The riddle of the dual expression of IgM and IgD

Introduction

The membrane form of immunoglobulins (mIg) is expressed on the surface of B lymphocytes from a very early developmental stage in the bone marrow (pre B cell) until the cell finally differentiates into a plasma cell. mIg associates with other transmembrane proteins to form the B-cell antigen receptor complex (BCR). Numerous studies from different laboratories have confirmed that signalling through the BCR not only steers the B cell through development, but also secures its survival in the periphery, as a fully matured cell. mIg exists in different forms (isotypes or Ig-classes) defined by the constant region of the molecule. During development in the bone marrow, mIg is restricted to the IgM isotype, but as soon as the B cell leaves the bone marrow to populate peripheral lymphoid organs, like spleen, lymph nodes or intestinal mucosa tissue, it starts to express a second isotype on the surface, IgD. Since its identification four decades ago¹ IgD has been intensely studied, leading to often controversial results concerning the role of IgD in B-cell physiology. Because the function of IgD is still enigmatic, we will summarize in this review recent findings in the field of IgD research and will develop a model highlighting IgD as a regulatory molecule for B cell development and homeostasis.

The BCR

In mammals, B cells descend from a common lymphoid progenitor cell and develop in the fetal liver and later in the bone marrow. Diversification of the antibody repertoire is obtained by the largely random recombination of V, D and J segments on the heavy chain locus and V and J segments on the light chain locus, resulting in the random generation of functional immunoglobulin genes and their subsequent expression as proteins in the pre-B cell stage of development. A functional immunoglobulin protein consists of at least two light chains and two heavy chains. In its membrane-bound form immunoglobulin is obligatorily associated with two other membrane proteins, CD79-a and CD79-b (Ig-α and Ig-β, respectively).² These proteins not only function as a chaperone to facilitate membrane expression, they also couple the antigen receptor to membrane-proximal signalling elements. Signals derived from this BCR complex regulate the further development of the B cell [reviewed in 3]. Summarizing, BCR signalling controls:

The precise mechanism of the initiation of BCR signalling is still matter of speculation. According to the prevailing model, a signal is initiated upon oligomerization of two isolated BCR complexes through engagement by antigen (cross-linking of the receptors). Alternative models explain the initiation of signalling by a conformational change within the receptor upon antigen encounter, or by the selective shuttling of engaged receptors into lipid rafts, which are special domains of the plasma membrane, optimised for the initiation of signals (for reviews see 4,5). Upon binding to antigen, the receptor is linked to the cytoskeleton and rapidly internalized. However, internalization is not necessary for signal transduction and B-cell activation. Rather, the route of internalisation is crucial for antigen processing and subsequent presentation to T cells to initiate an immune response.⁶

Despite the controversies, it is believed that the different immunoglobulin isotypes initiate signalling in a similar way. Differences in the amino acid composition of the transmembrane domains and cytoplasmic tails of the different isotypes may, however, account for the recruitment of distinct signalling components in addition to Ig-α and Ig-β.⁴ Remarkably, the receptor isotypes IgM and IgD are structurally similar and both are expressed on the surface of mature B cells, in humans as well as in rodents. Thus, the question has to be raised whether these isotypes have a similar function and if not, what are the differences between them.

IgD evolution, structure and expression

All extracellular domains of immunoglobulin molecules show a particular secondary protein structure, the immunoglobulin fold, which makes the immunoglobulin molecule the prototype of a large family of proteins. A common feature of such immunoglobulin-like domains is their use in mediating protein interactions, mostly in those involved in cell–cell contact. So it is likely that in evolution the immunoglubulins have descended from immunoglobulin-like proteins by gene duplication and diversification as the proteins of choice for the recognition of antigenic material.⁷ Evolutionarily, immunoglobulins first appeared in jawed vertebrates (gnathostomes), which developed about 500 million years ago. Gnathostomes feature an immunoglobulin locus with repetitive units of variable and constant region gene segments, arranged in clusters.⁸

Different constant regions – and thus immunoglobulin classes or isotypes – are thought to have developed from a single ancient isotype by gene duplication and diversification.⁹ The ability to diversify the constant regions by class switch recombination (CSR) appeared with the tetrapods (amphibians, reptiles, birds and mammals) about 350 million years ago. In CSR the isotype of the immunoglobulin is changed without affecting the antigenic specificity by a deletional mechanism that recombines a specific region 5′ of the IgM constant region gene (switch region) with a similar switch region 5′ of another constant region gene, thereby excising the intervening DNA (cf. the ‘jumped’′ isotype(s)). The γ and ε isotypes most likely evolved from a common precursor, called υ (first expressed in amphibians).⁹

IgD seems to be expressed only in teleosts (i.e. bony fish) and in mammals. Interestingly, two different mechanisms are used for the expression of the IgD isotype: alternative splicing and CSR. In teleosts, as well as in humans and rodents, a true CSR signal (switch region) between the µ and δ genes is lacking and IgD is generated by alternative splicing of a large primary mRNA transcript, containing mRNA for the µ and δ heavy chains. The recombined VDJ gene segment is spliced to the first of three exons coding for the δ constant region. In rodents, the exon coding for the second constant domain of δ has been replaced in evolution by a short hinge region.⁹ The alternative splicing process is tightly regulated during B-cell development and differentiation. In early stages of B-cell development, only µ is expressed. Splicing of the VDJ region gene segment onto δ constant domain exons first starts when the B cell leaves the bone marrow and populates secondary lymphoid organs. The mechanism that underlies the alternative splicing is poorly understood. Evidence has been brought forward for the existence of a specific region between the µ and δ gene, which attenuates the transcription complex and forces the preferential production of mRNA for membrane IgM by default. Binding of lineage specific regulatory proteins to this region may alter the ratio of processing of the primary mRNA transcripts.^10,11

IgD expression can also be caused by CSR. In humans, only a rudimentary switch region between the µ and δ gene is present and, consequently, CSR from µ to δ is a very rare event. Nonetheless, some human lymphomas are switched to δ, with deletion of µ exons.¹² In addition, it was recently reported that in artiodactyls¹³ a short switch-region-like sequence is located immediately upstream of the δ gene and directs regular µ–δ class switch recombination in the cow.¹⁴

IgD expression is not only regulated by the two mechanisms described above. Surface expression is also regulated by the selective turnover of mRNA and membrane proteins. In humans and rodents, the expression of mIgM and mIgD varies, but the ratio of membrane expression of either isotype does not necessarily reflect the abundance of the transcripts or the splicing efficacy. Indeed, Yuan et al. reported that in resting B cells the amount of mRNA for µ exceeds the amount of mRNA for δ about sevenfold, although the actual surface levels of protein show more IgD than IgM.¹⁵ This is because of a higher turnover rate of the IgM-containing receptors. However, after B-cell activation by antigen, mRNA levels for δ are drastically decreased and the half-life of µ mRNA is enhanced significantly.^15,16 How the turnover rate of the proteins is regulated is still poorly understood. IgD was shown to be much more stable at the cell surface than IgM, which may be attributed to a higher degree of glycosylation. However, in solution IgD shows a higher sensitivity to proteolytic enzymes than IgM and thus to degradation.^17,18

The abundance of the membrane expression of a protein can also be affected by differences in receptor endocytosis.¹⁹ IgM seems to be rather efficiently endocytosed in an antigen-independent fashion. This is dependent on the associated Ig-α/Ig-β sheath.²⁰ To our knowledge the rate of constitutive IgD endocytosis and recycling has not been determined. Yet it would be interesting to know whether isotype- and subset-specific differences exist. Indeed, in mature follicular B cells the expression for IgM is low and for IgD high, whereas in marginal zone and peritoneal B1 cells a reversed expression pattern is seen: IgM is high and IgD is low. In tolerized, deactivated cells, very little IgM is expressed.

Mechanistically, the level of surface expression could reflect association with so-called scaffolding proteins. A recently identified group of proteins, the cbl family proteins, strongly contributes to the regulation of surface expression of membrane proteins and these proteins are key factors in the endocytosis machinery. Cbl proteins (especially cbl-b) can associate with the BCR and thereby target ubiquitin ligase and endophilin to the BCR complex. Ubiquitin ligase attaches mono-ubiquitin to the BCR which mediates internalization for either degradation or recycling of the BCR complex. Endophilin helps to bud off membrane vesicles for subsequent internalization.^21,22 For the case that cbl proteins regulate the number of receptors on the B cell, they additionally might influence signalling.⁶ It would be interesting to analyse whether cbl-b is preferentially recruited to a certain BCR isotype, thus contributing to the distinct ratios of surface IgM and IgD levels.

The BCR as a signalling device

B cells develop daily in the mouse bone marrow (2 × 10⁷), but only a few percentage arrive as mature B cells in the peripheral long-lived B-cell pool. Hence, the developing B cells underlie a selection process during which the gross majority of the cells is lost. The only sensible criterion for selection is the antigen-specificity of the receptor: as a minimal requirement, high-affinity self-reactive clones have to be prevented from participating in the immune response. Three forms of selection can be envisioned (and are found in the selection processes that shape the immunocompetent T cell populations:²³ (i) ‘negative’ selection – a B cell is actively deleted, e.g. by apoptosis, from the pool; (ii) selection by ‘neglect’– the B cell is left alone (‘death by neglect’); or (iii) ‘positive’ selection – the cell is actively recruited into the pool of surviving cells.

First it was assumed that B cells, also autoreactive clones, were normally generated, but upon encounter with antigen, the autoreactive clones were made unresponsive, without apparent deletional processes.²⁴ These conclusions were based on studies, in which mice were carrying immunoglobulin transgenes coding for high affinity antibodies to a particular antigen, while the tolerizing antigen, also encoded by a transgene, was present in the serum of the mouse in a soluble form. Further studies showed, however, that B cells were deleted, when the tolerizing antigen was presented in a membrane-bound form.^25,26 These and later studies showed that the immature B cell in the bone marrow, as the first B cell in development to carry rearranged heavy and light chains, is initially protected from deletion. Yet, BCR-derived signals do play a role in this phase of development; cells expressing a BCR with autoreactive specificity get a second chance by a renewed shuffling of the V region genes (‘editing’^27,28). In the next stage of development, B cells leave the bone marrow and populate the spleen and other peripheral lymphoid tissues. This stage of development has been called transitional type 1.^29,30 In this window of development B cells are extremely sensitive to negative selection. At the end of this stage, B cells start to express IgD as the second immunoglobulin-receptor type (with the same specificity!), they also become more resistant to negative selection. It is assumed that by this time the process of negative selection is completed and the repertoire purged of autoreactive specificities. Polyclonal stimulation of B cells however, belays that this purging is not absolute: low-affinity self-reactive clones are easily detected after stimulation.

B cells of the transitional type I can develop in the spleen to transitional type II and further to mature follicular B cells and marginal zone (MZ) B cells. In recent studies it was shown that these steps in development are not default processes, but guided by BCR engagement and cytokines as costimulating agents, in particular B cell activating factor belonging to the TNF family (BAFF) of the tumour necrosis factor (TNF) family of cytokines.^31–33 Also the survival of mature, follicular B cells is dependent on the BCR. Lam et al.³⁴ found that the induced deletion of a transgenic heavy chain in mature B cells caused enhanced apoptosis. This implies that a low level of signalling through the BCR is required for the survival of both mature and immature B cells. The nature of this signal is not known. It could either be the result of a tonic, and thus ligand-independent signalling of the BCR as proposed by Pracht et al.³⁵ or of a low-affinity interaction with self structures. Recent experiments have shown, that the downstream target of this signalling is the transcription factor nuclear factor-κB (NFκB), a known survival inducer.³⁶

Levine and coworkers modified this model and stated, that the transitional type II B cells were positively selected in dependence of a distinct set of heavy and light chain pairs.³⁷ Hence, they argued that the selection process had to be based on a specific interaction with self-tissue or minimally, a distinct ligand.

Also earlier in development, during the pre-B-cell stage self-antigens can influence positive selection of B cells: the pre-B-cell receptor, containing the rearranged µ heavy chain and the so-called surrogate light chain, seems to interact with self determinants to trigger proliferation of the pre-B cells. In the periphery, peritoneal B cells and marginal zone B cells, both characterized by a low expression of mIgD and a high expression of mIgM, are reported to be positively selected based on low-affinity interactions with glycosylated self antigens and by distinct combinations of heavy and light chain pairs. Probably the selection for glycoprotein could enhance the probability of the specific recognition of sugar residues on bacterial cell walls.³⁸

In line with these observations, as reviewed elsewhere, various defects in molecules involved in BCR signalling can abrogate transition from immature to mature B cells. A extensive review of these data can be found in reference 39.

The developmental steps discussed above take place without a change in the binding specificity or affinity of the BCR. New autoreactive specificities should not arise and the degree of autoreactivity of the BCR is apparently sufficient for selection without notable autoaggression. Yet deviations in the strength and duration of the interaction processes inevitably lead to autoaggression: for instance BAFF overexpression³¹ or aberrant IgM signalling in T1 B cells (and therefore deficient negative selection)⁴⁰ both lead to autoimmune phenomena. During bona fide immune responses affinity maturation takes place. If the B cell undergoing affinity maturation was autoreactive with low affinity or the affinity maturation process caused new autoreactive clones to arise, further checkpoints would be necessary. Indeed, a phase of sensitivity for negative selection has been found in preplasmablasts.⁴¹

In the steps of the transitional B cells towards maturity BCR engagement is essential. Remarkably, in this process, IgD appears as the second immunoglobulin-receptor, with the same specificity as IgM. In T2 transitional B cells both IgM and IgD are expressed at high levels, allowing for a high signalling capacity. In this phase, the B cell is not sensitive to negative selection,^29,33,42 but protected by antiapoptotic factors.^29,33,42 In this phase, the BCR is also coupled differently to the signalling pathway, with involvement of Btk and protein kinase Cβ.⁴³ However, it is not known whether these new elements in the coupling pertain to both receptors, or preferentially to one.

Also, other data show that the levels of BCR surface expression, and therefore signalling potential, regulate the differentiation into the different B-cell subsets.⁴⁴

A very special state in tolerance induction is the anergic state. It is now generally accepted, that engagement of the BCR without a second, costimulatory signal, leads to apoptosis. Yet, stimulation of peripheral B cells with nominal antigen in the absence of costimulation, as can be done in transgenic models, does not always lead to cell death, but rather to a state that is called anergy. For example, transgenic mature peripheral hen egg lysozyme (HEL)-specific B cells became anergic when HEL production was induced in adult mice. This was accompanied by a characteristic down-regulation of mIgM, but not of IgD.⁴⁵ The nature of the anergic state is still enigmatic. One hypothesis suggests that anergy is based on an uncoupling of membrane immunoglobulin from the proximal signal transducers, here Ig-α and Ig-β, a process termed receptor desensitization.⁴⁶

Desensitized cells, which can be mature and immature, are functionally very similar to anergic cells, however, they do not exhibit the characteristic downmodulation of IgM.^47,48 Goodnow and colleagues propose that the anergic state is the result of the selective coupling of the BCR to a tolerosome, a signal transducing complex that inhibits activation of the cell and induces targeting of BCRs to the endosome.⁶ The cbl family proteins are thought to play a major role in the recruitment of the tolerosome.

Interestingly, in the above-mentioned transgenic mouse model, anergic B cells showed a markedly shortened lifespan in the periphery,⁴⁹ and never entered the long-lived pool of B cells. This led the authors to the conclusion that the distinction between anergy and deletion is a relative one. In conclusion, also ‘anergic’ B cells are excluded from the follicles in secondary lymphoid organs,⁵⁰ ensuring the elimination of antibody-producing cells, that have a high affinity for self antigens and are thus potentially harmful.⁵¹

As mentioned before, also BCR-independent processes contribute to the selection of type II transitional B cells, while the type I transitional B cells are relatively insensitive for other exogenous signals. Receptors belonging to the TNF receptor family, like CD40,⁵² the BAFF receptor (BAFF-R), B cell maturation antigen (BCMA), and the inhibitory receptor transmembrane activator and calcium signal-modulating cyclophilin ligand interactor (TACI) all contribute to a delicate balance between apoptotic and survival signals in the type II transitional B cells.³² A strong signal, in particular derived from BAFF-R can lead to the production of autoantibodies and autoimmunity.³¹ Interference with BAFF-R signalling can prevent autoimmunity.³² Mechanistically, recruitment of TNF receptor associated factor (TRAF) after engagement of receptors belonging to the TNF-R family of receptors causes activation of NFκB.⁵³ But also other surface receptors on B lymphocytes are coupled to NFκB activation, e.g. members of the Toll-like receptor (TLR) family. In particular, TLR4, as the receptor for lipopolysaccharide (LPS), can induce a polyclonal B-cell activation, in particular in Type II transitional B cells.⁴³ Also TLR9, which was originally described as a B-cell specific receptor for bacterial DNA, has now been implicated as a pathogenic factor for autoimmune anti-DNA antibodies.^54,55

These data are strong evidence for the existence of B-cell clones with autoreactivity in the periphery. These clones may be beneficial to the host, they certainly also form a potential hazard. Activation signals for B cells have to be finely tuned to prevent autoagression. It is in this context that a function for IgD has been sought.

Searching the role of IgD

Initially, it was assumed that the two receptor isotypes IgM and IgD would have clear functional differences. However, when B cell lines and immunoglobulin-transfectants were examined the results were quite contradictory. For example, Kim and Reth⁵⁶ propose that IgM, but not mIgD is under the control of a negative feedback loop, leading to enhanced apoptosis through IgM. Thus, the signals initiated from either isotype have to differ concerning kinetics and intensity. These data are consistent with the findings that engagement of IgM on immature cells in vivo and in IgM-transfectants in vitro results in apoptotic death, whereas engagement of mIgD fails to do so.^29,57 Peckham et al.⁵⁸ obtained the reverse results: they found that a minimal stimulation with anti-δ, but not with anti-µ antibodies induced apoptosis of mature resting B cells, and correlated this with an abortive cell cycle entry. Reasons for the discrepancies in the results might be that signalling through the BCR is largely dependent on the developmental stage of the B cell and thus on the cell line used for the studies⁴ and on the receptor density, which is again dependent on the developmental stage of the B cell.⁴⁴

Remarkably, the IgD molecule can be expressed on the surface in two alternative ways, causing the involvement of different signalling cascades. In the canonical way, IgD is associated with Igα and Igβ. In the alternative way, IgD can be post-translationally processed and linked to membrane lipids via a glycosyl-phosphatidylinositol (GPI) linkage.⁵⁹ Normally, only a minor percentage of IgD is GPI-linked. However, the GPI-linked isoform of mIgD selectively activates cAMP-dependent signalling pathways,⁶⁰ which synergistically support Ca²⁺-dependent signalling from the canonically sheathed mIgM and mIgD receptors.

On the other hand, early experiments with transgenic mice indicated that the δ heavy chain could fully substitute a µ heavy chain in early B-cell development.⁶¹ Also, in vivo, the BCR of either isotype seems to be able to compensate for the loss of the other because mice deficient for the µ or δ heavy chain showed only weak phenotypes.^62–64 IgD deficiency in mice had no apparent effect on the development and function of B lymphocytes. The antibody response in δ-deficient mice was only slightly delayed compared with normal mice, and the IgD deficient animals had a slightly reduced number of peripheral mature B cells, leading to lymphopenia. In contrast, Yuan et al. report that increased expression of IgD in transgenic mice impairs the activation of memory B cells.⁶⁵ Furthermore, in immunoglobulin-transgenic mice carrying either HEL-specific mIgM or mIgD, the response to HEL was comparable to that of the double transgenics in both tolerance induction and activation.⁶⁶ Hence, it seems that in mice the IgM receptor is able to mimic the IgD receptor and vice versa.

In some respects, IgD is drastically different from IgM. IgD is present in very low quantities in serum and does not seem to play a role in humoral defence mechanisms.

Further, IgD binds with relatively high efficiencies to certain bacterial proteins. Binding is not established by the antigen-binding site, but through sugar residues on the constant domains.^67,68 It is not clear what the function of this binding is, but as a result of binding, B cells can be found that express mIgD in the virtual absence of mIgM, whereby the VDJ regions bear numerous somatic mutations. These mutations are so extensive, that antigen binding can be excluded. Apparently, binding results in activation, also when the binding is not V-region dependent, and sufficient costimulation is present to induce somatic hypermutation. Possibly, costimulation is achieved by engagement of TLRs, which recognize pathogen-associated molecular patterns, e.g. LPS, bacterial DNA, peptidoglycans, flagella, etc.

Finally, we recently observed that engagement of mIgM strongly influences the simultaneous internalisation of mIgD, in dependence of the quality and strength of the mIgM engagement, but not vice versa. This effect was of short duration.⁶⁹

From these data, it becomes hard to draw a simple picture for the role of IgD in immune defence. All BCR-dependent functions (activation, receptor desensitization, apoptosis induction and tolerance induction) were induced by either of the two isotypes or by both isotypes in combination. So it seems likely that IgD rather plays a role in homeostasis and fine-tuning of the B cell response.

A model for IgD-dependent fine tuning of BCR signalling

Important for our hypothesis are the following premises:

The presence of auto-specific B-cell clones in the periphery allows two conclusions: (1) it indicates that a negative selection process is not absolute or (2) it indicates the existence of a positive selection process. These possibilities are not mutually exclusive, but surely both harbour the same inherent danger: After appropriate stimulation, the selected clones could produce autoaggressive antibodies. While the negative selection process takes place in a window of differentiation in which mIgD is not expressed, the proposed positive selection takes place in a window of differentiation in which IgD is highly expressed (type II transitional B cells) and costimulation is present in the form of BAFF and a high expression of BAFF-R. It is not far fetched to assume, that mIgD, although an additional signalling molecule, helps to protect the selected clones. One therefore has to assume that the added signalling capacity of both receptors during interaction with the autoantigen is too little to deactivate or delete the cell, too little to activate the cell, but enough to provide the ‘nourishing’ signalling to sustain survival and guide differentiation. Were the receptors fully equal, the combined signalling capacity of both receptors would be enormous and could easily surpass a threshold level for full activation. Therefore, one possibility is that the receptors are somewhat different, and by varying the ratio between them, B cells can actively regulate the expression of survival factors (e.g. Bcl2-Xl and A1/Bfl-1) and evade elimination. At the other hand, the full complement of the two receptors may be necessary to sufficiently activate B cells in the transitional stage II, assuming only a low affinity of the receptors.

The level of expression of either isotype on mature follicular B cells seems to be set during the late phases of the transitional stage. One mechanism for setting the ratio could be based on BCR specificity. A second mechanism could be antigen-independent and based on costimulation. Support for the first view comes from work with transgenic mice of known BCR specificity. For example, transgenic mice with rearranged VDJ regions specific for HEL exhibit a relatively high ratio of IgM to IgD surface expression, although they belong to the mature follicular B-cell compartment.⁷⁰ The HEL-specific B cells are of poor crossreactivity towards endogenous antigen, so possibly the low affinity for self antigens accounts for a high surface expression of IgM on follicular B cells. Furthermore, B cells from transgenic mice that bear self-specific BCRs are either deleted^25,26 or anergized,⁷¹ or they modulate surface expression of IgM.^72,73

Support for the second mechanism comes from the study of autoimmune-disease-prone mice. The New Zealand Black mouse (NZB) displays a broad array of autoimmune disease symptoms: It develops immunopathologies such as lupus nephritis, mesangial glomerulonephritis and systemic lupus erythematosus (SLE) characterized by an abnormal expansion and activation of self-specific B cells.^74,75 Studies on these mice have lead to the identification of various susceptibility loci, many of which are related to costimulatory molecules or activation markers. B cells from NZB mice exhibit a high expression of surface IgM.⁷⁶ Remarkably, they show a defect in negative selection in the transitional type I stage. Instead of a deletional process, engagement of mIgM in this phase leads to the induction of anti-apoptotic factors and an expansion of the type I transitional B-cell pool in NZB mice.⁴⁰ Apparently, signalling in NZB B cells is qualitatively or quantitatively different, allowing a higher expression of mIgM without inducing a deletional process.

As depicted in Fig. 1, we think that diminishing the signal strength by modulation of IgM levels enables the rescue of self-specific B cells. Modulation of IgM below a certain minimum level leads to anergy and finally negative selection of the cell. Probably, IgD can minimize the amount of surface IgM needed for positive selection. As reported by Soulas et al. pathogenic self-specific B cells in a transgenic mouse model for rheumatoid arthritis are only positively selected when they express IgM together with IgD, and not when IgM is expressed in the absence of IgD.^77,78

We therefore propose that the IgD molecule is able to finely tune the average signal strength triggered from the BCR upon antigen engagement. In this model, IgD warrants the maturation of B-cell clones that would undergo clonal deletion in the absence of IgD because of its specificity for self antigen. The IgD-mediated maturation of self-reactive B-cell clones could be advantageous in combating pathogenic determinants that mimic self-antigen, although it bears the risk of autoimmune disorders. We therefore favour the model that IgD modulates the signal strength of B cells upon encounter with antigen in order to increase the threshold for negative selection. Two mechanisms could be postulated for the role of IgD in the positive section of self-reactive B cell clones: (i) small differences in intensity and kinetics between IgM- and IgD-mediated signals are sufficient to warrant positive selection of critically self-reactive clones; or (ii) IgD interferes with IgM signalling leading to modulation of signals generated from the BCR. We recently described the modulation of signalling through IgD by IgM engagement in a B-cell line.⁶⁹ Probably this mechanism contributes to a balanced signalling between IgM and IgD in order to set the right threshold for selection, maintenance or activation. However, both models account for the role of IgD in B cell maturation and can explain why the bulk of B cells which are not critically self-specific, are not perturbed in their maturation in the absence of either of the isotypes. Indeed, a role for IgD in positive selection of B cells would account for the lymphopenic phenotype reported in IgD knockout mice.

Conclusion

The humoral immune system has evolved as a powerful weapon to combat invading pathogens and their toxins by antibodies. The secreted antibodies have a specific effector function determined by the constant region. In contrast to the other immunoglobulin isotypes, the IgD class is mainly expressed on the surface of B lymphocytes; in the serum of rodents and primates it is found only in very low levels. This special characteristic of IgD has shaped the first hypotheses on the function of IgD: they centred around a basic signalling function, which was indispensable for B-cell activation or deactivation. However, as IgD functions were analysed more closely it became apparent that IgD is more a modulator in B-cell homeostasis than a key player in deciding B-cell fate. Depending on the differentiation and activation stage of the B cell, the BCR exhibits different functions and a distinct behaviour, which makes it hard to obtain a precise understanding of the functioning of the BCR. In conclusion, the view that IgD has a simply definable function should be replaced by the assumption that IgD fine tunes humoral responses, modulates B-cell selection and homeostasis and thus shapes the B-cell repertoire, defining IgD to be a key modulator of the humoral immune response.

Acknowledgements

Experimental work and publication charges were supported by the Austrian Science Foundation (P-19017) and the Austrian National Bank (OENB grant: 11710).