INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

The immunoglobulin (Ig) µ heavy-chain gene enhancer is a cell-specific transcription regulatory sequence. Located in the J_H-C_µ intron, it is necessary for the expression of a rearranged IgH gene in transfection as well as transgenic assays. Moreover, when taken out of its normal context, the µ enhancer is sufficient to target a heterologous transgene (1, 14, 18, 27) for expression at the appropriate differentiation stage in the B-lymphocyte lineage. In addition to activating transcription from a V_H gene promoter that has recombined into the C_µ locus, this enhancer has also been implicated in the initiation of V(D)J recombination at the IgH locus. This is based on two experimental observations: transgenic recombination substrates were shown to be activated by the µ enhancer (7), and genetic deletion of the endogenous enhancer was shown to suppress IgH recombination (4, 31). At present, it is not clear whether the recombination activation and transcription activation properties of the enhancer are directly related; however, it is likely that µ enhancer-mediated chromatin reorganization resulting in increased accessibility of the IgH locus to polymerases and recombinases is important for both processes.

B-cell-specific µ enhancer function is determined by multiple trans-acting nuclear proteins that bind to specific sites within the enhancer (6, 19). Proteins that bind to the enhancer can be broadly classified into two categories: those that are more restricted in their tissue distribution, such as µA, µB, and octamer binding proteins, and those that are ubiquitously expressed in most cell types, such as the µE1, µE2, µE3, and µE5 binding proteins. However, no µ enhancer binding protein identified to date has an expression pattern that correlates perfectly with the cells in which the IgH gene is expressed. Based on the expression pattern of proteins binding to the µA and µB sites, we have previously proposed that tissue specificity of the enhancer is the result of two essential enhancer binding proteins that have overlapping tissue distributions, with the enhancer being active only in those cells where both factors are coexpressed (20).

Tissue specificity may also be achieved, in part, by negative regulation of the enhancer. In particular, µE4 and µE5 motifs have been implicated in suppressing enhancer activity in non-B cells (30, 32) by binding the zinc finger protein ZEB (8). The µE2 to µE5 motifs contain a consensus CANNTG motif that binds the basic helix-loop-helix (bHLH) family of transcription factors (12, 17). Mutational analysis of µE motifs has shown that loss of single elements does not significantly affect µ enhancer activity (13, 15, 23, 35). This observation has been interpreted to indicate that there is redundancy among the µE elements; that is, absence of one µE element can be functionally compensated for by other µE elements present in the enhancer. Although sequence similarity among the µE elements suggests that they have similar roles in enhancer function, there is no experimental evidence in favor of or against this proposition.

To reduce the complexity of the enhancer, we have previously identified a core domain, which we refer to as the minimal enhancer, that has no redundant elements (20). The minimal enhancer contains the µA, µB, and µE3 motifs, and mutation of any one of these elements abrogates enhancer activity in B cells. We reasoned that µE3 plays an essential role in this context because there are no other compensatory µE motifs in the fragment. Two interesting features of the minimal enhancer should be noted. First, it is also composed of tissue-restricted (µA and µB) and ubiquitous (µE3) elements like the full enhancer, suggesting that it is a good starting point for the analysis of the µ enhancer. Second, multimerization of single elements from this enhancer does not produce a B-cell-specific transcription activator, suggesting that B-cell specificity of the tripartite enhancer is determined by a combinatorial mechanism.

Analysis of the minimal enhancer has led to the following model. This enhancer can be transactivated in nonlymphoid cells by coexpression of PU.1 and Ets-1 (µB and µA binding proteins, respectively). As observed in B cells, enhancer activity requires the intervening µE3 site to which, presumably, an endogenous protein is recruited in the presence of the transfected ETS proteins (26). Interestingly, a deletion mutant of PU.1 that lacks a previously identified transactivation domain (TD) retains its ability to activate the enhancer, whereas deletion of the TD of Ets-1 is functionally deleterious (5). We have proposed that the ETS domain of PU.1 may be sufficient to activate the enhancer because it serves a structural role that includes DNA bending and direct interactions with Ets-1 (21). A TD is required on the µA binding Ets-1 protein. Furthermore, the Ets-1 TD does not activate when tethered to the µB site, indicating that correct location of the domain on the enhancer is essential for function. Our working hypothesis is that µA and µE3 binding proteins present a composite activation domain to the basal transcription machinery (5).

Despite the insights gained from characterization of the minimal enhancer, a crucial aspect of the overall enhancer organization is still missing; this concerns how multiple E motifs contribute to enhancer activity. We show in this report that µE2 and µE3 elements have distinct functions, that µE2 and µE3 sites synergistically activate transcription, and that the synergy is mediated indirectly via protein binding to the µA element that lies between them. Our studies reveal a hitherto-unrecognized organization of µE motifs of the enhancer and a novel mechanism of "through-protein" cooperation between transcription factors. Notably, we provide evidence indicating that ETS domain proteins may mediate transcriptional synergy between bHLH proteins.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Plasmids. The PstI-BamHI fragments (nucleotides 376 to 433) derived from wild-type or mutant (µA, µE3, or µB) enhancers were first subcloned into pSP72 digested with the same enzymes. (Note that the BamHI site in the enhancer was previously introduced by site-directed mutagenesis and shown not to affect enhancer activity.) The nucleotide sequences at mutant sites have been described previously (20). For the µE2 mutation, the sequence was changed from CAGCAGCTGG to CATGCTCTGG, creating an SphI site but destroying a PstI site present in the wild-type enhancer. The SphI-BamHI mutant fragment was subcloned into pSP72 cut with SphI and BamHI. The wild-type and mutant enhancer fragments were isolated as HindIII-Asp718 fragments from the pSP72 subclone, treated with Klenow fragment, and cloned into the 56CAT reporter plasmid at the SalI site. Plasmids with dimeric inserts were identified and sequenced to confirm orientation. All reporters contained the enhancer inserts in the B orientation as defined by Nelsen et al. (20). A mammalian expression vector for TFE3 (28) was provided by Kathryn Calame (Columbia University, New York, N.Y.). E47 protein was expressed from pRC.E47, which contains the full-length E47 cDNA cloned in the pRC/CMV vector.

Transfection assays. S194 cells were grown in RPMI medium supplemented with 5% fetal bovine serum, 5% calf serum, and 50 µg each of penicillin and streptomycin per ml. Reporter plasmids (5 µg) were transfected into S194 cells by the DEAE-dextran method (20), and whole-cell extracts were prepared 48 h later by three rounds of freezing and thawing.

Expression and purification of recombinant proteins. (i) Ets-1 derivatives. The full-length Ets-1 protein (26) and Ets-1 167 (5) were expressed as hexahistidine-tagged proteins in pET vectors. For protein expression, the plasmid was transformed into the BL21 bacterial strain. Protein production was induced with 0.5 mM isopropyl--D-thiogalactopyranoside (IPTG), and His-tagged protein was purified from bacterial extracts by nickel affinity chromatography as described by the manufacturer (Novagen, Inc.).

(ii) GST.TFE3 and GST.E47. Bacterial expression vector for GST.TFE3 was provided by Kathryn Calame (Columbia University), and full-length E47 cDNA was cloned into pGEX.2T for expression of the protein. Glutathione S-transferase (GST) fusion proteins were purified as described previously (33). Briefly, expression of the fusion protein was induced by IPTG to a final concentration of 0.5 mM for 2.5 to 3 h. The bacteria were collected and resuspended in 6 ml of cold NETN buffer (10 mM Tris-Cl [pH 8.0], 100 mM NaCl, 1 mM EDTA, and 0.5% Nonidet P-40). The cell extract was made by sonicating at 0°C. After centrifugation, the supernatant was added to glutathione-agarose beads and incubated for 1 h. The beads were washed three times with 30 ml of NETN, and the absorbed proteins were eluted with elution buffer (50 mM Tris-Cl [pH 8.0], 10 mM reduced glutathione). Proteins were dialyzed against buffer D (20 mM HEPES [pH 7.9], 100 mM KCl, 0.2 mM EDTA, 0.5 mM dithiothreitol, 20% glycerol).

DNase I footprinting assays. The HinfI-DdeI fragment (bp 345 to 518) from the µ enhancer was treated with Klenow fragment and cloned into pSP72 cut with EcoRV. The plasmid was linearized with EcoRI, dephosphorylated, and radiolabeled with [-³²P]ATP and polynucleotide kinase. After digestion with BglII, the labeled enhancer fragment was purified by electrophoresis through 8% polyacrylamide gels. Fifty-microliter footprinting reaction mixtures contained 15,000 cpm of probe, 100 ng of poly[d(I-C) · d(I-C)], 4% polyvinyl alcohol, 20 µg of bovine serum albumin, and various amounts of bacterial proteins. After incubation, an equal volume of 10 mM MgCl₂-5 mM CaCl₂ was added, followed by DNase I treatment at a final concentration of 17 µg/ml for 1 min. The reaction was quenched with 20 mM EDTA-1% sodium dodecyl sulfate-0.2 mM NaCl-250 µg of yeast tRNA per ml. The DNA was purified by one extraction with phenol-chloroform (1:1), precipitated, and analyzed by electrophoresis through 8% denaturing polyacrylamide gels containing urea. The gels were dried on 3MM paper and exposed to X-ray film or phosphorimager screens as required.

Electrophoretic mobility shift assay (EMSA). The wild-type µ enhancer probe was isolated from the pSP72 subclone containing the PstI-BamHI fragment of the enhancer. Binding reaction mixtures (20 µl) contained 20,000 cpm of probe, 100 ng of poly[d(I-C) · d(I-C)], 2 µl of 10× lipage buffer (100 mM Tris [pH 7.5], 0.5 M NaCl, 100 mM -mercaptoethanol, 10 mM EDTA, and 40% glycerol), and various amounts of recombinant proteins. After 15 min of incubation on ice, the reaction mixtures were electrophoresed through 4% polyacrylamide gels which were visualized by autoradiography.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

µE2 function requires an intact µE3 element. We previously defined a minimal domain of the Ig µ intronic enhancer (µ70) that contains three sequence elements, µA, µE3, and µB (Fig. 1A). To examine how additional E motifs affected the activity of this minimal enhancer, we incorporated the µE2 motif and assayed the transcription activation properties of an enhancer fragment containing µE2, µA, µE3, and µB (µ57 [Fig. 1A]). Note that the new fragment containing four motifs is shorter than the previously defined tripartite enhancer because we have eliminated sequences 3' of the µB site that have no known function. For these studies a dimer of this fragment was cloned into the vector 56 fos CAT, which contains a bacterial CAT gene expressed from the c-fos gene promoter. S194 murine plasma cells were transiently transfected by using DEAE-dextran, and CAT protein expression was assayed by ELISA. Inclusion of the µE2 motif raised the transcription activation potential of the enhancer fragment significantly (three- to fourfold) compared to the activity of the µ70 enhancer (Fig. 1B). Mutation of the µE2 element in this context left an enhancer whose activity was indistinguishable from that of µ70 (Fig. 1B), confirming that sequences present in the µ70 fragment 3' of µB did not contribute significantly to transcription activation in B cells. Consistent with our previous suggestion of a critical role for the µA and µB elements, mutation of either motif in this context substantially reduced enhancer activity (Fig. 1B). Interestingly, mutation of the µE3 element also abrogated transcriptional activity (Fig. 1B). Because the µ57 µE3 fragment contained an intact µE2 element, we concluded that juxtaposition of µE2, µA, and µB elements did not constitute a functional enhancer, unlike the placement of µA, µE3, and µB elements in the µ70 enhancer. Thus, µE2 and µE3 are not functionally equivalent despite their close sequence similarity. Furthermore, because µE2 function required an intact µE3 element, these observations reveal a previously undetected communication between E motifs of the µ enhancer.

View larger version (18K): [in this window] [in a new window]   FIG. 1.   Analysis of a four-motif µ enhancer in B cells. (A) Schematic representation of the immunoglobulin µ heavy chain gene enhancer and fragments used in this study. The µ170 enhancer contains four µE motifs (squares). Proteins binding to these elements can be detected in nuclear extracts from lymphoid as well as nonlymphoid cells. µA and µB sites (ovals) bind tissue-restricted ETS domain proteins. The µA site binds Ets-1 and several other ETS proteins such as Fli-1 and Erg-3. The µB site binds the B-cell- and macrophage-restricted protein PU.1. The three small circles indicated as core 1, 2, and 3 have sequence similarity to the simian virus 40 enhancer core sequence. Previous mutational analyses show that both µA and µB are essential for B-cell-specific activity of the µ enhancer, whereas mutation of individual µE motifs partially reduces enhancer activity. Mutation of each core site suggests that they do not contribute to enhancer function. µ70 refers to a previously identified minimal enhancer domain. We refer to this as a tripartite enhancer because the three sites µA, µB, and µE3 are all required for enhancer activity. In this context as well, the core 1 site does not contribute to enhancer activity. The µ70 enhancer has weak transcription activity as a monomer, and we usually assay it as a dimer. µ57 refers to an enhancer fragment that contains four sequence motifs. Its shorter length compared to µ70 is because sequences 3' of the µB site that are present in µ70 were deleted in this fragment. In this study, µ57 activity was also assayed with the fragment cloned as a dimer. The last line shows the sequence of the µ57 fragment to indicate the positions of the various sequence elements. The µE2 and µE3 elements contain a core CANNTGG sequence, which is characteristic of the binding sites of bHLH transcription factors. The µA and µB elements contain the core GGAA sequence characteristics of ETS domain protein binding sites. However, recent structural and biochemical analyses of the ETS domain indicate that these proteins make several additional contacts 5' of the core GGAA. Therefore, the underlined region for both sites is shown to extend five nucleotides upstream of the GGAA sequence. In this interpretation the 5' end of the µA site significantly overlaps the 3' end of the µE2 site. Enhancer mutations used in this study were as follows: µE2 changes the second GCAG within this site to TGCT, µA changes the GGA within this site to TCG, µE3 changes the TGG within this site to CAT, and µB changes the TTT within this site to CCC. (B) Transcriptional activity of the µ57 enhancer in B cells. CAT reporter plasmids containing wild-type or mutated µ57 dimers were transiently transfected into S194 plasma cells by using DEAE-dextran. CAT enzyme activity in whole-cell extracts was determined by an ELISA (Boehringer Mannheim) and is shown normalized to the activity of the wild-type (WT) fragment. The previously analyzed µ70 dimer-containing reporter was used as the positive control, and the enhancerless reporter (56) was used as the negative control. µ57 derivatives mutated at each of the four elements are indicated as µA, µE3, µB, and µE2. The results shown were obtained by averaging three sets of transfections carried out in duplicate. Error bars represent standard errors.

Inhibitory interactions between µE2 and µA binding proteins. The ability of µE2 to act as a transcriptional activator only in the presence of µE3 prompted us to more closely examine DNA-protein interactions on this segment of the enhancer, which is shown in greater detail in the sequence in Fig. 1A. Although the binding site of ETS domain proteins contain a GGAA sequence (at the right end of the µA sequence underlined in Fig. 1A) (22), it is clear that ETS proteins make additional contacts several nucleotides 5' of the GGAA (36). Therefore, the µA bracket is shown extending 5 bp upstream of the core sequence, which makes it overlap with the µE2 element. We used full-length E47 and Ets-1 in in vitro binding assays.

View larger version (39K): [in this window] [in a new window]   FIG. 2.   Analysis of Ets-1 binding to the µA sites of the enhancer. (A) The noncoding strand of the µ enhancer was radiolabeled and used in DNase I footprinting studies with bacterially expressed proteins. Lane 1, no added proteins; lanes 2 to 6, binding reactions carried out with increasing amounts of His-tagged Ets-1 protein prior to DNase I treatment. A G+A ladder to identify the locations of the various sites is shown preceding lane 1, and the locations of the motifs are indicated to the right of the gel. Ets-1 binding to µA is visualized best by an increased intensity of the upper band compared to the lower band of the doublet that is marked by asterisks to the right of the gel within a bracket labeled µA. The ratio of the intensity of the upper band to that of the lower band is approximately 1.4 in the absence of any added proteins (lane 1). With increasing Ets-1 binding, the intensity of the upper band is increased while that of the lower band is decreased, resulting in an increase in the ratio of the upper to lower bands. Note that there is very little protection of the µB site even at the highest levels of Ets-1. In the experiment shown, 80, 160, 320, 640, and 960 ng of Ets-1 (lanes 2 to 6, respectively) were used. (B) Quantitation of the intensities of the two bands marked by asterisks in panel A. The intensities of the indicated bands were estimated after exposure of the gel shown in panel A to a phosphorimager screen. The ratio of the intensity of the upper band to that of the lower band is plotted against the amount of Ets-1 protein used. The increasing ratio with higher Ets-1 concentrations reflect an increase in the intensity (hypersensitivity to DNase I) of the upper band and a decrease in the intensity (protection against DNase I) of the lower band.

View larger version (72K): [in this window] [in a new window]   FIG. 3.   Inhibition of E47 and Ets-1 binding to µ enhancer DNA. DNase I footprint analysis of E47 binding to the µ enhancer in the absence or presence of Ets-1 is shown. Lane 1, no protein added; lanes 2 to 6, increasing amounts of GST-E47; lane 7, Ets-1 alone; lanes 8 to 12, a constant amount of Ets-1 (equal to that present in lane 7) with increasing amounts of E47 (equal to those present in lanes 2 to 6, respectively). E47 binding to the µE2 site is visualized by the loss of a major band located in the middle of the bracket marked µE2 and several other bands located both above and below the major band. E47 protein was used at 60, 160, 300, 600 and 1.2 µg in lanes 2 to 6, respectively. For Ets-1-plus-E47 binding, Ets-1 was present at 640 ng and E47 was varied as described above.

View larger version (59K): [in this window] [in a new window]   FIG. 4.   E47 and Ets-1 binding to the µ enhancer. (A) Full-length Ets-1 and E47 were expressed in bacteria as GST fusion proteins and used in EMSAs with wild-type (WT) or mutated µ enhancer probes as indicated. Lane 1, no proteins; lanes 2 to 4, E47 alone (70, 140, and 280 ng, respectively); lane 5, Ets-1 alone (320 ng); lanes 6 to 8, a constant amount of Ets-1 (as in lane 5) with increasing amounts of E47 (as in lanes 2 to 4, respectively); lanes 9 to 12, same proteins as in lanes 5 to 8 with a µE2 probe; lane 13, no protein; lanes 14 to 16, Ets-1 (160, 320, and 640 ng, respectively); lane 17, E47 (140 ng); lanes 18 to 20, a constant amount of E47 (as in lane 17) with increasing amounts of Ets-1 (as in lanes 14 to 16, respectively); lanes 21 to 24, same proteins as in lanes 17 to 20 with a µA probe. (B) EMSA analysis with DNA binding domains of E47 and Ets-1. Wild-type or mutated µ enhancer probes as indicated were used in binding assays with the bHLH domain of E47 and a truncated (231) Ets-1 derivative. Lane 1, no proteins; lanes 2 to 5, E47 bHLH alone (0.25, 0.5, 1, and 2 ng, respectively); lane 7, Ets231 alone, 400 ng; lanes 8 to 11, a constant amount of Ets231 (as in lane 7) with increasing amounts of E47 bHLH (as in lanes 2 to 5, respectively) (a double-occupancy complex is indicated by the arrow); lanes 12 to 17, same protein conditions as in lanes 6 to 11 with a µE2 probe; lanes 18 to 23, same protein conditions as in lanes 6 to 11 with a µA probe.

TFE3 enhances Ets-1, but not E47, DNA binding. Transfection assays (Fig. 1B) indicated that the µE2 site contributed significantly to enhancer function only in the presence of µE3. To address how µE3 contributed to revealing the µE2 transcriptional potential, we included µE3 binding proteins in the in vitro analyses. Because the µE2 and µE3 sites are separated by 20 nucleotides and therefore lie on the same side of the DNA helix, one mechanism by which µE3 and µE2 may cooperate is by direct interactions between µE2 and µE3 binding proteins. Such interactions may be reflected in cooperative DNA binding. In DNase I footprint assays, TFE3 protein (2, 29) generated a footprint centered over the µE3 element (Fig. 5A, lanes 1 to 5) that extended into the µA element that lies directly 5' of µE3. In addition to protecting residues within the µA element, TFE3 binding also increased the intensity of the two bands (asterisks) that were discussed above with regard to Ets-1 binding. However, phosphorimager quantification of these bands showed that the ratio of the intensity of the upper band to that of the lower band was unchanged upon TFE3 binding (Fig. 5B). This is in contrast to the observation with Ets-1, where the ratio of this doublet increases with Ets-1 binding (Fig. 2B).

View larger version (52K): [in this window] [in a new window]   View larger version (13K): [in this window] [in a new window]   View larger version (23K): [in this window] [in a new window]   FIG. 5.   Effect of TFE3 on proteins binding to the µE2 and µA elements in vitro. (A) Protein binding to the µE2 and µE3 sites. TFE3 binding to the µE3 site was assayed by DNase I footprinting in the absence or presence of E47. Lane 1, no proteins added; lanes 2 to 5, increasing amounts of TFE3 (100, 200, 400, and 600 ng, respectively); lane 6, 120 ng of E47; lanes 7 to 10, increasing amounts of TFE3 (as in lanes 2 to 5, respectively) in the presence of a constant amount of E47 (as in lane 6). Positions of the enhancer motifs are indicated on the right. Asterisks mark the positions of bands that are affected significantly by Ets-1 binding. Note that the intensity of both bands increases with TFE3 binding, whereas the ratio of the intensities of the bands changes with Ets-1 binding only, as described in the text and shown in panel B. (B) Quantitation of the asterisk-marked bands in the presence of TFE3 alone or TFE3 together with either E47 or Ets-1. Bands were quantitated by phosphorimager analysis, and the ratio of the upper band to the lower band is shown as a function of TFE3 concentration. Despite the increased intensities of these bands in the presence of TFE3 or of TFE3 plus E47, there is no discernible change in the ratio of their intensities. Quantitation of the autoradiograph shown in panel C demonstrates that in the presence of a constant amount of Ets-1, additional TFE3 results in a significant increase in this ratio (TFE3+Ets). The increased ratio at 0 ng of TFE3 in the TFE3+Ets curve is because 160 ng of Ets-1 alone (panel C, lane 1) results in weak but detectable binding to µA. (C) Binding of Ets-1 plus TFE3 in vitro. DNase I footprint assays were carried out with Ets-1 and TFE3 as indicated. Lane 1, Ets-1 alone (160 ng) at a level that shows weak, but detectable, enhancement of the upper asterisk-marked band; lanes 2 to 5, a constant amount of Ets-1 (as in lane 1) together with increasing amounts of TFE3 (100, 200, 400, and 600 ng, respectively).

View larger version (69K): [in this window] [in a new window]   FIG. 6.   EMSAs of Ets-1 and TFE3 binding to the µ enhancer. A µ enhancer probe, containing µE2-µB (Fig. 1A), was used in in vitro binding assays with increasing amounts of GST.TFE3 (lanes 2 to 4) or Ets167 (lane 5) or a constant amount of Ets167 (equal to that used in lane 5) together with increasing amounts of GST.TFE3 (same range as used in lanes 2 to 4) (lanes 6 to 8, respectively). The complex generated with TFE3 alone is indicated by bracket A, and the complexes generated in the presence of TFE3 plus Ets-1 are marked by bracket B. TFE3-plus-Ets-1 binding reactions were done with 1 µl of anti-His antibody (lane 9) or 1 µl of anti-Sp1 antibody (lane 11). Binding reactions also were done with either anti-His or anti-Sp1 antibody alone (lanes 10 and 12, respectively). A supershifted complex in the presence of anti-His antibody is indicated by the arrow. Ets167 is an N-terminal truncation lacking the first 167 amino acids of Ets-1. The binding reaction mixture contained 200 ng of Ets167 when indicated, and GST.TFE3 was used at 50, 100, and 150 ng (lanes 2 to 4 and lanes 6 to 8, respectively).

Ets-1-TFE3 interactions enhance formation of a three-protein-DNA complex. To investigate whether Ets-1-TFE3 interactions affected E47 binding to the µE2 site, we carried out footprinting studies with all three proteins. Increasing amounts of E47 protein were added to binding reaction mixtures that contained no other proteins (Fig. 7, lanes 1 to 5), only Ets-1 (Fig. 7, lanes 6 to 10), or both Ets-1 and TFE3 (Fig. 7, lanes 11 to 15). Because the goal was to examine the effect on E47 when Ets-1 was bound either alone or together with TFE3 to the DNA, we used conditions that gave comparable Ets-1 DNA binding in the two-protein (Fig. 7, lanes 6 to 10) or three-protein (lanes 11 to 15) analyses. Specifically, the ratio of the upper to lower asterisk-marked bands, which indicates Ets-1 DNA binding, was kept the same in the two sets. (Note that this ratio actually underestimates the extent of Ets-1 binding in the third set, because TFE3 increases the intensity of the lower band, thus decreasing the ratio.)

View larger version (74K): [in this window] [in a new window]   FIG. 7.   Binding of Ets-1, TFE3, and E47 to the µ enhancer. Increasing amounts of E47 protein were used in binding reactions containing no other proteins (lanes 1 to 5), a constant amount of Ets-1 (lanes 6 to 10) or a constant amount of Ets-1 plus TFE3 (lanes 10 to 15). Lane 1, no proteins added; lanes 2 to 5, increasing amounts of E47 alone (60, 120, 180, and 270 ng, respectively); lane 6, Ets-1 alone (480 ng); lanes 7 to 10; increasing amounts of E47 (as in lanes 2 to 5, respectively) in the presence of a constant amount of Ets-1 (as in lane 6); lane 11, Ets-1 (140 ng) plus TFE3 (400 ng); lanes 12 to 15 Ets-1 plus TFE3 (as in lane 11) with increasing amounts of E47 (as in lanes 2 to 5, respectively). Different concentrations of Ets-1 were used in lanes 6 to 10 than in lanes 11 to 15 to maintain similar levels of Ets-1 binding in the two sets, as estimated by the ratio of the upper to lower band of the asterisk-marked doublet. Phosphorimager quantification showed an insignificant change of the doublet ratio between lanes 11 to 15, indicating that Ets-1 occupancy was not reduced with increased E47 concentration.

Transcriptional synergy between E47 and TFE3 requires the intervening µA site. Our transfection studies with B cells indicated that endogenous µE2 and µE3 binding proteins synergistically activated transcription in the context of the µ57 enhancer. For the in vitro analyses described in the preceding section, we used E47 and TFE3 as µE2 and µE3 binding proteins, respectively. To test whether these proteins could activate the µ enhancer in nonlymphoid cells, we carried out the following transfection studies. The µ57-containing reporter plasmid was cotransfected into COS cells together with vectors directing expression of E47 and TFE3. Expression of either E47 or TFE3 alone activated transcription of the reporter weakly (Fig. 8, first three bars). Both of these genes have been previously shown to contain transcription activation domains, which are presumably responsible for this activity (2, 25). However, coexpression of both E47 and TFE3 resulted in synergistic activation of the reporter plasmid (Fig. 8). We also noted that the activity in the presence of both proteins was approximately fourfold greater than the sum of the activity due to each gene product. The extent of synergy is reminiscent of the three- to fourfold-greater activity of the µ57 enhancer (which contains µE2 and µE3) compared to the µ70 enhancer (which contains only µE3) in B cells. These results provide direct evidence for functional synergy between the bHLH (E47) and bHLH zipper (TFE3) families of transcription factors, which has also been noted by others (3).

View larger version (14K): [in this window] [in a new window]   FIG. 8.   Transcription activation by E47 and TFE3. A µ57 dimer reporter plasmid (2 µg) was transfected into COS cells by using calcium phosphate together with expression vectors for E47 (1 µg) and TFE3 (1 µg) as indicated below the graph. At 48 h after transfection, CAT expression was assayed by ELISA. In transfections containing only one of the transactivator plasmids, the total DNA was kept constant by the addition of pEVRF expression vector containing no insert. The roles of individual elements in transcriptional activation were assayed by using mutated µ57 reporter plasmids (indicated as µE2, µA, µE3, and µB). The results shown are the averages of three experiments carried out in duplicate, normalized to the expression of the wild-type (WT) plasmid in the presence of both transactivators. Error bars represent the standard error between measurements.

Ets-1 enhances transcriptional synergy between E47 and TFE3. Next, we directly tested whether the µA binding protein, Ets-1, could mediate transcriptional synergy between µE2 and µE3. The effect of Ets-1 protein on transcriptional activation by E47 and TFE3 was examined in COS cell transfection assays. As described above, E47 plus TFE3 transactivated the wild-type µ57 enhancer (Fig. 9, first four bars). This activity was significantly increased by coexpression of Ets-1 (Fig. 9, fifth bar). Enhancers carrying mutated µE2 or µA motifs were inactive under these conditions, indicating that the exogenously expressed Ets-1 raised E47- and TFE3-dependent transcriptional synergy.

View larger version (12K): [in this window] [in a new window]   FIG. 9.   Ets-1 mediates E47-TFE3 transcriptional synergy. Reporter DNAs (2 µg) were cotransfected into COS cells with expression vectors for E47, TFE3, and Ets-1 as indicated, and CAT enzyme activity was assayed by ELISA 48 h after transfection. The data shown are normalized to the activity of the wild-type µ57 reporter in the presence of TFE3 and E47, which is assigned the value 100. µE2 and µA refer to reporters containing mutated µE2 and µA motifs, respectively, in the context of the µ57 enhancer fragment (see also the legend to Fig. 1). 286 refers to an expression vector encoding an N-terminal truncation mutant of Ets-1 that contains the DNA binding ETS domain and extends to the C terminus (5). The total amount of DNA in each transfection was kept constant at 5 µg by using the empty expression vector pEVRF-0. All assays used 1 µg each of E47, TFE3, and Ets-1 expression vectors. The results shown are based on two transfection experiments carried out in duplicate. Error bars represent standard errors.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Functional differences between µE elements. In this work, we studied the mechanisms by which bHLH proteins and ETS proteins cooperate to activate the Ig µ enhancer. First we showed that inclusion of an additional µE element, µE2, in the previously defined tripartite enhancer significantly increased enhancer activity in B cells. These results indicate that sequential addition of µE3 and µE2 raises the transcriptional activity of the µA-µB combination, which by itself is not significantly active. Unlike µE3, which together with µA and µB produces a transcriptional activator, a DNA fragment containing µE2, µA, and µB did not activate transcription. Thus, µE2 and µE3 appear to have distinct functions in the µ enhancer. Furthermore, the µE2 element contributes to transcription activation only when the µE3 element is intact, indicating functional synergy between these two µE elements.

View larger version (18K): [in this window] [in a new window]   FIG. 10.   A model for through-protein cooperation between E motifs of the µ enhancer. Line 1, region of interest within the µ enhancer. The double-headed arrows represent the partially palindromic (CANNTG) µE2 and µE3 motifs, which bind bHLH proteins E47 and TFE3, respectively. The single-headed arrow represents the nonpalindromic µA site that binds ETS domain proteins. Partial overlap between the µE2 and µA sites (see Fig. 1A for nucleotide sequence) is indicated by the µE2 arrow lying partially within the µA arrow. Line 2, mutual inhibition of DNA binding by E47 and Ets-1. For simplicity, Ets-1 is shown bound to DNA (the circle represents the DNA binding ETS domain) with other parts of Ets-1 obstructing access to the µE2 site. However, our data does not directly address the mechanism by which these proteins are mutually inhibitory. Therefore, the indicated steric inhibition model should be viewed only as one of several possibilities. Line 3, TFE3 binding to the µE3 element enhances DNA binding by Ets-1. This is probably mediated by contacts between TFE3 and Ets-1 that relieve intramolecular inhibition of Ets-1 DNA binding. An altered Ets-1 conformation is indicated by the changed position of the non-DNA binding domains of Ets-1, which are shown directly interacting with TFE3. Line 4, Stabilization of Ets-1 binding by TFE3 reduces the inhibitory effect of Ets-1, allowing E47 to bind and generate a three-protein-DNA complex. One possibility is that reconfiguration of Ets-1 by interaction with TFE3 exposes the µE2 half-site shown to be blocked in line 2. Furthermore, transfection studies with COS cells indicate that coexpression of E47 and TFE3 is not sufficient for efficient transcriptional activation; maximal transcription activation requires, in addition, an ETS protein bound between the two bHLH factors. Based on these observations, we propose that functional synergy between µE2 and µE3 is the result of two interactions, both of which involve an intermediate ETS protein. First, the mutual inhibition of DNA binding by Ets-1-E47 interaction is relieved by TFE3, allowing all three proteins to bind the enhancer. Second, the ETS protein couples the transcription activation potential of the DNA-bound bHLH factors.

µE3-dependent µE2 activity. The possible role of µE3 in recruiting the transcription activation potential of µE2 was also inferred from in vitro DNA binding assays. We found that in the presence of the µE3 binding protein, TFE3, Ets-1 binding to µA was significantly enhanced. Furthermore, under these conditions E47-Ets-1 interference was markedly reduced, resulting in increased occupancy of the µE2 site. Because the DNA binding domains of E47 and Ets-1 can bind simultaneously to the µE2 and µA sites, we concluded that the mutual inhibition of binding by the full-length proteins must be due to other domains in these proteins. In principle, this inhibition can be removed if conformational alterations in either protein realign these domains so that they no longer obstruct DNA binding. We propose that Ets-1-TFE3 interaction alters the conformation of Ets-1 so that it no longer inhibits E47 binding to the µE2 site (shown schematically in Fig. 10, lines 3 and 4). Moreover, Ets-1-TFE3 interactions may also stabilize Ets-1 DNA binding, thereby bringing Ets-1 into a complex where it plays a critical role in transcription activation, as described below. In these models, enhancement of E47 transcriptional activation potential by TFE3 is mediated indirectly via Ets-1 protein that binds between them and not by direct interactions between TFE3 and E47. The mechanism of µE box cooperation on the µ enhancer is therefore distinct from the cooperative interactions between factors forming multiprotein assemblies on the beta interferon promoter (34) or the T-cell receptor -chain gene enhancer (9).

ETS protein-dependent transcriptional synergy between bHLH proteins. We anticipated that the collaboration between µE2 and µE3 would be explained by the facilitation of E47 DNA binding via TFE3-Ets-1 interactions as described above. Presumably, when both proteins were bound to the DNA, their TDs would be brought into close proximity and thereby enhance transcription. However, when we directly assayed the combined transactivation potential of E47 and TFE3 by cotransfection, synergistic transactivation was observed only when the intervening µA site was intact. Thus, transcriptional synergy between E47 and TFE3 requires an intermediate protein. One possibility that cannot be ruled out unequivocally is that the two proteins cannot bind to the enhancer under these conditions. However, we consider this to be unlikely because each protein individually transactivated the reporter at low levels, indicating that each could access its site on the reporter plasmid. Taken together with the observation that both proteins bound simultaneously to their respective sites in vitro, it is likely that both sites are also occupied in the transfection experiment. We infer that the transcription activation domains of these factors cannot cooperate unless the µA element is also occupied. Transfection assays with Ets-1 provided direct evidence that ETS proteins can enhance the transcription activation potential of bHLH proteins such as TFE3 and E47. Furthermore, the DNA binding domain of Ets-1 was not sufficient for this purpose, implicating a role for the transcription activation domain of Ets-1 in the three-protein complex, as discussed below. These observations provide a novel example of "through-protein" cooperation between the µE2 and µE3 elements of the µ enhancer.

A model for µ enhancer function. These observations can be incorporated into a model for µ enhancer function as follows. The binding and cotransfection assays discussed above deal with the possible interactions between the µE2, µA, and µE3 sites. However, enhancer activity in B cells also requires the µB site (Fig. 1), which is located 3' of the µE3 element. Our earlier studies, in particular those altering the spacing between µA and µB sites, have suggested that these two elements act as a unit (21). Furthermore, the ETS domain of PU.1 is sufficient to synergize with Ets-1 to activate the three-component µA-µE3-µB enhancer in nonlymphoid cells (5). Based on these observations, we have proposed that the ETS domain of PU.1 serves two functions: first, it may facilitate formation of a functional nucleoprotein complex by bending the DNA, and, second, direct interactions with Ets-1 may alter the structure of either Ets-1 or PU.1 (or both) and thereby affect enhancer activity. We suggest that PU.1 binding (via the mechanisms discussed above) is necessary to allow the formation of a transcription activation complex at the µE2, µA, and µE3 sites but that it does not participate directly in transcription activation. Stabilization of Ets-1 binding (to µA) by TFE3 (bound to µE3) is a critical aspect of the subsequent assembly. It is reflected in the transcriptional activity of the µA-µE3-µB enhancer as well as in rendering the µE2 site accessible to appropriate bHLH proteins. In the tripartite (µA-µE3-µB) enhancer, Ets-1 bound at the µA site provides a transcription activation domain which works together with an activation domain in TFE3. An additional, unexpected role for the µA binding protein was revealed in the studies presented here. Transcriptional synergy between µE2 and µE3 required an intact µA site, suggesting that ETS proteins mediate functional interactions between E47 (bHLH) and TFE3 (bHLH zipper) proteins bound to these sites. In addition to describing a novel mechanism of transcription activation by ETS proteins, these observations provide the first insights into how tissue-restricted elements, such as the µA motif, combine with ubiquitous elements, such as the µE2 and µE3 motifs, to generate tissue-specific enhancers.