Dendritic Cell Subsets in Intestinal Immunity and Inflammation

The mammalian intestine is a complex environment that is constantly exposed to an array of Ags derived from food and the microbiota. It is estimated that we ingest over 100 g of foreign protein per day in our diet (1). Added to this potential antigenic load, the human intestine is colonized by ∼10¹³ bacteria (2). Physical and immunological barriers in the intestine prevent these Ags from triggering potentially detrimental immune responses. The physical barrier is composed of mucus and glycocalyx that coat a single layer of epithelial cells (3). The immunological barrier includes intraepithelial lymphocytes and immune cells residing in the intestinal lamina propria (LP) and the GALT. LP- and GALT-resident dendritic cells (DC) have been particularly implicated both in the maintenance of tolerance toward the commensal microbiota and food and in the generation of protective immune responses against pathogens. This impressive flexibility in function is due in part to the ability of DC to sense and integrate signals from their local environment, thus shaping the ensuing immune response. In this Brief Review, we review the different types of classical DC (cDC) located in the LP and the GALT, as well as discuss how cDC subsets balance tolerance and inflammation in both mice and humans.

DC are composed of a largely heterogenous population of APCs that can be mostly divided up into cDC, plasmacytoid DC (pDC), and monocyte-derived DC. Although it should be noted that all of the mentioned DC types can be found in the intestinal LP and GALT and they each exhibit unique functions, this review focuses in particular on cDC.

cDC exhibit superior capacity for taking up, processing, and presenting Ags to naive T cells (4). Bona fide cDC express high levels of CD11c and MHC class II (MHC-II), lack the expression of macrophage-associated markers CD64 and F4/80, and express the transcription factor ZBTB46 (5). cDC can be further grouped into type 1 cDC (cDC1) and type 2 cDC (cDC2): the former are superior at initiating cytotoxic CD8⁺ T cell and Th1 responses, whereas the latter excel at inducing and maintaining Th2, Th17, and T regulatory (Treg) responses (6). Within gut lymphoid tissues, cDC1 can be identified as CD4⁻CD8α⁺CD11b⁻ and cDC2 can be identified as CD4⁺CD8α⁻CD11b⁺ (Table I). In gut nonlymphoid tissues, cDC1 are usually classified as CD103⁺CD11b⁻ whereas cDC2 comprise both CD103⁺CD11b⁺ and CD103⁻CD11b⁺ (Table I). It is important to note that because macrophages are a major component in the CD103⁻CD11b⁺ population, macrophage markers (CD64, F4/80, CX₃CR1) should be included to differentiate true DC versus macrophages (7).

Because of the demonstration of nonredundant roles between the CD103⁺ and CD103⁻ cDC2 subsets and their differential reliance on certain transcription and metabolic factors, we have attempted to discern which particular cDC2 subset contributes to a described phenotype wherever possible. For example, cDC subsets can be identified based on their developmental requirement for certain transcription factors including IRF8 and BATF3 (cDC1) or IRF4 (all cDC2) and RBP-J (CD103⁺CD11b⁺ cDC2). Furthermore, although the field is adapting a new marker system to classify cDC across species and tissues using SIRPα and XCR1 (8), for the purposes of this review we will use the CD103/CD11b “historical” markers to differentiate cDC subtypes because the majority of the literature is based on this nomenclature.

DC functions can be studied in vitro and in vivo. To assess DC functionality in vitro, DC subsets are typically sorted based on their surface markers (Table I) and then cultured with danger- or pathogen-associated molecular patterns, Ags, cytokines, and transgenic reporter T cells to read out priming capacity. To isolate DC from the intestinal LP, an enzymatic strategy is usually needed because of the dense network of extracellular matrix proteins in this location. Different combinations of enzymes such as collagenase and dispase can be used to optimize the DC yield. It should be noted, however, that enzymatic treatment may result in loss of cell-surface molecules, and the yield varies from batch to batch (13). Lymphatic DC migrating from the intestine to the mesenteric lymph nodes (MLNs) can be collected by mesenteric lymphadenectomy and thoracic duct cannulation (14). To study human DC function, monocytes are isolated from the peripheral blood and cultured with conditioning media (15). DC from human intestinal biopsies can be obtained either by enzymatic digestion of the tissue (16) or using a walkout method (cells migrating out of biopsies in culture) (17).

To dissect the functions of DC in vivo, one may eliminate cDC using constitutive knockout strains, or deplete particular cDC subsets engineered to express the diphtheria toxin receptor using diphtheria toxin administration (Table II). This approach complements reporter systems that can be used to visualize cDC or cDC subsets (Table II). Because cDC2 are a functionally and phenotypically heterogenous population, care must be taken in interpreting results generated using these tools. We list some commonly used mouse models in Table II. Recent reviews also provide detailed comparisons of different mouse models studying DC and DC subsets (18, 19).

The organized structures of the GALT and the gut-draining lymph nodes (LNs) are the principal locations for priming adaptive immune cells in the intestine. The GALT is comprised of Peyer’s patches (PPs), caecal patches, and colonic patches. In addition, smaller lymphoid aggregates (isolated lymphoid follicles and cryptopatches), collectively termed solitary isolated lymphoid tissues (SILTs), are distributed along both the small and large intestine (51). As in other tissues, cDC play an important role in mounting appropriate immune responses against intestinal Ags. The individual features of the intestinal LP, gut-draining LNs, and PPs are discussed below, together with the cDC populations residing in each compartment.

The intestinal LP is located under a single layer of intestinal epithelial cells (interspersed with intraepithelial lymphocytes) and is enriched in lymphocytes, which include numerous plasma cells, and myeloid cells. Three main cDC subsets have been identified in mouse and human intestinal LP: CD103⁺CD11b⁻ cDC1, CD103⁺CD11b⁺ cDC2, and CD103⁻CD11b⁺ cDC2 (Fig. 1, Table I). Along the length of the mouse intestine, there are marked differences in the ratio of CD103⁺CD11b⁺ cDC2 versus cDC1, with cDC2 comprising the majority of cDC in the small intestinal LP but being rare in the colonic LP. By contrast, cDC1 are the major CD103⁺ cDC subset in the colonic LP. Interestingly, the cDC2/cDC1 ratio reflects the local concentration of vitamin A (51, 52). This differential distribution of cDC is also mirrored in the gut-draining LNs that drain the small intestine and colon (12, 53). Unlike in the intestinal LP, SILTs that are distributed along the intestine contain cDC clusters that are absent of cDC2 and enriched in cDC1 and pDC (54). These SILT cDC express CXCL13 to support the recruitment of B cells and consequently formation of B cell follicles (54).

MLNs, duodenopancreatic LNs, and caudal LNs drain different segments of the intestine (12, 51). Collectively, MLNs, which drain the small intestine, cecum, and the ascending colon, are the largest LNs in the body (12). Both migratory cDC and resident cDC can be found in the MLNs. Migratory cDC are MHC-II^hi with a mature phenotype, whereas resident cDC are MHC-II⁺ with an immature phenotype (55) (Table I). After activation, small intestinal cDC1 and CD103⁺ cDC2 carrying luminal Ags migrate to MLNs and present these Ags to naive T cells.

PPs are large lymphoid structures composed of aggregated lymphoid follicles surrounded by the follicle-associated epithelium (FAE) that forms the interface between the GALT and the luminal microenvironment. The FAE contains specialized cells called microfold (M) cells. M cells transport luminal Ags and bacteria toward underlying immune cells that inhibit or activate the immune response, leading to either tolerance or an inflammatory immune response. Morphologically, PPs are separated into three main domains: the follicular area, the interfollicular region, and the FAE (56). The follicle is surrounded by the subepithelial dome, which contains B cells, T cells, macrophages, and DC. cDC1 are localized within the T cell–rich interfollicular region, whereas cDC2 are present under the FAE in the subepithelial dome (57) (Fig. 1, Table I). A recent paper describes the complex network of PP-resident DC, which teases out the identity of CD11c⁺CD11b⁺ DC based on function and localization into CD11b⁺ cDC2, monocyte-derived lysozyme expressing DC, and CD4⁺ macrophages (58). Furthermore, the paper also identifies a CD11b⁻CD8α⁻ DC subset that possibly represents an immature cDC2 (58).

The anatomical distribution of cDC subsets may contribute to maintaining the necessary balance between tolerogenic and proinflammatory immune responses required in different anatomical compartments of the intestine. For example, LNs draining the duodenum are found to contain cDC with lower expression levels of inflammatory cytokine receptors and higher expression of tolerogenic factors (e.g., Aldh1a2) compared with LNs draining the ileum and colon (53). In contrast, the caudal and iliac LNs that drain the distal colon and rectum contain CD103⁻CD11b⁺ cDC2 that express cyclooxygenase-2 (responsible for generating PGE₂) and are sufficient to drive colonic tolerance (24). However, it is worth noting that cDC exhibit considerable functional plasticity that is not necessarily hardwired by anatomical location. For example, the ability of CD103⁺ cDC to promote Treg differentiation has been shown to be abrogated during intestinal inflammation, unlike during steady-state conditions (59). Therefore, the functionality of intestinal cDC to be either tolerogenic or inflammatory, independent of anatomical location, appears to also rely on environmental cues such as inflammation. Ascribed functions and cases of functional plasticity for different gut cDC subsets are discussed in the following sections.

Intestinal environmental factors contribute to imprint cDC identities and functions. The most well-studied dietary factor is retinoid acid (RA), a metabolite derived from vitamin A. Intestinal stromal cells, epithelial cells, and DC have the capacity to convert vitamin A into RA via intrinsic activity of retinaldehyde dehydrogenase (RALDH) (60–62). RA has pleiotropic effects in the intestinal immune system. In the context of DC, RA supports the development of both intestinal cDC1 and cDC2 transcriptional programs, including the upregulation of their own RALDH machinery and α4β7 surface expression on pre-DC for homing toward the intestine (62–64). Furthermore, RA preferentially skews toward cDC2 representation rather than cDC1 (63, 65).

In addition to RA, other environmental factors also play a role in dictating DC phenotype and function. In particular, short-chain fatty acids derived from dietary fiber are able to promote cDC1 and CD103⁺ cDC2 differentiation, partially via induction of RALDH expression in intestinal epithelial cells, which in turn increases Treg numbers, thus promoting oral tolerance (66, 67). Both cDC1 and CD103⁺ cDC2 express Raldh and have the capacity to imprint gut-homing capacity to Tregs by inducing the receptor CCR9 on Tregs, thus endowing them with the capacity to migrate from the MLN to the intestinal LP (63, 68, 69). Furthermore, short-chain fatty acids such as butyrate and propionate exposure endow DC with the capacity to induce Foxp3 expression in CD4⁺ T cells via histone deacetylase inhibition, although the authors did not indicate which type of DC mediate this effect (70).

The gut mucosa represents the largest reservoir of IgA-producing plasma cells (71). IgA secreted across the gut epithelium into the lumen via the polymeric IgA receptor is important for mucosal barrier protection, and influences the composition of the gut microbiota (72). IgA can be produced in a T cell–independent manner or via coordinated interactions with T follicular helper cells (Tfh) within germinal centers in the GALT that select for high-affinity B cell clones. This complexity has resulted in multiple pathways, both DC intrinsic and DC extrinsic, being identified as important for IgA production.

In the case of T-dependent IgA responses, cDC are thought to orchestrate high-affinity IgA Abs to toxins and pathogens (73). PP cDC regulate IgA class switch recombination (CSR) as well as the expression of gut-homing receptors on IgA-producing cells (74, 75). At steady-state, cDC can promote Bcl6 expression and differentiation of Tfh via IL-12 production (76) or conversion of Foxp3⁺ Treg to Tfh via CD40-CD40L interactions (77). The lymphotoxin β receptor (LTβR) pathway is particularly important for IgA production because LTβR-, LTα-, or LTβ-deficient mice exhibit profound reductions in fecal IgA levels (78), and cDC2-intrinsic LTβR signals have been implicated in this process (75).

In the case of T cell–independent IgA responses that typically target the commensal microbiota, these are mostly generated in the SILTs within the intestinal LP (79–81). Although cDC-intrinsic LTβR signaling regulates IgA CSR in the PPs (75), other cDC-extrinsic LTβR signals likely contribute to IgA CSR at steady-state such as stromal cells (78, 82). IgA responses to rotavirus do not require DC-intrinsic LTβR signals or, for that matter, Zbtb46-dependent cDC (83, 84), but rather stromal cell intrinsic LTβR signals (85). In addition to cDC, pDC from the intestine are able to induce IgA CSR in vitro (86) and Ag-specific IgA responses in mice infected with rotavirus (87).

The usual response to harmless Ags or nutrients is to induce tolerance, which prevents unnecessary inflammation and hypersensitivity. The state of hyporesponsiveness to fed Ags is known as oral tolerance (88). Intestinal cDC are likely integral in ensuring that pathological immune responses to harmless Ags do not develop. cDC that constitutively traffic out of the intestinal LP have been shown to deliver Ags from both commensal bacteria and apoptotic epithelial cells to the MLNs (89, 90). MLN stromal cells are imprinted for high Treg-inducing capacity soon after birth and contribute to lifelong homeostatic intestinal tolerance by constantly modulating functional properties of cDC1 and CD103⁺ cDC2 (60, 91). MLN cDC1 and CD103⁺ cDC2 express αvβ8 integrin, which converts latent TGF-β into its active form (92), as well RALDHs and thymic stromal lymphopoietin to foster Treg development (93–95). Signaling pathways within cDC, such as Wnt, MAPK p38, TNF receptor-associated factor 6, and TGF-βR, induce intestinal DC to express RALDH, IL-10, and TGF-β, all of which encourage Treg induction while suppressing T effector cells (96–100). Both intestinal cDC1 and CD103⁺CD11b⁺ cDC2 are able to induce Foxp3⁺ Treg in vitro (28, 101), with cDC1 expressing the highest levels of Aldh1a2, Tgfb2, and Itgb8 by RNA sequencing (28). Moreover, whereas mice lacking either cDC1 or CD103⁺CD11b⁺ cDC2 have normal Treg number in the intestine, huLangerin-DTAxBatf3 mice, which lack both CD103⁺ cDC subsets, exhibit reduced numbers of Tregs in the small intestinal LP but not in the MLNs (48). Therefore, cDC1 and CD103⁺CD11b⁺ cDC2 are mutually redundant in maintaining Treg numbers in the small intestinal LP in vivo.

In addition to the MLN stroma, mechanisms exist within the intestinal LP to maintain tolerance. Microbial components such as polysaccharide A and zymosan promote Treg function and differentiation via cDC (102, 103). In addition, exposure of cDC1 and CD103⁺ cDC2 to mucin-2, a major host-derived mucus component, subdues cDC responses to microbe-derived signals and promotes the capacity to induce oral tolerance (98). GM-CSF derived from type 3 innate lymphoid cell (ILC) promotes intestinal cDC to secrete RA, which, as mentioned, is critical for the induction of oral tolerance (104). In summary, intestinal cDC can provoke immune tolerance to self and innocuous environmental Ags in the steady-state (105).

Th1 and CTL responses are crucial for protection against intracellular pathogen challenges, such as rotavirus (84, 106), norovirus (107), Listeria monocytogenes (43), Citrobacter rodentium (108), Toxoplasma gondii (109), and Tritrichomonas musculis (110). Uncontrolled Th1 responses can be harmful and can lead to Th1-mediated colitis. By providing the polarization signal IL-12, intestinal cDC1 control Th1 and CD8⁺ T cell at steady-state and during infections (27, 32, 84). At a molecular level, TLR ligation is sufficient to induce IL-12 production from DC through a mechanism involving transcription factors IRF8 and MyD88 signaling (111–113). IFN-γ stimulation and ligation of CD40 can synergize with TLR signals in IL-12 production from DC (114, 115). In addition to cDC1, cDC2 can cross-present IgG immune complex via the neonatal Fc receptor to CD8⁺ T cells during colorectal cancer (116).

Th2 cells, whose cytokines (IL-4, IL-5, and IL-13) direct IgE- and eosinophil-mediated destruction of pathogens, are effective at controlling helminths and responsible for allergic diseases. DC, particularly the cDC2 subset, have been found to induce Th2 responses in vivo and in vitro (22, 117–119). In response to Nippostrongylus brasiliensis, Trichuris muris, and Schistosoma mansoni, cDC2-deficient mice exhibit a reduction in IL-4, IL-5, and IL-13–producing Th2 cells in the MLNs and the intestinal LP (120, 121). Interestingly, the small intestine is a major reservoir for eosinophils, which have been reported to control cDC2 activation and Th2 priming, suggesting the microenvironment created by eosinophils licenses DC activation (122, 123). In addition to eosinophils, ILC2-derived IL-13 has been shown to promote migration of activated lung DC into the mediastinal LN during the primary allergic response (124) and to elicit production of the Th2 cell–attracting chemokine CCL17 by CD103⁻CD11b⁺ cDC2 during allergic recall responses (125). Whether ILC2 play a critical role in licensing DC to promote Th2 responses in the gut has not been investigated. Recently it has been reported that CD103⁺CD11b⁺ cDC2 induce Th2 responses in the small intestine, whereas CD103⁻CD11b⁺ cDC2 perform this role in the colon, revealing a division of labor among intestinal DC in inducing Th2 responses (121). RA, in contrast, inhibits allergic responses to oral Ags by preventing MLN CD103⁻CD11b⁺ cDC2 from inducing IL-13–producing inflammatory Th2 cells (126).

Intestinal Th17 maintenance is dependent on signals (such as IL-1β) from the microbiota (127), with segmented filamentous bacteria being a prominent contributor (128). Although DC are important for the induction of Th17 cells by segmented filamentous bacteria (129), macrophages also play a key role, with perhaps both DC and macrophages compensating for each other (130). During pathogenic infections, Th17 cells provide protection against the fungus Candida albicans or the bacteria Salmonella typhimurium, C. rodentium, and Yersinia enterocolitica (131–134). The induction and differentiation of Th17 cells requires TGF-β together with IL-6 or IL-21 (94, 135), whereas the expansion and maintenance of Th17 requires IL-23 (136). Interestingly, both intestinal LP CD103⁺CD11b⁺ and CD103⁻CD11b⁺ cDC2 can induce Th17 cells in vitro (52, 137, 138). However, a specific reduction in CD103⁺CD11b⁺ cDC2 in the small intestinal LP due to DC-specific loss of Notch2 results in a defect in the homeostatic maintenance of small intestinal LP-resident Th17 cells (139). Moreover, intestinal IRF-4–dependent CD103⁺CD11b⁺ cDC2 rather than CD103⁻CD11b⁺ cDC2 are required for the differentiation of homeostatic Th17 in an IL-6–dependent manner in the MLNs. Of importance, when compared side by side in vitro, both CD103⁺CD11b⁺ and CD103⁻CD11b⁺ cDC2 can achieve this, thus the in vivo context is important (23). Finally, selective depletion of CD103⁺CD11b⁺ cDC2 in the small intestinal LP and MLNs of huLangerin-DTA transgenic mice results in a reduction in Th17 cells in the small intestinal LP, and the homeostasis of these Th17 cells was independent of cognate CD4/MHC-II interactions (48). However, other DC may compensate for the lack of CD103⁺CD11b⁺ cDC2 in huLangerin-DTA transgenic mice when it comes to the induction of IL-22 by ILC3 in response to flagellin or infections with C. rodentium or Salmonella enterica (48). This is in contrast to findings in mice that lack Notch2 in cDC2 in which the IL-22 response to C. rodentium is impaired due to a reduction in cDC2 derived IL-23 (21). It is possible that other Notch2-dependent DC subsets could account for this observation or, alternatively, that there are other unknown vivarium-dependent or model-dependent variables that differ between these studies. Nevertheless, it seems clear that, when studied in their native in vivo environment, CD103⁺CD11b⁺ cDC2 rather than CD103⁻CD11b⁺ cDC2 are largely responsible for the differentiation and maintenance of Th17 cells in the MLNs and small intestinal LP. The molecular requirements for activation of CD103⁺CD11b⁺ cDC2 versus CD103⁻CD11b⁺ cDC2 in promoting a Th17 response may be partially dependent on the distinct requirement of the MyD88 signaling pathway. Specifically, induction of Th17 by CD103⁺CD11b⁺ cDC2 is independent of the MyD88 signaling pathway, whereas MyD88 is required for CD103⁻CD11b⁺ cDC2 to promote Th17 cells (140). Understanding the differential requirements of MyD88 and other similar pathways between cDC2 subsets will have important implications for the design of therapies that activate or inhibit Th17 responses.

Human intestinal DC, like DC in mice and in human blood, exhibit superior T cell stimulation capacity compared with macrophages (141). The phenotype, maturation status, and migratory activity of human intestinal cDC was first reported by the Stagg group in 2001 (17). Similar to mouse intestinal cDC, human intestinal cDC can be divided into several subgroups based on the expression of CD103 and SIRPα (23, 142). CD103⁻SIRPα⁺ cDC2 can be further divided based on the expression of CCR2 (137). The characterization of human intestinal cDC is summarized in Table III. Recently, mass cytometry (CyTOF) and unsupervised high-dimensional analyses were used to align DC subsets across human, macaque, and mouse tissues, enabling direct comparison of intestinal DC between different species (8).

Inflammatory bowel disease (IBD) is a chronic inflammatory condition of the gastrointestinal tract encompassing two main clinical entities: Crohn's disease and ulcerative colitis. Both forms of IBD are associated with multiple pathogenic factors including environmental changes, an array of susceptibility gene variants, an abnormal gut microbiota, and a broadly dysregulated immune response (143). In the context of IBD, DC contribute to the inflammatory response through TLR2/4-mediated production of cytokines, such as IL-12, IL-6, and IL-23 (144, 145). In terms of cDC subsets, CD103⁺ cDC (both CD141⁺ and CD1c⁺) are reduced in the inflamed intestine, and their RALDH activity is reduced in ulcerative colitis patients compared with control and Crohn's disease patients (146). Early in Crohn's disease, levels of mesenteric fat and production of leptin are observed to be elevated (147). Leptin may promote cDC maturation by increasing CCR7 surface expression on DC to facilitate their migration into mesenteric fat, suggesting a link between mesenteric obesity and inflammation (148). In contrast, intestinal cDC from Crohn's disease patients that are CD83⁺DC-SIGN⁺ express lower levels of CCR7 and higher levels of TNF-α compared with controls (149). These data suggest that intestinal inflammation disrupts normal DC trafficking patterns and leads to dysregulated T cell responses and tissue damage.

The capacity to generate a potent gut-trophic effector T cell response during a mucosal challenge relies on cDC1 and CD103⁺ cDC2 that can migrate to the MLNs (150). Therefore, vaccination strategies have employed the use of adjuvants that can promote LP cDC1 and CD103⁺ cDC2 migration toward the MLN such as flagellin and oral cholera toxin (151, 152). One study described the use of RA as an adjuvant to boost effector and memory T cell responses in mucosal sites during a viral challenge (153). During an intestinal challenge, it is reasonable to assume that RA can be used as an adjuvant to promote DC-mediated IgA production or promote the homing of intestinal cDC to mucosal LNs. As discussed earlier, there are several factors that can modulate the ability of intestinal cDC to generate Th1, Th2, and Th17 responses and should be considered when selecting an adjuvant during the engineering of a vaccine. Future studies aimed at dissecting the different subsets of cDC responsible for passive immunity will shed further light on vaccine design.

An additional modulator of oral vaccine-induced immunity is age. Neonatal mice exhibit a marked deficit in cDC1 and CD103⁺ cDC2 during the first week of life, perhaps due to the lower production of DC-attracting chemokines by neonatal intestinal epithelial cells. This relative paucity of cDC1 and CD103⁺CD11b⁺ cDC2 in neonatal mice renders them susceptible to Cryptosporidium parvum infection (154). Despite their scarcity, cDC1 can provoke “adult-like” CD8⁺ T cell responses in clearing intestinal viral infections in neonatal mice (84). Understanding the differences in how intestinal cDC initiate immune responses during the neonatal versus adult phases of life and between human and mice can potentially provide insight on the timing of vaccine administration.

The gut is a unique environment where nonresponsiveness to harmless or tolerated Ags needs to be continually maintained while retaining the ability to briskly respond to enteric pathogens. cDC are at the center of this immunological détente due to the different subsets that exist and their accompanying functions, thus providing the immune system with considerable flexibility and nimbleness. With the age of single-cell RNA sequencing upon us, we will undoubtedly learn about new divisions of labor among the already identified cDC subsets. Moreover, we know very little about the stromal cells that underpin the activity of intestinal cDC, but new reporter/depleter mice along with single-cell RNA sequencing application is rapidly shedding light on this question. In the future, therapies that quiet excessive inflammation will go beyond the effector phase of the disease (e.g., blockade of TNF-α) but rather will tackle cDC dysfunction as a means of getting to the route of disease etiopathology.

We apologize to all colleagues whose work was not cited owing to space constraints.