B-cell receptors and heavy chain diseases: guilty by association?

Immunoglobulin (Ig), as paired heavy chain (HC) and light chain (LC), occurs in a secreted form as antibody. It also exists in a membrane form (HC + LC) sheathed by the Igα/Igβ heterodimer^1,2  forming the B-cell receptor (BCR). Here it plays a major role in B lymphocyte activation³  and survival.^4,5  Heavy chain diseases (HCDs) are rare B-cell proliferative disorders characterized by the presence in the serum of monoclonal, incomplete, Ig HCs not associated with LCs.⁶  Alterations in the Ig HC gene leading to HC molecules have been identified in HCD, with deletions in the variable (V) domain in all but one case studied to date.⁷  This makes the HC gene mutation a recurrent alteration in a tumor, which tends to suggest a role in pathogenesis. Whereas much attention has been paid to the secreted form of the Ig in these diseases, it now appears that the membrane form of the truncated HC, as part of the BCR, has abnormal aggregation and signaling properties that may contribute to disease development. Here we review the current knowledge on the functions of surface HC and on the alterations of these proteins in HCD.

The definition of HCD is usually a biochemical one:⁸  the presence of an abnormal HC not associated with LC in the serum. However, this biochemical definition has 2 caveats: (1) there are lymphoproliferations where abnormal HCs are not secreted, although produced;⁹  and (2) a biochemical alteration is not a disease by itself. In this review, we focus on HC protein alterations associated with lymphomas, whether they meet the traditional biochemical criteria or not.

The clinical aspects of HCD have been recently reviewed elsewhere¹⁰  and are summarized here.

α-HCD,¹¹  the most common HCD with more than 400 reported cases,¹²  is almost equivalent to immunoproliferative small intestinal disease (IPSID),¹³  although there are gastric and pulmonary forms of α-HCD, and IPSID can also be associated with γ-HCD. IPSID, a variant of extranodal marginal zone lymphoma, is characterized by the presence of diffuse mucosal infiltrates,¹⁴  involving a large part of the intestine, thus producing malabsorption.¹⁵  It affects mainly younger adults in the Middle East and the Mediterranean region. Treatment of IPSID at an early phase with antibiotics causes remission in some patients,¹⁶  which may be the result of their effects on Campilobacter jejuni, a bacterium strongly associated with IPSID.¹⁷ 

γ-HCD is a heterogeneous condition but usually presents as an aggressive form of lymphoplasmacytic lymphoma, whereas μ-HCD, the rarest form of HCD, typically presents as a chronic lymphocytic leukemia, frequently with vacuoles in the malignant lymphocytes, and the presence of light chains in the urine.¹⁰ 

It is of interest to note that all of these B-cell proliferations, even if they have a large plasma cell component, are lymphomas rather than plasma cell proliferations. The latter are known not to require an HC for survival.¹⁸ 

All classes of HC can be made in a membrane-bound form, as a consequence of the production of a distinct mRNA comprising membrane exons, from the same gene that gives rise to the secreted transcript.¹⁹  It follows that the genetic alterations affecting secreted HCs will also impact on surface HCs. The membrane-bound HC is part of the BCR (Figure 1), which is responsible for transmitting signals to the cell on antigen binding, as well as for internalizing antigen, facilitating peptide presentation to T cells.²⁰  The BCR is required for clonal selection of cells recognizing exogenous antigens; and although the precise role of BCR signaling in clonal selection is not known, it is worth noting that aggregation-induced BCR signaling stimulates calcium flux²¹  and tyrosine phosphorylation,²²  both processes linked to cell activation and proliferation. Although T-cell help is the limiting factor for clonal selection in the germinal center,²³  BCR stimulation leads to up-regulation of major histocompatibility complex class II and costimulatory molecules²⁴  and thus is likely to influence the efficiency of T-cell help. BCR signaling is also responsible for negative selection of cells encountering autoantigens.²⁵  The fate of B cells after BCR stimulation by antigen binding (negative selection vs clonal selection) seems to depend on the maturity of the cell (BCR signals tolerize immature B cells) and from the simultaneous presence of associated signals.²⁶  More recently, another essential role of unligated BCR in B-cell survival, acting through the phosphatidylinositol 3 kinase pathway, has been demonstrated.^4,5,27  This basal constitutive activity is sometimes referred to as “tonic signaling.”²⁸ 

The membrane-bound HC is also part of the pre-BCR (Figure 1), in which the heterodimeric surrogate LC, composed of Vpre-B and λ5, associates with the HC at a developmental stage where the conventional LC is not yet produced.^29,30  λ5-deficient mice show a dramatic decrease in B-cell development,³¹  whereas mutations in the human λ5 gene result in agammaglobulinemia,³²  as do germline mutations in μ-HC³³  or in the Igα/Igβ heterodimer.^34,35 

HCs and conventional LCs are produced from genes that recombine during development, a process that generates huge variability but is error prone. The role of the pre-BCR seems to be in ensuring that HC recombination has occurred successfully before an LC is produced: signaling through the pre-BCR allows proliferation only of cells having rearranged a functional HC.³⁶  It might also inhibit further rearrangement at the HC locus,³⁷  securing HC allelic exclusion. However, there is also evidence that this can occur in the absence of surrogate light chains,^31,38  and even of any type of light chains,^39,40  whereas the membrane form of HC seems to be mandatory for this process.⁴¹ 

Figure 2 (adapted from Buxbaum and Alexander⁷ ) illustrates the various alterations that have been found in HCD proteins. As can be seen, all but one show complete or partial deletion of the V region, with CH1 also frequently missing. The unique case with a full-length V displays several alterations in the V sequence,⁴²  including additional cysteine residues and glycosylation sites. In this case, a CH1 deletion accounts for secretion in the absence of LC. Various genetic alterations, including deletions, insertions, and point mutations, putatively generated in the course of somatic hypermutation,⁴³  are responsible for the production of secreted HCD proteins.⁷  In addition, it has been shown that deletions preventing the production of the secreted form of α-HC may account for nonsecretory α-HCD in which membrane-bound HC alone is produced.⁴⁴ 

The most obvious consequence of deletions within the HC is the ability of truncated HCs to be secreted, and of membrane HCs to be exported,⁴⁵  in the absence of LCs. However, other consequences, namely, self-aggregation and ligand-independent signaling of the BCR, probably play a major pathogenic role. Although the recurrence of a genetic alteration usually implies a selective role, the role of abnormal BCR in this selective advantage was neglected for 3 main reasons:

1. The HC gene is frequently involved in chromosomal translocations in lymphomas, and its contribution to pathogenesis in these situations does not seem to involve HC protein production, but rather provision of HC regulatory elements for another gene.⁴⁶  This, however, is not the case in HCD, where the HC gene responsible for HCD protein production is not included in a chromosomal translocation. In one instance, a t(9;14)(p11;q32) translocation occurs on the nonproductive allele.⁴⁷ 

2. It has been suggested that HCD protein production is only required for the cells to cope with LC loss,⁴⁸  although there are several μ-HCD cases⁴⁹  and at least one γ-HCD case⁵⁰  with free LC production (Bence-Jones protein). This model will be examined in the light of recent results obtained in LC-deficient mice.⁵¹ 

3. The deletion of the antigen-binding domain could have been interpreted as an indication that the BCR is inactive; the selective advantage would then be related to the escape from idiotypic control.⁴⁸ 

Whereas mutant surface receptors with a deletion in the ligand-binding domain could, at first sight, be viewed as loss-of-function mutants, there is now ample evidence that they can actually be constitutive mutants, with the ability to generate signals in the absence of ligand. Mutants activated in such a way include oncogenic growth factor receptors with protein tyrosine kinase activity,^52,–54  glycoprotein hormone receptors,^55,56  and cytokine receptors.^57,58  It has been demonstrated, for most of these receptors, that, in the normal situation, activation involves receptor dimerization on binding of their cognate ligand,^59,60  explaining why some antibodies capable of aggregating these receptors behave as agonists. In many instances, it has been shown that constitutively activated mutants can adopt a dimeric form in the absence of ligand^61,,–64  (Figure 3).

Mice transgenic for a human V-less μ-HC similar to the monoclonal protein found in 3 cases of μ-HCD have been created.⁶⁵  The membrane form of this μ-HCD protein is expressed on the B-cell surface without associated LC. Although production of HCD protein in humans is thought to occur only at a later stage in B-cell development, because of somatic hypermutation-type genetic events,⁴³  early expression of μ-HCD protein in mice provides information indicating an unusual behavior: the truncated protein inhibits mouse HC gene rearrangements and, more importantly, allows pre-B-cell development to proceed in the absence of the λ5 component of the surrogate light chain.⁶⁶  Although this clearly indicates a ligand-independent activity of the surface-μ-HCD-composing receptor, whether this activity could be ascribed to new properties brought about by truncation or solely to the fact that truncated HCs can be expressed on the surface in the absence of surrogate LCs has remained controversial.^67,,–70  Indeed, evidence for a constitutive activity of normal BCR has been obtained,^4,5,71  and the existence of a pre-BCR ligand is disputed.^{29,67,72,–74}  The finding that recombinant pre-BCR Fab-like fragments form dimers suggests that aggregation can occur without any pre-BCR ligand.⁷⁵  It is now widely accepted that pre-BCR activity is brought about by λ5-mediated aggregation,^76,77  and recently it has been shown that this is, at least in part, the result of the binding of the λ5 tail to a conserved asparagine (N)–linked glycosylation site in the CH1 domain of μ-HC.⁷⁸  Self-aggregation of surface μ-HCD proteins in the absence of λ5 has been evidenced in transgenic mice.⁶⁶  Because self-aggregation is not observed with a normal BCR, this property is related to V-deletion. Furthermore, B cells of μ-HCD protein transgenic mice have a phenotype similar to that of BCR transgenic B cells developing in the presence of the cognate BCR ligand,^79,,–82  or surrogate LCs expressing transgenic B cells.⁸³  Therefore, deletion of the V region of the μ-chain brings new properties to the BCR, in addition to its normal, low level of constitutive activity.

The HCD-BCR study has provided some clues about how truncation-induced BCR aggregation is achieved. One of the initial difficulties in analyzing the system was the inability of normal HCs to be expressed on the cell surface in the absence of LCs.⁸⁴  However, this also suggests a mechanism: normal HCs, in the absence of LCs, interact with the chaperone immunoglobulin binding protein (BiP), which retain them in the lumen of the endoplasmic reticulum⁸⁵  (Figure 4); in contrast, HCD proteins are exported to the cell surface in the absence of LCs, indicating that they do not form stable interactions with BiP. The major function of chaperones is to assist in folding of newly synthesized polypeptide chains and to prevent them from aggregating.⁸⁶  In addition, it has been shown that HCs, when separated from LCs, tend to aggregate.⁸⁷  In terms of signaling, this would indicate that LC inhibits constitutive signaling by associating with HC. Evidence supporting this hypothesis has been shown in a cell system using the adaptor protein SLP-65.⁷⁷ 

It is worth noting that BCR signals activate the nuclear factor-κB pathway,⁸⁸  a pathway involved in the cellular response to stress and in some cancers, but also in the control of Ig LC expression,⁸⁹  as its name (nuclear factor-κB) indicates, which could be the basis of a regulatory feedback.

What is the mechanism making HC when not associated with LC prone to self-aggregation on the cell surface? Although they could be collectively referred to as misfolding, one can envision 2 distinct mechanisms: in one case, the clustering would be caused by a lack of steric hindrance from the amino-terminal domain, which would be lost when this domain or the LC is removed, whereas in the other case, exposure of hydrophobic residues would be the major determinant. Interestingly, hydrophobicity has been proposed as a major determinant for the initiation of inflammatory responses.⁹⁰ 

The model developed here for explaining ligand-independent signaling rests on the current model of BCR activation where BCR oligomers trigger signaling (crosslinking model).⁹¹  In addition to the large body of evidence generally cited in most reviews, strong support for this model stems from the study of Nagata et al, showing that crosslinking of Igβ on proB cells can elicit differentiation signals similar to those delivered by the pre-BCR in B-cell development.⁹²  Recently, Yang and Reth proposed an alternative model for BCR activation, in which most BCRs exist on the B-cell surface in “signaling-inactive” dimers (oligomers) in equilibrium with a small number of “signaling-active” monomers.⁹³  However, even this model has to acknowledge the notion that antigen acts by “clustering” BCR, which is not clearly conceptually distinct from “crosslinking” or “oligomerization.” In addition, a major point of concern in this model is the percentage of “signaling-inactive” dimers, whose measurement is biased in the Yang and Reth study⁹³  by the method they use, which involves irreversible association of 2 Igα half-domains. The high percentage of dimers they find contrasts with the results of Pierce et al,⁹¹  using different methods, which indicate that only a small fraction of BCR in resting B cells exist as dimers and that antigen binding results in the formation of dimers or oligomers. Therefore, although there is some indication from the work of Yang and Reth⁹³  that the crosslinking model might be refined in some way, it is too early to elaborate on their model as a new paradigm.

New evidence^4,5  for the requirement of a BCR for B-cell survival has given some support to a model where HC truncation is secondary to LC loss, allowing the cell to survive by presenting an HC on the cell surface in the absence of LC. The creation of LC-deficient (L^−/−) mice⁹⁴  and the spontaneous production of HC-only antibodies in these animals have allowed this model to be examined.⁵¹  Plasma cells harboring somatic gene deletions in γ, α, or μ are selected, in agreement with the model. However, all the mutants analyzed were found to have an intact VH region, with the deletion removing CH1, CH1 and hinge, or CH1 and CH2, suggesting that deletion of the antigen-binding domain is not the consequence of LC loss.^51,95  The secreted HCs in LC-deficient mice were similar to those produced in HC deposition disease, a kidney disease secondary to plasma cell dyscrasia.^96,97  More surprisingly, in L^−/− mice and in another model of LC deficiency, B cells were found to express high levels of a full-length μ-chain on their surface.^98,99  It is not known whether surface HC is expressed alone or in association with an LC-like molecule. These results indicate that LC loss does not result in the selection of cells with V domain HC deletion. In addition, V domain deletion can occur without LC loss in μ-HCD, supporting the idea that HC truncation occurs first.

As BCR aggregation is a signal for somatic hypermutation in germinal center-like B cells,^100,101  we favor the hypothesis that LC loss is secondary to mutations induced on HCD-BCR ligand-independent aggregation. This is further supported by the extensive somatic hypermutation observed on the LC gene in γ-HCD.¹⁰²  The difference between μ-HCD and γ- or α-HCD in terms of LC production could be accounted for by the rate of somatic hypermutation, which is known to occur preferentially on switched isotypes.

In addition to indicating that LC loss is not followed by V gene deletion, LC-deficient mice provided new evidence for the role of LC in preventing HC aggregation. Most of the plasma cells of LC-deficient mice were shown to be Mott cells, containing Russell bodies filled with aggregated CH1 or CH1/CH2-deleted HC.⁹⁹  By expressing similar mutant HC in LC-proficient mice, it was possible to demonstrate that Ig aggregation leading to Russell body formation was favored by the absence of LC. However, it is important to consider that these results indicate subtle differences between HCD and HC deposition disease-type protein in alterations of BCR signaling because they lead to different phenotypes, both in humans and in mice.

Although the past few years have provided indications on how the loss of the antigen-binding domain affects BCR activity, there is currently no animal model for any of the HCDs. Transgenic animals that express the μ-HCD protein do not have lymphoproliferative disease, and their B cells are similar to B cells developing in the presence of their cognate antigen (ie, anergic).⁷⁹  The precise timing of BCR alteration, thought to occur later in development in HCD, may explain the lack of B-cell proliferation because BCR stimulation of newly formed B cells is tolerogenic. Depending on timing, normal BCR signaling can either increase¹⁰³  or decrease¹⁰⁴  myc-induced lymphomagenesis. Moreover, other mutations, some of them corresponding to clonal cytogenetic abnormalities found in HCD might be required. The stimulus provided by C jejuni infection could be involved at the early stage, in addition to the genetic alterations. These other factors might compensate for the lack of cognate T-cell help,²³  which might be expected as a consequence of the inability of HCD cells to recognize, internalize, and present a specific antigen.

Contrasting with V-less BCR, T-cell receptor-β with a variable region deletion is oncogenic in transgenic mice.¹⁰⁵  Tumors develop from double-positive T cells at a high incidence, and unexpectedly, enforced expression of the Bcl2 oncogene reduces lymphomagenesis.¹⁰⁶  It should be noted, however, that, although a V-less T-cell receptor-β can be produced by retroviral insertional mutagenesis in cat and in mouse lymphomas,^107,108  it has not been associated with a human disease.

Other molecules of the BCR signaling pathway have been identified as targets of genetic alterations in B-cell lymphomas. Among them, 3 molecules are key regulators of the nuclear factor-κB signaling pathway.^109,–111  Both MALT1 and Bcl10 are independently associated with chromosomal translocations in mucosa-associated lymphoid tissue lymphoma, whereas the CARD11 gene displays activating point mutations in a percentage of diffuse large B-cell lymphomas. Interestingly, constitutive nuclear factor-κB activation seems to be related to spontaneous aggregation of the mutant CARD11 proteins.¹¹¹  Neither of these genes has been shown to have oncogenic activity in mice¹¹² ; and on the contrary, Bcl10 transgenic mice display enhanced apoptosis in spleen.¹¹³  Somatic mutations affecting the immunoreceptor tyrosine-based activation motifs of CD79B and CD79A have been observed in the activated B cell–like subtype of diffuse large B-cell lymphoma,¹¹⁴  but it is not known whether they lead to enhanced signaling, and they have not been yet tested in an animal model.

The mechanism suggested for HCD-BCR constitutive activation (ie, ligand-independent receptor aggregation and activation) as a consequence of misfolding, resulting from escape from chaperone binding, might also be involved in other diseases. In activated B cell–like–diffuse large B-cell lymphoma, IgM-bearing receptors are clustered, which is correlated to “chronic active” signaling.¹¹⁴  However, no mechanism has been found explaining clustering and chronic active BCR signaling, and it has been shown that it does not depend on the CD79 mutations present in a proportion of cases.¹¹⁴  Misfolding can occur on apparently normal proteins. For instance, impaired glycosylation and folding of the μ-HC have been demonstrated in chronic lymphocytic leukemia, a very frequent disease with clinical features similar to those of μ-HCD.¹¹⁵  Whether this misfolding leads to enhanced BCR ligand-independent signaling (or “chronic active”) might be an important issue, as BCR signaling inhibitors are being used in clinical trials.¹¹⁴  However, it is not yet clear whether BCR signaling inhibitors are more efficient than anti–B-cell therapy. It would be interesting to get some information from HCD patients.

In addition to a cell-growth promoting role in cancer with abnormal receptors, misfolding-induced signaling might be a cause of cytotoxicity in what have been termed conformational diseases,¹¹⁶  a large group of pathologies characterized by protein aggregation. Indeed, there is increasing evidence that, for most of these diseases, pathogenicity is not correlated to large aggregates but could be better explained by the formation of dimers or oligomers.¹¹⁷  It is therefore expected that information obtained from HCD pathogenesis would help with the understanding of the mechanisms of these apparently unrelated diseases.

This work was supported by Inserm, the Biotechnology and Biologic Sciences Research Council, and European Science Foundation (exchange grant FFG-1233).

Contribution: D.C. wrote the review; L.S.M. and M.J.O. edited the manuscript; and all authors performed some of the research reported and approved the final version.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

The current affiliation for M.J.O. is Recombinant Antibody Technology Ltd, Babraham Research Campus, Cambridge, United Kingdom.

Correspondence: Daniel Corcos, Inserm U955, Faculté de Médecine de Créteil, 8 Rue du Général Sarrail, 94010 Créteil Cedex, France; e-mail: daniel.corcos@inserm.fr.