The B cell receptor complex (BcR) is essential for normal B lymphocyte function, and surface BcR expression is a crucial checkpoint in B cell development. However, functional requirements for chains of the BcR during development remain controversial. We have used retroviral gene transfer to introduce components of the BcR into chicken B cell precursors during embryonic development. A chimeric heterodimer, in which the cytoplasmic domains of chicken Igα and Igβ are expressed by fusion with the extracellular and transmembrane domains of murine CD8α and CD8β, respectively, targeted the cytoplasmic domains of the BcR to the cell surface in the absence of extracellular BcR domains. Expression of this chimeric heterodimer supported all early stages of embryo B cell development: bursal colonization, clonal expansion, and induction of repertoire diversification by gene conversion. Expression of the cytoplasmic domain of Igα, in the absence of the cytoplasmic domain of Igβ, was not only necessary, but sufficient to support B cell development as efficiently as the endogenous BcR. In contrast, expression of the cytoplasmic domain of Igβ in the absence of the cytoplasmic domain of Igα failed to support B cell development. The ability of the cytoplasmic domain of Igα to support early B cell development required a functional Igα immunoreceptor tyrosine-based activation motif. These results support a model in which expression of surface IgM following productive V(D)J recombination in developing B cell precursors serves to chaperone the cytoplasmic domain of Igα to the B cell surface, thereby initiating subsequent stages of development.
The Ag receptor complex on most B lymphocytes (BcR)3 includes two Ig μ H chains, each covalently bound to an Ig L chain (IgL). By contrast, the pre-BcR, expressed by B lineage cells immediately before IgL rearrangement, includes two Ig μ-chains each bound to a surrogate L chain consisting of the VpreB and λ5 polypeptides (1). Whereas the ligand binding domains of the BcR, the VH and VL domains of μ and L, respectively, define the specificity of the B cell, it is currently unclear whether the μ-chain VH domain confers a functional specificity to the pre-BcR complex when paired with VpreB. Nonetheless, the μ-chains of both the BcR and the pre-BcR are both noncovalently associated with the Igα/β heterodimer that contains cytoplasmic signaling motifs (2, 3, 4). Critically, the association between μ and Igα/β is required for signal transduction in response to BcR or pre-BcR ligation (5).
In the absence of any of the components of the BcR or pre-BcR, the residual complex is retained in the endoplasmic reticulum (6). Moreover, surface BcR expression requires association between the Igα/β and μ extracellular and transmembrane domains (7). Thus, mutations in μ that disrupt the association of μ with Igα/β result in a failure to form a functional surface BcR (5).
In mice lacking μ, VpreB, λ5, Igα, or Igβ, B cell development is blocked at the pre-B cell stage (8, 9, 10, 11, 12). As with the requirement for surface expression of the BcR, association between μ and Igα/β is also required for the normal B cell development (13). Therefore, surface expression of intact BcR and pre-BcR complexes regulates critical checkpoints in mammalian B cell development.
Although surface BcR expression has been conserved during evolution as a checkpoint in B cell development (14), pathways of B cell lymphopoiesis have diverged. Although B cell development in rodents and primates occurs primarily in the bone marrow, other species use a variety of gut-associated lymphoid tissues as their primary B lymphoid organ (15). In particular, avian B cell development occurs in the bursa of Fabricius: a discreet gut-associated organ in which B cell development is segregated from the development of other hemopoietic lineages.
The bursal microenvironment is not required for B lineage commitment or Ig gene rearrangement. The bursal mesenchyme is colonized by a single wave of B cell precursors (16), some of which migrate across the bursal epithelial basement membrane and subsequently undergo clonal expansion within epithelial buds (17, 18). Only those B cell precursors that have undergone productive Ig gene rearrangement and can therefore express an intact BcR complex are selected for expansion in the bursa (19). Because rearrangement of chicken Ig (chIg) VH and VL genes occurs stochastically (20, 21), there is no apparent role for the expression of a pre-BcR in chicken B cell development, and chicken homologues to VpreB and λ5 have not been identified. Nonetheless, rearrangement of chIg genes generates a BcR of very limited diversity, which could be considered structurally analogous to the murine pre-BcR. Thus, in chicken, as in mammals, surface expression of BcR-related receptor complexes provides a critical checkpoint in B cell development.
Igα- and/or Igβ-mediated signal transduction is required to support B cell development (13, 22). Although the signaling potential of the Igα/β heterodimer remains to be fully characterized, the cytoplasmic domains of both Igα and Igβ contain immunoreceptor tyrosine-based activation motifs (ITAMs) as well as other residues implicated in signal transduction (2, 3, 4). Nonetheless, given the divergence of signaling pathways downstream of Igα and Igβ, it remains unclear as to which of these pathways are required for the normal progression of B cell development.
Surface expression of BcR complexes lacking V(D)J-encoded determinants (truncated Ig μ-chain (Tμ)) supports the early stages of murine and chicken B cell development (23, 24, 25). Under these circumstances, receptor ligation could still occur between a ligand expressed in the environment and the residual extracellular domains of the truncated receptor complexes, including residual domain(s) of the Tμ molecule, or the extracellular domains of the associated Igα/Igβ complex, resulting in a signal required for B cell development. Alternatively, the μ-chains of the receptor may act as a chaperone to target the signaling domains of Igα/β to the cell surface, supporting the induction of basal levels of signaling in the absence of receptor ligation. Therefore, while BcR and/or pre-BcR expression is critical for B cell development, it remains unclear whether receptor ligation is also required or whether receptor expression per se is sufficient.
We have therefore used productive retroviral gene transfer in vivo to introduce modified BcRs into developing chicken B cell precursors under circumstances whereby precursors expressing such receptors are in direct competition with B cells expressing endogenous BcR. Consequently, this allows us not only to assess whether modified receptors support B cell development, but more importantly, directly determine the efficiency with which they support development relative to endogenous surface Ig (sIg).
The extracellular and transmembrane domains of murine CD8 (mCD8) were used to target the cytoplasmic domains of chIgα and chIgβ to the cell surface. Targeting the cytoplasmic domain of Igα to the membrane is not only necessary, but is sufficient to support the early stages of B cell development in the absence of endogenous sIgM expression, supporting the contention that BcR ligation is not required for B cell development. Moreover, membrane-proximal expression of the cytoplasmic domain of Igα, but not the cytoplasmic domain of Igβ, supports B cell development as efficiently as the intact BcR complex.
Materials and Methods
Generation of replication competent avian leukosis virus with splice acceptor (RCAS)-mCD8:chIg chimeric proteins
The extracellular and transmembrane domains of mCD8α and mCD8β were cloned by PCR amplification using the primer combinations CD8α5′-GGAGCGGCCGAGCCACAGGCACCCGAA with CD8α3′-GTGGTAGACAATTGCAGTGATGATCAAGGACAG, and CD8β5′-AGCTCGGCCGTCATTCAGACCCCTTCG with CD8β3′-AAAGTAGACAATTGCTCCGAGGAATGCCAGCAG, respectively, from either a mCD8α-containing plasmid provided by R.-P. Sekaly (University of Montreal, Montreal, Quebec, Canada) or, in the case of CD8β, from thymic cDNA. The cytoplasmic domains of chIgα and chIgβ were PCR amplified using the primer combinations Igαcyt5′-GGAATGCTGGCAATTGTCAGGAAGCGCTGG with Igαcyt3′-CCAAGCTTCTCAGGGTTTCTCCAGGCC and Igβcyt5′-ATGCTCGCAATTGTAGAAAAGGGTGACAGA with Igβcyt3′-GTGAAGCTTCCCTCACTCCTCTCCTGG, respectively, from bursal cDNA. In the above primers, BstXI sites are indicated in bold, MfeI sites are underlined, and HindIII sites are indicated in italics. Sequences encoding mCD8α or mCD8β were fused to sequences encoding chIgα or chIgβ by ligation following MfeI digestion. Gel-purified products were subsequently digested with BstXI and HindIII and cloned into the BstXI and HindIII sites of the CLA-12-leader plasmid, containing the chIgVH leader sequence (23, 26). The mCD8α:chIgαF2 mutant was constructed by PCR amplifying the mCD8α:chIgα sequence with the primer combination CD8α5′ with IgαF23′-CGAGATGTCTTCGAACATGGAGCA and IgαF25′-TGCTCCATGTTCGAAGACATC with Igα3′ (BstBI sites underlined). Amplified products were digested with BstBI and ligated. Gel-purified products were cloned into the CLA-12-leader plasmid, as above. Chimeric constructs were excised from their respective CLA-12 plasmids with ClaI and cloned into the unique ClaI site of either the RCAS (bryan polymerase (BP))A- or RCAS(BP)B-productive retroviral vectors (26).
RCAS-mCD8:chIg virus production and in vivo gene transfer
The RCAS-based plasmids were used to transfect line 0 chick embryo fibroblasts (CEFs; Regional Poultry Research Laboratories, East Lansing, MI) by calcium phosphate precipitation (22). Within 7 days of transfection, essentially all CEFs expressed the retrovirally transduced mCD8-containing chimeric protein(s). DT40 bursal lymphoma cells were infected with the RCAS-mCD8:chIg viruses by coculture with transfected CEFs and subsequently subcloned by FACS single cell deposition (27). Day 3-incubated SC line (Hy-Vac International, Adel, IA) chick embryos were inoculated with 1 × 106 CEFs transfected with RCAS-based plasmids.
Abs and flow cytometry
Surface expression of CD8 chimeric proteins was detected using anti-mCD8α (53-6.72) and/or anti-mCD8β (53.8.84) Abs (provided by P. Hugo, MetrioGene Biosciences, Montreal, Quebec, Canada), followed by FITC- or PE-conjugated anti-mouse Ig isotype Abs (Southern Biotechnology Associates, Birmingham, AL). ChB6, chIgμ, and L determinants were detected, as before (23). Samples were analyzed on a FACSCalibur or sorted on a FACSVantage (BD Biosciences, Mississauga, Ontario, Canada).
Changes in cytosolic calcium concentrations were detected in INDO-1 (Molecular Probes, Eugene, OR)-loaded cells, as before (28), except that samples were assayed on a FACSVantage. Calcium mobilization was assessed following exposure of cells to anti-CD8α (53-6.72), or anti-IgL (6E1).
Protein tyrosine phosphorylation
DT40 clones were stimulated for 2 min with Abs or 50 μM of pervanadate (29) at 37°C before centrifugation, lysis by boiling in SDS-PAGE loading buffer, and electrophoresis. Electrophoresed samples were blotted onto Nytran membranes (Amersham, Arlington Heights, IL), and blots were probed with anti-actin Abs (Sigma-Aldrich, St. Louis, MO; AC-40), followed by goat anti-mouse Ig conjugated to HRP (Bio-Rad, Hercules, CA). After detection of HRP by chemiluminescence (Amersham), blots were stripped and reprobed with PY72 anti-phosphotyrosine Ab (30), followed by HRP-conjugated anti-mouse Ig and detection by chemiluminescence.
Cell cycle analysis
PCR amplification and cloning of V gene sequences
Neonatal bursal cells from chicks inoculated with RCAS-based viruses were stained, and aliquots of 1500 cells of defined phenotype were sorted directly into PCR tubes. Cells were incubated in 15 μl of 10 μg/ml proteinase K for 1 h at 50°C, followed by heat inactivation at 85°C for 20 min. VJL, VDJH, and DJH rearrangements were amplified using the primer combinations described elsewhere (23). All amplification reactions also contained the RAG5′ and RAG3′ combination of primers to amplify the single-copy RAG2 sequence from genomic DNA as an internal standard. PCR products were electrophoresed, and bands were quantitated by scanning densitometry. Complementarity-determining region 3 length of VJL sequences (23) and gene conversion (32, 33) were assessed, as before. VJL sequences were cloned into PCR2.1 (Invitrogen, San Diego, CA) and sequenced (York University, North York, Ontario, Canada).
Samples of bursal tissue were frozen, sectioned, and stained, as before (34), using Abs to ChB6, chicken μ, and mouse CD8α, followed by biotinylated goat anti-mouse Ig and detection using the ABC kit (Vector Laboratories, Burlingame, CA) revealed with Vector VIP substrate kit, according to manufacturer’s instructions.
Targeting the cytoplasmic domains of Igα and Igβ to the B cell surface
The extracellular and transmembrane domains of murine CD8α and/or CD8β were used to target the cytoplasmic domains of chIgα and chIgβ to the cell surface independent of endogenous sIgM expression and in the absence of any extracellular sIg complex domains. Sequences encoding the chimeric proteins (Fig. 1 A) were cloned into the RCAS(BP)A- and RCAS(BP)B-productive avian retroviral vectors that support efficient gene transfer in vitro and in vivo (23, 26).
Surface expression of the mCD8α:chIgα chimeric protein was observed in transfected CEFs. In contrast, the mCD8β:chIgβ chimeric protein was not expressed at the surface of singly transfected CEFs. However, cotransfection of mCD8β:chIgβ together with mCD8α:Igα resulted in surface expression of both proteins (data not shown), consistent with the requirement for mCD8α expression in the surface expression of mCD8β (35).
Igα and Igβ chimeric receptors support signal transduction in DT40 cells
The chicken bursal lymphoma DT40 supports sIg-mediated signal transduction and was therefore used to analyze the signaling capacity of mCD8:chIg chimeric receptors (27, 36, 37). DT40 cells express endogenous sIgM, as detected with Abs against μ and L. In contrast, an L-negative variant DT40 cell line (DT40L−) does not express sIgM. Infection of the DT40L− cell line with the RCAS(BP)-mCD8:chIg viruses yielded expression patterns of the CD8 chimeric proteins equivalent to those seen in CEFs (Fig. 1 B).
Exposure of DT40L− cells expressing both mCD8α:chIgα and mCD8β:chIgβ to either anti-mCD8α (Fig. 1,C) or anti-mCD8β Abs (data not shown) resulted in a rapid induction of calcium mobilization. Similarly, exposure of DT40L− cells expressing mCD8α:chIgα alone to anti-mCD8α Abs resulted in a similarly rapid induction of calcium mobilization (Fig. 1,C). In contrast, exposure of DT40L− cells expressing mCD8α:chIgβ to anti-CD8α did not induce a significant elevation of intracellular calcium ion concentrations (Fig. 1 C).
BcR cross-linking on DT40 cells resulted in a rise in intracellular protein tyrosine phosphorylation not observed in the DT40L− variant cell line (Fig. 1 D). DT40L− cells coexpressing mCD8α:chIgα and mCD8β:chIgβ exposed to anti-CD8 Abs showed increased protein tyrosine phosphorylation close to levels observed in sIg+ DT40 cells stimulated with anti-IgL Abs. Similarly, the mCD8α:chIgα chimeric receptor alone supported the induction of protein tyrosine phosphorylation following cross-linking on the surface of DT40L− cells. Although quantitative and qualitative differences were observed, increased protein tyrosine phosphorylation was also seen in cells expressing CD8α:chIgβ.
Equivalent results were observed following pervanadate stimulation. Thus, mCD8α:chIgα, either alone or in combination with mCD8β:chIgβ, reconstituted protein tyrosine phosphorylation in DT40L− cells following pervanadate stimulation (Fig. 1 D). As seen following receptor cross-linking, DT40L− cells expressing the mCD8α:chIgβ chimeric receptor exhibited quantitative and qualitative different patterns of protein tyrosine phosphorylation following pervanadate stimulation. Thus, each of the chimeric receptor complexes supported the induction of basal signal transduction, as revealed by pervanadate exposure.
The cytoplasmic domains of chIgα and chIgβ together support bursal colonization in the absence of endogenous μ
The mCD8α:chIgα and/or mCD8β:chIgβ chimeric proteins were introduced into developing hemopoietic precursors by inoculating day 3 chick embryos with appropriately transfected CEFs. Under these circumstances, retroviral penetrance is high and the inoculated CEFs themselves do not become detectably chimeric within the recipient embryo (23). At hatch, bursal cells were analyzed by flow cytometry for the expression of endogenous sIg and chimeric receptors.
All bursal B cells express high levels of the pan-B cell surface Ag ChB6 (ChB6+) (Fig. 2,A). All ChB6+ bursal cells from normal chicks or from chicks infected with control RCAS vectors (Fig. 2,F) express endogenous sIgM, consistent with earlier conclusions that sIgM expression is required for productive bursal colonization (19). In chicks infected with control RCAS vectors, levels of follicular colonization, proportion of bursal cells expressing sIg, size of follicles, development of follicular structure, and the emigration of B cells to the periphery are all indistinguishable from normal uninfected chicks. Conversely, in chicks infected with the combination of RCAS(BP)A-mCD8α:chIgα and RCAS(BP)B-mCD8β:chIgβ viruses, a significant proportion of bursal B cells lacked the expression of endogenous sIgM (Fig. 2 B). These cells were all mCD8α+, suggesting that surface expression of the mCD8:chIg chimeric protein(s) is sufficient to support the early stages of B cell development in the absence of endogenous BcR expression. The proportion of mCD8α+/IgM− bursal B cells was 10–35% from 22 chicks assessed.
The cytoplasmic domain of Igα is sufficient to support bursal colonization
Strikingly, not all sIgM− bursal B cells from chicks infected with the combination of RCAS(BP)A-mCD8α:chIgα and RCAS(BP)B-mCD8β:chIgβ viruses coexpressed mCD8β:chIgβ. Typically, only ∼50% of mCD8α+/sIgM− bursal B cells from these chicks coexpressed mCD8β:chIgβ (Fig. 2,C). This suggested the possibility that expression of mCD8α:chIgα alone was sufficient to support the progression of B cell development. This was confirmed in neonatal chicks singly infected with the RCAS(BP)A-mCD8α:chIgα virus. Such chicks contained a substantial population of mCD8α+/sIgM− bursal B cells (Fig. 2 D). Indeed, 5–85% of bursal B cells in 28 RCAS(BP)A-mCD8α:chIgα-infected chicks assayed expressed mCD8α:chIgα in the absence of endogenous sIgM expression.
Although both sIgM+ and sIgM− bursal cells are found within the embryo bursal mesenchyme, the translocation of sIgM+ cells across the basement membrane and their subsequent oligoclonal expansion in follicles are required for productive bursal colonization (16, 17, 18, 27). The morphology of bursal sections from chicks infected with RCAS(BP)A-mCD8α:chIgα was normal (Fig. 3), and was indistinguishable from the bursal morphology of control chicks infected with RCAS vector alone. The presence of follicles containing either a mixture of mCD8α+/IgM− cells and mCD8α−/IgM+ cells or exclusively mCD8α+/IgM− cells is consistent with the normal oligoclonal colonization of bursal follicles (17, 18) and demonstrates that mCD8α+/IgM− cells translocate across the bursal basement membrane. Furthermore, cell cycle analysis demonstrated that both mCD8α+/IgM− (Fig. 3,G) and mCD8α−/IgM+ (Fig. 3 H) bursal B cells were proliferating at the high rates characteristic of neonatal bursal cells.
The high levels of staining with anti-CD8α Abs outside the follicles of bursae from RCAS(BP)A-mCD8α:chIgα-infected chicks (Fig. 3 C) reflect the penetrance of the RCAS virus in the nonlymphoid cells that make up the interfollicular connective tissue. Broad tropism of the RCAS virus has been documented elsewhere (26).
The cytoplasmic domain of Igβ is not sufficient to support bursal colonization
Fusion of the cytoplasmic domain of Igβ with the extracellular and transmembrane domains of mCD8α generated a mCD8α:chIgβ construct that was expressed at the surface of CEFs (data not shown) and DT40L− cells (Fig. 1,B) in the absence of the cytoplasmic domain of Igα. However, while mCD8α:chIgα supported bursal B cell development, replacement of the Igα cytoplasmic domain with the Igβ cytoplasmic domain was not effective (Fig. 2 E). Thus, bursal cells from neonatal chicks infected as day 3 embryos with RCAS(BP)A-mCD8α:chIgβ contained no mCD8α+/sIgM− bursal B cells.
Reduced levels of V gene rearrangement in mCD8α+/sIgM− bursal cells
To formally demonstrate that expression of the cytoplasmic domain of Igα is sufficient to support B cell development in the absence of endogenous Ig expression, we assessed the state of the Ig genes in mCD8α+/sIgM− bursal cells. Because DJH rearrangement only occurs in B lineage cells in the chicken (21), the high levels of DJH rearrangement confirmed that mCD8α+/IgM− bursal cells are indeed B lineage.
Reduced levels of V gene rearrangement in mCD8α+/IgM− bursal B cells from RCAS(BP)A-mCD8α:chIgα-infected chicks were observed both at the IgH and IgL loci. Specifically, V gene rearrangement was reduced to ∼20% of normal levels at the H chain locus and to 25% of normal levels at the IgL locus (Fig. 4 A).
Moreover, in contrast to IgM+ cells, there was no selection for productive VJL rearrangements in mCD8α+/IgM− bursal B cells (Fig. 4,B). The heterogeneous distribution of complementarity-determining region 3 lengths is characteristic of B lineage cells before selection based on sIgM expression (19, 23) and was confirmed by analysis of VJL sequences from mCD8α+/IgM− bursal cells (Fig. 5 B). Thus, while essentially all VJL junctions isolated from control IgM+ bursal B cells were in frame, consistent with results obtained elsewhere (19, 20), only 48% of VJL junctions isolated from mCD8α+/IgM− bursal B cells were in frame. Taken together with the low frequency of VH and VL rearrangement, these results obviate the possibility that mCD8α+/IgM− cells have been selected on the basis of endogenous sIgM expression.
mCD8α:chIgα supports Ig gene diversification
Colonization of bursal follicles and subsequent B cell expansion coincide with the induction of Ig V gene diversification by gene conversion (19). V(D)J rearrangement in the chicken generates limited Ab diversity. At the IgL locus, there is only one functional VL segment, which rearranges to the unique JL segment. Similarly, IgH rearrangement uses one functional VH segment, a cluster of highly conserved DH segments and a unique JH segment. Diversity is generated by gene conversion events in which sequences derived from upstream pseudo-VDH and pseudo-VL genes replace homologous sequences in the rearranged VDJH and VJL genes, respectively (32, 38, 39).
Before the induction of gene conversion, cleavage sites for the endonucleases KpnI and SmaI are present in the functional VL gene. Typically, following gene conversion, these cleavage sites are modified, and rearranged gene-converted VJL segments become resistant to KpnI and/or SmaI digestion (32, 33). IgM+ and mCD8α+/IgM− bursal B cells from RCAS(BP)A-mCD8α:chIgα-infected chicks demonstrated comparable levels of resistance to both KpnI and SmaI digestion (Fig. 5,A). Moreover, sequence analysis of VJL segments isolated from mCD8α+/IgM− bursal B cells had undergone gene conversion at the same frequency as control IgM+ bursal B cells (Fig. 5 B). Importantly, levels of gene conversion were independent of whether the V gene rearrangement had resulted in an in-frame junction. Thus, expression of the mCD8α:chIgα chimeric protein was sufficient to support the induction of gene conversion among the minority of such cells that had undergone endogenous VJL rearrangement.
A functional ITAM is required for Igα-mediated early B cell development
Signaling from the BcR complex has been attributed to ITAM contained within the cytoplasmic domains of the Igα/Igβ heterodimer (2, 3, 4). Consequently, the C-terminal tyrosine residue of the Igα ITAM was mutated to phenylalanine. The resulting mCD8α:chIgαF2 mutant was expressed in CEFs (data not shown) and DT40 cells (Fig. 1,B) at levels equivalent to those observed for the mCD8α:chIgα construct. The mCD8α:chIgαF2 mutant failed to support calcium mobilization following cross-linking with anti-CD8α Abs (Fig. 1,C). However, both pervanadate treatment and receptor cross-linking resulted in increased protein tyrosine phosphorylation (Fig. 1 D), although at levels substantially lower than those seen following cross-linking of the wild-type mCD8α:chIgα receptor.
Neonatal chicks infected as day 3 embryos with the RCAS(BP)A-mCD8α:chIgαF2 mutant virus contained bursal B cells expressing mCD8α:chIgαF2. Critically, however, all such cells coexpressed endogenous sIgM (Fig. 2 G). Consequently, disruption of the ITAM in the cytoplasmic domain of Igα rendered the cytoplasmic domain of Igα incapable of supporting the early stages of bursal B cell development.
Moreover, in chicks infected with the combination of RCAS(BP)A-mCD8α:chIgαF2 plus RCAS(BP)B-mCD8β:chIgβ, all bursal B cells expressing mCD8 also expressed endogenous sIgM (Fig. 2, H and I). Consequently, expression of the cytoplasmic domain of Igβ failed to support the progression of bursal B cell development in the absence of a functional Igα ITAM.
We have used retroviral gene transfer in the developing chick embryo to assess the role of surface IgM in supporting early B cell development. Although a Tμ supported early B cell development as efficiently as endogenous sIgM (23), this construct did not allow us to determine whether such support was a consequence of receptor expression or receptor ligation. We therefore expressed the cytoplasmic domains of Igα and/or Igβ in the absence of any extracellular domains associated with the Ig receptor complex in developing B cell precursors. Because mCD8α is expressed on the cell surface either as a homodimer, or as a heterodimer in association with mCD8β (35, 40, 41), it was possible to target the cytoplasmic domains of chIgα or chIgβ to the membrane either singly or together.
Expression of either mCD8α:chIgα alone or in conjunction with mCD8β:chIgβ resulted in the generation of bursal ChB6+ cells that lacked the expression of endogenous sIgM. In the avian system, DJH rearrangement is restricted to B lineage cells and therefore provides a definitive marker of B cell lineage commitment (21). The levels of DJH rearrangement in bursal B cells expressing the mCD8α:chIgα chimeric protein, in the absence of endogenous sIgM, are equivalent to the levels of DJH rearrangement in sIgM+ bursal cells. Thus, mCD8α+/IgM− bursal cells from RCAS(BP)A-mCD8α:chIgα-infected chicks are indeed committed to the B cell lineage, and expression of mCD8α:chIgα, either alone or with mCD8β:chIgβ, therefore supported all checkpoints in early B cell development. In this regard, expression of the CD8 chimeric receptor is functionally indistinguishable from expression of the Tμ receptor (23) or indeed the intact sIgM receptor complex during early B cell development.
The interaction between CD8 and MHC class I Ags is highly species specific (42). Comparison of the amino acids responsible for association of mouse MHC class I and mouse CD8 with the equivalent sequences in chicken revealed several modifications in the positioning of charged residues. Positional modifications of charged residues in chicken CD8 correlated with changes in the positions of charged residues in chicken MHC class I (42, 43). Thus, mouse CD8 would not be expected to associate with chicken MHC class I. Support for this notion comes from the observation that expression of either mCD8α:chIgα alone, or in conjunction with mCD8β:chIgβ, in DT40 cells did not affect basal levels of signaling in the absence of receptor ligation despite expression of chicken MHC class I on the DT40 B lymphoma (44). Therefore, mouse CD8 is an appropriate means of targeting the cytoplasmic domains of Igα and Igβ to the cell surface.
The requirement for surface IgM expression in normal avian B cell development is analogous to the requirement for pre-BcR and BcR expression in murine B cell development. It remains debated as to whether murine B lineage progression requires solely the expression of the pre-BcR and BcR or whether these receptors require ligation by an extracellular ligand. Observations that a soluble murine surrogate L chain (VpreB/λ5) in association with a C-terminal truncated μ-chain binds to bone marrow-derived stromal cells (45) together with the identification of galectin 1 and stroma cell-associated heparan sulfate as ligands for the human (46) and murine (47) surrogate λ-like L chains, respectively, have been taken as support for the presence of an extracellular ligand for the μ/VpreB/λ5 complex.
In contrast, we have previously demonstrated that a Tμ complex, lacking V(D)J-encoded determinants and IgL, supports early avian B cell development (23). Similar Tμ complexes supported early murine B cell development (24, 25, 48). This is not compatible with an obligate requirement for BcR or pre-BcR ligation by an extracellular ligand(s) specifically recognizing IgL, VpreB, or λ5. The remaining possibility, that early B cell development requires ligation of residual extracellular domains of the Tμ/Igα/β complex, or may be a consequence of self-aggregation of truncated μ-chains (48), has been ruled out by the experiments reported in this work.
In an alternative approach, targeting a fusion protein containing the cytoplasmic domains of Igα and Igβ to the cell surface by palmitylation supported the development of B lineage precursors into immature B cells in mice (49). In this model, however, cells expressing the fusion protein did not compete with cells expressing endogenous sIg. Thus, it remained unclear whether BcR or pre-BcR expression in the absence of ligation is necessary and sufficient to provide efficient support for the early stages of B cell development. In this work, we show that surface expression of the cytoplasmic domains of the Igα/β heterodimer, in the absence of any extracellular domains associated with the sIg receptor complex, supports early B cell development as efficiently as the endogenous sIg receptor complex.
Our results extend observations in murine models suggesting that expression of Igα and Igβ is each independently sufficient to support early B cell development (13, 50, 51, 52). In mice containing either Igα or Igβ cytoplasmic domain truncations, V-DJH rearrangement occurs and a pre-B cell population is generated (51, 52). However, in these mice, the extracellular domains of the sIg complex remain intact; as a consequence, it is not possible to assess and distinguish the role(s) of the cytoplasmic domains of Igα and Igβ from any contribution made by the extracellular domains of the BcR. This was also the case in which fusions of the μ-chain with the cytoplasmic domain of Igα (IgM-Igα) or Igβ (IgM-Igβ) supported early B cell development in Rag-1−/− mice (13, 50), suggesting that expression of the cytoplasmic domain of Igα or Igβ supported the generation of an immature B cell development.
Support of B cell development in mice containing Igα or Igβ cytoplasmic domain truncations or the IgM-Igβ fusion protien was observed under conditions in which B cell development induced by a normal BcR or pre-BcR complex was compromised. It is therefore not possible to determine from these experiments whether such support of B cell development was of comparable efficiency to the intact sIg receptor complex. Thus, cells harboring the Igα or Igβ truncation or the IgM-Igβ fusion protein may survive and differentiate due to a lack of competition from cells expressing an endogenous receptor complex.
In contrast, our system does not possess such inherent limitations because the penetrance of the RCAS(BP)A-mCD8α:chIgα virus is not 100% in vivo. Thus, in RCAS(BP)A-mCD8α:chIgα-infected chicks, B cell precursors expressing mCD8α:chIgα receptors compete with cells expressing the wild-type receptor complex for space within the homeostatic B cell population. Cells expressing mCD8α:chIgα receptors competed efficiently with cells expressing the endogenous BcR complex by all criteria that define early B cell development. In bursae of RCAS(BP)A-mCD8α:chIgα-infected chicks, follicles containing a mixture of both mCD8α−/IgM+ and mCD8α+/IgM− cells were observed. In addition, follicles containing exclusively mCD8α−/IgM+ B cells or exclusively mCD8α+/IgM− B cells were present, both of which were of comparable size and morphology. Moreover, the rate of cell division of both populations was equivalent, and similar levels of gene conversion were present in both B cell populations. Thus, we conclude that the cytoplasmic domain of Igα is sufficient to drive all the early stages of B cell development with the same efficiency as the entire receptor complex.
The distribution of mCD8α−/IgM+ and mCD8α+/IgM− cells in bursal follicles also supports the contention that similar numbers of B cell precursors productively colonize the bursa in chicks infected with the RCAS(BP)A-mCD8α:chIgα virus, as in normal chicks. Because bursal morphology, size, and number of bursal follicles are normal, any increase in the total number of B cell precursors in RCAS(BP)A-mCD8α:chIgα-infected chicks would be reflected in increased numbers of precursors colonizing each follicle. Under these circumstances, the frequency of follicles containing exclusively mCD8+/IgM− or mCD8−/IgM+ cells would be minimal. The identification of follicles containing exclusively mCD8+/IgM− or mCD8−/IgM+ cells therefore provides strong support for the oligoclonal colonization of bursal follicles in RCAS(BP)A-mCD8α:chIgα-infected chicks, as is seen in normal chicks.
The Epstein-Barr viral protein latent membrane protein 2A (LMP2A), when expressed as a transgene in mice, provided efficient support for sIgM− B cell development in mice that also contain B cells expressing an intact BcR (53). LMP2A-mediated support of B cell development required the presence of a functional ITAM, suggesting the possibility that LMP2A expression may mimic in some way expression of pre-BcR and BcR during B cell development (54). Nonetheless, the complexity of the LMP2A protein, which contains 12 membrane-spanning regions, and its self-association in the plasma membrane make direct comparisons with the BcR difficult.
In mammals, pre-BcR expression down-regulates further rearrangement at the H locus and sets the stage for L rearrangement. In the chicken, by contrast, a single wave of Ig gene rearrangement occurs during embryogenesis during which VH and VL genes undergo rearrangement simultaneously (20, 21). Allelic exclusion of chicken V gene rearrangement is not regulated by the products of rearrangement (55). Consequently, it is not surprising that expression of mCD8α:chIgα does not completely inhibit rearrangement of endogenous VH and VL genes. Nonetheless, V gene rearrangement in such cells is partially inhibited, to ∼20% of the levels of rearrangement in sIgM+ cells. This is equivalent to the levels of VDJH and VJL rearrangement in bursal cells that express Tμ in the absence of endogenous IgM (23).
We have previously argued that while surface expression of the BCR complex, including Igα/β, is not the determining factor mediating halotype exclusion in peripheral pre-bursal B cells, BcR-mediated signals may be required to initiate the migration of sIg+ B cell precursors into bursal follicles in which subsequent gene rearrangements are inhibited (55). The low levels of endogenous rearrangement seen in mCD8α:chIgα-expressing cells would therefore reflect rearrangements that had occurred before migration of mCD8α:chIgα-expressing cells into bursal follicles.
Our results support a model in which basal signaling through Igα in the absence of ligation supports B cell development. Thus, disruption of the ITAM obviated the ability of the cytoplasmic domain of Igα to support B cell development, although the mCD8α:chIgαF2 protein was expressed on bursal cells at high levels. The cytoplasmic domain of murine Igα has the potential to interact with Src family kinases such as Lyn, Fyn, and Blk; the Syk kinase; and the SLP-65 adaptor protein (37), and deletion of these signaling proteins in mammals compromises the normal development of B cells (54). chIgα shows greatest homology with mammalian Igα at these binding sites, suggesting that the same sites are functionally important across this range of species (27). Consequently, multiple signaling pathways downstream of the pre-BcR and BcR complex are probably critical in normal B cell development, although the precise role of each remains unclear (56).
Igα and Igβ associate with distinct signaling proteins (57), and residues within the ITAMs of Igα and Igβ (aspartic acid-cysteine-serine-methionine and glutamine-threonine-alanine-threonine, respectively) have been implicated in the differential signaling capacity of Igα and Igβ (3). A conserved tyrosine residue within the cytoplasmic domain of Igα, outside the ITAM, mediates the interaction of Igα, but not Igβ, with SLP-65 (58). SLP-65 is involved in the initiation of a signaling cascade that results in Ca2+ mobilization and NF-κB activation (59). However, while cooperativity between the cytoplasmic domains of Igα and Igβ may be required for later stages of development, we show in this study that expression of the cytoplasmic domain of Igα is not only necessary, but is sufficient for the early stages of B cell development.
We have shown that while expression of the cytoplasmic domain of Igα is sufficient to support early chicken B cell development, expression of the cytoplasmic domain of Igβ is not. In contrast, murine truncation mutants in which the cytoplasmic domains of Igα or Igβ have been truncated both support some level of pre-B and B cell development (51, 52). Although this might reflect systemic differences between mouse and chicken, in that the latter species does have a defined pre-B cell stage of development (14, 15), it nonetheless points to clear functional differences between the cytoplasmic domains of chIgα and chIgβ. This is similar to the activities of the cytoplasmic domains of murine Igα and Igβ, in which expression of the cytoplasmic domain of Igα supports B cell development to the immature B cell stage (51), while expression of the cytoplasmic domain of Igβ in the absence of Igα results in marked reductions in pre-B and immature B cell levels (52). Consequently, in both mouse and chicken, there appear to be functional differences between the cytoplasmic domains of Igα and Igβ.
Such regulation would be analogous to the regulation of T cell development. A truncated pre-TCR supported the progression of thymocyte differentiation. Indeed, while surface expression of the pre-TCR is sufficient to induce differentiation and proliferation of double-positive thymocytes (60), cross-linking of the pre-TCR blocks the differentiation of double-positive T cells (61). Similarly, exposure of bursal cells in the chicken embryo to Abs that cross-link sIg results in their elimination (62, 63).
To conclude, we have designed a system that enables us to identify the roles of the components of the BCR complex in regulating B cell development. We have presented the first direct evidence that the cytoplasmic domain of Igα is sufficient to drive the early stages of B cell development with the same efficiency as the entire receptor complex. Strikingly, under conditions in which B cell precursors have to compete with cells expressing endogenous sIg, the cytoplasmic domain of Igα efficiently supports the early stages of B cell development.
We thank Ken McDonald and Gisele Knowles for expert assistance with flow cytometry.
This work was supported by the Canadian Institutes for Health Research (MT10040) and by a studentship to K.A.P. from the Fonds de la recherche en santé du Québec.
Abbreviations used in this paper: BcR, B cell receptor complex; BP, bryan polymerase; CEF, chicken embryo fibroblast; chIg, chicken Ig; ITAM, immunoreceptor tyrosine-based activation motif; mCD8, murine CD8; RCAS, replication competent avian leukosis virus with splice acceptor; sIg, surface IgM; Tμ, truncated Ig μ-chain; LMP2A, latent membrane protein 2A.