Volume 117, Issue 5, 28 May 2004, Pages 663–676
Stepwise Reprogramming of B Cells into Macrophages
- Department of Developmental and Molecular Biology, Albert Einstein College of Medicine Cancer Research Center, 1300 Morris Park Avenue, Bronx, NY 10461 USA
- Received 15 December 2003, Revised 23 March 2004, Accepted 31 March 2004, Available online 27 May 2004
- Published: May 27, 2004
Abstract
Starting with multipotent progenitors, hematopoietic lineages are specified by lineage-restricted transcription factors. The transcription factors that determine the decision between lymphoid and myeloid cell fates, and the underlying mechanisms, remain largely unknown. Here, we report that enforced expression of C/EBPα and C/EBPβ in differentiated B cells leads to their rapid and efficient reprogramming into macrophages. C/EBPs induce these changes by inhibiting the B cell commitment transcription factor Pax5, leading to the downregulation of its target CD19, and synergizing with endogenous PU.1, an ETS family factor, leading to the upregulation of its target Mac-1 and other myeloid markers. The two processes can be uncoupled, since, in PU.1-deficient pre-B cells, C/EBPs induce CD19 downregulation but not Mac-1 activation. Our observations indicate that C/EBPα and β remodel the transcription network of B cells into that of macrophages through a series of parallel and sequential changes that require endogenous PU.1.
Introduction
Blood cell formation is one of the classical systems in which to study mechanisms of vertebrate lineage determination. Hematopoietic lineages are specified in a stepwise process of binary decisions, starting with multipotent progenitors, which branch into a common lymphoid and a common myeloid progenitor that differentiate in turn into additional intermediate progenitors Akashi et al. 2000a and Akashi et al. 2000b. Each lineage exhibits a distinct gene expression pattern that is laid down and maintained by a set of more than a dozen lineage-restricted transcription factors that are part of a “transcription factor network” Orkin 2000 and Sieweke and Graf 1998. Since random combinations of several of these factors could, at least in principle, lead to far larger numbers of phenotypes than observed, there must be mechanisms that ensure a highly coordinated control of transcription factor network remodeling during commitment. This assumption is supported by enforced transcription factor expression experiments where, for example, ectopic expression of GATA-1 (a Zn finger protein) in transformed as well as normal myeloid precursor cells reprograms them into megakaryocytic/erythroid (MEP) cells, eosinophils, or mast cells, lineages in which GATA-1 is normally expressed Heyworth et al. 2002 and Kulessa et al. 1995. Likewise, expression of GATA-1 in common lymphoid progenitors suppresses their lymphoid colony-forming capacity and induces their capacity to form MEP colonies (Iwasaki et al., 2003). These experiments also showed that altering the balance of lineage-restricted transcription factors can reverse commitment and that reprogramming involves not just activation of novel gene expression programs but also extinction of the original ones. At least in some cases, this involves direct transcription factor interactions such as those between GATA-1 and PU.1 (an ETS family factor) Cantor and Orkin 2001 and Graf 2002.
Which transcription factors determine the decision-making process between lymphoid and myeloid cell fates and how they do it remain largely unknown. Unlike in adult bone marrow, where lymphoid and myeloid compartments branch early, the fetal liver contains bipotent B cell/macrophage progenitors (B/M) (for review, see Katsura [2002]), suggesting that these two lineages are closely related. It has also been shown that certain oncogene-immortalized B cell lines can be reprogrammed into macrophages by the raf/ras oncogenes (Klinken et al., 1988) and by an activated form of the M-CSF receptor (M-CSFR) (Borzillo et al., 1990). Although the cell conversion frequencies appeared to be low, in both studies, the resulting myelomonocytic cells had the same immunoglobulin rearrangements as the original B lineage cells. These changes must have therefore involved a complete shutdown of the B cell-specific gene expression program and its replacement by a macrophage-specific gene expression program. How does this happen?
B cell differentiation is initiated by the transcription factors E2A (a helix-loop-helix protein) and EBF, in whose absence B cells are arrested at a stage before DJ rearrangements. Together, these genes induce the transcription of many B cell-specific genes, including that of the B cell commitment factor Pax5. From the pro-B cell stage onward, Pax5 (a paired domain transcription factor) activates the expression of genes such as CD19 and blnk while suppressing lineage-inappropriate genes such as M-CSFR (c-fms) and Notch-1 Kee and Murre 2001 and Schebesta et al. 2002. C/EBPβ (a bZip family transcription factor) may also play a role in B cell differentiation, since ablation of this factor, which is expressed predominantly in mature cells of the B lineage (Cooper et al., 1994), leads to a decreased number of B cells in the bone marrow (Chen et al., 1997). A transcription factor that is important for both B cell and macrophage formation is PU.1, since mice defective in this gene lack both lineages (reviewed in Schebesta et al. [2002]). This factor, which is expressed at lower levels in B cells than in macrophages (DeKoter and Singh, 2000), has been implicated in B/M lineage decisions, since low levels of PU.1 expressed in PU.1−/+ fetal liver precursors promote B cell differentiation, while high levels promote macrophage differentiation (DeKoter and Singh, 2000). No transcription factors have yet been shown to be uniquely required for macrophage formation, although C/EBPβ−/− macrophages exhibit functional defects such as a decreased production of the inflammatory response cytokine IL-12 (Screpanti et al., 1995). In contrast, C/EBPα−/− mice lack granulocytes (Zhang et al., 1997), a defect that can be rescued if C/EBPβ is inserted into the C/EBPα locus (Jones et al., 2002). In short, B cell fate is determined by E2A, EBF, Pax5, and PU.1, while myelomonocytic cells (which include both macrophages and granulocytes) are determined by elevated levels of PU.1 as well as by C/EBPα and/or C/EBPβ.
Here, we show that enforced expression of C/EBPα and β in B cells leads to their rapid and efficient reprogramming into macrophages. This occurs through a complex process that is initiated by the coordinated inhibition of Pax5 activity, resulting in the downregulation of CD19, and the synergistic action between C/EBPα and PU.1, resulting in the activation of macrophage-specific genes.
Results
Enforced C/EBPα Expression Induces Mac-1 Expression in Primary B Cell Precursors
A retroviral approach was used to express myeloid transcription factors in primary B cell precursors. As shown in Figure 1A, the vectors encode either GFP or hCD4 alone (“control virus”) or C/EBPα together with GFP/hCD4 (“C/EBPα virus”). To obtain B cell progenitors, CD19+ cells were purified from the bone marrow of C57Bl/6J mice using magnetic bead selection, typically yielding >99% CD19+ cells (Figure 1B). Control virus-infected cells, grown under conditions that support both B cell and myeloid cell development, remained CD19+Mac-1−(Figure 1C). In contrast, cells infected with C/EBPα downregulated CD19 and upregulated Mac-1, with ∼60% of the cells becoming CD19−Mac-1+ after 4 days (Figure 1D). No changes were seen in the uninfected portion of the sample. The effect is also C/EBPα-specific, since cells infected with virus expressing the irrelevant FOG-1 transcription factor showed no changes (data not shown). The efficiency of the observed phenotypic conversion was dependent on the level of C/EBPα, since the proportion of “nonresponders” was 4% within the population of GFP high cells and 60% within that of GFP low cells (Figure 1E).