The TAL1/SCL Transcription Factor Regulates Cell Cycle Progression and Proliferation in Differentiating Murine Bone Marrow Monocyte Precursors

ABSTRACT Monocytopoiesis involves the stepwise differentiation in the bone marrow (BM) of common myeloid precursors (CMPs) to monocytes. The basic helix-loop-helix transcription factor TAL1/SCL plays a critical role in other hematopoietic lineages, and while it had been reported to be expressed by BM-derived macrophages, its role in monocytopoiesis had not been elucidated. Using cell explant models of monocyte/macrophage (MM) differentiation, one originating with CMPs and the other from more committed precursors, we characterized the phenotypic and molecular consequences of inactivation of Tal1 expression ex vivo. While Tal1 knockout had minimal effects on cell survival and slightly accelerated terminal differentiation, it profoundly inhibited cell proliferation and decreased entry into and traversal of the G1 and S phases. In conjunction, steady-state levels of p16(Ink4a) mRNA were increased and those of Gata2 mRNA decreased. Chromatin immunoprecipitation analysis demonstrated the association of Tal1 and E47, one of its E protein DNA-binding partners, with an E box-GATA sequence element in intron 4 of the Gata2 gene and with three E boxes upstream of p16(Ink4a). Finally, wild-type Tal1, but not a DNA binding-defective mutant, rescued the proliferative defect in Tal1-null MM precursors. These results document the importance of this transcription factor in cell cycle progression and proliferation during monocytopoiesis and the requirement for direct DNA binding in these processes.

Mononuclear phagocytes serve as critical effectors of both innate and adaptive immunity and have important functions in the maintenance of tissue homeostasis. Macrophages derive from common myeloid precursors (CMPs) that differentiate in the bone marrow (BM). Monocytopoiesis is driven by specific growth factors that include interleukin-3 (IL-3) and macrophage colony-stimulating factor (M-CSF) and by transcription factors, including PU.1 and C/EBP␣ (10). This generates monoblasts, promonocytes, and finally monocytes, which are released into circulation (10,12). While monocytes are capable of phagocytosis and cytokine secretion, their functional capabilities are enhanced with differentiation to macrophages in tissues.
The T-cell acute lymphoblastic leukemia 1 (TAL1) gene, also known as stem cell leukemia (SCL), encodes a class II basic helix-loop-helix (bHLH) transcription factor that regulates both embryonic and adult hematopoiesis (25,28,39,42,43,45). In the adult, TAL1 has a recognized role in the differentiation of erythroid, megakaryocytic, and mast cell progenitors (1,3,13,14,31,41,54) and is expressed in vascular and lymphatic endothelial cells under both physiological and pathological conditions (49). Heterodimers of TAL1 and ubiquitous bHLH proteins known as E proteins regulate transcription by binding to specific regulatory regions in target genes in association with the zinc finger transcription factor Gata1 or Gata2. The sequence element recognized by those complexes includes an E box (CANNTG) that binds a bHLH heterodimer and a WGATAR motif recognized by a GATA protein (34,35,56,58,60). TAL1/E protein heterodimers can also bind E boxes to effect transcriptional repression (8,11,19,20,55). In murine erythroleukemia and T-cell acute lymphoblastic leukemia cell lines, TAL1 had been shown to interact with a number of nuclear corepressors, including mSin3A and HDAC1 (21,23), Brg1 and HDAC2 (59), and Mtg16/Eto-2 and Mtgr1 (6,44).
Although an important role for TAL1 has been demonstrated in other hematopoietic lineages, a function in mononuclear phagocyte production, while suggested by a number of observations, had not been established. First, TAL1 had been found to contribute to specific E box DNA-binding complexes in the M1 monocytic leukemia cell line (55). While studies with M1 cells and the human bipotent cell line TF-1 suggested that it was downregulated during monocytic differentiation (18,50,55), TAL1 mRNA or protein was detected in mouse peripheral blood mononuclear cells and BM macrophages (25) and zebrafish (62) and human (41) macrophages. In addition, higher and earlier expression of IL-6 receptor (Il6-r) and colony-stimulating factor 1 receptor (Csf1r) mRNA (our unpublished data) was observed in Tal1-transduced primary mouse BM monocyte/macrophage (MM) precursors and M1 cells induced to differentiate with mouse M-CSF (mM-CSF) and IL-6 (mIL-6), respectively. In contrast, knockdown of TAL1 in human cord blood CD34 ϩ cells by short hairpin RNA reduced the numbers of both myeloid and erythroid cell progenitors (5). Finally, while deletion of Tal1 in adult hematopoietic stem cells (HSCs) did not affect granulocyte-macrophage or macrophage progenitor numbers (16, 30), 5-fluorouracil (5-FU) treatment of Tal1 Ϫ/Ϫ mutant mice resulted in a transient but significant delay in the recovery of circulating monocytes (David Curtis, unpublished data).
In this study, we establish that Tal1 is expressed at all stages of monocytic differentiation from CMP to macrophage and report a previously unrecognized function of this hematopoietic transcription factor in cells of this lineage. Using an ex vivo method of Tal1 gene deletion, knockout of Tal1 in BM mononuclear cells was found to suppress cellular proliferation at a relatively late stage of monocytopoiesis and delay entrance into and progression through S phase. Finally, we identify two genomic targets of TAL1 in differentiating MM precursors whose altered expression likely contributed to the abnormal phenotype of knockout cells.
periods, including 12, 24, 36, and 48 h. These cells were processed for flow cytometry analysis identically to the others.

RESULTS
Tal1 is expressed continuously during murine monocytopoiesis. To characterize Tal1 expression in monocyte progenitors, we made use of a previously developed method for isolating CMPs from mouse BM and inducing their differentiation (48). Using quantitative reverse transcription (RT)-PCR analysis, Tal1 mRNA was detected at all stages from the common myeloid progenitor to the postmitotic macrophage (Fig.  1A). Although Tal1 message fluctuated in abundance, it was detectable throughout, as well as in LPS-and IFN-␥-activated macrophages, albeit at lower levels than in unactivated cells (Fig. 1A). Macrophage differentiation was verified by morphological analysis of Wright-Giemsa-stained cytospin preparations (Fig. 1B), quantitative RT-PCR analysis of well-  (48), and flow cytometry analysis of the cell surface antigens CD31 and Ly6C. As previously described (48), CMPs (CD31 ϩ Ly6C Ϫ ) differentiated sequentially to CD31 ϩ Ly6C ϩ , CD31 Ϫ Ly6C ϩ , and CD31 Ϫ Ly6C Ϫ cells (data not shown). Expression studies were also carried out with MM precursors explanted at a later stage of differentiation. Using methods developed by Stanley and colleagues (15,47,53), terminal differentiation of mouse BM monoblasts and promonocytes was induced in vitro with recombinant murine M-CSF. Western blot analysis of cellular extracts prepared after 1, 4, 6, and 8 days of culture revealed only modest changes in Tal1 expression over this period ( Fig. 1C and D).
Loss of Tal1 impairs proliferation of mouse MM precursors. For subsequent studies, monoblasts and promonocytes were isolated from mouse BM using established procedures (15,47,53) and Tal1 expression was manipulated in them in culture. In the absence of a suitable promoter to direct Cre expression to monocytic progenitors in vivo, a strategy was employed to delete the gene in MM precursors isolated from mice with a floxed Tal1 allele. To that end, BM mononuclear cells from SCL loxP/LacZ mice were transduced with a retrovirus, MSCV-GFP-Cre, expressing Cre recombinase to render them nullizygous for a functional Tal1 gene and these cells were then induced to differentiate in the presence of recombinant M-CSF (15,53). Semiquantitative PCR analysis of genomic DNA from day 8 Tal1 Ϫ/Ϫ cells revealed that the loxPflanked allele was almost totally deleted by retroviral delivery of Cre, although low levels of the floxed allele were detected at later times in culture, likely due to a competitive advantage of the SCL loxP/LacZ cells over SCL ⌬/LacZ cells (data not shown). In any case, RT-PCR analysis showed essentially complete abolition of Tal1 expression in gene-targeted cells ( Fig. 2A).
Cell counts at timed intervals in culture revealed a severe proliferative defect in Tal1 knockout cells compared to vectortransduced cells (Fig. 2B). While the number of cells in cultures infected with the control vector increased approximately threefold, Cre-transduced cells showed little or no change. Toxic effects of Cre expression and retroviral infection (data not shown) were ruled out by transduction of wild-type cells with Cre DNA and the empty vector, respectively, with both populations found to proliferate at rates similar to those of nontransduced cells (Fig. 2C). In contrast, Tal1-overexpressing cells had a higher proliferative capacity than wild-type cells (Fig. 2D). Finally, heterozygous knockout cells accumulated in slightly lower numbers than wild-type cells, although this was not apparent until late in culture (days 6 and 8) (Fig. 2D). Thus, loss of Tal1 in differentiating MM precursors significantly impaired cellular proliferation, with Tal1 gene expression correlating closely with proliferative potential.
Dye dilution analysis of transduced cells using the fluorescent membrane-intercalating dye PKH26 showed both an absolute reduction and a delay in cell division in SCL ⌬/LacZ cells compared to SCL loxP/LacZ cells (Fig. 2E). With Ͼ98% of the day 1 cells of both genotypes assigned to the first generation (Fig. 2E, i and iv), a model in which three (or more) generations of cells appeared over the 6-to 8-day culture period best fit the data. This would correspond to a fourfold [or 2 (3 Ϫ 1) ] increase in cell number over the first generation, slightly higher than what was observed for control SCL loxP/LacZ cells. After 6 days, 34.8% of the SCL ⌬/LacZ cells were still first generation, 46.8% were second generation, and only 17.8% could be classified by dye content as third generation (Fig. 2E, v). Over the same time period, 13.8% of the control SCL loxP/LacZ cells were first generation, 21.1% were second generation, and 62.8% were third generation (Fig. 2E, ii). While slightly less than half of the SCL ⌬/LacZ cells (48.9%) were second generation after 8 days, most of the SCL loxP/LacZ cells were third generation, with some fourth generation cells (6.5%) also represented (Fig. 2E, iii and vi). Similar results were obtained when mean fluorescence intensities were compared (data not shown).
Tal1 knockout MM precursors show reduced viability and accelerated differentiation. To determine whether the reduced accumulation of SCL ⌬/LacZ cells resulted from decreased viability, apoptosis was analyzed using annexin V and 7-AAD staining, with nonapoptotic cells characterized by the ability to exclude both stains (Fig. 3A). The percentage of viable cells in the SCL ⌬/LacZ group was very slightly decreased relative to that of SCL loxP/LacZ cells at days 4 and 6 but was not different at days 1 and 8 of culture. These results show that Tal1 gene loss had minimal effects on cell survival and that apoptosis could not account for the marked reduction in cell number observed.
Cellular differentiation was then assessed by flow cytometry analysis of cell surface antigens CD31 and Ly6C (Fig. 3B). A higher percent of the most mature CD31 Ϫ Ly6C Ϫ cells was noted in the SCL ⌬/LacZ population (59.2% at day 2 and 91.2% at day 4) than in the SCL loxP/LacZ population (52.2% at day 2 and 82.6% at day 4) at early times in culture. In contrast, less differentiated CD31 Ϫ Ly6C ϩ cells made up a smaller proportion of SCL ⌬/LacZ cultures (26.9% at day 2 and 8.5% at day 4) than of SCL loxP/LacZ cultures (36.1% at day 2 and 16.4% at day 4). Expression of another widely used macrophage marker, F4/80, did not reliably discriminate between different stages of monocytopoiesis in our studies, with approximately 90% of the cells in both groups becoming F4/80 positive after only 24 h in culture (data not shown). In sum, these data indicate that SCL ⌬/LacZ cells lacking a functional Tal1 gene differentiated slightly more rapidly than SCL loxP/LacZ cells.
SCL ⌬/LacZ cells are defective in cell cycle progression. To better characterize the proliferative defect in Tal1 knockout cells, a BrdU pulse-chase analysis was carried out. To that end, GFP-expressing cells were cultured for 3 days, labeled with BrdU for 45 min, and followed through at least one cell cycle by flow cytometry (Fig. 4A). Analyzed immediately after pulsing, the majority of the BrdU-labeled cells from both the SCL loxP/LacZ and SCL ⌬/LacZ groups were in S phase (94.9% and 87.3%, respectively), with smaller percentages in early S phase or late G 0 /G 1 (5.1% and 12.7%, respectively; shaded dark in the 0-h population, Fig. 4A, i and v). At later times, SCL loxP/LacZ cells progressed rapidly to G 2 /M and then to G 0 /G 1 , with a concomitant decrease in the percentage in S phase (Fig. 4A, ii, iii, and iv). In contrast, while some SCL ⌬/LacZ cells advanced to late S phase 1 h after pulsing, as evidenced by the two peaks of labeling (Fig.  4A, vi), these cultures had a larger proportion of cells in G 0 /G 1 after 2 and 4 h (18.1% and 35.2%, respectively), lacked a distinct G 2 /M population (0% and 18.8%, respectively), and contained a higher percentage of cells still in S phase (81.9% and 46.0%) than did the control cells (Fig. 4A, vii and viii). These results provide evidence of slowed cell cycle progression 2184 DEY ET AL. MOL. CELL. BIOL.
in Tal1 knockout cells, with delayed entry into and traversal of S phase. As only cells able to incorporate the BrdU label can be analyzed by the pulse-chase approach, MM precursors were also cultured with BrdU for longer times (12,24,36, and 48 h) before analysis by flow cytometry (Fig. 4B). As could have been predicted, the SCL ⌬/LacZ population contained a higher proportion (27.5% versus 7.6% for SCL loxP/LacZ cells, Fig. 4B, iv  (Fig. 4B, vii and viii), consistent with their increased tendency toward apoptosis (Fig. 3A). Finally, and most directly, double staining with Hoechst 33342 and pyronin Y showed a threefold higher percentage of cells in G 0 in SCL ⌬/LacZ than in SCL loxP/LacZ cultures (data not shown). Together, these studies suggest a role for TAL1 in cell cycle progression in MM precursors that can largely account for the proliferative defect associated with homozygous Tal1 gene loss. Tal1 knockout MM precursor cells contain lower Gata2 and higher p16(Ink4a) mRNA levels. Tal1 gene loss would be expected to impact the transcription of its target genes, especially direct targets like those coregulated with E proteins. Accordingly, real-time PCR analysis was used to evaluate expression of genes known to have a role in MM differentiation, including putative TAL1 targets, with comparisons made among SCL WT/WT , Tal1-overexpressing, and SCL ⌬/LacZ cells. The same bicistronic retroviral vector used to deliver Cre to MM precursors for inactivation of the floxed Tal1 allele was also employed in enforcing Tal1 expression (Fig. 5A). These studies revealed that Gata2 mRNA was reduced seven-and fourfold, respectively, at days 1 and 4 in SCL ⌬/LacZ relative to SCL WT/WT cells (Fig. 5B). Although the level of Gata2 expression decreases physiologically with differentiation of MM precursors ( Fig. 5B and reference 48), Tal1-null (SCL ⌬/LacZ ) cells showed even lower Gata2 expression than wild-type controls at days 6 Because of the delay in cell cycle progression observed with Tal1 loss (Fig. 4A and B) and the previously published evidence of their regulation by E proteins, expression of the genes for several cyclin-dependent kinase (CDK) inhibitors was analyzed. p16(Ink4a), which was shown to be a target of repression by Tal1 (17,33,37), was upregulated approximately threeto fourfold in SCL ⌬/LacZ cells and reduced, albeit to a lesser extent, in Tal1-overexpressing cells (Fig. 5C). In contrast, expression of p21(Cip1), another putative target of Tal1 and E proteins (29,37,40), was only slightly increased in SCL ⌬/LacZ cells compared to SCL WT/WT cells (data not shown). While elevation of p16(Ink4a) and p21(Cip1) could relate to the cell cycle delay in Tal1 knockout cells, especially given the phase of the cell cycle affected, the only slight reduction in these mRNAs cannot explain the increased proliferation by Tal1-overexpressing cells (Fig. 2D). Finally, expression of Csf1r, the receptor for M-CSF and itself a marker of macrophage differentiation, was slightly elevated in SCL ⌬/LacZ cells (data not shown) and could, potentially, have contributed to their accelerated differentiation (Fig. 3B).
Gata2 intron 4 and p16(Ink4a) upstream regions are occupied by a Tal1-E protein complex in MM precursors. The above-described studies revealed that knockout, and to a lesser extent overexpression, of Tal1 affected the expression of Gata2 and p16(Ink4a). Previous studies defined an enhancer in intron 4 of the Gata2 gene (kb ϩ9.5) with a consensus E box-GATA DNA-binding motif that is occupied by Tal1, E proteins, and GATA proteins in endothelial and fetal liver cells (27,57). Moreover, this DNA sequence element appeared to be essential for the proper expression of Gata2 in endothelial cells (27).
To ascertain whether Tal1 occupies this region in MM precursors, a ChIP analysis was carried out on sonicated chromatin from day 4 and day 7 wild-type (SCL WT/WT ) cells with polyclonal antibodies to Tal1 and E47. The amount of DNA immunoprecipitated was then quantified in a real-time PCR analysis using region-specific primers and expressed as a percentage of the input (Fig. 6A and B).
Antibodies to Tal1 and E47, but not normal rabbit IgG, were able to precipitate this intronic fragment from day 4 cells (Fig.  6A) but not day 7 cells (Fig. 6B), concordant with the decline in Gata2 expression over this same time period. Primer sequences designed for the Gata2 3Ј UTR did not support any amplification (data not shown), indicating specific occupancy of this intronic region in MM precursor cells.
Scanning a 12-kb region upstream of the p16(Ink4a) translational start site identified five preferred E box sequences (CATCTG or CAGATG) for TAL1/E protein binding (bases Ϫ13, Ϫ2592, Ϫ4638, Ϫ5106, and Ϫ6932). PCR primers complementary to each of these regions and to the gene's 3Ј UTR were then designed, and a quantitative ChIP analysis was carried out. Three of these E boxes, at Ϫ2592, Ϫ4638, and Ϫ5106, were shown by this approach to be occupied by Tal1 and E47 in day 4 cells (Fig. 6A). In contrast, no binding was detected at Ϫ13, Ϫ6932, or the 3Ј UTR at any time point (data not shown), and another, noncanonical E box in the promoter region (Ϫ437) likewise was not occupied (data not shown). Finally, Tal1 and, unexpectedly, E47 occupancy of the Ϫ2592 and Ϫ4638 elements was very low and essentially absent at the Ϫ5106 E box at day 7 (Fig. 6B), despite the increase in p16(Ink4a) expression with differentiation (Fig. 5C). This suggests that E proteins other than E47 are involved or that other mechanisms for p16(Ink4a) gene transactivation operate at these later times. p16(Ink4a) expression increases with IL-6-induced differentiation (2) of the murine myeloid leukemia cell line M1 in parallel with a decrease in Tal1 expression and Tal1/E protein DNA-binding activity (55). As this phenocopies the results of Tal1 gene knockout, a ChIP analysis was also carried out on chromatin from untreated and IL-6-treated M1 cells. As predicted from its drastic decline in expression (55), Tal1 occupancy was detected only in uninduced cells (Fig. 6C). In contrast, E2A association with these same E box elements was detected in both IL-6-treated and untreated cells, indicating that its E12 and/or E47 protein products acted to augment the expression of this gene unopposed by TAL1 when these cells were induced to differentiate (Fig. 6D).
Tal1 function in MM precursor cell proliferation requires direct DNA binding. TAL1 has been shown to regulate transcription through direct binding to DNA and as a non-DNAbinding cofactor (34,56,58). To determine whether the proliferative defect in SCL ⌬/LacZ cells could be rescued by a wild-type Tal1 cDNA, SCL loxP/LacZ MM precursors were infected with MSCV-GFP-Cre and MSCV-YFP-Tal1, while control cells were transduced with the corresponding parental vectors. Equal numbers of GFP-and YFP-expressing cells were then sorted, and cell numbers were determined at specific times. These studies showed that the defective proliferation of Tal1 knockout cells could be completely rescued by the Tal1 cDNA, with the growth curve of Tal1-transduced SCL ⌬/LacZ cells virtually identical to that of control cells (Fig. 7). Since Tal1 deletion resulted in the downregulation of Gata2 expression, a Gata2 cDNA was also tested in rescue experiments. However, cells transduced with the Gata2 cDNA were nonviable after only 1 day in culture, demonstrating an adverse effect of Gata2 overexpression in a Tal1-null background (data not shown). Finally, to investigate whether DNA-binding activity is required for TAL1 function in MM proliferation, SCL ⌬/LacZ cells were transduced with a cDNA for a well-characterized DNA binding-defective mutant form of Tal1, Tal1 T192P (21). Significantly, cells transduced with this protein behaved identically to SCL ⌬/LacZ cells and showed no recovery of proliferative capacity (Fig. 7). This result indicates that, similar to its actions in erythroid model systems (26), Tal1 DNA-binding activity is required for outgrowth of murine MM precursors in explant cultures.
FIG. 6. ChIP analysis of Tal1 and E47 occupancy of the p16(Ink4a) and Gata2 genes in wild-type MM precursors cultured for 4 (A) or 7 (B) days. DNA derived from chromatin fragments immunoprecipitated with antibodies to Tal1 or E47 or normal rabbit IgG was quantified by real-time PCR analysis using region-specific primers. Each bar represents the mean abundance Ϯ the standard deviation of each fragment relative to the input as determined from triplicate PCR measurements. ChIP analysis of Tal1 (C) and E2a (D) occupancy of the p16(Ink4a) and Gata2 genes in M1 cells cultured with or without IL-6 for 48 h. DNA derived from chromatin fragments immunoprecipitated with antibodies to Tal1 or E2a or normal rabbit IgG was quantified by real-time PCR analysis using region-specific primers. Each bar represents the mean abundance Ϯ the standard deviation of each fragment relative to the input as determined from triplicate PCR measurements.

DISCUSSION
Previous studies have established the importance of TAL1 in erythroid, megakaryocytic, and mast cell differentiation. Although derived from the same progenitor cell as these, the CMP, there is conflicting evidence for TAL1 function in MM production. First, work carried out with the TF-1 (18) and M1 (55) cell lines showed that TAL1 expression declined with macrophage differentiation and that enforced expression of this transcription factor inhibited this process (51), although monocytic differentiation of TF-1 cells was assessed with the CD45 and Fc␥RII (CD32) markers, neither of which is macrophage specific. Regardless, TAL1 expression has been detected in primary mouse macrophages by several groups and in mouse peripheral blood mononuclear cells and BM macrophages (25) and in zebrafish (62) and human (41) macrophages. In addition, the murine Tal1 cDNA was originally cloned from a BM macrophage cDNA library (4).
We investigated here whether the abundance of Tal1 mRNA was altered during in the differentiation of CMPs to monocytes and whether Tal1 knockout affected the survival, proliferation, or differentiation of mouse MM precursors. While deletion of Tal1 in HSCs in IFN-inducible MxI-Cre mice did not change the granulocyte-macrophage progenitor number significantly (16), loss of Tal1 in cells also nullizygous for the related Lyl1 gene did reduce myeloid colony-forming activity, raising the possibility of genetic redundancy (46). Deletion of the Tal1 gene ex vivo, as done in these studies, may also have brought out an abnormal phenotype, either because of the acuteness of gene loss or because growth in culture simulates stress monocytopoiesis. In support of the latter notion, SCL ⌬/LacZ mice showed a small but significant delay in the recovery of circulating monocytes after 5-FU treatment (Curtis, unpublished).
The complete loss of Tal1 caused proliferative and cell cycle defects in macrophage precursors, and wild-type and heterozygous knockout cells also differed in their growth rates. This was most apparent at later times in culture (Fig. 2D), however, suggesting that more differentiated cells such as monoblasts or promonocytes are more sensitive to Tal1 gene dosage. In contrast, overexpression of Tal1 in MM progenitors resulted in a substantial increase in cell number (Fig. 2D), similar to what was reported for the 32D and HL-60 cell lines (7). As enforcement of Tal1 expression in mouse MM precursors was associated with Il6-r and Csf1r upregulation (data not shown), the increased proliferation of these cells could reflect a heightened sensitivity to cytokines.
The genes for several CDK inhibitors, including p16(Ink4a) and p21(Cip1), have been shown to be positively regulated by E proteins, while Tal1/E protein heterodimers were suggested to repress their transcription (8,17,19,33,37,40). Further, restoration of E2A activity in the Jurkat cell line, in which Tal1-E2A function is low, profoundly depressed growth and increased apoptosis (36). In Tal1 knockout MM precursor cells, abrogation of Tal1 inhibition of E protein action was likely a major contributor to both the upregulation of p16 (Ink4a) and p21(Cip1) expression and the impairment of cell cycle progression and proliferation observed. The gradual increase in p16(Ink4a) mRNA abundance (Fig. 5C) in the face of nearly invariant Tal1 expression during late-stage monocytopoiesis ( Fig. 1A and C) suggests that while Tal1 may be required for initial suppression of p16(Ink4a) transcription, it is less important for maintenance of gene repression. This is also supported by the decreased Tal1/E protein occupancy of E boxes upstream of the p16(Ink4a) gene at day 7 of culture compared to day 4 ( Fig. 6A and B). The further upregulation of p16(Ink4a) expression in Tal1 Ϫ/Ϫ Lyl1 Ϫ/Ϫ CMPs (Curtis, unpublished) suggests that there is, in addition, genetic redundancy in these bHLH proteins, at least for this target gene. In contrast, p16(Ink4a) mRNA abundance was only slightly decreased in Tal1-overexpressing cells (Fig. 5C), making it unlikely to account for their increased proliferative rate (Fig.  2D). Regardless of the timing of its actions, the direct association of Tal1 with three preferred E box sites in the p16(Ink4a) upstream region in both primary MM precursors and the M1 cell line provides the best evidence that this gene is a target of this transcription factor in cells.
While p16(Ink4a) transcription was inhibited by Tal1 in differentiating cells of this lineage, the Gata2 gene appeared to be transactivated by Tal1. A region in intron 4 of Gata2 was shown previously to have strong enhancer activity in endothelial and mouse fetal liver cells (27,57), and the present data are compatible with its also being active in mononuclear phagocytes. Indeed, the slight acceleration of differentiation in Tal1 knockout cells could be explained by their reduced levels of Gata2, as a similar finding was reported for Gata2-deficient juvenile myelomonocytic leukemia cells (61) and brown adipocytes (32,52). Furthermore, Gata2 knockdown reduced G 1 -S transition while increasing myeloid gene expression in G1ME cells (24), and the further decrease in Gata2 expression in SCL ⌬/LacZ cells could also have contributed to their abnormal cell cycle progression. In contrast to p16(Ink4a) transcription, Gata2 transcription was not affected by Tal1 overexpression and was reduced only with Tal1 nullizygosity, reflecting a difference in the quantitative requirement for Tal1 in genes activated and repressed by this transcription factor.
Finally, the proliferative defect in SCL ⌬/LacZ cells was fully rescued by the reintroduction of a Tal1 cDNA, which confirms that this phenotype is attributable to Tal1 loss and is not an indirect effect of retroviral transduction or Cre expression. The failure of a DNA binding-defective mutant to rescue prolifer- ation in these cells, in contrast, shows that DNA-binding activity is as important for Tal1 function in this lineage as in erythroid progenitors (26).
In summary, these studies demonstrate that the bHLH transcription factor Tal1 is expressed throughout monocytopoiesis and that loss of Tal1 impairs the proliferation of late-stage MM precursors. This was associated with decreased exit from G 0 , delayed entrance into S from G 1 , and slowed traversal of S phase, resulting in part from the altered expression of a direct Tal1 target, the CDK inhibitor gene p16(Ink4a).