Abstract
The transcription factor Pax5 is essential for B cell commitment and development. Although the detailed Pax5 expression pattern within the hemopoietic system is still largely unknown, we previously reported that Pax5 is monoallelically transcribed in pro-B and mature B cells. In this study, we have investigated the expression of Pax5 at single-cell resolution by inserting a GFP or human Cd2 indicator gene under the translational control of an internal ribosomal entry sequence into the 3′ untranslated region of Pax5. These insertions were noninvasive, as B cell development was normal in Pax5ihCd2/ihCd2 and Pax5iGFP/iGFP mice. Transheterozygous Pax5ihCd2/iGFP mice coexpressed GFP and human CD2 at similar levels from pro-B to mature B cells, thus demonstrating biallelic expression of Pax5 at all stages of B cell development. No reporter gene expression could be detected in plasma cells and non-B cells of the hemopoietic system. Moreover, the vast majority of common lymphoid progenitors and pre-pro-B cells in the bone marrow of Pax5iGFP/iGFP mice did not yet express GFP, indicating that Pax5 expression is fully switched on only during the transition from uncommitted pre-pro-B cells to committed pro-B cells. Hence, the transcriptional initiation and B cell-specific expression of Pax5 is entirely consistent with its B cell lineage commitment function.
Hemopoietic stem cells give rise to B lymphocytes by first differentiating to lymphoid-primed multipotent progenitors (MPP)4 (1), which subsequently develop to common lymphoid progenitors (CLP) with their characteristic B, T, and NK cell potential (2). The CLP enters the B lymphoid lineage by differentiating via the pre-pro-B cell stage to early pro-B cells, which undergo B cell lineage commitment followed by development along the B cell pathway (3, 4, 5).
The entry of lymphoid progenitors into the B cell lineage and the subsequent developmental progression to terminally differentiated plasma cells is controlled by a multitude of transcriptional regulators (3, 6). Pax5 (BSAP) is one of the earliest expressed transcription factors that is essential for B cell development, as its mutation arrests B lymphopoiesis at an early progenitor B cell stage in the bone marrow (7, 8). Pax5−/− pro-B cells possess an extensive self-renewal and broad developmental potential (9, 10, 11), correspond to lymphoid progenitors with a latent myeloid potential (12) and, in this regard, resemble the uncommitted pre-pro-B cells of wild-type mice (4, 5). The restoration of Pax5 expression suppresses the multilineage potential of Pax5−/− pro-B cells, while simultaneously promoting their development to mature B cells (9). Conversely, the loss of Pax5 expression by conditional gene inactivation converts committed pro-B cells with a restricted B lymphoid potential into lymphoid progenitors with a broad developmental potential (13). Together, these data identified Pax5 as the critical B cell commitment factor. Pax5 expression is continuously required to maintain B cell lineage commitment even at later stages of B cell development, as conditional Pax5 deletion leads to loss of the identity and function of mature B cells (14, 15). At the molecular level, Pax5 fulfills its commitment function by repressing B cell lineage-inappropriate genes and activating B cell-specific genes, which results in the shutdown of inappropriate signaling systems and the simultaneous promotion of B lymphoid signal transduction (15, 16, 17).
The expression of Pax5 was initially studied in cultured cell lines representing different stages of B cell development, indicating that Pax5 is expressed from pro-B to mature B cells, but not in plasmacytoma and other hemopoietic cell lines (18, 19). Given the essential role of Pax5 in B cell lineage commitment, the question arises when and how Pax5 transcription is initiated at the onset of B cell development. RT-PCR analyses of Pax5 expression in FACS-sorted progenitors have generated conflicting results. In two studies, the Pax5 gene is shown to be expressed in pre-pro-B cells already at ∼40% (5) or close to 100% (20) of the level observed in committed pro-B cells. Other reports identified even the earlier CLP cell as the developmental stage where Pax5 transcription is efficiently initiated (21, 22). At face value, these results are incompatible with the B cell lineage commitment function of Pax5, as both the CLP and pre-pro-B cells are uncommitted progenitors despite their apparent Pax5 expression (2, 4, 5, 20). To study how Pax5 transcription is initiated in early B cell development, we previously took advantage of the fact that expression of the two Pax5 alleles can be individually monitored in B lymphocytes of heterozygous Pax5lacZ/+ mice carrying a lacZ reporter gene insertion in the mutant Pax5 allele (7, 23). Flow cytometric analyses revealed that Pax5 is predominantly expressed from only one allele in pro-B and mature B cells, whereas it switches to a biallelic transcription mode in pre-B and immature B cells (23, 24). As we could confirm these results in genetically unmanipulated B lymphocytes by single-cell RT-PCR and RNA-FISH analyses, we proposed that Pax5 is initiated in a stochastic manner from only one allele at the onset of B cell development (23, 24). Our hypothesis of stage-specific monoallelic expression of Pax5 was subsequently challenged by Rhoades et al. (25), who provided evidence for biallelic Pax5 expression in unfractionated (immature and mature) B cells of the spleen by single-cell RT-PCR analysis.
To study the regulation of Pax5 at single-cell resolution in vivo, we inserted a GFP or human Cd2 indicator gene under the translational control of an internal ribosomal entry site (IRES) element into the 3′ untranslated region of the endogenous Pax5 gene. B cell development was normal in Pax5ihCd2/ihCd2 and Pax5iGFP/iGFP mice, indicating that the two noninvasive reporter gene insertions can be used to monitor the expression of individual Pax5 alleles. Transheterozygous Pax5ihCd2/iGFP mice coexpressed GFP and human CD2 at similar levels from committed pro-B cells in the bone marrow to mature B cells in the spleen, thus demonstrating biallelic expression of Pax5 at all stages of B cell development. No reporter gene expression could, however, be detected in plasma cells and non-B cells of the hemopoietic system. Pax5 is therefore exclusively expressed within the B lymphoid lineage and is repressed during terminal plasma cell differentiation. Importantly, the vast majority of CLP and pre-pro-B cells in Pax5iGFP/iGFP mice did not express GFP, whereas a minor fraction of these uncommitted progenitors started to up-regulate GFP expression. These data demonstrate that the onset and B cell specificity of Pax5 expression are entirely consistent with the B cell lineage commitment function of Pax5.
Materials and Methods
Generation of Pax5 knock-in mice
The Pax5ihCd2 and Pax5iGFP targeting vectors were assembled in a pSP64 plasmid containing a polylinker with appropriate loxP and restriction sites. A 1.6-kb XhoI-SalI PCR fragment and a 4.5-kb BamHI fragment of the Pax5 gene were inserted as homology arms together with a floxed puromycin resistance gene (positive selection) and an IRES linked to either an intracellularly truncated human Cd2 gene (hCd2t) or to a codon-improved (enhanced) GFP gene. The HSV-tk and DT-A genes (negative selection) were inserted upstream of the short homology arm. FspI-linearized DNA (15 μg) was electroporated into HM-1 ES cells (1 × 107) followed by selection with 2.5 μg/ml puromycin and 2 μM gancyclovir. PCR-positive clones were verified by Southern blot analysis of BamHI-digested DNA before injection into C57BL/6 blastocysts and the generation of Pax5+/puro-ihCd2 and Pax5+/puro-iGFP mice, respectively. The floxed puromycin gene was deleted in the germline by crossing Pax5+/puro-ihCd2 and Pax5+/puro-iGFP mice with the Mox2-cre line (26) to generate Pax5+/ihCd2 and Pax5+/iGFP mice, respectively. The Pax5ihCd2 and Pax5iGFP alleles were backcrossed for seven generations or more into the C57BL/6 genetic background. The Pax5ihCd2 and Pax5iGFP alleles were genotyped by PCR amplification of a 327- and 448-bp DNA fragment, respectively. These fragments were amplified from the human Cd2 and GFP gene with the forward primers 5′-GAGACAAGAGCCCACAGAGTA-3′ and 5′-CCGACCACTACCAGCAGAACA-3′, respectively, and the common reverse primer 5′-ACAGAGGAGCAAGACGGTATT-3′ in Pax5 exon 10.
Abs and flow cytometry
The following biotin-, PE-, CyChrome-, allophycocyanin-, PE Cy7-, or allophycocyanin Cy7-coupled Abs were used for flow cytometry: anti-B220 (RA3-6B2), CD4 (GK1.5), CD5 (53-7.3), CD8α (53-6.7), CD11b/Mac1 (M1/70), CD11c (HL3), CD19 (1D3), CD21 (7G6), CD23 (B3B4), CD44 (1M7), CD117/c-Kit (ACK4), F4/80, Gr1 (RB6-8C5), IgD (1.19), IgM (M41.42), Ly6C (6C3), Pan-NK (DX5), Sca1/Ly6A (D7), TCR-β (H57-597), and Ter119 Abs from BD Pharmingen and anti-CD25/IL-2Rα (PC61), CD127/IL-7Rα (A7R34), and human CD2 (RPA-2.10) Abs from eBioscience. Unspecific Ab binding was suppressed by preincubation with CD16/CD32 Fc block solution (BD Pharmingen). Single-cell suspensions were stained with the respective Abs and analyzed on a Canto flow cytometer (BD Biosciences). A wide forward and side light scatter gate was used for analysis of the different hemopoietic tissues. Bone marrow cells and thymocytes were stained with the following PE-coupled lineage marker Abs for identifying MPPs (LSK), CLPs, and double-negative (DN)1/2 cells (anti-B220, CD4, CD8α, CD11b/Mac1, CD11c, CD19, Gr1, Ly6C, Pan-NK, TCR-β, and Ter119) and pre-pro-B cells (same Ab mixture but lacking anti-B220 and CD4). Subsequently, the cells were washed once, incubated with anti-PE beads (Miltenyi Biotec), and lineage-depleted by MACS sorting (Miltenyi Biotec). The progenitor-enriched Lin− cells were then stained with the corresponding Abs for identifying the MPPs (LSK) as Lin−Sca1highc-KithighIL-7Rα− cells, the CLP as Lin−Sca1lowc-KitlowIL-7Rα+ cells, and the pre-pro-B cells as Lin−c-KitlowB220low cells. DN1 and DN2 thymocytes were identified as Lin−c-KithighCD44highCD25− and Lin−c-KithighCD44highCD25+ cells, respectively. Plasma cells were defined as CD138highCD28+Lin− (CD4−CD8α−F4/80−CD21−) cells in the bone marrow and spleen as described (15). For isolating B-1a cells, 6-wk-old mice were i.p. injected with 10 ml of prewarmed RPMI 1640 medium. After massaging the filled peritoneum, the B-1a cell suspension was aspirated, washed, and stained with the corresponding Abs.
Immunization with sheep RBC
Sheep RBC were washed in PBS and resuspended at 109 cells/ml followed by i.p. injection of 150 μl into an adult mouse.
Results
Generation of two Pax5 alleles with distinct reporter gene insertions
The expression of Pax5 was so far determined in cultured B cell lines and sorted lymphocyte fractions with methods that allowed quantification of Pax5 expression only at the cell population level (5, 18, 19, 20). To be able to measure allele-specific regulation of Pax5 at single-cell resolution, we generated two Pax5 knock-in alleles expressing each a distinct reporter gene from the Pax5 promoters. To this end, we used homologous recombination in ES cells to insert either a GFP or intracellularly truncated human Cd2 indicator gene under the translational control of an IRES element into the 3′ untranslated region (exon 10) of Pax5 (Fig. 1,A). Blastocyst injection of correctly targeted ES cells resulted in Pax5+/puro-ihCd2 and Pax5+/puro-iGFP mice, which were crossed with the Mox2-cre line (26) to achieve germline deletion of the puromycin resistance gene (Fig. 1,B). Heterozygous Pax5+/ihCd2 and Pax5+/GFP mice were intercrossed to obtain homozygous Pax5ihCd2/ihCd2 and Pax5iGFP/iGFP mice (Fig. 1,B), which were born at Mendelian frequency and showed no apparent abnormalities. Importantly, flow cytometric analysis revealed similar numbers of B220+CD19+ B cells in the bone marrow and spleen of Pax5ihCd2/ihCd2 and Pax5iGFP/iGFP mice compared with wild-type control mice (Fig. 1 C). We conclude therefore that both reporter gene insertions are noninvasive, as they do not interfere with normal expression and function of Pax5.
Similar expression of both Pax5 alleles throughout B cell development
To monitor the expression of each Pax5 allele, we analyzed the bone marrow of transheterozygous Pax5ihCd2/iGFP mice, which express human CD2 from one and GFP from the other Pax5 allele. We used the pan-B cell marker CD19 to separate bone marrow cells into non-B cells (CD19−) and B lymphocytes (CD19+) by flow cytometric analysis and then displayed human CD2 and GFP expression for both cell populations. As shown in Fig. 2,A, the GFP and human CD2 proteins were coexpressed in B lymphocytes, but not in non-B cells, which demonstrates exclusive expression of Pax5 within the B cell lineage of the hemopoietic system. We next investigated Pax5 expression during B lymphopoiesis by monitoring GFP and human CD2 expression at different B cell developmental stages (Fig. 2,B). Both Pax5 alleles were coexpressed in pro-B (CD19+c-Kit+), pre-B (CD19+CD25+IgM−), immature B (B220lowIgM+), and recirculating B (B220+IgDhigh) cells of the bone marrow. Flow cytometric analysis of Pax5ihCd2/iGFP splenocytes also revealed coexpression of GFP and human CD2 in immature (IgMhighIgDlow) and mature (IgMlowIgDhigh) B cells as well as in transitional 1 (B220+CD21−CD23−), transitional 2 (B220+CD21highCD23high), marginal zone (B220+CD21highCD23low/−), and follicular (B220+CD21intCD23high) B cells (Fig. 3,A). Likewise, all B-1a cells (IgM+CD5+Mac1+) in the peritoneum of Pax5ihCd2/iGFP mice coexpressed the human CD2 and GFP indicator proteins (Fig. 3,B). Direct comparison of GFP expression revealed similar levels of Pax5 expression at all stages of B lymphopoiesis except for marginal zone B cells, which expressed Pax5 at a ∼2-fold higher level (Figs. 2 C and 3C). Together these data unequivocally demonstrate that Pax5 is expressed within the B lymphoid lineage at similar levels and in a biallelic manner from the pro-B to the mature B cell stage.
Loss of Pax5 expression during terminal plasma cell differentiation
We next studied Pax5 expression in late B lymphopoiesis by analyzing germinal center B cells, which undergo Ig class switch recombination and somatic hypermutation before differentiation into memory B or plasma cells (27). Eleven days after immunization with sheep RBC, the germinal center B cells (B220+CD19+PNAhigh) of homozygous Pax5iGFP/iGFP mice expressed similar GFP levels (Fig. 4,A) as the mature B cells in the spleen (Fig. 3,A). The terminal differentiation of germinal center B cells generates short-lived plasma cells in the spleen as well as long-lived plasma cells that home to the bone marrow (27). We have recently described the unambiguous identification of plasma cells as CD138highCD28+Lin− (CD4−CD8α−F4/80−CD21−) cells by flow cytometric analysis (15). The long-lived plasma cells in the bone marrow of immunized Pax5iGFP/iGFP mice failed to express GFP, whereas the short-lived plasma cells in the spleen retained a very low GFP level, possibly due to the long half-life of the GFP protein (Fig. 4 B). The loss of Pax5 expression is also reflected by the down-regulation of CD19 on plasma cells (28), consistent with the fact that Pax5 is absolutely required for transcription of its target gene Cd19 (16). In summary, these results indicate that Pax5 expression is lost during the development of germinal center B cells to terminally differentiated plasma cells.
Initiation of Pax5 expression in lymphoid progenitors
Turning our attention to early lymphopoiesis, we were particularly interested to see whether the initiation of Pax5 expression would be consistent with the previously described B cell lineage commitment function of Pax5 (9, 10). To determine Pax5 expression in early hemopoietic progenitors, we first eliminated lineage-positive cells from the bone marrow of homozygous Pax5iGFP/iGFP and wild-type control mice followed by staining with appropriate Abs to identify the different progenitor cell populations (Fig. 5,A). No GFP expression could be detected in early MPPs (LSK; Lin−IL-7Rα−c-KithighSca1high), which also include the hemopoietic stem cells of Pax5iGFP/iGFP mice (Fig. 5,A). The majority (93.5%) of CLPs (Lin−IL-7Rα+c-KitlowSca1lowB220−CD19−) and most (83.7%) of the pre-pro-B cells (Lin−c-Kit+B220+CD19−) did not yet express GFP. A small percentage (3.5%) of CLPs and a slightly larger fraction (7.8%) of pre-pro-B cells expressed GFP at a low level, indicating that these cells started to up-regulate Pax5 expression (Fig. 5,A). Similar GFP expression levels as found in committed pro-B cells were observed only in 3% and 8.5% of CLPs and pre-pro-B cells, respectively (Fig. 5,A). In marked contrast, all pro-B cells (c-Kit+B220+CD19+) homogeneously expressed GFP both in Pax5iGFP/iGFP (Fig. 5,A) and transheterozygous Pax5ihCd2/iGFP mice (Fig. 2 B). CLPs and pre-pro-B cells (also referred to as CLP2 or EPLM) are known to have a broad lymphoid and latent myeloid developmental potential in contrast to the B cell lineage-committed CD19+ pro-B cells (2, 4, 5, 20). As GFP expression of the tagged Pax5 allele is fully switched on only in committed pro-B cells, we conclude that the initiation of Pax5 expression is entirely consistent with the B cell lineage commitment function of Pax5.
In addition to the lymphoid progenitors in bone marrow, the earliest CD4−CD8− DN progenitors (DN1) in the thymus are also uncommitted and retain some B cell developmental potential (29, 30). No GFP expression could, however, be observed in DN1 (Lin−c-KithighCD44highCD25−) or DN2 (Lin−c-KithighCD44highCD25+) prothymocytes of Pax5iGFP/iGFP mice (Fig. 5 B). Moreover, a subset of plasmacytoid dendritic cells was shown to exhibit low-level expression of lymphoid genes including Pax5 (31). However, plasmacytoid dendritic cells (CD19−B220+Ly6C+CD11c+) in the bone marrow of Pax5iGFP/iGFP mice failed to express detectable levels of GFP, indicating that these cells do not transcribe significant amounts of Pax5 mRNA (data not shown). Hence, the absence of Pax5 expression in uncommitted progenitors and hemopoietic cell types other than B lymphocytes is fully compatible with the known role of Pax5 in controlling the B cell identity.
Discussion
The transcription factor Pax5 is essential for B cell commitment and subsequent development to the mature B cell stage (9, 10, 14). In this study we have analyzed the hemopoietic expression of Pax5 by monitoring the transcription of individual Pax5 alleles, which was achieved by noninvasive insertion of different reporter genes into the two Pax5 alleles. The Pax5 gene was shown to be exclusively expressed within the B lymphoid lineage of the hemopoietic system from the committed pro-B to the mature B cell stage. Both Pax5 alleles were similarly expressed at all stages of B cell development, demonstrating that Pax5 is biallelically transcribed in contrast to our previous report suggesting stage-specific monoallelic expression of Pax5 (23).
Genes coding for receptors, which convey clonal specificity to a cell, are known to be subject to allele-specific regulation in addition to imprinted and X-chromosomal genes. This class of monoallelically expressed gene codes for Ag receptors (32), NK cell receptors (33), odorant receptors (34), and cadherin-related neuronal receptors (35). Allele-specific regulation has also been described for certain cytokine genes. For instance, the closely linked Il4 and Il13 genes are monoallelically transcribed in CD4+ Th2 cells, which was demonstrated by three independent methods. First, single nucleotide polymorphisms were used to discriminate, by RT-PCR analysis, the allelic origin of Il4 and Il13 mRNA in CD4+ T cell clones generated from BALB/c × CAST/Ei F1 hybrid mice (36, 37). Second, nascent transcripts at one or both alleles of the same cytokine genes were visualized by RNA-FISH in nuclei of Th2 cells (38). Finally, FACS analysis was used to monitor the expression of the two Il4 alleles in Th2 cells of mice that carried an inactivating insertion of either a human Cd2 or GFP reporter gene in one of the two Il4 alleles (39, 40). All three methods produced concordant results for the Il4 gene, thus unequivocally demonstrating that this gene is subject to stochastic allele-specific regulation. In contrast, FISH and single-cell RT-PCR analyses revealed tight monoallelic regulation of the Il2 gene in mature CD4+ T cells (41), whereas the majority of CD4+ T cells expressed both Il2 alleles in mice containing an Il2GFP knock-in allele (42). These discrepancies question whether negative results obtained by FISH and single-cell RT-PCR analyses always reflect the failure of an allele to be expressed or rather result from the limitation of these methods in reliably detecting the expression of both alleles, as discussed by Rhoades et al. (25).
Based on single-cell RT-PCR and RNA-FISH analyses of genetically unmodified B cells as well as FACS analysis of Pax5lacZ/+ B lymphocytes, we previously concluded that the Pax5 gene is monoallelically expressed in pro-B and mature B cells during early and late B cell development (23, 24). In an attempt to verify this hypothesis with a more reliable method for detecting allele-specific gene transcription, we tagged the two Pax5 alleles with IRES-hCd2 or IRES-GFP insertions. Analysis of transheterozygous Pax5ihCd2/iGFP mice revealed that Pax5 is biallelically expressed throughout B cell development. Our new data thus confirm and significantly extend the single-cell RT-PCR results of Rhoades et al. (25), who demonstrated biallelic Pax5 expression only in unfractioned splenic B cells. Hence, the negative data of our published single-cell RT-PCR and RNA-FISH analyses have misled us despite careful control experiments (23, 24). We analyzed Pax5 expression by RT-PCR only in those B cells displaying biallelic expression of the equally rare Cd19 mRNA, and the transcription factor gene Ikaros revealed biallelic expression under the same RNA-FISH conditions used for the Pax5 gene (23, 24). Our recent identification of a potent B cell-specific enhancer in intron 5 of Pax5 provides a likely explanation why the lacZ gene is infrequently transcribed from the endogenous Pax5 promoters in pro-B and mature B cells of heterozygous Pax5lacZ/+ mice (23). This Pax5 enhancer may efficiently interact and thus specifically service the promoter of the neomycin resistance gene in intron 2 of the targeted Pax5 allele (7), as this promoter is located more closely to the enhancer than the endogenous Pax5 promoters, thus resulting in a transgenic artifact. Interestingly, five of the nine mammalian Pax genes are associated with mouse developmental mutant and human disease syndromes, as heterozygous loss-of-function mutations in these genes lead to haploinsufficient phenotypes (43). The Pax6 gene, which is associated with aniridia in humans and Small eye (Sey) in mice, was shown to be biallelically expressed during lens placode development, thus ruling out the possibility that monoallelic Pax6 expression in this tissue causes the haploinsufficient eye phenotype of Sey mice (44).
The B cell-specific expression of Pax5 from the pro-B to the mature B cell stage is entirely consistent with the B cell identity function of Pax5, as conditional Pax5 inactivation in pro-B or mature B cells results in loss of the B cell phenotype and retrodifferentiation to uncommitted lymphoid progenitors (13, 14) (C. Cobaleda and M. Busslinger, unpublished data). Moreover, the loss of Pax5 expression during plasma cell differentiation is in agreement with the fact that Pax5-repressed, B cell lineage-inappropriate genes are reactivated and Pax5-activated, B cell-specific genes are down-regulated in plasma cells (15) (A. Schebesta and M. Busslinger, unpublished data). Finally, the initiation of Pax5 expression at the onset of B cell development fully supports a critical role for Pax5 in B cell lineage commitment. The Pax5 gene is initiated only in a small subset of the uncommitted CLP and pre-pro-B cell progenitors, whereas it is highly expressed in all committed pro-B cells at the next B cell developmental stage. Interestingly, the Pax5 mutation arrests B cell development in the bone marrow at an early pro-B cell stage that resembles the pre-pro-B cells (also known as CLP2 or EPLM) of wild-type mice (4, 5, 20) in several aspects including their strong lymphoid and latent myeloid potential (9, 10, 12). Hence, the Pax5 mutation blocks adult B lymphopoiesis precisely at the developmental transition, when expression of the Pax5 gene is fully switched on to give rise to committed pro-B cells.
Acknowledgment
We thank C. Theussl for blastocyst injection.
Disclosures
The authors have no financial conflict of interest.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by Boehringer Ingelheim and the Austrian Industrial Research Promotion Fund.
Abbreviations used in this paper: MPP, multipotent progenitor; CLP, common lymphoid progenitor; IRES, internal ribosomal entry site; DN, double negative; LSK, Lin−Sca1highc-Kithigh progenitor.