A reappraisal of evidence for probabilistic models of allelic exclusion

Abstract

B cell development requires the coordinated rearrangement of Ig heavy (IgH) and light chain loci (IgL). Most mature B cells express a single B cell receptor of unique specificity, and a central question in immunology concerns the mechanisms that prevent the productive rearrangement of >1 IgH and IgL allele per cell. Probabilistic models of allelic exclusion maintain that simultaneous rearrangement of both alleles is rare, because the likelihood of undergoing rearrangement is low for a given Ig allele. Strong support for this idea came from studies in which a GFP marker was inserted into the Igk locus. In this system, the probability of high-level germ-line transcription and subsequent locus rearrangement appeared to be low in pre-B cells. Readdressing the validity of GFP expression as a reporter for the level of germ-line transcription, we found a striking discordance between GFP transcript and protein levels at the pre-B cell stage, which is explained at least in part by the developmentally regulated usage of 2 alternative Igk-J germ-line promoters. These results question the validity of the kappa-GFP system as evidence for probabilistic models of allelic exclusion.

B lymphocyte development is characterized by a series of genomic rearrangement events at the Ig heavy and light chain loci (IgH and IgL) that are required to produce a mature antigen receptor. Productive rearrangements of Ig loci are usually restricted to 1 allele, ensuring allelic exclusion and monospecificity of the antigen receptor. Successful rearrangement of the IgH locus at the pre-BI cell stage (also referred to as pro-B cell stage) and pairing of the IgH chain with the surrogate light chain proteins VpreB and λ5 (encoded by the Vbreb1 and Igll1 loci) results in the expression of a functional pre-B cell receptor (BCR). Pre-BCR signaling drives allelic exclusion of IgH, cell cycle entry, and a transient phase of proliferation, described here as the large pre-BII cell stage. When pre-BCR signaling is down-regulated in a process that requires the adapter proteins SLP-65 and linker for activation of T cells (LAT) (1–3), pre-BII cells drop out of cycle to become small pre-BII cells, which initiate Ig kappa (Igk) light chain rearrangement. Because germ-line transcription facilitates rearrangement (4, 5), reports of biallelic Igk germ-line transcription (6) raised the issue how monoallelic Igk rearrangement is achieved. One explanation could lie in developmentally regulated epigenetic events that set the 2 alleles apart and the finding of monoallelic CpG (DNA) demethylation of Igk alleles (7), as well as allelic nonequivalence for other epigenetic parameters (8–10) are consistent with such models. No developmentally regulated distinction between the alleles is required in probabilistic models of allelic exclusion. Here, the likelihood of locus rearrangement is low, making the simultaneous rearrangement of both loci a rare event. An elegant experimental system based on the introduction of GFP into the Igk locus has led to a general acceptance of probabilistic models of allelic exclusion (11). To provide a reporter for Igk germ-line transcription, Liang et al. (11) inserted a GFP cDNA into the J-kappa locus to generate kappa-GFP^+/− mice. According to their report, ≈5% of heterozygous pre-BII cells were GFP positive. When GFP⁺ cells were isolated by flow cytometry and analyzed after overnight culture, they contained Igk rearrangements. The reported data suggested that high-level Igk germ-line transcription and subsequent rearrangement occurred with a probability of 5% (or 1 in 20 alleles). According to this interpretation, the likelihood of simultaneous rearrangement of both Igk loci would be sufficiently low to account for allelic exclusion. To investigate the genetic and epigenetic factors that set the probability of Igk locus activation, we have interrogated Igk germ-line transcripts. Contrary to expectations, there was a profound discordance between undetectable GFP protein expression and active germ-line transcription at the small pre-BII cell stage. This dissociation is explained at least in part by our finding that the activity of the 2 Igk-J germ-line promoters is developmental stage-specific. Pre-BII cells predominantly use the 5′ promoter, and the resulting transcripts do not encode functional GFP. Our results cast doubt on the validity of the kappa-GFP system as a cornerstone of probabilistic models of allelic exclusion.

Results

Developmentally Regulated Usage of Igk Germ-Line Promoters.

Two germ-line promoters are known within the Igk joining (J) region. The 5′ promoter (5′k^o) is located 3.6-kb upstream of Jk1. A primary transcript of 8.4 kb reads through to the Igk Constant region (IgkC), and is spliced to IgkC after 394 bp. The 3′ promoter (3′k^o) is just upstream of Jk1 (−52 to −126 bp), and initiates a 4.7 kb transcript, which is spliced to IgkC downstream of Jk1 (Fig. 1A). The resulting mature transcripts span 1.1 and 0.8 kb, respectively (12, 13). To determine 5′k^o and 3′k^o promoter usage during development, we used flow cytometry to sort developing B cells from bone marrow and mature B and T cells from spleen. We isolated B220^lo CD43⁺ IgM⁻ pre-BI cells, B220^lo CD43⁻ IgM⁻ pre-BII cells separated into large and small based on forward scatter, as well as splenic B (B220^hi) and T cells (TCR⁺) (Fig. 1B). Real-time RT-PCR showed that both transcripts increase at the small pre BII stage when Igk rearrangement takes place. Interestingly, the activity of the 5′ promoter declined after the small pre-BII cell stage, whereas the 3′ promoter remained active in mature B cells, providing evidence for differential usage of the Igk-J germ-line promoters during development [Fig. 1C; for the PCR strategies used to detect Igk germ-line transcripts in this study, see supporting information (SI) Fig. S1]. Cd43, Igll1, and Igkc are included as transcripts of known lineage and developmental stage specificity (Fig. 1C).

Fig. 1.

Developmental regulation of J-kappa germ-line promoter usage. (A) Schematic representation of the Igk locus showing the location of the variable (V), J and C regions, the Igk-J 5′ and 3′ promoters 5′k^o and 3′k^o with their spliced transcripts designated 1.1 and 0.8 kb. (B) Flow cytometry was used to isolate B220^lo CD43⁺ IgM⁻ pre-BI cells and large as well as small B220^lo CD43⁻ IgM⁻ pre-BII cells from bone marrow and B (B220^hi) and T cells (TCR⁺) from spleen. (C) The 5′k^o and 3′k^o promoter usage as assessed by real-time RT-PCR analysis of the cell populations isolated in B. The 1.1 and 0.8 kb transcripts are shown relative to the house keeping genes Yhwaz and Ubc (mean ± SE, n = 2). Cd43, Igll1, and Igk-C are included as transcripts of known lineage and developmental stage specificity.

The Kappa-GFP Model.

To provide a reporter for Igk germ-line transcription, Liang et al. (11) created kappa-GFP^+/− mice in which a GFP cDNA is inserted into the Jk locus. Their data indicated that GFP expression was developmentally regulated: although pre-BI cells lacked detectable GFP expression, ≈5% GFP⁺ pre-BII cells were reported, and the frequency of GFP⁺ cells increased during B cell differentiation to reach ≈50% in mature B cells, which is close to the frequency of kappa-GFP reporter alleles that remain in germ-line configuration in mature B cells (11). Based on the knowledge of developmentally regulated J-kappa germ-line promoter usage (Fig. 1), we decided to reappraise GFP expression in the kappa-GFP system.

We used multi parameter flow cytometry to determine GFP fluorescence during B cell development in kappa-GFP^+/− mice (courtesy of Mark Schlissel). We analyzed pre-BI cells (B220^lo CD43⁺ IgM⁻), small pre-BII cells (B220^lo CD43⁻ IgM⁻) and immature (B220⁺ IgM⁺ IgD⁻) bone marrow B cells, and transitional (B220^hi IgM⁺ IgD⁻) and mature (B220^hi) splenic B cells. Our flow cytometric analysis confirmed GFP expression by ≈8% of immature, 30% of transitional, and 50% of mature B cells (see GFP histogram overlays in Fig. 2A). However, in contrast to the original report (see ref. 11), we detected very few (<1%) GFP⁺ cells at the small pre-BII cell stage (Fig. 2A). To determine whether the lack of GFP fluorescence was due to the absence of GFP protein, we sorted preBI, small preBII GFP^pos mature B cells, and GFP^neg mature B cells from kappa-GFP^+/− mice and stained them with antibodies against GFP. GFP was readily detected in mature B cells, but not in pre-BI or small pre-BII cells (Fig. 2B). To allow immunoblot analysis of GFP protein, we first expanded pre-B cells by culture of bone marrow pre-B cells on ST2 stromal cells in the presence of IL7 (14). Extracts of wild-type and kappa-GFP^+/− pre-B cell cultures and kappa-GFP^+/− mature B cells were separated by protein gel electrophoresis, blotted, and probed with antibodies to GFP or histone H3 as a control for equal loading. No GFP signal was detected in pre-B cell lysates, even though serial dilutions of mature B cell lysates showed that the equivalent of 5% GFP⁺ cells was readily detectable (star symbol in Fig. 2C).

Fig. 2.

Developmental regulation of GFP expression in kappa-GFP^+/− reporter mice: the case of the missing GFP⁺ pre-BII cell population. (A) Bone marrow and spleen cells from kappa-GFP^+/− mice were analyzed by flow cytometry to distinguish pre-BI (BM: B220^lo CD43⁺ IgM⁻), small pre-BII (BM: B220^lo CD43⁻ IgM⁻ FCS^lo), immature B (BM: B220^lo IgM⁻ IgD⁻), transitional B (spleen: B220^hi IgM⁺ IgD⁻), and mature B cells (spleen: B220^hi IgM⁻ IgD⁺). Histogram overlays show GFP fluorescence (mean % ± SD, n = 3–6). (B) Sorted pre-BI, small pre-BII, and GFP⁺ and GFP⁻ mature B cells from kappa-GFP^+/− mice were stained with anti-GFP-Alexa Fluor 488 (green, omitted from the “staining control” to show the lack of GFP fluorescence at this exposure) and DAPI (blue). (C) Protein extracts of wild-type and kappa-GFP^+/− pre-B cell cultures and mature kappa-GFP^+/− splenic B cells were analyzed by Western blotting with anti-GFP or anti-histone H3 as a loading control. Extracts were serially diluted and a star indicates the equivalent of a 5% GFP⁺ population.

We remade kappa-GFP^+/− mice from kappa-GFP^+/− ES cells (kindly provided by Mark Schlissel) to address the possibility that the absence of GFP protein from small pre-BII cells was due to the progressive silencing of the targeted allele, or to changes in genetic background that could have occurred in the original kappa-GFP^+/− mouse line. We monitored GFP expression in founder mice and in 3 subsequent generations on different strain backgrounds chosen to reflect the origin of the ES cells (129SV), the backcross strain for the original kappa-GFP^+/− mouse line (C57BL/6), and the strain of Cre-transgenic mice used to delete the neomycin selection cassette from the original kappa-GFP^+/− mouse line (FVB/N; in our own experiments, the neomycin cassette was removed by in vitro transfection of ES cells with a Cre expression vector, see Fig. 3A). All generations and strain backgrounds recapitulated our findings described above for the original kappa-GFP^+/− line, namely a progressive increase in GFP⁺ expression from the immature B cell stage onwards and a lack of GFP⁺ small pre-BII cells (Fig. 3B). This analysis ruled out strain background or progressive transgene silencing as potential causes for the lack of GFP expression by small pre-BII cells. Because the 5% GFP expressing pre-BII cell population had provided the experimental basis for models in which allelic exclusion is explained by a finite probability of Igk locus activation (11), these data question the validity of the kappa-GFP^+/− reporter as evidence for stochastic models of allelic exclusion.

Fig. 3.

Kappa-GFP^+/− reporter mice remade: genetic background or generation-dependent transgene silencing do not account for the lack of GFP⁺ small pre-BII cells. (A) ES cells were transfected with pTurboCre to delete the floxed neomycin marker from IgκJ1. PCR of genomic DNA form wild-type, untransfected, and transfected kappa-GFP^+/− ES cells. (B) Flow cytometric analysis of the first 3 backcross generations of kappa-GFP^+/− mice on the indicated genetic backgrounds (mean percentage GFP⁺ ± SD, n = 3).

Igk Germ-Line Transcripts Are Abundant in Small Pre-BII Cells in the Absence of GFP Protein.

We next addressed whether the absence of GFP⁺ small pre-BII cells in kappa-GFP^+/− mice was because of the lack of Igk germ-line transcripts. Poly(A) RNA from wild-type and kappa-GFP^+/− pre-B cell cultures and kappa-GFP^+/− mature B cells was subjected to Northern blot analysis by using probes specific for GFP and for Hprt, which served to normalize the abundance of 4 distinct transcripts (labeled A to D in Fig. 4A) by densitometry. Transcript A was found in mature B cells but not in cultured pre-B cells, and transcript B was present at both developmental stages. Transcripts C and D were most abundant with 2-fold higher expression of transcript C in pre-B cells, and 6.8-fold higher expression of transcript D in mature B cells (Fig. 4A).

Fig. 4.

Pre-BII cells express high levels of Igk-J germ-line transcripts in the absence of GFP protein: origin, splicing, and coding potential. (A) Northern blot analysis of poly(A) RNA from wild-type (lane 1) and kappa-GFP^+/− (lane 2) pre-B cell cultures, and kappa-GFP^+/− mature B cells (lane 3). The abundance of 4 distinct GFP transcripts A to D was normalized to Hprt by densitometry. (B) RLM-RACE of poly(A) RNA from wild-type (lane 1) and kappa-GFP^+/− (lane 2) pre-B cell cultures, and kappa-GFP^+/− mature B cells (lane 3). Sequence data are shown in Table S1. The weak band seen in the wt (lane 1) was also sequenced and was not derived from Igk. Arrows indicate predicted ORFs (red) and PCR primers (black), a blue box indicates the RNA adaptor. (C) Expression of transcripts C and D 48 h after transfection into 293T cells. GFP fluorescence is shown versus DsRed to control for transfection efficiency (Upper). RT-PCR was used to verify the presence of transcripts C and D (Lower, wedges indicate a 3-fold increase in cDNA input between lanes). (D) RT-PCR of total GFP RNA and transcripts C and D in the indicated populations from kappa-GFP^+/− mice. Cd43, Igll1, and Igk-C transcripts are shown as controls (mean ± SE, n = 2). (E) GFP transcript abundance in cultured pre-B cells by real-time RT-PCR (Left, the scale shows RNA expression relative to pre-B cells) and GFP fluorescence of pre-B cells 48 h after transfection with GFP, transcript D, or with control vector (Right).

We used RNA linker-mediated rapid amplification of cDNA ends (RLM-RACE) to further characterize transcripts C and D. RLM-RACE confirmed the predominance of transcript C in pre-B cells and transcript D in mature B cells, and sequencing of the RLM-RACE products revealed that transcript C and a minor variant (C') originated from 5′k^o, whereas transcript D originated from 3′k^o (Fig. 4B; for sequences, see Table S1).

GFP Coding Potential of Igk Germ-Line Transcripts in Kappa-GFP^+/− Pre-B Cells and Mature B Cells.

Importantly, only transcript D, which originates from 3′k^o and is the predominant form in mature B cells, contained full-length GFP cDNA. Transcripts C and C', which originate from 5′k^o, lacked 58 nt from the 5′ end of the GFP coding sequence, apparently due to an aberrant splicing event (Table S1). Also, we found translation initiation and stop codons upstream of the GFP sequence in transcripts C and C', whereas transcript D contained a single ORF (red arrows in Fig. 4B). We cloned transcripts C and D into the mammalian expression vector pcDNAIII to directly test their coding potential. Transcript D, but not transcript C, conferred GFP fluorescence after transfection into 293T cells (Fig. 4C).

The developmental regulation of transcripts C and D was assessed by real-time RT-PCR analysis of sorted pre-BI cells (B220^lo CD43⁺ IgM⁻), small pre-BII cells (B220^lo CD43⁻ IgM⁻), and immature (B220⁺ IgM⁺ IgD⁻) bone marrow B cells, and transitional (B220^hi IgM⁺ IgD⁻) and mature (B220^hi) splenic B cells (Fig. 4D; for PCR strategies, see Fig. S1). Total GFP transcripts showed a broad developmental distribution. In contrast, transcript C peaked in large pre-BII cells, and transcript D was maximally expressed later at the immature B cell stage (Fig. 4D). Developmental stage-specific expression was validated by the levels of Cd43, Igll1, and Igk-C transcripts (Fig. 4D), and is consistent with our previous analysis of Igk germ-line transcripts in wild-type mice (Fig. 1). RT-PCR analysis may underestimate the differential expression of transcript D (6.8-fold more abundant in mature B cells than in pre-B cells by Northern blotting, see Fig. 4A) due to the presence of unspliced transcripts originating at 5′k^o (band B in Fig. 4A and data not shown). We conclude that the developmentally regulated usage of Igk germ-line promoters and the inability of transcripts originating from 5′k^o to encode GFP contribute to the lack of GFP⁺ small pre-BII cells in the kappa-GFP^+/− system.

Comparable Polysome Association of Igk Germ-Line Transcripts in Pre-B Cells and Mature Kappa-GFP^+/− B Cells.

To ask whether Igk germ-line transcripts are subject to differential regulation at the level of translation, we compared their association with polyribosomal complexes (polysomes) in pre-B and mature B cells (Fig. S2). Sucrose gradients were used to separate free RNA from RNA with increasing ribosomal association, and real-time RT-PCR was used to monitor the representation of RNA in each fraction (SI Methods). Messenger RNA for the housekeeping gene Hprt is efficiently translated and was predominant in the polysome fraction in both pre-B and mature B cells. Addition of EDTA dissociates polysomes and removed this peak. Compared with Hprt RNA, total GFP RNA showed a broad distribution throughout the gradient without a pronounced peak in the polysome fractions. The distribution of transcripts C and D was indistinguishable from total GFP RNA, and similar in pre-B and mature B cells. No further enrichment in the polysome fraction was recorded for fully spliced Igk germ-line transcripts. In contrast, productively rearranged V-J kappa RNA isolated from mature B cells was enriched in the polysomal fractions (Fig. S2). Hence, differential polysome association does not appear to account for differential GFP protein expression in pre-B and mature kappa-GFP^+/− B cells.

GFP Protein in Pre-B Cells Overexpressing 3′k^o Germ-Line Promoter Transcripts.

To test whether GFP expression in pre-B cells was limited by the amount of GFP-encoding Igk germ-line transcripts, we transfected cultured pre-B cells with pcDNAIII containing GFP, either with or without the 5′ UTR of transcript D. Quantification by real-time PCR showed that transfected pre-B cells contained ≈80-fold more transcript D than kappa-GFP^+/− pre-B cells (Fig. 4E Left). Transfected pre-B cells produced abundant GFP protein (Fig. 4E), regardless of the presence of the 5′ UTR or the SV40 intron, which forms part of the GFP transgene and appears to spliced less efficiently in pre-B cells than in mature B cells (data not shown; for PCR strategies, see Fig. S1). We conclude that, given sufficient expression levels, GFP-encoding transcripts originating from 3′k^o are translated in pre-B cells.

A role for Pre-BCR and BCR Signaling in Regulating Differential Igk-J Promoter Usage?

The transition from the pre-BI to the pre-BII stage is controlled by the pre-BCR, which initiates a proliferative phase that is terminated by the down-regulation of cell surface pre-BCR and the transcriptional silencing of Igll1 and Vpreb1. The termination of pre-BCR signaling and the initiation of Igk rearrangement require the adaptor protein SLP-65 (1–3). To test the role of pre-BCR signals in regulating Igk-J germ-line promoter usage, we used SLP-65 deficient pre-B cells that stably express a tamoxifen-inducible ERt2-SLP-65 fusion protein. Treatment with tamoxifen induced cell surface expression of Ig-kappa by an increasing proportion of cells over time (Fig. 5 A and B), which was further enhanced by reducing the concentration of IL7 in the culture medium (Fig. 5B). By using this system, we evaluated the abundance of transcripts emanating from the 5′k^o and 3′k^o Igk-J germ-line promoters. SLP-65 reconstitution and IL7 withdrawal increased the abundance of both transcripts by 1 to 2 orders of magnitude, but did not significantly change the ratio of transcripts from 5′k^o and 3′k^o (Fig. 5C).

Fig. 5.

Inducible pre-BCR signaling activates Igk rearrangement and both 5′k^o and 3′k^o Igk-J germ-line promoters. (A) SLP-65 deficient pre-B cells containing tamoxifen-inducible ERt2-SLP-65 were treated with tamoxifen (HT) in IL7^lo conditions and the time course of Ig-kappa cell surface expression was monitored by flow cytometry (black, Ig-kappa; gray, staining control; mean ± SE, n = 2). (B) ERt2-SLP-65 cells were treated with tamoxifen (HT, red bars) or carrier (control, black bars) in high (solid bars) or low concentrations of IL7 (hatched bars) and stained for Ig-kappa surface expression 72 h later (mean ± SD, n = 3). (C) ERt2-SLP-65 cells were treated as in B, and 1.1 and 0.8 kb Igk-J germ-line transcripts were quantitated by real-time RT-PCR (mean ± SD, n = 3).

Immature B cells undergo BCR editing by initiating secondary Ig rearrangements in response to BCR stimulation (15). To explore a potential role of Igk germ-line promoters during BCR editing, we isolated immature (B220^lo) B cells transgenic for the H2K^b-specific 3–83 BCR (Fig. 6A). As expected (15), immature B cells down-regulated surface BCR expression when exposed to fibroblasts expressing the stimulatory H2K^b antigen, but not in response to fibroblasts of the neutral H2^d haplotype (Fig. 6A). Real-time RT-PCR analysis demonstrated that Rag1 and Rag2 transcripts were up-regulated as an indication of receptor editing in response to BCR activation of immature B cells, but this did not significantly alter the relative abundance of 1.1 and 0.8 kb Igk-J germ-line transcripts (Fig. 6B).

Fig. 6.

Igk-J germ-line promoter usage during receptor editing in response to BCR stimulation of immature B cells. (A) Immature (B220^lo) B cells transgenic for the H2K^b-specific 3–83 BCR were sorted (Left) and cultured on H2^d (neutral) or H2^b (stimulatory) fibroblasts. Surface IgM expression was analyzed by flow cytometry 24 h later (Right). (B) 3–83 BCR transgenic immature B cells were cultured as in A, and analyzed by real-time RT-PCR. Rag1 and Rag2 transcripts were up-regulated, but Igk-J germ-line transcripts were largely unchanged in response to activation with H2^b (mean ± SE, n = 2).

Discussion

Our data show that the alternative usage of Igk-J germ-line promoters is developmentally regulated. The lack of synchrony of B cell development in vivo makes it difficult to tease out the physiological stimuli that determine alternative promoter usage. Therefore, we have used model systems to recapitulate defined steps of development, namely the progression from the large to the small pre-BII cell stage (by reconsituting SLP-65 signaling) and BCR signaling in immature B cells (by providing stimulatory ligands to BCR transgenic cells). Although these experimental systems have not provided clear leads toward understanding the regulation of differential Igk-J promoter usage, they do have limitations as models for B cell development. It is also possible that we have not found all relevant transcripts, and for that reason underestimate the role of Igk-J germ-line promoters in pre-BCR or BCR signaling, editing, or other biological events. Genetic dissection of Igk-J germ-line promoters has revealed impaired Igk rearrangement, but no specific involvement of either promoter. A deletion encompassing 5′k^o and much of the sequence between 5′k^o and 3′k^o impaired Igk rearrangement (16). Similarly, mutation of the KI and KII regions just upstream of 3′k^o impaired Igk rearrangement but had no apparent effect on the 0.8 kb transcript, suggesting KI and KII are not critical for regulating the 3′k^o promoter (17).

The developmental stage-specific activity of Igk-J germ-line promoters correlates with differential GFP protein expression in the kappa-GFP^+/− model. Importantly, Igk-J germ-line transcripts originating from the 5′k^o promoter lack GFP coding potential. These data explain at least in part the increase in GFP expression during B cell development in kappa-GFP^+/− mice (11). Due to the distance between 5′k^o and the downstream J segments, transcripts originating from 5′k^o may not be equivalent to 3′k^o transcripts in directing posttranslational histone modifications conducive for Rag-mediated rearrangement. In this sense, GFP-encoding 3′k^o transcripts might unwittingly predict Igk rearrangement. However, this argument does not account for the discrepancy that Liang et al. (11) reported GFP fluorescence in 5% of small pre-BII cells, whereas we have no evidence for the existence of GFP⁺ small pre-BII cell population either in the original kappa-GFP mouse line or in mouse lines that we have newly generated from the original ES cells and analyzed over subsequent generations on 3 different genetic backgrounds. Whereas the previously postulated 5% probability of Igk locus activation (11) might have been compatible with B cell development and the existence of mature B cells with 2 productive Igk rearrangements, the diminutive frequency of actual GFP⁺ pre-B cells argues that GFP expression is not an indicator of Igk locus activation in the kappa-GFP model. The dissociation between active Igk germ-line transcription and a lack of GFP protein in small pre-BII cells is important because it calls into question the original interpretation (11) that individual Igk alleles have a limited likelihood of activation, which had formed a cornerstone of probabilistic models of allelic exclusion.

Materials and Methods

Cells and Tissue Culture.

Primary pre-B cells were established from mouse bone marrow in accordance with the Animals (Scientific Procedures) Act under Project Licences granted by the Home Office, United Kingdom, on irradiated ST2 stromal cells in IMDM, 10% FCS, 1% penicillin-streptomycin, 2 μM 2-mercaptoethanol and 5 ng/mL IL7 (“high”) or 0.1 ng of IL7 (“low”). Splenic B cells were isolated by flow cytometry, activated with anti CD40, IL4 and anti-Igkappa for 24h before RNA isolation (see below); 293T cells or Bcl2 transgenic pre-B cell cultures were transfected (Amaxa) with 20 μg of empty pcDNAIII vector or pcDNAIII containing the indicated transcripts under CMV promoter control. SLP-65 deficient pre-B cells containing a fusion protein of SLP-65 and the estrogen receptor hormone binding domain (ERt2-SLP-65) were the kind gift of Hassan Jumaa (3).

Immunofluorescence Staining, Microscopy, and Flow Cytometry.

B220-PE, CD43-FITC, Igkappa-biotin (PharMingen), IgD-biotin (Southern Biotech), and IgM-Cy5 (Caltag) were used for surface staining. Cells were analyzed and sorted by using FACS instruments (Becton Dickinson). Immunofluorescence staining and microscopy were as described (2), by using anti-GFP (Clontech) and goat anti-rabbit IgG-Alexa Fluor 488 (Molecular Probes).

Real-Time RT-PCR.

Total RNA was isolated by using RNAbee (Tel-Test) and reverse transcribed. PCR reactions included 2x SYBR PCR Master Mix (Qiagen), 300 nM primers, and 2 μL of cDNA as a template in 50 μL of reaction volume. Cycle conditions were 94 ^οC 8 min, 40 cycles of 94 ^οC 30 sec, 55 ^οC 30 sec, 72 ^οC 1 min, and plate read. All primers amplified specific cDNAs with >95% efficiency. Data were normalized to the housekeeping genes Ywhaz and Ube (2). Primer sequences were (5′ to 3′):

• Ywhaz fw CGTTGTAGGAGCCCGTAGGTCAT rev TCTGGTTGCGAAGCATTGGG

• Ube fw AGGAGGCTGATGAAGGAGCTTGA rev TGGTTTGAATGGATACTCTGCTGGA

• Igll fw GGACTTGAGGGTCAATGAAGCTC rev GTGGGATGATCTGGAACAGGAG

• Cd43 fw GCTCCAAGTACCTCTGAAGCCC rev CCAGCAGAAGTCTCCAAAGAAGAC.

Additional primer sequences are listed in Fig. S1.

Northern Blotting.

PolyA⁺ RNA was enriched by using Oligotex (Qiagen), denatured in glyoxal buffer, separated on nondenaturing agarose gels, blotted (Hybond N+, Amersham Biosciences), and prehybridized (UltraHyb, Ambion). In vitro translated (MAXIscript, Ambion), purified riboprobes were hybridized for 16 h at 65 °C, visualized by PhosphorImager, and analyzed by using ImageQuant (GE Life Sciences).

RLM-RACE.

RNA was reverse transcribed with random decamers. Products amplified with the FirstChoice RLM-RACE Kit (Ambion) were gel purified, cloned into pCRII by using a TA cloning kit (Invitrogen), and transformed into competent cells with 100 mM IPTG and 40 mg/mL X-gal. Single white colonies were grown with 100 μM ampicillin for 2 h at 37 °C, 1 μL was PCR amplified (M13 primers) and sequenced (T7 and SP6 primers).

Immunoblotting.

Immunoblotting was done by using rabbit anti-histone H3 (Abcam), rabbit anti-GFP (Clontech), and donkey anti-rabbit IgG-HPR (Amersham Biosciences).