Fortifying B cells with CD154: an engaging tale of many hues.

Abstract

Although the original antibody to what is now defined as CD40 was raised against and shown to react with bladder carcinoma cells, the major insights into the functional attributes of this important receptor molecule – now known to be found on a diverse range of cell types – have arisen from studies on B lymphocytes. This 50 000 MW glycoprotein member of the tumour necrosis factor receptor (TNF-R) family, expressed constitutively at high density throughout B-cell differentiation, provides the focus for T-cell delivered cognate help to the B cell via the capacity to engage its inducible counterstructure – CD154; or ‘CD40 ligand’ (CD40L). 1-5 The aim of this article is to review the multiple stages at which the cognate CD40–CD40L pairing might serve to direct and modify the B-cell response with a focus on data generated from human in vitro studies set within an in vivo context gleaned primarily from observations in the mouse. The consequences of engaging CD40 on lymphoma B cells are discussed while, finally, we address how the quantity and quality of the CD154 being offered to its receptor may dictate the functional outcome of the cell in question. Absence, or blockade, of the CD40–CD40L interaction results in gross impairment of the physiological and molecular B-cell pathways that are unique to T-dependent (TD) antibody responses. Thus in patients with X-linked Hyper–immunoglobulin M (IgM) Syndrome (who carry a functionally crippled CD40L gene), 6 in genetically engineered CD40- or CD154- deficient mice,7, 8 or in mice administered with antibodies that impede CD40–CD154 interaction, 9 there is a failure of germinal-centre (GC) formation accompanied by a marked reduction in immunoglobulin (Ig) isotype switching and a lack of affinity maturation that would normally arise from the processes of somatic hypermutation and selection that occur in GC responses. Central and exclusive to B cells responding to TD antigens is their entry, expansion, and differentiation that occur within the follicles of secondary lymphoid tissues. Prior to their follicular development, B cells bearing receptors specific for a challenge antigen need to engage TH cells primed to peptides from that same antigen within the context of interdigitating dendritic cells (IDC), the potent antigen-presenting cells (APC) located in the T-rich zones that surround the follicles. 10 Appropriate signalling among this cellular triumvirate will, independently of any follicular involvement, drive the extrafollicular (EF) expansion of the B cells and generate plasma cells producing some IgG but primarily IgM of low-to-moderate affinity for the antigen that initiated the response. While clearly not essential, it remains open as to whether CD40 engagement might modify the initial EF phase of B cells responding to TD antigen. This possibility was first highlighted by the study of Wheeler et al. where it was shown that co-ligating CD40 with IgM on the surface of resting (primarily EF) tonsillar B cells lowered the threshold requirement for signalling through the B-cell receptor (BCR) by approximately 1000-fold. 11 Unlike the well-established threshold lowering of the BCR signal seen on co-engaging the CD19/CD21 complex, 12 the CD40-dependent enhancement of proliferation that was observed with low levels of antigen surrogate (anti-Ig) was not a consequence of amplifying the BCR-dependent Ca2+ response (K. Wheeler & J. Gordon, unpublished obser-vations). Despite the dramatic outcome of CD40/BCR co-ligation in vitro, it was unclear as to when this combination of signals might be encountered physiologically: as na?ve B cells fail to turn on na?ve T cells13, 14 it would be necessary to implicate the IDC – either directly or indirectly – in the provision of CD154 during an early phase of the response. Following their migration into 2° lymphoid tissue, Langerhans cells that have captured and processed antigen in the periphery turn off their antigen-processing machinery 15 and may thus hold residual antigen in a native configuration to be recognized by the B cell. This possibilty has recently been addressed by Wykes et al. 16 with in vivo studies in the rat. When spleen DC were transferred from rats immunized with haptenated keyhole limpet haemocyanin (KLH) to na?ve control animals, antibody responses to both hapten and carrier protein were observed in the recipients. 16 This strongly suggests that DC can retain intact antigen in vivo. Importantly, Wykes et al. were also able to conclude that such retention could reach 36 hr, a time-frame compatible with the reported kinetics for activation of both TH cells and B cells during primary antibody responses in vivo.16 Therefore, as long as it were still retained at the site of priming, a CD154-expressing TH cell would have the opportunity to signal – via CD40 engagement – an antigen-captured (and BCR-triggered) B cell ( Fig. 1). The triumverate model of B-cell activation. Having first taken up and processed antigen outside of secondary lymphoid tissue dendritic cells then migrate to the T cell-rich zones ready to present the foreign peptides to specific TH cells traversing these sites (1). Tissue-resident IDC constitutively express high-level major histocompatibilit complex (MHC) class II, CD40 and ligands for CD28. In this model, peptide/costimulatory molecule activation of the TH cell induces CD154 expression (2) that on docking with IDC-CD40, promotes expression of IDC-CD154 (3). Residual unprocessed antigen on the IDC allows receptor-specific B cells to be trapped and engaged (4) not only by BCR but concomitantly via CD40 engagement of the IDC-expressed CD154; CD154 can also be delivered to the B cell via the T-cell-expressed ligand. Thus an environment for BCR/CD40 co-ligation is established for efficient B-cell activation in response to a low concentration of captured native antigen. One difficulty with the concept of a physical ‘ménage-?-trois’ (as evoked in Fig. 1) is the sheer implausibility of the three cell types, each infrequent and motile, being together in the same place at the same time. A way out of this conundrum comes from the mounting evidence that implicates DC directly as providers of co-stimulatory signals to B cells, including that of CD40L. 17 Clark and co-workers have shown that ligating CD40 on human DC leads to an induction of both CD154 mRNA and functional protein: in this setting, CD40 and BCR could be simultaneously engaged by ligands presented at the DC surface. 18 This would closely mimic the in vitro system established by Wheeler and colleagues where B cells are activated by antibodies to CD40 and BCR presented on the surface of mouse fibroblasts. 11 Thus, the amount of native antigen needed to be held by a DC in order to stimulate a na?ve B cell into the proliferative cycle would be greatly diminished where CD40L had been induced: perhaps as a consequence of CD40–CD154 interactions generated during prior union of the DC with a specifically primed TH cell ( Fig. 2). This would be analogous to the model proposed by Matzinger and colleagues for the activation of cytotoxic T cells, whereby a CD40-positive DC is ‘licensed’ by a CD154-expressing TH cell to stimulate – at some later juncture – an antigen-specific TC cell. 19 Licensing of IDC for B-cell activation. (a) Foreign peptide-specific TH cells are primed by IDC as described in Figure 1. (b) Induction of CD154 on IDC (3) by the above interaction licenses the IDC to subsequently activate B cells bearing BCR specific for residual native antigen (4) even once TH cells have left the site of initial priming. Again, conditions for joint BCR/CD40 ligation are established, equipping the B cell with co-stimulatory molecules and processed peptide (5) for subsequent cognate engagement with primed TH cells. Cytokines released by the activated IDC, among which will be IL-12, contribute to ensuing B-cell development. Given the capacity of CD40L to stimulate na?ve B cells directly, the possibility then exists of bystander activation for any B lymphocyte that happened to be traversing sites of priming in an ongoing TD response. However, Nature seems to have tailored an elegant fail-safe mechanism to militate against such a potentially disastrous outcome. Garrone and colleagues first reported the rapid induction of Fas (CD95) that occurs following the ligation of CD40 on resting tonsillar B cells: the appearance of Fas sensitized the B cell to Fas-mediated apoptosis manifesting as an inhibition of the CD40-dependent B-cell growth and differentiation that would otherwise develop in the presence of stimulating cytokines. 20 In an elegant extension to this work, Defrance and co-workers demonstrated that additional engagement of BCR – while not suppressing CD40-triggered induction of Fas on either na?ve or memory B cells – nonetheless, engendered complete protection from Fas-mediated death. 21 Thus, while all B cells engaging CD40L at sites of priming – either by design or accidentally – would be sensitized for Fas-mediated elimination, those that had been recruited into the response by virtue of their antigen specificity would be uniquely spared this fate. Dendritic cells are a rich source of the TH1-driving cytokine interleukin-12 (IL-12) 22 with CD40 ligation providing an important signal for its induction. 23 It has recently been asked whether, and how, B cells receiving TD signals – potentially at DC surfaces – respond to IL-12 exposure. For human B cells primed by CD40 and BCR ligation, IL-12 was found to drive them to a phenotype, characterized by CD38 hyper-expression, that was indistinguishable from that achieved with the prototypical TH1 cytokine interferon-γ (IFN-γ). 24 The effects of IL-12 were shown to be wholly dependent upon the induction of autocrine IFN-γ. Interleukin-10 was capable of dampening IL-12's actions by suppressing endogenous IFN-γ production but failed to perturb the effects of exogenously added IFN-γ. The distinct phenotype of CD23 hyper-expression promoted in B cells primed with TH2 signals was left untouched by the presence of IL-12. It was concluded that the results support a model for the progressive and hierarchical commitment of B cells to polarization during a developing TD response dominated at the level of the TH cell rather than that of the DC. 24 This was considered consistent with recent findings from the mouse demonstrating that B cells participating in TD responses in vivo manifest polarization along a TH1 or TH2 pathway shortly after antigen challenge with the direction taken being dependent upon the nature of the immunogen. 25 Interestingly, Gray and colleagues have shown that DC-derived IL-12 encourages CD40-primed murine B cells to convert conditions that normally drive T cells along a TH1 pathway to those suitable for TH2 development: 26 this lends support to the notion that IL-12-producing DC do not – by default – establish an exclusive TH1 environment at sites of priming in T-dependent B-cell responses. What makes a na?ve B cell being recruited into a TD response choose between a follicular or extrafollicular route of development? Could it simply be stochastic? Or, as suggested by some, reflect heterogeneity amongst na?ve B cells with distinct precursors individually committed to primary and memory pathways? Or, is it more a reflection of the signals received? As interrupting the CD40–CD40L interaction is known to abrogate the follicular recruitment of B cells participating in TD responses, 27 attempts to model this pathway in vitro to address such questions have naturally included the provision of a CD40 signal. In 1996, two papers reported on how EF B cells react to co-ligation of their CD40 and BCR by developing a phenotype with features at least partially reminiscent of GC cells. Wheeler and Gordon found that dual signalling of EF B cells through CD40 and BCR led to: increased CD19 and CD20 expression; low-level expression of CD23; a down-regulation of sIgM and sIgD with the appearance of a minor, but significant population, of IgG-expressing cells; the up-regulation of CD38. 28 While all these features are compatible with a GC phenotype, others were not. Thus CD5 was induced on almost half the cells while the entire population exhibited CD44 hyper-expression: GC centroblasts are CD44neg while centrocytes are CD44low. A similar outcome was noted by Galibert and colleagues who not only confirmed the induction of CD95 as previously described but also reported the appearance of the GC-selective enzyme carboxy-peptidase-M in EF B cells following BCR/CD40 cross-linking. 29 Again, however, the failure to down-regulate CD44 and an inability to induce CD10 highlighted the incomplete conversion of EF B cells to a GC phenotype. A suggestion that engagement of CD44 may itself contribute toward the development of a GC population recently came from studies by Ingvarsson et al. in which provision of a CD44 signal was reported to enhance the actions of BCR/CD40 co-ligation in this regard. 30 Reasoning that the strength of the CD40 signal might be a crucial factor in shaping whether a B cell takes the follicular route or expands extrafollicularly, McCloskey and colleagues established cultures of EF tonsillar B cells on transfected fibroblasts carrying different levels of CD154. 31 By this approach they were able to demonstrate – for the first time – high-level de novo induction of CD77/globotriaosylceramide, a defining antigen for B cells located within GC. Critically, induction of CD77 to levels comparable to those found on freshly isolated GC B cells required a high threshold expression of CD154: a 50% reduction in the amount of CD40L carried by the fibroblasts was sufficient to ablate CD77-inducing capacity almost completely. Induction was rapid, being detected as early as 12 hr post-stimulation. While the physiological role of CD77 remains unclear, the induced molecule was shown to be functional in that subsequent exposure of expressing cells to E. coli O157-derived Verotoxin-1 – a known ligand for the globotriaosylceramide – resulted in their toxin-elicited death. 31 Importantly, the findings of McCloskey et al. established that while the engagement of an EF B cell with cell membranes carrying relatively modest levels of CD40L was sufficient to promote optimal entry into the cell cycle, a higher threshold of CD40 engagement needed to be reached before it could acquire the defining GC characteristic of CD77 expression. It is likely that such a situation would be achieved physiologically only once a critical level of T-cell priming had been attained. Thus the availability of induced CD40L on the TH cell – or even, as discussed above, on the DC following its prior conditioning during T-cell priming – would dictate the pathway to be taken by an antigen-holding B cell ( Fig. 3). An elegant study by Wilson and colleagues using mice carrying a conditional CD154 transgene came to an identical conclusion by showing that CD40L abundance was a critical limiting factor in dictating the level of high-affinity IgG antibodies (a process requiring GC formation) developed in response to immunization by TD antigens: small percentage increases above normal amounts of CD40L were sufficient to drive large shifts in the quantity of mutated and selected antibodies produced. 32 Deciding whether to develop inside or outside of the follicle. IDC-primed, antigen-specific B cells and TH cells can now make productive cognate interactions via MHC-carried peptide and TCR (1) that could, in turn, further up-regulate CD154 on the TH cell leading to additional signalling through B-cell CD40 (2). Early shaping of the TH-cell cytokine profile (3) will impinge on the polarization of the ensuing B-cell response. The quantity of CD154 available (as represented here on the TH cell but with a potential contribution additionally from the IDC) fashions whether the subsequent route taken by the B-cell remains extrafollicular or, instead, leads to its development inside the follicle (4). High CD40 occupancy directs the B cell into a neighbouring follicle where it is joined by activated TH cells (5) to undergo extensive proliferation and ultimately to create a germinal centre. Relatively low occupancy of CD40 drives an extrafollicular response culminating in the generation of plasma cells (PC) producing antibody (primarily IgM but with some IgG) of low-to-intermediate affinity (6). B cells that have experienced either low- or high-level CD154–CD40 engagement are induced to Fas/CD95 expression but are protected from FasL-dependent elimination – either within or outside of the follicle – if authorized by prior BCR signalling. Following a massive wave of intrafollicular expansion, B cells that have gained entry into this pathway then differentiate and compartmentalize to form the germinal centre 33 (see Fig. 4, upper panel). Dark zone centroblasts provide the targets for somatic V-gene hypermutation whilst light zone centrocytes provide the pool from which antigen trapped on follicular dendritic cells (FDC) selects the high-affinity mutations: selected cells emerge from the GC either to establish memory or to become plasmablasts providing an immediate source of high-affinity antibody. Checkpoints and outcomes in the germinal centre. Upper panel: Schematic view and taxonomy of the germinal center highlighting: centroblasts (CB) forming the dark zone; centrocytes (CC) contained within the light zone where follicular dendritic cells (FDC) are now also concentrated; tingible body macrophages (M) scattered throughout the GC in readiness to engulf apoptotic B cells; foreign peptide-specific TH cells polarized to the light zone and abundant within the outer zone that interfaces the follicular mantle comprising uninvolved B cells which were occupying the resting follicle prior to its colonization by antigen-responsive lymphocytes. Lower panel: Selection, maintenance, and exit from the germinal centre. Centroblasts proliferate within the dark zone possibly with homotypic interactions facilitating autocrine communication (1). These cells provide the targets for V-gene-directed hypermutation (2) with emerging centrocytes expressing altered BCR. Centrocytes bearing mutated receptors with high affinity successfully compete for the antigen that initiated the response – now trapped as immune complexes on FDC (3). Unsuccessful centrocytes succumb to apoptosis (4). FDC-selected centrocytes offer BCR-captured (and processed) antigen to CD45RO TH cells whereupon pre-formed cytoplasmic CD154 is made available for surface presentation to B-cell CD40 (5). At high concentration, this provides a permissive signal for centrocytes to return to the dark zone for further rounds of somatic mutation (6). A centrocyte that has mutated towards ‘self’– while capable of engaging and processing endogenous antigen – on failing to recruit cognate T-cell help, would rapidly be aborted (7). Additional signals from cytokine-expressing TH cells (possibly coupled with a diminution in available CD154) could serve to direct selected centrocytes out of the GC: bifurcation towards plasma cells providing an immediate source of high-affinity antibody (8) with others establishing a memory pool for a swift and effective response to future antigen encounters (9). The early observation that CD40 monocolonal antibody (mAb) delivered a potent survival signal to GC B cells ex vivo34 was given a sound physiological rationale when CD40L was first cloned and identified as an inducible surface molecule of TH cells. 35 Numerous studies have subsequently cemented the central role of the CD40–CD40L interaction both in maintaining GC responses and in the recruitment of centrocytes into the memory pool; 1-5 (see Fig. 4, lower panel). Whilst the studies with CD40 mAb highlighted the emergence of out-of-cycle cells from the GC population – compatible with differentiation towards memory – the possibility that CD40 signalling might also be involved in maintaining a centroblast pool was intimated by studies of Holder et al. 36 Here it was shown that GC B cells placed on membranes from cells transfected with CD40L not only continued to cycle but also, notably, remained Bcl-2 negative: the turn-off of Bcl-2 is a phenotype that among mature B cells is not only unique to those residing within the GC but is also a necessary component of their ability to be selected for incorporation into the memory pool. 37 As CD154-carrying membranes appear to support cycling of the total GC B-cell population in vitro, it has been suggested that centrocytes encountering strong CD40 signals might be encouraged back into the dark zone to undergo further rounds of mutation and selection. 36 Ex vivo, GC B cells can be maintained on fibroblasts presenting a CD40 signal for a maximum of 3–4 days. 36 In the GC microenvironment there would be the opportunity for other growth-promoting factors to be encountered being delivered by, for example, FDC or the TH cells located in the light zone. A study of candidate cytokines that might prolong the CD40-dependent growth of GC B cells in vitro identified sets of triple combinations effective in this regard. 38 Among five different cytokine cocktails synergistic for GC B-cell growth, IL-10 was common to all. Other cytokines demonstrating activity included: IL-1β, IL-2, IL-3, IL-6 and IL-7. Interestingly, IL-3 + IL-7 + IL-10 – an effective combination for supporting CD40-dependent GC B-cell growth – had previously been reported by Saeland et al. 39 as providing optimal conditions for maintaining long-term DNA synthesis in human bone marrow B-cell precursors. Several of the cytokines found to be active in supporting GC B cells in vitro have been identified at the mRNA level in GC T cells. 40 IL-1β has been shown at both the mRNA and protein level in tonsillar FDC 40 while Kopf and colleagues have reported a requirement for IL-6 in the development of murine GCs: low-density cells of dendritic morphology, possibly FDC, were candidates for the local source of IL-6 in this setting. 41 Finally, de Saint-Vis et al. have described a unique population of GC–DC containing mRNA for IL-7 and IL-10 42 suggesting that this DC subset may also contribute to the propagation of GC B cells in situ. Although GC B cells could be maintained for extended periods (up to 4 weeks) with selected cytokine combinations, sustained growth demanded both a constant CD40 signal and the continuous presence of stroma: importantly, populations recovered from such long-term cultures retained features of GC B cells with a significant proportion being CD38hi/CD44neg, the characteristic phenotype of centroblasts. The factor(s) being provided by the fibroblast feeder layer remain uncharacterized but they could be substituting for an activity normally supplied by FDC: for example, Li et al. recently described a 282 amino acid protein on FDC that supports the CD40-dependent growth of GC B cells. 43 A similar dependence on persistent CD40 signalling for sustaining cytokine responsiveness has been noted for ovine B cells isolated from lymphoid follicles of Peyer's patches. 44 In addition, Kelsoe and colleagues have demonstrated a continued requirement for CD154 in maintaining murine GC responses in vivo. They showed that administration of anti-CD40L to mice even as late as 6–10 days into a TD response led to an abrupt dissolution of the GC reaction established. 9 Using modifications to the culture system described above, it was asked whether selective cytokine cocktails – in combination with CD40 signals – could support the active V-gene mutation that occurs within centroblasts in vivo. Single GC B cells were cultured by limiting dilution (to allow detection of mutations arising during proliferation in vitro) with two of the cytokine combinations capable of supporting similar levels of CD40-dependent growth: IL-10 + IL-1β + IL-2 and IL-10 + IL-7 + IL-4. 45 Analysis of VH3 genes from cultured cells by reverse transcription–polymerase chain reaction (RT-PCR)-based single-strand conformation polymorphism indicated that the combination IL-10 + IL-1β + IL-2 promoted active V-gene mutation whereas IL-10 + IL-7 + IL-4 was ineffective. This was confirmed by sequencing which also revealed that the de novo generated mutations were located in the complementarity-determining and framework regions and shared characteristics with those arising in V-gene hypermutation in vivo. This led to the conclusion that somatic mutation in the target GC B-cell population may therefore be actively cytokine-driven and not simply a consequence of continued proliferation. 45 The active contribution of CD40 signalling to the hypermutation process is presently unclear although the ability of the IL-10 + IL-1β + IL-2 combination to sustain mutational activity was, de facto, CD40-dependent as GC B cells could not be maintained in the absence of CD40 engagement. Probably one of the most crucial roles conferred on CD154 resides in its capacity to check out the possibility that – among the random mutations generated during the centroblast stage – some centrocytes may now be encoding receptors for self. CD154 exists pre-formed, located in the cytoplasm of CD45RO TH cells scattered within the centrocyte-rich light zone and concentrated in the outer zone of GC (where they are ideally placed to act as sentinels to check emigration of B cells): 10 on activation, this CD154 is rapidly mobilized to the cell surface. 46 Gray and colleagues have demonstrated that GC B cells are capable of presenting processed antigen to GC T cells 47 while, more recently, it was shown that the T cells colonizing a follicle post-immunization with TD antigen are clonally related progeny of the TH cells activated extrafollicularly during the priming phase and thus share specificity for the processed immunogen that initiated the reaction. 48 Taking the above in toto, a train of events can be established that – if operable – would militate against the harmful consequences of recruiting autoreactive mutants of possibly high affinity into the memory pool, with CD154 serving as the central arbiter. Thus, centrocytes that have successfully competed for foreign antigen held in the form of immune complexes on FDC process then present the peptides to GC T cells, the resulting T-cell receptor (TCR) engagement leading to the swift induction of CD40L that on docking with the B cell's CD40 delivers survival and differentiation signals; any centrocyte recognizing and capturing self antigen would be denied such intimate contact with the immunogen-specific T cells and, by default, succumb to apoptosis in the absence of engaging CD154. Again, to ensure that such a fail-safe mechanism remains that way, it has been proposed that – as during the initial EF response – any by-stander CD40-activated centrocytes (that could include autoreactive members) are eliminated as a consequence of Fas–FasL interactions for which they would now be primed. 49 Following CD154-dependent selection in the light zone, what is the centrocyte's next move? The possibility that some may return to the dark zone to reactivate the mutational machinery has been discussed. A variety of in vitro studies, although primarily on naı¨ve B-cell populations, have suggested that CD40L provides a potent permissive signal for cytokine-dependent plasma cell differentiation.50, 51 Other experiments have indicated that the converse may be true – i.e. that ligating CD40 actively inhibits B cells from differentiating toward antibody-secreting cells. A detailed study aimed to address this controversy resulted in firm support of the latter thesis. Thus, Randall et al. showed that both in B-cell lines and isolated GC B cells from mice, membrane-presented CD154 actively suppressed antibody secretion by repressing the steady-state levels of differentiation-specific, but not activation-specific, genes. 52 Such inhibition could not be relieved by cytokines or BCR signalling. Only on disruption of the CD154–CD40 interaction could B cells be coaxed to terminal differentiation. It was suggested that the data accounted for the observed lack of plasma cells within GC, a prerequisite for the selection of high-affinity antibody-forming cells. The conclusions are consistent with earlier observations indicating that CD40 ligation encourages the emergence of cells with a memory phenotype from GC populations. They are also supported by the lack of plasmacytoid differentiation noted in CD40-dependent cultures of GC B cells maintained with cytokine combinations among which included factors (e.g. IL-6 and IL-10) that are more typically associated with promoting antibody secretion. 38 The focus of this review has been firmly on CD154 as a stimulating activity delivered to the B cell via an appropriately recruited T cell. Indeed, central to its pivotal role in antigen-specific T-dependent B-cell responses is the tenet that CD40L availability is restricted both temporally and spatially during cognate T–B interaction; first at the EF priming phase then during the selection stage in the GC. However, the existence of DC-expressed CD154 has already been mentioned while platelets, basophils, and mast cells have also been described as providers of CD40L. As long as CD40L expression is tightly controlled, which it appears to be, its occurrence on these different cell types is not overly disturbing: but what if B cells themselves could express CD154? It seems that – under certain conditions – they can. Induction of CD154 after polyclonal activation has now been described for both human53, 54 and murine B cells. 55 The levels expressed do, however, appear to be low. Thus, in t