Role of lymphotoxin and homeostatic chemokines in the development and function of local lymphoid tissues in the respiratory tract

Summary

Secondary lymphoid organs are strategically placed to recruit locally activated antigen presenting cells (APCs) as well as naïve, recirculating T and B cells. The structure of secondary lymphoid organs - separated B and T zones, populations of specialized stromal cells, high endothelial venules and lymphatic vessles - has also evolved to maximize encounters between APCs and lymphocytes and to facilitate the expansion and differentiation of antigen-stimulated T and B cells. Many of the general mechanisms that govern the development and organization of secondary lymphoid organs have been identified over the last decade. However, the specific cellular and molecular interactions involved in the development and organization of each secondary lymphoid organ are slightly different and probably reflect the cell types available at that time and location. Here we review the mechanisms involved in the development, organization and function of local lymphoid tissues in the respiratory tract, including Nasal Associated Lymphoid Tissue (NALT) and inducible Bronchus Associated Lymphoid Tissue (iBALT).

INTRODUCTION

The respiratory tract is constantly exposed to inhaled environmental antigens and pathogens. In turn, the immune system must respond appropriately to eliminate pathogens without triggering excessive inflammatory responses that may damage the lung or permanently alter normal breathing physiology. The respiratory tract uses a variety of mechanisms and tissues, including specialized lymphoid tissues in the nose and lung, to detect inhaled antigens, promote pathogen clearance and limit the magnitude of local immune responses. Two important tissues that are strategically located to confer protection against aerial pathogens are the Nasal Associated Lymphoid Tissue (NALT) and the Bronchus Associated Lymphoid Tissue (BALT). The cellular organization of NALT and BALT is similar to that of conventional lymphoid organs, such as lymph nodes (LNs) and Peyer’s patches, suggesting that the molecular mechanisms involved in maintaining the architecture of these organs are shared with other mucosal lymphoid tissues. However, the development of NALT and BALT differs substantially from that of both LNs and Peyer’s patches, suggesting that these organs may also use unique mechanisms to initiate immune responses that protect the respiratory tract.

NALT

Structure of NALT

NALT was first identified in rodent species as a paired mucosal lymphoid organ, located on the roof of the soft palate just at the entrance of the pharyngeal duct (1–3). NALT is structurally similar to and was thought to be the rodent equivalent of Waldeyer’s ring in humans, which includes the palatine tonsil. However a mucosal lymphoid tissue more similar to NALT was recently identified in the nasal passages of humans (4). Thus, NALT in rodents probably serves the same function as several mucosal lymphoid tissues in the nasal pharyngeal passages of humans.

In cross-section, NALT appears to contain a single B cell follicle surrounded by a smaller T cell zone directly underneath the nasal epithelium (5, 6). However NALT is actually composed of a series of B cell follicles that extend along the nasal passages. These follicles are surrounded and separated by interfollicular T cell areas. Like other mucosal lymphoid organs, NALT is covered by a follicle-associated epithelium (FAE) that contains specialized membranous (M) cells that sample antigens and pathogens in the lumen of the nasal passages (7–10). In fact, several pathogens, including group A Streptococcus (7, 11), use M cells to gain entry into the body. The FAE of NALT consists of ciliated cells intermixed with clusters or individual M cells and a few goblet cells (3, 8, 12). In addition, a population of subepithelial dendritic cells (DCs) is positioned directly beneath the FAE of NALT (5, 13), where they are poised to receive antigens and pathogens that are transported by M cells from the lumen of the nasal passages. As defined in other mucosal lymphoid tissues (6, 14–17), these subepithelial DCs are often in direct contact with the M cells in an M cell “pocket”. Although the DC populations in NALT have not been rigorously characterized, the T cell zone contains CD11c⁺ as well as CD11b⁺Ia⁺ DCs (18) and there is a population of CD11c⁺ subepithelial DCs (5, 13). Thus, NALT has the ability to sample antigens from the nasal passages and initiate immune responses.

The B cell follicles in the NALT of naïve mice primarily contain IgD⁺IgM^lo follicular B cells centered on networks of follicular dendritic cells (FDCs)(5, 13, 19). However, NALT often contains germinal center B cells after intranasal immunization or infection (13, 19–22). Like other mucosal lymphoid organs, the B cells in NALT predominantly switch to IgA (23–27), which is an ideal isotype to protect mucosal surfaces due to its ability to bind the polymeric Ig receptor and to be transported across the mucosal epithelium (28, 29). The ability to promote IgA responses is ideal for intranasal vaccines against mucosal pathogens. For example, live attenuated influenza vaccines administered intranasally are highly effective, yet elicit relatively little neutralizing IgG in the serum (30–33). Instead they are thought to elicit primarily IgA responses in the upper respiratory tract.

T cells in NALT are predominantly CD4 T cells (6) and some CD8 cells, including CD8αβ⁺ and CD8αα⁺ T cells (6) as well as CD8γδ⁺ T cells (34). NALT also contains regulatory T cells that express toll like receptor 2 (TLR2)(35). These cells are thought to maintain tolerance to innocuous inhaled antigens, but are repressed in the presence of TLR2 agonists to allow responses to pathogens (35). Depending on the types of adjuvants used with intranasal antigens, NALT can support the differentiation of naïve CD4 T cells into Th0, Th1 or Th2 cells. For example, bacterial cell wall components administered together with cholera toxin typically induce type 2 responses (18, 25, 36), which promote IgA production in nasal passages as well as in distal mucosal sites, including the respiratory, intestinal and genito-urinary tracts (37, 38). In contrast, intranasal vaccination with Mycobacterium bovis bacillus Calmette-Guerin (rBCG) expressing foreign antigens induces a classic Th1 polarization (39).

Despite the obvious similarities of NALT with other mucosal lymphoid organs, particularly Peyer’s patches (6, 14), the high endothelial venules (HEVs) in NALT primarily express the peripheral lymph node addressin (PNAd) rather than Mucosal Addressin Cell Adhesion Molecule (MAdCAM)(40), which is expressed on HEVs in Peyer’s patches and mesenteric lymph nodes (41, 42). In fact, T cells primarily use CD62L (the receptor for PNAd) to home to NALT (40). This raises the question of whether the function of NALT can truly be extrapolated based on our knowledge of Peyer’s patches.

Another interesting issue is the excellent communication between NALT and distal mucosal sites. Several studies have shown that vaccination through the intranasal route elicits antigen-specific IgA responses in the pulmonary and vaginal tract (43, 44). This extraordinary connection between mucosal sites is due to the characteristic expression of CCR10 and α₄β₁ integrin on IgA-producing cells that are attracted to sites in the respiratory and genito-urinary tract by the local production of the corresponding ligands, CCL28 and VCAM-1 (45, 46). The homing properties of T and B cells in Peyer’s patches are imprinted by Peyer’s patch DCs (47, 48) and it is likely that DCs in NALT will have similar functions. This makes intranasal vaccination an ideal way to elicit immunity to mucosal pathogens.

Development of NALT

The development of most LNs and Peyer’s patches is dependent on the lymphotoxin (LT) signaling pathway (49–51). Despite this, NALT develops in the absence of LTα, LTβ and TNFα as well as the receptors TNFR1 and LTβR (5, 52). Moreover, NALT develops normally in mice treated in utero with both soluble LTβR and soluble TNFR1 (52), even though these mice do not develop any LNs or Peyer’s patches (53). These data argue that none of the ligands that bind to either the LTβR or the TNFR1, including the LTα homotrimer, the LTαβ heterotrimer, TNFα or LIGHT, are essential for NALT development.

TNFR and LTβR are coupled to the NFκB signaling pathway (54) and as a result, animals that have defects in the NFkB pathway also have defects in the development of secondary lymphoid organs (55). However, mice with mutations in the NFkB signaling pathway, including aly/aly, Nfkb1^−/−, Nfkb2^−/− and Relb^−/− mice, all have NALT to some degree (52, 55). These data suggest that the NFkB signaling pathway is also not essential for the development of NALT, even though it is essential for the development of LNs and Peyer’s patches.

The current model for LN and Peyer’s patch development maintains that Lymphoid Tissue inducer (LTi) cells are instrumental for the initiation of lymphoid organ development (56–58). LTi cells are derived from the fetal liver and can be distinguished from other lymphocytes based on their surface phenotype (CD3−CD4⁺CD45⁺IL-7R⁺c-Kit⁺)(59). LTi cells also express the chemokine receptors, CCR7 and CXCR5 (60), which promote their homing to sites of future lymphoid organ development and lead to their interactions with local mesenchymal VCAM⁺ICAM⁺ lymph node organizer cells (61–67). LTi cells express LT and trigger the LTβR on the lymph node organizer cells. In turn, the local mesenchymal cells produce homeostatic chemokines, including CCL19, CCL21 and CXCL13, which accelerates the recruitment of LTi cells as well as other lymphocytes and promotes the formation of lymphoid architecture, including the differentiation of stromal cells and high endothelial venules (HEVs) as well as the formation of segregated T and B cell zones (59, 61, 68).

CD4⁺CD3⁻ LTi cells are among the first to appear at the site of NALT organogenesis just after birth, suggesting that they are likely to be involved in the initiation of NALT formation (21, 43). However, the role of LTi cells in NALT development remains unresolved. For example, the importance of LTi cells in the development of secondary lymphoid organs is illustrated in gene-targeted mice that lack the transcription factors, retinoic acic related orphan receptor γ (RORγ) (58, 69, 70) or inhibitor of DNA binding 2 (Id2) (71). These transcription factors are necessary for the development of LTi cells and, as a result, mice lacking these factors lack LTi cells and fail to develop LNs and Peyer’s patches (69, 71). Despite this, Rorc^−/− mice have NALT (5), suggesting that LTi cells may not be required for NALT development. In contrast, Id2^−/− mice completely lack NALT (52). Moreover, the adoptive transfer of LTi cells from normal mice into Id2^−/− mice during the first week after birth partially restores NALT formation (52), suggesting that at least some types of LTi cells are required for NALT development. Alternatively, the difference in NALT development in Rorc^−/− and Id2^−/− mice may reflect a requirement for additional cell types to promote NALT development. For example, Id2^−/− mice lack some populations of DCs as well as NK cells (72, 73) and these cell types may play important, although currently undefined, roles in NALT development. This would be consistent with other data showing that NK cells participate in the stabilization of LNs during development (74).

The discrepancy between Rorc^−/− and Id2^−/− mice may also reflect diversity in the LTi population -with some LTi cells being formed independently of RORγ. Consistent with this idea, the LTi cells found in NALT express very low levels of the chemokine receptors, CCR7 and CXCR5, relative to LTi cells from Peyer’s patches (21). These receptors are normally required for the recruitment of LTi cells to sites of LN and Peyer’s patch development and as a result, mice lacking these receptors or their chemokine ligands are unable to form most LNs and Peyer’s patches (60, 75). In contrast, NALT develops in the absence of CXCL13, CCL19 and CCL21 and normal numbers of LTi cells migrate to the NALT of Cxcl13^−/− or plt/plt mice just after birth ((13, 21) and unpublished data). These data demonstrate that conventional LTi cells are probably not involved in NALT development and further suggest that additional, as yet undefined chemokines attract LTi cells to the NALT anlage.

Additional evidence suggesting that LTi cells may not be required for NALT development is that NALT is formed independently of IL-7R and TRANCE receptor signaling (5). IL-7 and TRANCE provide survival and proliferation signals for LTi cells (76, 77). Interestingly, LNs and Peyer’s patches are selectively dependent on either IL-7 or TRANCE (60, 65, 67, 76). As a result, LNs fail to form in TRANCE^−/− mice (76), even though Peyer’s patch development is normal. Conversely, the development of Peyer’s patches is much more dependent on IL-7 than on TRANCE (60, 65, 67). Interestingly, NALT develops in TRANCE^−/− mice and even appears hyperplastic (5). This may be due to the altered bone structure in TRANCE^−/− mice, which lack osteoclasts (78, 79), resulting in unusual skull morphology including alterations to structure of the nasal passages surrounding NALT. Although the development of NALT in TRANCE^−/− mice may not be surprising, since these mice form Peyer’s patches, NALT also develops in IL-7R^−/− mice, which lack Peyer’s patches (67). Despite its ability to form in IL-7R^−/− mice, the structure of NALT is abnormal, since IL-7R^−/− mice have impaired T and B cell development resulting in lymphopenia (5). Regardless, NALT development can occur in the apparent absence of LTi cells or factors that promote LTi cell survival.

One potential explanation for why NALT development may occur independently of LTi cells is that NALT forms in the postnatal period in the first couple of weeks after birth (21, 43, 52). In contrast, most LNs develop during embryogenesis between days E11 and E19 (53, 80). Since LTi cells are one of the few cells that express LT during embryogenesis, they are essential to initiate the development of LNs and Peyer’s patches. However, conventional lymphocytes also express LT and begin to circulate after birth. Thus, these cells may replace the function of LTi cells. Given that LT is not required for the development of NALT, the expression of LT on LTi cells or other cells may be relatively unimportant compared to other signals delivered by LTi cells. For example, agonistic antibodies to the LTβR do not promote LN development in Rorc^−/− mice (58, 69), suggesting that LT is not the only signal provided by LTi cells. Thus, any cell type that may substitute for LTi cells in the developmental program of NALT would likely have to deliver this unknown signal as well.

LT and the maintenance of lymphoid architecture

In addition to the development of lymph nodes and Peyer’s patches, LT signaling is also important for the maintenance of lymphoid architecture (81–84). For example, the spleens of Lta^−/− mice are unable to form segregated B and T cell zones (81), lack marginal zones (51, 85) and cannot form germinal centers (83, 86). LT is also important for the differentiation of stromal cell populations, including FDC (87), gp38+ stromal cells and BP3+ stromal cells (88, 89). Local LT production is also important to maintain steady state number of DCs in secondary lymphoid organs (90). Furthermore, expression of peripheral lymph node addressin (PNAd) and mucosal addressin cell adhesion molecule 1 (MAdCAM-1) on blood vessels from SLO is totally dependent on LT (91, 92). Consistent with these results, the few B and T cells present in the NALT of Lta^−/− mice are not segregated into separate areas (5). Moreover, the NALT of Lta^−/− mice lack FDCs and BP3+ stromal cells (Figure 1)(5, 13). Moreover, the development of functional HEVs is impaired in the absence of LT. For example, LT is crucial for the induction of the high endothelial cell sulfotransferase (HEC-6ST), which is needed for the expression of PNAd on HEVs (43). LT is also required for the expression of the protein GlyCAM-1, which is modified on sialyl Lewis × by HEC-6ST to generate PNAd (43, 93). Although PNAd-expressing HEVs appear nearly normal in the NALT of Cxcl13^−/− and plt/plt mice (13), they have reduced chemokine expression, particularly CCL21, and are somewhat smaller than those in normal mice, possibly due to reduced trafficking of cells. Thus, even though NALT is formed in the absence of LT, LT is still necessary for the maintenance of NALT architecture.

Fig 1. LT expression in NALT is important for stromal cell differentiation and HC production.

FDC differentiation is compromised in LT−/− (C) and Cxcl13 −/− mice (D) (CD21: Brown, B220: Blue). BP3+ stromal cells do not develop in the absence of LT (G) and CXCL13 (H). FDC- and BP3 stromal cell networks are only detected in NALT from B6- (A, E) and plt/plt mice (B, F). mRNA expression measured by quantitative PCR. CXCL13 is important for LT expression in NALT and LT is required for production of CXCL13 and CCL19.

For the most part, the structural defects in the NALT of Lta^−/− mice are not developmental and can be repaired by the reconstitution of Lta^−/− mice with normal bone marrow (BM)(5). For example, BM reconstitution restores the structure of B cell follicles, FDCs, PNAd⁺ HEVs and even subepithelial DCs (5). This points out an important difference between NALT and other lymphoid organs, since Peyer’s patches and LNs are never restored in Lta^−/− mice, even after BM transfer (not shown and (53, 86)). BM reconstition also restores the function of NALT (5). In fact, antigen-specific CD8 T cells and germinal center B cells are formed at normal frequencies and antigen-specific IgM and IgG are produced at normal levels after intranasal immunization or infection of Lta^−/− mice that have been reconstituted with normal BM.

Homeostatic chemokines and the structure of NALT

The architecture of most secondary lymphoid organs is maintained by homeostatic chemokines, including CCL19, CCL21 and CXCL13, which are constitutively expressed in secondary lymphoid organs and direct the recruitment and placement of lymphocytes and DCs (75, 94–97). These chemokines are also expressed in the NALT and their expression pattern fits with a role for these molecules in organizing the structure of NALT (13). For example, CXCL13 is located in the center of the B cell follicle where it can recruit B cells, CCL21 is located around the follicle in the T cell area as well as on HEVs, where it can attract recirculating B and T cells and CCL20 is expressed in the dome epithelium, where it can attract subepithelial DCs (13). As in other lymphoid organs, LT controls the expression of these homeostatic chemokines (13). In fact, the structure of NALT in Cxcl13^−/− × plt/plt mice (which lack CXCL13, CCL19 and CCL21) appears very similar to the structure of NALT in Lta^−/− mice, with very few B or T cells and no FDCs or germinal centers (Figure 1)(13).

In fact, there is a positive feedback relationship between LT expression on lymphocytes and homeostatic chemokine expression, particularly CXCL13 expression, by stromal cells (75). Consistent with this, Lta^−/− and Cxcl13^−/− mice lack well-organized CD21⁺ and BP3⁺ stromal cell networks (13), suggesting that interactions between LT expressing cells and CXCL13 producing cells is crucial for differentiation of stromal cell populations and the maintenance of well-populated B cell zones in the NALT. For example, the overall structure of NALT is similarly compromised in the absence of either LTα or CXCL13 (13). In addition, both LTα and CXCL13 are necessary for the differentiation of FDCs as well as BP3⁺ and ERTR7⁺ stromal cells in NALT (13). This is consistent with the idea that CXCL13 is necessary for primary follicle formation and FDC differentiation under steady state conditions in the spleen as well as the idea that CXCL13 is involved in a positive feedback loop with LTαβ and TNFα (75, 89). In fact, the stromal cell defects in the NALT of Lta^−/− and Cxcl13^−/− mice are most likely due to the interruption of this loop and the failure of LTαβ and TNFα to promote the differentiation of stromal cell precursors. Thus, the structural defects in the NALT of Lta^−/− mice and Cxcl13^−/− appear to be linked by the co-dependent expression of CXCL13 and LTαβ.

Although the expression of CCL19 and to a lesser extent, CCL21 is also controlled by LT signaling in NALT (13), the loss of these chemokines in plt/plt mice has less dramatic consequences for the structure of NALT than does the loss of CXCL13 in Cxcl13^−/− mice (13). In part, this is because the large central B cell follicle of NALT including the FDC network, is minimally affected by the loss of CCL19/CCL21. However, T cell recruitment is substantially impaired in the NALT of plt/plt mice (Figure 1)(13, 21).

CCL20 should also be considered a homeostatic chemokine, as it is constitutively expressed in the FAE of mucosal lymphoid organs (15, 17, 98). In Peyer’s patches, CCL20 is thought to control the recruitment and placement of subepithelial DCs (15). This is a critical function, as subepithelial DCs are believed to process antigen that is transported across the epithelium by M cells (15, 99, 100). The current model is that M cells in Peyer’s patches express CCL20 and possibly other chemokines, like CXCL16 and CCL9 (15, 17, 101, 102) and that these chemokines attract CCR6⁺ sub-populations of DCs and lymphocytes to the area below the dome epithelium, where they are positioned to receive antigen from the M cells and to interact with naïve lymphocytes entering Peyer’s patches via HEVs. Cells in the FAE of NALT also express CCL20 (13), although it is not known whether they also express CXCL16 and CCL9. The expression of CCL20 in NALT, like the expression of CXCL13, CCL19 and CCL21, is also dependent on LTα(13). Although the role of CCL20 in NALT development and recruiting LTi cells has not been directly tested, Epithelial cells overlaying the NALT of Lta^−/− mice do not express CCL20 and DCs are not recruited to the subepithelial dome region of Lta^−/− NALT.

The relationship between LT and homeostatic chemokine expression was clarified by the demonstration that LTβR uses both the canonical and an alternative NFκB signaling pathway (54). The canonical NFκb pathway culminates with the translocation of p50/relA or p52/RelA and leads to the expression of inflammatory chemokines like macrophage induced protein 1 (MIP 1β) and macrophage induced protein 2 (MIP2)(54, 55). In contrast, the alternative NFκB signaling cascade goes through NFκB inducing kinase (NIK) and leads to the translocation of p52/RelB, which promotes the production of the homeostatic chemokines, CCL19, CCL21 and CXCL13 (54). Alymphoplasia or aly/aly mice have a mutation in NIK, which leads to altered stromal cell differentiation and abnormal production of homeostatic chemokines and ultimately to the failure of LN and Peyer’s patch development (103).

Chemokines and the generation of local immune responses in NALT

The function of homeostatic chemokines in fully developed secondary lymphoid organs is to promote immune responses by recruiting naïve B and T cells from the blood and facilitating their interactions with activated DCs and with each other (94). In the T cell zone, CCL19 and CCL21 bring together mature DCs and naïve T cells to initiate T cell priming (94, 104). In the B cell area, CXCL13 coordinates the interaction of B cells and T follicular helper (T_FH) cells and promotes B cell differentiation that culminates with the production of antigen-specific, antibody producing plasma cells (95, 97, 105–108).

The function of NALT is similarly dependent on chemokines. For example, LTα and CXCL13 are both required for the formation of germinal centers in NALT after influenza infection (5, 13, 21). Although a lack of germinal centers would normally lead to impaired B cell responses, isotype switching to IgA does occur in the NALT of both Lta^−/− and Cxcl13^−/− mice after influenza infection, albeit at reduced levels (13). Despite this reduction, IgA-secreting influenza-specific plasma cells are observed in the nasal mucosa of Cxcl13^−/− and Lta^−/− mice at normal or above normal numbers (13). These results are surprising, given that switching to IgA is thought to be impaired in Lta^−/− mice (109–111). However, the production of influenza-specific IgA in serum is normal in Lta^−/− mice (5), suggest that Lta^−/− mice actually make more IgA than normal mice in the NALT and nasal mucosa after influenza infection. Thus, not all IgA responses are dependent on LTα. Consistent with this, the administration of LTβR-Ig reduced, but did not prevent, IgA responses to tetanus toxin in the intestine (112), demonstrating that LT signaling, FDCs and germinal centers are not required for IgA switching. Interestingly, previous studies suggested that switching to IgA may occur via unusual mechanisms, such as the direct development of IgA cells from bone marrow (113, 114), the differentiation of IgA cells outside of lymphoid organs (115–117) and the production of IgA independently of cognate interactions with T cells (118). However, other studies showed that IgA switching occurs only in lymphoid tissues, such as NALT, Peyer’s patches and isolated lymphoid follicles (119). Thus, it appears that functional lymphoid tissues are important to facilitate switching to IgA, however LTα is not essential for this function in NALT.

LTα is also important for CD8 T cell responses in NALT, as CD8 T cell responses to influenza are delayed in the NALT of Lta^−/− mice (13, 120). In part, this is due to the poor expression of CCL21 and CCL19 in the NALT of Lta^−/− mice, since CD8 T cell responses are also impaired in the NALT of plt/plt mice (13). However, the delayed T cell responses in Lta^−/− mice are probably due to more than just the impaired expression of CCR7 ligands. In fact, it is likely that the impaired expression of CCL20 in the dome epithelium of Lta^−/− NALT disrupts the recruitment of sub-epithelial DCs to the follicular dome of the NALT (13). Therefore, the process of antigen acquisition from the nasal lumen is likely to be impaired in the NALT of Lta^−/− mice as a consequence of a reduced CCL20 production by epithelial cells. In addition, it was recently demonstrated that LTα is required for homeostatic proliferation of DCs in the spleen (90). Thus, the delayed activation of influenza specific CD8 T cells in the NALT of Lta^−/− mice may result from a combination of deficiencies in DC activation and antigen retrieval from the dome epithelium of the NALT.

BALT

Structure of BALT

BALT was first defined by Bienenstock in the lungs of pigs and rabbits as a mucosal lymphoid tissue located just beneath the bronchial epithelium and adjacent to a pulmonary artery (121–125). Like NALT, BALT is structurally similar to Peyer’s patches, with prominent B cell follicles and interfollicular T cell areas underneath a specialized FAE (121–125). BALT also has a population of subepithelial DCs beneath the FAE that can be identified with antibodies against S-100 and MHC class II antigens (126). T cells are also found in this same environment (126). The strategic positioning of these cells in the dome structure suggests that the subepithelial DCs are ready to receive antigens obtained by M cells from the luminal side of the epithelium and to present these antigens to local T and B cells (126). Unlike Peyer’s patches, the HEVs of BALT express PNAd and not MAdCAM (127, 128), suggesting that NALT and BALT are more closely related to each other than to lymphoid tissues of the intestine. Unike HEVs in conventional lymphoid organs, HEVs in BALT also express VCAM-1 (128). In fact, lymphocytes use L-selectin/PNAd, α4β1 integrin/VCAM-1 and LFA-1 to enter BALT (128). The use of this combination of receptors is unique among secondary lymphoid tissues and may recruit specific populations of effector or memory cells to the BALT. Thus, BALT probably functions as a mucosal lymphoid tissue in the upper bronchi of the lungs that primes and expands T cells that are specialized to protect the lung.

BALT can be detected in several species, such as rabbits, rats, pigs and guinea pigs, under normal physiological conditions and without antigenic stimulation (129). However, it is absent in most healthy cats, mice and humans (130). Moreover, BALT is not found in the lungs of pigs living in pathogen free conditions or humans who died from non-pulmonary causes (130). In fact, the sporadic appearance of BALT-like areas in humans and mice has led some investigators to doubt whether iBALT is an important secondary lymphoid tissue in these species (131). Despite this concern, it is clear that tissues structurally similar to BALT are formed in the lungs of mice and humans in response to inflammation and infection (127, 130, 132–134). We have termed these tissues inducible BALT or iBALT (127). Unlike classically defined BALT, which develops in during embryogenesis and is generated and maintained in the absence of antigen (121–124), iBALT appears to be formed only after infection or inflammation in both mice and humans (127, 130, 132, 135, 136) and is found in numerous places along the upper and lower bronchi and even in the interstitial areas of the lung (137). In addition, BALT is detected upon autopsy of human fetal and infant lungs, mainly in episodes of amnionitis or intrauterine pneumonia without any signs of infection (138). Moreover, transgenic mice that express various cytokines in the lung develop iBALT (139, 140), as do mice and humans prone to autoimmune diseases (128, 141) or humans with recurrent or chronic respiratory infections (126, 130, 141, 142). These areas of iBALT are consistent with Sminia’s description of multiple Bronchus Associated Lymphoid Units (BALUs) (125). The areas of iBALT are also similar to tertiary lymphoid tissues detected at peripheral locations in several autoimmune and chronic inflammatory diseases (143) (144) (145) (146) (147) (148, 149). A common component of each of these models is chronic inflammation, which is implicated in the development of ectopic or tertiary lymphoid areas in other sites, such as rheumatoid joints (150–153). Interestingly, the most well organized iBALT (with well-defined B cell follicles, T cell areas, FDCs and HEVs) is observed in mice in which respiratory inflammation is present from birth (51, 128, 139, 140). These data suggest the possibility that the development of iBALT, like the development of other lymphoid organs, may be preferentially initiated during a specific developmental window during the neonatal period.

One difference between classically defined BALT and iBALT is the presence of the FAE. Classic BALT is a true mucosal lymphoid organ underneath a specialized FAE that contains M cells for the transport of antigen (154–156). In contrast, most examples of iBALT do not have a clearly visible FAE and in many cases they are in perivascular areas or parenchymal areas of the lung with no discernable access to antigens from the lumen of the respiratory tract (127, 141). Despite the apparent lack of access to antigen, most iBALT follicles are surrounded by lymphatic vessels (141). However, it is unclear whether these are afferent or efferent lymphatics. Afferent lymphatics could bring antigen from distal parts of the lung to the iBALT, while efferent lymphatics would connect iBALT with the draining lymph node. Transport of antigen and DCs through afferent lymphatics can have a positive impact in T- and B cell priming in the LN. Also, the presence of efferent lymphatics can be associated to the exclusive stimulation of lymphocytes in tissue domains that produce self-antigens (141). This is an important issue that needs to be resolved.

Although most reports of iBALT in humans have associated the presence of this tissue in the lungs with local pathology (126, 130, 135, 141, 157–159), it is clear that iBALT also participates in lymphocyte priming and differentiation, formation of germinal centers, B cell isotype switching, recruitment of effector cells during infection and maintenance of local memory cells in mice (127, 136, 137, 160–166). In these cases, iBALT appears to be beneficial and can even reduce the immunopathology associated with influenza infection (127, 137). In contrast, iBALT hypertrophy in humans is often associated with persistent viral infections, such as HIV, EBV or KSHV (135, 157), and is also associated with autoimmune diseases, such as Systemic Lupus Erythematosus, Rheumatoid Arthritis and Sjogren Syndrome (126, 167). In addition, iBALT is associated with various forms of interstitial lung disease, such as chronic obstructive pulmonary disease (157, 168), which leads to decreased lung function. In these cases, it appears that the formation of iBALT and the recruitment of lymphocytes to the lung result in pathology (141, 157, 168). Thus, iBALT functions like many other secondary lymphoid organs and promotes beneficial immune responses that help clear infection, but also serves to amplify pathologic immune responses against autoantigens or chronic infections.

Development of iBALT

It is becoming increasingly clear that iBALT in mice and humans is not a constitutive secondary lymphoid organ, but is an ectopic or tertiary lymphoid tissue that is formed in response to antigen or inflammation (127, 130, 138). The development of tertiary lymphoid tissues is termed lymphoid neogenesis (150, 153) and is thought to use mechanisms similar to those used during the programmed development of secondary lymphoid organs. For example, the ectopic expression of LT or homeostatic chemokines in the pancreas leads to the development of a fully formed tertiary lymphoid tissue in the pancreas (153, 169). Given the central importance of LT in lymphoid organ development it seems that LT should be involved in iBALT development as well. However, this issue is not been fully resolved. For example, Lta^−/− mice form large iBALT-like areas in the lungs (51, 170). These iBALT areas are not fully developed as they lack segregated B and T cell zones, FDCs and PNAd⁺ HEVs (51, 127). However, they do seem capable of promoting DC/T cell interactions and priming naïve T cells (171, 172). Lta^−/− mice are also capable of generating primary T and B cell responses to influenza and other respiratory viruses (120, 127, 173) and at least some of the response appears to be initiated directly in the lung. Interestingly the B and T cell responses are delayed in Lta^−/− mice (120), suggesting that although the iBALT areas in these mice are functional, they are not particularly efficient. This poor efficiency probably reflects the poor organization of iBALT in Lta^−/− mice. However, Lta^−/− mice do eventually accumulate large numbers of B cells in the lung, many of which had germinal center characteristics (120). Again, this is surprising, as germinal centers are not formed in the spleens of Lta^−/− mice (51, 84, 86, 120). Like it does for NALT, the reconstitution of Lta^−/− mice with normal BM restores the structure of iBALT - including segregated B and T cell zones, networks of FDCs and PNAd⁺ HEVs (127). These areas of organized iBALT are much more efficient at inducing immune responses to influenza than the disorganized lymphoid accumulations in Lta^−/− mice. In fact, splenectomized, BM-reconstituted Lta^−/− mice generated B and T cell responses to influenza with normal kinetics (127). These data argue that although LT is clearly involved in the organization and efficient function of iBALT, it is not absolutely required for the development of iBALT.

Areas of iBALT that are formed in response to influenza infection contain large B cell zones with a central FDC networks and germinal center B cells (127). Consistent with its ability to organize follicular structures, CXCL13 is found in the center of B cell follicle in a reticular pattern that appears to co-localize with the FDC network (127). CCL21 is detected in the T cell zone in a reticular pattern and more strongly on PNAd⁺ HEVs (127). The spatial location of chemokine expression in iBALT suggests that these molecules are central for the organization of iBALT. The role of homeostatic chemokines in the lung is not restricted to simply organizing iBALT. CCL21 expression in the lung helps to recruit activated T cells to sites of inflammation (174) and CCL21 expression on lymphatics helps to retrieve CCR7⁺ T cells from the lung and return them to the draining LNs (175, 176). Thus, homeostatic chemokine expression in the lung serves numerous functions.

Based on the paradigm for conventional lymphoid organs, the expression of these chemokines should be dependent on LT. This would make sense, since iBALT is disorganized in Lta^−/− mice (51, 127). However, CXCL13, CCL19 and CCL21 are all inducibly expressed in the lung during influenza infection in the complete absence of LTα and TNF (127). This suggests that there may be other TNF family members, such as LIGHT, that can compensate for the absence of LT and TNFα and induce the expression of homeostatic chemokines (177). Alternatively, there may be very different mechanisms that control the expression of homeostatic chemokines under conditions of acute inflammation. If this is the case, it may explain how the development of ectopic lymphoid follicles is initiated. An initial infection or inflammation could trigger the acute expression of CXCL13, CCL21 and CCL19, which would lead to an influx of lymphocytes. The lymphocytes would then express LT and promote the differentiation of local stromal cell precursors into mature lymphoid stromal cells. Chemokine expression by these stromal cells would then be maintained under homeostatic conditions by LT.

Clues to the type of inflammation that is required for iBALT formation can be found in the literature. For example, mice deficient in the NAD(P)H/NRH:quinone oxidoreductases, NQO1 and NQO2, develop iBALT (178), suggesting that oxidative stress may be involved in iBALT formation. Conversely, mice lacking iNOS do not develop iBALT (179). The effects of oxidative stress could be direct, via activation of local mesenchymal cells, or indirect via the production of cytokines like TNF, IL-6, and IL-1β (178). Supporting a role for cytokines is the recent observation that Tregs are instrumental in the negative regulation of iBALT (180). However, it is not clear whether Tregs are acting to suppress other T cells from making cytokines or whether Tregs are acting on APCs, such as macrophages, and suppressing the production of inflammatory mediators.

One open question is whether LTi cells are involved in the development of ectopic lymphoid follicles, such as iBALT. Given that CXCL13 is expressed in the lungs just a few days after influenza infection (127) and that inflamed lungs also express VCAM (128), it seems likely that LTi cells expressing CXCR5 and α4β1 integrin could home to sites of inflammation in the lung and be involved in the formation of iBALT. However, it is not clear whether LTi cells exist in adults or if they do exist, whether they are functionally similar to those in the developing embryo (181, 182). Moreover, LTi cells are not required for the development of ectopic lymphoid follicles in an experimental model of thyroiditis (183, 184). In this model, interactions between infiltrating DCs and activated CD4 T cells trigger formation of ectopic follicles (183). This seems reasonable because dendritic cells express the LTβR and can express CCL19, setting up the conditions for the attraction for populations of CCR7⁺ B and T cells that can continue with the compartmentalization of the ectopic follicle. Again, LT may not be the primary inducer of iBALT, since some aspects of iBALT can be found in Lta^−/− mice (51). Thus, the cellular and molecular events that initiate iBALT formation are still incompletely understood.

Once iBALT is generated during acute inflammation or infection, the structure is maintained for several months. For example, iBALT is found in murine lungs at least 90 days after influenza infection (137). However, iBALT progressively becomes smaller over time in the absence of continued stimulation (137). However, it is not clear if iBALT ever completely resolves since a subsequent infection seems to rapidly replenish the areas of iBALT (unpublished data). Moreover, iBALT probably serves as a niche to maintain and reactivate antigen specific memory cells that recirculate through the lung (137). Thus, there is a dynamic process of recirculation of cells in and out of iBALT that probably helps to maintain local immune surveillance.

The role of iBALT in lung pathology

Data from our laboratory show that B and T cell immune responses to influenza can be initiated directly in iBALT without involvement of conventional lymphoid organs(127). This suggests that iBALT can serve a protective function. In fact, mice lacking conventional lymphopid organs survive higher doses of influenza significantly better than normal mice(127). These data are very surprising and suggest that priming in iBALT may elicit a different type of immune response than priming in draining LN. This would be consistent with other data showing that T cells primed in situ in the lung are skewed towards a Th2 response (171, 172). Although Th2 responses are not normally associated with anti-viral immunity, the lymphocytes responding to influenza in iBALT may produce anti-inflammatory cytokines or accelerate B cell responses that promote viral clearance without excess pathology. This would fit with the idea that iBALT is a mucosal tissue that generates immune responses specialized to protect the lung. These possibilities are currently being investigated.

IBALT is also formed in response to other infectious pulmonary diseases (185) and is prominent in tuberculosis (186). Pulmonary infection with tuberculosis leads to the formation of granulomas, which consist predominantly of lymphocytes and macrophages. Granulomas are highly organized and have characteristics of iBALT, including B cell follicles, HEVs, FDCs and expression of the homeostatic chemokines CXCL13 and CCL19 (186). The CCR7 ligands, CCL19 and CCL21, contribute to the formation and effectiveness of tuberculosis granulomas and iBALT (186). Again, the proper formation of iBALT functions protectively in response to tuberculosis.

On the other hand, iBALT in humans is most often associated with pulmonary pathology. For example, iBALT has been identified in humans with a variety of interstitial lung diseases (187), including hypersensitivity pneumonitis (159, 188), desquamative pneumonia (135), nonspecific interstitial pneumonia, bronchioalveolar carcinoma and particularly in patients with autoimmune diseases such as rheumatoid arthritis and Sjogren syndrome (141). Interestingly, iBALT was never observed in patients with idiopathic pulmonary fibrosis (141), suggesting that particular types of disease preferentially lead to the formation of iBALT. Given the extensive formation of iBALT in patients with tuberculosis or autoimmune disease (141, 186), we believe that prolonged exposure to antigen from pathogens or even self-antigens in the lung triggers iBALT formation (132). In patients with rheumatoid arthritis and Sjogren syndrome, there is a general association between the progression of autoimmunity and the complexity in the organization of iBALT (141). Again, this suggests that antigen may be the driving force behind the maintenance of iBALT.

The structure of iBALT in patients with autoimmune diseases resembles iBALT associated with infectious disease - with separated B and T cell zones, FDCs, germinal centers, HEVs and lymphatics (Figure 2)(126, 141). Interestingly, LT-expressing T cells can be found in the T cell zone, suggesting that activated T cells may contribute to the organization of T cell zone in BALT by triggering the LTβR on stromal cells and inducing production of CCL21 and CCL19. LT-expressing T cells may also promote the differentiation of HEVs (91). LT is also expressed on B cells and is crucial for lymphangiogenesis in reactive LNs (189), making the traffic of antigen and DCs into the LN more efficient. Thus, it is likely that antigen-activated B and T cells maintain the structure of iBALT via LT expression.

Fig. 2. Cellular and molecular features of human iBALT.

A) H&E stain of a lymphoid follicle like structure in a lung biopsy from a RA patient. B) Proliferating B cells (PCNA/CD20). C) FDC network (CD21). D) Parafollicular area (PFA) containing APCs (CD68) and T cells (CD3). E) LT producing cells are found in PFA. F) Specialized vasculature: HEVs (PNAd) and lymphatics (M2A) develops in PFA. G) FDC like cells are producing CXCL13 in the middle of a germinal center (CXCL13/CD68). H) CCL21 is produced by vessels outside the B cell follicle and close to T cell zone (CCL21/CD3).

In addition to supporting the structure of iBALT, the activated B and T cells in the lungs of autoimmune patients may also contribute to disease. For example, the lungs of rheumatoid arthritis patients contain increased levels of peptidyl arginine deiminase 2 (PADI2)(141), an enzyme involved in the postranslational transformation of arginine into citrulline (190, 191). Since citrullinated proteins are a primary autoantigen in rheumatoid arthritis (190, 192), this enzyme probably contributes to pathology. In fact, citrullinated proteins are easily detected in the lungs of rheumatoid arthritis patients surrounding areas of iBALT (141). Moreover, plasma cells producing autoantibodies against citrullinated fibrinogen are also found surrounding areas of iBALT in the lungs of rheumatoid arthritis patients (141). Finally, there is a positive correlation between iBALT complexity and the production of anti-cyclic citrullinated peptide antibodies in the bronchial alveolar lavage (141), suggesting that autoantibodies are produced locally in the lung. Since immune complexes lead to the release of inflammatory mediators that can activate fibroblasts, the local production of autoantibodies is likely to lead to collagen deposition and pulmonary fibrosis. Thus, iBALT may be contributing to pulmonary pathology in patients with autoimmune disease.

Autoimmunity is not the only disease associated with the formation of iBALT. For example, iBALT formation is also associated with the most severe stages of chronic obstructive pulmonary disease (168). Moreover, iBALT if formed in the lungs of heavy smokers (133, 157, 193). Thus, some types of pulmonary inflammation, particularly inflammation associated with oxidative stress (178), lead to iBALT formation and pathology in the absence of easily identified antigens or autoantigens.

SUMMARY AND CONCLUSIONS

Like other lymphoid organs, NALT and iBALT have evolved to promote the interactions of lymphocytes and APCs and to efficiently elicit antigen-specific B and T cell responses to infectious agents. In this regard the roles of lymphotoxin and the homeostatic chemokines play similar, if not identical roles in all secondary lymphoid organs. However, the details of the development and function of NALT and iBALT differ from those of conventional lymphoid organs. This probably reflects the specialized requirements of the respiratory tract to promote immunity without excessive inflammation. However, in pathological conditions like autoimmunity, LT-induced homeostatic chemokines participate in the organization of iBALT and have a detrimental impact in lung physiology, promoting formation of local cellular niches with favorable conditions for the expansion and activation of autoreactive and bystander lymphocytes and attraction of additional cell populations that can contribute to the induction of severe lung pathology.

Given the positive impact of iBALT on immunity to infectious agents and the negative impact of iBALT on respiratory function in chronic pulmonary diseases, it is important to discover the mechanisms that regulate the formation and function of these tissues. However, many important questions about iBALT remain open. For example: do adult LTi cells participate in iBALT formation? What molecular pathways initiate homeostatic chemokine expression in the lung and how do these mechanims impact iBALT formation? Finally, can we manipulate the mechanisms of iBALT formation and function to promote immunity to respiratory infections and prevent pulmonary disease associated with autoimmunity? The answers to these questions will require research in both murine and human systems and should allow us to intervene in a variety of pulmonary diseases.