The expression of different sets of immunoglobulin specificities by fetal and adult B lymphocytes is a long-standing puzzle in immunology. Recently it has become clear that production of immunoglobulin μ heavy chain and subsequent assembly with a surrogate light chain to form the pre-B cell receptor complex is critical for development of B cells. Here we show that instead of promoting pre–B cell progression as in adult bone marrow, this complex inhibits pre–B cell growth in fetal liver. Curiously, we identify a fetal-associated VH11 μ heavy chain that allows continued pre-B proliferation in fetal liver. Interestingly, this heavy chain does not associate efficiently with a surrogate light chain, providing a previously unrecognized mechanism for skewing the expression of distinctive VH genes toward fetal through early neonatal life.
Immune responses during the neonatal period show significant differences from those in the adult, with striking deficiencies in the ability to respond to certain antigens (1, 2), although the mechanism for this shift is not yet completely understood. Consistent with their restricted ability to respond to antigenic challenge, fetal and neonatal B cell precursors use a restricted set of heavy chain VH (Ig heavy chain variable region) genes, preferentially from segments proximal to the Ig diversity element (D), such as VH81X in mice (3, 4). The expansion from a restricted set of genes used by neonatal B cells to the wide variety employed in the adult (3–5) is referred to as repertoire maturation. Although the phenomenon has been appreciated for many years, the molecular and cellular mechanisms that result in this change are not yet understood. Since fetal B cells preferentially express VH segments proximal to D, ordered accessibility to recombination has been suggested as a possibility. Yet this cannot be the full explanation, since preferential rearrangement of D-proximal VH segments is also observed in B precursor cells in adult bone marrow (6). In addition, productive rearrangements of such genes predominate in fetal cells but become infrequent in adult precursors (7–9). Furthermore, some VH genes not proximal to D, such as VH11, also show expression biased to B cells generated from the fetal stage through the early neonatal stage (fetally/neonatally) (10, 11). These observations imply that mechanism(s) in addition to rearrangement accessibility, possibly the differential control of B cell development by particular VH genes, shape the distinctive repertoires of fetal and adult times.
Early B lineage development is critically dependent on expression of Ig μ heavy chain at the pre-B stage, where it associates with a surrogate light chain (SLC), composed of two molecules, λ5 and VpreB (12, 13), which collectively form a complex known as the pre–B cell receptor (BCR). The importance of the pre-BCR in this process is illustrated by the developmental arrest induced upon elimination of any of these components in gene-targeted mice (14–16). Successful pre-BCR assembly induces several hallmark events associated with progression from the pro-B to large pre-B stage in the bone marrow (17), including downregulated expression of genes involved in Ig rearrangement, such as terminal deoxynucleotidyl transferase (TdT) (18, 19) and the recombinase-activating genes (Rag), Rag-1 and Rag-2 (20, 21). These changes coincide with a sharp proliferative expansion in bone marrow, which interestingly occurs at precisely the stage where representation of the fetally biased VH81X shows a profound decrease (7). This alteration of VH representation at the stage of μ-dependent proliferation suggests to us that differences in the V region of the μ heavy chain itself may critically influence the growth of B cell precursors during fetal and adult development.
By establishing VH11 transgenic (Tg) mouse lines expressing two different levels of μ heavy chain, in comparison with other VH μ Tg lines, we demonstrate in this paper a distinctive in vitro proliferative response by fetal pre–B cells to pre-BCR assembly, with clear in vivo consequences on subsequent B cell development. We propose that this differential response by fetal pre–B cells to SLC-μ provides a previously unrecognized mechanism for skewing the expression of distinctive VH genes toward fetal/neonatal life.
Materials And Methods
A functionally rearranged VH11 gene was cloned from a CD5+ B cell–derived hybridoma cell line (2-2G4) secreting an IgM antibody reacting with bromelain-treated mouse red blood cells (10). The rearranged gene, known to reside on a 4.8-kb EcoRI/EcoRI fragment, was ligated into λ phage arms and used for generating a library. Positive clones were identified on colony lifts by hybridization with pJ11, and the λ phage was converted into a phagemid vector. The EcoRI 4.8-kb fragment was gel purified, modified by trimming the 3′ end with XhoI, and then ligated into the pICEM19H Cμ plasmid vector. Appropriate VH orientation was established by mapping several clones with appropriate restriction enzymes. The 17-kb VH11-Cμ construct was gel isolated and used for microinjection. Litters so generated were screened by PCR amplification for rearranged VH11-JH1 (Ig heavy chain joining segment) using DNA made from tail fragments. Tg+ founders were mated to C.B-17Icr, allowing discrimination of the transgene Cμ allotype (Igha) from the endogenous Cμ (Ighb). Two lines, high copy BR1 and low copy BR5, were backcrossed more than six times to C.B-17 and used for analyses. Human (Hu) μ Tg mice originally generated by M. Nussenzweig (22) and Rag-1− mice generated by E. Spanapoulou and D. Baltimore (15) were provided by E. Spanopoulou (Mt. Sinai School of Medicine, New York) and maintained in our animal facility. 3H9 Tg animals (23) were provided by M. Weigert (Princeton University, Princeton, NJ).
Cell Staining and Culture.
Single cell suspensions of bone marrow (from 3-mo-old animals) or fetal liver (from animals at day 16 of gestation) were stained, analyzed, and sorted as previously described (17, 18) with anti-CD45R(B220) (allophycocyanin-6B2), anti-CD43 (fluorescein-S7), and anti-CD24/HSA (phycoerythrin-30F1). Reanalysis of sorted fractions consistently showed purities >95%. The FLST2 stromal line–dependent proliferation assay was performed as previously described (15, 17), except that cultures were supplemented with 100 U/ml of recombinant human IL-7 (gift of S. Gillis, Immunex Corp., Seattle, WA). 2–5 × 104 cells (CD43+HSA+B220+) were sorted per 1 ml well. Cells were harvested after 4 d, stained as above, and analyzed by flow cytometry.
Digitonin lysates prepared from 105–106 cells were immunoprecipitated with antibodies to μ (M41), λ5 (LM34-8E), VpreB (VP245), or κ (187.1), and then proteins were resolved by SDS-10% PAGE as previously described (19). After electrotransfer to PVDF membranes (Dupont-NEN, Boston, MA), filters were blocked with milk protein, probed with horseradish peroxidase–conjugated M41 rat anti–mouse μ mAb, washed, developed using Supersignal Substrate (Pierce, Rockford, IL), and exposed to Reflection film (Dupont-NEN). In some experiments, μ immunoprecipitates were first deglycosylated by treatment with anhydrous trifluoromethanesulfonic acid according to the manufacturer's protocol (“GlycoFree”; Oxford GlycoSystems, Bedford, MA) before electrophoresis. Freshly isolated thymocytes were stimulated for 5 min at 37°C with 0.1 mM pervanidate to induce ubiquitination of TCR-ζ (24). The TCR complex was immunoprecipitated from lysates of treated or untreated thymocytes using anti–TCR-β mAb, resolved by SDS-11% PAGE, and transferred to PVDF membrane. VH11-μ immunoprecipitate was loaded in an adjacent lane. Filters were immunoblotted with an antiubiquitin antibody (MAB1510; Chemicon International, Temecula, CA).
Reverse Transcriptase PCR Analysis.
DNA constructs were introduced into the ret 02/1 cell line by electroporation as previously described (19). The SP6μ plasmid contains a neor gene derived from pSV2-neo in addition to the heavy chain gene (25). For transfections with transgenic constructs (VH11, VH11/Vκ9, VH3H9 , and VH81X ), cells were cotransfected with pSV2-neo to allow selection with G418. A rearranged Vκ9Jκ2Cκ light chain gene, present on an 8.4-kb BamHI fragment, was cloned from the 2-2G4 hybridoma (10). BamHI-digested DNA of this size was ligated into λ phage arms, the library was screened with nick-translated pECk probe, and a positive clone was ligated into pBluescript II KS (Stratagene, La Jolla, CA).
Results And Discussion
Transfection of μ heavy chain into pro–B cell lines results in the downregulation of TdT expression (19), providing a model system for investigating the effect of different VH segments on pre-BCR assembly. Expression of several μ heavy chains with different VH (VHDJH) regions, such as Sp6 and 3H9, in the pro–B cell line ret02/1 results in diminished message levels of TdT as well as Rag-1 (Fig. 1). These changes mimic those normally seen in the differentiation of pro–B cells to the early pre–B cell stage in vivo (18–21). However, not all μ heavy chains are equally efficient in this assay: transfectants with VH81X or VH11 μ chains do not show significant TdT or Rag-1 downregulation when compared to the parental line (Fig. 1). Since VH81X μ chains often do not associate well with λ5 (27), it was reasonable to ask whether pre-BCR assembly was inefficient in VH11 transfectants.
To assess the extent of pre-BCR assembly with VH11 μ, we performed immunoprecipitation experiments. SDS-PAGE analysis revealed that the VH11 μ protein expressed in transfected cell lines exists as two forms, a predominant species with an Mr 20 kD greater than other μ heavy chains tested, and a minor species with near normal mobility (Fig. 2,a). These μ species exhibited different extents of association with SLC: although the conventionally size μ pairs with λ5 and VpreB, this is not true of the other species. Only about half of the larger μ species is complexed with λ5, and even less (∼5%) is associated with VpreB (Fig. 2 a). Since both λ5 and VpreB are required for pre-BCR function, these data demonstrate that almost none of the larger μ protein is assembled into complete pre-BCR complexes. Moreover, because the larger μ form makes up 80% of VH11-μ, the assembled pre-BCR is decreased by fivefold in these transfectants.
Significantly, the aberrant molecular form of VH11-μ was not detected in cells cotransfected with Vκ9 light chain (Fig. 2,b), the light chain found frequently in mature VH11+ B cells (10, 11). This suggests that the higher Mr μ species represents a posttranslationally modified form, possibly targeted for degradation due to incomplete association with SLC, since a light chain capable of assembling with VH11-μ eliminates the aberrant band. Deglycosylation of VH11-μ (Fig. 2,c, left ) did not collapse the double band. Furthermore, the aberrant species was not detected in an antiubiquitin immunoblot (Fig. 2 c, right). The nature of modification responsible for the slowly migrating μ species is still under study, and failure to completely assemble with SLC does not always generate it, since VH81X chains that failed to associate with λ5 showed conventional size μ (27). Nonetheless, our data suggest that the downregulation of TdT and Rag gene expression in μ heavy chain transfectants depends on efficient assembly of a complete pre-BCR and therefore suggests that both VH81X-μ and VH11-μ are inefficient in mediating these changes.
To test whether normal nontransformed pre–B cells developing in vivo also show a comparable dependence on particular VH μ, we examined two lines of VH11 μ transgenic mice representing low (BR5) and high (BR1) copy number, comparing them with several other μ transgenic mouse lines. To eliminate any effect by endogenous μ expression and to restrict our analysis to the pre–B cell stage, we used Rag-1− mice bearing Ig Tgs. As Fig. 3 a shows, a differential ability to downregulate Rag-2 in adult bone marrow was clearly evident when comparing low copy number BR5 μ with 3H9 μ Tg mice, in agreement with data from the cell line transfection experiments. In addition, a human μ Tg (Hu μ), previously shown to promote B cell development in mice (15, 22), also induced downregulated Rag-2 expression. Thus, the extent of bone marrow pre–B cell progression in Ig transgenic mice, as monitored by changes in gene expression that we measured, appears to be dependent on VH-mediated pre-BCR assembly, with VH11 being particularly ineffective.
Unexpectedly, the high copy number BR1 VH11 μ Tg mice showed Rag-2 downregulation, different from the transfection data (Fig. 3,a); however, analysis of the μ protein in these two lines provides a potential explanation. Although the novel and conventional μ species are both generated in developing B lineage cells in these Tg mice, the conventionally sized μ predominates in pre-B cells from the high copy number BR1 mouse, whereas BR5 pre–B cells showed predominance of the aberrant size μ (Fig. 3 b), similar to the transfectants. Importantly, while BR1 pre-B cells express more total μ protein than do pre-B cells from non-Tg mice, the μ levels in the BR5 line are closer to those in wild-type mice. Thus, we consider that the BR5 line, which expresses physiological levels of μ and fails to promote Rag-2 downmodulation, provides a realistic picture of VH11 function in vivo. Moreover, the downregulation of Rag-2 in the BR1 line likely reflects super-physiologic μ expression that is presumably able to compensate for the inefficient assembly of VH11 with SLC.
To further pursue how the efficiency of pre-BCR assembly influences pre-B and subsequent B cell development, we next tested the growth response of fetal and adult pre–B cells to μ expression in stromal cell culture (Fig. 4), since a proliferative burst is another characteristic associated with early pre–B cell progression in the bone marrow. As Fig. 4 a shows, analysis of short term cultures of pre–B cells sorted from bone marrow of competent (Hu μ or BR1 μ) and incompetent (BR5 μ) adult Tg mice (on a Rag-1− background) revealed that Tg expression had relatively little effect on cell growth, with any enhancement in proliferation largely balanced by differentiation and exit from cell cycle. Strikingly, however, analysis of the comparable pre-B fraction isolated from fetal liver of the same transgenic mouse lines revealed that Hu μ or BR1 μ expression arrested cell growth. In contrast, the BR5 VH11 line showed little inhibition of fetal liver B-lineage proliferation, and instead allowed continued pre–B cell growth.
Furthermore, consistent with in vitro analysis, the frequency of pre–B cells in liver of newborn Rag-1− Tg mice was significantly reduced (relative to nontransgenic mice) in Tg animals with SLC-associating VH genes, such as Hu μ, but showed near normal levels in BR5 Tg mice (Fig. 4,b). Importantly, when analyzed on a Rag-1+ background, the majority of BR5 B220+ B-lineage cells in spleen of neonatal (1 wk) BR5 mice were surface VH11 Tg+, without endogenous μ surface expression (Fig. 4,c). In contrast, VH11 μ did not promote efficient B cell development in adult BR5 mice, since B cells in these mice comprised predominantly cells bearing exclusively endogenous μ. As with the Rag-2 downregulation analysis, the BR1 line appeared normal, showing decreased pre–B cell development early in ontogeny (Fig. 4,b) and predominance of Tg+ B cells in the adult (Fig. 4 c), likely due to superphysiologic VH11 μ levels. In summary, cells in fetal liver expressing SLC-nonassociating VH regions show a growth advantage compared to cells with SLC-associating VH segments. In contrast, the reverse is true in adult bone marrow where successful pre-BCR assembly is important for B lineage progression.
Our analysis suggests that a bias in VH representation can occur after successful VH-DJH rearrangement due to interaction (or lack thereof) with SLC, and this assembly-mediated response differs between fetal and adult pre–B cells. We suggest that skewed representation of B cells expressing certain VH μ, such as VH81X or VH11, generated during fetal time can result from active inhibition of clonal expansion of other B cells bearing SLC-associating μ. This model predicts a significant (and different) change in VDJ representation at the late pre–B cell stage during both fetal and adult B cell development. Most VH81X and VH11 productive heavy chain sequences reveal a fetal/neonatal origin reflected by a low level of CDR3 diversity (28, 29), due to low levels of TdT during the fetal stage, which results in little addition of extra nucleotides at heavy chain V–D and D–J junctions (18, 30). This has led some to suggest that the representation of certain VH genes (such as VH81X and VH11) is biased by a requirement for nucleotide homology at the junctions (28, 31), such that TdT-mediated addition in the adult would result in a preponderance of nonproductive joints. However, recent analysis of VH81X sequences generated from adult bone marrow pre–B cell cultures has revealed considerable CDR3 diversity in productive rearrangements (9), demonstrating that these can occur in the presence of TdT. Thus, the difference in TdT levels with ontogeny cannot account for the skewed repertoire difference.
As animals mature from the fetal to the adult stage, the potential to generate a more diverse VH μ repertoire increases due to TdT expression. This increasing heterogeneity of Ig heavy chain, important for generating more diversity in the adult, may require a more intricate mechanism for selection of appropriate VH μ, such as screening VH structures for an ability to pair with light chains. If this is the case, then SLC may provide a template for an “average” light chain structure. This process would be less important during fetal B lymphopoiesis where TdT is absent, resulting in expression of a more restricted set of VH regions. Thus a dependence of the pro-B to pre-B transition on efficient pre-BCR assembly in adult mouse bone marrow could be viewed as an elaboration on a simpler mode of B cell development represented in fetal liver.
Whether B cell development with these SLC-nonassociating VH genes is completely SLC-independent remains to be determined, since SLC is expressed at the highest levels on the surface of B lineage cells at a developmental stage before heavy chain expression (32), where it could conceivably provide a growth signal independent of μ. Regardless of a possible role for SLC in the pro–B receptor, it appears that the pre-BCR phase of fetal B lymphopoiesis is quite distinctive compared to that in adult bone marrow, allowing selective expansion of cells with certain VH genes. We speculate that over evolution, useful VH regions have been selected into the germline repertoire and that the distinctive response of fetal pre–B cells to SLC association provides a mechanism for their preferential expression. Determining why fetal B cells should express this primordial repertoire remains an interesting subject for future studies.
We thank Drs. M. Weigert, J. Kearney, and R. Perry for providing heavy chain constructs. We also thank Dr. T. Iwamoto for providing the ret02 line and Dr. H. Karasuyama for anti-15 and anti-VpreB antibodies. We appreciate critical reading of this manuscript by Drs. D. Kappes, M. Bosma, and A. Singer.
This work was supported by grants from the National Institutes of Health (AI-26782, AI-40946, CA-06927) and the American Cancer Society (DHP-82594), and by an appropriation from the Commonwealth of Pennsylvania.
R. Wasserman's current address is Division of Oncology, Children's Hospital of Philadelphia and Department of Pediatrics, University of Pennsylvania School of Medicine, Philadelphia, PA 19104. Y.-S. Li's current address is Beijing Sai-Yin-Si Institute of Biotechnology, Beijing 100024, China. C.E. Carmack's present address is Molecular Dynamics, 928 East Arques Ave., Sunnyvale, CA 94086-4520.
Address correspondence to Dr. Richard R. Hardy, Institute for Cancer Research, Fox Chase Cancer Center, 7701 Burholme Ave., Philadelphia, PA 19111. Phone: 215-728-2463; Fax: 215-728-2412; E-mail firstname.lastname@example.org