Extended Duration of DH–JH Rearrangement in Immunoglobulin Heavy Chain Transgenic Mice: Implications for Regulation of Allelic Exclusion

Ig transgenic lines of C.B-17 scid mice hemizygous for the H chain tgs, M54 (24), 3H9 (25), or the kappa L chain tg, Vκ8 (26), have been described previously (20). M54/Vκ8 and 3H9/Vκ8 scid mice were obtained by crossing individual tg lines and typing offspring for the presence of the respective tgs. Ig transgenic control mice, heterozygous for the scid mutation (denoted as scid/+ or s/+ mice), were obtained by crossing the above transgenic scid lines with C.B-17 wild-type mice. Mice used in this study were between 8 and 12 wk of age.

Bone marrow cells were flushed from femurs with staining medium using a syringe and 22 gauge needle. The cells were then dispersed by gentle pipetting, treated with 0.165 M NH₄CL, washed, and resuspended in staining medium and passed through a sterile nylon screen. B cell hybridomas were obtained by fusing unstimulated splenic cells from adult M54/Vκ8 (or 3H9/Vκ8) scid and scid/+ mice in a manner previously described (27) using Ag8.563 (28) as the cell fusion partner.

Bone marrow cell suspensions were analyzed for the presence of B lineage cells representing different stages of development (17). In brief, cell suspensions were stained with Cy5 (Biological Detection Systems, Inc.) or allophycocyanin (APC; PharMingen) conjugated anti–CD45(B220), FITC-conjugated anti–CD43, or biotinylated anti–IgM. Binding of biotinylated antibodies was revealed by Texas red conjugated Streptavidin (Southern Biotechnology). B220⁺CD43⁺IgM⁻, B220⁺CD43⁻IgM⁻, and B220⁺CD43⁻IgM⁺ cells were enumerated or sorted by multiparameter flow cytometry using a dual laser FACStar Plus^® (Becton Dickinson & Co.). Forward and light-angle scatter gates were set to include lymphoid cells only. Dead cells were identified by propidium iodide staining and excluded from analysis. To distinguish early B lineage subsets, B220⁺ CD43⁺-gated cells were stained with phycoerythrin-conjugated anti–BP1 (29) and biotinylated anti–heat stable antigen (HSA) (17) (both reagents were provided by R. Hardy, Fox Chase Cancer Center, Philadelphia, PA).

Genomic DNA was prepared from sorted cell subsets (0.5–1.0 × 10⁶ cells) as described previously (30) and dissolved in water at a concentration corresponding to 10⁵ cell genome equivalents/μl.

Ligation-mediated PCR (LM-PCR) (31–33) was used to assay DNA samples for double strand breaks (DSBs) resulting from the initiation of H chain gene rearrangement. Initiation of V(D)J rearrangement results in site-specific DSBs at the recombination signal/coding borders of V, D, and J elements: two kinds of broken DNA molecules are generated; covalently closed (hairpin) coding ends and blunt signal ends (34, 35). We assayed for broken molecules with signal ends; specifically, those with JH signal ends and those with 5′ or 3′ DHfl16.1 (DHfl) signal ends. We also assayed for signal joints (by inverse PCR), completed DH-to-JH rearrangements and unrearranged JH loci as scored by the retention of germline sequence immediately upstream of JH1.

Assays were performed as follows. A double strand linker was ligated to DNA (equivalent to ∼4 × 10⁵ cell genomes). The linker was constructed according to Roth et al. (32) by annealing two oligonucleotides, DR19 (5′-CACGATTCCC-3′) and DR20 (5′-GCTATGTACTACCCGGGAATTCGTG-3′). After ligation, different dilutions of the ligation reaction (input DNA) were used to perform PCR amplifications of one or more of the following: (a) linkered JH signal ends using DR20 and an oligonucleotide (MB221) complementary to a sequence immediately 5′ of JH1 (5′-TCTCTTGTCACAGGTCTCACTATGC-3′); (b) linkered 5′ DHfl signal ends using DR20 and an oligonucleotide (MB222) complementary to a sequence 5′ of DHfl (5′-GCCTTCCACAAGAGGAGAAG-3′); (c) linkered 3′ DHfl signal ends using DR20 and an oligonucleotide (MB241) complementary to a sequence 3′ of DHfl (5′-TGGGTCAGTGGTCAAGACTCG-3′); (d) signal joints resulting from the joining of JH1, JH2, or JH3 signals to the 3′ DHfl signal using MB221 and MB241; (e) DJH coding joints using a DHfl/DHsp primer MB109 (5′-CCGAATTCGTCCTCCAGAAACAGACC-3′) and a primer (MB 92) complementary to a JH4 sequence (5′-GCCGGATCCCTTGACCCCAGTAGTCC-3′); (f) retained sequence immediately upstream of JH1 using MB221 and MB92; and (g) α actin sequence using actin-specific primers (17). The level of amplified α actin product served as a control for the amount of input DNA.

DNA was amplified in a 50-μl reaction volume containing each primer at a concentration of 0.5 μM, 200 μM MgCl₂, 10 mM Tris-HCL, pH 8.3 at 25°C, 50 mM KCL, 0.001% gelatin, 200 μM (each) dTTP, dGTP, dATP, and dCTP (Pharmacia LKB Biotechnology), and 1 U Taq polymerase (Perkin-Elmer Cetus Corp.). The PCR reaction was carried out for 26 cycles of 94°C for 1 min, 60°C for 45 s (or 70°C for 45 s for amplification of JH signal ends), and 72°C for 90 s, followed by a 5-min extension at 72°C. Ligation and PCR amplification with different primers were performed at the same time to minimize experimental variation. Each assay included positive controls and was done several times with independent preparations of DNA. Amplification of PCR products was approximately proportional to the input DNA at several different dilutions. PCR products were separated by electrophoresis and analyzed by Southern blot analysis.

Blots were hybridized with: (a) pJH6.3 (36) to reveal H chain gene rearrangements, DJH coding and signal joints, and unrearranged JH alleles; (b) a genomic fragment corresponding to DHfl and its surrounding region (amplified by PCR using MB222 and MB 241) to reveal LM-PCR–amplified 3′ and 5′ DHfl signal ends; and (c) pActin (17) to reveal PCR-amplified α actin. Probes were labeled with α-[³²P]dCTP using a Prime-It II kit (Stratagene Inc.).

Fig. 1 is a schematic representation of the effects of μ and μ/κ tgs on scid B cell development. The different stages of B cell development are designated with the letter code of Hardy et al. (17); the alternative nomenclature of Rolink and Melchers (37) is shown for comparison. As indicated, B cell development in scid bone marrow is blocked at stage C, shortly after B lineage cells initiate H chain gene rearrangement. Relief from this block can be achieved by introduction of a μ tg into the scid genome. In μ-tg scid mice (22), developing pro-B cells appear to bypass stage C and develop directly into early pre-B cells, denoted as C′. Most cells in subset C′ are in cycle (17) and show downregulated RAG expression (23). Cells of subset C′ give rise to the D subset. At this late stage of pre-B cell differentiation, RAG expression is again upregulated (23) and L chain gene rearrangement is initiated (17, 18, 38). Differentiation does not proceed beyond stage D in μ-tg scid mice, presumably because scid pre-B cells are unable to repair DNA DSBs resulting from the initiation of κ gene rearrangement (22, 30). Complete relief from the scid block can be achieved in double tg scid mice, bearing both a μ and κ tg (20). In μ/κ-tg scid bone marrow, B cell development proceeds to stage E and appears to do so very rapidly, as evidenced by the near-normal percentage of B cells and virtual absence of pro- and pre-B cells.

Data supporting the above model are illustrated in Figs. 2 and 3. Fig. 2 shows the effect of two different μ tgs, M54 (24) and 3H9 (25), on scid B cell development before the late pre–B cell stage (stage D). Members of subsets C (BP1⁺HSA^dull) and C′ (BP1⁺HSA^bright), both positive for the early B lineage marker BP1 (29), are distinguished by their level of staining for heat stable antigen (17). Note that subset C, which is present in scid mice, appears to be replaced by subset C′ in M54 and 3H9 scid mice. Note also that the BP1⁻HSA⁺ cell fraction, which consists exclusively of HSA^dull cells in scid mice, includes both HSA^dull and HSA^bright cells in M54 and 3H9 scid mice. We designate HSA^dull and HSA^bright cells in the BP1⁻HSA⁺ fraction as B and B′, respectively. The upregulation of HSA when μ-tg expressing cells transit from stage B to B′ to C′ presumably reflects μ chain– dependent signaling. The effect of both μ and κ chain tgs on scid B cell development is shown in Fig. 3. As indicated, scid mice bearing M54 (or 3H9) and the L chain tg, Vκ8 (26), have near-normal percentages of B (B220⁺IgM⁺) cells in their bone marrow, but are severely deficient in early B lineage (B220⁺CD43⁺ and B220⁺CD43⁻) cells comprising subsets B–D.

Since pro-B cells (subsets B and B′) in μ-tg scid mice appear to differentiate directly or very rapidly into early pre-B cells deficient in RAG expression (subset C′), we suspected that many developing B lineage cells in these mice might not initiate DH–JH rearrangement until the late pre–B cell stage (D) when RAG expression is again upregulated. To assay for the initiation of DH–JH rearrangement, we tested for DSBs at JH recombination signal/coding borders in FACS^®-sorted B220⁺ CD43⁺ (CD43⁺) and B220⁺CD43⁻ (CD43⁻) bone marrow cells (CD43⁺ cells would include stages B-C′ and CD43⁻ cells would correspond to stage D). Broken DNA molecules with JH signal ends were detected by LM-PCR (31, 33). We also tested for completed DH-to-JH rearrangements and for retention of JH germline alleles (see Materials and Methods for details).

As shown in Fig. 4 and Table I, JH signal ends resulting from the initiation of DH–JH rearrangement in M54 scid mice were much more abundant in the late pre-B (CD43⁻) cell fraction than in the CD43⁺ fraction containing pro-B and early pre-B cells. In M54 scid/+ mice as well, JH signal ends were more abundant in the CD43⁻ than CD43⁺ cell fraction (Fig. 4 and Table I). We conclude that initiation of DH–JH rearrangement in M54 mice occurs predominantly at the late pre–B cell stage. Despite the relatively low abundance of JH signal ends in the CD43⁺ cell fraction of M54 scid/+ mice, alleles with (completed) DH–JH rearrangements were readily detectable in this cell fraction (Fig. 4). This is not surprising as DH–JH rearrangements would be expected to result in DJH complexes that have a much longer half-life than JH signal ends, especially in μ-tg–expressing pro-B cells that fail to rearrange their VH elements (20).

Non-tg scid and scid/+ control mice showed widely different levels of JH signal ends; i.e., JH signal ends were more abundant in the CD43⁺ cell fraction of scid than scid/+ bone marrow (Fig. 4 and Table I). As discussed later, a possible explanation for this difference is that initiation of DH–JH rearrangement continues unabated in scid mice, whereas in scid/+ mice, DH–JH rearrangement is limited by the onset of VH–DJH rearrangement. JH signal ends were in low abundance in the CD43⁻ cell fraction of scid/+ mice, consistent with a low retention of germline JH alleles (Fig. 4) and the completion of H chain gene rearrangement. The high retention of germline JH alleles in the CD43⁺ cell fraction of scid/+ (and scid) mice is presumed to reflect in part the presence of early B lineage cells not yet expressing RAG protein and the known contamination of this fraction with non–B lineage cells (39).

To test for ongoing DH–JH and VH–DJH rearrangement in the CD43⁺ and CD43⁻ cell fractions of scid, scid/+, and M54 scid/+ mice, we assayed for DSBs at both the 3′ and 5′ signals of the DHfl16.1 (DHfl) element. DHfl is the most upstream DH element (40) and is used in ≥50% of DH–JH rearrangements (41–44). Broken DNA molecules with 3′ DHfl signal ends signify initiation of DH–JH rearrangement, whereas 5′ DHfl signal ends can be taken to reflect initiation of VH–DJH rearrangement (33).

Scid and scid/+ mice showed striking differences in their levels of 3′ and 5′ DHfl signal ends (Fig. 5). In the CD43⁺ cell fraction of scid mice, 3′ but not 5′ DHfl signal ends were abundant, whereas, in the corresponding cell fraction of scid/+ mice, 5′ but not 3′ DHfl ends were abundant. Thus, in the CD43⁺ cell fraction of scid mice, initiation of DH–JH rearrangement predominates over that of VH–DJH rearrangement, whereas the reverse is true in the CD43⁺ cell fraction of scid/+ mice. In the late pre-B (CD43⁻) cell fraction of scid/+ mice, neither 3′ nor 5′ DHfl signal ends were detectable, indicating that H chain gene rearrangement is normally completed before this stage, which is in agreement with the results of Fig. 4. In contrast, in the CD43⁻ cell fraction of M54 scid/+ mice, DH–JH rearrangement was ongoing, as indicated by the abundance of 3′ DHfl signal ends (Fig. 5). Note that 5′ DHfl signal ends were not detectable in the CD43⁻ (or CD43⁺) cell fraction of M54 scid/+ mice. Therefore, even though DH–JH rearrangement is ongoing in late pre-B cells of M54 scid/+ mice, initiation of VH–DJH rearrangement does not evidently occur in these cells. This apparent inability of the V(D)J recombinase system to target VH elements in late pre-B cells of μ-tg mice is consistent with the early findings of Yancoupolus and Alt (45). These investigators found that VH558 transcripts are detectable in μ⁻ but not μ⁺ lines of transformed pre-B cells and concluded that VH elements in μ⁺-transformed pre-B cells are not accessible to the V(D)J recombinase system.

To test whether initiation of DH–JH rearrangement occurs at a normal frequency in μ/κ-tg mice, we sorted B220⁺IgM⁻ bone marrow cells from scid, 3H9/Vκ8 scid, and 3H9/Vκ8 scid/+ mice, and then assayed for the level of JH signal ends. The B220⁺ IgM⁻ cell population would include B lineage subsets (B–D) before the immature B cell stage (E). We also assayed for circular DNA molecules with signal joints resulting from the joining of the JH1, JH2, or JH3 signals with the 3′ DHfl signal (see Materials and Methods). Signal joint formation, in contrast to coding joint formation, is not impaired in scid mice (46, 47). Also, we would expect circular DNA molecules to have a longer half-life than broken molecules with JH signal ends, thus making signal joint formation a sensitive assay for attempted DH–JH rearrangement in scid mice.

As shown in Fig. 6, JH signal ends were more abundant in the B220⁺IgM⁻ cell fraction of non-tg scid mice than in the corresponding cell fraction of 3H9/Vκ8 scid and 3H9/ Vκ8 scid/+ mice. Thus, the initiation of DH–JH rearrangement is clearly reduced in the presence of these tgs. This is also apparent from the reduced level of signal joints in 3H9/Vκ8 mice compared with control non-tg scid mice (Fig. 6). The level of JH2 signal joints in 3H9/Vκ8 scid and 3H9/Vκ8 scid/+ mice was estimated to be ∼10 and 60%, respectively, the level in non-tg scid mice (see Fig. 6, legend). We suggest later that the lower level of signal joints in 3H9/Vκ8 scid than 3H9/Vκ8 scid/+ mice may be attributable to premature death of developing scid B cells resulting from persisting DSBs at DH and JH coding elements.

Given that most developing scid B cells fail to rearrange their D and J elements successfully (15, 48, 49) and die with persisting DSBs (33, 35), the cells most favored to become B cells in μ/κ-tg scid mice would be those in which DH–JH rearrangement is not attempted. To test this prediction, we generated and cloned B cell hybridomas from the spleen of M54/Vκ8 and 3H9/Vκ8 scid mice, and then examined these hybridomas for the status of their H chain alleles. Control hybridomas were obtained from M54/Vκ8 and 3H9/Vκ8 scid/+ mice. Representative results are illustrated in Fig. 7 for 9 3H9/ Vκ8 scid/+ hybridomas and 10 3H9/Vκ8 scid hybridomas. Note that one or two H chain gene rearrangements were clearly evident in all but one of the scid/+ hybridomas. In contrast, none of the scid hybridomas showed a rearranged allele.

44 hybridomas from μ/κ-tg mice were analyzed and the results are summarized in Table II. 10 scid/+ hybridomas showed one allele to be rearranged with the other allele in germline configuration; 11 (10 of which came from 3H9/ Vκ8 scid/+ mice) showed both alleles to be rearranged and 5 showed a single rearrangement with the other allele missing or undetectable. Two scid/+ hybridomas showed germline H chain alleles only. As normal B cells and their precursors show H chain rearrangements at both alleles (3, 8, 50), the retention of at least one germline H chain allele in ∼40% of the scid/+ hybridomas demonstrates significant transgene-mediated reduction of DH–JH rearrangement. These results are in agreement with the results of Fig. 6 and with previous reports showing that the frequency of endogenous H chain rearrangement is reduced in B lineage cells of M54 (10, 51) and 3H9/Vκ8 (25) wild-type mice. In contrast to the scid/+ results, all M54/Vκ8 and 3H9/ Vκ8 scid hybridomas showed germline H chain alleles only (Table II). The absence of detectable H chain gene rearrangement in the scid hybridomas indicates that cells that succeed in becoming B cells in μ/κ-tg scid mice do not attempt DH–JH rearrangement.

The preceding results show a striking alteration in the representation of B lineage subsets and duration of DH–JH rearrangement in bone marrow of μ-tg and μ/κ-tg scid mice compared with non-tg scid and scid/+ control mice. Specifically, late pro-B cells (subset C) appear to be missing in μ-tg scid mice and DH–JH rearrangement occurs predominantly at stage D, the late pre–B cell stage. In μ/κ-tg scid mice, early B lineage subsets (B–D) are grossly under represented and initiation of DH–JH rearrangement is less frequent than in non-tg control mice. Further, pro-B cells that succeed in reaching the B cell stage in μ/κ-tg scid mice do not attempt DH–JH rearrangement. Interestingly, in non-tg control mice, we found initiation of DH–JH rearrangement to be greater or more sustained in scid mice than in scid/+ mice. The implications of these findings are discussed below.

In μ-tg mice initiation of DH–JH rearrangement was found to occur most frequently at the late pre-B cell stage (stage D). As DH–JH rearrangement is normally completed before this stage (17, 18), initiation of H chain gene rearrangement appears to be somewhat delayed in μ-tg mice. To explain this result, we suggest the following model: μ-tg- and RAG-expressing pro-B cells (subsets B and B′) rapidly differentiate into recombinase-deficient early pre-B cells (subset C′), such that many cells do not have time to initiate or complete DH–JH rearrangement at both alleles until the late pre-B stage (subset D) when RAG expression is again upregulated.² Rapid progression of pro-B cells to the C′ stage would presumably result from premature expression of a pre-BCR containing a tg-encoded μ chain, surrogate light (SL) chain and the signal transducing chains, Igα and Igβ (reviewed in 13, 52–54). Consistent with this model is the known early expression of μ tgs (30) and the genes for SL chain (55–57), the apparent absence of subset C in μ-tg scid mice (see Fig. 2), the shortened duration of the pro-B stage in μ-tg mice (57a), and the finding that the majority of cells corresponding to subset C in non-tg mice contain nonproductive VDJH (VDJH⁻) rearrangements (58). The latter finding has been interpreted to suggest that pro-B cells containing a VDJH⁺ rearrangement quickly exit the subset C compartment (58).

Applying the above model to non-tg mice, we suggest that pro-B cells that make a VDJH⁺ rearrangement on the first attempt may exclude VH–DJH rearrangement at the other allele by rapidly progressing to the RAG-deficient C′ stage, and then to stage D, at which rearrangement of VH elements is no longer permissible. For allelic exclusion to occur in this model, a pre-BCR need only signal developmental progression. This notion is consistent with previous reports showing that exclusion of VH–DJH rearrangement is tightly linked with progression of pro-B cells to the pre-B stage (14, 59–61). Such linkage is even observed in μ-tg mice that express a truncated μ chain, which results in a pre-BCR complex lacking (specificity) a μ variable region and surrogate light chain (62, 63). However, pro- to pre-B progression and VH–DJH rearrangement are both blocked in μ-tg mice that express a mutated μ chain that precludes assembly of a pre-BCR complex with the signal transducing Ig α/β chains (64–67). Interestingly, the few B lineage cells that reportedly escape the above developmental block show allelic exclusion (67), consistent with our proposed model.

Ongoing initiation of DH–JH (Fig. 4) and Vκ–Jκ rearrangement (30) in late pre-B cells of M54 scid mice could help explain why these mice uniformly lack B cells (20) and appear no more leaky than non-tg scid mice (68). If attempted rearrangement of DH and JH elements in developing M54 scid pre-B cells is initiated before that of Vκ and Jκ elements, some cells might be expected to succeed in making a DH–JH rearrangement. Indeed, DH–JH rearrangements were recovered from late pre-B cells of M54 scid mice (see Fig. 4). However, the chance of a scid cell surviving attempted rearrangements at both H and L chain loci would seem unlikely, consistent with the absence of detectable VJκ coding joints in late pre-B cells of M54 scid mice (22, 30).

In μ/κ-tg mice, we found initiation of DH–JH rearrangement was less frequent than in non-tg scid mice. Signal joints resulting from the initiation of DHfl to JH rearrangement in 3H9/Vκ8 scid and 3H9/Vκ8 scid/+ mice were estimated to be present at ∼10 and 60%, respectively, the level observed in non-tg scid mice. Based on the difference in level of recovered signal joints in 3H9/Vκ8 scid/+ and non-tg scid mice, we estimate that initiation of DH–JH is ∼40% less frequent in μ/κ-tg than non-tg mice. This estimate agrees favorably with the observed frequency of germline H chain alleles in B cell hybridomas from 3H9/Vκ8 (6/35 alleles) and M54/Vκ8 (6/14 alleles) scid/+ mice (Table II). The much lower level of signal joints in 3H9/Vκ8 scid than 3H9/Vκ8 scid/+ mice is taken to reflect loss (death) of scid cells that attempt DH–JH rearrangement. This could account for the absence of rearranged H chain alleles in B cell hybridomas recovered from μ/κ-tg scid mice (Table II).

To explain the reduced level of DH–JH rearrangement in μ/κ-tg mice, we suggest that expression of a tg-coded BCR in early pro-B cells promotes very rapid progression of these cells to the B cell stage, such that there is little time to initiate DH–JH rearrangement. Consistent with this notion, (a) μ and Vκ8 tgs are known to be expressed early in B cell development (30), (b) μ/κ-tg scid/+ mice contain near-normal percentages of B cells in bone marrow but markedly reduced percentages of pro- and pre-B cells (20), and (c) μ/κ-tg scid mice show near-normal percentages of B cells in bone marrow but sharply reduced percentages of pro-B cells compared with non-tg scid controls and virtually no pre-B cells (≤1%; Fig. 3).

In non-tg control mice, we found that early B lineage cells in the CD43⁺ cell fraction from scid mice showed a much higher level of JH signal ends than the corresponding cell fraction from scid/+ mice. As scid does not impair the joining of signal ends (46, 47), one cannot attribute the relatively high level of JH signal ends in scid mice to a blockage in signal joint formation. What scid does impair, however, is the processing of coding ends before their being joined (33, 35). Thus, developing B lineage cells in scid mice do not often succeed in joining DH and JH coding ends (48, 49) and would not be expected to initiate the second step of H chain gene rearrangement (VH–DJH rearrangement). Indeed, 5′ DHfl signal ends, signifying the initiation of VH–DJH rearrangement, were not detectable in CD43⁺ scid cells (Fig. 5). We suggest that in the absence of VH–DJH rearrangement, initiation of DH–JH rearrangement continues unabated in CD43⁺ cells, resulting in a high level of JH signal ends. On the other hand, in the CD43⁺ cell fraction of scid/+ mice, initiation of VH–DJH rearrangement was prominent and that of DH–JH rearrangement barely evident (Fig. 5). This implies that initiation of DH–JH rearrangement in scid/+ mice may be limited to the earliest stage of pro-B cell development, consistent with the idea discussed below, that onset of VH–DJH rearrangement may preclude further DH–JH rearrangement.

In wild-type or scid/+ cells, a DH–JH rearrangement may be followed by rearrangement of a VH element to the resulting DJH complex or the complex may be replaced by the joining of an upstream DH element to a downstream JH element (41). The latter event, DJH replacement would seem counterproductive to efficient assembly of VH, DH, and JH elements. Thus, it makes sense, as originally postulated by Alt et al. (69), that after DH–JH rearrangement VH rather than DH elements are preferentially rearranged. How might this happen? Recent evidence suggests that initiation of VH–DJH rearrangement is associated with a shift in the targeting of the V(D)J recombinase activity from the 3′ to the 5′ side of DH elements (70). Targeting of the recombinase to signals on the 5′ side of DH elements would minimize DJH replacement and limit the duration of DH–JH rearrangement to the earliest stage of pro-B cell development. Although DJH complexes can be readily detected in late pro-B cells (subset C) (17, 18, 38), this does not necessarily reflect ongoing DH–JH rearrangement at this stage; the observed DJH complexes could have been formed earlier in cells of subset B.

In sum, DH–JH rearrangement in non-tg mice is normally initiated and completed at the early pro-B stage. In μ-tg mice, DH–JH rearrangement may begin at the pro-B stage, but it appears to continue and occur most frequently at the late pre-B stage. Based on the altered representation of pro-B subsets in μ-tg scid mice, we suggest that the extended period of DH–JH rearrangement in these mice primarily reflects rapid progression of μ-tg–expressing pro-B cells to the recombinase-deficient early pre-B cell stage. Thus, many cells may not have time to initiate DH–JH rearrangement until the late pre-B stage when RAG expression is again upregulated. In addition, ongoing DH–JH rearrangement (including DJH replacement) at the late pre-B stage would not be limited by initiation of VH–DJH rearrangement, as the latter does not apparently occur in late pre-B cells of μ-tg mice. Finally, rapid progression of μ/κ-tg–expressing pro-B cells to the recombinase-inactive B cell stage could explain why in μ/κ-tg mice we find a reduced initiation of DH–JH rearrangement compared with non-tg mice and a striking deficiency of pro- and pre-B cells despite near-normal numbers of B cells.

We thank Norman Ruetsch for help with Fig. 1 and Randy Hardy and Susan Shinton for providing anti– HSA and anti–BP1 reagents for the FACS^® analyses in Fig. 2. We also thank David Allman, Randy Hardy, Dietmar Kappes, N. Klinman, Pamela Nakajima, Norman Ruetsch, David Wiest, and Martin Weigert for review of the manuscript, and Roseanne Diehl for assistance in typing the manuscript.

This work was supported by grants from the National Institutes of Health (CA06927 and CA04946) and an appropriation from the Commonwealth of Pennsylvania.

BCR

B cell receptor

DSB

double strand break

HSA

heat stable antigen

LM-PCR

ligation-mediated PCR

RAG

recombinase activation gene

tg

transgene