Reversion of B Cell Commitment upon Loss of Pax5 Expression

See allHide authors and affiliations

Science  05 Jul 2002:
Vol. 297, Issue 5578, pp. 110-113
DOI: 10.1126/science.1067518


The transcription factor Pax5 is essential for initiating B cell lineage commitment, but its role in maintaining commitment is unknown. Using conditional Pax5 inactivation in committed pro-B cells, we demonstrate that Pax5 is required not only to initiate its B lymphoid transcription program, but also to maintain it in early B cell development. As a consequence of Pax5 inactivation, previously committed pro-B cells regained the capacity to differentiate into macrophages in vitro and to reconstitute T cell development in vivo in RAG2 −/− mice. Hence, Pax5 expression is continuously required to maintain B cell lineage commitment, because its loss converts committed pro-B cells into hematopoietic progenitors with multilineage potential.

The hematopoietic stem cell gives rise to all lymphoid lineages by differentiating through the intermediary stage of the common lymphoid progenitor (1). Entry of this progenitor into the B cell pathway depends on the transcription factors E2A, EBF, and Pax5 (BSAP), which control two separate transcriptional programs (2). The regulators E2A and EBF coordinately activate the expression of B cell–specific genes (3) and are essential for the generation of the earliest B cell progenitors (pro-B cells) (4,5). However, mere activation of the B lymphoid expression program is not sufficient for commitment to the B cell lineage, because Pax5-deficient pro-B cells still retain the broad developmental potential of early hematopoietic progenitors (6,7) despite normal expression of E2A and EBF (8). Instead, Pax5 restricts the multilineage potential of early progenitors to the B cell pathway by repressing lineage-inappropriate genes and further activating B cell–specific genes (6, 7).

A fundamental aspect of organogenesis concerns whether the commitment of tissue-specific stem cells to different developmental pathways is a reversible or irreversible process. This question has recently been investigated for the neural crest stem cell that generates neurons or glial cells during peripheral nervous system development. In this case, transient Notch activation was sufficient to irreversibly commit neural crest stem cells to the glial cell fate (9). Similarly, we were interested in determining whether B cell lineage commitment requires only the transient activation or continuous expression of Pax5 during early B cell development. To address this issue, we took advantage of the Cre-loxP system for conditional inactivation of a floxed (F)Pax5F allele carrying loxP sequences on either side of the paired domain exon 2 (10).

B cell development is normal inPax5F/F mice, whereasPax5 Δ/Δ mice, generated by germ line deletion (Δ) of exon 2 (10), are phenotypically indistinguishable from Pax5 −/− mice (11). For conditional Pax5 inactivation in pro-B cells, we crossed Pax5F/F mice with the transgenic line CreED-30, which expresses the Cre-EBD(G521R) fusion protein from the simian virus 40 promoter and Eμ enhancer in B lymphocytes (12). The Cre-EBD(G521R) protein consists of the Cre recombinase linked to a mutant estrogen receptor ligand-binding domain (EBD containing a glycine to arginine substitution at position 521) that is responsive to the synthetic antagonist 4-hydroxytamoxifen (OHT) instead of β-estradiol. TheCreED-30 transgene therefore facilitates OHT-mediated induction of Cre activity in B lymphocytes (12). Pro-B cells were sorted from the bone marrow of Pax5F/FCreED-30 mice and cultured in vitro on stromal cells in the presence of interleukin-7 (IL-7), which promotes the expansion of pro-B cells while preventing their further differentiation (13). Flow cytometric analysis revealed normal expression of the B cell markers B220 and CD19 on these pro-B cells (Fig. 1A, F/F). Because CD19transcription is strictly dependent on Pax5 function (8,14), these data demonstrate that the CD19+ Pax5F/F pro-B cells express normal levels of Pax5 and have undergone commitment to the B lymphoid lineage. OHT treatment of these pro-B cells (13) resulted in loss of CD19 expression and thus in efficient Pax5inactivation under the in vitro culture conditions used (Fig. 1A, Δ/Δ). The proliferation of pro-B cells remained largely unaffected by OHT treatment (15), indicating that Pax5inactivation did not result in an appreciable cell loss or in outgrowth of a minor subpopulation of Pax5 Δ/Δ pro-B cells.

Figure 1

Gene expression changes followingPax5 inactivation. (A) Loss of CD19 expression.Pax5F/F CreED-30 pro-B cells were analyzed by flow cytometry (13) for CD19 and B220 expression before (F/F) or 5 weeks after (Δ/Δ) treatment with OHT (1 μM). (B) Kinetics of Pax5 exon 2 deletion in OHT-treated Pax5F/F CreED-30 pro-B cells (13). (C) Loss of Pax5 protein. Whole-cell lysates of OHT-treated Pax5F/F CreED-30pro-B cells were analyzed by sequential immunoblotting with antibodies to Pax5 and tubulin (13). (D) RT-PCR analysis. The indicated transcripts were analyzed at different days after OHT treatment (13). FL and ΔE2: full-length and truncatedPax5 transcripts, respectively; rEST1-3: Pax5-repressed (r) transcripts of unknown function (15). The 5′ ends of the μm and δm transcripts (amplified only from constant to transmembrane exon) may correspond to the Iμ, μo, or Dμ transcript. HPRT, hypoxanthine phosphoribosyltransferase.

Time-course analyses revealed that the Pax5F/FCreED-30 pro-B cells deleted Pax5 exon 2 (Fig. 1B) and expressed the exon 2–truncated Pax5 transcript (Fig. 1D) within 1 day of OHT treatment. Concomitant with the loss of Pax5 protein (Fig. 1C), the expression of activated Pax5 target genes started to decline at day 4, as revealed by reverse transcriptase–polymerase chain reaction (RT-PCR) analysis (13) (Fig. 1D). These target genes encode the cell surface proteins CD19 and Igα (mb-1) (14), the transcription factor CIITA (10), and the alternatively spliced δm transcript of theIgH gene (10). Conversely, Pax5inactivation led to reexpression of the repressed Pax5 target genes encoding the secreted J chain (16), the macrophage colony-stimulating factor (M-CSF) receptor (6), and transcripts of unknown function (rEST1-3) (15) (Fig. 1D). However, loss of Pax5 did not affect the expression of control genes encoding hypoxanthine phosphoribosyltransferase and the B cell–specific EBF, B29(Igβ), and μm transcripts (Fig. 1D). In summary, the expression pattern ofPax5F/F pro-B cells was indistinguishable from that of Pax5+/+ pro-B cells and could be converted by Pax5 inactivation into that ofPax5 −/− pro-B cells. Hence, the Pax5-dependent gene expression program is reversible in pro-B cells undergoingPax5 inactivation.

Because the Pax5 Δ/Δ pro-B cells were derived from committed Pax5F/F pro-B cells, we next investigated whether thePax5 Δ/Δ pro-B cells remained committed or had again adopted a broad developmental potential characteristic of early hematopoietic progenitors. We assessed the potential of thePax5 Δ/Δ pro-B cells to differentiate into macrophages by culturing them in the absence of IL-7 on stromal ST2 cells (13), which produce the myeloid cytokine M-CSF (6). Within 3 weeks, thePax5 Δ/Δ cells down-regulated the B cell surface protein B220, expressed the macrophage markers Mac-1 and F4/80 (Fig. 2A), displayed the morphology of enlarged vacuolar macrophages (Fig. 2B), and phagocytosed fluorescently labeled Escherichia coli (Fig. 2C). Hence, thePax5 Δ/Δ pro-B cells were able to differentiate into mature macrophages, as observed withPax5 −/− pro-B cells (6).

Figure 2

Reacquisition of multilineage potential uponPax5 inactivation. (A) Macrophage differentiation. Pax5 Δ/Δ pro-B cells were cultured for 18 days on ST2 cells with or without IL-7 before flow cytometric analysis (13). (B) May-Grünwald-Giemsa staining ofPax5 Δ/Δ pro-B cells grown for 3 weeks on ST2 cells with or without IL-7. (C) Phagocytosis. In vitro–differentiated Pax5 Δ/Δmacrophages were incubated with fluorescein isothiocyanate–labeledE. coli, followed by DAPI (4′,6-diamidino-2-phenylindole) staining of the nuclei (13). (D) Thymus reconstitution by Pax5 Δ/Δ pro-B cells (106) that were injected into 5.5 Gy–irradiatedRAG2 −/− mice (13). Three weeks after transfer, the thymus consisted of 80 × 106cells (compared with 5 × 106 cells of uninjectedRAG2 −/− mice). The CD4+CD8+ double-positive (DP) and the CD4+ and CD8+ single-positive (SP) thymocytes were analyzed by flow cytometry for TCRβ expression (13). Stippled lines indicate control RAG2 −/−thymocytes. (E) Absence of IgM+B220+B cells in the spleen of the same injectedRAG2 −/− mouse analyzed in (D). (F) B cell lineage commitment of Pax5F/F pro-B cells. B220+ pro-B cells sorted fromPax5F/F CreED-30 bone marrow were cultured for 2 weeks (without OHT treatment) before injection of 5 × 105 cells into 5.5 Gy–irradiatedRAG2 −/− mice. After 3 weeks, the thymus and spleen were analyzed by flow cytometry (13). The Thy1.2+TCRβ thymocytes correspond to pro-T cells of the RAG2 −/− host.

We next examined the potential ofPax5 Δ/Δ pro-B cells to develop into T lymphocytes after intravenous injection into sublethally irradiated, T cell–deficient RAG2 −/− mice (13,17). Three weeks after cell transfer, the cellularity of the thymus was massively increased owing to the presence of CD4+CD8+, CD4+CD8, and CD8+CD4 thymocytes, in which expression of the T cell receptor (TCR) β was similar to that in wild-type thymocytes (Fig. 2D). Hence, thePax5 Δ/Δ pro-B cells could fully reconstitute T cell development but were unable to generate immunoglobulin M–positive (IgM+) B cells in the spleens ofRAG2 −/− mice (Fig. 2E), in analogy toPax5 −/− pro-B cells (7). This result was obtained regardless of whether the pro-B cells were cultured after Cre recombinase activation for only 4 days or for up to 5 weeks before injection into RAG2 −/− mice (Fig. 2D; figs. S1 and S2) (13). However,Pax5F/F pro-B cells did not give rise to TCRβ+ thymocytes; instead, they developed into IgM+ B cells in the spleens of injectedRAG2 −/− mice (Fig. 2F). ThePax5F/F pro-B cells are therefore fully committed to the B cell pathway. Hence, the inactivation ofPax5 alone is sufficient to revert the narrow B lymphoid potential of Pax5F/F pro-B cells to the broad multilineage potential characteristic of early hematopoietic progenitors.

The heterozygous Pax5 +/− mouse provided a second opportunity to investigate the reversibility of Pax5-dependent B cell lineage commitment, because the pro-B cells of this mouse preferentially express only one of the two Pax5alleles (18). During in vitro culture, the heterozygous CD19+ pro-B cells consistently switch expression from the wild-type Pax5 gene to the mutant allele, which carries an in-frame lacZ gene insertion (18, 19). This allele switching results in loss of expression of the Pax5 target gene CD19 and simultaneous activation of the lacZgene of the targeted Pax5 locus (18,19) (Fig. 3, A and B). Therefore, we derived CD19 β-galactosidase–positive (β-Gal+) pro-B cell clones (13) fromPax5 +/− bone marrow (Fig. 3A) and demonstrated by RT-PCR analysis (13) that these pro-B cells switched off the expression of activated Pax5 target genes and derepressed genes normally suppressed by Pax5 (Fig. 3B) (15). Hence, the CD19 β-Gal+ pro-B cells acquired the gene expression pattern of Pax5-deficient pro-B cells, although, as genotypically heterozygous cells, they retained the potential to reactivate the wild-type Pax5 allele. These CD19 Pax5 +/− pro-B cells were also decommitted, because they differentiated, upon IL-7 withdrawal, into functional Mac-1+F4/80+ macrophages with phagocytic activity (Fig. 3, C and D) (15). To study the T lymphoid potential of CD19 Pax5 +/− pro-B cells, we first infected these cells with a human (h) CD2-expressing retrovirus before injecting them into sublethally irradiated RAG2 −/− mice (13). Three weeks after cell transfer, the thymus contained CD4+CD8+, CD4+CD8, and CD8+CD4 T cells of donor origin (hCD2+), indicating that the CD19 Pax5 +/− pro-B cells could reconstitute T cell development (Fig. 3E). Thymic reconstitution was still observed 11 weeks after cell transfer owing to long-term engraftment of the CD19B220+ Pax5 +/−pro-B cells in the bone marrow, from where these progenitors continuously seed the thymus (Fig. 3F). The bone marrow additionally contained CD19+B220+ B lymphocytes of donor origin (hCD2+), which had reentered the B cell pathway through activation of the wild-type Pax5 allele (Fig. 3F). Finally, a third population of hCD2+CD19B220 cells was present (Fig. 3F), which consisted of mature CD4+ and CD8+ T lymphocytes (15). In conclusion, heterozygous Pax5 +/− pro-B cells not only regained the broad developmental potential of early hematopoietic progenitors upon switching to the mutant Pax5 allele, but at the same time maintained their competence to undergo B cell development by reactivating the wild-type Pax5 allele in vivo.

Figure 3

Decommitment of Pax5 +/−pro-B cells upon switching expression from the wild-type to the targeted Pax5 allele. (A) Generation of CD19 Pax5 +/− pro-B cell clones. CD19+ pro-B cells from Pax5 +/− bone marrow were cultured with ST2 cells and IL-7 for 8 weeks (w) before single-cell cloning (13). The pro-B cell pool and a derived clone were analyzed by flow cytometry for CD19, B220, and β-Gal expression at the indicated times after the start of in vitro culture (13). The stippled line denotes wild-type pro-B cells. (B) RT-PCR analysis of the indicated transcripts in wild-type (+/+), homozygous (−/−), and heterozygous (+/−)Pax5 mutant pro-B cells. The pool (p) and a clone (c) of Pax5 +/− pro-B cells were analyzed after 1 and 13 weeks of in vitro culture, respectively. (C andD) Macrophage differentiation. Cloned CD19 Pax5 +/− pro-B cells were cultured for 3 weeks on ST2 cells with or without IL-7 and analyzed for B220 and Mac-1 expression (C) and phagocytic activity (D). (E) Thymus reconstitution. The cells of a CD19 Pax5 +/− pro-B clone were infected with a retrovirus expressing a truncated human (h) CD2 protein, injected into 5.5 Gy–irradiated RAG2 −/− mice, and analyzed by flow cytometry at 3 weeks after transfer (13). CD4 and CD8 expression is shown for the hCD2+ thymocytes of donor origin. (F) Long-term engraftment. Eleven weeks after cell transfer, the bone marrow of reconstitutedRAG2 −/− mice was analyzed by flow cytometry; expression of CD19 and B220 is displayed for hCD2+ donor cells.

In summary, we have shown by conditional Pax5inactivation and allele switching experiments that Pax5 is continuously required to maintain its B lymphoid transcription program and B cell lineage commitment during early B cell development. These findings raise the intriguing possibility that wild-type pro-B cells may differentiate into other hematopoietic cell types by down-regulatingPax5 expression, thus reversing B cell commitment in response to external signals. Because the target gene CD19is totally dependent on Pax5 for its transcription (8,14), we argued that CD19 expression is not only an indicator of Pax5 activity (18), but is also a decisive marker of B cell lineage commitment (6,19). Unexpectedly, however, CD19+ progenitors of the bone marrow were shown to differentiate upon exposure to appropriate cytokines, into dendritic cells (20), macrophages (21), or osteoclasts (22), which was accompanied by down-regulation of CD19 expression (20). These observations, together with our data, suggest that committed B cell progenitors can again broaden their developmental potential by repressing Pax5 transcription under physiological conditions.

Finally, the generation of multipotent hematopoietic progenitors from wild-type pro-B cells is also of medical relevance. Ribozyme, antisense RNA, or RNA interference strategies could be used to inactivate or down-regulate PAX5 expression in wild-type pro-B cells isolated from human leukocyte antigen (HLA)–matched donors (23). In analogy to the mouse model (7), these human Pax5-deficient pro-B cells should have regained the potential to restore T cell development in patients with AIDS or inherited immunodeficiency syndromes (24).

  • * These authors contributed equally to this work.

  • Present address: Department of Biochemistry, Institute of Medical Biology, University of Tromsø, N-9037 Tromsø, Norway.

  • To whom correspondence should be addressed: E-mail: busslinger{at}


View Abstract

Navigate This Article