Immunoglobulin A (IgA), as a major serum immunoglobulin and the predominant antibody class in the external secretions that bathe mucosal surfaces, plays key roles in immune protection. Indeed, the body expends considerable energy in producing IgA, such that the daily production of IgA exceeds that of all the other antibody classes combined. This rate of production suggests that, at least from an evolutionary standpoint, the benefits provided by IgA in terms of immune defence must be considerable, in order to outweigh such energy costs to the body.
IgA, at concentrations of about 2–3 mg/ml, is the second most prevalent antibody in serum after IgG, which is normally present at about 12 mg/ml. Since serum IgA is metabolized some five times faster than IgG, the production rates of serum IgA and IgG must be similar. Whilst serum IgA is predominantly monomeric in nature, the IgA in secretions (secretory IgA, S-IgA) is chiefly polymeric, comprising mainly dimeric forms (Figure 1). S-IgA serves a variety of functions to protect the vast surface area (approximately 400 m2) occupied by mucosal surfaces, such as the linings of the respiratory, gastrointestinal, and genitourinary tracts. Together, these mucosal surfaces represent an enormous area of potential exposure to inhaled and ingested pathogens. As the major class of antibody present at these sites, S-IgA can be considered an important first line of defence against many invading pathogens.
Like all immunoglobulins, the monomeric structural unit of IgA comprises two identical heavy chains and two identical light chains arranged into two Fab regions and an Fc region, separated by a flexible hinge region (Figure 1). The paired variable regions at the tips of the Fab arms are responsible for antigen recognition, while the Fc region mediates interaction with various receptors and effector molecules. The heavy and light chains fold up into globular domains, each with a characteristic tertiary structure of anti-parallel β sheets termed the immunoglobulin fold. With the exception of the upper domains of the Fc region (the Cα2 domains), the domains are arranged in pairs, stabilized by numerous non-covalent trans interactions.
In humans, two subclasses of IgA, termed IgA1 and IgA2, exist, each the product of a separate gene. Numerous sequence differences are found in their heavy chain constant regions (ie Cα1, hinge, Cα2, and Cα3) (Figure 1). A major difference between the subclasses lies in the hinge region, which is greatly extended in IgA1. Two allotypic variants of IgA2, named IgA2m(1) and IgA2m(2), have been well characterized (Figure 1). Their heavy chain constant region sequences differ at a number of points along their length. The most notable difference relates to the fact that while IgA2m(2) has the disulphide bridges linking light and heavy chains typical of most immunoglobulins, these are generally lacking in IgA2m(1). Instead, the light chains bond to each other and the association with the heavy chains is stabilized through non-covalent interactions. A third possible allotype, IgA2n, has also been described 1.
Both subclasses carry a number of N-linked carbohydrates, contributing 6–7% of molecular mass in IgA1 and 8–10% in IgA2 2 (Figure 1). The hinge region of IgA1 also bears three to five short O-linked sugars attached to Ser and Thr residues 3, 4. A recent report indicates that 5–10% of serum IgA1 carries a sixth O-linked sugar 5.
Early electron microscopy studies provided the first insights into the shape and size of the different forms of IgA 6, 7. More recently, molecular models for human IgA1 and IgA2m(1) based on X-ray and neutron scattering data have been generated 8, 9 (Figure 2). IgA1 may have a more extended ‘reach’ than IgA2m(1) because the models predict that the tips of the Fab arms (ie antigen binding sites) of IgA1 can be spaced at much greater distances apart than those of IgA2m(1). Hence IgA1 may be able to interact simultaneously with two antigen molecules separated by a considerable distance, while IgA2m(1) may have a more limited capability in this respect. Such a capacity may afford IgA1 advantages in higher avidity recognition of repeated antigenic structures spaced along the surface of certain pathogens.
A more detailed view of the Fc region of IgA1 has recently emerged from the solved X-ray crystal structure of the complex of human IgA1 Fc with the extracellular domains of the human IgA-specific receptor, FcαRI 10 (Figure 3). Overall, the IgA1 Fc structure resembles that of IgG and IgE, but the positions of N-linked sugars and disulphide bridge arrangements are different.
The majority of IgA in the secretions is polymeric, mainly in the form of dimers comprising two IgA monomers linked to one molecule of J (joining) chain, a 15 kD polypeptide also present in polymeric IgM (Figure 1). J chain forms disulphide bridges to both IgA monomers via the penultimate Cys residues of their tailpieces, the C-terminal 18 amino acid extensions which are present on IgA heavy chains but lacking from the heavy chains of Ig classes that do not polymerize (eg IgG) 11–13. An additional component of S-IgA is secretory component (SC), a polypeptide of ∼80 kD comprising the extracellular proteolytic fragment of the receptor responsible for transport of dimeric IgA into the secretions, namely the polymeric immunoglobulin receptor (pIgR) (Figure 1). This receptor is expressed as an integral membrane protein on the surface of epithelial cells lining mucosal membranes and binds newly synthesized polymeric IgA at the basolateral surface. The receptor–IgA complex is endocytosed and moves across the cell through a series of vesicles to be delivered to the apical surface (Figure 4). The extracellular, IgA-binding portion, which is disulphide-bonded to the polymeric IgA and comprises five Ig-like domains, is then cleaved and the whole complex is released as S-IgA 14.
Electron microscopy, in the early 1970s, revealed that S-IgA has a double-Y shape, comprising two IgA monomers linked end to end via their Fc regions. The angle between the Fab arms was seen to vary widely, consistent with significant flexibility in the molecule. While there remains a lack of structural information on J chain, recently the crystal structure of the first N-terminal domain (domain 1) of human pIgR was determined 15. In the first step in binding of the receptor to dimeric IgA, this domain is known to make a non-covalent interaction with the Cα3 and possibly the Cα2 domain of dimeric IgA 16–19. The domain has a structure similar to that of Ig variable domains, and residues implicated in IgA binding lie in a conserved helical turn in the equivalent of the first complementarity determining region (CDR1) 20–22.
Sites of IgA synthesis
S-IgA is the product of local synthesis at the mucosal surfaces, which serve as the main source of antigenic material for the body. The mucosal immune system such as the gut-associated lymphoid tissue (GALT) is highly specialized and functions largely independently of the systemic immune system. Although all parts of the mucosal immune system share common characteristics, the actual lymphoid structures vary from tissue to tissue; for example, Peyer's patches are major components of GALT in humans.
Although the importance of IgA in mucosal secretions is well established, it is clear that in humans a proportion of the IgA is secreted directly into the blood and plays no role at mucosal surfaces. Serum IgA is predominantly (∼90%) monomeric IgA1 that is produced in the bone marrow, while in external secretions most of the locally produced IgA is polymeric with a relative increase in the proportion of IgA2. The lymphocytes that produce monomeric or polymeric IgA of IgA1 or IgA2 subclass are characteristically distributed in various lymphoid and non-lymphoid tissues. The differential interaction of monomeric and polymeric IgA molecules with various cells leads to their selective distribution in body fluids and possibly to differences in their effector functions. Secretory and serum IgA are therefore molecules with different biochemical and immunochemical properties produced by cells with different organ distributions. Different methods of immunization can induce serum or secretory IgA responses or a combination of both.
It appears that in mice, secretory IgA in the gut comes from two sources 23, 24. Around 75% is from B2 lymphocytes in organized germinal centres of mucosal lymphoid tissues such as Peyer's patches. This IgA production is T lymphocyte-dependent. A second source, possibly contributing around 25% of the secretory IgA, is produced by B1 lymphocytes that develop in the peritoneal cavity and are distributed diffusely in the intestinal lamina propria. This IgA might represent a primitive T lymphocyte-independent source of IgA recognizing gut bacteria. However, there are important differences between mice and humans in this respect. In humans, there is no evidence that precursors of human IgA-secreting plasma cells are induced or expanded in the lamina propria 25. Thus, in humans, intestinal secretory IgA originates only from organized gut-associated lymphoid tissues.
The very high concentration of secretory IgA in human colostrum and milk suggests strongly that it must play an important role in the passive (and possibly active) immune protection of the newborn and there is overwhelming evidence that this is the case 26. Breast-feeding markedly decreases infant death not only from gastrointestinal, but also from respiratory infection. Mothers' milk contains S-IgA against a wide variety of microbial antigens and these antibodies are able to neutralize toxins and viruses. Since there is very little transfer of intact protein across even the newborn gut, it appears that the role of milk-derived S-IgA must be as a molecular ‘paint’ neutralizing the effect of micro-organisms by prevention of adherence and other processes. This passive protection against a range of micro-organisms has also been demonstrated in adults and in many animal studies 27.
The same passive protection by IgA secreted by the mucosal system must, of necessity, play a central role in the protection of mucosal surfaces in general. The levels of complement proteins and the numbers of leucocytes are generally low in secretions. As a result, the powerful immune effector functions that they mediate, and which are so important in protecting internal tissues and fluids, are much less important at normal mucosal surfaces, though this might not be the case in inflammatory situations and where there is internal bleeding. There is increasing evidence that serum IgA is able to trigger effector functions that have the potential to destroy micro-organisms and mammalian cells. It has long been recognized that compared with IgM and IgG, IgA is a poor activator of complement. IgA does not activate the classical pathway and its role in activation of the alternative pathway remains controversial. Recently, it has become apparent that polymeric IgA upon binding mannan binding lectin (MBL) can activate the lectin pathway 28. The significance of complement activation by IgA in vivo remains unclear. Since IgA is a poor activator of complement, in a situation where there is limited antigen, IgA can actually inhibit complement activation by blocking binding of IgG or IgM antibodies that would be more potent at activating complement.
Receptors specific for IgA are expressed on a diverse range of cell types. On epithelial cells, pIgR is critical for the transport of polymeric IgA into mucosal secretions, while other cell types including neutrophils, macrophages, monocytes, and eosinophils express receptors for the Fc portion of IgA. IgA has also been reported to bind to T cells, B cells, and natural killer (NK) cells, but the nature of the receptors involved remains unresolved.
Human FcαRI (CD89) is constitutively expressed on neutrophils, monocytes, eosinophils, some macrophages, interstitial dendritic cells, and Kupffer cells. In blood and tissues, neutrophils make up the majority of FcαRI-positive cells 29. Peritoneal fluid and bronchoalveolar lavage fluid were found to contain newly emigrated CD89-positive macrophages, while during inflammation of the intestine there are major influxes of CD89-positive cells, chiefly neutrophils 29.
Although the ligand-binding (α) chain of FcαRI is structurally related to the α chains of IgG Fc receptors (FcγR) and the high-affinity IgE receptor (FcεRI), its gene maps to chromosome 19, separate from the cluster on chromosome 1 where the genes for these other Fc receptors are located. The FcαRI gene lies alongside genes for NK cell receptors (KIRs) and leucocyte Ig-like receptors, and shares closer homology with these molecules than with other Fc receptors 30. CD89 orthologues have been identified in chimpanzees, macaques, cattle, horses, and rats 31–34. The lack of an equivalent gene in mouse is explained by disruption of the gene during a translocation event.
The α chain of FcαRI comprises two Ig-like extracellular domains, a transmembrane region, and a short cytoplasmic tail, devoid of any recognized signalling motifs. The X-ray crystal structure of the extracellular portion has been solved in isolation 10, 35 and in complex with the Fc region of human IgA1 10. The membrane distal domain (D1) of the receptor binds to the interface of the Cα2 and Cα3 domains of IgA Fc (Figure 3) (reviewed in ref 36).
The FcαRI α chain associates with a homodimer of a signal transducing polypeptide termed FcR γ chain, which also associates with other Fc receptors. The γ chain carries an immunoreceptor tyrosine-based activation motif (ITAM), comprising a sequence of two tyrosine-containing boxes (Tyr-X-X-Leu) separated by seven amino acids. When the FcαRI α chain is engaged by ligand and effectively becomes cross-linked, the γ chain ITAM tyrosines are phosphorylated, as the first step in a signalling cascade that culminates in the assembly of a multimolecular adapter complex capable of regulating signalling. Cross-linking of FcαRI drives relocation into ‘rafts’ or specialized lipid microdomains within the plasma membrane that serve as platforms for the initiation of signal transduction 37, 38.
Binding of antigen-complexed IgA to FcαRI elicits a wide range of biological responses depending on the cell type involved. These processes include antibody-dependent cell-mediated cytotoxicity (ADCC), phagocytosis, release of cytokines, superoxide generation, calcium mobilization, degranulation, and antigen presentation (reviewed in refs 39 and 40). On the whole, the responses emanate from Tyr kinase activation via the γ chain, but the integrin Mac-1 (CR3; CD11b/CD18) has also been implicated as a co-receptor in some of the functions mediated by FcαRI 41, 42.
Polymorphisms in some classes of Fc receptor have been associated with increased susceptibility to certain diseases. There are three known polymorphisms in the FcαRI promoter region, one of which is reportedly associated with an increased disposition to IgA nephropathy 43, 44. Single nucleotide polymorphisms have also been documented in the coding region of the gene, two in D1 and a third in the cytoplasmic tail. Of these, an A-to-G transition at nucleotide 363 within the D1 exon (numbering according to GenBank accession X54 150) is reported to influence susceptibility to aggressive periodontitis 45. However, since the transition results in no change in amino acid sequence, the possibility that the differential susceptibility could be explained by another gene polymorphism located near the FcαRI D1 domain cannot be ruled out. No association with any polymorphism was found when all six polymorphisms were screened in allergic asthma patients and controls 44.
Other receptors for IgA
A receptor termed Fcα/µR which binds IgA and IgM has been identified in mice and humans. The mouse receptor, while constitutive on B cells and macrophages, is also expressed on lymph nodes, intestine, appendix, kidney, and oligodendrocytes 46–48. The membrane distal portion of the extracellular region of this type I receptor shares structural features with Ig-like domains. In particular, this region has strong (43%) identity with domain 1 of pIgR. The presence of conserved residues at key positions suggests that it adopts a similar structure to that of pIgR domain 1, and may well share a similar mode of interaction with IgA 15. Although the human receptor has not been extensively studied as yet, mouse Fcα/µR has been shown to mediate B-cell endocytosis of IgM-opsonized bacterial targets. Presumably the receptor can equally trigger uptake of IgA-coated targets, and is envisaged to contribute to defence against various micro-organisms.
Initially known as a receptor for transferrin, CD71 (or TfR) has more recently been described as a receptor for IgA, binding human IgA1 (polymeric rather than monomeric) but not IgA2 49. Binding of IgA does not compete with that of transferrin. CD71 expression has been demonstrated on fetal mononuclear haematopoietic cells and certain lymphocytic and myeloid cells, but the receptor is not readily detected on adult mononuclear or polymorphonuclear cells. However, the receptor is present on mesangial cells, and the up-regulated expression seen in IgA nephropathy patients has led to suggestions that this receptor may be associated with IgA deposition in the kidneys 50, 51. Recently, interstitial-type dendritic cells (DCs) were also shown to express CD71. In immature but not activated monocyte-derived DCs, the receptor can mediate endocytosis of IgA complexes 52.
The potency of S-IgA to trigger degranulation and superoxide production in eosinophils is attributable to the expression of both FcαRI and a receptor specific for SC 53. The latter receptor has not been fully characterized, although a 15 kD protein has been implicated. The finding that a CD18-specific antibody could ablate S-IgA-mediated superoxide generation by eosinophils may point to some role for β2 integrins in this response 54.
Over recent years, there has been a growing appreciation that pIgR is not only capable of delivering polymeric IgA into mucosal secretions, but it can also transport locally-formed pIgA-containing immune complexes from the lamina propria out across the epithelium 55, 56 (Figure 4). Thus, antigens originating from the diet, environment or gut microbiota or generated during mucosal infections can be trapped by pIgA and excreted into the lumen. More than this, pIgR can mediate transcytosis of whole viruses and bacteria that have been recognized by specific pIgA 57–59. Other in vitro evidence points to the ability of pIgA undergoing pIgR-mediated transport to neutralize intracellular viruses such as influenza, Sendai, measles, and human immunodeficiency viruses within epithelial cells 57, 60–64 (Figure 4).
The interaction of dimeric IgA with pIgR is a complex one to unravel (reviewed in ref 65). However, available evidence indicates that domain 1 of pIgR plays a principal role in binding non-covalently to the Cα3 domain (and possibly also the Cα2 domain) of the IgA dimer 16, 18, 19 (Figure 3). In addition, Cys467 in domain 5 of pIgR forms a disulphide bond to Cys311 on the surface of the IgA Cα2 domain 66. The apparent overlap of part of the pIgR site with that bound by FcαRI 19 (Figure 3) may account for the inability of S-IgA to promote phagocytosis via FcαRI. However, S-IgA is capable of stimulating a respiratory burst via FcαRI 67, presumably by co-ligating Mac-1 (CD18/CD11b) as a means to overcome the decreased binding of S-IgA to FcαRI 41, 42.
Subversion of IgA and pIgR function by bacteria
The fact that numerous major pathogens have evolved means to perturb IgA function underlines the importance of IgA in immune defence. Such mechanisms provide pathogens with opportunities for more efficient invasion. Two important examples are the secretion of proteases that specifically cleave IgA and the blockade of IgA function by IgA-binding proteins that compete with host Fc receptors. In an alternative strategy, some strains of Streptococcus pneumoniae co-opt the pIgR transcytosis machinery as a means to facilitate adherence and invasion 68. These examples will be discussed further below.
Highly specific IgA1-cleaving proteases are produced by a number of important bacterial pathogens that are able to colonize mucosal surfaces and may invade mucosal tissues 69. Most notably, the bacteria frequently associated with acute bacterial meningitis (Neisseria meningitidis, Haemophilus influenzae, and Streptococcus pneumoniae) all secrete IgA1 proteases. The fact that the enzymes appear to have arisen by convergent evolution, as evidenced by their variety of structural and mechanistic features, indicates that the ability to disrupt IgA function bestows significant advantages on the bacteria concerned. The proteases are associated with virulence 70, 71; closely related strains that lack IgA1 protease are non-pathogenic.
Each IgA1 protease cleaves a specific Pro–Ser or Pro–Thr peptide bond within the extended hinge region of IgA1 (Figure 5). IgA2 remains resistant to cleavage because the susceptible hinge region sequence is missing. Cleavage of IgA1 into Fab and Fc fragments renders the antibodies unable to link Fab-mediated antigen recognition to Fc-mediated elimination mechanisms. In addition, because the Fab fragments may still bind specific antigens on the surface of the bacterium, they prevent access by any intact immunoglobulins. Thus, the bacterium evades the protective functions of mucosal immunoglobulins.
Mutagenesis studies, in which defined mutations are introduced into the IgA1 hinge region, have begun to reveal the sequence requirements for substrate recognition and cleavage by the various proteases 72–75. Surprisingly, given their remarkable specificity for particular bonds in wild-type IgA1, the enzymes can be impelled to cleave alternatives to their preferred post-proline bond if such a bond is unavailable. The enzymes also display requirements for the susceptible bond to be suitably positioned relative to the Fc region 75, and for structural elements outside the hinge region 76. Thus, in many cases, the proteases appear to recognize suitable substrates through a combination of hinge sequence recognition and more global recognition of the context in which the hinge region lies. It is hoped that this type of detailed analysis may ultimately aid the development of specific IgA1 protease inhibitors.
Bacterial IgA-binding proteins
Proteins that bind specifically to IgA (IgA-BPs) are expressed by many strains of group A Streptococcus (Streptococcus pyogenes) and group B Streptococcus. These bacteria are major human pathogens, the former associated with throat and skin infections, toxic shock, and sequelae of rheumatic fever and acute glomerulonephritis, and the latter with life-threatening bacterial septicaemia, meningitis, and pneumonia in neonates. IgA-BPs permit these bacteria to evade IgA-driven eradication processes.
Important examples of IgA-BPs include proteins Arp4 and Sir22 of group A Streptococcus that are members of a family of proteins, termed M proteins, associated with virulence 77, 78, and β protein, an unrelated protein expressed by group B Streptococcus79, 80. All these three IgA-BPs were found to interact with the same region on IgA Fc, in the inter-domain region between the Cα2 and Cα3 domains 81. The site recognized is essentially the same as that bound by FcαRI (Figure 3), and the IgA-BP can inhibit the binding of IgA to FcαRI and the triggering of an FcαRI-mediated respiratory burst 81. Such blockade would therefore allow the bacterium to evade elimination mechanisms that would normally be elicited by IgA through interaction with FcαRI.
Recently, the superantigen-like protein SSL7 from Staphylococcus aureus has been shown to bind monomeric IgA1 and IgA2 and S-IgA 82. The same study also showed that SSL7 could block binding of IgA to FcαRI. Since defence against S. aureus depends on preventing adherence to mucosal surfaces by S-IgA and other opsonins, the ability to perturb IgA-mediated clearance in this way presumably provides the bacterium with increased opportunities for colonization.
A polymorphic surface protein of Streptococcus pneumoniae known as SpsA, CbpA or PspC binds specifically to human pIgR and SC. A hexapeptide motif on SpsA mediates interaction with a site localized to domains 3 and 4 of pIgR 83–85. It has been shown in a cell culture model of bacterial invasion that in the absence of free SC or S-IgA, the SpsA–pIgR interaction mediates adherence and internalization of S. pneumoniae into epithelial cells 59. Thus, the pIgR–SpsA interaction may be of pathogenic significance by allowing the adherence of pneumococci to pIgR-positive nasopharygeal cells, as a first step in dissemination.
IgA in disease
IgA is by far the most abundant immunoglobulin in many mammals because of its mucosal secretion. In humans and higher apes, serum IgA would appear to have special significance because of the unique presence of monomeric IgA1. It is therefore remarkable that IgA deficiency is one of the most common primary immunodeficiencies, with up to 1 in 400 individuals affected in some ethnic groups, the majority of whom show no obvious signs of vulnerability to infection 86.
Although this apparent contradiction remains unresolved, there are a number of important observations gained from study of IgA deficiency that shed some light on the role of IgA. The deficiency is caused by changes in genes controlling IgA production rather than mutations of the IgA genes themselves. It is a heterogeneous deficiency. The inheritance is often sporadic, though large families have been identified often with other family members affected by other immunoglobulin deficiencies, usually common variable immunodeficiency (CVID). CVID itself is a much less common (incidence of 1 in 30–60 000), though equally heterogeneous deficiency. IgA deficiency may in fact represent one end of a spectrum of deficiencies defined as CVID. Individuals with IgA and even minor IgG deficiency, for example of one subclass, are usually symptomatic.
In some IgA-deficient families, certain MHC haplotypes are more common 87. The frequency of IgA deficiency is very different in different ethnic groups, being most common in certain Caucasian groups but possibly never found in some Asian groups. In some cases, IgA deficiency develops into CVID. It can appear or disappear spontaneously. A number of non-HLA candidate genes for IgA deficiency have been identified 88.
The reason for the lack of increased susceptibility to infection in IgA-deficient individuals is assumed to result from increased secretion of IgM, though the evidence is somewhat lacking. Some individuals with IgA deficiencies show marked susceptibility to infection. The most common infections are gastrointestinal, particularly Giardia lamblia, Campylobacter, Clostridium, Salmonella, and rotavirus. Respiratory tract infections can also occur. Although most individuals with H. pylori show strong serum and secretory IgA responses to the organism, individuals with IgA deficiency do not appear to have a higher incidence of H. pylori89.
It is well documented that IgA-deficient individuals have an increased incidence of autoimmune disease, particularly that associated with the gastrointestinal tract. In susceptible ethnic groups, up to 1% of individuals with coeliac disease can be IgA-deficient. Coeliac disease is an autoimmune disease triggered by an inappropriate immune response to wheat proteins, particularly gliadins. The mechanism of this disease is now clearly defined. The disease is caused by an immune response (both T cell and B cell) to proteins that have been modified by the enzyme tissue transglutaminase, which converts certain glutamine residues in some proteins to glutamic acid. This important post-translational modification is common in wheat proteins but also occurs in mammalian tissues including the gut. This process links the response to wheat with autoimmunity to the gut 90.
Antibodies to gliadin, to the endomysium, and to tissue transglutaminase are characteristic of this disease. Interestingly, it is the IgA autoantibodies that are most diagnostic. IgA antibodies against a wide range of food antigens are also commonly found in patients with coeliac disease. Here is apparently yet another contradiction, that a disease diagnosed by the presence of specific IgA antibodies is more common in IgA-deficient individuals.
Although IgA antibodies can be useful in the diagnosis of coeliac disease, the disease itself is T-cell-mediated. However, the IgA autoantibodies associated with coeliac disease are also associated with dermatitis herpetiformis, where IgA is deposited in affected skin. Neutrophils found at the sites of IgA deposition are believed to cause the disease. Dermatitis herpetiformis and coeliac disease are often associated 91.
Serum antibodies to Saccharomyces cerevisiae (ASCA) are gaining some acceptance as a tool for the diagnosis of Crohn's disease. Again, the IgA antibodies are most specific. The antigenic determinants are specific mannan carbohydrates that are found in both S. cerevisiae and other micro-organisms, raising the possibility that other micro-organisms might be a trigger for this disease. Although the antibodies are occasionally also found in patients with liver disease who have increased IgA production, they are seldom found in other inflammatory bowel diseases 92.
There can be no doubt that a significant number of individuals with IgA deficiency do have disease of the gastrointestinal tract. These and other observations suggest an as yet undefined link between IgA deficiency and inter-related infectious, inflammatory, autoimmune, and neoplastic disease in the gut. Taken together, these findings indicate a role for IgA in the general well-being of the gut. Recently, there has been an explosion in interest in the possibility of health improvement by alteration of the flora of the gut by a change in diet or by ingestion of ‘friendly bacteria’ 93. Although many of the claims have yet to be substantiated, there are reports that prebiotic and probiotic foodstuffs do increase levels of mucosal IgA in rodents 94. However, as far as we are aware, there are no rigorous studies as yet to show that similar effects occur in humans.
The liver has been known for many years to play a key role in IgA metabolism, though it is not yet clear what its function might be 95. Unfortunately, many early studies came to the wrong conclusions because of differences in the binding of human and rat IgA to the asialo-glycoprotein receptor (a receptor which recognizes terminal galactose residues on serum glycoproteins and internalizes them for intracellular degradation), and differences in the hepatic expression of the polymeric Ig receptor in man and rodents. It has been recognized for many years that patients with severe liver damage have a marked rise in serum IgA concentration, including polymeric and secretory IgA.
FcαRI expression has been demonstrated on Kupffer cells in the liver of CD89 transgenic mice and in human liver 96. In the transgenic mice, Kupffer cells expressing FcαRI were shown to mediate in vivo phagocytosis of E. coli that had been precoated with serum IgA or secretory human IgA prior to injection. The data suggest that in pathological conditions of the gut, characterized by a defective mucosal barrier and production of inflammatory mediators, expression of FcαRI will be induced in Kupffer cells. These phagocytes might remove IgA-opsonized bacteria from the portal blood before full septicaemia can ensue.
The significance of increases in serum IgA associated with liver damage caused by alcohol or by inflammatory bowel disease is not known. A number of diseases, including AIDS, which are linked to increased levels of IgA, are associated with aberrant expression of FcαRI and IgA deposition in tissues. Interestingly, CD89 transgenic mice in which the transgene is driven by a myeloid-specific promoter conferring high FcαRI expression on monocytes/macrophages have been shown to spontaneously develop IgA nephropathy 97. IgA nephropathy is the most common form of glomerulonephritis, characterized by deposition of IgA in the glomerulus. The mechanism in the mouse model is believed to involve binding of mouse IgA dimers to FcαRI with subsequent release of soluble FcαI–IgA complexes that deposit in the kidney. However, there are some apparent inconsistencies in this putative mechanism as controversy remains over the ability of mouse IgA to bind human FcαRI. Mutagenesis of human IgA in the FcαRI interaction site to mimic the mouse IgA sequence ablates receptor binding, so it is unclear if this model provides a true reflection of IgA immune complex formation in the human system.
The cause of IgA nephropathy, which is variable in the severity of its presentation, remains unknown. It occurs as primary disease with no apparent trigger but it can be associated with liver cirrhosis. In both cases, the IgA deposited in the kidney is primarily polymeric IgA1. Although there are many reports (sometimes conflicting) that the circulating IgA in patients with the disease is abnormal in terms of either its molecular size, specificity (to food antigens or infectious organisms) or aggregation state, the reason for the deposition is not known. Interestingly, in IgA myeloma where IgA levels can be even higher, IgA deposition in the kidney is not a common feature.
There is currently much emphasis on the analysis of differences in the glycosylation of the circulating IgA in IgA nephropathy. The O-linked carbohydrate chains present in the hinge region of IgA1 appear to be different from normal in patients with IgA nephropathy 98–101. This lack of specific galactose residues could have marked effects on the properties of the molecule 102. The trigger for the abnormality is not known. Although there is a large amount of evidence for impaired gut mucosal immunity in IgA nephropathy, the IgA deposited in the kidney may not originate from mucosal sites.
Although the IgA is abnormal in IgA nephropathy, others have shown that IgA receptors on leucocytes or on liver cells, which might remove abnormal circulating IgA, and IgA receptors on kidney mesangial cells, which might bind the IgA, are also abnormal 103.
Potential for use of IgA in therapy
Increasing numbers of IgG monoclonal antibodies are being licensed for clinical use, and interest in the possibility of using IgA in such a setting is growing with the emergence of some promising examples 104, 105. A secretory IgA-based antibody against a Streptococcus mutans adhesion protein was shown to offer protection against oral streptococcal recolonization, suggesting that such antibodies may have potential in caries control 106. IgA-based antibodies against other bacterial pathogens including S. pneumoniae107, 108, N. meningitidis109, and Bordetella pertussis110, 111 have also shown efficacy in laboratory-based studies. Other studies have suggested that FcαRI-directed therapies offer new possibilities for the treatment of certain cancers 112–115.
Systemic immunization, for example with conjugate Hib or meningococcal C vaccines, can lead to the production of both serum IgG (and IgA) and mucosal IgA (and IgG) associated with protection 116. However, there has long been a goal to develop generic technologies for mucosally administered vaccines designed to elicit mucosal responses. The earliest evidence for effective mucosal immunization in humans came from the oral polio vaccine, where efficacy correlated with the production of mucosal anti-polio IgA 117. Since then, a wide range of technologies have been used to try to develop mucosal immunity, with variable success 118. Oral cholera and other oral vaccines using a variety of vectors induce mucosal IgA and show promising protection in animal models. The recently approved trivalent live attenuated nasal influenza vaccine, FluMist, could herald a new generation of mucosal vaccines 119.
A growing appreciation of the importance of the mucosal immune system, coupled with an improved understanding of the structural and functional properties of IgA facilitated by advances in antibody engineering, has stimulated renewed interest in this previously rather neglected antibody class. Continued research into IgA seems likely to open up possibilities for novel uses in therapeutic settings, for improved mucosal vaccination, and for new interventions in a variety of disease states associated with abnormal IgA function.
JMW gratefully acknowledges funding from the Wellcome Trust.