Ca²⁺-dependent Ras/Erk signaling mediates negative selection of autoreactive B cells

Abstract

Signaling via the Ras/Erk pathway has long been recognized to be critical in lymphocyte development and function, yet the mechanisms that control the distinct functional outputs of this pathway in different cellular contexts remain poorly understood. Our recent results have demonstrated unexpected involvement of Ras/Erk signaling in the sensitization of B cells to apoptosis in order to eliminate autoreactive cells. Increases in cytosolic Ca²⁺ are necessary and sufficient to induce activation of this Ras/Erk pathway, and the biochemical events involved in its activation are different from the ones involved in diacylglycerol (DAG)-mediated Ras/Erk activation. Developmental regulation of upstream mediators of these distinct pathways contributes to their predominant activation at different stages of B cell development. These findings have revealed a mechanism by which antigen stimulation can activate distinct Ras/Erk pathways at different developmental stages to mediate appropriate functional outputs that control the selection, development and activation of B cells. Despite these recent findings, however, much remains to be learned about the molecular mechanisms that confer functional specificity to common Ras/Erk signaling modules.

This article refers to:

The Ras pathway has long been recognized as an important pathway in lymphocyte development and activation.1 Ras proteins were originally identified on the basis of their oncogenic potential, and have since been implicated in the regulation of important signaling pathways in many cell types. Ras proteins have intrinsic GTPase activity and exist in a dynamic equilibrium between the GTP-bound state (active) and the GDP-bound state (inactive); RasGTP production is catalyzed by Ras guanine nucleotide exchange factors (RasGEFs), while hydrolysis to RasGDP is accelerated by Ras GTPase-activating proteins (RasGAPs). Ras signaling typically involves activation of a GEF, such as SOS, RasGRF or RasGRP, which triggers RasGTP production. RasGTP then activates Raf proteins; activated Raf phosphorylates MEK and MEK phosphorylates Erk. Erk has the potential to phosphorylate many targets, making this a remarkably promiscuous pathway, and its activation has been implicated in a wide variety of cellular processes in multiple cell types. Therefore, the mechanisms that convey specificity to this pathway have been an active field investigation. Current models suggest that scaffolding of Ras/Erk signaling complexes plays a crucial role in determining the functional output of the signals.2,3 By bringing together the upstream activators of the pathway, the components of the Ras/Erk cascade and the appropriate substrates, scaffolds ensure that the pathway is activated by the proper stimuli and at the proper location to gain access to appropriate substrates while being sequestered away from other non-specific substrates.

Stimulation of lymphocytes via the antigen receptor triggers uniquely high levels of RasGTP production,1,4,5 and recent work has elucidated the mechanisms by which such elevated levels of RasGTP are generated. Lymphocytes express RasGEFs from at least two different families, namely SOS and RasGRP. RasGRP proteins contain two Ca²⁺-binding EF hands and a diacylglycerol (DAG)-binding C1 domain; expression of RasGRP1 and RasGRP3 is largely restricted to neuronal and hematopoietic cells. DAG binding to RasGRP has been implicated in its activation, as has PKC-mediated phosphorylation of a threonine residue (T184 in RasGRP1, corresponding to T133 in RasGRP3).6–8 SOS proteins, in contrast, are ubiquitously expressed and their activation is regulated by recruitment to sites of receptor or adaptor phosphorylation. In addition, SOS proteins contain an allosteric pocket that, upon binding RasGTP, increases their GEF activity 75-fold.9 A current model proposes that, upon antigenic stimulation, RasGRP is initially activated by DAG, leading to production of an initial pool of RasGTP; this RasGTP can then bind to the allosteric pocket of SOS, increase its catalytic activity and thereby drive a feed-forward loop for RasGTP production. The synergistic activation of these two GEFs would then account for the unusually high RasGTP production observed in lymphocytes.10,11

Although DAG production has been extensively associated with activation of Ras/Erk signaling in lymphocytes, Ras/Erk pathway activation in lymphocytes has been thought to be largely Ca²⁺-independent. This notion comes from early studies demonstrating that DAG analogs (such as PMA), but not Ca²⁺ ionophores, could potently induce the Ras activation in lymphocytes.4 The involvement of RasGRP and PKCs in Ras/Erk activation in lymphocytes explains the early observations that showed that DAG analogs could potently induce Ras activation. However, in other cell types, perhaps most notably in the nervous system, it has become increasingly clear that Ca²⁺ can lead to Ras/Erk activation by means of Ca²⁺-responsive GEFs (in addition to RasGRP, neuronal cells express RasGRF, which is potently activated by Ca²⁺).12

Our recent results now demonstrate that Ca²⁺ can potently induce Erk activation in immature and transitional bone marrow B cells. Our unexpected findings linking Ca²⁺ to Erk activation were first obtained while examining a B cell line overexpressing Stim1. Stim1 is an ER transmembrane protein that contains a Ca²⁺-binding EF hand domain localized within the lumen of the ER. Upon ER Ca²⁺ store emptying, Stim1 becomes unbound from Ca²⁺, aggregates into punctate structures, and moves toward the plasma membrane,13–16 where it interacts directly or indirectly with plasma membrane channels (Orai proteins) and causes them to open.17–22 Extracellular Ca²⁺ then enters the cell following a ten thousand-fold gradient, a process known as store-operated Ca²⁺ entry (SOCE), thereby sustaining increases in cytoplasmic Ca²⁺ concentration. DT40 avian B cells that were made to overexpress Stim1 displayed augmented amplitude and duration of SOCE, which was sufficient to sensitize the cells to antigen receptor-induced apoptosis. In this Stim1-overexpressing cell, Ca²⁺-dependent Erk activation required PKCδ and RasGRP, but not SOS, proteins, and disrupting this pathway by means of RasGRP deletion was sufficient to rescue the cells from antigen receptor-induced apoptosis. Strikingly, this pathway is also active in primary B cells, where this Ca²⁺-Erk pathway is selectively activated at the bone marrow immature and transitional stages of B cell development. These developing B cells undergo apoptosis in response to antigen, one critical mechanism that eliminates self-reactive B cells to prevent later autoimmunity (Fig. 1). We also demonstated that PKCδ and RasGRP1 protein expression is highest at these developmental stages. Moreover, PKCδ^-/- B cells display a competitive advantage over wild type cells at this developmental stage concomitant with a marked defect in activation of the Ca²⁺-Erk pathway, and the PKCδ^-/- mice subsequently develop severe B cell autoimmunity. In contrast, developing bone marrow B cells that overexpress Stim1 display increased propensity to negative selection, an effect that depends on PKCδ.23

The involvement of PKCδ in this pathway was unexpected and revealed additional mechanisms that modulate RasGRP1 function that were not previously known. We identified a new phosphorylation site on RasGRP1, S332, which is a putative PKCδ target site and selectively modulates the responsiveness of RasGRP1 to Ca²⁺. Phosphorylation at S332 is not required to activate Erk via the DAG pathway, but it is absolutely required for Ca²⁺-mediated Erk activation. Further, although phosphorylation of T184 on RasRGP1 (which corresponds to T133 on RasGRP3) had previously been identified and shown to be crucial for optimal Erk activation via the DAG pathway,7,8,24 phosphorylation of this site was largely dispensable for Ca²⁺-dependent Erk activation. These results demonstrated that RasGRP1 activity can be modulated by multiple mechanisms. These mechanisms may involve Ca²⁺, DAG or both second messengers, and likely involve at least two phosphosites, T184 and S332. Different cellular perturbations at different developmental stages may therefore induce RasGRP1 activation by unique mechanisms that contribute to its functional specificity.

By homology modeling, we predicted that S332 lies in close proximity to the Ras-binding site within the CDC25 domain of RasGRP1. A phosphorylation event at this site may create electrostatic interactions that influence Ras binding. Indeed, it is possible that such interactions may change the affinity of different isoforms of Ras or Ras-like GTPases for RasGRP1. Because different GTPases have unique trafficking and localization properties, a phosphorylation event that alters the choice of Ras molecule to activate could very well have an impact on its functional output. Alternatively, phosphorylation of this site could establish electrostatic interactions with other charged residues within the CDC25 domain, particularly in flap2, which is predicted to abut the phosphosite. Such interactions could thereby coordinate the enzymatic activity of the exchange factor. Our results indicate that mutation of this site to alanine disrupts the responsiveness of RasGRP1 to Ca²⁺. However, a BCR-induced DAG signal is sufficient to trigger activity of this mutant, albeit at lower levels than wild type. These results demonstrate that mutation of this site does not sufficiently alter the conformation of RasGRP1 to disrupt the enzymatic activity of the protein, but rather modulates its responsiveness to Ca²⁺. However, based on the reduced activity observed in response to the BCR-induced DAG signal, it is possible that subtle conformational changes in the CDC25 domain induced by phospho-S332 may improve the enzymatic function of the protein.

It is noteworthy that a phosphomimetic mutation at this site, S332E, RasGRP1 behaves like WT RasGRP1, indicating that phospho-S332 is permissive for activation of both the Ca²⁺- and the DAG-induced Erk pathways. However, this phosphomimetic residue does not lead to constitutive activation of either pathway. Therefore, while phosphorylation of RasGRP1 on S332 is required for Ca²⁺-dependent Erk activation, it is not sufficient to activate the pathway, implying that other Ca²⁺-dependent events may be at play. As mentioned above, RasGRP proteins contain both a DAG-binding C1 domain and two Ca²⁺-binding EF hands. Mutagenesis studies have uncovered an important role for the C1 domain in DAG-mediate Erk activation, but the role of the EF hands is still not well understood.7,8,24 Interestingly, and consistent with our results, one report demonstrated that RasGRP1 can sensitize a WEHI-231 immature B cell line to antigen receptor-induced apoptosis, and this effect seemed to depend on the EF hands.25 This implies that Ca²⁺ binding to RasGRP1 via the EF hands may be important for its ability to mediate apoptotic signals. Combined with our findings, these results suggest that Ca²⁺ binding may either alter the conformation of RasGRP1 to facilitate phosphorylation of S332, or may synergize with this phosphorylation event to induce pro-apoptotic signals in B cells. However, at this point we cannot rule out that other Ca²⁺-dependent events are at play in this pathway, possibly involving additional Ca²⁺-modulated proteins.

Another mechanism that may play an important role in establishing specificity in these pathways is control of subcellular localization of signaling complexes by Ca²⁺ and DAG. It is widely recognized that gradients of these second-messengers can recruit proteins to subcellular locations where the second-messengers are generated. As mentioned above, developing B cells that are susceptible to negative selection display increased amplitude and duration of SOCE and lower levels of DAG in response to antigen stimulation as compared with their mature counterparts.26–29 It is therefore possible that these second messengers compete to recruit active Erk complexes to particular subcellular compartments where their downstream substrates will vary. While speculative, this hypothesis is consistent with previous reports that demonstrate that RasGRP proteins can relocalize within cells in response to stimulation.30 In addition, the fact that Ca²⁺ and DAG induce Erk activation via distinct biochemical pathways that rely upon proteins that contain Ca²⁺ and DAG binding domains suggests that these second messengers may act, not only to promote activation of one pathway, but also to simultaneously suppress activation of the second pathway, a model that makes a lot of sense when diametrically opposite cell fates are at stake (e.g., proliferation vs. apoptosis). Based on this, it is not far-fetched to imagine that, while Ca²⁺-induced phosphorylation of S332 and possibly Ca²⁺ binding to the EF hands of RasGRP1 promote apoptosis, DAG binding to the C1 domain might alter its conformation and/or localization sufficiently to disrupt signaling from this Ca²⁺-activated complex, while promoting signaling from a newly-formed DAG-dependent complex. This idea becomes even more compelling when we consider the fact that PKCδ is a non-classical PKC, which contains only a DAG binding C1 domain but no ability to bind Ca²⁺, yet it is this PKC isoform that is involved in RasGRP1 modulation downstream of Ca²⁺ elevations. Therefore, it is possible that other Ca²⁺ dependent events (perhaps Ca²⁺ binding to RasGRP1 and/or phosphorylation of S332 RasGRP1) act to promote the PKCδ-RasGRP1 interaction, while DAG binding could sufficiently alter the conformation and localization of these proteins to disrupt their interaction. This idea is also consistent with published observations that suggest that retention of PKCδ at the plasma membrane by DAG may be a mechanism to inhibit its proapoptotic functions.

The subcellular localization of Ras/Erk signaling modules is increasingly viewed as a crucial mechanism that confers specificity to these signaling pathways. This model proposes that scaffold proteins that localize to different subcellular compartments recruit Ras/Raf/Mek/Erk signaling modules to a particular location where the signaling cascade can be activated by appropriate stimuli and simultaneously gain access to appropriate substrates. This model for signaling specificity is not exclusive to MAPK signaling, and one other well characterized example of functionally distinct signaling complexes that contain the same effector molecules are the mTOR-containing complexes mTORC1 and mTORC2.31–33 Examples of Ras/Erk scaffolds are the KSR proteins, CNK, Sef and MP-1,2,34–38 but it is unclear whether any of these proteins play a role in the induction of either of these Erk pathways in B cells. Studies in T cells have shown that Ras/Erk signaling can occur at the plasma membrane as well as in endomembranes, most prominently in the Golgi, and RasGRP1 has also been found to shuttle between these compartments in response to stimulation.30,39,40 In DT40 B cells, RasGRP1 has been found to accumulate at the plasma membrane in response to BCR crosslinking, and this effect seems to require EF1, but not Ca²⁺.41 Previous data demonstrate that PLCγ is recruited to the site of antigen stimulation, and that a gradient of DAG from this site is generated and contributes to the recruitment of proteins toward this site.42,43 In contrast, the migration of Stim1-containing punctae toward the plasma membrane in cells stimulated with thapsigargin or soluble antibody is not constrained to a particular section of the membrane, implying that CRAC channels likely open throughout the plasma membrane.13,15 Although this does not appear to be the case when the stimulation depends on cell-cell contact and an immunological synapse is formed, in these cases Stim1 punctae have been shown to accumulate not only at the synapse, but also at the distal cap of the cells.44,45 These observations suggest that perhaps spatial sequestration of DAG and Ca²⁺ plays an important role in controlling the localization of Erk signaling: DAG-mediated restriction of Ras/Erk modules to sites of antigen stimulation might protect cells from apoptosis, while Ca²⁺-mediated Ras/Erk signaling in endomembranes might confer pro-apoptotic signals (Fig. 2). However, this is in direct contrast with what has been observed in T cell selection: in this system, it was found that thymic selection threshold correlated with compartmentalized Ras/MAPK signaling, and plasma membrane localization of the signals was correlated with thymic negative selection, while localization to endomembranes correlated with positive selection.46 It will therefore be essential to further investigate how the subcellular localization of these signaling modules is controlled by Ca²⁺ and DAG, how these changes in location alter their functional output, what scaffolds control their spatial distribution, and what signals are triggered downstream of Erk at these different locations to mediate proliferation or apoptosis.

Another possible means of regulation of these pathways is by MAPK phosphatases (MKPs). These proteins belong to the dual-specificity family of protein phosphatases (DUSPs), which have the ability to dephosphorylate both phosphotyrosine and phosphoserine/phosphothreonine residues within the same substrate.47 There are ten members of the DUSP family known to act as MKPs, each of which displays unique substrate specificity for Erk, JNK or p38. Expression of these proteins is tissue-specific and developmentally regulated. In some cases, they can act as inducible/early response genes, thereby triggering negative feedback loops over activated MAPKs. Recent studies have uncovered important roles for DUSP1, DUSP2 and DUSP10 in the control of inflammatory responses,48–54 but how these proteins affect lymphocyte development is a question that remains largely unexplored. Interestingly, publicly available data through the Immunological Genome Project website (www.immgen.org/index_content.html) demonstrates that DUSP2, DUSP6 and DUSP10 expression is substantially regulated during B cell development and is much higher in immature and transitional bone marrow cells than in any other B cell compartment. Because this is the same stage where Ca²⁺-dependent Erk activation occurs, it is provocative to speculate that the expression of these particular DUSPs could shape the outcome of an Erk response by preferentially inhibiting Erk signaling from one, but not the other Erk pahway. It will therefore be interesting to examine whether B cell development and/or function is affected in mice lacking these proteins, and whether the subcellular localization of these phosphatases could relate to the differential outcome of Erk signaling downstream of Ca²⁺ or DAG.

An additional layer of complexity in the regulation of these pathways relates to the unique biochemical and developmental roles of the different RasGRP proteins. RasGRP1 and RasGRP3 are both expressed in B cells, yet we have found that their expression is regulated very differently during stages of B cell development. Neither of these RasGRP proteins is expressed early in B cell development. RasGRP1 expression is robustly upregulated in immature and transitional B cells, and substantially downregulated in mature B cells. In contrast, RasGRP3 expression levels are relatively low in immature and transitional B cells,23 but are very high in mature B cells.23 Interestingly, although PKCδ is highly expressed in early B cell development, its expression is further increased at the immature and transitional stages, followed by a substantial decrease in mature B cells.23,55 The pattern of expression of these proteins is intriguing and suggests unique functions of the different RasGRP isoforms during development. Indeed, the B cell phenotypes of RasGRP-deficient mice are consistent with this idea. Although it is not yet clear if these phenotypes are B cell intrinsic, RasGRP1-deficient mice develop B cell lymphoproliferation and autoantibody production, while RasGRP3-deficient mice do not.56–58 These characteristics of the RasGRP1^-/- mice are similar to what is observed in PKCδ^-/- mice. Further, RasGRP1/RasGRP3 double deficient mice do not show B cell lymphoproliferation, suggesting that RasGRP3 is required for the peripheral expansion of RasGRP1-deficient cells.56 These data suggest that different biochemical mechanisms are at play for these two exchange factors to mediate different roles during development, where RasGRP1 could play a role in B cell negative selection, while RasGRP3 may predominantly be involved in proliferative Erk signaling in mature B cells.

How might such an observation relate to the biochemical regulation of RasGRP signaling vis-à-vis the different RasGRP isoforms? One interesting observation is that, although serine 332 is conserved between RasGRP1 and RasGRP3, arginine 335 is not. This residue makes S332 a PKCδ consensus target site on RasGRP1 but not on RasGRP3. Further, this arginine is predicted to lie close enough to the putative phosphate on S332 to establish electrostatic interactions that may significantly change the Ras-binding pocket of RasGRP1 relative to RasGRP3. Indeed, differences in the ability of these two RasGRP isoforms to mediate nucleotide exchange of different GTPases have been documented,59,60 and justify the need to explore the mechanisms by which such specificity is obtained and its functional consequences. It is also not known what role SOS proteins may have during B cell development, and how the interplay between these two families of exchange factors may be regulated during B cell selection checkpoints.

The implication that Ras/Erk signaling may participate in B cell negative selection is interesting, given that recent reports implicate Erk as an important regulator of cell death in neurons and other cell types.61–63 In those systems, it has been observed that differences in duration and localization of Erk signaling correlate with distinct neuronal cell fates, such as proliferation, differentiation or apoptosis. Interestingly, many of the stimuli that can induce such responses, such as NGF or EGF, also induce activation of PLC and subsequent DAG production and Ca²⁺ entry. Moreover, neurons express proteins from another RasGEF family, namely RasGRF proteins, which are highly sensitive to regulation by Ca²⁺ via their Ca²⁺-calmodulin-binding IQ motif.12,64,65 Therefore, it is possible that neurons display analogous Ras/Erk regulatory mechanisms to the one we have uncovered in B cells.

Our recently published findings shed light on the mechanisms by which developmental regulation of SOCE and DAG production alters the functional output of antigen encounter in B cells. They implicate biochemically and functionally distinct regulatory pathways upstream of Ras/Erk in determining B cell fate during development, and raise many more questions concerning the regulation of these pathways. How is the specificity achieved? How does subcellular location of Ras/Erk signaling complexes relate to our findings? What is the role of scaffolding proteins, and which scaffolds are important for each of these pathways? What are the downstream effectors of Erk signaling that determine the proliferation vs. apoptosis decision? To what extent does the developmental control of protein expression alter the configuration of these pathways during development? These are all intriguing and important questions that need to be addressed in order to fully understand the mechanisms that confer functional specificity during B cell selection. Because development of appropriate lymphocyte repertoires is crucial to prevent autoimmune and lymphoproliferative disorders, understanding these mechanisms will be essential in efforts to therapeutically target one pathway with relatively minor or no consequences on closely related but functionally distinct pathways.

Figures and Tables

Figure 1 Antigen encounter triggers receptor editing or apoptosis in developing cells, but leads to activation and proliferation in mature B cells. Precursor B cells undergo VDJ recombination in order to generate a B cell repertoire that is immensely diverse, where each B cell carries a recombined receptor with unique specificity. Because many of these receptors recognize self, antigen encounter during the immature or transitional stages of B cell maturation leads to receptor editing or apoptosis in order to eliminate these autoreactive cells. Once B cells mature, antigen encounter generally triggers proliferative responses leading to an immune response.

Figure 2 Model for activation of functionally distinct Ras/Erk pathways. The model depicts spatial sequestration of Ca²⁺- or DAG-dependent signaling complexes as a potential means to couple Ras/Erk signals to appropriate downstream effectors. Ca²⁺-dependent Erk activation involves PKCδ-mediated phosphorylation of RasGRP1 on S332 and its activation is favored in immature and transitional B cells. DAG-dependent Erk activation may be favored in mature B cells, and likely requires activity of RasGRP3 and phosphorylation of T133 (corresponding to T184 on RasGRP1). Developmental regulation of RasGRP1, RasGRP3 and PKCδ protein levels, as well as changes in the amplification or negative regulation of Ca²⁺ and DAG, may contribute to the preferential activation of these pathways at different stages in B cell development.

Extra View to: Limnander A, Depeille P, Freedman TS, Liou J, Leitges M, Kurosaki T, et al. STIM1, PKCδ and RasGRP set a threshold for proapoptotic Erk signaling during B cell development. Nat Immunol 2011; 12:425 - 433; PMID: 21441934; http://dx.doi.org/10.1038/ni.2016

 Google Scholar