IgM from an Eµ-deficient allele can drive B cell development
It has been assumed that one of Eµ's important roles within the Igh locus is to activate transcription of newly assembled IgH genes. Other enhancers that can drive IgH gene transcription lie at the far 3' end of the locus, but it has not been established whether or not this regulatory region, the 3' regulatory region (3' RR), is functional immediately after IgH gene formation in pro-B cells.

If transcription of newly formed IgH genes were Eµ dependent, expression of the V_H^a allele would be delayed until other transcription control elements (e.g., the 3' RR) became functional. This would arrest B cell development at the pro-B to pre-B transition in homozygotes but not in mice heterozygous for the V_H^a allele because, in heterozygotes, VDJ recombination and IgH gene expression could take place on the other WT Igh allele. If transcription of newly formed IgH genes were Eµ-independent, B cell development would be normal both in homozygotes and heterozygotes and, because of feedback inhibition of V-DJ recombination, the B cells of V_H^a/WT^b animals would express only the V_H^a allele.

B lineage cells were first examined in V_H^a/V_H^a animals and directly compared with V_HEµ^a/V_HEµ^a animals, a mouse strain which carries the same V_H gene knockin but with Eµ left intact (12). As shown in Fig. 2 A, the proportion of B cells in the splenic lymphocytes of V_H^a/V_H^a mice was normal and did not differ from that in V_HEµ^a/V_HEµ^a mice (comparisons of multiple mice with these genotypes are summarized in Table S1, available at http://www.jem.org/cgi/content/full/jem.20081202/DC1). IgM^a levels on the surface of resting splenic B lymphocytes was also the same, whether the B cells' Igh loci contained or lacked Eµ (Fig. 2 B). Consistent with surface IgM levels, quantitative measurements of Igµ messenger RNA (mRNA) produced by splenic B cells showed that the Eµ-deficient genes produced as much (or more) Igµ mRNA as Eµ-containing genes (Fig. 2 B).

View larger version (45K): [in this window] [in a new window] [Download PPT slide]   Figure 2. Spleen and BM B cell profiles in homozygous VHEµa and VHa mice. (A) Flow cytometry profiles of splenic lymphocytes from adult mice stained for B220 (B cell marker) and CD3- (T cell marker). (B) Surface IgMa and relative Igµ transcript levels in splenic B cells. Left, histogram of spleen cells stained with antibody to IgMa. Right, quantitative RT-PCR results, using primers for B1-8i µ mRNA and normalized to hgprt1 mRNA (see Materials and methods). µ mRNA from VHEµa B cells are set as 1.0 and WTb/WTb is the negative control. (C) BM B cell subsets. Top, BM cells gated for B220+ cells and analyzed for CD43 and IgM expression. Bottom, BM B220+ cells analyzed for IgD and IgM. Data shown are representative of three animals of each genotype.

Allelic exclusion is dramatically compromised in the absence of Eµ
It was shown previously that B cells in mice heterozygous for the V_HEµ^a (knockin) allele expressed only that allele, displaying what appeared to be perfect allelic exclusion (12). Because Igµ from the Eµ-deficient V_H^a allele could drive normal B cell development in homozygotes, we expected that B cells from mice heterozygous for this locus (V_H^a/WT^b mice) would similarly express only the V_H^a allele and not the alternate WT^b allele.

As shown in Fig. 3 A, IgM-positive BM cells from WT^a/WT^b mice expressed either IgM^a or IgM^b but not both. V_HEµ^a/WT^b BM B cells expressed only the V_HEµ^a allele, which is consistent with earlier reports (12). Quite clearly, however, BM B cells from V_H^a/WT^b mice included a subpopulation (10% of IgM⁺ cells) that broke the rules of allelic exclusion, displaying both alleles on the cell surface. This was even more striking in splenic B cells (20% of IgM⁺ cells) and in B cells of the peritoneal cavity (50% of IgM⁺ cells; Fig. 3 A). In stark contrast, V_HEµ^a/WT^b mice had no distinct population of double producers in the spleen, and such cells, although present in the peritoneal cavity, were much lower in number than in the V_H^a/WT^b mice (7% vs. 50%; Fig. 3 A).

View larger version (41K): [in this window] [in a new window] [Download PPT slide]   Figure 3. IgM allotype expression in VHEµa/WTb and VHa/WTb mice. (A) Flow cytometry profiles of lymphocytes in BM, spleen, and the peritoneal cavity, stained with antibodies to IgMa and IgMb. (B) IgM allotype expression on immature and mature B cells of BM (enriched for B220+ cells). Plots are of cells gated for IgMa alone (left), cells that were IgMa+ and IgD– (middle; immature B cells), or IgMa+ and IgD+ cells (right; mature B cells) and then analyzed for IgMa and IgMb. In both A and B, the number in top right quadrant = (percentage of double-positive cells)/(total IgM+ cells). (C) Percentage of IgMa+b+/IgM+ cells at different stages of development and in different tissues. n = number of mice analyzed. Error bars show SD.

To determine the frequency with which double producers were arising in the BM of adult mice, we used IgD/IgM expression to distinguish newly arising immature B cells (IgM⁺IgD^–) from recirculating mature B cells (IgM⁺IgD⁺). Double producers made up a mean of 5% of the immature B cells (Fig. 3, B and C). These expanded to a mean of 15% of mature BM B cells and underwent further expansion in the spleen (Fig. 3, B and C). In summary, analyses of the lymphoid tissues of V_H^a/WT^b mice revealed a profound defect in allelic exclusion, distinguishing these from V_HEµ^a/WT^b mice which carried an identical V_H insertion but with Eµ intact.

To confirm at the molecular level that cells staining with both antiallotype reagents in V_H^a/WT^b animals carried a successfully assembled Igµ gene on the Igh^b allele, splenic IgM^a+b+ cells were isolated and their genes examined by PCR. A primer specific for the WT Igh allele (J_H4 primer; Fig. 4 D) and primers specific for three V_H families were used to amplify and clone assembled V_H genes on the WT^b allele. 14 independent clones were sequenced, and all revealed a successfully assembled (productive) V_H gene on the WT^b allele (unpublished data).

View larger version (37K): [in this window] [in a new window] [Download PPT slide]   Figure 4. V-DJ rearrangements on the WT Ighb allele of VHa/WTb mice. (A) Igh locus. Probe C detects a sequence deleted upon V-DJ joining. Probe D detects a sequence retained on both alleles in all cells. (B) Assay for V-DJ rearrangement on the Igha and Ighb alleles in WTa/WTb and VHa/WTb mice. Genomic DNA extracted from liver and from sorted IgMa–b+ spleen cells (WTa/WTb; left) and IgMa+b– spleen cells (VHa/WTb; right). DNAs were digested with BamHI and hybridized with probes C (top) and D (bottom). (C) Quantitative analysis (ImageQuant) of blot shown in B. A repeat experiment was done using liver and spleen cells from VHa/WTb animals. Band intensities in liver were set to 100%. Normalization was to probe D for WTa/WTb mice; normalization was to the Igha fragment in VHa/WTb mice. Error bars show SD. (D) Igh locus maps indicating PCR primers and probes for detecting V-DJ rearrangements on Ighb allele. 5' primers specific for individual VH gene families (VH family primer) were used in combination with a 3' primer (JH4 primer), which was present on the Ighb allele but missing on both the VHEµa and VHa alleles. The representative blot demonstrates allele-specific amplification. H2O, no DNA template control.

To estimate the frequency of such rearrangements, we took advantage of a BamHI restriction fragment length polymorphism within the V-D interval. A probe lying within this interval (Fig. 4 A, probe C) detects a 9.6-kb BamHI fragment in liver DNA from 129/Ola (Igh^a) mice and a 7.5-kb BamHI fragment from C57BL/6 (Igh^b) mice. Because of their location, these BamHI fragments would be deleted upon fusion of any V_H with any D_H gene segment on the Igh^a and Igh^b alleles, respectively.

Splenic B cells were sorted from WT^a/WT^b F₁ animals and from V_H^a/WT^b mice. IgM^a–b+ cells from WT^a/WT^b mice almost entirely lacked the BamHI fragment derived from the WT^b allele (residual fragment results from contaminating cells in sorted populations) and retained only 30% germline (liver) levels of the larger BamHI fragment from the WT^a allele (Figs. 4. B and C). A BamHI fragment mapping 3' of the Igh loci (Fig. 4 A, probe D) was used to normalize DNA loading in these experiments. These results are consistent with earlier studies and demonstrate the inefficiency of the V_H assembly process such that many B cells (in this case, 70%) have undergone unsuccessful VDJ gene assembly on the unexpressed Igh^a allele before they attempt and succeed at assembly on the Igh^b allele (13, 26).

In B1–8 V_H knockin mice, the B1–8 V_H gene was inserted upstream of Cµ on the Igh^a allele and all germline J_H genes were simultaneously removed. As noted earlier, the inserted B1–8 V_H coding sequences were modified to inhibit V_H gene replacement events, so the 9.6-kb BamHI fragment lying between the V_H and D_H genes on this Igh^a chromosome should not be lost in any B cells from these mice (early studies reported no evidence for V-D rearrangements [reference 13]). As shown in Fig. 4 B, this fragment did, in fact, remain in DNA isolated from IgM^a+b– B cells from V_H^a/WT^b mice. The smaller BamHI fragment from the WT^b allele was also present in these cells and was present at 85–95% germline levels (Fig. 4 C). Quantification of the data showed no statistically significant difference from levels of this fragment in liver DNA (Fig. 4 C). We conclude that most cells expressing only the V_H^a allele have not undergone V-DJ rearrangement on the WT^b allele.

To address this issue further, we turned to a more sensitive measure of V-DJ rearrangement. Semiquantitative PCR was used to quantify VDJ rearrangements on the WT^b allele of IgM^a+b– splenic B cells (single producers) using V_H family primers and the WT Igh-specific primer described earlier. As shown in Fig. 5 A, V-DJ rearrangements on the WT^b allele were readily detectable in IgM^a+b+ double producers from the V_H^a/WT^b mice (similar in abundance to Igh rearrangements in the IgM^a+b– cells of WT mice). PCR products were also generated with each of the V_H family primers when DNA was isolated from the single producers (IgM^a+b– B cells) from V_H^a/WT^b spleen, but these were somewhere between 1/5 and 1/25 as abundant as those seen in double producers or in the IgM^a+b– B cells from WT mice. Importantly, this small but detectable level of V-DJ rearrangement on the WT^b allele was also seen in the IgM^a+b– spleen cells of V_HEµ^a/WT^b mice (Fig. 5 B). Consistent with the surface phenotype, these rearrangements, cloned from both animals, were nonproductive (18/18 from V_HEµ^a/WT^b mice and 18/19 from V_H^a/WT^b mice; unpublished data).

View larger version (88K): [in this window] [in a new window] [Download PPT slide]   Figure 5. V-DJ rearrangements on the WT Ighb allele of VHEµa/WTb and VHa/WTb mice. (A) Splenic B cells from WTa/WTb and VHa/WTb, sorted on the basis of phenotype (IgMa+b– or IgMa+b+). PCRs were performed on fivefold serial dilutions of DNA templates (four lanes/cell type). PCR strategy detects V-DJ rearrangements on only the Ighb allele in VHa/WTb mice (both alleles in WTa/WTb mice). VDJ rearrangements involving each JH gene segment are indicated (JH1, JH2, etc.). VH family primers are indicated to the right of blots (VHJ558, etc.). PCR of HS4 (a 3' RR element) was normalized for DNA input. (B) Like A, but a primer designed to anneal to all VH families (promiscuous VH 46) was included. (C) D-J rearrangements on the Ighb allele of pre-B cells (B220+CD43–IgM–) sorted from six mice of each genotype. The DHL primer anneals to most DH genes (47); DQ52 anneals only to DHQ52. Allele-specific 3' primer (JH4) was used as in A. Liver DNA from WTa/WTb mice and spleen cell DNA from homozygous VHEµa and VHa mice were included as controls. H2O, no template control. (D) Pre-B cells were isolated as in C and PCR reactions using the VH gene family primers were performed as described in A. (E) Pre-B cells were isolated as in C. Sample dilutions determined to be in the linear range (Fig. S4, available at http://www.jem.org/cgi/content/full/jem.20081202/DC1) were prepared in triplicate for PCR. Representative data is shown. (F) Quantitative analysis of VDJ rearrangement blots. For each PCR reaction with a given VH primer, signals for JH1, JH2, JH3, and JH4 rearrangements were quantified and summed by ImageQuant. Values obtained for pre-B cells from VHEµa/WTb and VHa/WTb mice were normalized to the corresponding values for pre-B cells of WTb/WTb mice. Data were obtained from two pools of two mice each for WTb/WTb and three pools involving a total of seven mice for each of the VHEµa/WTb and VHa/WTb mouse lines. Error bars show SD.

To increase the resolution of these analyses, we used twofold dilutions of template, confirmed linearity (Fig. S4, available at http://www.jem.org/cgi/content/full/jem.20081202/DC1), and performed triplicate PCRs of diluted template samples. Pre-B cells from seven mice of the V_HEµ^a/WT^b genotype and seven of the V_H^a/WT^b genotype were examined in three independent pools. Assembly of the WT^b allele in these pre-B cells was compared with that of the two WT (WT^a and WT^b) alleles in WT mice (two pools of two mice each). Representative data are shown in Fig. 5 E and a quantitative analysis of all data is provided in Fig. 5 F.

V-region gene assembly on the WT^b allele in the pre-B cells of both V_HEµ^a/WT^b and V_H^a/WT^b mice was reduced relative to the WT^a and WT^b alleles in pre-B cells of WT mice (Fig. 5 F). The reduction, relative to WT mice, was greater for the more distal V_HJ558 family than for the more proximal V_H7183 family. We conclude that the V_H^a and V_HEµ^a alleles were inhibiting, but not entirely prohibiting, V-DJ recombination on the allelic chromosome, and the feedback inhibition was largely, if not entirely, Eµ independent. Most importantly, however, the data showed that there was no significant difference in the amount of V-region assembly taking place on the WT^b alleles of V_HEµ^a/WT^b and V_H^a/WT^b pre-B cells. Assuming that some of these rearrangements were productive (confirmed by cloning as described in the next paragraph), both populations of pre-B cells included precursors to double-producing cells.

An Eµ-dependent checkpoint for allelic exclusion at the pre-B to immature B cell transition
Because the phenotypic outcome of gene assembly on the WT^b allele was so dramatically different in the V_H^a/WT^b and V_HEµ^a/WT^b mice (double producers evident in the immature and mature B cells of the former but not the latter mouse strain), we asked whether the V_H gene rearrangements in the pre-B cells of these two mouse lines differed in some fundamental way. Pre-B cells were harvested from mice of both genotypes, their DNA was isolated, and assembled V_H genes from the WT^b allele were cloned and sequenced. 15/36 unique V_H7183-DJ rearrangements cloned from pre-B cells of the V_HEµ^a/WT^b genotype were both in frame and lacked a stop codon (productive rearrangements), as were 12/36 unique V_H7183-DJ rearrangements cloned from V_H^a/WT^b pre-B cells. Closer examination of these productive rearrangements showed no fundamental differences among them in the D_H segments used, the 7183 V_H family members used, or the overall size of the junctions (Table S2, available at http://www.jem.org/cgi/content/full/jem.20081202/DC1). N nucleotides were found in the V-D junctions in >70% of the sequences, and 40% of these had three or more N nucleotides at this junction (mean length, 3.5 nt in V_HEµ^a/WT^b mice and 4.3 nt in V_H^a/WT^b mice). Notably, this contrasts with light chain genes where N nucleotides are much more rare (10% of cells) and are found in much lower numbers/junction (generally 1 or 2 nt/junction) (27). We conclude that the precursors to double producers can be found in BM from both V_H^a/WT^b and V_HEµ^a/WT^b mice and that V_H assembly on the WT^b allele has likely occurred before the pre-B cell stage (see Discussion).

In an attempt to understand the mechanism through which Eµ might influence the development and selection of precursor B cells, we compared levels of µ mRNA generated from the V_HEµ^a and V_H^a alleles in pre-B cells. To enrich for pre-B cells, V_HEµ^a and V_H^a mice were backcrossed to Rag1^–/– mice to generate V_HEµ^a/WT^b and V_H^a/WT^b mice that lacked Rag-1 activity. In both of the resulting strains, B cell development was arrested at the pre-B cell stage because Ig light chain gene assembly was blocked. Age-matched mice of each genotype were killed and B220⁺ BM cells isolated (see Materials and methods). As expected, no IgM⁺ cells were present in these BM cells, and the bulk of cells were smaller and expressed lower levels of CD43 than comparable cells from WT^b/WT^b Rag1^–/– littermates, which is consistent with their having progressed to the pre-B cell stage (Fig. 6 A). Total RNA was isolated from the B220⁺ BM cells of Rag1-deficient V_HEµ^a/WT^b and V_H^a/WT^b mice (WT^b/WT^b; Rag1^–/– littermates were controls) and Igµ mRNA was quantified by real-time RT-PCR (5' primer for unique VDJ junction of V_HB1-8 and 3' primer for C_H1 exon of Cµ; see Materials and methods).

View larger version (41K): [in this window] [in a new window] [Download PPT slide]   Figure 6. Igµ transcription and cytoplasmic Igµ protein levels in pre-B cells of mutant mice. (A) B220+ lymphocytes from BM of Rag1-deficient WTb/WTb, VHEµa/WTb, and VHa/WTb mice were analyzed for size (forward scatter) and for CD43 expression by FACS. Data shown are representative of three individual mice of each genotype analyzed. (B) Igµ mRNA levels in cells shown in A. Data were generated by quantitative RT-PCR, normalized to hgprt1 mRNA, and included two experiments, analyzing a total of five individual animals of each genotype. Negative controls were mRNA isolated from pro-B cells of a WTb/WTb Rag1–/– littermate and mRNA from C57BL/6 heart tissue (WTb/WTb, Rag1+/+). Statistical significance (P = 0.027) was obtained by a two-tailed Student's t test. Error bars show SD. (C) Cytoplasmic Igµ levels in B220+ BM cells of Rag1–/– mice. Left, histograms of cytoplasmic Igµ. Right, mean Igµ fluorescence in pre-B cells from multiple mice (n = number of mice analyzed). Error bars show SD.

Evidence for greater selective pressure on the developing B cells of V_H^a/WT^b mice
The reduced Igµ in V_H^a/WT^b pre-B cells might be suboptimal for preBCR-mediated clonal expansion and/or BCR-mediated selection into the immature B cell pool. Survival of precursors expressing both alleles (double producers) in V_H^a/WT^b mice might reflect a need for the increased Igµ levels achieved by biallelic expression. If this were true, IgM^a+b– cells might circumvent this problem through alternate means, perhaps by expressing light chains that yielded BCRs with superior signaling properties. This would likely require light chain receptor editing. Receptor editing can be achieved through successive V-J rearrangements in which V-J joins involving the more upstream J gene segments (J1 and J2) are replaced by joins to the more downstream J segments (J4 and J5). A degenerate V primer (VD) (28) was used in combination with a primer downstream of J5 (J5e) to examine J gene usage in splenic B cells of V_HEµ^a/WT^b and V_H^a/WT^b animals. As shown in Fig. 7 (A and B; and Fig. S5, available at http://www.jem.org/cgi/content/full/jem.20081202/DC1), the IgM^a+b– single producers of V_H^a/WT^b mice uniquely showed a decrease in V-J joins involving J1 and J2 gene segments and a concomitant increase in joins involving J5. The IgM^a+b+ double producers of V_H^a/WT^b mice were not similarly affected; these cells arose from precursors that would have been expressing both the V_H^a allele and the WT^b allele where Eµ remains intact.

View larger version (44K): [in this window] [in a new window] [Download PPT slide]   Figure 7. Light chain editing in IgMa+b– cells of VHa/WTb mice. (A) V-J rearrangement analyses. PCR products detected with J5 probe (see Materials and methods). (B) Quantitative analysis (ImageQuant) of blot shown in A and a repeat experiment. Detected V-J4 and V-J1 and 2 products were normalized to those for V-J5. *, P < 0.05 by a two-tailed Student's t test. Pie charts in Fig. S3 (available at http://www.jem.org/cgi/content/full/jem.20081202/DC1) show the relative increase in V-J5 rearrangements in IgMa+b– B cells from VHa/WTb animals. Error bars show SD. (C) + splenic B cells in WT and mutant mice. Spleen cells were stained for both Ig and Ig and the percentage of + cells was calculated as /(+). Genotypes are provided below bars. Data were pooled from four experiments. n = number of mice analyzed. P-values were calculated by a two-tailed Student's t test. Error bars show SD. (D) Histogram of cells stained for µa, µb, and . Left, single producers (IgMa+b–) from VHa/WTb mice; right, double producers (IgMa+b+) from VHa/WTb mice.

Ig

Ig

	DISCUSSION

Top ABSTRACT RESULTS DISCUSSION MATERIALS AND METHODS REFERENCES

 
The present study has shown that the intronic enhancer Eµ serves a critical function even after V_H gene (VDJ) assembly is complete. In mice heterozygous for an Eµ-deficient but productively rearranged IgH gene (V_H^a/WT^b), large numbers of IgH double-producing cells arose in peripheral tissues, breaking the rules of allelic exclusion. In contrast, a matched mouse line with the same V_H knockin but with Eµ present (V_HEµ^a/WT^b) lacked these double-producing cells. In both animals, there was evidence that the Igµ product of the assembled IgH gene was signaling a shutdown of V_H assembly on the Igh^b allele, which is the primary mechanism postulated to ensure IgH allelic exclusion (Figs. 4 and 5). There was no measurable difference in the efficiency of this feedback inhibition whether Eµ was present or absent.

This latter finding was unexpected. Eµ was initially discovered as a transcriptional enhancer and believed to serve as such in newly formed IgH genes (1–3). In that capacity, Eµ could dictate preBCR levels and, thereby, signaling strength in a developing B lymphocyte. Even in Eµ's absence, however, pre-B cells were dominated by those that had undergone D-J, but not V-DJ, rearrangement on their Igh^b alleles, demonstrating that feedback signals inhibiting this last step in recombination were no less efficient in the absence of Eµ.

Both V_H^a/WT^b and V_HEµ^a/WT^b mice contained rare pre-B cells with V-DJ rearrangement on their Igh^b alleles. Sequence analyses of these rearrangements revealed abundant N nucleotides in the V-D junction (Table S2), suggesting that they occurred in pro-B cells, the only developmental stage at which terminal transferase (TdT) is expressed (27, 33). Chromatin remodeling studies have documented that pro-B cells are the only developing B cells with contracted Igh alleles facilitating V-DJ rearrangements, especially those involving distal V_H gene segments (e.g., V_HJ558) (34). V_H J558-DJ rearrangements on the Igh^b allele were easily detected in pre-B cells of both V_H^a/WT^b and V_HEµ^a/WT^b mice. These results suggest that subsequent to V_H gene assembly, there is sufficient delay in the assembly of, or signaling through, the preBCR to allow RAG-mediated rearrangements on the second allele in a minority of cells before transition to the pre-B cell stage.

Signaling through the preBCR is required not only to down-regulate the RAG genes (terminating V-DJ recombination) but also to promote cell proliferation and differentiation to the pre-B cell stage (for review see reference 31). If the reduced Igµ chain level in V_H^a/WT^b pro-B cells resulted in preBCR signals that were suboptimal for promoting this transition, we might expect that precursors to double producers (expressing Igµ from two alleles, one with Eµ present) would be enriched in the pre-B cell population of V_H^a/WT^b mice. Similarly, if expression of the V_H^a were delayed relative to expression of the V_HEµ^a allele, a higher proportion of pre-B cells in V_H^a/WT^b mice would carry a functionally assembled V_H gene on the WT^b allele, as this would be required in the absence of V_H^a allele expression for transit to the pre-B cell stage. This was not the case, however; these precursors were present at the same or close to equal numbers in V_H^a/WT^b and V_HEµ^a/WT^b mice (Fig. 5 and Fig. S4). Quantitative comparisons of V-DJ assembly on the WT^b allele showed no significant difference between V_H^a/WT^b and V_HEµ^a/WT^b pre-B cells, and cloned V_H gene sequences from the WT^b allele had the same ratio of productive to nonproductive rearrangements in both pre-B cell populations. In RAG-1^–/– mice, the bulk of B-lineage cells in both mice assumed a pre-B cell phenotype, further supporting the idea that signaling through the Igµ chain (preBCR) to promote the pro- to pre-B cell transition was largely unaffected by loss of Eµ and lower Igµ levels (Fig. 6).

The transition from pre-B to immature B cell requires successful assembly of a functional Ig light chain gene, resulting in replacement of the preBCR with a BCR. The BCR on immature B cells serves two opposing purposes. On the one hand, it is required for survival (BCR ablation at any B cell stage tested leads to cell death), and it has been suggested that this BCR-transmitted tonic signaling is antigen independent (positive selection) (35). On the other hand, if the BCR signal is too strong (e.g., autoreactive BCR), the cell is induced to modify the antigen specificity of the BCR through receptor editing and, failing that, is induced to die (negative selection) (for review see references 30, 31). The nature and strength of the BCR signal in immature B cells, therefore, profoundly affects cell fate. We propose that reduced Igµ levels in emerging immature BM B cells of V_H^a/WT^b mice result in BCR signaling that is generally below the threshold for positive selection. In these animals, the rare cells that express both Igh alleles (double producers) would have a selective advantage because the increased tonic signaling through the BCR could then reach the required threshold for survival.

If, in the absence of Eµ, BCR density is generally not optimal for effective tonic signaling, then how do the single producers in V_H^a/WT^b animals reach the immature B cell stage? Both density and signaling strength can be affected by light chain through its effects on BCR stability (the strength of heavy/light chain association) and Ag specificity. B cells starting with a disability (low µ-chain levels) would have more stringent light chain requirements than normal, needing either light chains that form a stronger and more stable association with the V_HB1-8-µ chain or those that form an innocuous antigen specificity with stronger basal signaling properties. Analyses of these cells were consistent with that prediction. The IgM^a+b– cells in V_H^a/WT^b and V_H^a/J_H mice showed greater use of downstream J gene segments (evidence of successive V-J rearrangements on a single Ig allele) and included a larger subpopulation of Ig-producing cells as compared with IgM^a+b– cells from V_HEµ^a/WT^b mice (Fig. 7). Also consistent with this model was the finding that light chain editing was increased neither in the IgM^a+b+ cells from V_H^a/WT^b animals (µ-chain emanates from both a WT allele and the V_H^a allele) nor in the B cells of V_H^a/V_H^a mice (µ-chain emanates from two copies of the V_H^a allele).

As described in the previous paragraph, we predicted that inferior BCR signals caused by reduced Igµ in IgM^a+b– cells from both V_H^a/WT^b and V_H^a/J_H mice would lead to light chain editing. Unable to sense that a light chain gene has been successfully assembled, the cell would continue to accumulate V-J rearrangements. A correlation between underexpressed BCRs and light chain editing has been described previously (36, 37). Reciprocally, strong autoreactive BCR signals can also lead to light chain editing. In this case, the cell uses light chain editing to replace the light chain component of an undesirable BCR.

A corollary to this model is that in V_HEµ^a/WT^b mice, neither single producers nor double producers have a below-threshold signaling problem. Rather, in these mice, threshold signaling is achieved whether expression is monoallelic (single producers) or biallelic (double producers). Cells producing two different heavy chains at equally high levels, however, would be at greater risk of creating a receptor with unacceptable autoreactive properties, placing double-producers in V_HEµ^a/WT^b mice at a selective disadvantage (i.e., subject to clonal deletion) relative to single producers. It has been estimated that 75% of newly arising immature B cells harbor an autoreactive BCR (38–40). A cell producing two different heavy chains, each combining with a single unique light chain would have, on average, only a 6% chance (as compared with the usual 25%) of avoiding expression of an autoreactive BCR. Moreover, autoantibody silencing through formation of an alternate light chain (light chain editing) would be compromised in such a cell because a new light chain might solve the problem with one of the heavy chains and yet create a new autoreactive receptor with the other.

In summary, feedback inhibition of V-DJ rearrangement on the WT^b allele is evident in the pre-B cells of both V_H^a/WT^b and V_HEµ^a/WT^b mice. Despite this inhibition, V_H^a/WT^b and V_HEµ^a/WT^b mice have equal or close to equal numbers of pre-B cells that harbor two functionally assembled Igh alleles. Potential antigen specificity for both single and double producers is exactly the same in both animals after assembly with randomly generated Ig light chains. Nevertheless, the newly arising B cells differ significantly between these two animals, with allelic exclusion compromised only in the V_H^a/WT^b mice. The ultimate effect on the mature B cell pool is profound. As described and shown in Fig. 3 (and Fig. S2), the 5% double producers among immature B cells in V_H^a/WT^b mice expanded to comprise 20% of splenic B cells and 50% of B cells in the peritoneal cavity. We suggest that the difference in outcome in the V_H^a/WT^b and V_HEµ^a/WT^b mice (the immature B cells arising/surviving from the precursors) is a result of selection pressures, both positive and negative, that favor double producers only in V_H^a/WT^b mice. Allelic exclusion with respect to the Igh locus, therefore, can be viewed as, at minimum, a two-step process involving both a largely Eµ-independent (feedback inhibition of V-DJ recombination) and an Eµ-dependent (pre-B to immature B selection) phase. More generally, these results suggest that one of Eµ's important functions is to ensure that monoallelic expression of most newly assembled IgH genes is sufficient to support both B cell development and entry into the mature B cell pool. Anything that reduces that expression (variants in promoter sequences, enhancer sequences, or signaling components) puts the system at risk for promoting the development of allelically included cells with potentially autoreactive properties.

Unlike cells expressing two different light chains in which one is generally expressed on the surface to the exclusion of the other (41, 42), the mature B cells expressing two different heavy chains in V_H^a/WT^b mice can be expected to express both types of receptor on the cell surface at comparable levels. It is likely of significance to issues of autoimmunity that expression of the V_H^a allele is low at the time of selection but high once cells reach the mature B cell compartment. Notably, many of the IgM^a+b+ double producers in these animals occupy the marginal zone of spleen and constitute most, if not all, CD5⁺ B1 B cells in the peritoneal cavity. It is not clear at this point how precursors of IgM^a+b+ double producers are selected into these mature B cell compartments. One possibility is that expressing the WT^b allele fulfills the antigen-specificity requirements associated with B1 B cell development (43). As CD5⁺ B1cells have been shown to require unusually strong and often autoreactive BCR signals for survival (43), it is also reasonable to speculate that monoallelic expression of the V_H^a allele, even with the assistance of light chain editing, is insufficient to signal development and/or maintenance of this B cell subset, leading to selective expansion of the double producers into this B cell pool.

	MATERIALS AND METHODS

Top ABSTRACT RESULTS DISCUSSION MATERIALS AND METHODS REFERENCES

 
Generating V_H mice
To generate the B1-8VDJEµ targeting vector, K. Rajewsky (Harvard Medical School, Boston, MA) and W. Muller (University of Manchester, Manchester, England, UK) provided us with a vector (B1-8iVDJ) (12) that we modified to replace the 3' homology arm with a 1.4-kb sequence between Eµ and Sµ (nt 3880–5295, GenBank accession no. J00440). Both the B1-8VDJEµ and B1-8iVDJ targeting vectors included 2-kb natural 5' flanking DNA to the B1-8V_H gene, and the 5' homology arm was comprised of 9 kb of DNA lying immediately upstream of D_HQ52 in the murine Igh locus. B1-8V_H coding sequences carried a TGT to TGC silent mutation at codon 92 to prevent V_H replacement events (12). The ClaI–NotI fragment, serving as 3'homology arm in the B1-8iVDJEµ targeting vector, was generated by PCR, using DNA from the 129P2/OlaHsd-derived ES (E14.1) cell line as template. Primers used were the following: 5'-CTCATCGATTCGGTTGAACATGCTGGTTG-3' (ClaI site underlined) and 5'-CTCGCGGCCGCAGTGTAGGCAGTAGAGTTTA-3' (NotI site underlined).

The B1-8VDJEµ targeting vector was used to transfect ES cell line E14.1, and three appropriately targeted clones were used to generate chimeric mice (Gene Targeting and Transgenic Service Laboratories, Rockefeller University). Successful germline transmission was achieved from one of the lines. Positive offspring that were heterozygous for the neo^RV_H allele were mated to EIIa-cre mice (C57BL/6J background; provided by H. Westphal, National Institutes of Health, Bethesda, MD) (25) to induce deletion of the neomycin resistance gene. The resulting mouse line was designated V_H.

V_HEµ mice were previously described as B1-8i mice (12) (supplied by K. Rajewsky, CBR Institute for Biomedical Research, Boston, MA). Mice carrying Igh alleles that lacked the J_H gene segments were obtained from The Jackson Laboratory (B6.129P2-Igh-J^tm1Cgn/J; stock #002438). The mutant allele lacking the J_H gene segments is called J_H in the present studies.

The V_H knockin for V_HEµ was also accomplished in ES 14.1 cells. Both V_HEµ and V_H mice were mated to C57BL/6J mice for allelic exclusion analyses. Progeny of intercrosses between V_HEµ^a/WT^b and V_H^a/WT^b mice were also analyzed, and double producers were strictly correlated with the V_H allele. Mice were bred and maintained in animal facilities at Hunter College, City University of New York, and all mouse experiments were approved by the Hunter College Institutional Animal Care and Use Committee.

Southern blots
Genomic Southern blot analyses were performed as previously described (44). 20 µg of restriction enzyme–digested genomic DNAs were size fractionated on 0.8% agarose gels and the DNA was transferred to nylon membrane (GE Healthcare). Blots were hybridized with ³²P-labeled probes generated by the random priming method (MegaPrime; GE Healthcare). Probe A (Fig. 1) is a 1.2-kb XhoI–HindIII fragment cloned from the B1-8iVDJ targeting vector (nt 25622029–25623228; GenBank accession no. NT_166318). Probe B (Fig. 1) is a 724-bp sequence that begins upstream of Cµ and covers C_H1 and 1/2 of C_H2 (nt 1214283–1215005; GenBank accession no. NT_114985). Probe B was generated by PCR, using E14.1 ES cell DNA as template. Primers used were the following: CµU1, 5'-CAAGGAAATAGCAGG GTGTAG-3'; and CµD1, 5'-CTTTGTTCTCGATGGTCACC-3'. Probe C (V-D interval; Fig. 4) is a 1,056-bp sequence 73 kb upstream of the most C_H-distal D gene fragment DFL16.1 and 25 kb downstream of the most C_H-proximal V gene fragment V_H81X (nt 7471–8526; GenBank accession no. AC073553). Probe C was generated by PCR from C57BL/6 mouse genomic DNA using the following primers: E1f, 5'-CATCCAGATACAGCACTCCCTTGTGTC-3'; and E1r, 5'-GAAGGCCAGGACCAAGGATTGAATAC-3'. Probe D (Fig. 4) is a 1,469-bp sequence derived from a region 70 kb 3' of the hs4 element within the Igh 3' RR (nt 25343884–25345352; GenBank accession no. NT_166318). Probe D was generated by PCR from C57BL/6J mouse genomic DNA with the following primers: Df, 5'-CTGAAGTTGGATGTAGGCCTGAAACTG-3'; and Dr, 5'-CCTCCCAATGCTAAGTAGAAACAGACG-3'. The JH4 probe is a 156-bp probe for J_H4 (nt 134365–134520; GenBank accession no. AC073553) that is used to detect DJ and VDJ rearrangements on the Igh^b allele. The J_H4 probe was generated by PCR from C57BL/6 mouse germline DNA with the following primers: JH4f, 5'-CTATGGACTACTGGGGTCAAGGAAC-3'; and JH4r, 5'-CAACTTCTCTCAGCCGGCTC-3'. The hs4 probe is a 601-bp probe for the hs4 element of the Igh 3' RR (nt 25418984–25419584; GenBank accession no. NT_166318).

Flow cytometry
Generally, 10⁶ cells were incubated at 4°C for 15 min in staining buffer (PBS, 5.6 mM glucose, 0.1% BSA, and 0.1% NaN₃) with monoclonal antibodies (BD or SouthernBiotech). Single-cell suspensions prepared from BM and spleen were treated with ACK buffer (0.15 M NH₄Cl, 10 mM KHCO₃, and 0.1 mM EDTA, pH 7.3) to lyse erythrocytes. Cells from the peritoneal cavity were recovered by flushing the cavity with cold RPMI-1640 media (Mediatech, Inc.), containing 5% bovine calf serum (Thermo Fisher Scientific). In most cases, propidium iodide was added before analyses to exclude dead cells.

Side and forward scatter were used to gate on lymphocytes. FACS instruments were used for flow cytometry and sorting (FACScan, FACSCalibur, and FACSVantage; BD). In some cases, BM B lineage cells were first enriched by positive selection (MACS B220 MicroBeads kit; Miltenyi Biotec). Data were acquired with CellQuest or Pro CellQuest (FACS instruments) and then further analyzed with FlowJo software (Tree Star, Inc.).

Monoclonal antibodies were obtained from BD, except where indicated, and conjugated to fluorescein-isothiocyanate (anti–mouse CD3-, anti-IgM, anti–mouse IgM^b, anti-IgD, anti–mouse CD21, and anti–mouse Ig), phycoerythrin (anti–mouse IgM^a, anti–mouse CD23, anti–mouse CD43, and anti–mouse Ig [SouthernBiotech]), allophycocyanin (anti–mouse B220 and anti–mouse CD5), or biotin (anti–mouse IgM^b, anti–mouse IgM^a, and anti–mouse IgD [Southern Biotech]). Cells were then washed in washing buffer (PBS, 5.6 mM glucose, and 0.1% NaN₃) and biotin-conjugated monoclonal antibodies were revealed with streptavidin-allophycocyanin (BD) or streptavidin-phycoerythrin (BD).

Cytoplasmic Igµ staining
BM cells were incubated with antibodies to surface antigens, washed, and then fixed and stained with rat anti–mouse IgM (eBioscience) using Cytofix/Cytoperm (BD).

PCR
Genotyping mice.
All genotyping PCR reactions were performed for 40 cycles with an annealing temperature of 60°C. PCR kits with HotStar Taq polymerase were used (QIAGEN), and the manufacturer's protocol was adhered to. Primers used were the following: number 2, 5'-CAGAGGGAGTTCACACAGAGCATG-3' (within the 3' homology region of B1-8iVDJEµ; nt 136031–136054; GenBank accession no. AC073553); number 4, 5'-TCTTTACAGTTACTGAGCACACAGGAC-3' (immediately 5' of the leader exon of B1-8V_H; nt. 312629–312655; GenBank accession no. BN000872); and number 8 (Eµ), 5'-CTTCCCTCTGATTATTGGTCTCCATTC-3' (nt 135733–135759; GenBank accession no. AC073553). These three primers were used together. Numbers 2 and 4 generate an 848-bp product from the targeted Igh^a locus, and numbers 2 and 8 generate a 322-bp product from the WT Igh^b locus. The PCR products generated by these three primers (numbers 4, 8, and 2) overlap the region of VDJ insertion.

Confirming neo^R deletion.
Primers used were the following: number 9, 5'-CCCACCATCACAGACCTTTCTCCATAG-3' (within the 5' homology region of the targeting vectors; nt 132108–132134; GenBank accession no. AC073553); and number 10, 5'-CTGAGGGCAGCAGTACAATGATGAGTC-3' (within the 5' flank of B1-8V_H and 280 bp 3' of loxP-flanked neo^R; primer differs by two nucleotides [underlined] from nt 311488–311514; GenBank BN000872). Primers number 9 and 10 can generate a product only from the targeted Igh^a allele. The PCR product from the neo^RV_H^a allele was 1,500 bp before neo^R deletion and, from the V_H^a allele after neo^R deletion, was 400 bp.

Detecting D-J and V-DJ rearrangements in B-lineage cells.
To analyze V-DJ rearrangements in isolated B-lineage cells, a previously described protocol (7) was followed with slight modification. Genomic DNA template was isolated from sorted cells, either by conventional procedures or by the following method: cells were lysed in 10 mM Tris-HCl, pH 8.0, and 0.1 mM EDTA (50 µl/10⁵ cells), proteinase K was added to a concentration of 0.5 mg/ml, and the mixture was incubated for 2.5 h at 50 and 95°C for 10 min. Serial dilutions of the purified DNA (or cell lysates) were used as a template to specifically amplify D-J and V-DJ rearrangements derived only from WT Igh alleles. Allele specificity was achieved by using a 3' primer that derived from sequences 3' of J_H4 and that were missing on the targeted Igh^a allele of both V_HEµ and V_H mice. The 5' primers were "degenerate" primers designed to anneal to the D_H genes, one of three V_H gene families (J558, Q52, or 7183 family), or all V_H genes. Primers for D_H families were the following: The primer for D_HL, 5'-GGAATTCGMTTTTTGTSAAGGGATCTACTACTGTG-3', is a degenerate primer that anneals to most murine D_H sequences (45); and the primer for D_HQ52, 5'-CCACAGGCTCGAGAACTTTAGCG-3', anneals to the most J_H-proximal murine D_H sequence, DQ52 (7). Primers for V_H families were the following: V_HJ558 family primer, 5'-GCGAAGCTTARGCCTGGGRCTTCAGTGAAG-3' (7); V_HQ52 family primer, 5'-GCCAAGCTTCTCACAGAGCCTGTCCATCAC-3' (7); V_H7183 family primer, 5'-GCGAAGCTTGTGGAGTCTGGGGGAGGCTTA-3' (7); and promiscuous V_H primer, 5'-GGGAATTCGAGGTGCAGCTGCAGGAGTCTGG-3' (46). Each of these primers was used in conjunction with the following primer, which lies 3' of J_H4 and cannot anneal to the targeted Igh^a alleles: 5'-AGGCTCTGAGATCCCTAGACAG-3' (nt 134530–134551; GenBank accession no. AC073553). PCR products were size fractionated on 1.5% agarose gels, blotted, and probed with a ³²P-labeled J_H4 probe (described in Southern blots). DNA amounts were normalized to a PCR product derived from the Igh locus 3' RR element hs4 present as one copy/haploid genome in all cells (as in reference 7). HS4 primers used were the following: 5'-CCAAAAATGGCCAGGCCTAGG-3' (nt 25419564–25419584; GenBank accession no. NT_166318); and HS4-3', 5'-AGGTCTACACAGGGGCTCTG-3' (nt 25418984–25419003; GenBank accession no. NT_166318). PCR conditions for detecting V-DJ and D-J rearrangements were 30 cycles at 94°C for 1 min, 60°C for 1 min, and 72°C for 3 min. The exceptions were for data shown in Fig. 5 (E and F) and Fig. S2. In these instances, HS4 was amplified for 20 cycles instead of 30.

V-J rearrangements.
The analysis of V-J rearrangements was the same as for IgH VDJ rearrangements except that different primers and probes were used. A degenerate V primer (VD) (28) was used to detect all V genes, and a primer downstream of J5 (J5e) was used to detect rearrangements, regardless of J fragment used. Primers used were the following: VD, 5'-GGCTGCAGSTTCAGTGGCAGTGGRTCWGGRAC-3' (a degenerate primer that anneals to most V gene sequences) (28); J5e, 5'-CTGACACTGTATGCCACGTCAACTG-3' (downstream of J5; nt 3168393–3168418; Gene ID: 243469; derived from GenBank NC_000072). The J5 probe used for detecting PCR products was generated by PCR, using the following primer pair: J5f, 5'-GCTCACGTTCGGTGCTGGGAC-3' (nt 3168251–3168271; Gene ID: 243469; derived from GenBank NC_000072); and J5r, 5'-ATAATGAGCCCTCTCCATTTTCTCAAG-3' (nt 3168367–3168393; Gene ID: 243469; derived from GenBank NC_000072).

Real-time RT-PCR
For analyses of Igµ transcripts in splenic B cells, B cells were enriched by negative selection (B cell isolation kit; Miltenyi Biotech). For analyses of Igµ transcripts in pre-B cells, enrichment was by positive selection for B220⁺ cells (B220 MicroBeads kit; Miltenyi Biotech). Total RNA was isolated with the RNeasy mini-prep kit (QIAGEN). Real-time RT-PCR analyses of Igµ mRNA transcribed from the V_H and V_HEµ^a alleles were performed with the QuantiTect SYBR Green RT-PCR kit (QIAGEN). Igµ transcripts were amplified with a 5' primer that annealed to the unique VDJ junction sequence of V_HB1-8 and a 3' primer that annealed to a sequence within the C_H1 exon of Cµ. The VDJ junction primer used was 5'-CGCAAGATACGATTACTACGG-3' and the Cµ primer used was 5'-GAAGACATTTGGGAAGGACTG-3' (nt 140198–140218; GenBank accession no. AC073553). Samples were normalized using primers and probes for mRNA from the housekeeping gene hgprt1 (hypoxanthine/guanine phosphoribosyl transferase; Applied Biosystems; TaqMan gene expression assay ID: Mm03024075_m1) and the TaqMan one-step RT-PCR kit (Applied Biosystems). PCR reactions were performed on the 7500 Real-Time PCR System (Applied Biosystems) and analyzed with a program supplied by Applied Biosystems (User bulletin #2; http://docs.appliedbiosystems.com/pebiodocs/04303859.pdf).

Online supplemental material.
Additional results show PCR-cloning results demonstrating that surface Ig phenotype correlated with cloned IgH genes from sorted spleen cells. Fig. S1 shows evidence of CRE-mediated neo^R deletion. Fig. S2 shows that double producers were enriched among marginal zone B cells in spleen and B1 B cells in peritoneal cavity. Fig. S3 shows analyses of newborn spleen and fetal liver cells for double producers. Fig. S4 shows establishment of quantification in V_H gene assembly PCR experiments. Fig. S5 shows pie charts quantifying relative use of J segments in V-J rearrangements in WT versus mutant mice. Table S1 shows absolute numbers of B cell subsets in BM and B and T cells of spleen in mutant and WT mice. Table S2 shows DNA sequences of assembled V_H genes cloned from WT and mutant mice. Online supplemental material is available at http://www.jem.org/cgi/content/full/jem.20081202/DC1.