INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Changes in gene regulation are critical for proper cellular differentiation. Many transcription factors play important roles in controlling the processes needed for lineage development. In the B-cell lineage, a variety of transcription factors are necessary for proper development including PU.1, Pip, NF-B, Ikaros, Blimp-1, E2A, EBF, and BSAP/Pax-5 (reviewed in reference 16). PU.1 is an Ets family transcription factor specifically expressed in erythroid, myeloid, and B cells (25, 46). Gene-targeting experiments of the PU.1 locus in mice indicate that PU.1 is necessary for the development of granulocytes, monocytes, B cells, and T cells (30, 58). Exogenous addition of PU.1 DNA binding site oligonucleotides inhibits hematopoietic colony formation in vitro, whereas overexpression of PU.1 in erythroid cells can lead to erythroleukemia (34-36, 57, 66). PU.1 also appears to play a role in terminal differentiation within the macrophage lineage (14, 21, 38, 44) and may be crucial for the proper tissue-specific and temporal regulation of immunoglobulin  chain (Ig) gene rearrangement (20, 22). Interestingly, high levels of PU.1 protein in hematopoietic precursors appears to favor the myeloid lineages while low levels favor the B-cell lineage (60). Therefore, any mechanism that regulates PU.1 function could have a profound effect on lineage development. Indeed, functional antagonism between PU.1 and the hematopoietic transcription factors GATA-1 and GATA-2 can result in repressed transcriptional activity and altered erythroid cell development (54, 72). It is possible that transcription factors expressed in other lineages also interact with PU.1 and modulate its function.

BSAP (Pax-5) is a bifunctional transcription factor capable of repressing or activating transcription (reviewed in reference 4). Genes activated by BSAP include CD19, mb1, blk, RAG2, N-myc, LEF-1, CD72, and the Ig germ line promoter (10, 13, 19, 26-28, 37, 42, 43, 63, 64, 67, 71, 73, 74). In contrast, the Ig heavy-chain 3' enhancer element, the Ig J chain gene, the PD-1 gene, and the hXBP gene are repressed by BSAP (39, 40, 42, 53, 55, 61). BSAP expression in the hematopoietic lineage is limited to B lymphocytes, although BSAP is also expressed in some nonlymphoid tissues. BSAP is expressed throughout B-cell differentiation until the mature B-cell stage and subsequently ceases at the plasma cell stage. BSAP-deficient mice fail to develop B cells due to a block at an early stage (65). In adult bone marrow, the arrest is at the pro-B-cell stage, but the block is earlier in fetal liver (37, 43). Interestingly, lack of BSAP enables immature pro-B cells to differentiate into multiple hematopoietic lineages (41). Therefore, BSAP expression appears to suppress lineage alternatives and to direct the cell to the B-cell lineage. Similar to PU.1, control of BSAP function could thus be very important for hematopoietic development.

The immunoglobulin  chain 3' enhancer binds PU.1 and BSAP as well as c-fos, c-jun, Pip, CREM, ATF1, E2A, YY1, and SP1 (7, 23, 31, 45, 48, 49, 51, 52, 56, 59). A subset of the above proteins (PU.1, Pip, c-fos, and c-jun) can form a specific enhanceosome complex over the enhancer, which can activate the enhancer in non-B cells (50). During early B-cell development (pro-B- and pre-B-cell stages), the 3' enhancer is silent, whereas later in development (B-cell and plasma cell stages), the enhancer is active. The  locus, which is fundamentally important for proper B-cell development, has a similar pattern of transcriptional activity. Thus, understanding the mechanisms of regulating enhancer activity may reveal mechanisms critical for B-cell development.

We show here that BSAP can repress Ig 3' enhancer activity. This repression is mediated by targeting the transcriptional function of PU.1. We show a physical interaction between the PU.1 Ets and BSAP paired-box DNA binding domains, but this interaction does not affect PU.1 DNA binding. Instead, we found the BSAP can repress PU.1 transactivation by at least two mechanisms. When PU.1 function was assayed in isolation using a reporter construct containing a multimerized PU.1 binding site, a portion of the PU.1 transactivation domain (residues 7 to 30) was the target of BSAP-mediated repression. This repression by BSAP required the carboxy-terminal BSAP inhibitory domain. Interestingly, the coactivator protein p300 could overcome this BSAP-mediated repression of PU.1. In the context of the Ig 3' enhancer and the other enhancer binding proteins, we found that BSAP repressed PU.1 by a distinct mechanism. In this case, repression did not require the PU.1 transactivation or PEST domains and could not be overcome by p300 expression. Finally, using an artificial BSAP-inducible reporter, we also showed that PU.1 can repress BSAP transactivation and that this repression could be partially overcome by p300. The potential roles of BSAP-PU.1 antagonistic interactions in hematopoietic development are discussed.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Plasmid constructs. To prepare various full-length or mutant BSAP plasmids, a BSAP clone (a gift from Barbara Birshtein, Albert Einstein Medical School) was used as a template for PCR. Primers contained either the BSAP start or stop codons with adjacent EcoRI restriction sites (FEcoBSAPATG or REcoBSAPUGA). PCR products were extracted with phenol-chloroform, precipitated with ethanol, and digested with EcoRI. Digested fragments were purified by agarose gel electrophoresis using the Qiagen Qiaex kit and then ligated with either EcoRI-cut Bluescript KS(+), cytomegalovirus CMV expression plasmid (CB6+), or GEX-2TK to produce KS+BSAP1-391, CMV-BSAP1-391, or GST-BSAP, respectively. For BSAP deletion mutants, PCR was performed with the FEcoBSAPATG forward primer and various 3' primers (with HindIII sites [Table 1]) that amplified the appropriate sequences. Amplified products were digested with EcoRI and HindIII and ligated into EcoRI-HindIII-cut Bluescript KS(+), or CB6+ to produce the appropriate deletion clones (BSAP1-143, BSAP1-319, BSAP1-357, and BSAP2-143). Plasmid CMV-BSAP1-385 plus a serine-to-glycine mutation at residue 283 was a gift from John Monroe (University of Pennsylvania). Internal BSAP deletion mutants (144-303 and 228-259) were created by amplifying a full-length BSAP template with the FEcoBSAPATG forward primer and an RBSAP144-303 or REcoBSAPAUG reverse primer and with a FBSAP144-303 forward primer in separate reactions. The two products were gel purified (Qiaex) and used as template for a second PCR using FEcoBSAPATG and REcoBSAPAUG primers. Amplified products were gel purified, digested with EcoRI, and ligated into EcoRI-cut CMV. Plasmid CMV-p300 was a gift from Paul Liberman (Wistar Institute). Plasmids CMVPip, CMV-Fos, CMV-Jun, (Fos/Jun)₄LBKCAT, (PU.1/Pip)₄LBKCAT, CMVPU.17-30, and CMVPU.133-100 were described previously (50). CMV-PU.1 was described in reference 52. CMVPU.1118-160 was described in reference 47. Bluescript plasmids PU.17-30, PU.130-100, PU.1119-160, and PU.1245-272 were described in reference 51, and plasmids PU.1201-272 and PU.1255-272 were described in reference 47. CORELBKCAT and GST-PU.1 were described in reference 48. CORETKCAT (construct E), (PU.1/Pip)₄TKCAT (oligo 5), and (Fos/Jun)₄TKCAT (oligo 2) were described in reference 49. CMVGAL-PU.1 was a gift from Tom Kadesch (University of Pennsylvania). Plasmids CMVGALPU.1 1-160 and CMVGALPU.1 1-200 were prepared by PCR using plasmid CMVPU.1 as a template. Following synthesis, PCR products were digested with EcoRI and XbaI and ligated into EcoRI-XbaI-cut CMVGAL (3). The GALE1bCAT reporter was a gift from Robert Ricciardi (University of Pennsylvania). GALYY1:1-143 was described previously (3). To prepare (BSAP)₄(PU.1/Pip)₄TKCAT and (BSAP)₄(Fos/Jun)₄TKCAT, a HindIII sticky-blunt DNA fragment with the multimerized BSAP binding site from the Ig3' enhancer was ligated into the blunted SalI-sticky HindIII sites of plasmid (PU.1/Pip)₄TKCAT and (Fos/Jun)₄TKCAT, respectively. These plasmids were cut with BamHI and HindIII to release the regulatory sequences and ligated into BamHI-BglII-cut LBKCAT to produce (BSAP)₄(PU.1/Pip)₄LBKCAT and (BSAP)₄(Fos/Jun)₄LBKCAT, respectively. BSAP₂CAT was a gift from John Monroe.

Cell culture and transfections. S194 plasmacytoma cells were grown and transfected by the DEAE-dextran procedure as previously described (49). Each transfection contained 1 µg of a -galactosidase-expressing plasmid to normalize transfection efficiencies, 3 µg of the appropriate reporter plasmid, and 3 µg of either empty expression vector (CMV) or CMV-BSAP. NIH 3T3 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and were transfected by the calcium phosphate coprecipitation method (18). Cells were harvested 36 to 48 h posttransfection. Transfection efficiencies were initially normalized using -galactosidase activity and chloramphenicol acetyltransferase (CAT) assays followed by thin-layer chromatography were performed as described by Gorman et al. (17). The percent CAT activity was calculated by scintillation counting of the acetylated product and substrate spots. Transfections were performed three to five times, and the data shown are the averages of all transfections.

Preparation of GST fusion proteins. Escherichia coli BL21 cells containing the appropriate plasmid were inoculated overnight and diluted the following morning. Cultures were induced with 1 mM isopropyl--D-thiogalactopyranoside (IPTG) for 2 to 3 h. The cells were centrifuged, subjected to a freeze-thaw cycle, and resuspended in NETN (100 mM NaCl, 1 mM EDTA, 20 mM Tris-HCl [pH 8], 0.5% Nonidet P-40) containing lysozyme. After incubation for 20 min on ice, the cells received five 15-s sonication bursts. The solution was centrifuged at 2,000 rpm in a Sorvall RTH-750 rotor for 10 min, and the remaining supernatant was incubated at 4°C overnight with 1.5 ml of glutathione beads. The following morning, the beads were centrifuged and stored at 4°C.

GST chromatography. A 20-µl volume of solution containing glutathione S-transferase (GST) fusion protein or an equivalent amount of GST protein alone was incubated with 5 to 15 µl of recombinant protein made by in vitro transcription and translation (TNT kit; Promega) in a 100-µl reaction mixture containing NETN. Samples were rocked for 1 to 3 h at 4°C and washed three to five times with 300 µl of NETN. The samples were loaded onto denaturing polyacrylamide gels, run for 1 to 2 h at 150 V, dried, and subjected to autoradiography overnight.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

BSAP represses Ig 3' enhancer activity. BSAP, or a BSAP-like protein, can bind to the Ig 3' enhancer (56, 59). However, the role of BSAP in controlling Ig 3' enhancer function has never been elucidated. BSAP expression inversely correlates with Ig 3' enhancer function, suggesting that BSAP may negatively regulate enhancer activity. Therefore, we tested the ability of BSAP to regulate the expression of a CAT reporter gene driven by the thymidine kinase promoter adjacent to the Ig 3' enhancer (enhancer residues 390 to 523 [32]). This reporter (CORETKCAT) was transfected into S194 plasmacytoma cells with either empty expression vector (CMV) or a BSAP expression plasmid (CMV-BSAP). CORETKCAT activity was repressed two- to threefold by CMV-BSAP compared to empty vector alone (Fig. 1, lanes 1 and 2). Therefore, BSAP expression in plasmacytoma cells, which normally lack BSAP, appears to repress Ig 3' enhancer activity.

View larger version (33K): [in this window] [in a new window]   FIG. 1.   BSAP represses Ig 3' enhancer activity. The CORETKCAT reporter plasmid was transfected into S194 plasmacytoma cells either with empty CMV expression plasmid (lane 1) or with CMV-BSAP expression plasmid (lane 2). CAT activities of cell extracts from transfected cells are shown. Repression averaged 2.0- ± 0.4-fold in six independent experiments.

View larger version (59K): [in this window] [in a new window]   FIG. 2.   BSAP represses enhancer activity by inhibiting PU.1 function. (A) BSAP represses Ig 3' enhancer activity in NIH 3T3 cells. NIH 3T3 cells were transfected with the CORELBKCAT reporter, and plasmids expressing PU.1, Pip, c-Jun and c-Fos. The presence (+) or absence () of CMV-BSAP is indicated above the lanes. Repression averaged 5.3- ± 1.8-fold in five independent experiments. (B) PU.1 must be present for BSAP repression. NIH 3T3 cells were transfected with the CORELBKCAT reporter and the various expression plasmids shown below the histograms. Transfections were performed in either the absence (cross-hatched columns) or the presence (dark shaded columns) of CMV-BSAP. Percent activity is plotted, with 100% defined as the amount of activity in the absence of BSAP for each transfection mix. Error bars indicate standard deviation. Activities in the absence of PU.1 or Pip averaged about 20% of the activity with all four proteins, while removal of c-fos had a minimal effect on overall enhancer activity. (C) Map of the TKCAT constructs used for transfection. Locations of the multimerized BSAP, PU.1/Pip, and c-fos/c-jun binding sites are indicated upstream of the TK promoter driving the CAT gene. (D) BSAP represses the activity of a PU.1/Pip-dependent reporter but not a c-Fos/c-Jun-dependent reporter. NIH 3T3 cells were transfected with the reporter plasmids diagrammed in panel C. Lanes 1 to 4 received CMV-PU.1 and CMV-Pip, while lanes 5 to 8 received CMV-Fos and CMV-Jun. The presence or absence of CMV-BSAP in the transfections is indicated above each lane. BSAP repression of constructs 1 and 2 averaged 10-fold, while constructs 3 and 4 and the TKCAT vector were minimally affected. (E) NIH 3T3 cells were transfected with the LBKCAT reporter plasmids diagrammed below the histogram. The locations of the multimerized BSAP, PU.1/Pip, and c-fos/c-jun binding sites are indicated upstream of the LBK promoter driving the CAT gene. Reporter constructs 1 and 2 were cotransfected with CMV-PU.1 and CMV-Pip, whereas reporter constructs 3 and 4 were cotransfected with CMV-Fos and CMV-Jun. Transfections receiving CMV-BSAP are designated by +, and those receiving empty CMV vector are designated by . Percent activity is plotted, with 100% defined as the amount of activity in the absence of BSAP for each transfection mix. Error bars indicate standard deviation. (F) BSAP represses the activity of a PU.1/Pip-dependent reporter plasmid in B cells. S194 plasmacytoma cells were transfected with a PU.1/Pip-dependent reporter plasmid (PU.1/Pip)4LBKCAT and either empty CMV vector (lane 1) or CMV-BSAP (lane 2). BSAP repressed activity eightfold.

BSAP repression targets PU.1 activity. We next sought to identify the protein target of BSAP repression. If BSAP targets a specific enhancer binding protein, removal of that component from the transfection mix should alleviate the repression. Thus, we performed transfections in which we systematically removed individual transcription factor components (PU.1, Pip, c-fos, or c-jun) from the transfection. This resulted in reduced, but still relatively robust transcriptional activity. Removal of c-fos had no effect on the ability of BSAP to repress enhancer activity (Fig. 2B, compare lane 1 with lane 4). Similarly, BSAP repressed enhancer activity in the absence of Pip (compare lane 1 with lane 3). However, removal of PU.1 greatly reduced the ability of BSAP to repress transcription (compare lane 1 with lane 2). Absence of c-jun resulted in complete loss of enhancer activity (data not shown), thus preventing us from determining its role in BSAP repression.

The PU.1 Ets domain physically interacts with BSAP. Since PU.1 was clearly a target of the BSAP-mediated repression but BSAP did not need to bind to DNA for repression, we sought to determine if BSAP could physically interact with PU.1. We tested whether PU.1 protein prepared by in vitro translation could interact with bacterially produced GST-BSAP. Indeed, a strong interaction was observed between BSAP and PU.1 (Fig. 3A, lane 6). Minimal PU.1 binding was observed with the GST beads alone (lane 5).

View larger version (47K): [in this window] [in a new window]   FIG. 3.   BSAP physically interacts with PU.1 through the PU.1 Ets domain. (A and B) PU.1 and various PU.1 deletion mutants were prepared by in vitro translation. Translated products were incubated with either GST or GST-BSAP as indicated above the lanes. The PU.1 samples used in each assay are indicated at the top. After incubation, samples were washed and subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis; 20% input samples are shown in lanes 1 to 4. (C) PU.1 residues 201 to 255 are necessary for interaction with BSAP. A diagram of each PU.1 mutant is shown. Sequences deleted are indicated on the left, and the ability to physically bind to GST-BSAP is indicated on the right.

BSAP requires the paired-box domain for in vitro interaction with PU.1. We conducted reciprocal protein-protein interaction assays to confirm the above results and to localize the region of BSAP involved in physical interaction with PU.1. Wild-type BSAP protein prepared by in vitro translation was incubated with bacterially produced GST-PU.1 protein. As expected, a specific interaction between BSAP and GST-PU.1 was detected (Fig. 4A, lanes 5 and 6).

View larger version (54K): [in this window] [in a new window]   FIG. 4.   The BSAP paired-box domain is necessary and sufficient for interaction with PU.1. (A) BSAP and various BSAP deletion mutants were prepared by in vitro translation. Translated products were incubated with either GST or GST-PU.1 as indicated above the lanes. The BSAP samples used in each assay are indicated at the top. Lanes 11 and 12 received unprogrammed rabbit reticulocyte lysate. (B) PU.1 prepared by in vitro translation was incubated with either GST-BSAP 1-391 or GST-BSAP 1-143. After incubation, samples were washed and subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis; 20% input samples are shown in lanes 1 to 4 of panel A and lane 1 of panel B. (C) Results of the interaction studies are summarized. The locations of the BSAP paired-box domain, homeodomain (HD), activation domain (AD), and inhibitory domain (Inh) are indicated. The ability of each BSAP construct to interact with PU.1 is shown on the right.

BSAP does not inhibit PU.1 DNA binding. Since BSAP interacts with PU.1 through the Ets DNA binding domain, we first sought to determine if BSAP inhibited PU.1 DNA binding. Electrophoretic mobility shift assays did not reveal an effect of BSAP on PU.1 DNA binding (data not shown), suggesting that displacement of PU.1 from the DNA was not part of the repression mechanism by BSAP. If the BSAP repression mechanism does not involve inhibition of PU.1 DNA binding, then a chimeric PU.1 protein consisting of PU.1 fused to a heterologous DNA binding domain should also be repressed by BSAP. We fused the GAL4 heterologous DNA binding domain (residues 1 to 147) to PU.1. The fusion protein was cotransfected with a GAL4-dependent reporter (GALE1bCAT), and activity was measured in the presence or absence of BSAP. Indeed, BSAP repressed GAL-PU.1 transactivation, indicating that DNA binding through the PU.1 Ets domain is not involved in the repression mechanism (Fig. 5A, lane 1).

View larger version (37K): [in this window] [in a new window]   FIG. 5.   BSAP repression does not involve the Ets domain and targets a portion of the PU.1 transactivation domain. (A) BSAP-mediated repression does not require the PU.1 Ets domain. NIH 3T3 cells were transfected with a GAL4-responsive reporter (GALE1bCAT) and various GAL-PU.1 or GAL-YY1 fusion constructs. Fold repression in the presence of BSAP is plotted. Histograms that fall below the horizontal line represent no repression by BSAP. Error bars indicate standard deviation. Basal activities of GALPU.1 1-160 and GALPU.1 1-200 were 13- and 7-fold higher, respectively, than that of GALPU.1 1-272. GALYY1 1-143 activity was 1.7-fold higher than that of GALPU.1 1-272. (B) BSAP repression targets a portion of the PU.1 transactivation domain. NIH 3T3 cells were transfected with the (PU.1/Pip)4LBKCAT reporter and plasmids expressing wild-type PU.1 or various PU.1 deletion mutants. Fold repression in the presence of BSAP is plotted. Error bars indicate standard deviation. PU.17-30 showed 20% of the activity of wild-type PU.1, whereas PU.133-100 and PU.1118-160 were equivalent to wild-type PU.1. (C) Summary of BSAP repression data. The positions of the PU.1 activation domain (AD), PEST domain, and DNA binding Ets domain are indicated. The ability of BSAP to repress each construct is indicated on the right.

The amino-terminal region of PU.1 is required for the BSAP repression mechanism. To identify the PU.1 sequences that are the target of BSAP repression, several deletion mutants were tested. These deletions (7-30, 33-100, and 119-160) can still activate a PU.1-dependent reporter, (PU.1/Pip)₄ LBKCAT, in NIH 3T3 cells. Wild-type or individual PU.1 mutants were transfected in either the presence or absence of BSAP, and the repression was measured. The activities of PU.133-100 and PU.1119-160 were repressed to levels similar to, or higher than, that of wild-type PU.1 (Fig. 5B, compare lane 1 with lanes 3 and 4). However, deletion of PU.1 amino acids 7 to 30 abolished the ability of BSAP to repress transcription (lane 2). These data indicate that PU.1 amino acids 7 to 30 are a target for the BSAP repression mechanism. The above results are summarized in Fig. 5C.

The carboxy-terminal portion of BSAP is necessary for repression. We sought to determine the BSAP sequences necessary for repression of PU.1 transcriptional activity. Using mutant BSAP constructs, transient transfections were conducted comparing the level of repression of these mutants to that of wild-type BSAP. Expression of full-length BSAP (WT BSAP 1-391) resulted in significant levels of repression (Fig. 6A, lane 2). Expression of the paired box domain alone (BSAP 1-143) was not capable of mediating BSAP repression, again confirming that physical interaction of PU.1 with BSAP is not sufficient for repression (lane 8). Deletion of amino acids 358 to 391, which constitute the inhibitory domain (lane 4), yielded a protein that had lost the ability to repress the activity of PU.1. However, another BSAP mutant protein which has an amino acid change (S to G) at residue 283 and a frameshift at amino acid 385 which deletes the C-terminal 6 amino acids but adds 25 heterologous residues (BSAP 1-385;S283G) repressed PU.1 activity (lane 3). A larger carboxy-terminal deletion removing amino acids 320 to 391 did not repress PU.1 activity (lane 5). An internal deletion (228-259) that deletes the internal homeodomain, and which was recently demonstrated by Eberhard and Busslinger (12) to interact with TBP and RB, also repressed PU.1 activity (lane 6). A larger internal deletion (BSAP144-303), which deletes all residues between the paired-box domain and the BSAP transactivation domain, still repressed PU.1 activity (lane 7). We also tested the repression properties of several BSAP constructs fused to the GAL4 heterologous DNA binding domain. Full-length BSAP (GAL-BSAP 1-391) repressed PU.1 transactivation (lane 9). Deletion of the BSAP inhibitory domain (GAL-BSAP 1-357) abolished BSAP repression (lane 10). Finally, an amino-terminal mutant which removes the paired-box domain (GAL-BSAP 144-391) repressed PU.1 transactivation (lane 11). Western blots of nuclear extracts isolated from transfected cells showed that the appropriate mutant BSAP proteins were expressed in vivo (data not shown). The BSAP constructs used above and their ability to repress PU.1 activity are summarized in Fig. 6B.

View larger version (4928K): [in this window] [in a new window]   FIG. 6.   (A) The BSAP inhibitory domain is necessary for repression of PU.1 activity. NIH 3T3 cells were transfected with the (PU.1/Pip)4LBKCAT reporter and CMV-PU.1. Various BSAP expression plasmids included in the transfection are indicated below each histogram. CAT activities of cell extracts from transfected cells are shown, and error bars indicate standard deviation. (B) Maps of BSAP constructs used for transfection. The positions of the BSAP paired-box domain, homeodomain (HD), activation domain (AD), and inhibitory domain (Inh) are indicated. The ability of each BSAP construct to repress PU.1 transactivation is shown on the right.

Coactivator protein p300 can reverse BSAP repression. Previously, it was shown that c-jun and CBP can function as coactivators of PU.1 activity (2, 70). Repression by BSAP could be due to BSAP targeting some function of a PU.1 coactivator protein. Our results described above showed that c-jun is not a target of BSAP repression (Fig. 2D and E), but coactivators with histone acetyltransferase (HAT) activity could possibly be part of the repression mechanism. We tested a variety of HAT proteins for their ability to function with PU.1 and found that p300 stimulated PU.1 transcriptional activity most strongly (Y. Bai and M. Atchison, unpublished data). Therefore, we tested whether the coactivator p300 could abolish the repression mediated by BSAP. Transient-expression assays were performed with PU.1 alone, PU.1 plus BSAP, or PU.1 plus BSAP and increasing doses of p300. Indeed, we found that p300 could completely reverse the inhibition mediated by BSAP (Fig. 7). Therefore, the repression mechanism of BSAP apparently involves disrupting the function of p300 with PU.1.

View larger version (31K): [in this window] [in a new window]   FIG. 7.   p300 can reverse the BSAP inhibition of PU.1 transactivation. (A) NIH 3T3 cells were transfected with the (PU.1/Pip)4LBKCAT reporter and the various expression plasmids indicated above each lane. BSAP represents clone BSAP 228-254. Lanes 4 to 6 contain 50 ng, 2.5 µg, and 5.0 µg of p300 expression plasmid, respectively. (B) Histogram of the p300 reversal of BSAP repression experiment in panel A. A duplicate experiment showed the same results.

PU.1 can repress BSAP transactivation. While our results described above indicated that BSAP could repress PU.1 function, in other systems PU.1 antagonizes the function of other transcription factors. For instance, PU.1 can inhibit the function of the transcription factor GATA-1, resulting in inhibition of erythroid cell differentiation (54, 72). Therefore, we set out to determine whether PU.1 could inhibit BSAP function. BSAP can activate the synthetic reporter construct (BSAP)₂CAT. We transfected the (BSAP)₂CAT reporter with BSAP in the presence or absence of PU.1. Indeed, we found that PU.1 repressed BSAP transcriptional activity (Fig. 8). Interestingly, deletion of the PU.1 target sequences necessary for BSAP-mediated repression of PU.1 (7-30) greatly reduced the ability of PU.1 to repress BSAP function (Fig. 8). This suggested that PU.1 could be repressing BSAP by targeting p300. If this is true, exogenous p300 should reverse the PU.1-mediated repression of BSAP activity. Cotransfection of p300 partially reversed the PU.1-mediated repression (Fig. 8, lane 4). Therefore, PU.1 and BSAP can antagonize each other by targeting p300 function.

View larger version (24K): [in this window] [in a new window]   FIG. 8.   PU.1 can inhibit BSAP transactivation. NIH 3T3 cells were transfected with the (BSAP)2CAT reporter and various expression plasmids shown below the histogram. The activity of the reporter plasmid and BSAP expression vector is defined as 100%. Error bars indicate standard deviation.

BSAP represses PU.1 function by a distinct mechanism in the context of the Ig 3' enhancer. PU.1 can function in a variety of contexts within various promoters and enhancers. Our results described above indicate that when PU.1 function is assayed in isolation (i.e., in the absence of other DNA binding proteins), BSAP repression requires PU.1 sequences 7 to 30 and involves disruption of p300 function. In the context of the Ig 3' enhancer, however, we previously found that PU.1 sequences 7 to 30 are dispensable for mediating enhancer activity (50). However, BSAP represses PU.1 function in the context of this enhancer (Fig. 1 and 2). This suggested that BSAP may repress PU.1 function by a distinct mechanism in the context of the Ig 3' enhancer. To test this, we performed transfections with the CORELBKCAT reporter and plasmids expressing Pip, c-jun, c-fos, and various mutant PU.1 proteins. As shown in Fig. 2B, BSAP repressed transcription in the presence of wild-type PU.1 (Fig. 9A). However, in contrast to our results with a multimerized PU.1 binding site, 3' enhancer activity was repressed by BSAP in the presence of PU.17-30 (Fig. 9A). Repression was also observed with PU.1 mutants 33-100, 2-118, and 118-168 (Fig. 9A). Thus, BSAP repressed PU.1 function even in the complete absence of the PU.1 transactivation domain (construct 2-118) or the PEST domain (118-168).

View larger version (40K): [in this window] [in a new window]   FIG. 9.   BSAP represses PU.1 function by a distinct mechanism in the context of the Ig 3' enhancer. (A) The PU.1 transactivation and PEST domains are dispensable for repression. NIH 3T3 cells were transfected with the CORELBKCAT reporter and plasmids expressing Pip, c-Jun, c-Fos, and various PU.1 mutants. The presence (+) or absence () of each transcription factor is indicated below the lanes. For each transfection mix, activity in the absence of BSAP is defined as 100%. In the presence of Pip, c-jun, and c-fos, the PU.1 mutants supported similar enhancer activity. Error bars indicate standard deviation. (B) Coactivator p300 cannot reverse BSAP repression in the context of the 3' enhancer. The presence (+) of each transcription factor is indicated below the lanes. Error bars indicate standard deviation. (C) The BSAP inhibitory domain is necessary for complete repression. The presence (+) of each transcription factor is indicated below the lanes. Error bars indicate standard deviation.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

We found that BSAP can repress the activity of the Ig 3' enhancer in plasmacytoma cells or in NIH 3T3 cells after stimulation by transcription factors PU.1, Pip, c-fos, and c-jun. PU.1 is a target of BSAP repression based on a variety of criteria. First, removal of PU.1 from the transfection mix greatly reduced BSAP repression. Second, a PU.1/Pip-dependent reporter, but not a c-fos/c-jun-dependent reporter, was repressed by BSAP. Third, GAL-PU.1 but not GAL-YY1 activation was repressed by BSAP. We found that BSAP sequences 358 to 385, which constitute the BSAP inhibitory domain (11), are needed for repression. Interestingly, we found that BSAP can repress PU.1 function by at least two different mechanisms depending upon the context of PU.1. PU.1 sequences 7 to 30, which include a portion of the PU.1 transactivation domain (24), are the targets of BSAP-mediated repression when PU.1 function is assayed in isolation (i.e., on a multimerized PU.1 binding site). This repression mechanism involves the coactivator protein p300. Furthermore, we found that PU.1 can repress BSAP transcriptional activation and that this repression also requires PU.1 sequences 7 to 30 and can be partially reversed by p300.

In the context of the Ig 3' enhancer, however, BSAP repression did not require the PU.1 transactivation or PEST domains and repression could not be reversed by p300. Based on transfections performed in the absence of Pip (Fig. 2B and 9B), this repression mechanism appears to disrupt synergy between PU.1 and c-jun/c-fos. Interestingly, Zhang et al. (72) also found that GATA-1 and GATA-2 can disrupt PU.1-c-jun synergy. Functional repression by GATA-1 and GATA-2 coincided with disruption of PU.1-c-jun physical interactions. Interestingly, PU.1 interacts with c-jun, GATA-1, and GATA-2 via the PU.1 Ets domain 3/4 region (72). This is the PU.1 region that we observed to bind to BSAP (Fig. 3). Therefore, it is likely that, similar to GATA-1 and GATA-2, BSAP can disrupt interactions between PU.1 and c-jun, resulting in transcriptional repression. However, it should be noted that the BSAP paired-box domain, which physically interacts with PU.1, is incapable of mediating repression. It is also interesting that, similar to our case with BSAP, the GATA factors do not inhibit PU.1 DNA binding (72).

BSAP has recently been shown to interact with regulatory proteins including TATA binding protein, the underphosphorylated form of the retinoblastoma protein, AML1, and Ets-1 (12, 15, 29). The TATA binding protein and retinoblastoma protein interactions require the internal homeodomain of BSAP. However, we found that these sequences are unnecessary for BSAP-mediated repression of PU.1 activity. In the case of the Ets-1 interaction, Fitzsimmons et al. (15) showed that in the context of the mb-1 promoter, BSAP can form a ternary complex with specific Ets family proteins and that this ternary complex is necessary for efficient promoter activity. Ternary-complex formation requires only the BSAP paired-box domain. Although we demonstrated that PU.1 can physically interact with the BSAP paired-box domain through its own Ets domain, PU.1 lacks the critical amino acids required for ternary-complex formation (69). In addition, the BSAP paired-box domain alone in insufficient for both of the repression mechanisms that we observed here. Instead, BSAP regulation of PU.1 transcriptional activity is distinct from the previously demonstrated BSAP regulatory mechanisms observed with other Ets proteins. Indeed, rather than activating expression, BSAP inhibits PU.1 function.

Dorfler and Busslinger (11) found that the BSAP inhibitory domain can inhibit or mask the activity of the BSAP transcriptional activation domain. The inhibitory sequences appeared to function as part of a unit including the adjacent activation domain, because inhibitory-domain repression was not transferable to a heterologous DNA binding domain (11). Our data, however, indicate an active role for the inhibitory domain. Rather than passively masking the BSAP activation domain, the BSAP inhibitory domain actively repressed PU.1 activity. Our cotransfection studies indicated that at least one functional target of the BSAP inhibitory domain is the coactivator protein, p300.

It is interesting that BSAP also negatively regulates the expression of the Ig J-chain gene promoter (55) and the activity of the Ig heavy-chain 3'  enhancer (33, 39, 40, 61). Both of these regulatory elements contain PU.1 binding sites which could be the targets of BSAP repression. Both of these regulatory elements exhibit expression patterns similar to the Ig 3' enhancer. The Ig heavy-chain 3'  enhancer, in particular, mirrors Ig 3' enhancer activity. Both enhancers are most active at late stages of B-cell development (plasma cell stage) and are inactive early in B-cell development (pro-B- and pre-B-cell stages). It will be very interesting to determine whether these two enhancers use similar mechanisms to control activity.

The functional competitions between PU.1, BSAP, p300, and c-jun that we observed here could be important for hematopoietic lineage development. PU.1 is necessary for myeloid and lymphoid cell development, whereas BSAP is necessary for B-cell development past the pro-B-cell stage (9, 14, 21, 30, 38, 44, 58). Interestingly, PU.1 does not appear to be necessary for myeloid lineage commitment but, rather, is needed for further development after commitment to the lineage has been initiated (21). BSAP may play a role in this lineage development by controlling the level of PU.1 activity and thereby reducing the plasticity of early B cells to differentiate into myeloid cells. Since PU.1 is known to influence macrophage proliferation (5), BSAP expression in early B-cell precursors could limit the expansion of cells into the myeloid pathway by inhibiting PU.1 function. Indeed, BSAP appears to reduce the clonal expansion of early myeloid cells (M. Chiang and J. Monroe, personal communication). On the other hand, PU.1 could limit the ability of BSAP to drive B-cell differentiation toward later stages. Absence of BSAP in pro-B cells enables cells at early stages of lymphopoiesis to "reverse" their differentiated state and to progress into other hematopoietic lineages (41). Inhibition of BSAP function by PU.1 could therefore drive cells toward the myeloid lineages. This would be analogous to the antagonistic interactions between PU.1 and GATA-1 during erythroid cell differentiation (54, 72). Therefore, functional interplay between PU.1, BSAP, p300, and c-jun could participate in the mechanisms for controlling hematopoietic lineage development.

A variety of regulatory options are possible for controlling the functional interactions between PU.1 and BSAP. Expression levels of either protein might be modulated. For instance, interleukin-2 can down-regulate BSAP expression, thereby relieving BSAP repression of the Ig J-chain gene (55). Similarly, BSAP expression is silenced in mice lacking the interleukin-7 receptor (6). BSAP or PU.1 could also be functionally regulated. The DNA binding function of BSAP may be altered by mitogenic stimuli, by reduction potential, or during B-cell development (59, 62, 68). Activity of the BSAP inhibitory domain, which we found is necessary for PU.1 repression, may also be regulated (11). This segment of BSAP is very rich in serine and tyrosine, and its function could potentially be differentially regulated by phosphorylation during B-cell differentiation. On the other hand, PU.1 functional interactions might be regulated. For example, phosphorylation of PU.1 serine 148 is known to regulate its ability to recruit transcription factor Pip to DNA (52). Similar posttranslational modifications might control the ability of PU.1 to interact with BSAP or p300. Thus, control of lineage development may depend upon signals that regulate the ability of either BSAP or PU.1 to repress one another's transcriptional activity. Additional experiments are necessary to elucidate these possible regulatory mechanisms.