signed blood samples were obtained in the first 24 h of the a consent form at the CBTB. PBMCs were isolated by postoperative period. Finally, we also included blood density gradient centrifugation. Then, PBMCs were re- samples from 11 healthy controls collected from the suspended in Roswell Park Memorial Institute (RPMI) blood donor service of La Paz University Hospital and Medium 1640 + GlutaMAXTM-1 (Gibco, Life Tech- Vall d’Hebron University Hospital. The Committee nologies, CA, USA) containing 10% human pooled for Human Subjects of La Paz University Hospital serum (One Lambda, ThermoFisher Scientific Brand, (PIE2392) and Vall d’Hebron University Hospital (PR Canoga Park, CA, USA), 100 units/ml penicillin, and (ATR)122/2019) approved the study, which was con- 100 mg/ml streptomycin, and cells were non-treated ducted in accordance with the ethical guidelines of (control) or treated with 10 ng/ml LPS from Escherichia the 1975 Declaration of Helsinki. All samples were in coli (O111:B4, Sigma-Aldrich, Darmstadt, Germany), compliance with the guidelines approved by the local and cultured in poly-HEMA (Santa Cruz Biotechnology, ethics committee, and all patients (sepsis, SIRS, and Dallas, TX, USA)-coated plates for 4 days. After that, healthy controls) received oral and written informa- monocytes were purified by cell sorting using the same tion about the possibility that their blood would be strategy as mentioned above. Alternatively, CD14+ mono- used for research purposes and signed informed cytes were isolated using positive selection with CD14 consent. magnetic Microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany) and cultured in the same conditions as PBMCs. Purification of human monocytes from patients with Purified monocytes were pelleted and stored until DNA sepsis, SIRS, and healthy control samples or RNA extraction. Supernatants were collected and Peripheral blood mononuclear cells (PBMCs) were ob- stored at − 80 °C until cytokine measurement. tained from blood by density gradient centrifugation using The endotoxin tolerance status of the septic patients lymphocyte isolation solution (Rafer, Zaragoza, Spain). and healthy donors was evaluated by exposing whole PBMCs were stained with CD14-PE (Miltenyi Biotec, blood to a stimulus of 5 ng/ml LPS, collecting super- Bergisch Gladbach, Germany), CD66b-APC (STEMCELL™ natant after 3 h to determine the state of innate immune Technologies, Vancouver, Canada), and Fixable Viability system (IIS) during initial infection. Dye eFluor™ 520 (ThermoFisher Scientific, MA, USA), and then, cells were fixed with 2% formaldehyde Cytokine measurements methanol-free (ThermoFisher Scientific). Pure monocytes The cytokine levels in whole blood were determined using were isolated as CD14+CD66b− cells using flow cytometry the cytometric bead array (CBA) Flex Set (BD Biosciences, sorting (MoFlo Astrios EQ, Beckman Coulter Spain, San Jose, CA, USA), following the manufacturer’s proto- L’Hospitalet de Llobregat, Barcelona, Spain). Purified sam- col. The collected data were analyzed with Flow Cytomet- ples were pelleted and stored at − 80 °C. ric Analysis Program (FCAP) Array Software v3.0 (BD Table 1 Summary of the patient cohorts in the study Healthy controls SIRS-cardio patients Septic patients p value (sepsis vs control) p value (sepsis p value (SIRS vs vs SIRS) control) N 11 4 14 Age (mean ± SD) 51 ± 11.8 67.8 ± 5.8 74.6 ± 14.5 0.0010*** 0.3662 0.0130** Sex (% female) 27.3 25 57.1 0.1353 0.2568 0.9299 SOFA 0 3.8 ± 2.2 3.9 ± 2.0 N/A 0.8226 N/A Basal cytokine levels TNFα (pg/ml) (mean ± SD)† 9.0 ± 17.0 2.7 ± 3.1 561.4 ± 2082.8 0.9774 0.7845 0.7396 IL6 (pg/ml) (mean ± SD)† 11.9 ± 13.1 123.8 ± 101.4 4273.4 ± 10,789.5 0.0002*** 0.7906 0.0050*** † IL10 (pg/ml) (mean ± SD) 5.8 ± 5.1 81.6 ± 41.4 55.6 ± 72.9 0.0005*** 0.1236 0.0050*** p values of continuous data were calculated using Mann-Whitney U test and categorical data were calculated using chi-squared test. Statistically significant tests are represented as **p < 0.01 and ***p < 0.005 † These correspond to the basal levels of the cytokines measured in the serum Lorente-Sorolla et al. Genome Medicine (2019) 11:66 Page 4 of 18 Biosciences). For in vitro experiments, the concentration libraries minfi, lumi, and limma. Data quality was of cytokines was measured from the cell culture superna- assessed using the standard pipeline from the minfi tants using an enzyme-linked immunosorbent assay package. The data were Snoob-normalized and, after (ELISA), according to the manufacturer’s instructions normalization, beta and M values were calculated. To (BioLegend, San Diego, CA, USA). exclude technical and biological biases, we developed a pipeline with several filters as removing CpGs with SNPs DNA methylation profiling using universal bead arrays, overlapped. To minimize the potential confounding in- bisulfite sequencing, and pyrosequencing fluence of age and gender, we used these parameters as Infinium MethylationEPIC BeadChip (Illumina, Inc., San covariates. Diego, CA, USA) array were used to analyze DNA In this study, we considered a probe to be differentially methylation. This platform allows > 850,000 methylation methylated if it had a methylation differential of 15% sites per sample to be interrogated at single-nucleotide (Δβ ≥ 0.15) and when the statistical test was significant resolution, covering 99% of reference sequence (RefSeq) [p < 0.01 and false discovery rate (FDR) < 0.05]. In genes. The samples were bisulfite-converted using EZ addition, we used the iEVORA algorithm [23] to desig- DNA Methylation-Gold™ Kit (Zymo Research, Irvine, nate a probe as differentially variable. This algorithm de- CA, USA) and were hybridized in the array following the tects the homogeneity of variances using Bartlett’s test manufacturer’s instructions. (FDR < 0.001) and then selects those probes whose t test Each methylation data point was obtained from a com- is significant (p < 0.05 and FDR < 0.05) in order to bination of the Cy3 and Cy5 fluorescent intensities from regularize the variability test which is overly sensitive to the M (methylated) and U (unmethylated) alleles. For single outliers. representation and further analysis, we used beta and M Spearman’s correlation was used to correlate methyla- values. Beta value is the ratio of the methylated probe in- tion changes with cytokine concentration. Spearman’s tensity to the overall intensity (the sum of the methyl- correlation coefficient is a nonparametric approach to ated and unmethylated probe intensities). The M value measuring the strength of association of two variables is calculated as the log2 ratio of the intensities of the being more reliable with non-linear data. We used the methylated versus unmethylated probe. Beta values parameters specified in each section for Spearman’s range from 0 to 1, in which 0 is no methylation and 1 is analysis. complete methylation, and were used to derive heatmaps and to compare DNA methylation percentages from bi- Gene ontology analysis, transcription factor (TF) sulfite pyrosequencing experiments. For statistical pur- enrichment analysis, and chromatin state discovery and poses, the use of M values is more appropriate. characterization (ChromHMM) Bisulfite pyrosequencing was used to validate CpG Gene ontology (GO) was analyzed using the Genomic methylation changes. DNA was isolated using ReliaPrep™ Regions Enrichment of Annotations Tool (GREAT, ver- FFPE gDNA Miniprep System (Promega, Madison, WI, sion 3.0.0) (http://great.stanford.edu/public/html/). USA) for methylation array samples and with Maxwell® GREAT assigns biological meaning to a set of non- RSC Cultured Cells DNA Kit (Promega, Madison, WI, coding genomic regions by analyzing the annotations of USA) for in vitro model. Bisulfite modification of genomic the nearby genes [24]. For gene identification, we DNA isolated from monocytes was performed using EZ assigned a window that extends 5 kb upstream and 5 kb DNA Methylation-Gold™ Kit (Zymo Research, Irvine, CA, downstream from the differentially methylated CpG site. USA) following the manufacturer’s protocol. Primers for This window allows the analysis of CpGs located in PCR amplification and sequencing were designed with the regulatory regions distant to a TSS. Enrichment is PyroMark® Assay Design 2.0 software (QIAGEN, Hilden, showed as –log10 raw binomial p values. Germany). See list of primers in Additional file 2: Table S2. We used the findMotifsGenome.pl program of the PCRs were performed with the IMMOLASE™ DNA HOMER suite to look for motifs that are overrepresented polymerase PCR kit (Bioline Reagents Limited, in the target set relative to the background set (software London, UK), and the success of amplification was v4.5). It was used to identify enrichment of TF binding assessed by agarose gel electrophoresis. PCR products motifs in the 500-bp window upstream and downstream were pyrosequenced with the PyromarkTM Q24 sys- of the differentially methylated CpG sites [25]. Annotated tem (QIAGEN, Hilden, Germany). CpGs in the EPIC array were used as background. Chromatin state discovery and characterization Quality control, data normalization, and detection of (ChromHMM algorithm) was used to analyze enrich- differentially methylated and variable CpGs ment of the different chromatin states for the corre- Methylation array data were processed in the statistical sponding CpG sites [26]. The enrichment among language R using methods from the Bioconductor chromatin states is defined using the 18-state Lorente-Sorolla et al. Genome Medicine (2019) 11:66 Page 5 of 18 ChromHMM model (Roadmap Epigenomics Integra- methylation levels (hypomethylated) than control mono- tive Analysis Hub, ChromHMM track of the UCSC cytes (Fig. 1c and Additional file 4: Table S3). Genome Browser) (http://www.roadmapepigenomics. We performed gene ontology (GO) analysis to deter- org/) based on six chromatin marks (H3K4me3, mine whether the differentially methylated genes were H3K4me1, H3K27ac, H3K36me3, H3K27me3, and associated with relevant biological processes in sepsis. In H3K9me3). A Fisher’s exact test was used to assign the hypermethylated set, there was enrichment of im- odds ratio and p value. portant GO categories such as MAPK signaling pathway, NF-κB signaling pathway, and inflammation mediated by Quantitative reverse transcription polymerase chain chemokine-cytokine signaling pathway. In the hypo- reaction (qRT-PCR) methylated group, the functional categories were also RNA was isolated by Maxwell® RSC simplyRNA kit (Pro- relevant in the context of immune cell biology, including mega, Madison, WI, USA) and reverse-transcribed using the MHC class II protein complex, genes involved in the Transcriptor First Strand cDNA Synthesis Kit interferon-gamma (IFN-γ) signaling, and phagocytic (Roche, Basel, Switzerland) according to the manufac- vesicle membrane (Fig. 1d). The analysis of the chroma- turer’s instructions. qRT-PCR was performed in tripli- tin states of differentially methylated CpG sites revealed cate using LightCycler® 480 SYBR Green Mix (Roche, the enrichment in active and weak enhancers (character- Basel, Switzerland). Expression values were normalized ized by H3K4me1 and H3K27ac) for both the hyper- against the expression of the endogenous gene controls and hypomethylated sets and also for transcription start as RPL38. See list of primers in Additional file 2: Table site (TSS) flanking regions in the case of the hyper- S2. methylated set (additionally marked by H3K4me3) (Fig. 1e). We then inspected the enrichment of TF binding mo- Statistical analysis tifs among the two sets of differentially methylated CpG Data were analyzed with Prism version 6.0 (GraphPad). sites. We observed a significant overrepresentation of Statistical analyses were performed using the Mann- binding sites of the interferon regulatory factor (IRF) Whitney test, except as indicated. The levels of signifi- and ETS TF families in hypermethylated regions in sep- cance were as follows: *p < 0.05, **p < 0.01, ***p < 0.001. tic monocytes (Fig. 1f). Previous reports have shown that ETS factors such as PU.1 can recruit DNA methyltrans- Results ferases [27]. Hypermethylation could also antagonize the Monocytes from individuals who have undergone sepsis function of these TFs in endotoxin response [28]. Se- display an aberrant methylation signature quences around hypomethylated CpG sites were We first performed DNA methylation screening on mono- enriched for binding motifs of the signal transducer and cytes, sorted from peripheral blood as CD14+CD66b− activator of transcription (STAT) family (Fig. 1f). Ac- cells (Fig. 1a and Additional file 3: Figure S1), from a co- cording to previous studies, the Janus kinase (JAK)/ hort of 14 septic patients (Table 1 and Additional file 1: STAT pathway plays a critical role in protective immun- Table S1) and compared it with those sorted from a co- ity during sepsis via controlling cytokine responses hort of 11 healthy controls. We also included 4 patients (reviewed in [29]). with systemic inflammatory response syndrome (SIRS) Inspection of individual genes among those containing following cardiac surgery. For the analysis, we used bead differentially methylated CpG sites made it possible to arrays to interrogate the DNA methylation status of > 850, identify some with essential functions in monocyte/ 000 CpG sites across the entire genome covering 99% of macrophage biology and function. These included genes RefSeq genes. In the analysis, to minimize the potential such as IL1A, CCL22, CCR2, and STAT3 in the hyper- confounding influence of age and gender (shown in methylated set and HLA-A, SOCS1, IL1R2, and CD46 in Table 1), we used these parameters as covariates. the hypomethylated set (Fig. 1g and Table 2). IL1A is a Principal component analysis (PCA) showed the two pro-inflammatory cytokine. Different pro-inflammatory groups of monocytes from patients with sepsis and con- cytokines induce JAK activation, as well as the phos- trols separated along the first principal component phorylation and activation of transcription activators (Fig. 1b), with the SIRS in between. Overall, we observed STAT3, STAT5, and STAT6. CCL22 and CCR2 associ- a wider heterogeneity in septic monocytes than in con- ate with leukocyte chemotaxis. HLA-A belongs to the trol and SIRS monocytes (Fig. 1b), perhaps due to the di- MHC class I and is involved in antigen presentation. versity of infective bacteria (Additional file 1: Table S1). SOCS1 encodes a member of the suppressor of cytokine Monocytes from sepsis patients display 595 CpG sites signaling (SOCS) family that is involved in negative with significantly higher methylation levels (hypermethy- regulation of JAK/STAT cytokine signaling. Another ex- lated) and 302 CpG sites with significantly lower ample is CD46, encoding a costimulatory factor for T Lorente-Sorolla et al. Genome Medicine (2019) 11:66 Page 6 of 18 Fig. 1 (See legend on next page.) Lorente-Sorolla et al. Genome Medicine (2019) 11:66 Page 7 of 18 (See figure on previous page.) Fig. 1 Global analysis of DNA methylation changes in septic monocytes. a Representative flow cytometry profiles indicating the sorting strategy and gates used in this study. Monocytes (MOs) (CD14+ CD66b−) were sorted from healthy controls and patients (SIRS and sepsis). b Principal component analysis (PCA) of methylation heatmap data for control, SIRS, and septic monocytes (in blue, green, and red respectively). c DNA methylation heatmap showing differentially methylated CpGs between controls (CON, blue) and patients with sepsis (SEP, red). The heatmap includes all CpG-containing probes displaying significant methylation changes (15% of differential of beta values, p < 0.01 and false discovery rate (FDR) < 0.05). A scale is shown at the bottom left ranging from − 2 (lower DNA methylation levels, blue) to + 2 (higher methylation levels, red). d Gene ontology (GO) analysis of genes associated with differentially methylated CpG sites showing the most relevant and significantly enriched categories resulting from the Genomic Regions Enrichment of Annotations Tool (GREAT). e Enrichment analysis of the different chromatin states for CpG sites corresponding to each methylation cluster (left to hypermethylation, right to hypomethylation). The relative enrichment of the different states is represented using the odds ratio. Dot size represents the FDR value. Tss, transcription start site; Enh, enhancer; Repr, repressed region; PC, Polycomb. f TF binding motif analysis of differentially methylated CpGs between control and sepsis. The panel shows fold change (FC), TF family and factor (selected TF with p ≤ 1e−05 for hypermethylated regions and p ≤ 1e−03 for hypomethylation). Motif logo is representative of the TF family. g Box plots showing β-values obtained from the DNA methylation array. We observed hypermethylation and hypomethylation in important immune system genes. The CpG sites are marked with a green line in the gene scheme placed on top of each graph, where the TSS is marked with a red arrow lymphocytes promoting T-regulatory 1 cells that sup- DNA methylation changes in monocytes from patients press immune response through IL-10. All these exam- with sepsis correlate with increased IL-10 and IL-6 levels ples highlight the relationship between the genes Many of the characteristics of endotoxin-tolerized mono- undergoing changes in DNA methylation and pathways cytes/macrophages resemble that of anti-inflammatory related to the acquisition of endotoxin tolerance. M2 macrophages [30, 31]. M2 macrophages show down- regulated inflammatory cytokines (e.g., IL-12, TNFα) but Septic monocytes also display increased DNA methylation upregulated anti-inflammatory cytokines (e.g., IL-10), variability scavenger receptor expression, and efficient phagocytosis. As indicated above, our PCA analysis indicated a higher het- We wondered whether the DNA methylation patterns ob- erogeneity in the DNA methylation profiles of monocytes served in monocytes from patients with sepsis might be from patients with sepsis. We used a recently developed al- associated with the generation of an immunosuppressive gorithm, named iEVORA [23], to determine significant dif- or tolerogenic environment in peripheral blood. ferentially variable CpG positions (DVPs) at an FDR < 0.05 To address this question, we first tested the levels of a between monocytes from patients with sepsis and healthy in- panel of cytokines associated with acute sepsis in cul- dividuals. By using this method, we not only could confirm tured PBMCs from patients with sepsis and compared it the occurrence of a higher number of DVPs in monocytes to those in control individuals. We identified signifi- from patients with sepsis (n = 6833) versus healthy controls cantly increased levels of IL-10 and IL-6 in patients with (n = 148) (Fig. 2a) but also determine that many of these sites sepsis (Fig. 3a). We also examined the levels of cytokines occur at genes that also display significant DNA methylation following the exposure to LPS. This analysis showed that changes (Additional file 5: Table S4). the majority of the septic patients had acquired toler- Similarly to our previous analysis of differentially ance, with decreased levels of pro-inflammatory TNFα, methylated sites, sepsis-associated DVPs mainly occur at IL-1β, and IL-6, following exposure to LPS, and in- enhancers and also TSS flanking regions (Fig. 2b). Se- creased secreted levels of anti-inflammatory IL-10, indi- quences surrounding these DVPs were highly enriched cating a higher degree of tolerogenic properties on in binding motifs of ETS and IRF families (Fig. 2c), again septic monocytes (Fig. 3a), as previously reported [32]. highlighting the role of these TFs in the acquisition of Given the greater heterogeneity of the methylation the tolerogenic phenotype of septic monocytes. profiles of septic samples with respect to those from GO analysis revealed an enrichment of sepsis-associated healthy controls, we investigated whether this could also DVPs in positive regulation of inflammatory response, be related to the wider range of cytokine levels, such as leukocyte chemotaxis, and defense response to bacterium those observed for IL-10 or IL-6. To this end, we first and toll-like receptor binding (Fig. 2d). We found sepsis- performed Spearman’s correlation between the DNA associated DVPs in genes important for immune response methylation data and the secreted IL-10 levels, for all pa- against infection and hyperinflammation such as IL1A and tients with sepsis and control individuals. This analysis TNF (Fig. 2e and Additional file 5: Table S4). revealed that there are 855 CpG sites that become All this suggests that a variety of factors encompassing hypermethylated and 389 CpG sites hypomethylated in bacterial infection and sepsis are driving DNA methyla- relation to increasing levels of IL-10 (r > 0.5; Δβ ≥ 0.15) tion and phenotypic changes in these cells in a similar (Fig. 3b and Additional file 6: Table S5). This analysis manner, regardless of the infecting bacteria or also showed that, for CpG sites associated with IL-10 individual-specific clinical outcome of the individual. levels, there is a significant difference of their median Lorente-Sorolla et al. Genome Medicine (2019) 11:66 Page 8 of 18 Table 2 List of differentially methylated genes in septic monocytes Lorente-Sorolla et al. Genome Medicine (2019) 11:66 Page 9 of 18 Fig. 2 Analysis of differentially variable CpG positions (DVPs) in monocytes from patients with sepsis. a Volcano plot showing the p value vs the variance ratio for healthy control, SIRS, and sepsis-associated DVPs. DVPs were identified using the algorithm iEVORA. b Enrichment of sepsis- associated DVPs at 18 distinct chromatin states using ChromHMM. c TF binding motif analysis for sepsis-associated DVPs. Bubbles are colored according to TF family. The p value is indicated by bubble size (selected TF with p ≤ 1e−07 for DVP regions). d GO categories resulted from GREAT analysis for sepsis-associated DVPs of section (a). e Representative examples showing beta values of DVPs DNA methylation levels between septic and healthy con- increasing levels of IL-6 (r > 0.5; Δβ ≥ 0.1) (Add- trols (Fig. 3c), reinforcing the notion of the existence of itional file 7: Figure S2A and Additional file 8: Table S6). differential methylation patterns between control and CpG sites displaying changes in methylation in relation sepsis in addition to a contribution of IL-10 to the ac- to IL-10 levels enriched for specific chromatin features quisition of such changes. Spearman’s correlation be- (Fig. 3d). Specifically, we found enrichment for enhancers tween the DNA methylation data and the IL-6 levels in both the hypermethylated and hypomethylated sets. identified 2492 CpG sites becoming hypermethylated We also investigated for this set of CpG sites the enrich- and 909 CpG sites hypomethylated in relation to ment in TF binding motifs for these differentially Lorente-Sorolla et al. Genome Medicine (2019) 11:66 Page 10 of 18 Fig. 3 (See legend on next page.) Lorente-Sorolla et al. Genome Medicine (2019) 11:66 Page 11 of 18 (See figure on previous page.) Fig. 3 DNA methylation changes in septic monocytes parallel the increase of IL-10 levels. a Cytokine measurement using cytometric bead array (CBA) from control and septic PBMCs before and after LPS stimulation (t = 3 h). Box and whisker plots show median values. Mann-Whitney test was used to determine significance (*p < 0.05, **p < 0.01, and ***p < 0.001). b DNA methylation heatmap of CpG changes in relation to IL-10 basal concentration (represented on the top of the heatmap as log scale). Spearman’s correlation was used with p < 0.01, r > 0.5, and Δβ ≥ 0.15. A scale is shown at the bottom, wherein beta values range from − 4 (lower DNA methylation levels, blue) to + 4 (higher methylation levels, red). c Violin plots corresponding to the 5mC-normalized data for control and sepsis presented in the heatmap in the previous section. The median and the interquartile range are represented. d Chromatin state characterization of differentially methylated sites for section b. The relative enrichment of the different state assignments is representing using the odds ratio. FDR is represented by the size of the dots. e Bubble scatterplot of TF enrichment for hypermethylated and hypomethylated CpGs. The x-axis shows the percentage of windows containing the motif, and the y-axis shows the fold enrichment of the motif. Bubbles are colored according to TF family. p value is indicated by the bubble size (selected TF with p < 1e−07 for hypermethylated and hypomethylation regions). f GO categories resulted from GREAT analysis of differentially methylated CpGs related to IL-10 concentration. g Genomic tracks representing the clusters of differentially methylated CpGs for protocadherins (PCDH, upper scheme) and human leukocyte antigen (HLA, lower panel). Blue and red lines represent the confidence intervals for each average values. A window of ± 50,000 bp was used methylated CpG sites. We observed that approximately to undergo aberrant DNA hypermethylation in cancer 25% of the hypermethylated CpG sites have motifs for and in other disorders [36, 37]. Moreover, recent bZIP TF factor family (Jun-AP-1) and 25% of the hypo- studies have identified mechanisms by which PCDHs methylated CpG sites display motifs for STAT family can regulate the Wnt pathway (reviewed in [38]), members (Fig. 3e), suggesting that methylation might be which further corroborates the Wnt pathway as a pu- driven by the stimulation of TLR by bacteria and the in- tative therapeutic target for the patient treatment. creased levels of IL-10, stimulating its receptor IL-10R. Regarding the hypomethylated CpG sites, we found an Interestingly, 15% of the hypomethylated sequences dis- enrichment in genes involved in the IFN-γ pathway, played enrichment of CTCF binding motifs, whose global which is essential for antimicrobial defense and restoring occupancy has been linked to differential DNA methyla- monocyte deactivation in patients with sepsis [39]. Re- tion [33]. Similar results were obtained when we analyzed markably, among the CpG sites displaying changes in the data for the Spearman’s correlation with IL-6 levels methylation in relation to IL-10 levels, we identified 23 (Additional file 7: Figure S2C and S2D). CpG sites within the HLA cluster, which is also induced GO analysis revealed distinct enriched biological pro- by the IFN-γ and JAK/STAT pathway [40] (Fig. 3g). cesses for hypo- and hypermethylated CpG clusters in relation to IL-10 levels, including MHC class II protein Monocytes exposed to LPS undergo DNA methylation complex and response to type I IFN, and fibroblast changes in parallel with the acquisition of endotoxin growth factor-activated receptor activity respectively tolerance (Fig. 3f). Of note, similar GO categories were also Our results suggest that TLR stimulation and the inflam- enriched for differentially methylated CpGs correlating matory environment generated in the context of sys- to IL-6 levels (Additional file 7: Figure S2B). temic bacterial infection are able to induce DNA Our data show that many members of Wnt signal- methylation changes in monocytes. First, the specific ing pathway (WNT3A, WNT6, and AXIN2 among DNA methylation profiles of monocytes from patients others) (Additional file 6: Table S5) display a gain of with sepsis associate with IL-10 and IL-6 levels. Second, methylation in septic monocytes compared with their there is an enrichment of binding motifs for AP-1 and healthy counterparts, highlighting the potential link STATs. This suggests that both the stimulation of TLRs, between aberrant DNA methylation and the Wnt the resulting inflammatory conditions, and subsequent pathway. In fact, cumulative evidence supports the anti-inflammatory signals, participate in shaping the role of the Wnt pathway in the regulation of the generation of aberrant methylation profiles which might macrophage-mediated inflammatory response in sepsis modulate and stabilize the phenotype of monocytes fol- [34], in which Wnt3a and Wnt6 reduce TNFα secre- lowing a septic episode. We therefore explored the tion and promote the differentiation towards an M2 possibility of inducing in vitro DNA methylation anti-inflammatory phenotype attenuating the immune changes observed in such conditions by exposing in response [35]. Furthermore, we found a large genomic parallel PBMCs and monocytes from healthy individ- region that also displayed DNA hypermethylation in uals to LPS, and compare it with PBMCs and mono- which it predominantly covered CpG sites in the cytes without such stimulation (Fig. 4a). We cultured three tandem gene clusters of protocadherin (PCDHA, the cells for 4 days and measured the acquisition of PCDHB, and PCDHG) (Fig. 3g) (Additional file 6: tolerance. In the case of PBMCs, we isolated Table S5). This region has previously been reported monocytes, sorted as CD14+CD66b− cells after these Lorente-Sorolla et al. Genome Medicine (2019) 11:66 Page 12 of 18 Fig. 4 In vitro exposure to LPS induces the acquisition of tolerance and DNA methylation changes similar to those observed in sepsis. a Schematic diagram depicting our in vitro model for sepsis. PBMCs were cultured with or without LPS during 4 days, and then, monocytes were sorted as CD14+CD66b− cells for subsequent analyses. In parallel, monocytes were isolated with magnetic CD14 antibody and cultured in the same conditions. b TNFα and IL-10 production, as determined by ELISA, from PBMCs or monocyte supernatants, following washout and resting following the 4 initial days (with/without LPS) and an 8-h exposure to LPS, as indicated in the scheme in a. c Bisulfite pyrosequencing of selected hypermethylated (IL15) and hypomethylated (IL1R2) genes in sorted CD14+CD66b− monocytes from PBMCs in the in vitro sepsis model. d IL1B, IL6, IL15, and IL1R2 mRNA levels were analyzed by quantitative RT-PCR using RPL38 as control in the same sorted monocytes. e Bisulfite pyrosequencing of IL15 and IL1R2 in CD14+ monocytes in the in vitro sepsis model. f IL1B, IL6, IL15, and IL1R2 mRNA levels analyzed by quantitative RT-PCR in CD14+ monocytes 4 days (Fig. 4a, top). These experiments were Specifically, these assays showed decreased levels of performed in the presence of poly-2-hydroxyethyl TNFα and increased levels of IL-10 in a second LPS methacrylate (poly-HEMA), an agent that restricts the exposure for those cells that had a first exposure to attachment of the monocytes to the plates and LPS (Fig. 4b). We then performed pyrosequencing of therefore their differentiation to macrophages, to a selection of hypo- and hypermethylated genes in prevent the occurrence of vast changes in DNA septic monocytes. Our analysis revealed that the methylation [19, 20]. Enzyme-linked immunosorbent in vitro stimulation with LPS is able to induce DNA assays (ELISA) revealed in both cases the acquisition methylation changes in these genes (Fig. 4c) with a similar of tolerance following the initial encounter with LPS. trend to the changes observed in patients with sepsis when Lorente-Sorolla et al. Genome Medicine (2019) 11:66 Page 13 of 18 PBMCs were exposed to LPS (followed by monocyte participation among the mechanisms leading to the gen- purification). In the case of monocytes directly exposed to eration of an aberrant phenotype of these cells. On the LPS, we were able to stimulate only demethylation of the other hand, correlation analyses of the DNA methylation IL1R2 gene, suggesting that perhaps signals from other profiles in relation to IL-10 and IL-6 levels, which are in- cell types are necessary to induce gains of methylation creased in patients with sepsis, suggest a potential mech- following LPS treatment. Interestingly, changes in DNA anism downstream to these cytokines participating in methylation at the aforementioned sites occurred in asso- the defective generation of DNA methylation alterations. ciation with changes in gene expression of these genes Furthermore, in vitro analysis of the influence of bacter- (Fig. 4d, e) in the two sets of samples exposed to LPS, and ial LPS and inflammatory context in determining the ac- not in controls. In addition, we also observed an increase quisition of DNA methylation alterations in monocytes in mRNA levels of pro-inflammatory cytokines (IL-1β and also shows how these changes associate with aberrant IL6) in LPS-treated cells. These results proved that both transcriptional levels of dysregulated genes. Finally, our gains and losses of DNA methylation and expression in analysis shows increasing changes in DNA methylation septic monocytes are the result of TLR stimulation and in relation to organ dysfunction. the generation of an inflammatory environment and are Epigenetic factors play a role in the acquisition of endo- associated with the acquisition of a tolerized state in toxin tolerance. In fact, seminal studies by the teams of monocytes. Netea, Logie, and Stunnenberg have shown that the tran- scriptional inactivity in response to a second LPS exposure Organ dysfunction associates with DNA methylation in tolerized macrophages is accompanied by a failure to de- changes posit active histone marks at promoters of tolerized genes We finally tested DNA methylation profiles in relation [17]. It has also been reported that leukocytes of patients to SOFA, the main score used to assess organ dysfunc- with sepsis have defects in important metabolism pathways tion. When using Spearman correlation, we determined and, interestingly, these immunometabolic defects were that there are 1890 CpG sites that become hypermethy- partially restored by therapy with recombinant IFN-γ [7]. lated and 1536 CpG sites hypomethylated in relation to The aforementioned studies paid less attention to increasing SOFA (p value < 0.01; r > 0.6) (Fig. 5a and DNA methylation changes, as they appeared to be less Additional file 9: Table S7). GO analysis revealed that prevalent than those at occurring in histone modifica- DNA methylation changes in relation to SOFA affect in- tions. However, DNA methylation is a relatively more flammatory response and antigen presentation in a simi- stable epigenetic mark than histone modifications. That lar manner than previous comparisons (Fig. 5b). We also property makes this modification worth of study: firstly, investigated for this set of CpG sites the enrichment in because it might have a long-term contribution to the TF binding motifs for these differentially methylated tolerized state of monocytes, and secondly, because it CpG sites. We obtained similar sets of TF binding motifs could potentially be used as a marker, if associations (Fig. 5c) like the ones observed for the correlations with with patient prognosis and/or progression were found. IL-10 and IL-6, suggesting the participation of these in- Many of the genes displaying differentially methylated flammatory cytokines in the acquisition of DNA methy- CpG sites between patients with sepsis and controls lation changes in relation to organic damage. Finally, occur within a limited number of pathways relevant to individual inspection of the lists revealed the association the stimulation associated with bacterial infection (Fig. 6). of several relevant molecules including HLA-A, IL19, For instance, genes experiencing changes in methylation IL15, and IL27 (Fig. 5d). Altogether, this analysis sug- such as IL1A, IL1R2, IL1R1, TAB2, TAB1, MAP 2K1, gested that the DNA methylation changes associated and MAP 3K1 occur within the same signaling pathway, with organic damage involve similar changes to those which is also merging from the signals downstream to observed in relation to inflammatory cytokines. TNF. On the other hand, several of the genes encoding for cytokines like IL27, IL23A, IL19, and IL22 also dis- Discussion play CpG sites undergoing methylation changes, as well Our study demonstrates for the first time the existence as genes encoding elements downstream to it, such as of DNA methylation alterations in human monocytes TYK2, JAK1, STAT3 and 4, and SOCS3 and SOCS5. The from individuals following a sepsis episode in relation to relationship between all these genes suggests two possi- the acquisition of a tolerized phenotype, paralleling data bilities: that DNA methylation changes at these sites obtained in a mouse model [41]. Most notably, changes have a causal effect in determining the activation or re- occur in genes relevant to the function of these cells in- pression of the associated genes or, alternatively, that cluding the interferon-gamma-mediated pathway and the monocyte methylome is acting as a sensor of the ac- MHC class II proteins. On the one hand, the observed tivation of these pathways through additional or alterna- methylation changes in patients with sepsis suggest their tive mechanisms. Lorente-Sorolla et al. Genome Medicine (2019) 11:66 Page 14 of 18 Fig. 5 DNA methylation changes in septic monocytes parallel organic damage. a DNA methylation heatmap of CpG changes in relation to the SOFA score (represented on the top of the heatmap). Spearman’s correlation was used with p < 0.01, r > 0.6. A scale is shown at the bottom, wherein beta values range from − 4 (lower DNA methylation levels, blue) to + 4 (higher methylation levels, red). b GO categories resulted from GREAT analysis of differentially methylated CpGs related to SOFA. c Bubble scatterplot of TF enrichment for hypermethylated and hypomethylated CpGs. The x-axis shows the percentage of windows containing the motif, and the y-axis shows the fold enrichment of the motif. Bubbles are colored according to TF family. p value is indicated by bubble size (selected TF with p ≤ 1e−06 for hypermethylated and hypomethylation regions). d Box plots showing β-values obtained of genes significantly correlating with the SOFA score Lorente-Sorolla et al. Genome Medicine (2019) 11:66 Page 15 of 18 Fig. 6 Scheme depicting important signaling pathways related to immunity and sepsis. Molecules whose encoding genes displayed DNA methylation alterations in this study are shown in red and blue for hypermethylation and hypomethylation respectively. The following proteins/ genes are represented in the figure: TLR, Toll-like receptor; MyD88, myeloid differentiation primary response 88; IRAK, interleukin-1 receptor- associated kinase; IL1R, interleukin-1 receptor; IL-1A, interleukin 1 alpha; TNFR, tumor necrosis factor receptor; TRADD, TNFR1-associated death domain; TRAF, TNF receptor-associated factor; TAK1, transforming growth factor (TGF) beta-activated kinase 1; TAB2, TGF beta-activated kinase 1 binding protein 2; MAPK, mitogen-activated protein kinase; IKK, IκB kinase; AP-1, activator protein 1; NF-кB, nuclear factor kappa B; CCL20, C-C motif chemokine ligand 20; CCL22, C-C motif chemokine ligand 22; IL-23A, interleukin 23A; IL-19, interleukin 19; IL-27, interleukin 27; IL-22, interleukin 22; JAK1, Janus kinase 1; STAT, signal transducer and activator of transcription; TYK2, tyrosine kinase 2; SOCS, suppressor of cytokine signaling. In this scheme, we have selected CpG sites/genes with a minimum 10% of differential of beta values, p < 0.01 and false discovery rate (FDR) < 0.05 In this respect, our results show that monocytes from the generation of a particular cytokine environment [42], patients with sepsis harbor an aberrant DNA methylation our results reinforce the notion that epigenetic changes signature that is related with the abnormal environment are related to the maintenance of the dysregulated im- derived from such inflammatory condition. Furthermore, mune response following an episode of sepsis. However, we have determined that DNA methylation changes cor- with the existing data generated in this study, we cannot relate with increased IL-10 and IL-6 levels and that those distinguish whether these epigenetic changes are a cause changes are functionally annotated to genes belonging to or a consequence: in other words, whether the DNA the Wnt and IFN-γ signaling pathways. In this respect, it methylation changes are caused by these individuals’ in- has been demonstrated, as mentioned above, that IFN- fection history, perhaps influenced by the environment, γ-based therapy can partially restore the defective or whether environmental factors cause the generation metabolic changes occurring in leukocytes from of the aberrant DNA methylation signature, which is patients with sepsis [7]. then accompanied by immune responses that are sec- Finally, our investigation demonstrates the sensitivity ondary to the sepsis. It is likely that the aberrant DNA of monocytes to translate environmental changes into methylation (and expression signature) generated as a more stable changes at the transcriptional level through result of the particular cytokine milieu generated under DNA methylation. Given that sepsis is associated with the sepsis episode contributes to perpetuating the Lorente-Sorolla et al. Genome Medicine (2019) 11:66 Page 16 of 18 tolerized state of these monocytes. Whether those Additional file 7: Figure S2. DNA methylation changes in septic changes also affect bone marrow monocyte progenitors monocytes parallel the increase of IL-6 levels. (a) DNA methylation heat- giving long-lasting reprogramming, as occurs with map of CpGs changes in relation to IL-6 basal concentration. Spearman’s correlation was used with p < 0.01, r > 0.5 and differential β-value ≥0.1. A trained immunity [43], remains to be investigated. Inter- scale is shown at the bottom, wherein beta values range from − 4 (lower estingly, the observed changes appear to be reflective of DNA methylation levels, blue) to + 4 (higher methylation levels, red). (b) the infection, as it suggests the data obtained for SIRS GO categories for differentially methylated CpGs of section (a). (c) Enrich- ment of differentially hyper- and hypo-methylated CpGs among chroma- patients (even if it is a small cohort), which are also tin states, defined using the 18-state ChromHMM model. (d) HOMER characterized by inflammation and organ dysfunction. motif analysis for methylation changes. The x-axis shows the percentage A potential limitation of our study is the size and char- of windows containing the motif and the y-axis shows the fold enrich- ment of the motif. Bubbles are colored according to TF family. p value is acteristics of the cohort. In future studies, it would be indicated by bubble size (TF with p < 1e−15 for hypermethylated regions necessary to use vaster cohorts, including patients with a and p ≤ 1e−5 for hypomethylation were represented). representative number of Gram-positive and Gram- Additional file 8: Table S6. List of hypermethylated and negative bacteria and patients at different stages follow- hypomethylated CpGs related to IL-6 cytokine. ing the sepsis episode. However, the size and features of Additional file 9: Table S7. List of hypermethylated and hypomethylated CpGs related to SOFA. our cohort, on the other hand, indicate that common changes in DNA methylation are associated with sepsis Abbreviations regardless of the infective bacteria. The identification of AP-1: Activator protein-1; CBA: Cytometric bead array; CTF/NF1: CCAAT box- specific DNA methylation markers associated with the binding transcription factor/nuclear factor I; DNMT: DNA (cytosine-5)- infecting organism of the clinical outcome of the patient methyltransferase; DVPs: Differentially variable CpG positions; ELISA: Enzyme- linked immunosorbent assay; FDR: False discovery rate; GO: Gene ontology; will surely be useful for predicting the evolution of the HDAC: Histone deacetylase; HLA: Human leukocyte antigen; IKK: IκB kinase; patients and perhaps their clinical management. IL: Interleukin; IFN-γ: Interferon-gamma; IRF: Interferon regulatory factor; JAK: Janus kinase; LPS: Lipopolysaccharide; MAPK: Mitogen-activated protein kinase; MHC: Major histocompatibility complex; MyD88: Myeloid Conclusions differentiation primary response 88; NF-кB: Nuclear factor kappa B; In the present study, we have shown that patients with PBMCs: Peripheral blood mononuclear cells; PCA: Principal component sepsis undergo widespread changes in the methylome of analysis; PCDH: Protocadherins; poly-HEMA: Poly-2-hydroxyethyl methacrylate; SNP: Single-nucleotide polymorphism; SOCS: Suppressor of their circulating monocytes in parallel with the acquisi- cytokine signaling; STAT: Signal transducer and activator of transcription; tion of endotoxin tolerance. Thousands of changes are TET2: Ten-eleven translocation methylcytosine dioxygenase 2; associated with the aberrant levels of IL-10 and IL-6, as TF: Transcription factor; TGF: Transforming growth factor; TLR: Toll-like receptor; TNF: Tumor necrosis factor; TRAF: TNF receptor-associated factor; well as with organ dysfunction. Stimulation of the Toll- TSS: Transcription start site; TYK2: Tyrosine kinase 2 like receptor in monocytes induces similar changes in DNA methylation and expression, concomitant with the Authors’ contributions acquired tolerance that points to a major role in the EB, CL-S, and DAE conceived the experiments; CL-S, AG-G, VT, and LC per- formed the experiments; AG-G and FC-M performed the biocomputing ana- stabilization of a tolerized phenotype through these al- lysis; VT, JA-O, CM-E, AM-Q, MM-G, AR-S, CR-R, RF-R, AGC, and EL-C provided terations. Our results open up possibilities not only to the samples; CL-S, AG-G, FC-M, VT, DAE, EL-C, and EB analyzed the data; EB use DNA methylation as a marker for disease but also wrote the paper. All authors read and approved the final manuscript. for understanding its role in the acquisition of the aber- Funding rant phenotype of these cells. We thank CERCA Programme/Generalitat de Catalunya for the institutional support. This work was funded by the Spanish Ministry of Economy and Competitiveness (MINECO; grant numbers SAF2014-55942-R and SAF2017- Supplementary information 88086-R) to E. B and by “Fondo de Investigación Sanitarias” and “Fondos Supplementary information accompanies this paper at https://doi.org/10. FEDER” (PIE 15/00065 and PI14/01234) to E.L.-C. 1186/s13073-019-0674-2. Availability of data and materials Additional file 1: Table S1. Clinical features of septic patients, SIRS and Methylation array data for this publication have been deposited in NCBI’s healthy controls included in this study. Gene Expression Omnibus. Additional file 2: Table S2. Primers sequences. Lorente-Sorolla C, García-Gómez A, Català-Moll F, Toledano V, Ciudad L, Avendaño-Ortiz J, Maroun-Eid C, Martín-Quirós A, Martínez-Gallo M, Ruiz- Additional file 3: Figure S1. Purification and quality of monocytes. (a) Sanmartín A, García del Campo Á, Ferrer-Roca R, Ruiz-Rodriguez JC, Álvarez- Flow cytometry profiles indicating the sorting strategy and gates used in Errico D, López-Collazo E, Ballestar E. Inflammatory cytokines and organ dys- the study. (b) Cell type deconvolution of the hybridized samples using function associate with the aberrant DNA methylome of monocytes in sepsis. Houseman algorithm. Methylation profiling by array of CD14+CD66bneg monocytes of 11 healthy Additional file 4: Table S3. List of hypermethylated and donors, 4 SIRS, and 14 septic patients datasets. GEO Series accession number hypomethylated genes in septic monocytes (р-value< 0.01; FDR < 0.05; GSE138074. Δβ ≥ 0.15). Additional file 5: Table S4. List of differentially variable CpG positions Ethics approval and consent to participate (DVPs). The Committees for Human Subjects of La Paz University Hospital (PIE2392) and Vall d’Hebron University Hospital (PR (ATR)122/2019) approved the Additional file 6: Table S5. List of hypermethylated and hypomethylated CpGs related to IL-10 cytokine. study, which was conducted in accordance with the ethical guidelines of the 1975 Declaration of Helsinki. All samples were in compliance with the Lorente-Sorolla et al. Genome Medicine (2019) 11:66 Page 17 of 18 guidelines approved by the local ethics committee. All participants provided 15. El Gazzar M, Yoza BK, Hu JYQ, Cousart SL, McCall CE. Epigenetic silencing of written informed consent to participate in the study. tumor necrosis factor during endotoxin tolerance. J Biol Chem. 2007; 282(37):26857–64. Consent for publication 16. El Gazzar M, Yoza BK, Chen X, Hu J, Hawkins GA, McCall CE. G9a and HP1 Written informed consent for publication was provided by the participants. couple histone and DNA methylation to TNFα transcription silencing during endotoxin tolerance. J Biol Chem. 2008;283(47):32198–208. Competing interests 17. Novakovic B, Habibi E, Wang SY, Arts RJW, Davar R, Megchelenbrink W, et al. The authors declare that they have no competing interests. β-Glucan reverses the epigenetic state of LPS-induced immunological tolerance. Cell. 2016;167:1354–1368.e14. Author details 18. Álvarez-Errico D, Vento-Tormo R, Sieweke M, Ballestar E. Epigenetic control 1 Epigenetics and Immune Disease Group, Josep Carreras Research Institute of myeloid cell differentiation, identity and function. Nat Rev Immunol. (IJC), 08916 Barcelona, Spain. 2Chromatin and Disease Group, Cancer 2015;15(1):7–17. Epigenetics and Biology Programme (PEBC), Bellvitge Biomedical Research 19. Garcia-Gomez A, Li T, Kerick M, Català-Moll F, Comet NR, Rodríguez-Ubreva Institute (IDIBELL), L’Hospitalet de Llobregat, 08908 Barcelona, Spain. 3Innate J, et al. TET2- and TDG-mediated changes are required for the acquisition of Immunity Group, IdiPAZ, La Paz University Hospital, 28046 Madrid, Spain. distinct histone modifications in divergent terminal differentiation of 4 Emergency Department, IdiPAZ, La Paz University Hospital, 28046 Madrid, myeloid cells. Nucleic Acids Res. 2017;45:10002–17. Spain. 5Immunology Division, Vall d’Hebron University Hospital and 20. Vento-Tormo R, Company C, Rodríguez-Ubreva J, de la Rica L, Urquiza JM, Diagnostic Immunology Research Group Vall d’Hebron Research Institute Javierre BM, et al. IL-4 orchestrates STAT6-mediated DNA demethylation (VHIR), 08035 Barcelona, Spain. 6Intensive Care Department, Vall d’Hebron leading to dendritic cell differentiation. Genome Biol. 2016;17:4. University Hospital, Shock, Organ Dysfunction and Resuscitation (SODIR) 21. Zhang Q, Zhao K, Shen Q, Han Y, Gu Y, Li X, et al. Tet2 is required to resolve Research Group, Vall d’ Hebron Research Institute (VHIR), Universitat inflammation by recruiting Hdac2 to specifically repress IL-6. Nature. 2015; Autònoma de Barcelona, 08035 Barcelona, Spain. 7Cardiac Post-Surgery Unit 525(7569):389–93. (UPCC), Vall d’Hebron University Hospital, 08035 Barcelona, Spain. 8Tumor 22. Shen Q, Zhang Q, Shi Y, Shi Q, Jiang Y, Gu Y, et al. Tet2 promotes pathogen Immunology Lab, IdiPAZ, La Paz University Hospital, 28046 Madrid, Spain. infection-induced myelopoiesis through mRNA oxidation. Nature. 2018;554: 9 Center for Biomedical Research Network, CIBEres, Madrid, Spain. 123–7. 23. Teschendorff AE, Gao Y, Jones A, Ruebner M, Beckmann MW, Wachter DL, Received: 18 July 2019 Accepted: 7 October 2019 et al. DNA methylation outliers in normal breast tissue identify field defects that are enriched in cancer. Nat Commun. 2016;7:10478. 24. McLean CY, Bristor D, Hiller M, Clarke SL, Schaar BT, Lowe CB, et al. GREAT References improves functional interpretation of cis-regulatory regions. Nat Biotechnol. 1. Singer M, Deutschman CS, Seymour C, Shankar-Hari M, Annane D, Bauer M, 2010;28:495–501. et al. The third international consensus definitions for sepsis and septic 25. Heinz S, Benner C, Spann N, Bertolino E, Lin YC, Laslo P, et al. Simple shock (sepsis-3). JAMA. 2016315(8):801-10 combinations of lineage-determining transcription factors prime cis- 2. Derek C. Angus, M.D., M.P.H., and Tom van der Poll, M.D. PD. Severe sepsis regulatory elements required for macrophage and B cell identities. Mol Cell. and septic shock. New Engl J Med 2013;369:840–851. 2010;38:576–89. 3. Van Der Poll T, Van De Veerdonk FL, Scicluna BP, Netea MG. The 26. Ernst J, Kellis M. ChromHMM: automating chromatin-state discovery and immunopathology of sepsis and potential therapeutic targets. Nat Rev characterization. Nat Methods. 2012;9(3):215–6. Immunol. 2017;17(7):407–20. 27. de la Rica L, Rodríguez-Ubreva J, García M, ABMMK I, Urquiza JM, Hernando 4. Quartin AA. Magnitude and duration of the effect of sepsis on survival. H, et al. PU.1 target genes undergo Tet2-coupled demethylation and Department of Veterans Affairs Systemic Sepsis Cooperative Studies Group. DNMT3b-mediated methylation in monocyte-to-osteoclast differentiation. JAMA. 1997;277(13):1058–63. Genome Biol. 2013;14(9):R99. 5. Cavaillon JM, Adib-Conquy M. Bench-to-bedside review: endotoxin 28. Pedchenko TV, Park GY, Joo M, Blackwell TS, Christman JW. Inducible binding tolerance as a model of leukocyte reprogramming in sepsis. Crit Care. 2006; of PU.1 and interacting proteins to the Toll-like receptor 4 promoter during 10(5):233. endotoxemia. Am J Physiol Lung Cell Mol Physiol. 2005;289:L429–37. 6. Shalova IN, Lim JY, Chittezhath M, Zinkernagel AS, Beasley F, Hernández- 29. Cai B, Cai J, Luo Y, Chen C, Zhang S. The specific roles of JAK/STAT signaling Jiménez E, et al. Human monocytes undergo functional re-programming pathway in sepsis. Inflammation. 2015;38:1599–608. during sepsis mediated by hypoxia-inducible factor-1α. Immunity. 2015; 30. Porta C, Rimoldi M, Raes G, Brys L, Ghezzi P, Di Liberto D, et al. Tolerance 42(3):484–98. and M2 (alternative) macrophage polarization are related processes 7. Cheng SC, Scicluna BP, RJW A, Gresnigt MS, Lachmandas E, Giamarellos- orchestrated by p50 nuclear factor kappaB. Proc Natl Acad Sci U S A. 2009; Bourboulis EJ, et al. Broad defects in the energy metabolism of leukocytes 106:14978–83. underlie immunoparalysis in sepsis. Nat Immunol. 2016;17(4):406–13. 31. Pena OM, Pistolic J, Raj D, Fjell CD, Hancock REW. Endotoxin tolerance 8. Gordon S, Taylor PR. Monocyte and macrophage heterogeneity. Nat Rev represents a distinctive state of alternative polarization (M2) in human Immunol. 2005;5(12):953–64. mononuclear cells. J Immunol. 2011;186:7243–54. 9. Biswas SK, Lopez-Collazo E. Endotoxin tolerance: new mechanisms, 32. Sfeir T, Saha DC, Astiz M, Rackow EC. Role of interleukin-10 in monocyte molecules and clinical significance. Trends Immunol. 2009;30(10):475–87. hyporesponsiveness associated with septic shock. Crit Care Med. 2001;29: 10. Avendaño-Ortiz J, Maroun-Eid C, Martín-Quirós A, Toledano V, Cubillos- 129–33. Zapata C, Gómez-Campelo P, et al. PD-L1 overexpression during endotoxin 33. Wang H, Maurano MT, Qu H, Varley KE, Gertz J, Pauli F, et al. Widespread tolerance impairs the adaptive immune response in septic patients via plasticity in CTCF occupancy linked to DNA methylation. Genome Res. 2012; HIF1α. J Infect Dis. 2018;217:393–404. 22:1680–8. 11. Foster SL, Hargreaves DC, Medzhitov R. Gene-specific control of 34. Houschyar KS, Chelliah MP, Rein S, Maan ZN, Weissenberg K, Duscher D, inflammation by TLR-induced chromatin modifications. Nature. 2007; et al. Role of Wnt signaling during inflammation and sepsis: a review of the 447(7147):972–8. literature. Int J Artif Organs. 2018;41:247–53. 12. Saeed S, Quintin J, Kerstens HHD, Rao NA, Aghajanirefah A, Matarese F, et al. 35. Villaseñor T, Madrid-Paulino E, Maldonado-Bravo R, Urbán-Aragón A, Pérez- Epigenetic programming during monocyte to macrophage differentiation Martínez L, Pedraza-Alva G. Activation of the Wnt pathway by and trained innate immunity. Science. 2014;345(6204):1251086. Mycobacterium tuberculosis: a Wnt-Wnt situation. Front Immunol. 2017;8:50. 13. Perkins DJ, Qureshi N, Vogel SN. A toll-like receptor-responsive kinase, 36. Dallosso AR, Hancock AL, Szemes M, Moorwood K, Chilukamarri L, Tsai H-H, protein kinase R, is inactivated in endotoxin tolerance through differential et al. Frequent long-range epigenetic silencing of protocadherin gene K63/K48 ubiquitination. MBio. 2010;1(5):e00239–10. clusters on chromosome 5q31 in Wilms’ tumor. van Heyningen V, editor. 14. Chan C, Li L, McCall CE, Yoza BK. Endotoxin tolerance disrupts chromatin PLoS Genet. 2009;5:e1000745. remodeling and NF-B transactivation at the IL-1 promoter. J Immunol. 2005; 37. Peek SL, Mah KM, Weiner JA. Regulation of neural circuit formation by 175:461–8. protocadherins. Cell Mol Life Sci. 2017;74:4133–57. Lorente-Sorolla et al. Genome Medicine (2019) 11:66 Page 18 of 18 38. Mah KM, Weiner JA. Regulation of Wnt signaling by protocadherins. Semin Cell Dev Biol. 2017;69:158–71. 39. Döcke WD, Randow F, Syrbe U, Krausch D, Asadullah K, Reinke P, et al. Monocyte deactivation in septic patients: restoration by IFN-gamma treatment. Nat Med. 1997;3:678–81. 40. Hussain S, Zwilling BS, Lafuse WP. Mycobacterium avium infection of mouse macrophages inhibits IFN-gamma Janus kinase-STAT signaling and gene induction by down-regulation of the IFN-gamma receptor. J Immunol. 1999; 163:2041–8. 41. DeMasellis G, Nahar M, Pappy A, Goodrich J, Cornell T, Shanley T, et al. 819. Crit Care Med. 2013;41:A204. 42. Munoz C, Carlet J, Fitting C, Misset B, Blériot JP, Cavaillon JM. Dysregulation of in vitro cytokine production by monocytes during sepsis. J Clin Invest. 1991;88(5):1747–54. 43. Mitroulis I, Ruppova K, Wang B, Chen LS, Grzybek M, Grinenko T, et al. Modulation of myelopoiesis progenitors is an integral component of trained immunity. Cell. 2018;172(1-2):147–161.e12. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.