Making the Case for Chromatin Profiling: a New Tool to Investigate the Immune-Regulatory Landscape

PERSPECTIVES

other to modulate gene expression over OPINION space and time. Indeed, chromatin profiling offers enormous potential to increase our Making the case for chromatin knowledge of the regulatory landscape in the immune system in both health and disease. profiling: a new tool to investigate Haematopoietic cells were among the first cell types for which the genome-wide organization and structure of chromatin were the immune-regulatory landscape profiled in humans, and these pioneering studies established the basic principles of epi- Deborah R. Winter, Steffen Jung and Ido Amit genomics that have since defined the field6–8. Abstract | Recent technological advances have enabled researchers to accurately In turn, chromatin profiling has provided insights into several key areas that are of the and efficiently assay the chromatin dynamics of scarce cell populations. In this foremost interest in immunology research Opinion article, we advocate the application of these technologies to central by addressing how crucial immune genes are questions in immunology. Unlike changes to other molecular structures in the cell, controlled in a cell-type-specific and tissue- chromatin features can reveal the past (developmental history), present (current specific manner in both health and disease. activity) and future (potential response to challenges) of a given immune cell type; For example, studies of resting and lipopoly- chromatin profiling is therefore an important new tool for studying the saccharide (LPS)-stimulated in vitro-cultured macrophages have shown that the majority of immune-regulatory networks of health and disease. enhancers regulating innate immune response genes are pre-bound by the transcription fac- All haematopoietic cells in the human cell type identity. Analysing RNA expression tor PU.1 and are, consequently, marked by a immune system have identical genomes levels provides only a partial view of the cell’s poised chromatin state before stimulation. By and, yet, each cell type has a unique role in terminal identity without giving a thorough contrast, a few latent enhancers are activated keeping the body healthy. Since the comple- indication of the underlying mechanisms. To de novo upon stimulation of the cell but retain tion of the Human Genome Project in 2001 learn more about the purpose of the genome a poised state after the stimulus is removed, (REFS 1,2), it has become clear that there is beyond the gene sequence, we turn to what thereby functioning as a form of cell memory considerably more to understanding the lies literally ‘over the genome’, the epigenome. in anticipation of future restimulation9,10. By human blueprint than simply sequencing In particular, this includes the structure and examining the chromatin state of genetic vari- the genome. This realization spawned many organization of chromatin within the cell’s ants at the BCL11A (B cell lymphoma 11A) global studies that aim to uncover the under- nucleus (BOX 1). Each cell type has its own locus in humans, an intronic enhancer lying regulatory networks, which include unique chromatin state, which reflects the required for erythroid-cell-specific expression both cis-acting elements encoded in the DNA regulatory network that defines the specific was found to influence fetal haemoglobin sequence and trans-acting binding factors that identity and function of that cell. The loca- levels11. In a follow‑up experiment using a together coordinate cell-type specification. tion and function of cis-acting regulatory mouse model, the authors confirmed that The past decade has seen massive progress in elements can be revealed through chromatin mutating this enhancer de‑repressed embry- the field of genomics, and a new approach to profiling assays. In fact, chromatin state is onic haemoglobin expression in erythroid understanding gene regulation has emerged: an important regulatory determinant for cells without influencing other cell types, namely, chromatin profiling. lineage specification, cellular response to suggesting a possible treatment for sickle cell In 2012, the Immunological Genome environmental stimuli and defining cell roles anaemia. Similarly, cross-referencing the Project (ImmGen) completed its first phase within the immune system. Whereas mRNA genome-wide association study (GWAS) data- by profiling the transcriptomes of 249 expression profiling provides a snapshot of base of disease-causing variants with genome- immune cell types3–5. ImmGen has provided the current state of a cell, understanding the wide chromatin profiles in specific immune the immunology community with a valuable epigenomic regulation can give perspective cells has highlighted the role of IKZF3 — resource, but the collection of transcribed on how this state was reached and how which encodes an IKAROS family transcription RNAs in a given cell type reveals only one cells might advance. With the common factor involved in lymphocyte differentiation — part of the larger regulatory network. With use of high-throughput sequencing, many in multiple sclerosis12. These and many only 1–2% of the human genome spanning chromatin assays have now been adapted other studies demonstrate the potential of protein-coding genes1,2, the remaining DNA for genome-wide application. Thus, one can chromatin profiling in guiding the develop- landscape incorporates an abundance of determine the specific catalogue of regula- ment of new therapies and extending our cis‑acting regulatory elements that contribute tory elements within a particular cell type basic understanding of how haematopoiesis to the establishment and maintenance of that develop, react and interact with each and immunity are encoded in our genomes.

NATURE REVIEWS | IMMUNOLOGY VOLUME 15 | SEPTEMBER 2015 | 585

In this Opinion article, we describe, with in everyday immunological research. We elements in the genome that are elucidated examples, how chromatin profiling can be discuss the approaches that have developed by chromatin profiling can also be used to used to make novel discoveries and reinforce from early chromatin studies in mouse cells infer the activity and binding of transcription classical immunological studies. We high- and cell culture models to gain insight into factors. We propose a path for translating light the recent advances in experimental the past, present, future and environmental chromatin studies into humans, taking into technologies that have made chromatin response of immune cell types. In many consideration the issue of genetic variation, profiling more practical and efficient for use cases, the global catalogue of regulatory which constitutes an added hurdle for clinical applications but also has the potential of unique rewards for direct translation to Box 1 | Chromatin disease therapy. We conclude with our vision of the future of immunological research, The DNA in the nucleus of all eukaryotic cells is organized into chromatin (see the figure). The basic including the regular use of chromatin pro- unit of chromatin is the nucleosome, which consists of approximately 147 base pairs of DNA filing and public epigenomic databases in wrapped approximately 1.7 times around a histone protein complex69. The histone complex consists studies of the immune-regulatory landscape. of two copies each of H2A, H2B, H3 and H4. Nucleosomes may change position, which alters the accessibility of the underlying DNA, or they can be post-translationally modified by proteins known We direct readers to the specific references as chromatin remodellers70. These histone modifications are designated according to their location throughout the text for exact methodological in the histone tail and the type of modification (such as methylation, acetylation, phosphorylation protocols and analysis, as well as to details or ubiquitylation)71: for example, H3K4me2 indicates that the H3 tail is dimethylated on the fourth of the current endeavours by consortia to lysine. Regions of the genome that are nucleosome depleted, known as ‘open chromatin’, are systematically map the epigenome (see enriched for functional regulatory elements, are accessible to transcription factor binding and Further information); our aim here is to typically coincide with active histone modifications17,72. Promoters, which are marked by H3K4me3, give the reader a meaningful overview and tend to be constitutive (found in all cell types); by contrast, enhancers, which are marked by impression of the potential of this approach. H3K4me1, form a more cell-type-specific regulatory landscape. Tightly bound chromatin, known as Although we cannot cover all of the note heterochromatin, often contains stably repressed, inaccessible genomic elements and is situated worthy studies that have profiled chromatin closer to the nuclear perimeter or lamina73,74. On a higher level, chromatin is arranged within the nucleus to form a 3D architecture through a process that is mediated by structural proteins. For in immune cells, we hope to convey the exten- example, two copies of CTCF bound at adjacent chromatin boundaries will come into contact sive benefits of this approach, the challenges with each other, as well as cohesin and other mediators . In this way, interacting elements (such as and the exciting possibilities ahead. enhancers) that are separated by a genomic distance can be brought into physical contact within DNA loops. Studies have shown that similarly regulated regions form large compartments including Advances in chromatin technology topologically associated domains (TADs) of active regulatory elements and tracts silenced by the Profiling the complete chromatin landscape polycomb-repressive complex (PRC)75,76. Each cell type has a distinct global chromatin state that of all cells in the haematopoietic system is a reflects the activity of its regulatory landscape. formidable task. Within the main cell lineages, Active regulatory loop many cell subtypes have been defined and Promoter each one of these must be individually isolated Histone and analysed. Taking into account external modiﬁcation influences and the sensitivity of cells to the Histone tail Mediators local environment, distinct cell populations — Lamina-associated Histone heterochromatin complex such as those in different tissues — should be analysed separately13,14. Furthermore, primary cells should be analysed immediately after Enhancer DNA DNA Nucleosome methylation isolation, as growth in culture following removal from their natural context is likely to influence the underlying chromatin regulatory Topologically landscapes. This was recently demonstrated associated domain for tissue-resident macrophages, which rapidly lost their tissue-specific chromatin signatures upon isolation and culture13. The general approach of chromatin studies is to isolate and enrich for genome fragments that are associated with the chromatin feature CTCF of interest (for example, nucleosomes, open chromatin, DNA methylation or histone modifications) to create libraries for sequencing (FIG. 1). The challenge of traditional genome- Nucleus wide studies is that due to the low efficiency PRC H3K27me3 of certain stages of the protocol, a large start- OFF ing population of cells is needed to produce the large number of unique sequences that are required to adequately sample all regulatory elements. In this way, the entire epigenomic Polycomb-repressed chromatin network can be constructed de novo without Nature Reviews | Immunology

586 | SEPTEMBER 2015 | VOLUME 15 www.nature.com/reviews/immunol

a Experimental protocol Cell sorting Chromatin protocol Speciﬁc antibody Histone modiﬁcation

ChIP

FACS machine laser DNA break Library of DNA fragments Macrophage ATAC

Tn5 Sequencing Open Open chromatin Sequencing adaptor chromatin and alignment Bisulﬁte sequencing CG CG CG Methylation UG Waste of DNA b Data interpretation Processing and normalization

Cell-type 1-speciﬁc enhancer Peak calling to identify AGGAAG genomic regions Transcription factor- enriched for a particular binding motif chromatin feature Reconstructed model of chromatin state and Cell-type 1 binding motifs

Diﬀerential regions of enrichment between cell types Cell-type 2

Shared Diﬀerential promoter enhancer Figure 1 | Representative chromatin profiling: from immune cells to the DNA methylation by converting unmethylated cytosines to uridines that regulatory landscape. a | Experimental protocol. In this example, the cell will be recognized during sequencing19. The fragments in these libraries type of interest (macrophages) is isolated from other cells by fluorescence- are then sequenced and can be aligned to theNature reference Reviews genome | Immunology using one activated cell sorting (FACS). Next, the chosen protocol for chromatin pro- of several open-access algorithms. b | Data interpretation. The next steps filing is carried out depending on the type of data desired. The purpose of are to process and normalize the aligned sequences so that the results can each protocol is to create a sequencing library enriched for genome be visualized (for example, in a genome browser). Bioinformatic analysis is sequences that associate with the particular chromatin feature of interest. carried out to interpret the data on a global level and to reconstruct the Shown here are chromatin immunoprecipitation (ChIP), assay for trans- original chromatin state including: identifying genomic regions that are posase-accessible chromatin (ATAC) and bisulfite sequencing. The ChIP enriched for a particular chromatin feature (peak calling); characterizing assay uses antibodies to locate a specific histone modification followed by, differential regions of enrichment between cell types; annotating the regu- typically, sonication to create sequencing fragments6,67,77. ATAC identifies latory elements in these regions; finding transcription factor-binding open chromatin regions through the binding of Tn5 transposase and motifs; and reconstructing the regulatory network. Dashed horizontal lines insertion of sequencing adaptors16. Bisulfite sequencing recognizes represent the threshold for peak calling.

prior knowledge of the important genes, and obtaining sample sizes sufficient for classical Fortunately, several common chromatin global patterns and key regulatory factors in analytical approaches has been exceedingly technologies have recently been successfully the cell can be identified. For most immune difficult. Therefore, until recently, it was optimized to considerably increase both effi- cell types in vivo (particularly precursors not possible to carry out high-throughput ciency and sensitivity15,16. This was achieved and rare subtypes, and for patient samples), chromatin assays on these cells. by improving the recovery rate for the

NATURE REVIEWS | IMMUNOLOGY VOLUME 15 | SEPTEMBER 2015 | 587

sequences of interest, as well as by decreasing the advances, much progress has been made in Inferring a spatiotemporal cell view inherent ‘noise’ of irrelevant genomic regions. data interpretation, and better algorithms Beyond the present activity of the cell, chro- Specifically, the iChIP approach — which for alignment, normalizing, visualizing, matin profiling can provide ‘behind-the- enhances the traditional high-throughput peak calling, motif finding and modelling scenes’ information about how a cell reached chromatin immunoprecipitation (ChIP) assay the regulatory network are now available. its current state and about its ability to for profiling histone modifications — uses a The community, in general, has raised the adapt to changing circumstances. Different double round of antibody enrichment (one threshold for the quality of both the raw chromatin features inform as to the cell’s general and one specific), as well as pooling data and analysis involved. A comprehensive developmental history, current regulatory of multiple samples, to increase signal and map of the chromatin and regulatory network and potential response to challenges, reproducibility. To replace assays for open network dynamics for the mammalian as well as to the impact of the local tissue chromatin that require an input of millions of immune system is now within reach. environment. cells, such as the DNase hypersensitivity assay17, In the next section, we describe various the assay for transposase-accessible chromatin epigenomic approaches and the information The past: developmental history. The past (ATAC) uses the Tn5 transposase to effi- that they provide about the cell’s regulatory history of a cell consists of numerous small- ciently fragment and barcode the genome16,18. network. The principles of these approaches scale decisions to turn genes on and off in a Both iCHIP and ATAC enable practical have been used to probe models for processes manner that will in combination lead to its in vivo genome-wide chromatin profiling as diverse as ageing and immune memory; current state. In haematopoiesis, these deci- of less than a thousand sorted immune cells such pioneering studies provide a glimpse of sions affect the maintenance of multipotency directly19. In parallel with these experimental the enormous potential of chromatin profiling. in haematopoietic stem cells (HSCs) and their differentiation into haematopoietic effector cells of the erythroid, myeloid or Glossary lymphoid lineage. Developmental decisions Assay for transposase-accessible chromatin Expression quantitative trait loci leave a lasting imprint on the chromatin state (ATAC). A method for identifying regions of open chromatin (eQTLs). Genomic loci that correlate with changes in that subsists over time. DNA methylation, by using Tn5 transposase to insert paired sequencing gene expression analogously with the relationship a crucial element of epigenomic regulation adaptors into accessible chromatin. between quantitative trait loci and phenotype. throughout development, is associated with 20 Chromatin Formaldehyde-assisted isolation of regulatory gene repression . For example, in HSCs, The 3D complex of DNA and proteins within the nucleus. elements methylation must be actively maintained Features of chromatin (including localization, structure, (FAIRE). A method for identifying regions of open to uphold the multipotent state by silencing interacting proteins, accessibility and modifications) chromatin by isolating DNA fragments that are not bound the expression of cell fate genes that regulate cell-type-specific gene expression. to DNA after crosslinking. would otherwise reduce proliferation21,22. Chromatin immunoprecipitation Genome-wide association study Interestingly, the inhibition of maintenance (ChIP). A method for identifying genomic regions that (GWAS). A statistical analysis comparing the occurrence methylation (through mutation of DNA associate with a particular protein, including specifically of multiple common genetic variants (usually methyltransferase 1 (DNMT1)) led to modified histones, through immunoprecipitation of the single-nucleotide polymorphisms) with a certain reduced lymphoid lineage commitment but crosslinked DNA fragments. outcome (a phenotype or disease) to identify causal had no effect on myeloid commitment, sug- variants. Chromatin interaction analysis using paired-end gesting that methylation is unnecessary for tag sequencing Hi‑C and 5C the latter21. Mutations in the gene encoding (ChIA-PET). A method for identifying pairs of regions High-throughput adaptations of the chromosome DNMT3A, the methyltransferase responsi- associated with a particular protein by combining conformation capture (3C) method that isolates DNA ble for de novo methylation, were found to chromatin immunoprecipitation with the isolation of fragments interacting with each other. 5C searches for interacting DNA fragments. interactions between many loci, whereas Hi‑C searches be common in HSCs of patients with acute for all possible interactions in the genome. myeloid leukaemia (AML); the mutated Cis-acting elements HSCs survived chemotherapy and even had Functional regulatory elements (such as enhancers) within Pioneer transcription factors a repopulation advantage over non-mutated the genome that regulate the transcription of genes. Sequence-specific DNA-binding proteins that target HSCs23. During the process of differentia- closed chromatin sites and recruit chromatin DNA methylation remodellers to transform these sites into an open tion from HSCs into mature immune cells, The addition of a methyl group to a nucleic acid in the chromatin state enabling the binding of additional DNA methylation functions in combination genome, typically to cytosine within a CpG pair. transcription factors. with histone modifications to generate the catalogue of functional and repressed regu- DNase hypersensitivity assay Promoter A method that relies on the preference of the enzyme A proximal regulatory element that is typically located latory regions that are necessary for each 24 DNaseI to digest unbound DNA to identify regions of upstream of the particular gene it regulates and includes stage of differentiation . A study measuring open chromatin. the binding sites of general transcription factors such as genome-wide methylation levels during RNA polymerase II. haematopoiesis identified genes with previ- Enhancers ously unknown roles in the decision between Distal regulatory elements that may function in Single-nucleotide polymorphisms combination with promoters or other enhancers to (SNPs). Mutations (substitutions, insertions or deletions) of myeloid and lymphoid lineage commit- influence the transcription of one or more genes through an individual nucleotide in the genome sequence that are ment — including Arl4c (ADP-ribosylation the binding of transcription factors. common in the population. factor-like 4C) and Jdp2 (JUN dimerization protein 2)22. Haematopoietic factors alter the Epigenome Trans-acting binding factors Modifications to the genome that do not change the Non-DNA molecules (typically referring to transcription methylation state as the cell matures so that DNA sequence, including DNA methylation, histone factors or chromatin factors) that regulate the transcription lineage-specific genes are activated and alter- modifications and rearrangements of chromatin structure. of genes. native fates are silenced20. As a result, each

588 | SEPTEMBER 2015 | VOLUME 15 www.nature.com/reviews/immunol

precursor intermediate in the differentiation with the most complex combinations of Alternatively, distal regulatory elements process involves a set of methylation and regulatory elements: the fact that each cell marked by H3K4me1 in the absence of demethylation events that ultimately leads type has many more enhancers than genes H3K27ac (poised enhancers) can have a to the mature cell type22. A study comparing suggested that different immune cell line- similar role in forecasting the activation HSCs from aged and young mice found that ages may use alternate enhancers to activate of lineage-specific genes, and they provide widespread shifts in chromatin state — such the same gene17. In fact, a study using chro- intriguing clues about the complex regula- as decreased methylation of binding sites for matin interaction analysis using paired-end tag tory dynamics of haematopoiesis28,34 (FIG. 2b). transcription factors associated with HSC sequencing (ChIA-PET) technology showed The ratio of poised to active enhancers in multipotency (for example, stem cell protein that many mouse genes, including Myc, a cell gives a measure of its differentiation (SCL; also known as TAL1), LIM domain- physically interact with an entirely different potential, with precursors tending to have binding protein 1 (LDB1) and runt-related set of enhancers in B cells than in embry- more poised enhancers than their succes- transcription factor 1 (RUNX1)) — led onic stem cells31. Moreover, comparison of sors. Comparing cell types throughout the to the reduced differentiation of HSCs the cis-acting regulatory elements in B cells mouse haematopoietic system indicated seen in ageing25. In the absence of further versus macrophages identified thousands of that ‘poising’ is a widespread phenomenon, interference, the methylation status of a differential regions bound by PU.1 that are with 32% of the enhancers that are active cell is preserved through division and thus likely to regulate overlapping sets of genes10. in differentiated cells having been initially provides a link to the cell’s past. These regulatory intricacies are examples of poised in lineage progenitors15. Such how chromatin profiling can reveal more cases of precise regulation can be crucial The present: current regulatory network. of the cell’s active regulatory state than can for health: S100a8 —which is marked by Chromatin profiling can also expand our be discerned from RNA profiling alone. H3K4me1 in common myeloid progenitors understanding of a cell’s present regulatory (CMPs) before the addition of H3K27ac state. In general, genome-wide measures of The future: response to challenge. The future and its expression in neutrophils and other chromatin accessibility — such as DNase activity of a cell, in terms of both develop- terminal myeloid cells15 — has an impor- hypersensitivity assays, formaldehyde-assisted mental pathways and the response repertoire tant role in the immune response, and its isolation of regulatory elements (FAIRE) and to stimuli, can also be viewed through chro- overexpression leads to chronic inflam- ATAC — which can be combined with matin profiling. Many regulatory elements mation such as that seen in patients with high-throughput sequencing (DNASE-seq, exhibit marks of potential, rather than cur- rheumatoid arthritis35. The enhancer near

FAIRE–seq and ATAC-seq), are useful rent, activity. Some of these regions, termed Cx3cr1 (CX3C‑chemokine receptor 1), in identifying which regulatory elements ‘bivalent domains’, are distinguished by a which encodes a chemokine receptor that are accessible at a particular time16,17,26. combination of active (H3K4 trimethylation is expressed during the development of all Specific histone modifications, as assayed (H3K4me3)) and repressive (H3K27me3) macrophages, is poised in CMPs but then by ChIP–seq, have been useful in classifying histone modifications, which are mostly reactivated only in intestinal macrophages diverse regulatory elements6,27. Acetylation, found in promoter regions6,32,33. Early on, and microglia14,15, and it is possibly actively particularly histone 3 lysine 27 acetylation bivalent domains were identified in embry- repressed by REV-ERB (also known as (H3K27ac), is further indicative of active onic stem cells near genes encoding tran- NR1D2) in most other tissue-resident enhancer elements and is closely connected scription factors (including SOX, forkhead macrophages36. Within the B cell lineage, with the expression of their associated box (FOX), Iroquois homeobox (IRX), POU poising of enhancers in pro‑B cells indicates genes28. Indeed, the activity of regulatory and paired box (PAX) family members) the future potential of genes such as Egr3 elements together with their proximity and in CD4+ T cells (such as the intergenic (early growth response 3) and Cd40 to be to transcription start sites may be used to region between RAB6IP2 and FBXL14)6,33. expressed in late B cell stages37. Moreover, match them to the particular genes that they Bivalent domains are most often observed poised enhancers persisting in mature regulate. A study from the Encyclopedia early in development, foretelling the role immune cells may indicate how they will of DNA Elements (ENCODE) consortium of the respective factors later in differen- respond to environmental stimuli. In macro suggested that the genes that are regulated tiation. Depending on the cell’s eventual phages and dendritic cells, previously poised by distal enhancers could be determined by differentiation fate, these regions may lose enhancers recruit signal-dependent tran- correlating the pattern of chromatin acces- their active or repressive marks to become scription factors of the nuclear factor‑κB sibility at an enhancer with that at the gene silenced or activated, respectively. For exam- (NF‑κB) and activator protein 1 (AP‑1) fam- promoter17. However, this approach assumes ple, when comparing CD133+ multipotent ilies upon LPS treatment and gain activating direct regulation in a linear genome. To take cells with differentiated CD36+ erythrocyte histone marks associated with the expression into account combinatorial interactions and precursors, about a quarter of the bivalent of immediate early genes9,38. Poised enhancers, the compacted structure of the genome, an domains identified in CD133+ cells remain together with de novo-activated or latent alternative approach assays the 3D archi- in this state in CD36+ cells, whereas half of enhancers, may retain activating marks after tecture within the nucleus by chromatin the bivalent domains lose only H3K4me3 the stimulation is removed in preparation for capture experiments, such as Hi‑C and 5C, active marks and most of the remaining future challenges through a process termed to identify physical interactions between regions lose H3K27me3 repressive marks as ‘trained immunity’ (REFS 9,39–41). For exam- genomic regions29,30. These approaches led well as gaining additional activating histone ple, in bone marrow culture-derived macro to the conclusion that not only can a single modifications as they differentiate32. The phages, the chromatin state was shown to enhancer regulate multiple genes but also combination of these active and repressive function as a memory of previous immune a single gene can be regulated by multiple histone marks can establish a poised state at challenges, enabling genes that continued to enhancers (FIG. 2a). Genes encoding proteins genes that define the possible lineages of the be inducible after Toll-like receptor (TLR) with immune functions were among those cell’s progeny. stimulation (non-tolerizable), including

NATURE REVIEWS | IMMUNOLOGY VOLUME 15 | SEPTEMBER 2015 | 589

a Cell-type-speciﬁc enhancers

H3K4me1 H3K27ac Transcription RNA polymerase II PU.1 EBF1 mRNA H3K4me3

B cell B cell enhancer Inactive region Promoter (shared)

MAF

Macrophage enhancer Macrophage b Poised enhancers in diﬀerentiation H3K4me1 No transcription

Macrophage GMP enhancer (poised)

H3K27ac H3K4me1 H3K4me3 Transcription Diﬀerentiation Macrophage Macrophage enhancer (active)

c Eﬀect of the tissue environment Synapse pruning Transcription Brain environment

Neuron Sall1 or Siglech Microglial cell MEF2C

Erythrocyte recycling

Liver environment

Clec4f Macrophage Erythrocyte Kupﬀer cell LXRα

Surfactant clearance Lung Lung macrophage environment

PPARγ Car4 or Chil3 Lung epithelial cell

Figure 2 | Multiple layers of chromatin regulation in immune cells. transcription factors, such as MAF. c | Effect of the tissue environment. a | Cell-type-specific enhancers. Alternative distal regulatory regions The tissue environment affects the regulatoryNature Reviews landscape | Immunology of a cell (enhancers) involved in lineage specification promote the transcription through the induction of specific transcription factors, leading to the of the same gene in different cell types. The cell-type-specific transcrip- expression of genes that are likely to be involved in the unique function factors of B cells (blue) and macrophages (orange) bind in combina- tional pathways of each tissue-specific cell type: Sall1 and Siglech (sialic tion with PU.1 to active enhancers marked by histone 3 lysine 4 acid-binding immunoglobulin-like lectin H) for neuronal synapse prun- monomethylation (H3K4me1; green) and H3K27 acetylation (H3K27ac; ing in microglia (brain-resident macrophages); Clec4f (C-type lectin red). Active promoters are indicated by H3K4me3 (purple) and RNA poly- domain family 4 member F) for erythrocyte recycling in Kupffer cells merase II binding. b | Poised enhancers in differentiation. Poised enhancers (liver-resident macrophages); and Car4 (carbonic anhydrase 4) and marked by H3K4me1 alone in the granulocyte–macrophage precursor Chil3 for surfactant clearance in lung macrophages 14. LXRα, liver X (GMP) become active upon cell differentiation to induce transcription in receptor-α; MEF2C, monocyte enhancer factor 2C; PPARγ, peroxisome macrophages through the co‑binding of macrophage-specific proliferator-activated receptor-γ.

590 | SEPTEMBER 2015 | VOLUME 15 www.nature.com/reviews/immunol

Fpr1 (formyl peptide receptor 1), Cnlp (also In this section on using chromatin pro- haematopoietic system include GATA1 for known as Camp) and Saa3 (serum amyloid filing to infer a spatiotemporal cell view, the erythroid lineage, CCAAT/enhancer- A3), to be prepared for future activation. we have described selected model systems binding protein-α (CEBPα) for the myeloid By contrast, tolerizable genes, including Il6 in order to best illustrate the diversity of lineage and FOXO1 for the lymphoid line- (interleukin‑6), Lipg (endothelial lipase) and chromatin assays and how chromatin pro- age15. At the apex of the transcription factor Cspg4 (chondroitin sulphate proteoglycan 4), filing data can be used to address various hierarchy, pioneer factors can work in had a reduced response to restimulation immunological questions — including how tandem with cell-type-specific transcription with LPS41. Thus, the chromatin landscape immune cells age, how haematopoiesis is factors to programme the cell’s chromatin offers clues about upcoming transitions in regulated, how immune memory is encoded landscape; in particular, PU.1 — the general cell state and regulatory network. and how the environment affects immune haematopoietic cell pioneer factor — can function — in a manner that could not bind with either B cell-specific transcrip- Space: the tissue environment. Recent studies be accomplished by other approaches. In tion factors (such as EBF1 (also known have shown that the chromatin landscape designing a specific research project, one as COE1)) or macrophage-specific tran- can also reflect another aspect of the cell: its must consider which assay or combination scription factors (such as MAF or IRF8) local environment13,14. Most tissue-resident of assays is required to provide the temporal to designate alternative enhancers10,14,56. macrophages are established prenatally, or spatial view of the regulatory network Likewise, cell-type-specific pioneer factors develop in tandem with the organ in which relevant to the research question. predetermine the enhancers that will recruit they reside and are maintained in the adult42. signal-dependent transcription factors to It was recently shown that the tissue environ Inferring transcription factor binding trigger a unique response to environmental ment has a marked effect on the resulting Importantly, the chromatin landscape of a and immune stimuli9,55. For example, a single function and identity of these macrophages cell, which defines the functional cis-acting enhancer regulating Il1a (interleukin‑1α) (FIG. 2c). Specifically, analysis of tissue- regulatory landscape, can also be used to expression in dendritic cells is bound by specific enhancers in these cells revealed that determine the binding of trans-acting regu- PU.1 and CEBPβ in unstimulated cells and signals from the local microenvironment latory factors. ChIP–seq assays for specific joined by RELA and signal transducer and shape the expression of unique tissue regula- transcription factors are used to identify activator of transcription 1 (STAT1) follow- tors — such as MEF2C in microglia, liver their binding sites across the genome. ing LPS stimulation38. Thus, by analysing X receptor-α (LXRα) in Kupffer cells and However, each cell type expresses hundreds the chromatin patterns and sequence of peroxisome proliferator-activated receptor-γ of transcription factors and it is not always cis-acting regulatory elements, one can (PPARγ) in peritoneal macrophages — and clear which of these are most relevant. further deduce the temporal and spatial thereby globally programme the chromatin Transcription factors are often crucial in role of trans-acting factors. landscape13,14. For example, the presence more than one cell type and are involved of retinoic acid in the peritoneum triggers in distinct tasks depending on their binding Translating to human health GATA-binding protein 6 (GATA6) binding partner: for example, interferon-regulatory Arguably, the overarching goal of immuno to widespread enhancers that regulate factor 4 (IRF4) and IRF8 collaborate with logical studies is to learn more about the the specific gene expression programme the ETS transcription factors PU.1 and SPIB aetiology of human disease. Towards this of peritoneal macrophages, including the at ETS–IRF motifs (EICEs) in myeloid and aim, we have to translate what we know transcription of Tgfb2 (transforming growth B cells, but with BATF at AP‑1–IRF motifs about methods and principles of chromatin factor‑β2), Serpinb2 (serine (or cysteine) (AICEs) in T cells47,48. Without experimen- dynamics in the mouse immune system into peptidase inhibitor clade B member 2) and tally testing each transcription factor, the the study of human specimens. Much has Alox15 (arachidonate 15‑lipoxygenase)14,43. binding of specific transcription factors already been learned from the preliminary The set of unique enhancers and the genes within cis-acting regulatory elements can studies of the Blueprint consortium, which that they regulate in each tissue are likely to be inferred from the presence of particular was established to assay chromatin in human be responsible for emerging tissue-specific sequence motifs to which they bind and the immune cells57,58. However, because of limit functions of each macrophage subpopulation; ‘footprint’ that they leave in the pattern of ations in the type and amount of cells that for example, synapse pruning by microglia, chromatin accessibility16,49–51. When compar- can be obtained from human donors, it is erythrocyte recycling by Kupffer cells and ing B cells with T cells, OCT1-, MEF2- and often difficult to retrieve all of the cell types surfactant clearance by lung macrophages44. GATA-binding motifs were found to be necessary for analysis and, with the exception Indeed, many immune cell types, including differentially enriched in the footprints of of vaccinations59,60, it is difficult to analyse regulatory T cells and natural killer cells, ATAC-seq peaks16. samples following defined in vivo perturba- are known to infiltrate the diverse tissues A special class of transcription factors tions. Thus, many human studies are carried of the body and take local residence45,46. function as pioneers that initiate chromatin out in culture, but this inherently requires Transcriptome studies suggest that these accessibility to enable the subsequent binding disturbing the natural environment of the cells may also receive a local imprint, but a of additional transcription factors52,53. cell. Accordingly, studies on primary cells comparison of chromatin state is required for These pioneer transcription factors, which in from small animal models, commonly mice a better understanding of the environmental many cases are also lineage-determining (such as those described above), are used to factors involved45,46. Notably, the observed transcription factors, have an important provide the basic principles for a working plasticity of regulatory networks even within role in lineage specification by selecting the map of the mammalian immune system. different populations of a single cell type precise catalogue of genomic regions that Although most mouse immune cell types demonstrates how chromatin profiling can form the cell’s regulatory landscape10,38,48,54,55. do have clear human counterparts, profiling determine the effect of the local environment Examples of pioneer factors that have been the human epigenome is required for a more on the gene regulatory network of a cell. inferred from chromatin profiling in the direct understanding of human disease, given

NATURE REVIEWS | IMMUNOLOGY VOLUME 15 | SEPTEMBER 2015 | 591

that the cis-acting regulatory landscape is a publicly available GWASs. However, the sequence variants to their mechanism of rapidly evolving feature61. In addition, human mechanisms through which these single- action and the genes that they influence samples have the combined advantage and nucleotide polymorphisms (SNPs) influence will help to decipher how they lead to dis- complication of being highly influenced gene expression are rarely known because ease and, thus, how to develop treatments. by genetic variation. As exemplified in an the vast majority of causal variants impli- As in the previously described case of the elegant analysis of different mouse strains62, cated in disease and selection, such as those BCL11A enhancer in erythroid cells, many the application of chromatin profiling associated with animal domestication63, are of the regions in which disease-causing across human immune cell types will enable found outside of protein-coding sequences. variants are found are specific to a certain genome-wide screening for the influence of In lymphoblastoid cell lines, studies cell type or condition, and thus their role in causal variants in cell-type-specific enhancers. found that sequence variation could lead disease may be evident only when investi- The high levels of genetic variation between to decreased chromatin accessibility and gating the relevant immune cell type in the individuals may create technical difficulties diminished RNA polymerase II‑mediated appropriate condition. A recent study com- in terms of sequence alignment and may transcription of nearby genes by disrupting pared GWAS-identified SNPs associated complicate characterization of the ‘prototype’ the binding of transcription factors, with autoimmune diseases with histone human epigenome. However, these dif- such as CTCF and NF‑κB family members, modification data in various immune cell ferences can also be manipulated through to cis-acting regulatory elements64–66. types to link enhancer variants to disease comparative analysis and in vitro assays to A high proportion of the SNPs affecting and thereby discover their underlying elucidate the precise activity of regulatory open chromatin — termed DNase hyper mechanism12. The results show a significant elements and their role in disease. sensitive quantitative trait loci (dsQTLs) — overlap between disease-causing variants With the aforementioned technological in lymphoblastoid cell lines were also and cis-acting regulatory elements and can advances in chromatin profiling, we can classified as expression quantitative trait loci be used to provide clues as to the specific now study large human cohorts to identify (eQTLs) and were associated with changes genes that are associated with disease in the regulatory regions that lead to disease to the binding of transcription factors such particular cell types. By integrating data on (FIG. 3). Currently, an immense library of as PU.1, basic leucine zipper transcription transcription factor binding sites, the pre- causal variants that affect countless diseases factor ATF-like (BATF), EBF1 and IRF4 cise disruption in the regulatory network and phenotypes exists in the results of (REF. 64). Connecting disease-causing can be inferred. In the example of IKZF3

Healthy cohort Disease cohort

Monocyte Altered monocyte

Active enhancer RNA polymerase II Transcription H3K4me1 H3K4me3 Inactive enhancer H3K27ac No transcription mRNA AP-1

SMAD3 promoter Intron SMAD3 promoter TGACTCA TGACTTA

Figure 3 | Association of human chromatin data and susceptibility healthy monocytes12. By comparing SNPs with chromatin profiles, we to immune disease. Cohort studies are designed to find sources of can determine whether they are located Naturein regulatory Reviews elements | Immunology on or genetic variation between the control and disease groups, such as the near transcription factor-binding sites in the relevant cell type. The single-nucleotide polymorphism (SNP) shown here located in an acti- resulting disruption of the chromatin state leads to altered gene tran- vator protein 1 (AP‑1)-binding motif within the SMAD3 intron (indi- scription and provides the mechanism of disease. H3K4me1, histone 3 cated by the red letter ‘T’). This SNP, which is associated with Crohn lysine 4 monomethylation; H3K27ac, H3K27 acetylation; H3K4me3, disease, disrupts the binding of AP‑1 to an enhancer that is active in H3K4 trimethylation.

592 | SEPTEMBER 2015 | VOLUME 15 www.nature.com/reviews/immunol

mentioned above, the SNP associated with histone deacetylases) to determine their Deborah R. Winter, Steffen Jung and Ido Amit are at multiple sclerosis was located in a degener- downstream effects, and clustered regularly the Department of Immunology, Weizmann Institute of Science, Rehovot 76100, Israel. ate MEF2‑binding motif within a B cell interspaced short palindromic repeats Correspondence to D.R.W. and I.A. enhancer that is also bound by RELA (an (CRISPR)-mediated activation or repres- e‑mails: [email protected]; 38,67 NF‑κB family member) and EBF1: in gen- sion of regulatory elements . With these [email protected] eral, regions bound by MEF2, NF‑κB and additional data, we will better understand D.R.W. and I.A. contributed equally to this work. EBF1 coincided with genome-wide causal the mechanism of gene regulation and 12 doi:10/1038/nri3884 SNPs for multiple sclerosis . Although how it can be influenced. Published online 14 August 2015 the actual mutation fell outside the binding Progress in charting the gene regulatory sites for RELA and EBF1, it is possible networks of immune cells is also important 1. International Human Genome Sequencing Consortium. for mutations in the binding site of one for advancing other areas of science and Initial sequencing and analysis of the human genome. Nature 409, 860–921 (2001). transcription factor to indirectly affect the medicine. The basic principles of gene 2. Venter, J. C. et al. The sequence of the human genome. binding of nearby factors. Alternatively, regulation in development, differentiation, Science 291, 1304–1351 (2001). 3. Heng, T. S., Painter, M. W. & Immunological Genome the IKZF3 SNP could function through signalling and response that can be investi Project Consortium. The Immunological Genome as yet undefined regulatory mechanisms. gated in the relatively contained immune Project: networks of gene expression in immune cells. Nat. Immunol. 9, 1091–1094 (2008). With continued research along these lines, system will be widely applicable. As we 4. Kim, C. C. & Lanier, L. L. Beyond the transcriptome: we will gain a better understanding of how identify the crucial elements involved in completion of act one of the Immunological Genome Project. Curr. Opin. Immunol. 25, 593–597 (2013). human genetic variation leads to altered maintaining health, we can better under- 5. Shay, T. & Kang, J. Immunological Genome Project and immune phenotypes and diseases. Applying stand the role of each cell population and systems immunology. Trends Immunol. 34, 602–609 (2013). chromatin analyses to clinical decisions will how their dysregulation leads to disease. 6. Barski, A. et al. High-resolution profiling of histone be complicated by the fact that each labora- We can use these data to understand how methylations in the human genome. Cell 129, 823–837 (2007). tory uses different protocols and analytical one immune cell population might replace 7. Schones, D. et al. Dynamic regulation of nucleosome practices; we suggest that the results of dif- another malfunctioning one: for example, positioning in the human genome. Cell 132, 887–898 (2008). ferent studies need to first be standardized bone marrow cells may be transferred to 8. Boyle, A. et al. High-resolution mapping and to enable a meta-analysis of results that the lungs to recapitulate the chromatin characterization of open chromatin across the genome. Cell 132, 311–322 (2008). could be used for practical applications. landscape and phenotype of lung macro 9. Ostuni, R. et al. Latent enhancers activated by We propose that any study of primary phages14,68. Again, studying human cohorts stimulation in differentiated cells. Cell 152, 157–171 (2013). haematopoietic cells should include is necessary to appreciate the level of inher- 10. Heinz, S. et al. Simple combinations of lineage- chromatin profiling with consideration ent variation. The pharmaceutical industry determining transcription factors prime cis-regulatory elements required for macrophage and B cell of natural human variation. will profit from the ability to link disease- identities. Mol. Cell 38, 576–589 (2010). causing SNPs with response to drug targets, 11. Bauer, D. E. et al. An erythroid enhancer of BCL11A subject to genetic variation determines fetal Vision for the future leading to more effective and precise treat- hemoglobin level. Science 342, 253–257 (2013). We believe that now is the time for ments. Autoimmune and inflammatory 12. Farh, K. K.‑H. et al. Genetic and epigenetic fine mapping of causal autoimmune disease variants. chromatin profiling to be integrated as diseases, such as multiple sclerosis and Nature 518, 337–343 (2015). a key feature of immunological studies. rheumatoid arthritis, are highly hetero 13. Gosselin, D. et al. Environment drives selection and function of enhancers controlling tissue-specific Laboratories across the field will benefit geneous. The clinical presentations of these macrophage identities. Cell 160, 351–352 (2014). from incorporating chromatin assays into diseases vary widely between patients, as 14. Lavin, Y. et al. Tissue-resident macrophage enhancer landscapes are shaped by the local microenvironment. their toolbox or collaborating with other does the response to current therapies. Cell 159, 1312–1326 (2014). researchers who already have. We have Drugs developed for rheumatoid arthritis 15. Lara-Astiaso, D. et al. Chromatin state dynamics during blood formation. Science 345, 943–949 (2014). demonstrated the possible advantages of and multiple sclerosis target different com- 16. Buenrostro, J., Giresi, P., Zaba, L., Chang, H. & chromatin profiling to individual studies, ponents of the immune system and include Greenleaf, W. Transposition of native chromatin for fast and sensitive epigenomic profiling of open but there is also a huge benefit to the cytokine inhibitors and antibodies directed chromatin, DNA-binding proteins and nucleosome community in general. The annotation of against various immune cells. Markers for position. Nat. Methods 10, 1213–1218 (2013). 17. Thurman, R. et al. The accessible chromatin landscape regulatory elements improves as chromatin disease classification, prognosis, therapeu- of the human genome. Nature 489, 75–82 (2012). profiles for each additional immune cell tic response and toxicity are lacking, as is 18. Cusanovich, D. A. et al. Multiplex single-cell profiling of chromatin accessibility by combinatorial cellular type are completed. From these combined a better understanding of the pathogenic indexing. Science 348, 910–914 (2015). data sets, we can produce a publicly avail- mechanisms that initiate and perpetuate 19. Greenleaf, W. J. Assaying the epigenome in limited numbers of cells. Methods 72, 51–56 (2015). able map of the regulatory landscape across the disease. We predict that chromatin 20. Cedar, H. & Bergman, Y. Epigenetics of the haematopoietic system in various spe- profiling techniques will eventually have a haematopoietic cell development. Nat. Rev. Immunol. 11, 478–488 (2011). cies and tissues. Then, chromatin analysis marked impact on how clinicians diagnose 21. Broske, A.‑M. et al. DNA methylation protects will become as commonplace as expression and target the appropriate causative cell hematopoietic stem cell multipotency from myeloerythroid restriction. Nat. Genet. 41, analysis, with easily accessible databases type and pathway on a patient-by‑patient 1207–1215 (2009). used for reference and meta-analyses. In basis. Instead of treating symptoms, or even 22. Ji, H. et al. Comprehensive methylome map of lineage commitment from haematopoietic progenitors. the future, researchers will be able to simply changes in downstream gene expression, Nature 467, 338–342 (2010). look up the enhancers regulating any given medication could be aimed at correcting 23. Shlush, L. I. et al. Identification of pre-leukaemic haematopoietic stem cells in acute leukaemia. gene in any condition. Of course, these the offending fault in the cell’s regulatory Nature 506, 328–333 (2014). maps should be further substantiated and network. Given the high value and infor- 24. Xie, W. et al. Epigenomic analysis of multilineage differentiation of human embryonic stem cells. improved with functional assays, such mation content of chromatin profiling and Cell 153, 1134–1148 (2013). as reporters for enhancer activity, knock- the recent development of cutting-edge 25. Sun, D. et al. Epigenomic profiling of young and aged HSCs reveals concerted changes during aging that out of specific chromatin remodellers technology, this technique is simply waiting reinforce self-renewal. Cell Stem Cell 14, 673–688 (DNMTs, histone acetyltransferases and to be exploited further. (2014).

NATURE REVIEWS | IMMUNOLOGY VOLUME 15 | SEPTEMBER 2015 | 593

26. Giresi, P., Kim, J., McDaniell, R., Iyer, V. & Lieb, J. 45. Burzyn, D., Benoist, C. & Mathis, D. Regulatory 66. McDaniell, R. et al. Heritable individual-specific and FAIRE (formaldehyde-assisted isolation of regulatory T cells in nonlymphoid tissues. Nat. Immunol. 14, allele-specific chromatin signatures in humans. elements) isolates active regulatory elements from 1007–1013 (2013). Science 328, 235–239 (2010). human chromatin. Genome Res. 17, 877–885 (2007). 46. Yokoyama, W. M., Sojka, D. K., Peng, H. & Tian, Z. 67. Mikkelsen, T. et al. Genome-wide maps of chromatin 27. Heintzman, N. et al. Distinct and predictive chromatin Tissue-resident natural killer cells. Cold Spring Harb. state in pluripotent and lineage-committed cells. signatures of transcriptional promoters and enhancers Symp. Quant. Biol. 78, 149–156 (2013). Nature 448, 553–560 (2007). in the human genome. Nat. Genet. 39, 311–318 (2007). 47. Glasmacher, E. et al. A genomic regulatory element 68. Suzuki, T. et al. Pulmonary macrophage 28. Creyghton, M. et al. Histone H3K27ac separates that directs assembly and function of immune-specific transplantation therapy. Nature 514, 450–454 active from poised enhancers and predicts AP‑1–IRF complexes. Science 338, 975–980 (2014). developmental state. Proc. Natl Acad. Sci. USA 107, (2012). 69. Richmond, T. & Davey, C. The structure of DNA in 21931–21936 (2010). 48. Ciofani, M. et al. A validated regulatory network for the nucleosome core. Nature 423, 145–150 29. Lieberman-Aiden, E. et al. Comprehensive mapping of Th17 cell specification. Cell 151, 289–303 (2012). (2003). long-range interactions reveals folding principles of 49. Boyle, A. et al. High-resolution genome-wide in vivo 70. Felsenfeld, G. & Groudine, M. Controlling the double the human genome. Science 326, 289–293 (2009). footprinting of diverse transcription factors in human helix. Nature 421, 448–453 (2003). 30. Sanyal, A., Lajoie, B., Jain, G. & Dekker, J. The long- cells. Genome Res. 21, 456–464 (2011). 71. Kouzarides, T. Chromatin modifications and their range interaction landscape of gene promoters. 50. Pique-Regi, R. et al. Accurate inference of transcription function. Cell 128, 693–705 (2007). Nature 489, 109–113 (2012). factor binding from DNA sequence and chromatin 72. Gross, D. S. & Garrard, W. T. Nuclease hypersensitive 31. Kieffer-Kwon, K.‑R. et al. Interactome maps of mouse accessibility data. Genome Res. 21, 447–455 (2011). sites in chromatin. Annu. Rev. Biochem. 57, 159–197 gene regulatory domains reveal basic principles of 51. Neph, S. et al. An expansive human regulatory (1988). transcriptional regulation. Cell 155, 1507–1520 lexicon encoded in transcription factor footprints. 73. Amendola, M. & Van Steensel, B. Mechanisms and (2013). Nature 489, 83–90 (2012). dynamics of nuclear lamina–genome interactions. 32. Cui, K. et al. Chromatin signatures in multipotent 52. Zaret, K. S. & Carroll, J. S. Pioneer transcription Curr. Opin. Cell Biol. 28, 61–68 (2014). human hematopoietic stem cells indicate the fate of factors: establishing competence for gene expression. 74. Grewal, S. & Moazed, D. Heterochromatin and bivalent genes during differentiation. Cell Stem Cell Genes Dev. 25, 2227–2241 (2011). epigenetic control of gene expression. Science 301, 4, 80–93 (2009). 53. Cirillo, L. & Zaret, K. An early developmental 798–802 (2003). 33. Bernstein, B. E., Mikkelsen, T. S., Xie, X. & Kamal, M. transcription factor complex that is more stable on 75. Dixon, J. et al. Topological domains in mammalian A bivalent chromatin structure marks key nucleosome core particles than on free DNA. genomes identified by analysis of chromatin developmental genes in embryonic stem cells. Mol. Cell 4, 961–969 (1999). interactions. Nature 485, 376–380 (2012). Cell 125, 315–326 (2006). 54. Ostuni, R. & Natoli, G. Lineages, cell types and 76. Rao, Suhas, S. P. et al. A 3D map of the human 34. Rada-Iglesias, A. et al. A unique chromatin signature functional states: a genomic view. Curr. Opin. Cell Biol. genome at kilobase resolution reveals principles uncovers early developmental enhancers in humans. 25, 759–764 (2013). of chromatin looping. Cell 159, 1665–1680 Nature 470, 279–283 (2010). 55. Winter, D. R. & Amit, I. The role of chromatin (2014). 35. Odink, K. et al. Two calcium-binding proteins in dynamics in immune cell development. Immunol. Rev. 77. Johnson, D. S., Mortazavi, A., Myers, R. M. & infiltrate macrophages of rheumatoid arthritis. 261, 9–22 (2014). Wold, B. Genome-wide mapping of in vivo protein- Nature 330, 80–82 (1987). 56. Vilagos, B. et al. Essential role of EBF1 in the DNA interactions. Science 316, 1497–1502 36. Lam, M. T. Y. et al. Rev-Erbs repress macrophage generation and function of distinct mature B cell types. (2007). gene expression by inhibiting enhancer-directed J. Exp. Med. 209, 775–792 (2012). transcription. Nature 498, 511–515 (2013). 57. Adams, D. et al. BLUEPRINT to decode the Acknowledgements 37. Treiber, T. et al. Early B cell factor 1 regulates B cell epigenetic signature written in blood. Nat. Biotechnol. The authors thank G. Brodsky for drafting the artwork. gene networks by activation, repression, and 30, 224–226 (2012). Research in the laboratory of I.A. is supported by transcription- independent poising of chromatin. 58. Martens, J. H. A. & Stunnenberg, H. G. the European Research Council (309788) and the Israeli Immunity 32, 714–725 (2010). BLUEPRINT: mapping human blood cell epigenomes. Science Foundation (712163), The Ernest and Bonnie Beutler 38. Garber, M. et al. A high-throughput chromatin Haematologica 98, 1487–1489 (2013). Research Program of Excellence in Genomic Medicine immunoprecipitation approach reveals principles of 59. Pulendran, B. Systems vaccinology: Probing humanity’s and the Center for Excellence in Genome Science from the dynamic gene regulation in mammals. Mol. Cell 47, diverse immune systems with vaccines. Proc. Natl National Human Genome Research I n s t i t u t e ( N H G R I; 810–822 (2012). Acad. Sci. USA 111, 12300–12306 (2014). 1P50HG006193). Research in the laboratory of S.J. is sup- 39. Monticelli, S. & Natoli, G. Short-term memory of 60. Brodin, P. et al. Variation in the human immune ported by the Israeli Science Foundation (887/11), the danger signals and environmental stimuli in immune system is largely driven by non-heritable influences. European Research Council (340345) and the Deutsche cells. Nat. Immunol. 14, 777–784 (2013). Cell 160, 37–47 (2015). Forschungsgemeinschaft (FOR1336). D.R.W. is grateful 40. Netea, M. G., Quintin, J. & van der Meer, J. W. M. 61. Ballester, B. et al. Multi-species, multi-transcription to the Azrieli Foundation for the Azrieli Fellowship award Trained immunity: a memory for innate host defense. factor binding highlights conserved control of tissue- and for support from the European Molecular Biology Cell Host Microbe 9, 355–361 (2011). specific biological pathways. eLife 3, e02626 (2014). Organization (EMBO; ALT766‑2014) and the European 41. Foster, S., Hargreaves, D. & Medzhitov, R. Gene-specific 62. Heinz, S. et al. Effect of natural genetic variation on Commission FP7 (Marie Curie Actions, EMBOCOFUND2012, control of inflammation by TLR-induced chromatin enhancer selection and function. Nature 503, GA‑2012‑600394). modifications. Nature 447, 972–978 (2007). 487–492 (2013). 42. Ginhoux, F. & Jung, S. Monocytes and macrophages: 63. Rosengren Pielberg, G. et al. A cis-acting regulatory Competing interests statement developmental pathways and tissue homeostasis. mutation causes premature hair graying and The authors declare no competing financial interests. Nat. Rev. Immunol. 14, 392–404 (2014). susceptibility to melanoma in the horse. 43. Okabe, Y. & Medzhitov, R. Tissue-specific signals Nat. Genet. 40, 1004–1009 (2008). control reversible program of localization and 64. Degner, J. et al. DNase I sensitivity QTLs are a FURTHER INFORMATION functional polarization of macrophages. Cell 157, major determinant of human expression variation. Blueprint consortium: http://www.blueprint-epigenome.eu/ 1–13 (2014). Nature 482, 390–394 (2012). ENCODE: https://www.encodeproject.org/ 44. Varol, C., Mildner, A. & Jung, S. Macrophages: 65. Kasowski, M. et al. Variation in transcription factor ImmGen: https://www.immgen.org/ development and tissue specialization. Annu. Rev. binding among humans. Science 328, 232–235 ALL LINKS ARE ACTIVE IN THE ONLINE PDF Immunol. 33, 643–675 (2015). (2010).

594 | SEPTEMBER 2015 | VOLUME 15 www.nature.com/reviews/immunol