key: cord-0009673-bid9bbe7
authors: Brandtzaeg, Per; Farstad, Inger Nina; Johansen, Finn‐Eirik; Morton, H. Craig; Norderhaug, Inger Natvig; Yamanaka, Takeshi; Brandtzaeg, Per; Farstad, Inger Nina; Johansen, Finn‐Eirik; Morton, H. Craig; Norderhaug, Inger Natvig; Yamanaka, Takeshi
title: The B‐cell system of human mucosae and exocrine glands
date: 2006-04-28
journal: Immunol Rev
DOI: 10.1111/j.1600-065x.1999.tb01342.x
sha: bdf816d79061ded899cce957e56bf3b5970347f2
doc_id: 9673
cord_uid: bid9bbe7

Summary: The mucosae and exocrine glands harbour the largest activated B‐cell system of the body, amounting to some 80–90% of all immunoglobulins (Ig)‐producing cells. The major product of these immunocytes is polymeric (p)IgA (mainly dimers) with associated J chain. Both pIgA and pentameric IgM contain a binding site for the polymeric Ig receptor (pIgR), or secretory component (SC), which is a requirement for their active external transport through secretory epithelia. The pIgR/SC binding site depends on covalent incorporation of the J chain into the quaternary structure of the polymers when they are produced by the local immunocytes. This important differentiation characteristic appears to be sufficient functional justification for the J chain to be expressed also by most B cells terminating at secretory effector sites with IgD or IgG production; they probably represent a ‘spin‐off’ from sequential downstream C(H) switching on its way to pIgA expression, thus apparently reflecting a maturational stage of effector B‐cell clones compatible with homing to these sites. Observations in IgA‐deficient individuals suggest that the magnitude of this homing is fairly well maintained even when the differentiation pathway to IgA is blocked. Certain microenvironmental elements such as specific cytokines and dendritic cells appear to be required for induction of IgA synthesis, but it remains virtually unknown why this isotype normally is such a dominating product of local immunocytes and why they have such a high level of J chain expression. Also, despite the recent identification of some important requirements in terms of adhesion molecules (e.g. integrin α4β7 and MAdCAM‐1) that explain the “gut‐seeking” properties of enterically induced B cells, the origin of regionalized homing of B cells to secretory effector sites outside the gut remains elusive. Moreover, little is known about immune regulation underlying the striking disparity of both the class (IgD, IgM) and subclass (IgA1, IgA2, IgGI, IgG2) production patterns shown by local iinmttnocytes in various regions of the body, although the topical microbiota and other environmental stimuli might be important. Rational design of local vaccines will depend on better knowledge of both inductive and migratory properties of human mucosal B cells.

Summary: The mucosae and exocrine glands harbour the largest activated B-cell system of the body, amounting to some 80-90% of all immunoglobulin {Ig)-producing cells. The major product of these immimocytes is polymeric (p)IgA (mainly dimers} with associated J chain. Both pIgA and pentameric IgM contain a hinding site for tbe polymeric Ig receptor (pIgR), or secretory component (SC), which is a requirement for their active external transport through secretory epithelia. The pIgR/SC binding site depends on covalent incorporation of the J chain into the quaternary structure of the polymers when they are produced by the local immunocytes. This important differentiation characteristic appears to be sufficient functional justification for the J chain to be expressed also by most B cells terminating at secretory effector sites with IgD or IgG production; they probably represent a "spin-off' from sequential downstream Qi switching on its way to pIgA expression, thus apparently reflecting a maturational stage of effector B-cell clones compatible witb boming to tbese sites. Observations in IgA-deficient individuals suggest tbai the magnitttde of tbis homing is fairly well maintaitied even wben the differentiation pathway to IgA is blocked. Certain microenvironmental elements such as specific cytokines and dendritic cells appear to be required for induction of IgA synthesis, but it remains virtually unknown why this isotype normally is such a dominating product of local immunocytes and why they have such a high level of J chain expression. Also, despite the recent identification of some important requirements in terms of adhesion molecules (e.g. integrin a4p7 and MAdCAM-1) that explain the "gut-seeking" properties of enterically induced B ceils, tbe origin of regionalized homing of B cells Co secretory effector sites outside the gut remains elusive. Moreover, little is known about immune regulation underlying tbe striking disparity of both the class (IgD, IgM) and subclass (IgAl, IgA2, IgGI, IgG2) production patterns shown by local iinmttnocytes in various regions of tbe body, althougb tbe topical microbiota and otber environmental stimuli migbt be important. Rational design of local vaccines will depend on better knowledge of both inductive and migratory properties of human mucosal B cells.

Secretory immunity is the best characterized part of the mucosal immune system, although the regulation of B cells forming its hasis is poorly understood. The interest in adaptive local immunity increased in the 1960s when it was reported hy several lahoratories that the predominant immunoglohulin (Ig) class in external hody fluids is IgA rather than IgG (1) . The secretory IgA (SIgA) molecule was found to be unique (2), both because of its polymeric structure and association with the secretory component (SC). This epithelial glycoprotein was initially termed "transport piece" to suggest that it might facilitate the entry of IgA into epithelial cells (3) -an interesting suggestion that would turn out to be true in an even more intricate manner than anticipated.

At about the same time, Crabbe et al. (4) demonstrated an isotype distribution of intestinal Ig-produdng immunocytes (B-cell blasts and plasma cells) strikingly different from that in peripheral lymphoid organs; mucosal IgA cells were reported to he at least 20 times more frequent than IgG cells. Despite the absence of a satisfactory immunoregulatory explanation, a general dominance of IgA immunocytes at all secretory effector sites has now been well established for a couple of decades (5) . Moreover, the first direct evidence showing that these cells are pecuhar in that they produce polymeric {p)IgA (mainly dimers) rather than monomeric IgA, was provided by our laboratory in the early 1970s (6) ; but again the mechanism(s) driving this additional characteristic of mucosal B cells remains elusive.

The J chain pla/s a key role in secretory immunity In the late 1960s we found that not only pIgA but also pentameric IgM is enriched in exocrine body fluids because of active external transport (7). This secretory IgM (SIgM) was subsequently shown to be associated with SC and to follow the same intracellular route through secretory epithelia as SIgA (8, 9), Therefore, an epithelial transport model common for pIgA and pentameric IgM was proposed by our laboratory in 1973-74 (6, 10, 11) . A few years earlier, the "joining" or J chain had been identified as a unique polypeptide of approximately IS kDa shared by the two Ig polymers (12, 13). Only the heavy chains of IgA and IgM have the extra 18 amino-acid tailpiece containing a penultimate cysteine residue employed for covalent bonding of the J chain (14). Over the next decade, we obtained evidence suggesting that the J chain and transmembrane SC represent the "lock and key" in receptor-mediated external translocation of pIgA as well as pentameric IgM (15), a notion that has recently been firmly established (16-18). Because of its affinity to both ligands, transmembrane SC is now often referred lo as the polymeric Ig receptor (pIgR).

A key role for J chain in secretory immunity accords well with the fact that B cells subjected to terminal differentiation at secretory effector sites show prominent J chain expression regardless of concurrent Ig class production (19, 20) . In teleological terms, one would like to think that the intricate cooperation between the mucosal B-cell system and secretory epi-thelia has developed during phylogeny and is preserved in mammals because secretory antibodies are necessary for survival of the species.

The latter view is challenged by the fact that approximately two-thirds of suhjects with selective deficiency of IgA remain healthy -apparently even when living under poor hygienic conditions (21) -and gastrointestinal infections are more common in patients with generalized B-cell deficiency than in those with a selective IgA defect (22, 23). Therefore, non-specific innate defence mechanisms together with T-cell-mediated immunity and serum antihodies often afford sufficient mucosal protection (24). In addition, a compensatory SIgM response is regularly seen in the gut when IgA is lacking (22, 25). However, the absence of SIgA antibodies does predispose to recurrent upper airway infections, allergic disorders, autoimmunity, certain gastrointestinal infections and coeliac disease (22, 23, 26) . Also, a strikingly increased frequency of Crohn's disease has recently been observed in selective IgA deficiency (L. Hammarstrom, personal communication).

Altogether, it is difficult to evaluate the clinical role of secretory antibodies at mucosal surfaces. This is true also in subjects with a completely normal immune system, because a superimposed protective effect of concurrent systemic cellular and humoral immunity must always be considered (27, 28). We have now generated a pIgR/SC knockout mouse that completely lacks secretory immunity to better evaluate its overall role in mucosal defence (29).

Induction of secretory immunity and its critical role in infancy Mucosa-associated lymphoid tissue (MALT) appears to be of central importance to the induction, unique regulation and special disseminatioti that take place for mucosal B cells (30). In the upper aerodigestive tract, the best known such organized lymphoepithehal structures are the tonsils. In the gut. MALT includes particularly the aggregated lymphoid foUicles of the ileal Peyer's patches (Fig. I) but also solitary follicles which are especially numerous in the appendix and distal large bowel (31). The mesenteric lymph nodes are sometimes considered to be part of the gut-associated lymph tissue (GALT) as well, and this may be justified because they do to some extent mirror the B-cell biology of Peyer's patches as discussed helow.

The appearance of secretory antibodies in breast milk is a reflection of the MALT-mammary gland axis of B-cell migration (32, 33), and the protective value of breast-feeding is highlighted in relation to infections in the newborn period, particularly in the developing countries. Mucosal pathogens are now a major killer of children below the age of 5 years, being responsible for more than 14 million deaths of children annu- Fig, 1 . Human Peyer's patches. A. Patches in the terminal ileimi of a 10-year-old girl. A specimen has been excised from one of the patches (*). B. Histology of normal hnman Peyer's patch (haematoxylin-eosin staining) containing several activated lymphoid follicles (F) with germinal centres beneath the specialized follicle-associated epithelium (FAE) that lacks villi. Magnification: x22. ally Diarrhoeal disease alone claims a coll of 5 miUion children per year, or about 500 deaths every hour These sad figures document the need for mucosal vaccines to enhance surface defence against common infectious agents, in addition to advocating breast-feeding. Convincing epidemioiogical documentation suggess that the risk of dying from diarrhoea is reduced 14-24 times in nursed children (34, 35) . Indeed, exclusively breast-fed infants are better protected against a variety of infections (34, 36), atopic allergy (34, 3 7) and coeliac disease (38). Moreover, recent experiments in neonatal rabbits strongly suggest that SIgA is a crucial protective component of breast milk (39).

This review wili focus on the functional characteristics of B cells induced in human MALT, and on the role they play in human secretory immunity by co-operating with plgR/SCexpressing epitheha. A brief outline of the complex interac-tions that take place between the various elements of the adaptive mucosal defence system will first be given as a basis for subsequent in-depth discussions.

Humoral defence of mucosae

Immune exclusion is a term coined for non-inflammatory surface protection mediated by antibodies in co-operation with innate non-specific factors and thus refers to the "first line" of mucosal defence (Fig. 2) . This mechanism has a formidable task because the mucosal surface area is approximately 400 m-in an adult human, and is mostly covered by a vulnerable monolayered epithelium. Antihody activities in the epithelial mucus layer and serous secretions are mainly provided by SIgA and SIgM, hut there may he some contribution by serum-derived or views 171 / J 999 locally produced monomeric IgA and IgG (40), which can reach the surface quite rapidly by paracellular diffusion through the epithelium, particularly after some sort of mucosal irritation (41). In fact, this passive external antibody transfer most likely plays an important protective role at surfaces where there is little proteolytic degradation of antibodies, such as in the respiratory (32, 42) and female genital (43) tracts. In addition, secretory antibodies might reinforce immune exclusion by capturing antigen during their pIgR/SC-mediated transport through epithelial cells. Thus, while continuously monitoring the vulnerable surface monolayer, pIgA and pentameric IgM could neutralize viruses intracellularly and carry the pathogens and their products back to the lumen, thus avoiding any damage to the epithelium through cytolysis (44-47).

Regtilation of mucosal immunity takes place hoth in the organized inductive MALT structures (Fig. 2) and at the diffuse secretory effector sites of the mucosae such as the intestinal lamina propria and epithehum (on the right in Fig. 2 ). Particulate and infectious antigens appear to be primarily taken up by MALT through the follicle-associated epithelium (FAE) which contains "membrane" (M) cells particularly designed for sampling and inward transport of luminal material (27, 48) . By cognate interactions with antigen-presenting cells (APCs), naive T and B lymphocytes are primed to become memory and effector cells for subsequent dissemination ("homing") to effector sites (30, 33).

After intracellular processing to provide immimostimulatory peptides, antigen is usually presented to CD4+ T helper (Th) lymphocytes by professional APCs, either macrophages or more speciahzed dendritic cells (DCs). In addition, luminal immtmogenic peptides might be presented directly to sub-and intraepithelial T lymphocytes by epithelial cells (49). In the human gut, mucosal APCs, the FAE of GALT outside the M cells, and the small intestinal villous epithelium express surface determinants encoded by loci present in the class II region of the major histocompatibility complex (MHC). particularly HLA-DR (27, 30). and also classical and non-classical MHC class I molecules (49). To elicit specific immune responses, these gene products are required for appropriate antigen presentation. Class Il-positive B lymphocytes that abound close to the M cells, may also present antigens efficiently to T cells in cognate inununostimulatory or downregulatory interactions (30).

T cells subjected to stimulation in MALT express co-stimulatory molecules and release mediator substances (cytokines) which act on other lymphocytes in the microenvironment. Most adjacent B cells primed by such "first signals" migrate rapidly via lymphatics to regional lymph nodes where they may be subjected to further stimulation; they then mostly reach

[mmunologicd Reviews 171/1999 Fij. 3, Schematic depiction of two major adaptive inuniine mechanisms indnced in the gut. (I) Immune exclusion limits epilhelial colonization of pathogen.s and inhibits penetration of harmful foreign material. This first line of defence is principally mediated by secretory antibodies of the IgA (and IgM) class in co-operation with various nonspecific innate protective factors (not shown). Secretory immunity is preferentially stimulated by particulate antigens and pathogenic infectious agents taken up through M cells as indicated (see Fig. 2 ).

(2) Penetrating solnble dietary antigens (magnitude of uptake indicated) and thc normal microbial flora are less stimulatory for secretory Immunity (broken arrows) but induce, instead, suppression of proinflammatory humoral immune responses (IgG and IgE antibodies) as well as delayed-type hypersensitivity (DTH) mediated by activated T helper cells (CD4*) of the interferon-7-producing Th 1 subset. This complex and poorly defined phenomenon is called oral tolerance: ii may exert down-regulatory effects botli locally and in the periphery. peripheral blood and are finally seeded into distant secretory effector sites such as the intestinal lamina propria where they develop into Ig-producing immunocytes (30, 33). This terminal differentiation requires "second signals" that are modulated by available cognate antigen, various local cell types expressing MHC class II molecules and regulatory T lymphocytes (Fig. 2) . Most B cells that home from MALT, apparently belong to memory or effector clones of an early maturation stage; this is indicated by their proneness to express cytoplasmic J chain regardless of concomitant isotype production, althotigh the IgA class is normally predominant (30). J chain-containing pIgA and pentameric IgM are finally translocated to the lumen as SIgA and SIgM by a regulated receptor-mediated (plgR/SC-dependent) epithelial transport mechanism (see below).

Immune elimination and mucosal tolerance Multiple mechanisms are involved in removal of foreign material that has penetrated the epithelial barrier. Such immtine elimination represents a "second line" of mucosal defence and depends partly on serum-derived or locally produced antibodies of various isotypes. probably often operating in comhination with antibody-dependent cell-mediated cytotoxicity (ADCC), T cells, nattiral killer (NK) cells, and various phagocytes and DCs (Fig. 2) . As discussed below, clearance of antigens from the lamina propria may be performed by a non-inflammatory pIgR/SC-dependent mechanism, but is likely to be enhanced by a variety of non-specific biological amplification systems of innate immunity. Thus, if satisfactory immune elimination is not rapidly achieved, inflammation and immunopathology may develop, thereby giving rise to clinically overt mucosal disease (27). Such "frustrated" immunological antigen clearance is apparently part of the evolving pathogenesis of various chronic disorders, including coeliac disease and inflammatory bowel disease (IBD). The purpose of the secretory immune system is to avoid this adverse development, particularly hy excluding infectious agents (Fig. 3 ).

Mucosally induced tolerance is another strategy employed by the immune system to avoid hypersensitivity reactions against innocuous soluble antigens that gain access to the subepithelial compartment (50. 51). In the gut, this suppressive regulatory mechanism is best known as "oral tolerance" against dietary proteins, which are prohably mainly taken up throtigh the extensive epithelial surface covering the mucosal effector tissue (Fig. 3) , Mucosally induced tolerance is poorly understood and might involve several immunological events, although rapid removal of penetrating antigens from the lamina propria by poorly stimulatory DCs appears to be one important facet (50, 52-55). Chronic mucosal disorders such as coeliac disease and IBD are considered to reflect abrogation of oral tolerance to gluten and certain components of the indigenous microbiota, respectively (56, 57), Excessive local production of IgA and IgG antibodies to innocuous luminal antigens occurs in both conditions, but the relationship between secretory immunity and mucosal tolerance remains elusive (50).

Distribution of B lymphocytes and ig-producing immunocytes All normal secretory effector sites in adult humans contain a remarkable preponderance of IgA-producing immunocytes, including B-cell blasts and plasma cells (Fig. 4) . This is particu- Paired immunofluorescence in situ staining showed that the small naive B lymphocytes were mainly derived from the follicular mantle-zone population (CD19"^CD20*sIgD"^IgM+) surrounding the germinal centres (Fig, 6) . Conversely, most small lymphoid cells in the lamina propria were T lymphocytes, and with IgA on the snrface and/or in the cytoplasm (Table I) IgG-, IgD-and IgE-producing cdls IgG immunocytes constitute 3-4% in normal adult intestinal inucosa, but a considerably larger contribution is found in gastric and nasal mucosa (Fig. 4) , which often shows some low- Expression of / chain and cyiopJosmic SC offinity IgA immunocytes at secretory effector sites clearly differ from those found iu lymph nodes, spleen and bone marrow by showing a much more prominent synthesis of pIgA than of monomeric IgA (6, 20) . The presence of pIgA in their cytoplasm becomes immunobistochemically apparent by incubating tissue sections with purified free SC (Fig. 9 ). The SG binding site (and therefore the polymeric Ig structure) can thu's be shown to be generated at tbe cytoplasmic level by incorpora- Altogether, therefore, no conclusive information exists with regard to the proportion of monomers actually secreted by the intestinal pIgA-producing immunocytes.

Tbere are likewise discrepant opinions as to the nature of intracellular IgA in the mucosal immunocytes. We have maintained that the diffuse cellular in vitro binding of free SC (Fig. 9 ), together with the immunohiscochemical requirement for unmasking of cytoplasmic J chain (Fig. 10) , constitutes direct as a target for IL-2 and IL-5 regulation, thereby contributing to tbe induction of J chain mRNA and Ig production (129). In addition to IL-2 and IL-5, IL-6 has also been suggested to contribute to murine J chain upregulation (130, 131), whereas IL-4 appears to have an opposing effect (132). A specific enhancer sequence was recently identified 7-8 kb upstream of the start site of the gene, and inductive interactions were shown to depend on an IL-2 signal; this event triggered opening of the enhancer chromatin, allowing binding of activated STAT5 to a specific DNA element (133), Human studies Our immunohistocbemical observations suggest that also the human J chain production increases as a function of plasmacell differentiation. However, molecular information concerning the regulation of J chain expression during B-cell develop-ment in humans has been obtained mainly from leukaemic and EBV-transformed cell lines. Contrary to tbe situation in mice, these studies have suggested that transcription of the human J chain gene is initiated during early .stages of B-lineage differentiation, even before Ig production takes place (134-137). More recent data on normal cells have supported the apparent difference in J chain regulation between the two species. Thus, J chain RNA could be detected in human foetal liver hefore (i chain expression (]38), This result was confirmed and extended with haematopoietic subpopulations from human foetal and adult tissues (139) , In the bone marrow, transcripts for J chain were detected at all B-lineage stages, including the progenitor (CD19-CD34+)-and pro-B (CD19+CD34+)-cell subsets.

Interestingly, J chain RNA was also detected during human fetal thymocyte development, including the double-negative (CD4-CD8-) and single-positive (CD4+ or CD8^) subpopulations, but the transcription was turned off in peripheral CD3+ T cells from both foetal and adult samples examined with the same molecular method (139) , Similar studies have apparently not been performed on the various inductive MALT compartments. Altogether, therefore, much remains to be learned about the regulation of the human J chain gene.

Mucosal immune responses are helieved to be generated primarily in organized MALT structures which lack afferent lymphatics but, instead, are designed to sample antigens from mucosal surfaces (Figs! &.2). In mechanistic terms, the best studied such inductive sites are the GALT structures of experimental animals, including the ileal Peyer's patches and the appendix (27, 48, 140). The chief function of GALT is to provide adaptive immune protection for the gut, but its primed effector cells also migrate to other exocrine tissues such as the upper airway mucosae and the lacrimal, salivary, and lactating mammary glands (33). Additional immune responses are presumably elicited in bronchus-associated lymphoid tissue (BALT), palatine tonsils and other parts of the Waldeyer's pharyngeal ring in humans, including nasal-associated lymphoid tissue (NALT) -particularly the nasopharyngeal tonsil, also called adenoids (124, 141, 142). Vaccine development can only to some extent be based on the functional integration of the so-called "common" mucosal immune system, hecause accumulating evidence suggests that regionahzed regulation and homing properties of B ceUs must also be considered (30, 33). ImmunoiogicoJ Reviews 171/1999 patches of mice (160) and humans (Fig, 14 ). This regional difference migbt be involved in GALT recruitment of memory B and T cells with bigh levels of tbe counter-receptor integrin a4p7 (the so-called "mucosal homing receptor", or LPAM-1),

whereas a dominance of VCAM-1 in normal tonsils and peripheral lymph nodes could favour localization of a4|il'" (predominantly a4p7'") ceUs (Fig. 15) . Alternatively, such differential (191) . Also, the increased number of y/S IBLs in coeliac disease (56) migbt explain the significantly enlarged population of pIgA-produciiig immunocytes in the lesion (192) .

Additional region-specific differences could exist, for example microenvironmental levels of IL-7, a cytokine mainly produced by goblet cells in tbe gut and known to be a growth factor for human intestinal lymphocytes (193) .

As discussed above, the various MALT structures are considered to be the chief inductive sites of the secretory immune system in wbich recirculating virgin B lymphocytes arriving from the bloodstream are initially stimulated (on the left in Fig. 15 ). Gertain adhesion molecules are more strongly expressed on the naive than on the primed (memory/effector) subsets, and viceversa, and some are relatively tissue specific in their function.

show tissue specificity (194). Thus, in GALT and mesenteric lymph nodes, but not in peripheral lymph nodes, higb endothelial venules (HEVs) abundantly express MAdCAM-1 (Fig. 14) , CDl la/CD18) that binds to ICAM-1 and ICAM-2 (CD120) on the endotbelium (Fig, 17) , Exit of primed B cdls from The interfollicular zones of GALT are not only the site of entrance for lymphoid cells, but also where they exit through draining microlymphatics (on the left in Fig. 15 ). 'We identified these vessels in human Peyer's patches and appendix as thin- Our studies suggested that the a4p7''' subsets identified at exit in GALT, reflect the first homing step to furnish mucosal effector sites with primed lymphoid cells (Fig. 15 ). Relatively would not be expected to adhere to PNAd and only subsidiarily to VCAM-1; they were therefore most likely destined primarily Anti-o4[^7 (red), anti-CD38 (green), anti-cytokeratin (blue) for the intestinal lamina propria (Fig, 15) ,

Characteristics of lamina propria effector cells After antigen-induced activation, proliferation and partial differentiation in GALT, it is assumed that primed lymphoid cells go rapidly to mesenteric lytnph nodes, from which (after some further differentiation) they follow the lymph inlo the bloodstream (Fig. 2) . in experimental animals, it has been directly shown that IgA-expressing plasmablasts mature as they tnigrate from Peyer's patches via mesenteric lymph nodes and the thoracic duct to the intestinal lamina propria (222); the relative fraction of cells containing cytoplasmic IgA was shown to increase from initially around 2% through 50% and 75% to 90%, respectively (223).

In mice, peritoneal B cells are composed of a special B-1 (CD5+) subset that can repoptilate the gut with IgA-producing immunocytes (224). Kroese and co-workers have determined that approximately 5 0% of the murine intestinal lamina propria plasma cells are derived from this subset. It is generally believed that B-1 cells produce so-called natural or polyreactive antibodies encoded by unmutated (germline) IgV region genes, but murine CD 5* B cells often show hypermutation as a sign of selection (224), Human intestinal plasma cells from the duodenum. Ueum and colon have highly mutated igV region genes, suggesting ihal their BCR is sliaped by persistent antigenic challenge in germinal centres (22S), This is true of both the IgA and IgM immunocytes (226), thus documenting that SIgM antibodies may be part of a secondary intestinal immune response. Also interestingly, sequences of igVH region genes from Peyer's patch germinal centre B cells are reported to be clonally related to those of ileal plasma cells (227) , which substantiates the presumed intestinal homing pattern of GALTderived primed B cells (Fig. 15) , Altogether, despite the presence of polyreactive SIgA antibodies in human exocrine secretions (see below). there is no evidence that these antibodies are produced by plasma cells originating from the peritoneal cavity (228),

As mentioned above, homing of primed lymphoid cells to the intestinal latiiina propria appears to be determined mainly by their high levels of c(4p7 in the absence of L-selectin (Fig, 17B) . This phenotype can bind to unmodified MAdCAM-i expressed on the lamina propria microvasculature (33. 194). and fits with the predominant adhesion molecule profile of specific antibody-producing cells present in hitman peripheral blood after intestinal stimulation (229-231). Conversely, circulating specific B cells generated by systemic immunization show preferential expression of L-selectin but relatively low a4p7 levels (229-231), At present, interactions of human MAdCAM-1 with Lseleciin have apparently not been explored to the same extent as for the murine counterpart. Nevertheless, the virtual absence of lymphoid ceUs bearing the latter adhesion molecule in human intestinal lamina propria (Table i) . strongly suggests that MAdCAM-1. when expressed ouiside organized GALT structures, does not bind L-seiectiii (33). Many large B cells retain high levels of a4p7 after extravasation in the human intestinal lamina propria, despite abundant coexpression of CD38 (Fig. 18) and cytoplasmic IgA as signs of terminal maturation (Fig. 15) 

Lictle is known about factors triggering terminal B-cell differentiation a: various secretory sites (Fig. 2) , alth<jugh IL-5, IL-6 and IL-10 have been suggested to be important as discussed above (Fig. 16) . Also the local antigen repertoire appears to play an important role in site-specific accumulation of effector B cells, but without imposing any selectivity on their initial emigration. Tbus, in experimental animals GALT-derived blasts home to presumably antigen-free neonatal intestinal mucosa (232), and to foetal gut grafted under the adult kidney capsnle (222, 233). However, after some bours, antigen deposition clearly influences the observed homing pattern by causing specific accumulation (retention, proliferation and terminal differentiation) of IgA-producing cells (see below). Therefore, it appears that primed immune cells generally tend to home preferentially to the effector site tbat corresponds to the inductive site where they were initially stimulated.

Tbe observed impact of local antigen on the IgA system is most likely mediated largely via "second signals" from activated T cells. In addition to cognate exogenous antigens, several otber factors probably contribute to site-specific retention of memory T cells and their restimulation as well as rescue from apoptosLS (234). Sucb regional variables are not well defined but one of them could be the MHC expression level since the magnitude of immune responses migbt be related to the density of such "self-antigens" (235); tbus, their abundant epithelial expression -in combination witb only trace amounts of foreign antigens or anti-idiotypic antibodies -could elicit sufficient "second signals" for some terminal B-cell differentiation to take place. Interestingly, in salivary and lactating mam-mary glands, the IgA immunocytes notably accumulate preferentially adjacent to HLA-DR-expressing ducts (236-238),

The epithelium migbt also mediate more non-specific stimulatory effects on the B cells. For instance, a soluble factor from bursal epitbeiial cells has been reported to induce IgA production (239), perhaps representing one of tbe cytokines mentioned above (Fig. 16) . Furtbermore, local DC subsets and matrix components, including adbesion molecules such as a4p7 in the gut (see above), probably contribute, together with higb CD44 post-germinal-centre levels (240), to local retention of primed immune cells by virtue <Df affinity for extracellular matrix components. Site-specific survival differences of Ig-producing cells might directly or indirectly (via activated T cells) be influenced by similar variables (171), including rescue from apoptosis by contact witb stromal cells (241). Fig. 2 ), Putative regionalization in commtinication between inductive and effector sites is indicated, rhe lieavier arrows representing preferential B-cell migration pathways. Homing from gut-associated lymphoid tissue (GALT) is beiieved to be determintjd mainly by integrin a4ii7 on primed cells, interacting witb mucosal addressin cell adhesion molecule-1 expressed on the microvascular endothelium in tbe intestinal lamina propria (see Fig. 17B ).

As discussed in the text, other adhesion molecules appear to be employed by immune cells primed i.n bronchus-associated lymphoid tissue (BALT) and nasal-associated lymphoid tissue (NALT), which ill humans appears to be constituted of the various lymphoepithelial structures in the Waideyer's ring, including the paiatine tonsils and tbe nasopharyngeal tonsil (adenoids), Tbe urogenital tract migbt employ similar molecular homing mechanisms as the upper aerodigestive tract and tberefore appears to receive primed immune cells from NALT, althougb probably to some extent from GALT as weil. Also lactating mainmary glands appear to receive primed celis from both types of inductive tissue. The model is based on both human and animal data as discussed in the text. Colour key: green, B cells and IgA-producing plasma cells; violet, T cells; orange, antigen-presenting cells; bright yeiiow, epithelium; faint yellow, antigentransporting epitbelium, including M cells.

tion between the appendix and the large bowel mucosa is suggested both in experimental murine colitis (249) and in human ulcerative colitis, as disctissed elsewhere (57). Notably, the immediate natural availability of environmental antigens at various secretory effector sites also appears to be reflected in the local density of Ig-producing immunocytes (238); thus, this is about seven times higher in the mucosa of the human colon (which has au enormous microbial load) than in tbe parotid and lactating mammary glands (Fig. 19) . Although MALT-related antigens theoretically should be required to provide secondary B-cell stimulation throughotit the mucosa! immune system, it is not likely that such live or dead foreign material normally gains direct access to tbe parotid and mammary gland, particularly not to tbe latter organ. Interestingly, the lacrimal gland, wbich is connected by many short ducts to an excessively aeroantigen-exposed mucosa (the conjunctiva), shows an IgA immunocyte density approaching that of the colon (Fig. 19) . In this context it is of further interest that rapid and widespread dissemination of fed antigen, presumably carried by DCs, and sequestration of circulating specific T cetts as well as iheir activation a: sites of cognate antigen presentation, were recently observed in TCR transgenic mice (54) . That study, together with previous results of adoptive B-cell transfer in syngeneic rats (250), suggests that local antigen-driven T-cell activation might provide important "second signals" for retention, expansion and terminal differentiation of mucosal Ig-producing immunocytes at secretory effector sites, even at a long distance from mucosal surfaces.

Heterogeneity in mucosal homing mechanisms Tbe molecular mechanisms tbat coordinate migration of primed B cells from GALT to secretory tissues beyond the gut are poorly defined, although they would form tbe functional basis for several desired oral vaccines. In contrast to such GALTderived dissemination of secretory immunity to extraintestinai effector sites, migration to the gut lamina propria of B cells indticed in NALT and BALT appears to be negligible in terms of generating intestinal SIgA antibodies. This notioti is based on boming results obtained by local immunization or infection experiments in rodents (251, 252) and pigs (253). On the other hand, considerable indirect evidence summarized elsewbere (142) , suggests that dissemination of primed pIgA precursor cells takes place from Waldeyer's pharyngeal lymphoid ring, or NALT, to regional secretory effector sites botb in humans and various experimental animals (Fig. 20) . In humans, sucb putative boming dichototny between the aerodigestive tract and the gut, is furthermore strongly supported by the apparent disparate distribution of sIgD+IgM-CD38"*^ B cells, as identified by tracing of their heavy-chain Cfi-and S|i,-gene deletion (30, 176). As discussed above, this unique tonsillar centroblast subset most likely contributes significantly to the relatively frequent occurrence of IgD-produciug immunocytes uot only in the palatine tonsils and adenoids, but also iu the sahvary aud lacrimal glands as well as nasal mucosa, especially iu IgAdeficient individuals (Fig. 4) , Furthermore, activated human tonsillar B cells transferred intraperitoneally to mice with severe combined iuimunodeficiency, migrated to the lung but not to the gut mucosa (254), Putative mucosal homing molecules operating outside the gut Regionalization of the mucosal immuue system might iu part be explained by disparity in adhesion molecules expressed on the local microvascular endothelium aud lymphoid cells primed in different MALT structtires, aud perhaps by different local chemokine profiles. Animal and human studies have both demonstrated that MAdCAM-1 is shared between GALT structures and the intestinal lamina propria (Fig. 17) , and it also shows considerable expression on HEVs in mesenteric lymph nodes (194, 196) . Other, as yet unidentified, regional endotheiial determinants might be shared between the immune-inductive sites aud exocrine effector sites in the upper aerodigestive tract (Fiq. 20) . Interestingly, very little or no MAdCAM-1 can be detected on HEVs in normal human tonsils (T. Yamanaka, M. J. Brisikim, P Brandtzaeg, unpubhshed results), and this molecule appears to be abseut from regional exocrine tissues such as lungs and salivary glands, as well as from mammary glands and the human and murine uterus (19S, 255). Moreover, some evidence has been reported suggesting that a4p7 is not an important homing receptor for lymphoid cells in the airways of humans (256), mice (257) or sheep (258). Indeed, circulating specific antibody-producing cells, detected after nasal immunization in htunans, were reported to show considerable coexpression of a4p7 and L-selectin, in contrast to the results obtained after intestinal immunization (230). However, the a4p7 level was apparendy not sufficient to make the cells gut-seeking, whereas antibody production was evoked in both adenoids and nasal mucosa (259). Likewise, when the slgD+IgM" putative primed tonsillar effector B-ceU subset (identified by its Cp-aud S|i-gene deletion; see above) is observed in the circulation, its adhesion molecule profile shows rather little a4|37 but a relatively high and uniform level of Lselectin expressiou (P. Brandtzaeg A non-iutestiual homing receptor profile might also explain effector B-cell migration from NALT to the urogeuital tract (Fig, 20) . As reviewed elsewhere (43), this putative link is reflected by particularly high levels of specific IgA and IgG antibodies in cervicovaginal secretions of mice and monkeys after intranasal immunization with a variety of antigens. Also BALT could be involved as inductive sites in these experiments, especially in anesthetized mice (260), Other studies have shown that stimulation of GALT, particularly via the rectal route, cau induce antibodies in the female genital tract of rodeuts (43) and humans (261, 262), although the human vaccination results have been variable (247, 263). This problem might be caused by uncontrolled accessibility to the inductive tissue of the human large bowel (Fig, 20) , which is constituted by numerous scattered lymphoid folhcles (31) and perhaps the appendix (see above), Interestingly, an immunological link has been suggested to exist between the upper respiratory tract and rectal mucosa (247), but no difference iu the expression of a4p7 or L-selectin was observed for circulatiug antibody-producing cells resulting frotn rectal compared with oral immunization (230, 247).

Regionalization of the mucosal immune system must be further evaluated and eventually taken into account in the development of effective local vaccines to protect mucosae beyond the gut, such as the airways, the eyes, the oral cavity and the urogenital tract (3 3), Notably, even without employing interactions between a437 and MAdCAM-1, homing of putative early effector B-cell clones with preferential expression of J chain and pIgA, is just as remarkable to secretory tissues in the upper aerodigestive tract as to the intestinal lamina propria (20, 30, 238). It is difficult to understand how L-selectin and its major counter-receptor (PNad) might contribute to selectivity in B-cell extravasation at such effector sites of the upper aerodigestive tract when the same adhesion molecules are operating in organized peripheral lymphoid tissue.

A similar problem relates to the possibility of a significant involvement of a4p 1 on primed mucosal B cells that migrate to secretory tissues outside the gut (Fig. 15) , The chief counterreceptor for this integrin is VCAM-1. which is expressed on microvascular endothelium both in human bronchial and nasal mucosa (264, 265), and also in the murine uterus (255). However, no direct evidence exists to suggest that high expression of pi integrin directs B cells to the upper aerodigestive tract, lungs or urogenital tract. Altogether, the remarkably selective B-cell homing to various secretory effector sites, combined with the regionalization ofthe mucosal immune system, cannot as yet be fully explained in molecular terms. are also indicated to show the strikingly preferential appearance of IgA and, to a lesser extent, IgM in extertial secretions. This selectivity is particularly reniarkahle for cokistruni, which is ohtained directly from the duct opening, thus avoiding suhstantial contamination of IgG from interstitial fluid by leakage through surface, epithelia. The same is true for parotid secretion (see Fig, 8 ).

[Modified from (278) (Fig. 22) . As a consequence, immunohistochemical staining reveals a cytoplasmic distribution of SC and the two pig classes that is congruent except in the Golgi region (Fig. 23) where pIgR accumulates selectively (Fig. 24 ) before it migrates basolaterally to meet its ligands (10), The epithelial staining for IgA is normally stronger than that for IgM (8), both because of che relatively abundant local pIgA production (Fig. 4) and because tbe pIgR is less accessible to pentameric IgM (see below).

The pIgR is an epithelial glycoprotein (transmembrane SC) of -100 kDa which belongs to the Ig supergene family (182, 266) . Tbe first of its five extracellular Ig-like domains (DI) initiates tbe non-covalent ligand interaction at the basolateral epithelial-cell membrane, and a loop similar to the first complementarity determining region (CDRl) of IgVic domains has been shown to be essential for pIgA binding (266) , This region of DI shows a remarkably higb degree of sequence conservation, but the ability of pIgR to interact with pentameric IgM is nevertheless differently preserved among various species (266) . Thus, buman pIgR binds both pIgA and pentameric IgM, while the rabbit counterpart has virtually no binding capacity for the latter ligand. We interchanged DI between buman and rabbit pIgR, and found that human D1 was essential for pentameric IgM binding, and sufficient to confer sucb biuding to a chimeric pIgR witb rabbit D2-D5 (171). Furthermore, exchange between the human and rabbit pIgR regions comprising the CDR-like loops in DI, showed that the CDR2- lumen as SIgA by this mechanism every day than the total daily IgG (~30 mg/kg) production (274) . Tberefore, the intestinal mucosa is quantitatively tbe most important effector organ of adaptive humoral immunity. During the transport process, thẽ -80-kDa extracellular part of pIgR is incorporated into the secretory antibody molecules as so-called bound SC, thereby being "sacrificed" to confer protection against non-specific proteolytic degradation; this is particularly true for SIgA in which SC becomes covalently linked (32, 93, 17S-177) .

Excess of unoccupied pIgR is released apically by proteolytic cleavage in the same manner as SIgA and SIgM to form so-called free SC (Fig. 22) , Variable amounts of this fragment (~80 kDa, identical to bound SC) are usually present in exocrine secretions (278) , By equilibrium with the bound component, free SC exerts a stabihzing effect on the quaternary structure of SIgM in whicb SC remains noncovalently linked (8). In various ways, free SC in secretions tnigbt also contribute to innate mucosal defence (see below).

In normal adults, external secretions contain mucb more SIgA tban SIgM (Fig, 21) ; tbis difference cannot solely be accounted for by the striking predominance of local pIgA-producing cells (Fig, 4) , Thus, in well-con trolled quantitative ,studies of jejunal fluid and parotid saliva, the concentration ratio of IgA-to-IgM is found to be 2.4-to 4,9-fold higber than the corresponding local immunocyte class production ratio (7, 75, 76, 278, 279) .

This estimate is based on the observation of a fairly similar syntbetic rate in IgA-and IgM-producing cells (280) . Moreover, as discussed above, mucosal IgA plasma cells release variable amounts of monomeric IgA in addition to pIgA, whereas IgMproducing cells normally are virtually restricted to pentamer secretion (281) . Altogether, our calculations suggest tbat tbe exterual transport of pIgA is favoured at least 6-to 1 2-fold over tbat of pentameric IgM on a molar basis (Fig, 8) , Significant biological variables of secretory immunity other tban the local immunocyte distribution tberefore appear to exist. Such variables could be differences in tbe ligand affinity for pIgR. in tbe efficiency of receptor-mediated epithelial transcytosis of the Anli-SC (red seperation) Anti-SC + anti-lqA (qreen Anti-I (red) + anti-A (green) Anti-I (red) + antJ-A (green) Possible ligand-enhanced transport The regulation of human pIgR appears to differ from that ofthe rabbit and rat receptors, for which transcytosis is stimulated by bound ligand. Thus, rabbit pIgR complexed with buman pIgA was transcytosed somewhat faster than unoccupied or anti-SC (Fab)-reacted receptor (283, 284) , Sucb ligand-mediated enhancement ofthe pIgR transport rate was also noted to operate for tbe rat receptor in vivo (285) . Contrary to these observations, we found that the human pIgR transcytosis rate was not influenced by binding of eitber pIgA or pentatneric IgM at concentrations reported to stimulate rabbit pIgR transcytosis (282). Thus, our results suggested tbat different levels of receptor saturation do not influence the rate of epithelial antibody transport mediated by the human pIgR (Fig. 26) , Altogether, species differences seem to exist with regard to regulation of pIgR transcytosis. It sbould be noted, however, tbat aithougb huinan pIgA was found to stimulate this process when mediated by the rabbit pIgR, the reported effect was rather small and most pronounced witb a mutant receptor tbat showed reduced efficiency of apical sorting (283, 284) , Tberefore, the biological significance of such ligand stimulation is questionable. Importantly substantial amounts of buman pIgR are continuously transcytosed across secretory epitbeliai cells in vivo even without any bound ligand (286), This observation has been further substantiated by comparing tbe level of SC in parotid saliva from IgA-deficient and hypogammaglobulinaemic subjects with normal coutrcils; the amount of free SC (plus IgM-bound SC) in the former two categories of secretion was found to be of tbe same magnitude as that of free and bound (mainly SIgA-associated) SC in normal controls (278) . 

Interactions of SlgA and free SC It has been claimed that SIgA can perform ADCC and promote phagocytosis via FcaRI present on macrophages and granulocytes. However, this suggestion is poorly documented, and it is difficult to assess the possible outcome in terms of an inflammatory reaction, for instance caused by TNF-a release as discussed above. Interestingly, potentially tissue-damaging eosinophils possess not only FcaRI but apparently also a receptor for free and bound SC (334), This probably explains that SIgA exhibits particularly strong eosinophil-degranulating properties (327, 335) and in this respect appears to be more potent than cross-linked IgE antibodies, which instead may be more involved in the recruitment and priming of these cells (336). Also, SIgA has been shown to induce degrannlation of IL-3primed basophils (33 7). Thus, by interacting with eosinophils and basophils (perhaps also mast cells?), SIgA provides a proinflammatory potential of the secretory immune system when immune exclusion fails, such as in parasitic mucosal infestations. However, it is at present difficult to know how the balance between SIgA and free SC may influence eosinophil activation, because free SC in solution (but presumably not on a surface) may be blocking in this respect (334, 335).

Circulating IgG antibodies to food antigens are often present in normal individuals, although with a remarkably varying subclass contribution (338), Because approximately 50% of the circulating pool of IgG is extravascularly distributed, these antibodies will he well represented locally in the intestinal mucosa, Parenteral immunization eliciting high levels of IgG antibodies may reduce mucosal antigen penetration (27), although some studies have reported increased uptake (339) , Our in vitro results showed that IgG antibodies could retard mucosal penetration of the corresponding antigen; conversely, the penetration of an unrelated bystander macromolecule was significantly enhanced (318) . A similar detrimental side effect of IgG antibody has been observed in vivo (340). This is probably explained by local formation of complement-fixing IgG-containing immune complexes (27) and activation of mast cells (339) . The proinflammatory potential of maternal IgG present in the intestinal mucosa ofthe newborns, or of locally produced IgG antibodies, may be less important in infants who are breast-fed because milk SIgA antibodies, which have little or no complement-activating capacity (292). will exert a non-inflammatory blocking effect (311).

As long as the mucosal homeostasis remains in control, IgG-mediated imtnune reactions occurring in the mucosa will not cause clinical symptoms and may be viewed as a "second line" of defence (Fig, 29) , However, in subjects with a pronounced regional IgE response to etivironmental or dietary antigens, immune-mediated degranulation of mucosal mast cells and activation of eosinophils are likely to result in a massive release of potent inflammatory mediators overruling the IgA-and tolerance-mediated local homeostatic mechanisms. Such "parasite-directed" pathotopic potentiation ofthe second line of mucosal defence is probably the hasis of overt atopic mucosal allergy (341), and may be considered as a breakdown of mucosally induced tolerance. In IBD (ulcerative colitis and Crohn's disease), mucosal homeostasis is severely altered with signs of immunological hyperactivation, downregulated J chain expression and substantial local overproduction of IgG, including antibodies to gut bacteria (57), Together, these features are believed to signify break of oral tolerance to antigens from the indigenous bacterial flora accompanied by less restricted extravasation of leukocytes.

This review has focused on the induction of mucosal B cells and their effector functions after terminal differentiation to immunocytes. However, it is becoming increasingly evident that B cells also can exert important immunoregulatory activities both as efficient stimulatory APCs (342) , tolerizing APCs (343) , and by secreting a variety of cytokines (344), In the mucosal immune system such B-cell characteristics remain elu-sive (30), Nevertheless, B cells located at the presumed sites of antigen entrance in GALT express the necessary costimulatory molecules lo function as APCs, perhaps leading to enhanced diversification of mucosal immutie responses, thereby explaining the observation that SIgA antibodies show broader specificity than comparable serum antibodies (30), Such putative diversification of secretory immunity is reflected by the delelions and insertions observed in the CDR regions of IgV genes analyzed from human intestinal immunocytes (226), Are local IgA production and mucosally induced tolerance Interrelated?

The relationship between induction of intestinal IgA responses and oral tolerance remains rather enigmatic (50). Experiments in CD8 knockout mice have suggested that this phenotype of T lymphocytes (the predominant IEL subset) is crucial for downregulation of the mucosal B-cell system (345). The tone of hyporesponsiveness in the intestinal immune system appears to be quite robust, because even a strong immunogen such as cholera toxin (CT) is unable to abrogate it although oral tolerance cannot be induced in the presence of CT (345), On the other hand, TGF-ji has been shown to be important to promote IgA switching (Fig. 16 ) also in mice immunized with CT (346), and this cytokine is in addition believed to be one ofthe major mediators of OT3.\ tolerance in murine test systems (5 1), It is as yet not possible to extrapolate such apparently contradictory information to the human mucosal immune system.

It is well established in the human gut that the mucosal immune system responds to infection with local IgA and IgM production (347) , and it appears that the level of this response may determine whether clinical symptoms will occur or not (348), In experimental animals, antibody-dependent immune exclusion has been shown to operate even for small molecules such as chemical carcinogens (349, 350), However, a rational basis for manipulation of local immunity by vaccines is still not satisfactorily established. Altogether, tbe following facts and cjpen questions can be summarized about the human mucosal B-cell system: 1) Secretory immunity depends on an intimate cooperation betvi'een mucosal B cells and the pIgR/SC-expressing epithelia. The obvious biological significance of the striking J chain expression shown by dissetninated MALT-derived immunocytes is that IgA and IgM polymers with high affinity for pIgR/SC can be produced at secretory effector sites and become readily available for active external transport. This important Immunological Reviews ] 71 /1999 functional goal in terms of cional differentiation, appears to be sufficient justification for the J chain also to be expressed by B cells terminating at such sites with IgG or IgD production; rhese imtnunocyles may be considered as a "spin-off from early effector clones tbat tbrough isotype switching are on their way to pIgA expression, 2) There is considerable evidence to support the notion that intestinal immunocytes are largely derived from B cells initially induced in GALT, However, insufficient knowledge exists concerning the relative importance of M cells, MHC class IIexpressing epithelial cells, B cells and other professional APCs in the transport, processing and presentation of luminal antigens that take place in GALT to accomplish the extensive and continuous priming and expansion of mucosal B cells. Also, it is not clear how the germinal-centre reaction in GALT, compared with other parts (jf MALT such as the tonsils, so strikingly promotes preferential isotype switching to IgA and a high level of] chain expression.

3) Although the B-cell migration from GALT to tiie intestinal lamina propria is guided by rather well-characterized adhesion molecules, the chemotactic stimuli involved in extravasation and microcompartmental distribntion of various B-cell subsets remain elusive. Also, the homing mechanisms of mucosal B cells appear to be remarkably regionalized, but they remain to be defined in the upper aerodigestive tract, 4) Retention and acctitnulation of B cells extravasated at secretory effector sites are influenced by antigen-driven local proliferation and differentiation. However, the relative roles of stromal T cells and IELs, MHC class Il-positive APCs and epithelial cells in providing the necessary stimulatory signals for proliferation and terminal differentiation of the local B cells are unknown, 5) The mucosal barrier normally allows some penetration of intact soluble antigens so there is probably always a need for immune elimination in the lamina propria. If immune exclusion is impaired (e.g. in IgA deficiency), or if there is too large an antigen load on the epithelial barrier (e.g, in chronic infection) , activated nonspecific amplification mechanisms involved in immune elimination may cause hypersensitivity which is observed clinically as mucosal disease. This immunopathological development may he mediated by proinflammatory IgG and IgE antibodies as well as by hyperactivated APCs and T cells. Although these immunopathogenic mechanisms are rather well understood in several types of intestinal disorders such as atopic food allergy and coeliac disease, the cause of their initiation, possibly involving abrogation of oral tolerance, generally remains unexplained. 6) Clinical observations in immunodeficient patients have shown that SIgA, SIgM and IgG are not the only important components of the intestinal mucosal defence system. It is becoming increasingly evident that innate immunity is crucial and much more complex than previously believed; the cooperation between innate and adaptive mucosal immunity needs exploration to better understand how homeostasis of mucous membranes normally is maintained. 6 

The IgA system, I. Studies of the transport and iminunochoniLstry of IgA in the saliva

Hereiiians JE The normal human intestinal mucosa as a major source of plasma cells coniainiug, A -i m m inio giobul in

Immunohistochemical studies of tlio formation and epithelial transport of immunogiobulins in normal and diseased human Iniestinal mucosa

RugtveitJ, Bakka A, Brandtzaeg P Differential distribution of B7,l (CD80) and B7,2 (CD86) CO-.stimulatory molectdes on mucosal macrophage subsets in human inflammatory bowel disease (IBD)

Induction of rapid T cell activation and tolerance by systemic presentation of an oraiiy administered antigen

Van der Heijden P), Stok W, Bianchi AT Contributionofimmunoglobulin-secreting cells in the murine small intestine to the total 'background' immunoglobulin production

Expression of iinmunoglobulin isotypes by lymphoid celis of mouse intestinal lamina propria

Selective loss of S-phase cells when making cell suspensions from iymphoid tissue

Distribution of IgAf andIgA2 plasma cells in various normal human tissues and in the jejunum of plasma IgA-deficient patients

Different subclass distribution of IgAproducing cells in human lymphoid organs and various secretory tissues

Increased jejunal secretory IgA and IgM in ankyiosing spondyhtis: normalization after treatment with sulfasalazine

Biological significance of IgAl proteases in bacterial colonization and pathogenesis: critical evaluation of experimental evidence

lL-2 and IL-5 both induce ji, and J chain mRNA in a cional B cell line, but differ in their cell-cycle dependency for optimal signaiing

Precursor B ceils transformed by Epstein-Barr virus undergo sterile plasma-cell differentiation: J-chain expression without immunoglobuiin

The J chain gene is transcribed dtiring B and T iymphopoiesis in humans

Brandtzaeg P Immunoiogical functions of adenoids

Brandtzaeg P Regionalized immune function of tonsils and adenoids

Eollicular dendritic ceil (FDC) precursors in primary lymphoid tissues

B Iympiiocytes induce the formation of foiiicuiar dendritic ceil clusters in a iymphotoxin a dependent fasiiion

Targeted gene disruption reveals a role for natural secretory IgM in (he maturation ofthe primary immune response

A putative ciiemokine receptor, BLRl, directs B ceil migration to defined lymphoid organs and specific anatomic compartments ofthe spleen

B lymphocyte che mo taxis reguiated in association with microanatomic localization, differentiation state, and B cell receptor engagement

Germinal center B cells rescued from apoptosis by CD40 ligation or attachment to follicuiar dendritic celis, but not by engagement of surface immunoglobuiin or adhesion receptors, become resistant to CD95-induced apoptosis

Memory B cells from human tonsils colonize mucosal epithelium and directiy present antigeti to T cells by rapid up-regulation of B7-i and B7-2

Casali P CD40 engagement triggers switching to IgAl and IgA2 in huinan B cells through induction of endogenous TGP-P; evidence for TGF-p but not IL-10-dependent direct Sft->Sa and sequential Sn-»Sy, Sy-^Sa DNA recombination

Compartmeiitahzation of ihe human mucosai immune system, A tonsiiiar B-cei! subset (sIgD+IgM-CD38-) homes to regional effector sites but not to the gut [Abstract |, Scandjlmmunol i998:47

A model for the differentiation of B iymphocytes wilh implications for the bioiogicai role of IgD

Brandtzaeg P Gluten-specific, HLA-DQ restricted T cells from coeliac mucosa produce cytokines with Th 1 or ThO profile dominated by interferon

Aanesen JP Brandtzaeg P Piienotypic distribution of T ceiis in human nasai mucosa differs from that in the gut

Brandtzaeg P Secretory immunity in ceiiac disease: ceiiuiar expression of immunogioi:)uhn A subclass and joining chain

Human mucosai addressin ceil adhesion molecule-1 (MAdCAM-1) demonstrates structurai and functionai simiiarities to the a4j37-integrin binding domains of murine MAdCAM-1, but extreme divergence of mucin-iike sequences

Brandtzaeg P Topographic distribution of homing receptors on B and T ceiis in iiuman gut-associated iympiioid tissue, Reiation of Lseiectin and integrin 0(4(57 to naive and memory piienotypes

Mice lacking expression of secondary lymphoid organ ciiemokine Iiave defects in iympiiocyte homing and dendritic cell localization

Sanchez-Madrid F. dc Landazuri MO, (x4P7 integrin mediates B ceii binding to fibronectin and vascular ceil adhesion moiecuie-i: expression and function of K4 integrins on human B lymphocytes

Siiaw S, Cords, channels, corridors and conduits: critical ardiicectural elements facilitating cell interactions in the iymph node cortex

Distribution of lymphatics in human palatine tonsiis; a study by enzyme-histochemistry and scanning electron microscopy of lymphatic corrosion casts

HubE, Rot A, Binding of RANTES, MCP-l, MCP-3, and MlP-la to celis in human skin

Lymphocyte migration into tissue: the paradigm derived Irom CD4 subsets

A fundamental subdivision of circulating lymphocytes defined by adhesion to mucosai addressin ceii adhesion moiecuie-1, Comparison with vascular ceii adhesion moiecuie-i and correlation with |57 integrins and memory differentiation

Intraperitorieal immunization of human subjects with tetanus toxoid induces specific antibody-secreting cells in the peritoneal cavity and in the circulation, but fails to ehcit a secretory IgA response

Antitoxic ciiolera immunity in mice: infiuence of antigen deposition on antitoxin-containing ceiis and protective immunity in different parts of tile intestine

Cellular kinetics of the intestinal immune response to cholera toxoid in rats

A recomhinant Salmonella typiiimurium vaccine induces locai immunity by four different routes of imm.unization

Differences in immune responses induced by oral and rectal immunizations with Salmonelia typiii Ty2 la: evidence for compartmentalization within the common mucosai immune system in humans

The roie of non-B ceiis in locahzing an IgA plasma ceii response in tiie intestine, Reg lmmunoi i990-9l:3:336-340, 251, McDerniottMR, Bienenstock J, Evidence for a common mucosal immunologic system. I, Migration ofB immunoblasts into intestinai, respiratory, and genitai tissues

Contribution of antibody-secreting celis induced in mucosai iympiioid (issues of pigs inocuiated with respiratory or enteric strains of coronavirus to immunity against enteric coronavirus chaiienge

Differential expression of homing moiecuies on recirculating lymphocytes from sheep gut, peripheral, and lung iymph

Colonization in the rectum and uterine cervix with group B streptococci may induce specific antibody responses in cervical secretions of pregnant women

Expression of endotheiial and leukocyte adhesion molecules interacellular adhesion molecule-1, E-selectin, and vascular ceii adiiesion moiecuie-1 in the bronchiai mucosa in steady-state and aliergen-induced asthma

Brandtzaeg P Reguiation ofthe formation and epitheiiai transport of secretory immunogiobuiins

IgA is complexed with secretory component (SC) on the surface of human intestinal epitiieiial ceils

A transmembrane precursor of secretory component, Tiie receptor for transceiiuiar transport of poiymeric immunogiobuiins

Brandtzaeg P Itnmimoliisiocheinicai studies on various aspects of glandular immutioglobuiin transport in man

Intestinal secretion of IgA and IgM; a hypothetical model

Specific and nonspecific iiumoral defense factors in the epithelium of normal and inflamed gastric mucosa, Immunohistochemical localization of immunogiobuiins. secretory coinponent, iysozyme, and iactoferrin

Fine-specificity of ligandbinding domain 1 in the polymeric immunoglobuiin receptor: importance of the CDR2-containing region for IgM interaction

Lindh E. Increased risistance of immunogiobulin A dimers to proteoiytic degradation after binding of secretory component

Secretory component deiays ihe conversion of secretory IgA into antigen-binding competent F(ah')2: a possibie implication for mucosal defense

Brandtzaeg et al • Tbe B-cell system of buman mucosae and exocrine glands 277. BerdozJ, BlancGT,ReinhardtM,Kraehenbuh] JP Corthesy B, in titro comparison ofthe antigen-binding and stability properties of the various molecuiar forms of IgA antibodies assembled and produced in CHO cells

Secretion of immunogiobuiins and piasma proteins from tiie colonic mueosa: an in vivo study in man

singie ceils from human tonsii

Brandtzaeg P Mechanism for eniianced external transfer of dimeric IgA over pentameric IgM, Studies of diffusion, binding to the human polymeric Ig receptor, and epitheliai transcytosis

Stimuiation of transcytosis ofthe polymeric immunogiobuiin receptor by dimeric IgA

Dimeric and tetrameric IgA are transcytosed equally by the polymeric Ig receptor

Secretory component in immunogiobulin deficiency and immun oelectron microscopic study of intestinal epithelium

IgM tiie secretory immunoglobulin of reptiles and amphibians

Evidence for a cutaneous secretory imm.une system in rainbow trout (Salmogairdneri)

Anti-inflammatory activity of human IgA antibodies and their Fab ci fragments: inhibition of IgG-mediated complement activation, EurJImmunoi i989:19

Constitutive and cytokine induced expres,sion of HLA molecuies, secretory component, and interceiiular adhesion moiecuie-1 is modulated by bucyrate in the colonic epithdiai cdl line HT-29

The poiymeric immunoglobulin receptor (secretory component) in a human intestinal epithelial cell line is up-regulated by interleukin-1

Brandtzaeg P Different regulatory pathways employed in cytokine-enhanced expression of secretory component and epithehal HLA dass I genes

Brandtzaeg P A composite DNA eiement in the promoter ofthe polymeric immunoglobulin receptor promoter regulates its constitutive expression

Characterization ofthe 5'-flanking region ofthe murine polymeric IgA receptor gene

Brandtzaeg P Bakiien K, Characterization of the IgA-immunocyte popuiation and its product in a patient witii excessive intestinal formation of IgA

Natural polyreactive secretory immunoglobulin A autoantibodies as a possible barrier to infection in humans

Protein Fv (Fv fragment binding protein); a mueosai human superantigen reacting with normal immunogiobulins

Free secretory component and Iactoferrin ol human milk inhibit the adhesion of ciiterotoxigenic Eschericiiia coii

SpsA, a novei pneumococcai surface protein with specific binding lo secretory immunoglobulin A and secretory component

Solid phase radioimmunoassay for detection of circuiating food protein antigens in human serum

Inhibition of receptordependent and receptor-independent generation ofthe respiratory burst in human neutrophils and monocytes hy human serum IgA

SoothilJ JR Transient IgA deficiency and pathogenesis of infantile atopy

Some intrinsic and extrinsic factors predisposing lo aliergy

Rag weed-specific antibodies in bronchoalveolar lavage fluids and serum before and after segmental lung ciiailenge: IgE and IgA associated with eosinopiiii degranulation

Rapid transepithelial antigen transport in rat jejunum: impact of sensitization and the hypersensitivity reaction

Do B cells drive the di versification of immune responses?

Lack of local suppression in orally tolerant CD8-dericient mice reveals a criiical reguiatory role of CD8' T ceils in the normal gut mucosa

Increased jejunai IgA synthesis in vitro during acute infectious non-bacterial gastroenteritis