Introduction
The health and survival of many organisms depends on the development and function of an immune system that can adapt to an enormous diversity of pathogens without responding to self antigens and causing autoimmunity. In developing lymphocytes, variable-(diversity)-joining (V(D)J) recombination, which is initiated by the lymphocyte-specific RAG-1 and RAG-2 endonucleases, assembles exons encoding antigen receptor V regions to generate the vast repertoire of antigen receptors1. Organisms have evolved strategies such as editing of the B cell antigen receptor (BCR) to tolerate the large fraction of developing lymphocytes that assemble and express autoreactive antigen receptors. During BCR editing, autoreactive BCRs signal the initiation of secondary rearrangements to replace previously assembled exons that encode autoreactive BCR chains2, 3. So far, little is known about the molecular components and mechanisms by which regulation of BCR editing is achieved. Two intriguing studies published in this issue of Nature Immunology, one by Herzog et al.4 and the other by Amin and Schlissel5, suggest that autoreactive BCRs signal the nuclear accumulation of Foxo transcription factors that directly activate transcription of Rag1 and Rag2 to promote receptor editing. These findings identify a new, albeit anticipated, molecular link between autoreactive BCRs and secondary rearrangements that provides stimulating insights into the mechanisms that facilitate B cell tolerance.
In mice, B lymphocytes develop through a differentiation program that includes regulation of both expression of Rag1 and Rag2 and the assembly of BCR chains1, 3. Pro–B cells express Rag1 and Rag2 and assemble immunoglobulin heavy-chain V-region exons from VH, DH and JH segments. The assembly of a productive (in-frame) VHDHJH rearrangement leads to expression of an immunoglobulin H (IgH) chain that can pair with 
5-VpreB surrogate light chains to form a pre-BCR. Cell surface expression of pre-BCRs generates intracellular signals that downregulate the expression of Rag1 and Rag2 and promote the survival and proliferation of pro–B cells and the differentiation of pro–B cells into pre–B cells. Pre–B cells reactivate Rag1-Rag2 expression and assemble immunoglobulin 
-chain V-region exons from germline V and J segments. The assembly of a productive VJ rearrangement leads to the expression of an immunoglobulin -chain that can pair with the IgH chain to form a BCR, which is expressed on the surface of these immature B cells. Expression of an innocuous (not autoreactive) BCR generates ligand-independent ('tonic') intracellular signals that downregulate Rag1-Rag2 expression and promote the survival and further differentiation of immature B cells2, 3. However, cell surface expression of an autoreactive BCR can also lead to antigen-BCR interaction, internalization of antigen-BCR complexes and activation of intracellular signals that upregulate Rag1-Rag2 expression, halt differentiation and initiate secondary V rearrangements that replace the primary VJ rearrangement2, 3. This BCR selection and editing process can repeat until all J segments are deleted.
Although tonic and autoantigen-induced BCR signals, respectively, require the PI(3)K-Akt kinases and the SLP-65 adaptor protein6, 7, the exact molecular pathways activated by either type of BCR have not been defined. In the article published here, Herzog et al.4 seek to further elucidate the signal-transduction cascades that regulate immunoglobulin -chain rearrangement in developing B cells. They find that signaling through PI(3)K and Akt negatively regulates immunoglobulin -chain rearrangement in SLP-65-deficient pre–B cell lines. The main targets of the Akt kinases are the Foxo proteins (Foxo1, Foxo3a, Foxo4 and Foxo6), which are members of the Fox family of transcription factors classified by a 'winged-helix' forkhead box DNA-binding domain8. Akt-mediated phosphorylation of Foxo proteins inhibits their transcriptional activity by promoting their cytoplasmic localization and degradation8.
The authors find that Akt activation promotes phosphorylation of Foxo3a, whereas a Foxo3a mutant lacking Akt phosphorylation sites increases steady-state Rag1 and Rag2 mRNA and promotes Igk rearrangements, even in the presence of constitutively active Akt. Specific deletion of Foxo3a in primary B lineage cells results in impaired development of mature B cells, a finding leading to the further observation that BCR-induced Akt activation can also promote Foxo1 phosphorylation. In addition, Herzog et al. find that SLP-65 'antagonizes' Akt-mediated phosphorylation of Foxo3a and Foxo1 in pre–B cell lines and after BCR crosslinking in BCR+ immature B cell lines, which suggests that SLP-65–mediated signals may be important for the initiation of receptor editing. Consistent with that interpretation, the authors show that either pharmacological inhibition of PI(3)K or expression of the Foxo3a mutant lacking Akt phosphorylation sites in a BCR+ SLP-65-deficient B cell line induces rearrangement of a transfected V(D)J recombination reporter as well as an increase in immunoglobulin -chain–negative and immunoglobulin -chain–high cells.
Properly regulated expression of Rag1 and Rag2 is required for normal lymphocyte development and the assembly of V-region exons encoding a vast repertoire of antigen specificities9, 10; however, little is known about the signaling pathways and transcription factors that control Rag1-Rag2 transcription. To gain further insight into such mechanisms, Amin and Schlissel5 initiated a functional screen for factors that might induce Rag1-Rag2 transcription during BCR editing. For this purpose, they generated Abelson mouse leukemia virus (AMuLV)-transformed pre–B cells from a 'knock-in' mouse with specific replacement of Rag1 with cDNA encoding green fluorescent protein (GFP) on one allele (Rag1-GFP mice) and infected them with a retroviral library of cDNA molecules generated from CD19+ bone marrow cells. With this approach, they isolated a cDNA molecule that induces Rag1-GFP expression and encodes the stress-response gene Gadd45a. Gene microarray analysis of wild-type and GADD45a-overexpressing AMuLV-transformed pre–B cells indicated that GADD45a induces Rag1-Rag2 transcription by activating Foxo transcription factors. To test that idea, the authors used a combination of short hairpin RNA (shRNA) and overexpression experiments to show that GADD45a activates Rag1-GFP transcription via Foxo1 but not via Foxo3a or Foxo4 in AMuLV-transformed pre-B cells. Retroviral expression of Foxo1 in primary B cells of Rag1-GFP mice leads to more Rag1-GFP expression, but expression of Foxo3a does not; in contrast, retroviral expression of constitutively activated Akt or Foxo1-specific shRNA results in less Rag1-GFP expression.
On the basis of those findings, Amin and Schlissel5 reason that tonic BCR signaling in immature B cells might suppress Rag1-Rag2 expression through Akt-mediated phosphorylation and inactivation of Foxo1, whereas binding of antigen to an autoreactive BCR would lead to antigen-BCR internalization and cessation of this regulation. Consistent with that idea, crosslinking of BCR on immature B cells expressing Foxo1-specific shRNA, but not those expressing control shRNA, interferes with the induction of Rag1 and Rag2 mRNA. Foxo1-mediated induction of Rag1 transcription does not require protein synthesis but does require the DNA-binding activity of Foxo1, which suggests that Foxo1 drives Rag1-Rag2 transcription by directly binding to sequences in the Rag1-Rag2 locus.
The observations presented by Herzog et al.4 and Amin and Schlissel5 indicate that in developing B lymphocytes, Foxo proteins function 'downstream' of the BCR as a 'molecular rheostat' that regulates Rag1-Rag2 transcription and, as a consequence, BCR editing. Collectively, their data provide evidence that 'innocuous' BCRs expressed on immature B cells signal through PI(3)K-Akt to phosphorylate and inactivate Foxo transcription factors, rendering them unable to drive Rag1-Rag2 transcription, whereas antigen engagement of autoreactive BCRs expressed on immature B cells initiates SLP-65-mediated signals that quench BCR-mediated tonic activation of PI(3)K-Akt and Foxo phosphorylation, which promotes nuclear accumulation of Foxo proteins to directly activate Rag1-Rag2 transcription (Fig. 1).
Figure 1: BCR-mediated regulation of Igk rearrangement.
(a) In immature B cells expressing an 'innocuous' BCR, tonic BCR-induced signals activate the PI(3)K-Akt kinases, which leads to the phosphorylation and degradation of Foxo transcription factors. This precludes the accumulation of Foxo proteins in the nucleus, which prevents transcriptional upregulation of Rag1-Rag2 and secondary V
-to-J rearrangements. (b) In immature B cells expressing an autoreactive BCR, antigen-BCR interactions lead to receptor internalization and activation of SLP-65, which suppresses the activation of PI(3)K and Akt by unbound autoreactive BCRs. This inhibits Akt-mediated phosphorylation of Foxo proteins, which allows their accumulation in the nucleus and leads to the transcriptional upregulation of Rag1-Rag2 and secondary V-to-J rearrangements. The possible involvement of Foxo factors in directly regulating Igk rearrangement has not been explored.
Kim Caesar
Full size image (73 KB)However, the data presented in these two manuscripts also raise important questions that will probably be the focus of future experiments. It is unclear why Herzog et al.4 find BCR-mediated phosphorylation of both Foxo3a and Foxo1, whereas Amin and Schlissel5 find that only Foxo1 functions in this way. Similarly, although NF-B transcription factors have been linked to BCR-mediated regulation of Rag1-Rag2 transcription11, Amin and Schlissel5 find that expression of a dominant negative regulator of NF-B activity has little influence on Rag1-GFP expression. As suggested by Amin and Schlissel5, different methods were used for these experiments, and further work is needed to clarify the functions of these transcription factors in modulating BCR-mediated signals.
Despite the important advances made by these two studies, more work is needed to fully understand the contribution of Foxo factors in the regulation of antigen receptor assembly and diversification. Further insights into the precise molecular mechanisms by which Foxo proteins regulate Igk rearrangement are required. Although both manuscripts suggest that Foxo proteins directly activate Rag1-Rag2 transcription, the relevant DNA sequences in the locus have not been identified. In this context, although the transcription factor Foxp1 activates Rag1-Rag2 transcription through a B cell–specific enhancer located in the Rag1-Rag2 locus10, 12, Amin and Schlissel5 find that Foxo1-regulated Rag1 expression is similar in wild-type B cells and B cells lacking this enhancer. As the binding of transcription factors to promoters and enhancers in antigen-receptor loci can promote Rag1-Rag2 accessibility and V(D)J recombination1, it will be important to determine whether Foxo factors also regulate BCR editing by directly binding to and activating the Igk locus (Fig. 1). The possible functions of Foxo transcription factors in regulating Rag1-Rag2 transcription during T cell development and the rearrangement of genes encoding T cell antigen receptors needs to be investigated in detail. Finally, future studies might also consider whether Foxo proteins similarly serve as a 'molecular switch' in mature B cells to integrate tonic and antigen-induced BCR signals to control transcription of the gene encoding activation-induced cytidine deaminase and to regulate somatic hypermutation and/or class-switch recombination.
