Introduction

Our bodies exist in symbiosis with a multitude of commensal microorganisms, collectively known as the microbiota, which colonize all epithelial and mucosal surfaces, but are particularly abundant in the gastrointestinal tract. In fact, humans are comprised of an estimated1013 human cells1 and 1014 microbial cells of thousands of species,2 which makes the definition of ‘self’ somewhat complicated. Commensal microbiota are essential for the digestion of food and generate essential nutrients and vitamins that are made available to the host.3 The microbiota also have profound effects on the immunological development of the host by stimulating the host immune system to respond more quickly and effectively to pathogen challenge and, via bacterial antagonism, inhibiting the colonization of mucosal surfaces by pathogenic microorganisms.4 Despite the beneficial properties of commensal microorganisms at mucosal surfaces, if they are not properly contained, they can cause local or systemic infections that can lead to chronic or acute inflammation and even death.5 Thus, although commensal colonization is clearly beneficial to the human host, the immune system must use noninflammatory mechanisms efficiently to prevent commensal encroachment into systemic sites, while simultaneously allowing their maintenance at mucosal surfaces.

In addition to providing an interface between commensal microbiota and the host, mucosal surfaces are often sites of pathogen entry. Viral, fungal, parasitic, and bacterial pathogens often colonize mucosal surfaces, use mucosal routes to gain entry to systemic sites, and shed their progeny into mucosal secretions to infect new hosts. Unlike commensals, pathogens are clearly detrimental to the host and the immune system must be capable of detecting them and mounting an immune response that is vigorous enough to clear the infection without unduly compromising the mucosal barrier. To make matters more complicated, the immune response must simultaneously discern the difference between commensals, pathogens, and inert antigens like those from food or airborne particles and respond appropriately,6, 7 as inappropriate responses to nonpathogens may lead to chronic inflammatory conditions, such as inflammatory bowel disease, asthma, or allergies.

Immune responses are facilitated by organized lymphoid tissues,8 which have evolved mechanisms to collect antigens from peripheral and mucosal tissues and efficiently present them to lymphocytes in ways that lead to rapid and robust immune responses to pathogens, as well as noninflammatory or tolerogenic immune responses to commensal organisms and inert antigens. The structure of lymphoid organs maximizes encounters between naive lymphocytes, antigens, and antigen-presenting cells, and provides a microenvironment that sustains the proliferation, differentiation, and survival of activated cells.9 Secondary lymphoid organs are strategically located within the drainage areas of most tissues in the body, including mucosal surfaces such as the respiratory tract, the gastrointestinal tract, and the reproductive tract. Importantly, each lymphoid organ supports immune responses that are appropriate for the tissue that they drain and for the type of antigen that they encounter. Here we will describe the different types of mucosa-associated lymphoid tissues, the pathways by which they differentiate, and how they control mucosal immunity and homeostasis.

Structure of mucosal lymphoid tissues

There are four predominant types of lymphoid organs that serve mucosal sites: encapsulated draining lymph nodes (LNs), mucosal lymphoid tissues, which are directly associated with mucosal tissues, fat-associated lymphoid tissues, and nonprogrammed, ectopically induced, tertiary lymphoid tissues. Each of these tissues has distinct developmental origins and structural characteristics (Figure 1) and each of these tissues performs specialized functions in the regulation of mucosal immunity. The encapsulated LNs, such as the mesenteric LNs that drain the intestines and the cervical and mediastinal LNs that drain the upper and lower respiratory tract, share similar structures. These bean-shaped organs collect antigen and antigen-presenting cells via afferent lymphatics that deliver their contents to the subcapsular sinus on the cortical side of the LN, where they are filtered through a layer of sinus-lining macrophages and dendritic cells (DCs) and percolate through the T-cell zone. B-cell follicles are found around the edge of the cortical region, whereas the T-cell zone is centrally located in the cortex. Lymphocytes enter the LNs through high endothelial venules (HEVs) that are found in the T zone and that are surrounding the B-cell follicles, and then exit the LN through the medullary cords, and subsequently through the efferent lymphatic vessels (Figure 1).

Figure 1
figure 1

Different types of lymphoid organs are present at mucosal sites. (a) Encapsulated lymphoid organs, such as lymph nodes, contain incoming afferent lymphatic vessels through which antigen and antigen-presenting cells enter the lymphoid organs. Efferent lymphatic vessels carry the lymph away from the lymphoid organ. (b) In nonencapsulated lymphoid organs, which are associated with the epithelium, antigen comes in through the follicle-associated epithelium, overlaying the B-cell areas. This epithelial layer contains specialized microfold cells (M cells). Dendritic cells are located right underneath this follicle-associated epithelium, enabling them to take up and process the antigen. Efferent lymphatic vessels are in place to transport lymph away from the lymphoid tissues. (c) Fat-associated lymphoid organs, such as fat-associated lymphoid clusters and milky spots, also lack a capsule. They can obtain antigen by lymphatic drainage directly from the intestine as well as from the peritoneal cavity. (d) Ectopic lymphoid follicles can form in association with the epithelium, as well as perivascular structures, which are away from the epithelium. HEV, high endothelial venules; B, B-cell area; T, T-cell area.

PowerPoint slide

Full size image

In contrast, mucosal lymphoid tissues, such as the Peyer’s patches in the small intestine, colonic patches in the colon,10 isolated lymphoid follicles (ILFs)11 and cryptopatches in both the small intestine and the colon,12 the nasal-associated lymphoid tissue (NALT) in the nasal passages,13 the tonsils and adenoids in the nasopharynx, and the tear duct-associated lymphoid tissue (TALT)14 in the eye, are not encapsulated and are embedded directly in the submucosa of the organs in which they are found. These tissues lack afferent lymphatics and instead collect antigens directly from the mucosal surface across a specialized epithelial layer known as the dome epithelium or the follicle-associated epithelium. This epithelial layer contains specialized epithelial cells termed M cells for the distinct microfold ridges on their apical surface.15 A distinct layer of DCs is found immediately below the dome epithelium16 and many of these cells are enfolded within the basolateral pocket of M cells, where they receive antigens transported from the mucosal surface by the M cells. B-cell follicles fill the dome region of mucosal lymphoid organs and the T zone is typically found to the sides of the B-cell follicles or between follicles (Figure 1). Like encapsulated LNs, the HEVs of mucosal lymphoid organs are found surrounding the B-cell follicles and in the T-cell zone. Also like LNs, mucosal lymphoid organs such as Peyer’s and colonic patches have efferent lymphatics, which are used as a point of egress for activated or recirculating cells. For ILFs, however, it has not been established whether efferent lymphatic vessels are in place to perform such a function.

The fat-associated lymphoid tissues are distinct from both LNs and classic mucosal lymphoid tissues in that they are not encapsulated and are not directly under a mucosal epithelium. Instead, the fat-associated lymphoid clusters (FALCs)17of the mesentery and the milky spots of the omentum18 are embedded in the adipose tissue. Although the FALCs may receive antigens and cells via the lymphatic drainage from the intestines,17 they are also able to acquire antigens directly from the peritoneal cavity via fenestrations in the mesothelial layer that overlays the mesentery (Figure 1). Similarly, the milky spots of the omentum acquire antigens and cells exclusively from the peritoneal cavity and appear to be a major site through which the peritoneal fluid is collected.19 Milky spots are dominated by large central clusters of B cells, but as they seem to lack follicular DC (FDC) networks in the B-cell area,18 they do not have B-cell follicles per se. In contrast, the T-cell zone in milky spots is somewhat limited and seems to surround the B-cell areas. HEVs are found around the follicles of milky spots, although unlike HEVs of Peyer’s patches and mesenteric LNs, the HEVs of milky spots express peripheral LN addressin and not mucosal addressin cell adhesion molecule.18 FALCs have even less of an organized structure than milky spots, but are similarly dominated by B1 cells, consistent with the idea that cells in the peritoneal cavity can easily translocate to these tissues.

Unlike the lymphoid tissues mentioned above, all of which develop their initial anlage independently of antigen or microbial colonization, ectopic lymphoid follicles, such as bronchus-associated lymphoid tissue (BALT) or gastric-associated lymphoid tissue, are formed after birth in response to inflammation or infection.20 Nevertheless, the structure of ectopic lymphoid tissues mimics that of conventional lymphoid tissue in many ways. For example, BALT often has large central B-cell follicles and separated T-cell areas, specialized stromal cells as well as HEVs and lymphatics (Figure 1). In some cases, well-developed BALT areas have a dome epithelium with identifiable M cells, similar to those in the Peyer’s patches or NALT. However, BALT may also be found in perivascular areas of the lung and lack a dome epithelium. Moreover, BALT can sometimes be observed with a well-developed B-cell follicle and even germinal centers, but lack a defined T-cell area. In fact, a wide range of ectopic lymphoid structures can be observed, from loose clusters of B cells to large multifollicular structures. However, unlike ILFs, which develop from cryptopatches at defined sites, ectopic lymphoid tissues seem to develop in a wide variety of locations from precursor structures that are not yet defined.

Mucosal lymphoid organ development

Lymphoid organ development requires reciprocal interactions between a variety of lymphoid tissue initiator or inducer (LTi) cells and lymphoid tissue organizer or stromal cells.21 LTi cells are innate lymphoid cells (ILCs) that require the transcription factors, RAR-related orphan receptor γt (RORγt), and Id2, and express CXC-chemokine receptor 5 (CXCR5) and CCR7, the receptors for CXC-chemokine ligand 13 (CXCL13) and CCL19/CCL21, respectively.22, 23, 24, 25 Importantly, LTi cells express lymphotoxin,26 which triggers the nuclear factor-κB signaling pathway in lymphotoxin β receptor-expressing lymphoid tissue organizer cells, and leads to the enhanced expression of homeostatic chemokines and cytokines, including CXCL13, CCL19, CCL21, interleukin-7 (IL-7), and RANK (receptor activator of NF-κB) ligand, which in turn attract and support the survival of LTi cells.27, 28 Ultimately, the lymphoid tissue organizer cells differentiate into the various stromal cell types, such as fibroblastic reticular cells29and FDCs,30 that form the scaffolding of the lymphoid tissue and help to compartmentalize lymphocytes and DCs into their appropriate anatomic locations. Once lymphoid tissues are formed, lymphotoxin continues to be expressed by adaptive lymphocytes, particularly B cells,31 which are important for maintaining the lymphotoxin:chemokine feedback loop, and as a result, the structure of the lymphoid organ.

Although individual mucosal lymphoid tissues develop following the same general program, each has unique characteristics. For example, the development of secondary lymphoid organs is temporally regulated—mesenteric LNs in mice begin to develop on embryonic day E9–10, brachial LNs on E13, axillary LNs on E15, inguinal LNs on E16, and popliteal LNs on E17.32 In contrast, Peyer’s patches are still developing up until birth32 and the TALT and NALT develop after birth.14, 33 The cryptopatches, ILFs, and milky spots of the omentum also develop after birth and are partially dependent on the presence of microbiota to attain their full maturation.34, 35 In addition to the developmentally programmed lymphoid organs, ectopic lymphoid tissues like BALT only develop following infection or inflammation.36, 37, 38

As described above, the lymphotoxin-driven expression of homeostatic chemokines like CXCL13 is important for the development of essentially all lymphoid tissues.39 However, submucosal lymphoid tissues like Peyer’s patches, ILFs, and NALT express additional homeostatic chemokines, such as CCL20 and CCL9,40 that attract populations of DCs and B cells to the subepithelial dome region. Like other homeostatic chemokines, CCL20 expression in the dome epithelium requires lymphotoxin signaling.41 The importance of these chemokines is illustrated in mice lacking CCR6, the receptor for CCL20, which is required on B cells for the proper maturation of ILFs from cryptopatches in the small intestine.42, 43 CCR6 is also expressed on putative LTi cells in cryptopatches,44 although CCR6 expression on LTi cells is not critical for cryptopatch formation.34, 35, 44 Interestingly, the requirement for the various chemokines in ILF formation is distinct between the small intestine and the colon; in other words, ILFs do not form in the absence of CCL20 or CXCL13 in the small intestine, but they are present in the colon of mice in which the CCL20 or CXCL13 signaling axis is disrupted, although with a disturbed architecture in CXCL13-deficient mice.34, 35 There is also a difference between tissues in the gut and the respiratory tract, as NALT development is not markedly disrupted in CCR6-deficient mice (TDR unpublished).

Role of LTi cells

Although the development of peripheral LNs is entirely dependent on the activities of LTi cells and lymphotoxin, many of the mucosal lymphoid tissues develop in the apparent absence of this pathway (Table 1). For example, NALT and TALT both develop independently of lymphotoxin, albeit with disrupted architecture.14, 33, 45 Furthermore, both TALT and NALT develop in RORγt-deficient mice, which lack the archetypical LTi cells, suggesting that the development of these tissues does not involve LTi cells.14, 33, 45 However, the development of NALT, but not TALT, does depend on Id2.14, 33 Given that Id2 is required for the differentiation of all ILCs, another ILC family member could be responsible for the inductive signals required for NALT formation46 (Table 1). Interestingly, although all ILCs require Id2, only the LTi-containing ILC3 subset requires RORγt, which promotes the expression of cytokines like IL-17 and IL-22.46 In contrast, other ILC family members, such as ILC1 or ILC2 cells, express interferon-γ or IL-5/IL-13, respectively.46 Although the involvement of these cytokines in lymphoid tissue formation has not been documented, the development of FALCs depends on natural helper cells (ILC2 cells), which produce IL-5 and IL-13.17

Table 1 Requirements for the formation of mucosal lymphoid tissues
Full size table

Another fat-associated lymphoid tissue, the milky spots of the omentum, also develops in the absence of lymphotoxin,18 although they do so with some structural defects. Their formation, however, does not require LTi cells, as milky spots are formed in both RORγt-deficient as well as Id2-deficient mice,18 suggesting that ILCs may not be involved at all (Table 1). Given that milky spots develop after birth at a time when adaptive lymphocytes are plentiful, it is likely that B cells, particularly B1 cells, promote their differentiation.47 Intriguingly, although they do posses a stromal cell network, milky spots do not have identifiable FDCs,18 suggesting that the functions of both LTi cells and lymphoid tissue organizer cells may be performed by different cell types during milky spot differentiation and may account for their unique developmental and functional characteristics.

Ectopic follicles in both the lung and the gut also appear to develop independently of LTi cells, although not necessarily independent of lymphotoxin.36, 48 For example, although RORγt-deficient mice lack LNs, Peyer’s patches, ILFs, and even cryptopatches owing to the loss of LTi cells, these mice develop ectopic follicles in the submucosa of the intestine.48 Interestingly, B cells seem functionally to replace LTi cells in this context and serve as an essential source of lymphotoxin that promotes the development of the microbiota-driven formation of ectopic follicles.48 Similarly, the ectopic development of BALT occurs independently of both RORγt and Id2,36 implying that LTi cells are not involved (Table 1). Nevertheless, lymphotoxin is important for the organization of BALT, although perhaps not its development. Mice lacking lymphotoxin or the lymphotoxin β receptor often develop extensive lymphoid aggregates in the lungs in the same locations one would normally observe BALT.49 However, these lymphoid aggregates fail to segregate B and T cells, do not form identifiable FDCs and fail to develop HEVs,49 suggesting that many of the architectural features of BALT are dependent on lymphotoxin, even though the initial expression of CXCL13 and the recruitment of lymphocytes can occur in its absence. Moreover, even though CXCL13 is required for the development of most lymphoid organs,39 mice lacking CXCL13 still form BALT with segregated B- and T-cell areas,50 although the B-cell areas lack FDCs (Table 1). Nevertheless, CXCL13 is important for the development of BALT, as the combined loss of CXCL13, CCL19, and CCL21 completely abrogates BALT development, whereas the loss of either CXCL13 or CCL19/CCL21 only leads to minor structural defects.50

Role of microbiota, infection, and inflammation in mucosal lymphoid organ development

Most lymphoid organs develop during embryogenesis in a sterile environment. However, some mucosal lymphoid organs complete or even initiate their development after birth and are markedly influenced by exposure to commensal microbiota.51 For example, Peyer’s patches, NALT, TALT, and cryptopatches all develop in germ-free mice.14, 21, 34 However, the cryptopatches mature in response to microbial colonization of the gut and accumulate B cells and ultimately transform into ILFs.12, 35 The requirements for the transition from cryptopatches to ILFs are different between the small and the large intestine: counterintuitively the maturation of ILFs in the small intestine depends on the microbiota, whereas the maturation of ILFs in the colon, where the microbial density should be the highest, occurs in a microbiota-independent, yet Myd88-dependent manner34, 35 (Table 1). Similarly, environmental exposure helps the maturation of NALT.52 In part, exposure to antigen causes some level of inflammation that leads to immune reactivity and expands germinal centers, just as it would in conventional lymphoid organs. However, microbial triggers also promote M-cell differentiation and maturation in both the gut53, 54 and the airways.55, 56 Given the interactions between M and B cells, it is not surprising that microbial exposure leads to the accumulation of B cells, the differentiation of a dome epithelium, and ultimately to the maturation of mucosal lymphoid tissues.

The microbiota also has an important role in the maturation of FALCs in the mesentery and milky spots in the omentum. The number and size of FALCs and milky spots are reduced in germ-free mice and their cellular composition is altered,57 suggesting that despite their placement in the peritoneal cavity, which under steady-state conditions should not be colonized with microbes, these sites are exposed to gut-associated commensals or their products. Moreover, upon oral administration of indomethacin or dextran sulfate sodium, both of which perturb the intestinal epithelium and lead to inflammation and leakage of intestinal contents, the number and size of milky spots is markedly increased.58 Acute peritoneal inflammation with sterile stimuli-like LPS also causes an immediate activation of peritoneal macrophages and B1 cells, which rapidly home to the milky spots.58 Even sterile peritoneal dialysis leads to increases in the number and size of the milky spots.19 Similarly, the FALCs are stimulated upon infection with N. brasiliensis, which leads to ILC2 activation, IL-5 and IL-13 expression, and parasite expulsion.17 Thus, the fat-associated lymphoid tissues in the peritoneal cavity are highly sensitive to both steady-state and inflammatory conditions in the intestinal lumen.

Persistent or repeated damage to the mucosal epithelium can allow microbes and their products to contact the underlying tissue, resulting in chronic inflammation or infection and leading to the development of ectopic or tertiary lymphoid tissues. For example, BALT can form in response to chronic inflammation, such as cigarette smoke,59, 60 exposure to allergens61 or autoantigens,62, 63 idiopathic pulmonary arterial hypertension,64 or following infection.37, 38, 65, 66, 67 A common feature of ectopic lymphoid tissue development is the expression of CXCL13, which is often not expressed in peripheral nonlymphoid organs, but is markedly increased at sites of tertiary lymphoid tissue development.65 Although CXCL13 expression in conventional lymphoid organs is controlled by retinoic acid signaling25and homeostatic signaling through the lymphotoxin β receptor, the expression of CXCL13 in the lungs can also be acutely triggered by IL-17,36, 64, 68 independently of lymphotoxin or retinoic acid. Although LTi cells express IL-17, the important source of IL-17 in BALT formation is initially γδT cells followed by αβT cells with a T follicular helper phenotype.36, 69 However, once enough activated B and T cells accumulate at the appropriate sites in the lung, they maintain CXCL13 expression independently of IL-17 via their homeostatic production of lymphotoxin, which is essential to maintain BALT structure.36, 37 BALT development has a similar requirement for CXCR5-expressing, IL-17-producing T cells following vaccination and challenge with Mycobacterium tuberculosis.69, 70 Ectopic lymphoid tissues in other locations, such as those in the gastric mucosa of the stomach upon Helicobacter pylori infection71, 72 and those in the small intestine following mucosal irritation or failure to contain commensal microbiota,48, 73 have similar requirements for CXCL13 and lymphotoxin.71, 72, 74 However, the initial inflammatory trigger that leads to the development of these ectopic tissues is not known. Given that IL-17- and IL-22-producing helper T cells are essential for protective immunity against Helicobacter,75 these cytokines may be involved.

Although IL-17 is clearly involved in the formation of ectopic follicles under some circumstances,36, 64, 68 it is not essential in all cases, as BALT and ectopic follicles in the gut both develop in RORγt-deficient mice,36, 48 which have essentially no capacity to produce IL-17. Moreover, BALT formation can be induced in adults in the absence of IL-17a and IL-17f by pulmonary infection with modified vaccinia Ankara virus.76 However, it is currently unclear which pathways are involved. Nevertheless, the role of DCs, which are required for the maintenance of BALT,37 should not be overlooked, as these cells are not associated with the development of conventional lymphoid organs, but are important for BALT formation. Finally, signaling via other innate inflammatory cytokines, such as IL-1, are also likely to be involved in the formation of ectopic follicles, as these cytokines are important in diseases like chronic obstructive pulmonary disease (COPD),77, 78 which feature the development of ectopic follicles. Thus, both microbial and inflammatory stimuli promote BALT formation via a variety of pathways that are not used in the development of conventional lymphoid organs.

Antigen acquisition in mucosal lymphoid tissues

The acquisition of antigen is essential for the function of any lymphoid organ. Conventional LNs acquire antigen through afferent lymphatics, which drain local nonlymphoid tissues and collect antigens and antigen-presenting cells that are activated by invading pathogens or even commensal organisms that have penetrated the mucosal barrier and elicited a local inflammatory response. In fact, the lamina propria of the mucosal epithelium is replete with tissue DCs that, when activated by microbial stimuli, rapidly migrate to the mesenteric LNs (from the gut), the mediastinal LNs (from the lower respiratory tract), or the cervical LNs (from the upper respiratory tract).7, 79 The migratory DCs include both the CD103+ DCs and the CD11bhi DCs, which reside in different locations and perform different functions,80, 81, 82, 83 depending on the type of antigen, the route of exposure, and the extent of inflammation.84

LNs that drain mucosal sites also have access to lumenal antigens under steady-state conditions, which requires mechanisms for transporting antigens across the mucosal epithelium. One such mechanism involves goblet cell-associated antigen passages or GAPs, which acquire soluble luminal antigens and transports them to underlying DCs, particularly CD103+ DCs.85 Given that CD103+ DCs are associated with the induction of immunological tolerance under steady-state conditions,86, 87 GAPs may be important for the maintenance of tolerance to food antigens. Consistent with this idea, the function of GAPs is linked to the secretory capacity of goblet cells85 and defects in goblet cells are linked to intestinal inflammation.88, 89 GAPS are observed in the small intestine, but not in the stomach, cecum, or colon, and have not yet been examined in the respiratory tract.85

In addition to the active sampling of luminal contents by specialized epithelial cells, some DCs can directly sample antigens by extending dendrites between epithelial cells. For example, CX3CR1-expressing DCs in the intestinal villi project dendrites between epithelial cells into the intestinal lumen and thereby capture antigens.90 CX3CR1 is necessary for these DCs to extend dendrites into the lumen and for bacterial uptake.91 These CX3CR1-expressing DCs are often perceived as nonmigratory92 and are suggested to represent immune effectors rather than immune inducers.93 However, recent data suggest that commensal microbiota actually restrain the migratory ability of CX3CR1+ DCs, but upon inflammation, these cells express CCR7, traffic antigens and bacteria to the LN and trigger T- and B-cell responses.94

M cells are another mechanism for antigen transport across the mucosal epithelium. Although M cells are most often associated with a dome epithelium overlaying mucosal lymphoid tissues,15, 95, 96 they are also found in the normal villi of the small intestine97 and the respiratory epithelium.98Unlike GAPs, which are limited to sampling low-molecular-weight antigens, M cells efficiently transcytose particulate antigens and even whole bacteria98—a process that is facilitated by a variety of receptors, including glycoprotein 2,54 uromodulin,99 and annexin V,100 all of which bind to bacterial components. In many cases, the organisms transported by M cells are commensal organisms that are part of the normal microbiota. However, some pathogens, including mouse mammary tumor virus,101 group A streptococcus,102 and Burkholderiapseudomallei,103 have taken advantage of M cells to hitch a ride across the epithelium. M cells have a basolateral pocket, which is often filled with a variety of hematopoietic cells, including CCR6-expressing B cells and DCs,43 which are poised to receive transported antigens. CX3CR1-expressing DCs are also found in the basolateral pocket of M cells and can extend dendrites through pores in M cells to sample directly antigens overlaying the dome epithelium of Peyer’s patches.104

Unlike antigen acquisition in LNs and mucosal lymphoid tissues, antigen acquisition in ectopic lymphoid tissues is poorly understood. Tissues like BALT are initially described as classic mucosal lymphoid organs with a dome epithelium and M cells.105, 106, 107, 108 However, BALT does not always have a dome epithelium and some areas of BALT are perivascular or are found in the lower airspaces that lack an association with either airways or arterioles.109 Moreover, ectopic lymphoid tissues in the intestine and gastric mucosa also lack a dome epithelium and are embedded in the tissue itself. Thus, these tissues must acquire antigens via alternative mechanisms.

One possible way in which ectopic lymphoid tissues collect antigens is via afferent lymphatics. Lymphatic vessels are observed surrounding ectopic follicles.20, 63 However, it is not clear whether these are afferent lymphatics that bring antigens and cells from distal regions or whether they are efferent lymphatics that return lymphocytes to the circulation. Histological evidence of DCs and macrophages, filled with diesel exhaust particles,110 silica particles,111 cigarette smoke particles,112 and other antigens within BALT,38 suggests that these cells may have acquired antigens and then migrate into BALT. Consistent with this possibility, live imaging of BALT shows that DCs can migrate from the lung airways into BALT,38 demonstrating that DCs can enter BALT via afferent lymphatics or directly across the epithelium.38 However, in many cases, ectopic lymphoid tissues seem to form directly surrounding pathogens like Helicobacter pylori or Mycobacterium tuberculosis,63, 66, 67, 68 in which case the antigen is already present inside the follicle. Thus, there may be multiple ways for ectopic lymphoid tissues to acquire antigen.

If pathogens, commensals, or even microbial products escape from the gastrointestinal tract into the peritoneal cavity owing to intestinal trauma or inflammation, they will be collected by the milky spots of the omentum or the FALCs.17, 19 These lymphoid clusters are open to the peritoneal cavity via fenestrations in the overlaying mesothelium.113 Peritoneal fluid is continually flowing from the peritoneal cavity through the milky spots and FALCs,19 which capture cell-free antigens and particles, as well as a wide variety of cell types that may be in the peritoneal cavity. For example, upon activation, peritoneal macrophages rapidly migrate to the milky spots and carry with them any organisms or antigens they have engulfed.58 Although peritoneal macrophages are unlikely to be a primary antigen-presenting cell, the milky spots also have a variety of DCs that are capable of priming and cross-priming T cells. B1 cells are also abundant in the peritoneal cavity and, upon activation by microbial signals, these cells also rapidly migrate to the milky spots.47, 58 Given the dependence of milky spot development on microbial colonization in the gut,57 it seems that they are probably continuously exposed to microbial products even under steady-state conditions.

Mucosal IgA responses

Considering the numerous ways in which mucosal antigens can be sampled, it is difficult to ascribe a particular function or activity to any individual type of mucosal lymphoid organ. However, the generation of B cells that have isotype-switched to immunoglobulin A (IgA) is a hallmark feature of mucosal tissues, of which Peyer’s patches serve as the prototypical model.114 IgA-expressing B cells dominate germinal centers in Peyer’s patches and the enzyme, activation-induced cytidine deaminase, which is required for isotype switching,115 is expressed in those cells as well.116, 117 Moreover, IgA switch circles, which are a by-product of isotype switching to IgA118 and postswitch IgA circle transcripts,119 are also abundant in Peyer’s patch B cells.

Despite their undisputed ability to promote the differentiation of IgA-expressing B cells, Peyer’s patches are not absolutely necessary for the production of IgA in the gastrointestinal tract, as ILFs can make considerable amounts of IgA in mice that lack Peyer’s patches and LNs.120, 121, 122 One important difference is that the generation of IgA-producing B cells is dependent on T cells in the Peyer’s patches, while this is a T-cell-independent process in ILFs.120 Switching to IgA may also occur in the lamina propria itself outside the organized lymphoid tissues.118 However, this observation is controversial,116, 117 owing to the difficulty of isolating lamina propria in the absence of any contaminating ILFs.

The milky spots are also proposed to contribute to intestinal IgA production. Both the peritoneal cavity and the omentum are filled with B1 cells, which efficiently switch to IgA in vitro when stimulated with microbial products and cytokines like tumor growth factor-β.58 Although B1 cells are unlikely to contribute to IgA elicited by T-dependent antigens, they may well contribute to “natural” IgA. Consistent with this idea, B1 cells also preferentially home to the peritoneal cavity and the small intestine,58, 123 implying that they may contribute to mucosal IgA responses. However, other studies demonstrate that B1 cells in the peritoneal cavity are primarily responsible for the production of natural IgM, but contribute only minimally to intestinal IgA under steady-state conditions.124

By analogy to canonical mucosal lymphoid organs like Peyer’s patch, one might expect that NALT would also promote the differentiation of IgA-producing B cells. While nasal immunization using cholera toxin as an adjuvant typically elicits IgA-secreting B cells in the NALT,125, 126, 127, 128, 129 nasal immunization with protein-containing microparticles primarily elicits IgG responses,130 which appear in the NALT before their appearance in cervical LNs. In contrast, nasal infections with viruses131, 132, 133 result in mixed B-cell responses consisting of both IgG- and IgA-secreting cells in the NALT. Nevertheless, nasal influenza infection leads to higher frequencies of IgA B cells in NALT than in mediastinal LNs, lung, or spleen.134, 135, 136 Moreover, NALT selectively promotes the development of high-affinity IgA memory B cells over IgG memory cells.137 Thus, although NALT is not exclusively biased toward IgA production, IgA is favored over IgG in the memory compartment. However, NALT is not absolutely necessary for adequate immune responses, as in its absence, nasal immune responses to influenza and protection against pneumococcal antigens can still occur.138,139 In contrast, cervical LNs are important to achieve nasal immune responses of a normal magnitude.139

Immunity in ectopic lymphoid tissues

Ectopic lymphoid tissues, like those in the lung (BALT) or those in the stomach or the gut also have a role in local immune responses. Given that these tissues are responding to mucosal antigens, one might expect that isotype switching in these sites would primarily result in IgA-producing B cells. However, IgG-secreting cells often predominate over IgA-secreting cells in BALT.106, 140 Nevertheless, IgA is produced in BALT under a variety of conditions,141 including after influenza infection.37 Moreover, BALT seems to generate IgA-secreting plasma cells that home to the lung as well as IgG-secreting plasma cells that home to the bone marrow.37 Importantly, BALT generate the long-lived isotype-switched plasma cells and maintain virus-neutralizing antibodies in the serum and bronchial lavage fluid that can completely prevent infection following a secondary challenge.142 Isotype switching in BALT is not restricted to IgG and IgA, as IgE is often found in BALT areas of patients with allergies143 or in experimental animals that have been sensitized to antigens under type 2 helper T cell-inducing conditions.144, 145 Ectopic follicles at other mucosal sites, including those in the small intestine responding to microbiota and those in the gastric mucosal responding to Helicobacter pylori, also preferentially promote IgG-producing rather than IgA-producing B cells, suggesting that immune responses in these tissues have more characteristics of systemic responses than mucosal responses.48, 71 Consistent with this idea, peripheral LN addressin is commonly expressed by HEVs in ectopic follicles, such as BALT, whereas mucosal addressin cell adhesion molecule is rarely observed.

As noted above, it is difficult to assign a particular immune function to specific lymphoid tissues. However, mice lacking spleen, LNs, and Peyer’s patches generate normal pulmonary B- and T-cell responses to influenza,65, 142, 146 suggesting that BALT is sufficient to initiate and expand some types of pulmonary immune responses. Similar studies using splenectomized lymphotoxin-deficient mice, which also lack LNs and Peyer’s patches, show that immune responses to pulmonary infection with Mycobacterium tuberculosis147, 148 and Leishmania major antigens149 are intact in the absence of conventional lymphoid organs and that antigen presentation occurs directly in the lung. More recent studies directly demonstrate that BALT can recruit and prime naive T cells.147 Importantly, T-cell priming does not occur if BALT is not induced previously.147 Similar studies show that following BALT formation, adoptively transferred ovalbumin-specific T cells are recruited to BALT and proliferate locally in response to intratracheally administered ovalbumin-pulsed DCs38 and long-term B-cell responses to influenza are initiated in BALT.37 Memory T cells are also maintained in BALT without support from other secondary lymphoid organs and can be locally reactivated,142 demonstrating that BALT supports both primary and secondary immune responses.

Ectopic follicles and pathology

Immune responses in ectopic follicles are not always beneficial. For example, although the presence of BALT does not correlate with either respiratory allergy or asthma,150, 151 the reactivity of BALT does increase in asthma patients.152 In addition, exposure to allergens, such as Aspergillus fumigatus, can lead to an asthma-like disease that is accompanied by the development of BALT with prominent IgE-expressing germinal centers.143 Similar results, including the formation of IgE-expressing germinal centers, are observed in mice following sensitization and pulmonary challenge with OVA.144

BALT development can also be induced by inflammatory stimuli like cigarette smoke.153 Moreover, smoking exacerbates BALT reactivity154 and promotes germinal center formation.152 Given the link between smoking and COPD, it is not surprising that BALT formation is most often observed in patients classified with GOLD (global initiative for chronic obstructive lung disease) stage 3 and stage 4 disease59 and strongly correlates with impaired lung function,59 suggesting the possibility that BALT has an important role in COPD pathogenesis. The germinal centers in the BALT of COPD patients have evidence of clonal B-cell proliferation and ongoing somatic hypermutation.112 In opposition to the idea that B cells and BALT contribute to the pathology of COPD, other studies show that exposure to cigarette smoke leads to the development of pulmonary emphysema to the same extent in normal and SCID mice, even though SCID mice completely lack B cells and BALT.155 Thus, although exposure to cigarette smoke promotes the formation of BALT in experimental model systems and BALT is associated with severe COPD in human patients, the causal link between BALT and COPD remains nebulous and it is unclear whether BALT is beneficial, harmful, or neutral to the progression of COPD.156

A subset of patients with rheumatoid arthritis (RA) and Sjogren’s syndrome develop pulmonary symptoms, including the formation of BALT.62, 63, 157, 158, 159 BALT in RA and Sjogren’s syndrome patients appears to contribute to local disease, as it often contains plasma cells producing IgG specific for citrullinated proteins and rheumatoid factor.63 Moreover, citrullinated proteins are observed in BALT areas of the lungs of RA patients,160 suggesting that autoantigens are located inside BALT areas. Interestingly, there is a strong association between RA and smoking,161 which may be explained by the ability of cigarette smoke to elicit the enzymes responsible for the citrullination process.

Although BALT is correlated with pathology in RA patients, it does not always exacerbate pathology in mouse models and even reduces pulmonary pathology in models of fibrosis,162 and infectious diseases,65, 163 suggesting that the presence of BALT actually reduces pulmonary pathology. Thus, it is not clear whether the BALT found in patients with RA is exacerbating pulmonary pathology or may actually be preventing more severe disease. Moreover, BALT sporadically forms in the vicinity of lung tumors,63 including nonsmall-cell lung carcinoma.164 In these patients, the development of BALT is associated with a better long-term prognosis.164

Summary and concluding remarks

Immunity at mucosal surfaces is regulated by a variety of mucosal lymphoid tissues and organs that cooperate to generate immune responses to pathogens, contain commensal microbiota, and prevent unwanted inflammatory responses to food and other innocuous antigens. The functions of these lymphoid tissues are linked to both their development and their interactions with both microbiota and pathogens. As a result, our understanding of infectious diseases that target mucosal tissues or inflammatory diseases that occur as a result of dysregulated immunity in mucosal sites will require a further understanding of how these tissues sense and respond to the mucosal environment.