The Transcription Factor NF-κB/p50 Interacts with the blkGene during B Cell Activation*

The B cell-specific transcription factor Pax-5 has been shown previously to interact with the promoter of theblk gene in vitro. blk encodes a tyrosine kinase associated with the B cell receptor, which is expressed during the early but not the final stages of B cell development. To investigate whether Pax-5 regulates expression of the blkgene in vivo during B cell development and/or activation,Pax-5a was overexpressed in B cell lines. Increases inblk promoter activity using a chloramphenicol acetyltransferase reporter gene system suggested a role for Pax-5a as a transcriptional activator. Subsequent site-specific mutagenesis studies showed that mutations of the Pax-5 binding site on blksignificantly alter promoter activity, although results suggested that other factors could bind to this region as well. Using mobility shift assays, we detected an inducible transcription factor that interacts strongly with a sequence overlapping the Pax-5 site on theblk promoter and identified this as a homodimer of NF-κB/p50, a member of the NF-κB/Rel family of transcription factors. This factor was present at high levels in lipopolysaccharide-activated normal B cells and in plasma cell lines but either at low levels or undetectable levels in resting normal B cells or pre-B or mature B cell lines. In contrast, lipopolysaccharide induction of a pre-B cell line (703/Z) induced a complex that contained both NF-κB/p50 and p65. These studies suggest that different NF-κB complexes are able to interact with a sequence overlapping the Pax-5 site on the blk promoter and that the relative levels of “bound” factor influence levels of blk expression. Since p50 homodimers and p50/p65 heterodimers of the NF-κB complex should have opposing effects on blk transcription, this could provide a mechanism to differentially regulate blkexpression during B cell development and activation.

The activation and subsequent proliferation and differentiation of mature B cells into plasma cells is perhaps the most critical event within the B cell differentiation pathway, because it leads to the production of antibodies necessary for the elimination of foreign antigens. B cell activation is typically initiated by cross-linking of the B cell antigen receptor complex, but activation signals can also be generated through binding of certain cytokines or the B cell mitogen lipopolysaccharide (LPS). 1 Activation involves a cascade of events associated with proliferation and differentiation, eventually resulting in the presence of plasma cells, which secrete large amounts of antibodies and memory B cells. In order to elucidate the molecular pathways involved in B cell development and activation, we have taken the approach of investigating specific transcription factors and the roles they play in the modulation of gene activation and/or repression during these processes.
A number of Src family kinases have been reported to be associated with the B cell receptor complex, including the B cell-specific Blk kinase, encoded by the blk gene (1). In this study, we use blk as a B cell-specific target gene whose expression is regulated, at least in part, by transcription initiation (1). Blk transcripts are present in all pro-B, pre-B, and mature B cell line studied but are undetectable at the plasma cell stage (1). As has been observed in many developmentally regulated genes, the 5Ј-flank of the blk gene contains at least four transcription start sites but no functional TATA box (1). Earlier work indicated that a 320-nucleotide (nt) promoter region of blk can promote transcription of a heterologous gene after transfection into B cell lines (2), although its activity is low. At least one B cell-specific transcription factor, the B cell-specific activator protein (BSAP) has been shown to interact specifically with a 5Ј blk sequence in vitro (2,3).
BSAP is encoded by the Pax-5 gene and is a member of the paired domain family of transcription factors. BSAP (also named Pax-5) protein has been detected in B cells, adult testis, and the developing midbrain (3)(4)(5). Within the B cell lineage, Pax-5 transcripts and its encoded products have been detected in pro-B, pre-B, and mature B cell lines, whereas either very low or undetectable levels were found in the various plasma cell lines studied (3)(4)(5). Pax-5 is alternatively spliced, and at least four isoforms have been detected in nuclear extracts of B cell lines, although the expression levels of the full-length form, Pax-5a, are higher then any of the alternatively spliced forms Pax-5b, -5d, or -5e (3). The amino acid sequence of isoform Pax-5a is identical to that of BSAP (3). 2 The in vivo function of Pax-5 has been investigated through targeted gene disruption in mice (6). Homozygous mutant mice display a complete developmental block at the late pro-B cell stage, suggesting essential regulatory functions in early B cell development. Many putative target sequences for Pax-5 have been reported, including the promoters of CD19, CD20, 5, V pre-B , J-chain, blk, and mb-1 (2,4,(7)(8)(9)(10)(11) as well as the 3Ј-␣ enhancers and switch regions of immunoglobulin (Ig) heavy chain genes (12,13). Based on the expression pattern of Pax-5 target genes, it is likely that Pax-5 is required during later B cell stages as well. Some evidence to support this comes from a recent report showing that overexpression of isoform Pax-5a in plasma cell lines can repress expression of two of its target genes, the J-chain and the Ig heavy chain gene (10,14,15).
Recent studies suggest that Pax-5a can act as an activator, a docking protein, or a repressor, depending on the target gene and possibly the developmental state of the B cell. For example, the expression patterns of the CD19 and Pax-5 genes in B cells are identical, which suggests that Pax-5 is a positive regulator of the CD19 gene (4). This possibility is further supported by the observation that CD19 expression has ceased completely in homozygous Pax-5 mutant mice (6). In the regulation of the B cell-specific mb-1 gene, Pax-5a functions as a docking protein to recruit a group of more widely expressed transcriptional activators, the Ets proteins (16). In contrast, a suppressor function has been suggested by studies of both J-chain and Ig heavy chain genes, although the exact mechanisms by which this is accomplished by Pax-5a are not well understood (10,12,14,17,18). In regard to the regulation of blk gene expression, it has not yet been established what in vivo role, if any, Pax-5a plays. Notably, the levels of blk RNA in the pro-B cells from homozygous Pax-5 knockout mice were reportedly identical to those of their wild type (WT) littermates, which either implies that Pax-5a does not regulate blk levels during the earliest stages of B cell development or that the absence of Pax-5a in knockout mice leads to an artificial situation of redundancy in which other factors can take over Pax-5a functions.
In this study, we wished to investigate whether Pax-5a influences blk expression during B cell development and/or activation. We show here that overexpression of Pax-5a in B cell lines leads to an increase of blk promoter activity as measured using a reporter gene system, which suggests that it acts as a transcriptional activator. Subsequent site-specific mutagenesis studies of the blk promoter showed that mutations of the Pax-5 binding site can significantly alter blk promoter activity in vivo, although the results suggest that other factors with binding sites close to or even overlapping with the Pax-5 site may influence blk transcription as well. Using normal B cells activated with LPS, we subsequently identified an inducible transcription factor that binds strongly to a sequence overlapping the Pax-5 site in activated B cells.
This inducible factor was identified as a homodimer of NF-B/p50, a member of the NF-B/Rel family of transcription factors that in mammals contains the protein members p50, p65, p52, c-Rel, and RelB; these subunits can form either homodimers or heterodimers (19 -22). NF-B is present in the cytoplasm during periods when it is not needed, forming an inactive ternary complex with the so-called inhibitory (IB) factor, which prevents NF-B from moving into the nucleus. The NF-B pathway can be activated in B cells by IgM crosslinking or by the addition of LPS (19 -22). In such situations, the IB protein is degraded, and the remaining (active) NF-B dimer is translocated to the nucleus, providing a system that is able to respond very rapidly to cell activation events. Interestingly, p50 homodimers lack a transactivating domain, and it has been implied that such dimers are involved in repression of transcription initiation (22)(23)(24)(25). In contrast, p50 heterodimers are thought to be involved in transcriptional activation (22)(23)(24)(25).
Unexpectedly, LPS induction of a pre-B cell line 703/Z did not induce formation of the p50 homodimer complex but instead induced a complex containing both p50 and p65. Further analysis in B cell lines representing different stages of B cell differentiation indicated that the relative levels of p50 homodimers, p50 heterodimers, and Pax-5a bound to the blk promoter strongly influence the levels of blk transcripts in these lines.
These studies suggest that different NF-B complexes are able to interact with the blk promoter and that these complexes compete for binding with Pax-5a. Because NF-B/p50 homodimers and heterodimers have opposite effects on transcription initiation, this may provide a mechanism to differentially regulate blk expression after cell activation, depending on the developmental state of the B cell.
DNA Constructs-The construction of the p(Ϫ191)blkCAT2E reporter construct has been described elsewhere (2). A promoterless chloramphenicol acetyltransferase (CAT) construct, ␦CAT2E, was derived from p(Ϫ191)blkCAT2 E by restriction digestion with the endonucleases StuI and SalI, resulting in the specific deletion of blk sequence only. The effector construct pcDNA3.Pax-5a was made by digesting both the expression vector pcDNA3 (Invitrogen) and the pBS-1.2 plasmid (2), containing the full-length cDNA sequence of Pax-5a, with the restriction endonuclease NotI, followed by gel purification of the 1.2kilobase pair Pax-5a insert. The NotI insert was then cloned directly into calf intestine phosphatase-treated, NotI-digested pcDNA3, and the orientations of the resulting clones were determined using dideoxy sequencing. This resulted in clone pcDNA3.Pax-5a, which expressed Pax-5a protein as determined by in vitro transcription and translation in the presence of [ 35 S]methionine followed by SDS-polyacrylamide gel electrophoresis and radiography. The pcDNA3 construct was used as a negative control effector construct, and the HBIICAT construct (2) was used as a control for transfection efficiency.
Transfections and CAT Assays-The B cell lines A20/2J, M12, and S194 were transiently transfected using DEAE-dextran, and chloramphenicol acetyl transferase was assayed as described elsewhere (2). The generation of clones S194-Tf1 and S194-Tf4, which are stable transfectants containing the full-length isoform Pax-5a in S194 (clone Tf1 expresses low levels of Pax-5a; clone Tf4 expresses high levels) has been described elsewhere (10,14). The relative CAT conversion was determined by calculating the ratio of the acetylated to the acetylated plus unacetylated counts per minute, subtracting the activity of the mock transfection (transfection without DNA), followed by subtraction of the promoterless construct (␦CAT2E) and normalization for protein content as determined by the Bradford assay. To enable comparison of different cell lines, values were normalized for transfection efficiency by dividing each activity by the relative CAT conversion of the HBIICAT control construct (see above). Relative CAT activity values for each transfection were calculated as -fold change compared with the activity of the wild type (Ϫ191)blkCAT2E construct.
In Vitro Transcription and Translation of Pax-5a-The plasmid (pBluescript) pBS-1.2 (containing full-length Pax-5a cDNA; see Ref. 2) was transcribed in the sense direction with T3 RNA polymerase. Translation was carried out using a rabbit reticulocyte lysate (Promega) according to the supplier's directions. Between 1 and 2 l of extract was used for mobility shift assays.
Antibodies-To generate an antibody against Pax-5a isoforms, the paired domain sequence of the Pax-5 gene was cloned downstream from the glutathione S-transferase sequence in the vector pGEX2T (Amersham Pharmacia Biotech). The fusion protein containing 148 amino acids of the paired domain sequence of Pax-5 was purified using glutathione-agarose beads. New Zealand White rabbits were immunized intradermally with 500 g of purified protein, followed by five booster immunizations of 100 g each that were administered every other week beginning 21 days after the initial immunization. Antiserum, named Ed-1, was obtained at 12 weeks, analyzed by enzyme-linked immunosorbent assay, and used at a 1:2000 dilution in Western blot analysis or at 1:5 or 1:10 dilution for the mobility shift assays. Antibodies to p65, p50, and Sp1 were purchased from Santa Cruz Biotechnology, Inc., and used at a 1:5000 dilution in Western blot analysis or 1:5 in mobility shift assays. The anti-p65, anti-Pax5, and anti-Sp1 antibodies were detected using a horseradish peroxidase-conjugated donkey-anti-rabbit secondary antibody (Amersham Pharmacia Biotech). The p50 antiserum was directed against the nuclear localization domain of p50, and a rabbit anti-goat, horseradish peroxidase-conjugated antibody was used as its secondary antibody in Western blot analysis (Amersham Pharmacia Biotech).
Isolation of Normal B Cells-B cells were isolated from spleens from 6 -8-week-old BALB/c mice. After removal of the spleens, cells were resuspended and washed in Hanks' balanced salt solution. Small resting B cells were isolated from a Percoll gradient (Amersham Pharmacia Biotech) according to the supplier's instructions. This population of cells, containing 75% Ig-positive cells, 24% Thy-1-positive cells, and 1% non-lymphocytes, was either processed immediately to obtain nuclear extracts or used for activation experiments by growing in complete RPMI medium (as above) including 20 g/ml LPS.
Western Blot Analysis-Nuclear and cytoplasmic extracts from B cell lines or normal B cells were prepared as described elsewhere (10). Extracts were separated on 12% SDS-polyacrylamide gels and electrophoretically transferred onto nitrocellulose filters (Schleicher and Schuell). Membranes were first incubated for 2 h in blocking solution of phosphate-buffered saline (154 mM NaCl, 1.9 mM NaH 2 PO4, 8.1 mM Na 2 HPO 4 , pH 7.3) containing 5% nonfat milk, followed by a 1-h incubation with the primary antiserum in blocking solution. Next, membranes were washed tree times for 10 min in phosphate-buffered saline, followed by a 1-h incubation in the presence of a secondary, horseradish peroxidase-conjugated antibody (1:10,000; Amersham Pharmacia Biotech). Membranes were then washed three times for 10 min in phosphate-buffered saline, the blots were developed with an enhanced chemiluminescence kit (ECL; Amersham Pharmacia Biotech), and bands were visualized on Kodak XAR5 film for 1 or 10 min.
Electrophoretic Mobility Shift Assays-Binding assays were carried out for 20 min at 30°C in 10-l reactions containing 60 mM KCl, 12 mM HEPES, pH 7.9, 4 mM Tris-Cl, pH 7.9, 1 mM EDTA, 1 mM dithiothreitol, 30 ng of BSA, 12% glycerol, 1-5 g of nuclear extract, 0.1 ng of 32 Plabeled DNA probe, and 1-5 g of poly(dI⅐dC). The F3 DNA probe was identical to the one used in earlier studies (2, 3) and spans nt Ϫ70 to ϩ136 of the murine blk promoter. The double-stranded blk.BSAP oligonucleotide probes (wild type and mutants) were made by annealing sense and antisense oligonucleotides spanning nt Ϫ70 to Ϫ39 of the murine blk promoter and labeled using [ 32 P]␣-dCTP as described previously (2,3). For shift assays involving antibodies or antisera, reactions were incubated for 10 min at 30°C in the presence of nuclear extract and antibody prior to the addition of the radioactive probe. Products were separated by gel electrophoresis through a 5% polyacrylamide gel in a buffer containing 30 mM Tris-Cl, pH 7.5, 29.3 mM boric acid, and 0.66 mM EDTA. Gels were dried and exposed to Kodak XAR5 film.

RESULTS
The murine blk promoter contains a Pax-5 binding site in the region covering nt Ϫ68 to Ϫ37, as determined previously in vitro. We wished to determine whether this sequence represents a functional site on the blk promoter. Initially, we tested whether overexpression of Pax-5a could increase transcription from a blk minimal promoter fragment driving expression of the CAT reporter gene. Transient co-transfections of the Pax-5a expression construct (pcDNA3.Pax-5a) and the blk reporter construct ((Ϫ191)blkCAT2E)) into the mature B cell lines M12 and A20/2J were performed using DEAE-dextran. The expression vector without insert (pcDNA3) was used in parallel in co-transfections as a negative control. Overexpressed Pax-5a was able to increase the activity of the blk promoter construct 4.6-fold (Ϯ0.55) and 2.8-fold (Ϯ0.41) in cell lines M12 and A20/2J, respectively, as shown in Fig. 1A.
As B cell lines generally have a low transfection efficiency, we performed these experiments in stable cell lines that expressed Pax-5a protein. A previous study had shown that clones from the plasmacytoma line S194 that had been stably transfected with Pax-5a were able to repress endogenous Jchain expression (10,14). These S194 clones were used to transiently transfect the blkCAT reporter construct using DEAE-dextran. CAT conversion of nontransfected S194 cells was compared with that of a stable clone that expressed low levels of transfected Pax-5a protein (S194-Tf1) and one clone that expressed high levels of Pax-5a (S194-Tf4), which resulted in a 6.6-fold (Ϯ0.5) and 44-fold increase (Ϯ3.9) of CAT activity, respectively. Results of the stable transfections are shown in Fig. 1B. Together, the transfection studies indicated that Pax-5a functions as an activator of transcription on the murine blk gene.
Next, we tested whether mutations within or in the vicinity of the Pax-5 binding site would affect the activity of the blk promoter in vivo. Earlier work (2) using mobility shift assays had already established that mutations at the 5Ј-end and center of the Pax-5 binding site (mutA and mutB), but not at the 3Ј-region (mutC), prevented Pax-5a from binding to the blk promoter in vitro. The location of the mutated blk sequences relative to the Pax-5 binding site is shown in Fig. 2A. The WT (Ϫ191)blkCAT2E or one of three mutated constructs (mutA, mutB, and mutC; see Fig. 2A), were transiently transfected into the highly differentiated, mature B cell line A20/2J (which expresses moderate levels of Pax-5a). Unexpectedly, CAT assays showed that mutations at the 5Ј-end of the binding site (mutA) led to a significant increase in promoter activity. As expected, mutation of the central region of the Pax-5 site (mutB) significantly reduced promoter activity, whereas mutation mutC had no significant effect on promoter activity. Results from the CAT assays are shown in Fig. 2B. These data at least partially agreed with the mobility shift assays mentioned above, because both in vivo and in vitro studies pointed toward the importance of the central region (mutB) but not the 3Јregion (mutC) of the Pax-5 binding site for interaction with Pax-5a. The observation that mutA causes an increase in blk activity will be discussed in more detail below. reporter construct in S194 clones stably transfected with an Pax-5a expression construct. Results of CAT assays are shown as -fold increase of CAT conversion Ϯ S.D. compared with non-Pax-5a-transfected S194 cells. Lane 1, S194 ϩ blkCAT; lane 2, S194-Tf1 (low Pax-5a) ϩ blkCAT; lane 3, S194-Tf4 (high Pax-5a) ϩ blkCAT.
Blk levels drop when mature B cells differentiate into plasma cells, and plasma cells have undetectable levels of blk transcripts (1). If Pax-5a is an activator of blk transcription, a decrease of promoter-bound Pax-5a would be expected when mature B cells are stimulated by LPS or anti-IgM. To obtain data on a range of B cell activation stages, we decided to use small resting B cells isolated from mice spleens using Percoll gradients. These cells were activated with LPS and collected at various time points after the LPS addition, and nuclear extracts were prepared. To investigate the levels of Pax-5 proteins in these samples, mobility shift assays were used with a labeled blk DNA fragment F3, which covers nt Ϫ70 through ϩ136 of 5Ј-blk. Results of the mobility shift assays are shown in Fig. 3A.
The data showed that the amount of Pax-5a interacting with the blk promoter decreased after 48 h of activation by LPS to very low levels at 73 h (Fig. 3A, arrow 2). To distinguish between a true decrease in total Pax-5a levels in the nucleus and a change in the amount of Pax-5a bound to the blk promoter, Western blot analyses were performed on the nuclear extracts. An anti-Pax-5 serum, Ed-1, which recognizes all four isoforms of the Pax-5 gene, was generated for this purpose (see "Materials and Methods"). Results of the Western blots are shown in Fig. 3B. Surprisingly, these data indicated that Pax-5a levels in the nucleus did not change significantly during this 73-h period.
This observation could suggest that blk-bound Pax-5a is replaced by a second transcriptional regulator, which is expressed after LPS induction. In agreement with this hypothesis, we noticed the presence of a new, slower migrating complex in activated cells (Fig. 3A, arrow 3). This complex was present at almost undetectable levels in resting B cell nuclei but became a major complex present on the blk probe after 24 -48 h of LPS stimulation. At the 73-h time point, this higher complex was still very prominent, in contrast to the Pax-5a complex. A second induced complex was found to bind very weakly to the blk promoter as well ( Fig. 3A; indicated by arrow 4).
A somewhat unexpected finding was that in nuclear extracts prepared from resting B cells, a protein that is likely to represent the alternatively spliced isoform Pax-5d bound to the blk promoter at levels almost identical to Pax-5a (see Fig. 3A, arrow 1). Interestingly, after 24 h of LPS stimulation, amounts of this protein had already dropped to a base-line level. Because no antisera specific to Pax-5d are yet available, we could not determine the exact nature of this protein. Although it was recognized by the anti-Pax-5 antiserum Ed-1, it could also be a proteolytic fragment derived from isoform Pax-5a. This possibility is still under investigation.
The LPS-inducible complex was studied in more detail using

FIG. 2. Effect of mutations of 5-blk on promoter activity using transient transfections of blkCAT reporter constructs.
A, overview of the three blk nucleotide substitution mutations, in mutA, mutB, and mutC. B, results from transfections in A20/2J with the WT (Ϫ191)blkCAT2E construct indicated as WT and three mutants (see Fig. 2A). The relative CAT conversion is shown as -fold increase of the WT reporter construct Ϯ S.D. and was calculated from four experiments (n ϭ 4) (see "Materials and Methods"). mobility shift assays with the F3 probe. For practical reasons, we used the presecretor B cell line BCL 1 instead of normal activated B cell extracts. BCL 1 has high levels of both Pax-5a and the unidentified complex and has a mobility shift pattern identical to that of normal B cells after 24 -48 h of LPS stimulation. Fig. 4A shows a sequence comparison of the various WT and mutant double-stranded oligonucleotides used in this and following competition shift assays. Using a doublestranded oligonucleotide covering nt Ϫ37 to Ϫ70 of blk as unlabeled competitor, we showed that this inducible complex (as well as Pax-5a) binds to this region of the blk gene (Fig. 4B,  arrow 2, lane 3). To narrow down the exact position of binding, additional competition assays with mutant double-stranded oligonucleotides were performed. From these experiments, we conclude that nt Ϫ57 to Ϫ50 (mutB) as well as nt Ϫ45 (mutH) were necessary for the binding of this unidentified factor, whereas mutations of nt Ϫ66 to Ϫ60 (mutA) had no effect on binding (shown in Fig. 4B (arrow 2) and/or summarized in Fig.  4A). This suggested that the new complex bound most strongly to a region overlapping with the central region of the Pax-5 site on the blk promoter (Fig. 4A). In addition, using oligonucleotides containing Pax-5 binding sites from the CD19 and J-chain promoters showed that only oligonucleotides containing the blk sequence can compete for binding with the induced complex (Fig. 4, A and B).
To identify the new complex, we tested antibodies specific for various putative factors and competitor oligonucleotides containing binding sites for such factors. Results from competition shifts with a double-stranded oligonucleotide containing the class I major histocompatibility complex NF-B site (20) as well as antisera against NF-B/p50 or NF-B/p65 suggested that the inducible complex in nuclei from activated normal B cells as well as from presecretor cell line BCL 1 contained the transcription factor NF-B/p50, but not p65, as shown in Fig. 5  (arrow 2).
NF-B proteins bind their targets either as homodimers or heterodimers. Based on the position of the inducible complex on shift gels (discussed with Fig. 6), as well as its inability to be recognized by p65 (see arrow 2 in Fig. 5), we hypothesize that this complex represents a homodimer of p50. One study showed that the optimal DNA binding site for p50/p50 homodimers is different from that of p50/p65 heterodimers (26). Preferential binding of p50 homodimers on the blk promoter is in agreement with the NF-B site on blk: 9 of 10 nt correspond to a p50/p50 homodimer consensus sequence, whereas only 8 of 10 nt correspond to a p50/p65 consensus site (see Fig. 4A). Because the p50 subunit lacks a transactivating domain, it has been implicated in repressor functions. The increased levels of p50 homodimers accessible to the blk promoter may provide a mechanism to down-regulate blk expression after B cells become activated.
In addition to this p50 homodimer, a faint, slower migrating complex was detected after cell induction (arrow 3). This complex was found to contain p50 but not p65 and could be competed by an oligonucleotide containing the NF-B binding site. The fact that both inducible complexes represent members of the NF-B family is perhaps not surprising, considering that such factors are normally induced in B cells by either IgM cross-linking or LPS.
The DNA binding sites for Pax-5 and NF-B/p50 are very close together and may in fact overlap (Fig. 4A). We wished to investigate whether the two factors bind to the blk promoter together, which would be indicative of a "docking" function for Pax-5a, similar to the situation on the mb-1 promoter (16). To address this possibility, co-immunoprecipitation experiments were performed using antibodies specific to both Pax-5a and NF-B/p50. Nuclear extracts from B cell presecretor line BCL 1 and pre-B cell line PD36 were incubated with antibody-coated beads overnight, the immunoprecipitated proteins were separated on SDS-polyacrylamide gels, and Western blot analysis was performed. We were unable to demonstrate any association between NF-B/p50 and Pax-5a in these experiments (data not shown). Some further evidence for a "replacement" but not a "cooperation" mechanism was supported by mobility shift assays in which the addition of excess in vitro translated Pax-5a to nuclear extracts from BCL 1 (containing both p50 and Pax-5a protein) resulted in increased levels of bound Pax-5a and a simultaneous decrease of bound p50 (data not shown). Together, these data suggested that at least in vitro, Pax-5a and NF-B/p50 compete for binding to the Ϫ65 to Ϫ45 region of blk rather then binding to the sites together.
To investigate whether the induction of an early B cell line would also result in binding of p50 homodimers to the blk promoter, we used the pre-B cell line 703/Z, which has been shown by others to be inducible by LPS (21). Cells were activated with LPS as described (above), and time points were collected at 0, 4, 48, and 120 h. Relative levels of p50 and Pax-5a from uninduced or induced cells were compared using mobility shift assays. As shown in Fig. 6A (lane 2), reasonably high levels of bound Pax-5a were present in unstimulated 703/Z cells, but no or very low levels of p50 homodimers were detectable. After 4 h of LPS stimulation, a higher molecular weight complex was detected, at a level equal to that of bound Pax-5a (Fig. 6A, lanes 3-5). The mobility shift pattern at 4 h of LPS stayed unchanged for the remainder of the 120 h (not shown). Interestingly, the induced complex appeared to run at a higher position in the gel compared with the p50 homodimer complex that we had previously identified in activated B cells and BCL-1 cells (Fig. 4B, arrow 2). Levels of p50 homodimers were not affected by LPS stimulation. Further analysis showed that the induced complex represents a heterodimer containing both p65 and p50 subunits of NF-B, which are normally involved in activation of transcription (Fig. 6A, lanes 6 and 7). Thus, at least in vitro, LPS stimulation of an early B cell line leads to interaction of the p50/p65 with the blk promoter, but not p50/p50, whereas in activated mature B cells or presecretor B cell lines, the p50/p50 homodimer binds preferentially.
To see whether this distribution reflected patterns present during B cell ontogeny, a number of B cell lines representing different stages of B cell development were analyzed for the presence of p50/p50 homodimers versus p50-containing heterodimer complexes on the blk promoter. Using mobility shift assays, we found that early B cell lines, including pro-and pre-B cell lines, have very low levels of promoter-bound p50 homodimers. Mature B cells, presecretors, and plasma cell lines have increasing levels of bound p50 homodimer. Levels of p65/p50 or other heterodimeric NF-B complexes varied in these lines. Results are shown in Fig. 6B.
Both p50/p65 and p50/Rel run at a similar position in the shifts, making it difficult to distinguish between the two unless anti-p65 antisera were used. To address this question, additional mobility shift assays as well as Western blot analyses were performed on a number of B cell lines including A20/2J, CH12, BCL 1 , SP2/0, and S194. Results from mobility shift assays (not shown) indicated that the slower migrating complex in these lines contained p50 but no, or only a relatively small amount of, p65. Interestingly, Western blot analysis showed that all tested lines had both p50 and p65 present in Competition shift assays using the blk.BSAP oligonucleotide as a probe; BCL 1 nuclear extracts and specific antisera (1:5 dilution) or control (pre-immune, PI, 1:5) serum as indicated; or a 500-fold molar excess of unlabeled NF-B oligonucleotides. The arrows represent the same complexes as in Fig. 4B. the nucleus, illustrating the difference between total levels of these factors present in the nucleus versus the amount bound to the blk promoter.
Interestingly, the experiments shown in Fig. 6B suggest that the ratio of bound p50 homodimers to bound Pax-5a strongly increases when B cells progress through their various developmental stages, from very low in pro-and pre-B cells to very high in plasma cell lines. These data agree with the patterns seen in LPS-activated normal B cells and may be indicative of a mechanism that regulates the levels of blk transcripts during B cell development. DISCUSSION In the studies described here, we sought to explore the functionality of the Pax-5 binding site on the blk promoter, which was shown in earlier studies to interact with Pax-5a and its alternatively spliced isoform Pax-5d in vitro. Our results indicate that the activity of the blk promoter is regulated by the relative level of at least three factors, including Pax-5a, the NF-B/p50/p50 homodimer, and the NF-Bp50/p65 heterodimer. Whether a fourth factor, Pax-5d, plays a functional role in transcription regulation of blk or other Pax-5 target genes is still under investigation.
Pax-5a Is a Transcriptional Activator of the blk Gene-Evidence for a role for Pax-5a as an activator was shown when Pax-5a protein was overexpressed in various B cell lines. In all transfected cell lines tested, activity of the blk promoter had increased significantly. This result is important, because it rules out the possibility that Pax-5a acts upon the blk promoter as a repressor. How exactly Pax-5a exerts its functions on the blk promoter is still unclear. From the experiments described here, we cannot distinguish between Pax-5a functioning as an activator of transcription through (indirect) interactions with the initiation complex, or alternatively, its functioning as a "gatekeeper" by keeping the blk promoter region accessible to other factors. Also, a role for Pax-5a as a docking protein cannot be ruled out, although it is unlikely that it would interact with NF-B proteins in this function (see below).
NF-B/p50 Homodimers Bind to the blk Promoter during Activation of Mature B Cells-An NF-B site was identified on the blk promoter, which overlaps with the Pax-5 binding site. The NF-B family contains at least five members, including p50, p52, p65, RelB, and c-Rel, which share a conserved Rel (DNA-binding) domain that interacts with NF-B sites on its target genes (20,26). Relatively high levels of p50 homodimers, but not p50 heterodimers, were found to bind to the blk promoter during the mature B, activated B, and plasma cell stages, although high levels of p50 heterodimers are reportedly present in the nucleus at these cell stages as well (21). One explanation for this could be that the blk sequence has a slightly higher homology with the NF-B/p50 homodimer consensus site as compared with the p50/p65 dimer consensus site (26), although the combination of other factors already binding to the blk promoter may influence this selective behavior as well. No evidence was found for interaction of the p52 subunit with the blk promoter, in agreement with the relatively low levels of this subunit during all B cell stages except the plasma cell stage (21).
The members of the NF-B family of transcription factors are important regulators of the immune and inflammatory responses in many cell types, including B cells (19,(27)(28)(29). We propose that the NF-B p50 homodimer plays an important role in the down-regulation of the blk gene when B cells are activated and differentiate into plasma cells. As has been shown by others, p50 proteins contain an intact DNA binding domain but no transactivation domain, suggesting that they act as repressors (22)(23)(24)(25)(26).
Relative Levels of Bound Pax-5 and NF-B Dimers Regulate blk Expression-Although a second transcription factor, NF-B/p50, binds to a site overlapping the Pax-5 site, no evidence pointed toward collaboration or interaction between Pax-5a and p50, using mobility shift assays and co-immunoprecipitations. In contrast, our experiments suggested that either Pax-5a or p50, can bind to this region on the blk promoter at a given time and that the relative concentration of the two fac- tors in the nucleus influences their binding frequency.
NF-B sites have been found in the vicinity of Pax-5 sites in other Pax-5 target genes. NF-B/p50 binds to two sites on the 3Ј-␣-enhancer of the Ig heavy chain genes, both the 3Ј␣-E (hs1,2) and 3Ј␣-hs4 sites. The two sites are not immediately adjacent, and it is unclear whether in these situations Pax-5a could prevent p50 from binding or vice versa. In this regard, two interesting studies support the hypothesis that binding of Pax-5a can somehow prevent other factors from binding in its vicinity, as shown for both the Ig heavy chain 3Ј-␣-enhancer (27) and the J-chain promoter. 3 Whether Pax-5a is able to prevent binding of NF-B factors and is relieved from this function after B cell activation is unclear. In any event, overlapping of two binding sites on the blk promoter allows the B cell to somehow regulate the activities of the involved factors during different stages of B cell differentiation.
Heterodimers of the NF-B Complex May Increase blk Promoter Activity-LPS activation of the pre-B cell line 703/Z induces the formation of a p50/p65 heterodimer complex on the blk promoter. This result is in agreement with studies by others who activated 703/Z cells using LPS (20). They found that in pre-B cell lines, both p50 and p65, but not p52, RelB, or Rel are induced after short term (4-h) LPS exposure. In addition, these authors report that in mature B cell lines, p50, p65, and Rel are present constitutively, whereas plasmacytomas mostly express the p52 and RelB subunits. Since the p50/p65 complex induced in 703/Z cells contains a functional activation domain, this would be expected to lead to increased levels of blk transcripts.
These data provide an explanation for earlier studies (1, 2) that had shown a remarkable variation in the amount of blk transcripts in different B cell lines, which did not correlate with the amount of promoter-bound Pax-5a. Results from the study described here provide an explanation for this variation. Generally, B cell lines that express high levels of blk RNA but possess low levels of Pax-5a have relatively high levels of p50 heterodimers (containing p65 or Rel). Because such heterodimers presumably are strong activators of blk expression, their interaction with the NF-B site on the blk promoter should result in high levels of blk. This is particularly clear in the mature B cell lines WEHI-231 and BFO3. In contrast, cell lines that express low levels of blk RNA in the presence of relatively high levels of Pax-5a usually have high levels of p50 homodimers but not p50 heterodimers bound to the blk promoter, for example the mature B cell lines B17.10 and the presecretor BCL 1 . Thus, both Pax-5a and the various NF-B complexes influence the level of blk expression.
The transfection experiments in mature B cell line A20/2J, although hampered by low activity of the blk promoter, may be indicative of a role for p50 heterodimers in increasing blk levels in mature B cells. This cell line represents a normal resting B cell, with low levels of p50 homodimers and moderate levels of Pax-5a. Mutation of a region on the blk promoter (mutA) that is necessary for binding of Pax-5a, but not p50, resulted in an increase in CAT activity. We hypothesize that, as a result of this mutation, other p50-containing complexes will now be able to bind and replace Pax-5a. Since both p50 and p65 subunits were detected in this line, they may form heterodimers and be responsible for the increase in CAT activity.
The co-transfection experiments in A20/2J also imply that p50-containing heterodimers are stronger activators of blk than Pax-5a. This is interesting in regard to the observation that in pro-B cells of Pax-5 knockout mice, blk levels had not decreased significantly. We speculate that this reflects the redundancy of transactivators that have affinity for the blk promoter. It is possible that although only low levels of p50/p65 heterodimers are present in the early B cell progenitors of the Pax-5 null mice, levels are high enough to replace Pax-5a functionally, by binding to the NF-B site in its absence. To test this hypothesis, it would be interesting to create double knockout mice that lack both Pax-5 and NF-B/p50; such mice would not be expected to have any detectable levels of blk in their pro-B cells or, alternatively, may have a developmental block prior to that of the pro-B cell stage.
Functional Significance of the p50 Homodimer-We did not detect a decrease in Pax-5a levels in nuclei of activated B cells, even after 73 h of LPS stimulation, which may be explained by the fact that B cells do not differentiate into plasma cells until 3-5 days after antigen exposure. It may be important that blk expression is down-regulated early in the B cell response; therefore, it may need to be shut off by p50/p50 and not by Pax-5a. This mechanism may be necessary because only a subset of Pax-5 target genes is down-regulated at this time, and the exact time point of down-regulation may vary among its target genes. Down-regulation is likely to depend on the unique function of each target gene during activation processes. It will be interesting to investigate blk levels in B cells of p50 knockout mice, which reportedly do not respond to LPS stimulation (27). Such mutant mice may have a delayed decrease in blk expression, which may prevent the cells from efficiently proliferating and differentiating into plasma cells. One report in agreement with this hypothesis suggests that blk is a growth inhibitor signaling molecule; the authors show that antisense oligonucleotides to blk prevent growth inhibition and apoptosis of an anti--chain activated B cell lymphoma (30).
Regulation of gene expression through differential usage of NF-B complexes has been shown to exist in T cells. The interleukin-2 promoter of resting T cells has been found to be associated with the p50 homodimer complex but not with other NF-B complexes (25,31). This homodimer complex decreased after full antigenic stimulation of the cells, whereas the amount of NF-B p50/p65 was sharply increased and could be correlated with increased expression of the interleukin-2 gene. These experiments reveal a physiological function for p50 in the T cell response to antigen in that p50 homodimers appear to inhibit expression of interleukin-2 by NF-B heterodimers.
In conclusion, our experiments show that the relative levels of Pax-5 and NF-B proteins provide a mechanism that differentially regulates blk expression during B cell development, activating blk transcription early in development and downregulating its levels during the final stages of B cell differentiation.