Transcriptional programmes underlying cellular identity and microbial responsiveness in the intestinal epithelium

Abstract

The intestinal epithelium serves the unique and critical function of harvesting dietary nutrients, while simultaneously acting as a cellular barrier separating tissues from the luminal environment and gut microbial ecosystem. Two salient features of the intestinal epithelium enable it to perform these complex functions. First, cells within the intestinal epithelium achieve a wide range of specialized identities, including different cell types and distinct anterior–posterior patterning along the intestine. Second, intestinal epithelial cells are sensitive and responsive to the dynamic milieu of dietary nutrients, xenobiotics and microorganisms encountered in the intestinal luminal environment. These diverse identities and responsiveness of intestinal epithelial cells are achieved in part through the differential transcription of genes encoded in their shared genome. Here, we review insights from mice and other vertebrate models into the transcriptional regulatory mechanisms underlying intestinal epithelial identity and microbial responsiveness, including DNA methylation, chromatin accessibility, histone modifications and transcription factors. These studies are revealing that most transcription factors involved in intestinal epithelial identity also respond to changes in the microbiota, raising both opportunities and challenges to discern the underlying integrative transcriptional regulatory networks.

Introduction

All differentiated cells and tissues in the animal body face the paradoxical challenge of maintaining their core identity and function while simultaneously retaining the plasticity to respond appropriately to changes in the environment. An archetypal example of this paradox is the intestinal epithelium, a rapidly renewing and remarkably resilient tissue that interfaces with a complex luminal environment. The contents of the intestinal lumen consist of a dynamic milieu of biochemical stimuli rarely encountered by most other cells in the body, including commensal and pathogenic microorganisms, dietary nutrients, xenobiotic chemicals and host-derived factors such as digestive enzymes and bile salts. The phenotypic plasticity that enables the intestinal epithelium to respond to this highly diverse array of inputs is defined by its core gene expression programmes.

These gene expression programmes can be divided into two major categories: those that shape the identity of the intestinal epithelium and those that mediate the sensitivity of the epithelium by responding to environmental cues. A comprehensive discussion of the diverse environmental cues encountered by intestinal epithelial cells (IECs) is a prohibitively broad topic, and, therefore, we focus here on the known effects of factors generated or modified by the communities of microorganisms that reside in the intestine, known as the gut microbiota. We define intestinal ‘identity’ as the hypothetical basal state and developmental programme encoded in an animal’s genome. These identity programmes are capable of specifying and replenishing a regionalized layer of diverse IEC subtypes, giving rise to the emergent functions of the intestine ^1–4. We define intestinal ‘sensitivity’ or ‘response’ programmes as those that work in parallel with established identity programmes to detect and respond to microbial cells and their products that are present in the luminal environment. In this Review, we use the terms ‘sensitivity’ and ‘responsiveness’ somewhat interchangeably to describe epithelial changes that result from microbial stimuli. However, their definitions have important distinctions in a biological context. ‘Sensitivity’ often describes the ability of host cells to initially perceive microbial signals. By contrast, ‘responsiveness’ describes the ability of host cells to translate those perceived signals into activation of downstream signal transduction pathways, including transcriptional programmes⁵. Here, we use both terms, but we use responsiveness most often as it is more easily and more frequently quantified experimentally.

The substantial phenotypic plasticity displayed by the intestinal epithelium is made possible by the transcriptional networks that define intestinal identity and sensitivity. For example, aspects of identity, such as cell type, and how many cells of a given type reside within a tissue affect the tissue’s capacity for sensitivity⁶. Although that and many other examples of interconnectedness are readily evident, the relationships between the transcriptional programmes that define identity and sensitivity remain unclear. In this Review, our working conceptual framework is that integrated regulatory networks define the identity and sensitivity of the intestinal epithelium. This framework is supported by emerging genome-wide chromatin and transcription assays, as described later, and also reflects the majority of the literature in the field, which often assesses the effects of transcriptional regulators on either identity or sensitivity, without exploring potential interactions between the two.

This Review focuses on transcriptional regulatory mechanisms in the intestinal epithelium of vertebrates, with an emphasis on humans, mice and zebrafish, in which the majority of vertebrate genetic and functional genomic analyses have been undertaken. However, it is important to remember that these species only represent a small fraction of vertebrate diversity. The vertebrate digestive tract typically includes a headgut involved in food capture and mechanical processing, a foregut consisting of an oesophagus and a stomach, and an intestine consisting of a midgut (that is, a small intestine) and a hindgut (that is, a colon)⁷. This anterior–posterior regional plan has undergone distinct adaptations in diverse vertebrate lineages to yield a wide range of differences in the length and anatomy of digestive tract regions. Different vertebrate species harbour their densest microbial communities in different parts of their digestive tract. For example, humans and mice harbour their densest communities of microorganisms in their colon, whereas diverse herbivorous mammals such as ruminants, macropod marsupials and the colobus and langur monkeys have modified forestomachs to serve that purpose⁸. Despite these differences, there are features of the intestine that seem to be broadly conserved among vertebrate species ⁸. For example, vertebrates as distantly related as mammals and fishes still share most of their major IEC types (stem cells, absorptive enterocytes, mucus-producing goblet cells and enteroendocrine cells) ^9–11 as well as a shared transcriptional regulatory network ¹².

Microbially responsive programmes can be general as well as confer distinct cell-specific and region-specific differences in intestinal gene expression that can vary depending on the presence and composition of the luminal microbiota ^3,4,13. Many of the genes involved in intestinal identity and microbial responsiveness are implicated in human intestinal diseases such as the inflammatory bowel diseases (IBD). Thus, there is a compelling need to understand the transcriptional regulatory programmes that underlie intestinal identity and microbial response in health and disease. In this Review, we summarize how intestinal identity is established on different spatial and temporal scales, and then discuss our current understanding of the epigenetic mechanisms and transcription factors that contribute to the identity and microbial responsiveness of the intestinal epithelium. Finally, we offer perspectives on the future directions of this rapidly changing field. Owing to the vast literature, we limit our discussion to key examples and provide a more comprehensive list of major intestine-associated transcription factors in Supplementary Table 1.

IEC identity: spatial and temporal scales

Although the intestinal epithelium consists of a continuous single-cell layer, its diverse functions require the establishment of appropriate cell types, tissue architecture and regional physiologies. Consequently, IEC identities vary across multiple scales of time and space ¹⁴. The location of an IEC along the anterior–posterior axis of the intestine, its position relative to the stem cell niche (or along the crypt–villus axis), and its cellular neighbors all affect its identity. Changes to intestinal identity occur on different temporal scales. Major developmental stages such as embryogenesis, larval or juvenile periods, adulthood and ageing all influence identity ¹⁵. The majority of cells in the intestinal epithelium are turned over every few days in adult humans, mice and zebrafish^9,16. Environmental factors including the gut microbiota can affect the rate of cell divisions and the pace of epithelial renewal ^17–19. Microbial effects on host intestinal epithelial metabolism is highly relevant to both identity and responsiveness programmes. In this Review, we do not comprehensively discuss metabolism, which likely merits its own comprehensive review, but key examples are cited.

Intestinal cell identities

In the mature intestinal epithelium, IECs begin as multipotent intestinal stem cells (ISCs) residing in specialized niches at the base of intestinal crypts or rugae. Niches maintain the undifferentiated state of ISCs and minimize cellular and DNA damage that might be caused by environmental insults²⁰. Host-derived signalling from the subepithelial stromal microenvironment and by adjacent cells in the epithelial microenvironment is important for niche maintenance ²¹. In the small intestine, Paneth cells provide ISCs with pro-proliferative growth factors and signals to maintain the ISC proliferative potential ^22,23. WNT signalling has a crucial role in the maintenance of ISCs and crypt compartments, with β-catenin and TCF transcription factors being the primary transcriptional mediators of the WNT signalling pathway ²⁴. An important transcription factor downstream of WNT signalling in intestinal stem cells is Achaete scute-like 2 (ASCL2), which controls ISC stemness and fate ^25,26. Intestinal progenitor cells have differing capacities to revert to ISC fate and renew the epithelium in response to damage, and these populations might express different transcription factors from those cells that do not revert (reviewed previously²⁷).

ISCs drive the rapid renewal of the intestinal epithelium, generating progenitor cells that differentiate into specialized IEC types ²⁸. When ISCs divide, one daughter cell remains an ISC in the niche compartment, and the other migrates along the epithelial layer through a transit-amplifying region where cell division occurs. In the transit-amplifying zone, transcriptional programmes that lead to the terminal differentiation of progenitor cells into the different IEC cell types are initiated²⁰. Differentiated epithelial cell types fall into two major lineages: secretory cells, which include Paneth cells, tuft cells, goblet cells and enteroendocrine cells; and absorptive enterocytes, which make up the vast majority of the villus epithelium.

In the mouse small intestine, secretory and absorptive cell lineages are specified in the transit-amplifying compartment by Notch signalling between neighbouring progenitor cells ^29,30. Notch signalling is central to the specification of secretory and absorptive cell types in the intestinal epithelium of arthropods as well as vertebrates ³¹, which suggests that Notch signalling was a part of the intestinal identity programme in their last shared common ancestor, over 600 million years ago ³². Notch ligands such as Delta activate translocation of the Notch intracellular domain (NICD) to the nucleus of neighbouring progenitor cells, where NICD associates with the transcription factor recombination signal binding protein for immunoglobulin κJ region (RBPJ) to activate target genes³³. High Notch signalling is achieved in a subset of progenitor cells through lateral inhibition, a process by which initially small differences in the amount of Notch ligand between neighbouring cells are amplified by a feedback loop in which the cells receiving more Notch signalling repress expression of the Notch ligand ³⁴. Notch signalling through Hes family BHLH transcription factor 1 (HES1) inhibits the activity of atonal BHLH transcription factor 1 (ATOH1), and ATOH1 promotes secretory cell differentiation^29,30. The high-Notch absorptive progenitor cells are more proliferative than the low-Notch secretory cells, which under homeostatic conditions results in an epithelial layer that is dominated by absorptive enterocytes ³⁵. Other transcription factors that are important for specific secretory lineages include SAM pointed domain ETS factor (SPDEF) for goblet and Paneth cells ³⁶, Neurogenin-3 (NEUROG3) for enteroendocrine cells ³⁷ and POU2F3 for Tuft cells ³⁸ (see ref.²⁰ for a more detailed review).

Epithelial cells continue to migrate from the crypt up to the villus tip through gradients of decreasing WNT signalling and increasing bone morphogenetic protein (BMP) signalling ³⁹. BMP signalling in the mouse small intestinal epithelium through BMP receptor type 1A (BMPR1A) negatively influences proliferation and differentiation of secretory lineage cells in the transit-amplifying region ⁴⁰. Different BMP levels have also been shown to induce the expression of different enteroendocrine cell hormones in mouse small intestinal organoids and in vivo⁴¹, and the BMP-regulated transcription factor SMAD4 participates in a feed-forward loop with the transcription factor hepatocyte nuclear factor 4α (HNF4A) to promote enterocyte fate ⁴², suggesting that BMP–SMAD signalling has diverse and important roles in epithelial cell differentiation. IECs end their journey from the crypt to the villus tip in the zone of extrusion, where mature epithelial cells undergo apoptosis and leave the epithelium through a process called anoikis⁴³.

Our knowledge of IEC diversity at different life stages is far from complete, and stage-specific identity programmes for different cell types might exist ⁴⁴. For example, a specialized population of lysosome-rich enterocytes (LREs) develop in the ileum of perinatal mammals but disappear at later stages upon the onset of peptic digestion ⁴⁵. Interestingly, LREs also develop in the corresponding ileal region of the zebrafish intestine during early larval development and are maintained into adult stages ⁴⁶. Cells with LRE morphology are also observed in a similar region of the posterior intestine and hindgut in adult jawless lamprey, suggesting that this poorly understood cell type has ancient origins and function⁴⁷. The transcription factors that govern the development and disappearance of LREs in mammals remain unknown, but mutation of the endocytic machinery used by LREs to uptake dietary protein leads to impaired growth and survival in zebrafish and mice⁴⁶. This work underscores the fact that there are still major IEC subtypes yet to be characterized, particularly at early life stages, and highlights the utility of complementary models such as the zebrafish in understanding their physiological roles.

Single-cell RNA sequencing of thousands of single IECs ¹ and RNA sequencing of regions of enterocytes along the crypt–villus axis ² from mice have provided an unprecedented amount of detail about gene expression patterns in villus IECs. For example, Moor and colleagues found zones along the crypt–villus axis where enterocytes express specific nutrient transporters ². These high-resolution sequencing studies have revealed with increased granularity the diversity of gene expression patterns within cell lineages, as well as new biomarkers of these populations. The field is now poised to uncover details about how chromatin architecture and transcription factor expression along both the crypt–villus and anterior–posterior axes of the intestine shape these nuanced cell identities.

Epithelial regional patterning and identity

During gastrointestinal development, transcriptional programmes specify functional regions along the anterior–posterior axis of the intestine. Regions of the intestine have specialized roles in the digestion and absorption of dietary nutrients. In mature animals, the composition and density of the intestinal microbiota also varies along the length of the intestinal tract ⁴⁸. Unsurprisingly, IECs display different gene expression programmes along the anterior–posterior axis of the intestine to execute these diverse functions ^3,4. Regional gene expression patterns are highly conserved between mammals and zebrafish, which last shared a common ancestor 420 million years ago ^12,49. Studies in adult and paediatric humans suggest that regional gene expression and gene regulatory features, including promoter and enhancer activity ⁵⁰, and chromatin architecture and accessibility ⁵¹ are perturbed in Crohn’s disease and ulcerative colitis. For example, gene expression profiling of biopsy samples from the descending colon of active Crohn’s disease or ulcerative colitis tissue revealed potential differences in the types of intestinal epithelial cells that made up the biopsy samples from the different IBDs ⁵⁰. In a separate study, gene expression and accessibility of non-inflamed colon tissue from patients with Crohn’s disease and healthy controls revealed distinct subtypes within the tissues of patients with Crohn’s disease linked to regional gene expression patterns. Biopsy samples of patients with Crohn’s disease had either a more normal colonic gene expression profile, or transformed, ileum-like gene expression, phenotypes that might be useful for disease classification and prognosis ⁵¹. Human and mouse ISCs harvested from different regions of the small intestine retain region-specific gene expression patterns even after long-term organoid culture, suggesting that in the adult intestinal epithelium, regional expression programmes are an intrinsic aspect of mature ISC identity ⁵².

Multiple transcription factors are important for specifying discrete regions of the intestinal epithelium (see ref.⁷ for an in-depth review of these factors). In mice and humans, regional patterning of the intestinal epithelium begins in utero, prior to the intestinal tract receiving environmental stimuli from the microbiota or diet. Expression of SRY-box 2 (SOX2) and caudal type homeobox 2 (CDX2) in the endoderm-derived inner layer of the gastrointestinal tract delineates the anterior, oesophageal–gastric epithelium from the posterior, intestinal epithelium ^53–55, and numerous gain-of function and loss-of-function experiments have demonstrated that SOX2 drives oesophageal–gastric cell fate while CDX2 drives intestinal cell fate ^53,56–62. These early steps of intestinal epithelial patterning support the subsequent expression of regional and lineage-specific transcription factors.

GATA transcription factors, named for their consensus DNA binding sequence (T/A)GATA(A/G), are important for further regionalizing the intestinal epithelium established by CDX2. IEC-specific gain-of-function and loss-of-function studies in mice have established GATA4 as an important driver of jejunal physiology ^63,64,65. ChIP sequencing of GATA4 in the jejunum revealed binding to activated jejunum-associated genes and to repressed ileum-associated genes, suggesting that GATA4 directly induces jejunal identity and represses ileal identity ⁶⁵. Conditional deletion of another GATA transcription factor, GATA6, in the intestinal epithelium revealed that it is required for normal morphology and enterocyte gene expression in the mouse ileum⁶⁶. GATA4 and GATA6 seem to function somewhat redundantly in the jejunum — loss of both factors results in increased numbers of goblet cells through reduced Notch signalling in progenitor cells ⁶⁶. Other transcription factors that influence regional patterning and physiologies continue to be revealed, including through single-cell sequencing of the small intestinal epithelium ^1,67.

Diverse model systems also enable the exploration of the natural variation in regional gastrointestinal anatomy that occurs among vertebrates, and how the roles of conserved transcription factors contribute to the evolutionary plasticity of intestinal identity. Stomach loss is common throughout teleost fish lineages, with 20–27% of teleost fish species estimated to be agastric ⁶⁸. Often, the absence of a stomach is accompanied by loss of pepsinogen and proton pump genes that are required to acidify the stomach ⁶⁹. In the stomachless zebrafish, the expression of sox2 extends past the oesophageal boundary into the proximal intestine, where it overlaps with the prospective stomach marker barx1 in the surrounding mesodermal tissue ⁷⁰. The presence of these gastric tissue-associated factors in the anterior of the zebrafish intestine could indicate a vestigial gene programme for stomach formation in this stomachless fish and might provide an inroad to understand the evolutionary mechanisms of gastrointestinal patterning.

Epigenetics in identity and sensitivity

Changes in the chromatin landscape are critical for the establishment and maintenance of intestinal cellular identity. DNA methylation, chromatin accessibility and histone modifications are each a separate layer of information that together regulate gene expression primarily by restricting or promoting transcription factor access to sequence-specific binding sites within cis-regulatory DNA regions (CRRs). CRRs can harbour binding sites for multiple activating or repressing transcription factors. CRRs are typically associated with nucleosome depletion resulting in accessible chromatin and specific post-translational modifications (PTMs) of histone proteins, for example, H3K4 monomethylation (H3K4me1) and H3K27 acetylation (H3K27ac), which mark regions associated with higher transcriptional activation ⁷¹. Collectively, these levels of chromatin regulation lead to differential utilization of CRRs by transcription factors, resulting in distinct identities and environmental sensitivities. Our understanding of the cis-regulatory landscapes upon which the transcription factors involved in intestinal identity and sensitivity work has advanced rapidly in the past decade through the application of genomic assays that enable us to understand the distribution of chromatin structure and modifications, transcription factor occupancy and gene transcription across the genome in different intestinal regions, cell types and environments (Table 1).

Table 1 |.

Transcription factors and microbial ligands

Transcription factor family	Ligand class and examples	Microbial influence?	References
HNF4	Fatty acids (such as palmytic acid and linoleic acid)	Possible microbial metabolites	142,143
FXR	Bile acids (primary and secondary)	Microorganisms modify through bile salt hydrolases	176,222
PPAR	Unsaturated fatty acids and eicosanoids (such as docosahexaenoic acid and prostaglandin J2)	Possible microbial metabolites	182,183
PXR	Diverse xenobiotics (such as prescription drugs, herbal medicines and environmental pollutants)	Unknown	195
CAR	Diverse xenobiotics (such as phenobarbital-like pesticides and androstenol)	Unknown	195
AHR	Diverse xenobiotics (such as dioxins, tryptophan, indoles and butyrate)	Some ligands are known microbial metabolites	204–209

DNA methylation

The most common form of DNA methylation in vertebrates is on cytosines of CpG dinucleotides, typically leading to suppressed transcription at nearby genes ⁷². These epigenetic marks are selectively deposited and maintained by DNA methyltransferases ⁷². The DNA methyltransferases DNMT1 and DNMT3A are both expressed in the crypt region of the intestinal epithelium, which suggests that DNA methylation might be an important aspect of ISC maintenance and/or proliferation ⁷³. Conditional IEC-specific knockout of Dnmt1, the maintenance DNA methyltransferase, results in crypt expansion and delayed IEC differentiation in mice ⁷³. In both zebrafish and mice, defects in DNA methylation machinery in the intestine lead to apoptosis of IECs, impaired barrier function and inflammation ^74,75. Identification of the specific mechanisms leading to these mutant phenotypes presents a major challenge because loss-of-function mutations in DNA methyltransferase genes result in highly pleiotropic phenotypes.

DNA methylation has also been examined directly by genome-wide bisulfite sequencing, revealing important potential roles for DNA methylation in different aspects of IEC identity^73,76,77. Multiple groups have compared DNA methylation in ISCs with that of differentiated IEC populations in both the mouse small intestine and colon. The findings are somewhat conflicting, with two groups noting only subtle differences in DNA methylation between ISCs and differentiated IECs ^76,77 and one group observing more pronounced differences ^73,76,77. Interestingly, when differentially methylated regions were observed between ISCs and differentiated IECs, they were concentrated specifically in CRRs and enriched for the binding sites of transcription factors that are known drivers of intestinal differentiation ^73,77. Multiple groups also found that DNA methylation patterns differ across anterior–posterior regions of the intestine ^78,79. Organoid cultures grown from different human intestinal regions retain region-specific methylation patterns even in long-term culture (11 passages; up to 3 months)⁸⁰, suggesting that regional DNA methylation patterns are stable and intrinsic features of the ISC genome.

DNA methylation might have different roles throughout intestinal development. In mice, DNMT1 is necessary for intestinal development in the perinatal period and is critical for post-embryonic maintenance of the crypt^73,74,76. Transcriptomic and epigenomic approaches have uncovered differences in fetal and mature IEC DNA methylation, suggesting that DNA methylation is dynamic during development ^78,81. Unlike the intestinal organoids derived from paediatric and adult human tissues, organoids derived from fetal tissue did undergo changes in their DNA methylation profile in long-term culture, potentially mirroring in vivo differentiation ⁸⁰. These findings fit with the general view of DNA methylation as an epigenetic strategy that once established can define transcriptional programmes long-term, and further suggest that changes in DNA methylation that occur during critical periods of development might have a lasting impact.

Although DNA methylation clearly has important roles in intestinal identity, there is also evidence that DNA methylation patterns in the mature intestinal epithelium can change in response to microbial stimuli. During the early postnatal period of intestinal development, the epithelium first adapts to changes in the luminal environment associated with feeding and introduction of the commensal microbiota. A comparison of DNA methylation between germ-free and conventionally raised mice revealed a substantial reduction in the methylation in germ-free animals of many sites across the genome that gain methylation in conventionally raised animals over the postnatal period ^76,82. These changes could be due to a decreased availability of microbially produced carbon metabolites that are necessary for methylation ⁷⁶ or due to a lack of microbial-associated molecular patterns or other microorganism-derived signals in germ-free mice.

In the context of intestinal inflammation and disease pathogenesis, sites across the genome are differentially methylated in IECs derived from patients with IBD compared with healthy controls^78,79. These differentially methylated sites are frequently found in CRRs associated with genes involved in intestinal development or function ^78,79. The precise mechanisms by which environmentally induced changes in DNA methylation might occur and how those changes contribute to inflammation or other intestinal disorders remain an ongoing area of investigation.

Cis-regulatory regions and chromatin accessibility

Chromatin accessibility broadly refers to the factors that affect the level of openness in a given genomic region. Accessible chromatin can generally be bound by transcription factors and utilized, in contrast to other states where chromatin is not open or accessible ⁸³. Accessibility is influenced by primary nucleosome positioning, histone PTMs and higher-order chromatin structures, such as heterochromatin⁸⁴. General metrics of chromatin accessibility can be queried using genome-wide assays such as ATAC-seq, PTM ChIP-seq, DNAse-seq, FAIRE-seq and MNase-seq ⁸⁴. Assessing chromatin accessibility using these approaches can identify CRRs in cell types of interest. In this section, we mainly discuss studies that utilize FAIRE-seq, ATAC-seq and DNAse-seq to define regions of DNA that are accessible in IECs, and more-specific histone PTMs are discussed in the subsequent section.

To determine whether changes in chromatin accessibility might be associated with cell-type specification prior to terminal differentiation, differences in chromatin accessibility between ISCs and enterocyte and secretory progenitors were analysed in mouse cells isolated from adult intestinal epithelial crypts^85,86. Surprisingly, chromatin profiles that signify accessibility and active enhancers were broadly similar between ISCs, secretory precursors and enterocyte precursors, suggesting that a baseline chromatin landscape is established early in progenitors and maintained until acted on by ATOH1 transcription factors during lateral inhibition and the terminal differentiation of enterocytes and secretory cells ^85,86. Interestingly, later studies did identify secretory cell progenitor-specific enhancers, and that these secretory cells specifically maintain the capacity to revert to an ISC-like cell in response to damage and loss of crypt ISCs ⁸⁷.

Overall, these studies indicate that prior to terminal differentiation, cell types are not predefined by nucleosome location or lineage-specific CRRs, and that there is a large amount of plasticity among progenitors undergoing differentiation. This observation suggests that the process of cell type-specific transcription factors binding within a relatively static, accessible chromatin landscape is a dominant mechanism of differentiation and also that environmental cues that influence these transcription factors have an opportunity to alter the process of IEC identity specification. Indeed, such a model is supported by evidence from zebrafish, in which the microbiota has been shown to influence secretory cell determination through modulation of Notch signalling in the intestinal epithelium ⁸⁸. By contrast, chromatin accessibility clearly varies in a region-specific manner along the anterior–posterior axis of the intestine, suggesting that chromatin remodelling, in addition to DNA methylation, is an important component of regional patterning during development ¹³.

Although chromatin accessibility seems to have a role in certain aspects of intestinal identity, it was unclear whether chromatin accessibility changed in response to microbial stimuli. A DNAse-seq comparison of germ-free mice colonized with a normal mouse microbiota (also known as conventionalized mice) and uncolonized, germ-free mice revealed that alterations in IEC gene expression induced by the microbiota were not accompanied by substantial changes in the position or level of openness of accessible chromatin sites across the genome ¹³. This observation suggests that alternative means of gene regulation, including differential transcription factor binding, are largely responsible for transcriptional differences between germ-free and colonized mice ^13,90. A study that used ATAC-seq to analyse cultured human colonocytes exposed in vitro to human faecal microbiota found that microbial communities induced changes in gene expression, and that a small subset of those gene loci also showed differences in chromatin accessibility, compared with colonocytes not exposed to microbiota⁹¹.

These studies suggest that non-terminally differentiated IECs within a given intestinal region retain a largely similar accessible chromatin landscape, and that sites of accessible chromatin do not change dramatically in response to diverse microbially-derived stimuli. Instead, it is likely that transcriptional changes in response to microorganisms are predominantly controlled via transcription factor binding or histone modifications. These changes in transcription factor binding or histone modifications might influence the terminal differentiation of cells or act in a cell type-specific manner in cells that are already terminally differentiated. In the future, a more-refined look at chromatin accessibility in distinct cell types in response to microbial colonization and other extrinsic environmental exposures could clarify the extent to which they influence the dynamic utilization of CRRs across the genome in IECs.

Histone modifications

Histones, the proteins that comprise nucleosomes, are functionally regulated in large part by PTMs. PTMs can be used to map the functional properties of chromatin, enabling the identification of CRRs and their activity. For example, nucleosomes flanking active enhancers are typically enriched in both H3K4me1 and H3K27ac, whereas poised enhancers have only H3K4me1 ^71,92. In the mouse small intestine, there are differences in active enhancers across developmental time, as embryonic epithelial cells harbour unique sites not seen in specified ISCs⁸¹Additionally, adult mouse IECs exhibit distinct active enhancer profiles across the different regions of the intestine ¹³. Similar to what was described earlier for open chromatin, H3K4me2 and H3K27ac sites were found to be broadly overlapping between ISCs, secretory progenitors and enhancer progenitors, suggesting that non-terminally differentiated cells in the mouse intestinal epithelium share many of the same active enhancers^85,86. When mouse ISCs and terminally differentiated IECs were compared, H3K27me3, a repressive histone mark, was enriched in IECs at genes that are active in ISCs but later silenced in the villus ⁸⁶. This finding suggests that repressive H3K27me3 might be particularly important in the silencing of stem cell-associated transcripts as differentiation proceeds. Polycomb repressive complex 2 (PRC2) controls the genomic locations of H3K27me3 modifications, and genetic evidence suggests that loss of PRC2 leads to defects in proliferation and enterocyte maturation ^86,93. In 2019, it was found that nine days after inducible PRC2-depletion in adult mouse intestinal epithelia, hypomethylated developmental enhancers are reactivated, regaining H3K27ac and H3K4me1, suggesting that reactivation or ‘memory’ of developmental programmes is possible in the intestinal epithelium via sites that maintain DNA hypomethylation status⁹⁴.

Histone modifications might also have a role in response to environmental cues. Comparing germ-free with conventionalized mouse IECs revealed that although overall chromatin accessibility was unchanged, H3K27ac and H3K4me1 are gained at loci that show increased expression in the presence of microbiota and lost at microbially-suppressed loci⁹⁰. This observation indicates that active enhancers are dynamically regulated in response to microbial colonization and are associated with gene expression changes. The enzymes responsible for adding and removing these histone modifications, histone acetyltransferases (HATs) and histone deacetylases (HDACs), respectively, work in tandem with other cofactors to modify PTMs on histone tails⁹⁵. IEC-specific Hdac1 and Hdac2-knockout mice each develop chronic intestinal inflammation and lose ISC maintenance, and double knockout mice are unable to be maintained as within days of genetic ablation, stem cells that escape knockout repopulate the intestinal epithelium ^96,97. These results establish the fundamental importance of histone acetylation in intestinal homeostasis ^96,97. Intestine-specific deletion of Hdac3 resulted in a substantial reduction in the number of Paneth cells and expression of Paneth cell-specific lysozyme, and impaired IEC function, phenotypes that were rescued when the mice were raised under germ-free conditions ⁹⁸, suggesting that histone modification via HDAC3 is essential for intestinal epithelial function in the presence of microorganisms. IEC gene expression responses to microbiota are also influenced by circadian rhythms⁹⁹. Emerging evidence suggests that the microbiota might regulate the diurnal expression of metabolic genes in small intestinal IECs in a Hdac3-dependent manner ¹⁰⁰.

Byproducts of gut microbial metabolism of dietary carbohydrates and proteins, such as short-chain fatty acids (SCFAs), can influence histone acetylation profiles in part by inhibiting HDACs¹⁰¹. Supplementing SCFAs in the diet of germ-free mice can recapitulate histone modification patterns associated with microbial colonization in the proximal colon and the liver, suggesting that the interface between diet, host and microbial metabolism might contribute to the epigenetic landscape of these cells ¹⁰². SCFAs produced by the gut microbiota can inhibit HDACs and conversely serve as cofactors for HATs ¹⁰³. This process seems to be particularly prevalent in colonic carcinoma cells, which can accumulate SCFAs in culture and inhibit HDAC activity ^103,104. However, these broad alterations in SCFAs do not explain the targeted and relatively balanced increases and decreases in H3K27ac that occur at microbially responsive loci ⁹⁰. These data suggest that targeted histone modifications might be a major mechanism of IEC gene regulation in response to the intestinal microbiota. However, the specific microbial cues and signalling pathways that lead to this targeted PTM remodelling remain poorly understood.

Lineage-specifying transcription factors

Specific types of transcription factors, termed ‘pioneer factors’, orchestrate the accessible chromatin landscape by initially displacing nucleosomes and recruiting histone-modifying enzymes to regulatory sites ^105–108. These processes enable the subsequent binding of other transcription factors to the sites of open chromatin that are also frequently marked with H3K27ac. The lineage-specifying transcription factor CDX2 might function as a pioneer factor, as it establishes regions of accessible chromatin in ISCs and progenitor cell types within the intestinal epithelium ^{57,85,109–111}. During a distinct developmental window (approximately embryonic day 14 (E14)), CDX2 drives intestinal fate at the expense of foregut (gastric-oesophageal) fate in mice, as it remodels chromatin accessibility to close foregut enhancers and open intestine-specific enhancers, restricting the fate of the epithelium ⁴⁴. This central role for CDX transcription factors in early intestinal epithelial specification seems to be deeply conserved among vertebrates, including zebrafish ¹¹². Changes in the expression or activity of lineage-specifying transcription factors such as CDX2 occur during progenitor differentiation and dedifferentiation to govern cell identity ¹¹³. Through activity at relevant CRRs, CDX2 (likely with other factors) opens accessible binding sites for specific transcription factors, such as HNF4A or RBPJ, to initiate the transcriptional programmes that drive intestinal differentiation and identity ¹⁰⁶ and to further define the chromatin architecture of specific cell types ⁴². Transcription factors such as CDX2 represent a clear source of potential feedback between transcription factor activity and chromatin accessibility.

Together, these data provide a working model for how the epigenetic landscape of the intestinal epithelium influences its identity and responsiveness to the microbiota. DNA methylation and chromatin accessibility seem to be important factors in the long-term specification of intestinal identity in precursor cell populations, including ISCs, by presenting an intestine-specific and region-specific landscape of accessible chromatin. Despite substantial transcriptional differences, the resulting landscape of CRRs does not appear to change much in response to microbial stimuli. However, microbiota-induced alterations in gene expression are often associated with PTM of histones at nearby CRRs. This observation suggests that transcriptional responses to microorganisms and other extrinsic environmental factors are determined largely by differential activation of transcription factors that then bind to a largely static landscape of accessible chromatin.

Transcription factors

Owing to the inherent interconnectedness of IEC identity and sensitivity, it is unlikely that many of the governing transcription factors function solely in one or the other. However, our current understanding of the roles of transcription factors in the intestinal epithelium treats them as largely distinct.

Intestinal identity

CDX2 and GATA transcription factors have clear and predominant roles in establishing identity of the intestinal epithelium, specifically regional patterning during development ⁷.

CDX2

CDX2 has been found to bind to distinct target genes in developing mice compared with adult mice ¹¹⁴, and further, CDX2 is required in the adult intestinal epithelium to maintain differentiated cell types ¹¹⁵. In addition, there is emerging evidence that CDX2 activity might also be sensitive to changes in the luminal environment. Increased bile acid concentrations in the oesophagus induce exogenous CDX2 expression through activity of the transcription factor nuclear factor κB (NF-κB) ¹¹⁶. This aberrant CDX2 expression establishes a chromatin architecture that promotes aberrant small intestinal epithelial transcription programmes and loss of oesophageal identity. Interestingly, the only naturally occurring bile acid found to induce CDX2 expression in the study was cholic acid, a secondary bile acid produced through microbial metabolism. Similarly, the microbial-derived SCFA butyrate stimulates CDX2 expression in the human adenocarcinoma cell lines Caco2 and HT29, which, in that context, might preserve epithelial identity and protect against oncogenesis ¹¹⁷.

GATA

By contrast, we are unaware of any published evidence suggesting that the activity of GATA factors in the vertebrate intestine are responsive to microbial stimuli. This observation might suggest that vertebrate GATA factors are remarkably well insulated from the influence of microorganisms, but this hypothesis remains to be tested directly. Indeed, multiple studies have implicated the Caenorhabditis elegans GATA homolog, ELT-2, in intestinal response to pathogenic and commensal microorganisms ^118–120. These observations provide early hints that even the transcription factors that seem to be most central to IEC identity might also be sensitive to microbial and other environmental factors.

Intestinal sensitivity

NF-κB

The NF-κB family of transcription factors are one example of a microbially sensitive transcriptional pathway. NF-κB transcription factors act rapidly downstream of several microbial-associated molecular pattern recognition receptors such as Toll-like receptors (TLRs; reviewed previously ¹²¹) and NOD-like receptors (NLRs; reviewed previously ¹²²), receptors that directly bind to microbial products. In mice, the expression of distinct TLRs on IECs is known to vary depending on the developmental stage, region of the intestine, crypt–villus axis position and IEC type ¹²³. Studies of transgenic mice and zebrafish harbouring reporters of NF-κB activity revealed cell type-specific and tissue-specific patterns of NF-κB activation, including dynamic responses to commensal microbial colonization and inflammation ^124–126. In mice, IEC-specific deletion of the NF-κB family member RelA (RelA^ΔIEC mice) reduced expression of antimicrobial genes and reduced Paneth cell numbers, a major cell type involved in small intestinal anti-microbial defense ^127,128. Strikingly, deletion of genes encoding all three NF-κB subunits containing trans-activating domains (RelA, RelB and c-Rel) in the intestine did not result in additional overt phenotypes in IEC differentiation or intestinal development and homeostasis compared with RelA^ΔIEC mice ¹²⁷, indicating a lack of compensatory activity by RelB and c-Rel in RelA^ΔIEC mice. Studies using human intestine-derived cells in culture have found that diverse microbial isolates modulate NF-κB activation, and in some cases commensal microorganisms can even suppress mediated NF-κB inflammation ^129–131. In human IBDs, inflammatory cytokines and TLRs that act upstream of NF-κB are associated with chronic inflammation, and suppression of inflammation is a major goal of many treatment strategies, including possibly suppressing aberrant NF-κB activity ¹³².

STAT

In response to direct or indirect microbial stimulation, host cells often produce cytokines and growth factors. These cytokines and growth factors activate corresponding plasma membrane receptors, potentiating a signalling cascade that results in the phosphorylation and activation of the signal transducers and activators of transcription (STAT) family of transcription factors ¹³³. Upon activation, phosphorylated STAT proteins form dimers and translocate to the nucleus, where they regulate transcriptional programmes involved in cell death, proliferation and innate immunity. In human cells in culture and mouse IECs in vivo, STAT6 is activated downstream of both IL-4 and IL-13 signalling, cytokines that are produced by immune cells during helminth infections or allergic reactions ^134,135. Constitutive activation of STAT6 in IECs increases IEC proliferation, secretory cell number and small intestinal length ¹³⁶, suggesting that it might influence intestinal cell or regional identity. Mice with STAT3-deficient IECs (Stat3^ΔIEC) have significantly more weight loss, increased recovery time and higher morbidity when treated with dextran sulfate sodium (DSS) to induce colitis compared with Villin-Cre negative Stat3^fl/fl control mice treated with DSS^137,138. STAT3 activates gene networks that restrict microbial translocation through the intestinal epithelium ¹³⁹, including that of antimicrobial peptides ^140,141, and activates pro-survival genes ¹³⁷ that restrict the extent of epithelial damage. Stat3^ΔIEC mice develop normally ¹³⁸ and do not have differences in IEC turnover rate or IEC differentiation compared with Villin-Cre negative Stat3^fl/fl control mice ¹³⁷. Thus, it seems that STAT3 does not substantially influence the establishment of intestinal identity, but rather the function of differentiated IECs in response to microbial stimuli.

Intestinal identity and sensitivity

HNF4

HNF4 transcription factors are well-established drivers of IEC identity, but work over the past few years has uncovered new roles for these transcription factors in IEC response to the microbiota. HNF4A (also known as NR2A1) and HNF4G (also known as NR2A2) are the two mammalian members of this nuclear receptor transcription factor family. Both mammalian HNF4 family members are expressed in digestive tissues including the intestinal epithelium ^142,143. Studies examining loss-of-function of each HNF4 factor separately revealed roles for both in cell type-specification, although neither the Hnf4a intestinal knockout nor the Hnf4g knockout mice resulted in mortality ^144,145. However, IEC-specific deletion of Hnf4a resulted in late-onset (6–12 months of age), spontaneous colonic inflammation ^144,146. These mice also display moderate defects in intestinal function including reduced villus size, impaired fatty acid absorption, decreased expression of junctional proteins and increased intestinal permeability ^{147,148,149,144,146}, and are also more susceptible to DSS-induced colitis¹⁵⁰. Hnf4g whole-animal knockout mice are also viable, but gain more weight than wild-type sibling controls on a normal diet ¹⁴⁵, likely owing to changes in the balance of cell lineages in the intestinal epithelium ¹⁵¹. Single-cell RNA sequencing and organoid models have identified HNF4G as a particularly dynamic transcription factor in IECs with an important role in promoting enterocyte fate ^1,67. Together, these studies suggest that the HNF4s individually promote intestinal homeostasis and identity.

HNF4A and HNF4G are thought to have similar DNA binding motifs ¹⁵², and the vast majority of binding sites can be bound by both transcription factors ^42,90. However, until 2019, the extent to which they function redundantly in the early intestinal epithelium to promote identity was unknown. An intestine-specific Hnf4a conditional knockout in tandem with an Hnf4g whole-animal knockout (Hnf4^DKO) revealed that HNF4s are not required for initial specification of the intestinal epithelium, nor are they required for the onset of crypt–villus axis formation. However, by E18.5 the intestines of Hnf4^DKO mice are severely underdeveloped, exhibiting shortened villi that lack expression of multiple differentiation markers ¹⁵³. When Hnf4^DKO mice are generated via tamoxifen inducible knockout during adult stages, they exhibit rapid weight loss requiring euthanasia within 5 days ⁴², a much stronger phenotype than either of the long-lived single mutants. Adult Hnf4^DKO mice also displayed decreased expression of many genes that are important for intestinal identity and involved in nutrient transport, lipid metabolic processes and brush border formation, and showed a loss of differentiated enterocytes. Further, SMADs and HNF4s were shown to coregulate genes important for enterocyte differentiation, and accordingly, loss of both factors shifted the balance of cell types in the intestinal epithelium, resulting in fewer enterocytes and more secretory cells ⁴². Additionally, emerging evidence suggests that HNF4s are important regulators of the metabolic state and renewal of ISCs by promoting the expression of genes involved in fatty acid oxidation ¹⁵⁴.

HNF4A and HNF4G have each been associated with human disease. Individuals that are heterozygous for mutations in HNF4A develop maturity onset diabetes of the young ¹⁵⁵. Three independent genome-wide association studies (GWAS) have identified variants of HNF4A that are associated with Crohn’s disease and ulcerative colitis ^156–158. Furthermore, HNF4A expression is reduced in intestinal biopsy samples from patients with ulcerative colitis and Crohn’s disease¹⁵⁹, and differentially active enhancers found in patients with IBD have predicted HNF4 binding sites ^160,161. Another study identified variants of HNF4G that were associated with ulcerative colitis ¹⁶². These studies suggest that HNF4s and their target genes might protect against inflammation and IBD, making them potential therapeutic target candidates. Together, these data establish major roles for the HNF4s in intestinal identity from late development through to adult homeostasis. Although significant redundancy between Hnf4a and Hnf4g in mice has been identified, their distinct human IBD associations and individual mouse mutant phenotypes underscore that they have different, non-redundant roles with important health implications.

In addition to specifying important aspects of intestinal identity, another study has revealed surprising roles for HNF4 transcription factors in response to microorganisms ^89,90. Colonization of germ-free mice with a conventional microbiota led to genome-wide reductions of HNF4A and HNF4G chromatin occupancy compared with germ-free controls ⁹⁰. Genetic analysis of hnf4a in zebrafish revealed that Hnf4a positively regulates many genes that are downregulated by the microbiota ⁹⁰. The identification of HNF4A as a novel mediator of IEC responses to the microbiota provides new insights into nuclear receptor biology, host–microbiota interactions and intestinal pathophysiology. Several studies have found that as nuclear receptor transcription factors, HNF4s bind, probably constitutively, to long-chain fatty acids such as palmitic acid, palmitoyl-CoA and linoleic acid ^{142,143,163–165}. These ligands are naturally occurring products of dietary lipid metabolism in the intestine and can also be products of bacterial metabolism ^166,167. One hypothesis as to how HNF4 activity is affected by microbial colonization is that microbial colonization alters the availability of these ligands. Other possibilities are that microorganisms might regulate access to HNF4 binding sites indirectly via epigenetic mechanisms, or that the activity of HNF4 co-activators, co-repressors or other interacting proteins and signalling pathways are affected by microbial colonization. Additional studies are needed to test these and other potential mechanisms.

FXR

In both the intestine and the liver, bile acids serve as an activating ligand for the nuclear receptor farnesoid X receptor (FXR; also known as NR1H4) ¹⁶⁸. Primary bile acids are synthesized in the liver and secreted from the gallbladder into the small intestine, where they emulsify dietary fats so that they can be digested and absorbed by IECs. Bile acids are then reabsorbed in the ileal region of the small intestine and recirculated to the liver. Bile acids induce FXR activity and transcription of the bile acid-binding protein FABP6 (also called IBABP) in the intestine ¹⁶⁹. Intestinal FXR is necessary for maintaining bile acid and fatty acid metabolism ¹⁷⁰, both of which are essential functions of the ileum, but broader roles for FXR in the specification of ileal identity remain to be fully established. The regulation of FXR by bile acids in the intestine provides one of the most direct links between the activities of the intestinal microbiota and intestinal transcription factors (reviewed previously ¹⁷¹). The intestinal microbiota modifies primary bile acids through bile salt hydrolases (BSH) and other enzymes, producing secondary bile acids and their derivatives ^172,173. Diverse primary and secondary bile acids can bind to FXR with varying capacities, and have agonistic or antagonistic effects on FXR transcriptional activation ¹⁶⁸. Consequently, microbial modification of bile acids can affect systemic metabolic processes such as fat storage ^173–177. Inhibition of bacterial BSH selectively suppresses intestinal FXR signalling ¹⁷⁸. Increasing the levels of BSH-retaining species in the microbial community (for example, Lactobacillus and Bifidobacterium species) increased bile acid deconjugation and excretion in an FXR-dependent manner, suggesting that the gut microbiota can selectively tune host FXR activity by modulating the levels of different secondary bile acids ¹⁷⁶. Although possibly indirectly, the gut microbiota lowers the total bile acid pool and changes its composition by influencing expression of genes in the bile acid synthesis, conjugation, and absorption pathways¹⁷³. Loss of FXR function in mice increases their susceptibility to DSS and 2,4,6-trinitrobenzene sulfonic acid (TNBS)-induced colitis, revealing the importance of FXR function for protection against chemical-induced injury of the intestinal epithelium ¹⁷⁹. Collectively, these studies show that FXR mediates intestinal response to microorganisms through differential responses to primary and secondary bile acids and that it might also mediate intestinal identity by promoting ileal transcriptional programmes.

Other nuclear receptor transcription factors

Nuclear receptors, such as HNF4 and FXR, might be ideal candidates for mediating transcriptional responses to diverse microbial stimuli as they are ligand-responsive transcription factors, some of which are known to bind to metabolites produced by microorganisms (Table 2)^180,181. The expression of multiple nuclear receptors is downregulated in IECs in response to microbial colonization, and many nuclear receptor-binding motifs are found near genes that are also downregulated upon microbial colonization ¹³. It remains unknown how microbial colonization causes this concerted response among nuclear receptors; however, these data suggest that they have an important role in tuning gene expression in response to intestinal microorganisms.

Table 2 |.

Layers of regulation in transcriptional programmes dictating identity and responsiveness

Mechanism of transcriptional control	Techniques	Identity		Microbial responsiveness
		Cell type differentiation	Regional patterning
DNA methylation	Bisulfite sequencing	Inconclusive evidence of changes in DNA methylation between ISCs and IECS 73, 76, 77 Differences more apparent at intestinal CRRs 73, 77	DNA methylation differs across anterior–posterior axis 78, 79 Organoids derived from different regions retain methylation signatures 80	Germ-free mice displayed reduced DNA methylation compared with conventionally raised mice in the postnatal period 76, 82 Differential methylation found in IECs of patients with IBD at CRRs important for intestine function 78, 80
Chromatin accessibility	ATAC-seq or DNase-seq	ISCs and progenitor cells have broadly similar CRRs prior to lineage commitment 85, 86	Chromatin accessibility and CRRs vary across intestinal regions 13	In vivo DNase indicated few substantial changes in conventionalized mice compared with germ-free mice 13 ATAC-seq data from cultured cells indicated that microbial communities influence accessible chromatin at a small number of genes 91
Histone PTMs	Histone PTM ChIP-seq	ISCs and progenitor cells have similar active enhancer profiles 85, 86 In IECs, repressive marks increased at ISC genes that are silenced in the villus 86	Active enhancer profiles differ across intestinal regions 13	Active enhancer marks correlate with differential transcriptional activation between conventionalized and germ-free conditions 90
Transcription factors	Transcription-factor specific ChIP-seq or functional genetic analysis	TCF4 is involved in maintenance of ISC identity 24 ATOH1 promotes secretory lineage identity 29 HNF4 and SMAD promote enterocyte identity 42	CDX2 important for early patterning of intestinal fate 44 GATA4 is important for jejunal identity 63 GATA6 promotes ileal Identity 66	HNF4 is suppressed by microbial colonization 90 Other liganded transcription factors are good candidates for mediating microbially-induce changes in the environment

PPARs

The role of the peroxisome proliferator-activated receptor (PPAR) family of transcription factors (PPARα (also known as NR1C1), PPARβ/δ (also known as NR1C2) and PPARγ (also known as NR1C3)) as regulators of lipid metabolism and insulin sensitivity has long been appreciated, but these factors have also been shown to have important roles in regulating intestinal genes involved in responses to the gut microbiota and inflammation. PPARs bind to fatty acids, which might originate from dietary or microbial sources as naturally occurring ligands ¹⁸² (a recent review covers PPAR–microorganism interactions in depth ¹⁸³). In the colon, PPARγ activity is thought to be anti-inflammatory: PPARγ contributes to the maintenance of an anaerobic colonic environment that favours butyrate producers and discourages the expansion of opportunistic facultative anaerobes such as Escherichia and Salmonella species¹⁸⁴. PPARγ is also an important regulator of anti-microbial genes such as the β-defensins in the colonic epithelium, and Pparg deficiency results in impaired microbicidal capacity of the mouse colonic mucosa¹⁸⁵. Induction of PPARγ activity by chemical agonists suppressed colonic inflammation in DSS-treated mice ^186,187. These data suggest that PPARγ is important for maintenance of the colonic microbial community.

Genetic variants of human PPARG are associated with increased risk of both ulcerative colitis and Crohn’s disease, and ulcerative colitis is associated with decreased PPARG expression ^188,189. Although the exact mechanism of PPARγ involvement in IBD remains unclear, there is mounting evidence of a link between microbial-induced TLR signalling pathways and PPARγ activity ¹⁸⁸. Mice that lack TLR4 specifically in the intestinal epithelium show hallmarks of metabolic imbalance, including a significant downregulation of PPARγ target genes ¹⁹⁰. Both reducing the intestinal microbial community by antibiotic administration and treatment with a PPARγ agonist in the TLR4-knockout mice prevents metabolic phenotypes, suggesting that a microbial–TLR4 signalling pathway affects metabolism through modulation of PPARγ activity. PPARγ suppresses TLR signalling by facilitating the export of the RelA subunit of NF-κB out of the nucleus thereby inhibiting transcription of proinflammatory genes ¹⁹¹. These findings suggest that there is dynamic interplay between TLR signalling and PPARγ activity. Ppara deficiency in mice also increases susceptibility to colonic inflammation, and pharmacological activation of PPARA reduces colonic inflammation ¹⁹². An earlier study also showed that Ppara expression is suppressed in colonized mice compared with germ-free mice and helps to mediate the microbiota’s influence on diurnal gene expression in the intestine ⁹⁹. Changes in diet also affect PPAR activity, either directly through dietary changes in fatty acid availability or indirectly through the microbiota. For example, high-fat diet caused decreased PPARγ gene expression and reduced expression of PPARγ target genes in mice ¹⁹³.

CAR and PXR

Constitutive androstane receptor (CAR; also known as NR1I3) and pregnane X receptor (PXR; also known as NR1I2) activate target genes by binding to xenobiotic response elements in the intestine and liver ¹⁹⁴. These nuclear receptors and their targets, which include multiple cytochromes and transporters, are critical for the metabolism of drugs and xenobiotics to which the organism is exposed ¹⁹⁵. CAR expression in the intestine has been shown to be affected by microbial colonization in mice and by IBD in humans. CAR-mutant mice were more susceptible to DSS-induced colitis than wild-type controls ¹⁹⁶, and additionally, activation of CAR through use of an agonist increased mucosal healing in the DSS model ¹⁹⁶. Indole metabolites produced by the commensal gut microbiota induce expression of PXR target genes in jejunal enterocytes ¹⁹⁷. PXR-knockout mice have impaired small intestinal barrier function ¹⁹⁷, and activation of PXR via a chemical agonist attenuated barrier defects observed in DSS-treated mice^198,199. Decreased PXR expression was also observed in biopsy samples from actively inflamed Crohn’s disease tissue ²⁰⁰. Microorganisms in the intestine can modify and metabolize drugs in diverse ways that are still being uncovered ²⁰¹. Because PXR and CAR sense xenobiotic and microbial factors and their target gene products promote the metabolism of these factors, they might have important roles in regulating intestinal homeostasis with apparently little effect on intestinal identity.

AHR

Aryl hydrocarbon receptor (AHR) is a type of liganded transcription factor belonging to the bHLH–PAS family that has been shown to have important roles in gastrointestinal physiology and in response to xenobiotic chemicals ²⁰² (reviewed previously ²⁰³), including microbial metabolites such as tryptophan, indoles and butyrate, which have been found to activate AHR signalling ^204–209. Upon ligand binding, AHR forms a heterodimeric transcription factor complex with AHR nuclear translocator (ARNT) proteins to promote gene expression ²¹⁰. In mice, intestinal AHR is important for intestinal barrier integrity — AHR has a protective role against DSS-induced colitis through upregulation of junctional proteins and downregulation of inflammatory responses ^210–212. The activity of AHR in the mouse intestine also modulates immune response, as AHR-dependent production of the cytokine interleukin-22 (IL-22) helps to dampen intestinal inflammatory responses ^213,214, and tryptophan metabolites in the colon can activate IL-10 expression in IECs, a process that is important for wound healing and protection from colitis ²¹⁵. Thus, AHR is an attractive therapeutic target for human IBD, and natural ^216,217 and synthetic AHR ligands ²¹⁸ are being tested as potential therapeutics. Together, these results indicate that AHR is a direct sensor of microbial and other environmental factors, and that the downstream transcriptional responses in IECs contribute to the maintenance of homeostasis and the control of inflammation in the intestinal epithelium.

Challenges and future directions

The literature reviewed here has greatly advanced our understanding of how chromatin organization, DNA methylation and histone modifications contribute to intestinal identity and sensitivity, and has underscored central roles for transcription factors in coordinating these aspects of intestinal biology. Gene expression in any tissue is not mediated by any one regulatory mechanism acting in isolation, but rather through complex and dynamic interactions between all layers of regulation. This field has been successful at identifying the individual modes and mechanisms of transcriptional regulation in the intestinal epithelium and is now well placed to investigate how they interact as a function of cell type, anterior–posterior location, developmental stage, disease status, and exposure to microorganisms and other environmental factors.

Many of the studies of microbial sensitivity that have been reviewed have relied on comparing wild-type animals and gene knockout models in the context of DSS-induced colitis. DSS treatment has been tremendously helpful as a model of intestinal injury; however, it does not faithfully recapitulate many aspects of chronic intestinal inflammation in human IBD. Furthermore, DSS administration in germ-free mice leads to more-severe disease than in colonized controls ^219,220, complicating its utility for probing host–microorganism relationships. Reliance on the DSS model has, therefore, in some ways impaired the field’s progress towards understanding the transcription factor pathways involved in IBD and the roles of the gut microbiota. The common use of antibiotics to test the role of the microbiota on host biology is also of limited use, as it does not distinguish between the substantial effects of microbiota suppression, persistence of antibiotic-resistant microorganisms or direct toxic effects of antibiotics on the host ²²¹. Thus, we see a need for studies in gnotobiotic and conventional animals at homeostasis and alternate disease models to decipher the transcription factor activities and chromatin dynamics underlying host–microorganism interactions.

Another challenge is developing a balanced understanding of transcriptional regulatory programmes governing identity and sensitivity along the length of the intestine. In contrast with previous studies of transcription factors that regulate intestinal identity, which have been largely balanced between small and large intestinal regions, analysis of intestinal response to the microbiota has been disproportionately focused on the large intestine. However, the microbiota differs in composition and density along the length of the intestine ⁴⁸, and findings within the past two decades have made it clear that microorganisms and other environmental factors exert a strong influence on gene expression programmes in all intestinal segments ³. Finally, untangling the contributions of different cell types to any given phenotype, and determining how the transcriptional networks within those specific cell types might contribute to intestinal homeostasis, is another challenge. Many of the genomic assays reviewed here have surveyed bulk IECs, without resolution to distinct cell types. Thus, the development of new approaches that enable higher resolution of gene expression and transcription factor genomic occupancy at the level of single cells or specific cell types should provide some insight.

Conclusions

Of the transcription factors described in this Review, most, if not all, that have important roles in aspects of intestinal identity are also regulated by the microbiota and/or other environmental factors. Based on available data, there emerges an apparent continuum ranging from transcription factors that have a major role in identity with no currently known roles in microbial response (for example, GATA factors) to those with a major role in microbial responsiveness with little or no known role in intestinal identity (for example, NF-κB). The majority of transcription factors described in this Review seems to have important roles in both intestinal identity and sensitivity. This observation suggests an intimate mechanistic relationship between intestinal epithelial identity and response to microbial colonization that helps to explain the developmental and physiological plasticity of intestinal tissue. Understanding how these different regulatory inputs are routed, prioritized, scaled and integrated within IECs to sustain gut health remains a major challenge of systems biology, one that is of central importance to the field of gastroenterology.

Transcriptional regulatory mechanisms have a dual role in specifying intestinal epithelial identity and enabling microbial responsiveness. In this Review, Rawls and colleagues describe what is currently known about the epigenetic patterning and transcription factors responsible for this duality.

Key points

• Regional and cell identities in the intestinal epithelium of vertebrates are patterned through interactions between changes in the chromatin landscape and transcription factors. DNA methylation and accessible chromatin in intestinal epithelial cells are relatively stable in response to the gut microbiota.

• Histone modifications and transcription factor activity respond dynamically to microbial colonization.

• Transcription factors often have dual roles, to various degrees, in specifying intestinal epithelial identity and microbial responsiveness.

• Nuclear receptors seem to be key mediators of intestinal epithelial responses to the gut microbiota.

Glossary

Intestinal microbiota

The microorganisms that colonize the lumen and mucosal surfaces of the intestine.

Intestinal epithelial cells (IECs)

The cells that comprise the columnar epithelial layer that lines the lumen of the digestive tract from the anterior small intestine to the rectum, serving multiple functions including as a physical barrier and an absorptive tissue.

Transcription factors

Proteins that regulate transcription (gene expression), typically by binding to specific DNA sequences.

Anterior–posterior regional intestinal identity

Differences in the physiological function and underlying cellular composition and gene expression patterns along the anterior–posterior axis of the gastrointestinal tract. This regionality is typically categorized into the major regions of the gastrointestinal tract (for example, duodenum, jejunum, ileum, colon, etc.).

Intestinal cell type or lineage identity

The intestinal epithelium is comprised of many different specialized types of cells (for example, absorptive enterocytes, enteroendocrine cells, goblet cells) that are specified partly through distinct gene expression programs.

Intestinal stem cells (ISCs)

Cells in the intestinal epithelium that undergo self-renewal and also give rise to all of the different intestinal epithelial cell types. They are stereotypically located at the base of intestinal crypts or rugae.

Secretory IECs

A major IEC lineage with diverse functions mediated partially through secretion of products into the lumen. The secretory cell lineage includes Paneth cells, tuft cells, goblet cells and enteroendocrine cells.

Absorptive IECs

A major IEC lineage that gives rise to absorptive enterocytes, the most abundant cell type in the intestine, which are also responsible for absorption of nutrients and other cargoes.

Lysosome-rich enterocytes

Specialized absorptive cell type found in the intestines characterized by a large lysosomal vacuole and involved in cellular digestion of macromolecular cargoes. These cells develop in diverse vertebrate species but are often overlooked in mammals due to the fact that they are only present during the ‘suckling’ or perinatal developmental stages.

Inflammatory bowel diseases (IBDs)

Conditions characterized by chronic inflammation of the intestine, most commonly ulcerative colitis and Crohn’s disease.

Cis-regulatory region (CRR)

A typically non-coding region of the genome often involved in modulating the transcription of a nearby gene. This term is used interchangeably with cis-regulatory element (CRE).

DNA methylation

The addition of methyl groups on DNA bases that serves as an additional layer of genetic information that can impact the expression of genes. Typically, methylation of a gene promoter results in suppression of that gene’s transcription.

Chromatin

The complex consisting of chromosomal DNA and associated protein and RNA within the nucleus.

Nucleosome

Structural unit of DNA organization in the nucleus, consisting of DNA wrapped around a complex of eight histone proteins.

Histone modifications

Post-translational modifications on the tails of histone proteins that can influence regional gene expression and other physical and enzymatic genome utilization dynamics.

Chromatin accessibility

The degree to which regions of chromatin DNA are available for transcription factor binding or other regulatory processes. Typically, chromatin accessibility is low due to nucleosome occupancy and can be modulated by post-translational modification of histone tails. This term is used interchangeably with chromatin openness.

Short-chain fatty acids

Short-chain fatty acids are products of microbial fermentation of non-digestible dietary fibre and protein, and a main source of energy for IECs in the colon.

Lineage-specifying transcription factors

Transcription factors that function early on in the differentiation of a cell lineage or tissue that have a role in shaping the chromatin landscape, setting the stage for further transcription factor binding and lineage specification. Lineage-specifying transcription factors are functionally similar to pioneer transcription factors but unlike pioneer transcription factors they have not definitively been shown to directly bind to condensed chromatin.

Nuclear receptors

A family of ligand-binding transcription factors with important roles in IECs that regulate the expression of genes associated with diverse processes from development to metabolism.

Bile acids

Cholesterol-derived acids that are secreted from the gallbladder into the small intestine. They assist in emulsification of dietary fats, act as ligands for multiple host receptors and transcription factors and can be chemically modified by the intestinal microbiota.

Xenobiotics

Chemicals that are foreign to the body.