| Nucleic Acids Research | Pages |
Mutational analysis of p50E4F suggests that DNA binding activity is mediated through an alternative structure in a zinc finger domain that is regulated by phosphorylation
Introduction
Materials And Methods
Plasmid DNA
Protein preparation
Gel shift assays
Dissociation rate assays
Results
The p50E4F zinc finger domain mediates sequence-specific DNA binding activity
p50E4F DNA binding activity requires dimerization through the zinc finger domain
Residues in the extreme E4F N-terminus contribute to DNA binding stability
Phosphorylation within the p50E4F zinc finger domain appears to regulate an alternative structure required for DNA binding activity
Discussion
The p50E4F zinc finger domain and dimerization
Regulation of p50E4F DNA binding domain structure by phosphorylation
Acknowledgements
References
Mutational analysis of p50E4F suggests that DNA binding activity is mediated through an alternative structure in a zinc finger domain that is regulated by phosphorylation
ABSTRACT
INTRODUCTION
One of the striking features of eukaryotic sequence-specific transcription factors is the modular design of their functional domains. Distinct and separate regions mediate DNA recognition and binding, factor dimerization, transcriptional activation or repression or other regulatory interactions and can often be grouped into distinct classifications based on common structural motifs that, in principle, function in the same way in different proteins. Thus far a relatively limited number of motifs, including the basic region/leucine repeat (bZIP), basic region/helix-loop-helix (bHLH), winged helix-turn-helix (WHtH), homeobox, POU and zinc finger domains, have been found to mediate sequence-specific DNA binding and, in some cases, requisite factor dimerization (1-9 and references therein). Often the interaction of transcription factors with DNA regulatory elements is a major control point in regulation of gene expression. Therefore, it is important to define and understand the domains within a factor that mediate or influence DNA binding, since the structural characteristics of these motifs will dictate the types of regulatory mechanisms the cell can use to regulate the factor.
p50E4F is a low abundance cellular factor that is regulated by adenovirus E1A oncoproteins. During early lytic infection E1A-induced phosphorylation of p50E4F protein stimulates its binding to a tandemly repeated element in the adenovirus E4 promoter to help activate high level transcription (10-14). In uninfected cells ectopic expression of p50E4F inhibits colony and focus formation in E1A-transformed NIH 3T3 and primary rat embryo fibroblasts respectively (E.R.Fernandes and R.J.Rooney, submitted for publication), suggesting a role in cell growth control. However, the cellular genes normally regulated by p50E4F are presently unknown.
The sequence element recognized by p50E4F, RTGACGTC/AAY, is also recognized by a number of relatively abundant bZIP factors, including ATF/CREB family members, AP-1, C/EBP, E4BP4 and NF-IL6 (13,15-22). One of these, ATF-2, can also mediate E1A-induced stimulation of the E4 promoter, through an independent mechanism that does not involve a change in ATF-2 DNA binding activity (23,24). However, mutational studies of E1A and the E4F binding site have indicated that a large portion of E1A-induced E4 promoter activity is p50E4F dependent (10,14,25). Experiments measuring protein-DNA complex dissociation have demonstrated that the stability of a p50E4F-DNA complex is one to two orders of magnitude greater than the stability of a CREB/ATF-1-DNA complex (14). We hypothesize that this greater stability enables p50E4F to out-compete more abundant factors for occupancy of its binding sites and, together with its stimulation by E1A, accounts for the apparent prominence of p50E4F in mediating E1A transactivation of the E4 promoter.
Previous experiments demontrated that a fragment containing the first 262 amino acids encoded by the E4F cDNA (E4F262) was functionally equivalent to endogenous p50E4F with respect to DNA binding, E1A regulation and transactivation of the E4 promoter (13). A number of motifs often found in transcription factors, including various zinc fingers, an amphipathic helix-turn-helix motif and a proline-rich region, are present in the E4F262 fragment. However, unlike the other factors that bind to the E4F site, there is no bZIP DNA binding domain. In this report a zinc finger region near the C-terminus of p50E4F was shown to mediate both phosphorylation-dependent sequence-specific DNA binding and factor dimerization. A separate region in the N-terminus was required for increased p50E4F DNA binding stability. Furthermore, mutational analysis of the DNA binding/dimerization region suggested that a central C2H2 zinc finger motif and portions of two other surrounding zinc finger motifs assume a phosphorylation-dependent alternative structure that mediates binding to the E4F recognition sequence.
MATERIALS AND METHODS
Plasmid DNA
Truncated sections of the E4F cDNA contained in pGST-E4F fusion constructs (as specified in the text) were synthesized by PCR and subcloned into pGEX-1 or pGEX-2T (Pharmacia). E4F-N1 and E4F-N4 truncations used for in vitro transcription/translation were cloned into pCITE-3b (Novagen). Single and double amino acid substitutions were created by PCR using 5[prime] primers that contained point mutations in the specified residues and a 3[prime] primer containing wild-type E4F cDNA sequence to residue 262 followed by a termination codon and introduced into the pGST-E4FN2 sequence by overlapping PCR. All cloned PCR products were sequenced in their entirety.
Protein preparation
To produce glutathione S-transferase (GST)-E4F fusion proteins pGST-E4F fusion plasmids were transformed into BL21(DE3)pLysS or NovaBlue (Novagen). Bacterial cultures were grown overnight at 30°C, diluted 1:10 in LB broth, grown for 1 h at 37°C and induced with 0.1 mM IPTG for 4 h. Induced cultures were pelleted and resuspended in 3 ml TSE buffer/l culture (TSE: 25 mM Tris, pH 8.0, 50 mM sucrose, 10 mM EDTA, pH 8.0, 0.5 mM PMSF, 100 µg/ml each aprotinin, leupeptin and pepstatin; Boehringer Mannheim). Bacteria were lysed by addition of 15 mg lysozyme/l culture and 5 min incubation at room temperature, adjustment to 10 mM MgCl2, 1 mM MnCl2, 10 µg/ml DNase I and a 15 min incubation at 37°C, followed by adjustment to 1× PBS, 1% Tween-20, 1% Triton X-100, 10 mM DTT and thorough mixing. After removal of cell debris by centrifugation at 4°C GST-E4F proteins were purified from the lysates by glutathione-Sepharose affinity chromatography as per the supplier's instructions (Pharmacia), dialyzed in buffer A (10 mM HEPES, pH 7.9, 50 mM KCl, 1.5 mM MgCl2, 10% glycerol, 0.1 mM PMSF, 100 µg/ml each aprotinin, leupeptin and pepstatin) and stored frozen at -85°C. Protein concentrations were determined by the Bradford colorimetric assay (BioRad).
For heterodimerization analysis 2 µg GST-WT and GST-N5 protein were mixed together or separately with 2 µg GST protein, diluted with an equal volume of 6 M guanidine HCl/NaPO4, pH 8.0, for 5 min and dialyzed for 12 h in buffer A. Dialyzed proteins were concentrated using a Centricon P-20 (Amicon), measured by Bradford assay and stored frozen at -85°C until use. All protein preparations were analyzed by 12% SDS-PAGE and silver staining (BioRad).
For in vitro synthesis of E4F-N1 and E4F-N4 proteins pCITE-E4F-N1 and pCITE-E4F-N4 plasmids were transcribed and translated in the STP (Novagen) rabbit reticulocyte lysate system containing T7 RNA polymerase and [35S]methionine (NEN). Products were analyzed by 10 or 12% SDS-PAGE.
To dephosphorylate GST-E4F proteins 1 µg protein (in 30 µl buffer A) was incubated with 10 µl alkaline phosphatase coupled to acrylic beads (Sigma) at 30°C for 20 min, after which the beads were removed by centrifugation (12,13). A second procedure, in which 1 µg protein (in 20 µl WCE dialysis buffer) was adjusted to a final volume of 50 µl in 1× gel shift binding buffer (20 mM HEPES, pH 7.9, 60 mM KCl, 1 mM MgCl2, 0.5 mM DTT, 4% Ficoll) containing 10 µM ZnCl2 and incubated with 5 U calf intestinal alkaline phosphatase at 30°C for 90 min, followed by addition of 2 µl 1 M NaF, was equivalent in reducing GST-E4F DNA binding activity.
Gel shift assays
The structure of the E4wt, E4pm1, E4pm2 and E4pm4 probes and competitor DNAs and the binding reactions and gel shift assays for E4F and ATF were as previously described (13,14). Reactions with GST fusion proteins contained 1 µg protein and 1 µg poly(dI·dC)·poly(dI·dC) (Pharmacia). Reactions with in vitro synthesized proteins contained 7 µl reticulocyte lysate and 3 µg poly(dI·dC)·poly(dI·dC). Competitor DNA amounts given in the figure legends are relative to 1 ng 32P-labeled probe/reaction.
Dissociation rate assays
Binding reactions containing GST-E4F proteins and 32P-labeled E4wt probe were scaled up 3-fold, incubated for 40 min to allow DNA-protein complexes to form and then challenged by addition of a 500-fold molar excess of unlabeled E4wt DNA. At the times indicated 10 µl samples were removed and loaded onto a 4% polyacrylamide gel running at 150 V. After addition of the last samples the gels were run at 280 V to completion (14).
RESULTS
The p50E4F zinc finger domain mediates sequence-specific DNA binding activity
Previously p50E4F sequence-specific DNA binding activity was localized to amino acids 84-262 of the E4F peptide sequence (13). Contained within this region are several motifs that could potentially serve as DNA binding domains (Fig. 1). Residues 120-153 have the potential to form two amphipathic [alpha]-helices separated by a triple glycine turn, reminiscent of helix-turn-helix motifs found in prokaryotic DNA binding proteins (26). Immediately preceding this motif is a proline-rich sequence (residues 82-113), a characteristic noted in a number of transcriptional regulatory domains (27-35). Residues 191-247 harbor two consensus C2H2 zinc finger motifs with Krüppel-like linker regions (36,37). Also, immediately following the two consensus C2H2 motifs in the E4F sequence is an abortive C2H2 motif or an atypical CCHC zinc finger motif (38). However, the C-terminal half of this putative structure is deleted by the C-terminal truncation to residue 262, which does not interfere with DNA binding and thus is not required.
Figure 1. Potential DNA binding motifs in p50E4F. Structural motifs characteristic of transcription factors that are present in the N-terminal 300 amino acids encoded by the E4F cDNA are illustrated. C2H2 and CCHC zinc finger motifs located between residues 193 and 272 are underlined; conserved cysteines and histidines, required for zinc coordination, are in bold type. A potential amphipathic [alpha]-helix-turn-[alpha]-helix motif lies between residues 119 and 153; hydrophobic residues in a 4-3 repeat are denoted by *. Prolines clustered between residues 82 and 113 are double underlined. A putative nuclear localization signal (NLS) is overstruck. The positions of residues (R) 84 and 262 are denoted. A series of GST fusion proteins containing further N- or C-terminally truncated E4F sequences were created to identify the regions responsible for specific binding to an E4F site-containing probe (Fig. 2). Proteins that were N-terminally truncated to residue 189 (GST-N1 to GST-N5) retained DNA binding activity and displayed the same sequence specificity previously demonstrated for HeLa cell p50E4F, thereby delimiting the N-terminal boundary to the zinc finger domain. C-Terminal truncation to residue 225 (GST-C1), which deletes most of the second C2H2 zinc finger motif, eliminated DNA binding activity, as did deletion of the entire domain (GST-C2). Figure 2. p50E4F DNA binding activity is mediated through a C-terminal zinc finger region. (A) Schematic depiction of GST-E4F wild-type and truncated fusion proteins; black bars indicate remaining p50E4F sequence. (B) Gel shift analysis of bacterially produced purified GST-E4F fusion proteins bound to 32P-labeled E4wt probe; a silver stained SDS-polyacrylamide gel of the purified GST-E4F fusion proteins is shown below (full-length fusion proteins are indicated by an arrow). (C) Gel shift competition analysis of GST-WT, -N1, -N2, -N3, -N4 and -N5 proteins. Binding reactions contained a 50-fold molar excess of unlabeled wild-type or mutant competitor probe DNA. E4wt site, ATGACGTAAC; E4pm1 site, ATAACGTAAC; E4pm2 site, ATGATGTAAC; E4pm4 site, ATGACGTATC; mutations are underlined. All three mutations reduce or eliminate p50E4F DNA binding (30,33). To better define involvement of the C2H2 zinc finger motifs, additional N-terminal truncations (Fig. 3) and substitution mutations (Fig. 4A) were introduced into this region. Truncation of the first five residues in the first C2H2 zinc finger motif had only a slight effect on DNA binding activity, whereas deletion of the entire first zinc finger motif abolished DNA binding (Fig. 3). Similarly, substitution of the conserved cysteines at positions 193 and 196 in the first zinc finger by serines did not diminish but instead increased DNA binding activity [sim]2- to 3-fold, whereas substitution of the conserved histidine at position 209 by alanine eliminated DNA binding (Fig. 4C). The increased DNA binding activity of the C193,6S mutant was consistently observed in multiple protein preparations. These results indicate that residues in the C-terminal half of the first C2H2 zinc finger motif are required for E4F site binding but that the zinc finger motif itself is not. Figure 3. Effect of deletions within the first C2H2 zinc finger motif. (Top) Schematic depiction of the N5, N6 and N7 deletions in the p50E4F zinc finger region; black bars indicate remaining p50E4F sequence. (Bottom left) Gel shift analysis of purified GST-N5, GST-N6 and GST-N7 proteins at 0.3 µg protein/lane and 1.0 µg protein/lane. (Bottom right) Gel shift competition analysis of GST-N6 protein (1 µg/lane) with a 100-fold molar excess of unlabeled E4wt, E4pm1, E4pm2 and E4pm4 probe DNAs. Below the gel shift competition is a silver stained SDS-polyacrylamide gel of the purified GST-N5, -N6 and -N7 proteins (500 ng/lane). Figure 4. Effect of amino acid substitutions in the p50E4F zinc finger region. (A) Schematic depiction of the specific substitutions introduced into the p50E4F zinc finger region. C2H2 zinc finger motifs have the characteristic sequence C--C-K/R-F/Y-----L--H---H. The relative DNA binding activity of each mutant protein is indicated: -, no activity; +, wild-type activity; ++, increased activity. (B) Gel shift analysis of purified GST-N2 proteins (1 µg/lane) containing specific substitutions in the C-terminal half of the p50E4F zinc finger region (left). Gel shift competition analysis and a silver stained SDS-polyacrylamide gel of the mutant proteins (right). (C) Gel shift analysis of purified GST-N2 proteins (1 µg/lane) containing specific substitutions in the N-terminal half of the p50E4F zinc finger region (top). A silver stained SDS-polyacrylamide gel of the mutant proteins (bottom). Consistent with a requirement for the second C2H2 zinc finger motif, as implied by the effect of the C1 truncation, a double substitution of Arg236 by leucine (in the `Z' position for specific base contact; 39-41) and conserved His237 by asparagine eliminated DNA binding activity, whereas double substitution of Lys231 by methionine and Gly232 by serine, located in non-conserved and non-DNA contacting positions, had no effect (Fig. 4B). Outside the C2H2 zinc finger motifs no effects on DNA binding activity were seen with mutations of Ser215 or Thr242 in the first and second Krüppel-like linker regions respectively or with mutation of the more C-terminal Lys254 (Fig. 4B). However, a double substitution of Lys248 by methionine and Cys249 by serine, at the N-terminal end of the truncated CCHC zinc finger motif, eliminated DNA binding (Fig. 4B). Taken together these data indicate that the second C2H2 zinc finger motif is required for sequence-specific DNA binding, whereas the putative zinc finger structures of the first consensus C2H2 motif and the non-consensus CCHC motif are not. Instead, the requirement for His209 in the first motif and Lys248 and Cys249 in the CCHC motif suggests that surrounding residues on both sides of the second C2H2 zinc finger motif participate in an alternative domain structure that mediates DNA binding activity.
p50E4F DNA binding activity requires dimerization through the zinc finger domain
The palindromic nature of the E4F recognition site suggests that p50E4F may bind as a dimer. As shown in Figure 5A, incubation of full-length GST-WT protein with the truncated GST-N5 protein did not produce any additional protein-DNA complexes other than those seen with each protein individually. However, when the mixed proteins were treated with guanidine HCl and subsequently renatured by dialysis prior to the DNA binding reaction protein-DNA complexes of intermediate mobility were observed, indicating heterodimerization. Similar treatment of each protein mixed with GST protein did not produce the intermediate complex. Complexes of intermediate mobility were seen with renatured mixtures of GST-WT + GST-N2 proteins and GST-N2 + GST-N5 proteins respectively (data not shown).
Figure 5. p50E4F proteins bind to DNA as a dimer. (A) Gel shift analysis of untreated GST-WT (WT) and GST-N5 (N5) proteins individually, untreated WT and N5 proteins mixed together, WT and N5 proteins mixed together and treated with guanidine HCl, WT protein mixed with GST protein and treated with guanidine HCl or N5 protein mixed with GST protein and treated with guanidine HCl, as described in Materials and Methods. The positions of homodimer and heterodimer complexes are indicated; the open and closed circles indicate GST-WT and GST-N5 proteins respectively. (B) Gel shift analysis of E4F-N1 and E4F-N4 proteins individually translated in vitro and E4F-N1 and E4F-N4 proteins co-translated together. Open and closed circles indicate E4F-N1 and E4F-N4 proteins respectively. E4F-N1 and E4F-N4 proteins contain the same p50E4F sequences as described for the GST-N1 and GST-N4 fusion proteins inFigure 1. In (A) and (B) E4wt was the DNA probe. |
 |
Evidence of dimerization was also observed with E4F proteins synthesized in vitro. E4F-N1 and E4F-N4 proteins were translated individually or together in reticulocyte lysates and analyzed by gel shift assay (Fig. 5B). Although DNA binding activity was low, a single gel shift complex with a distinct mobility was observed for each protein translated individually. In contrast, three gel shift complexes resulted from co-translation of the two proteins, two corresponding to the complexes formed by each protein individually and a third complex of intermediate mobility, showing that heterodimerization was not an artifact of the GST moiety in the fusion proteins. Mixing of the individually translated proteins did not produce the intermediate complex, indicating that co-translation of the two proteins was required for its appearance (data not shown). These results demonstrate that a p50E4F-DNA complex consists of a p50E4F dimer bound to the E4F recognition site and that residues present in the N5 protein (residues 189-262) are sufficient for proper dimer formation and DNA contact.
Residues in the extreme E4F N-terminus contribute to DNA binding stability
DNA-protein complexes containing purified HeLa cell p50E4F were previously shown to be much more stable than complexes containing the major ATF activity purified from HeLa cells, CREB/ATF-1 (42). In that experiment a side-by-side comparison of the relative dissociation rates of p50E4F-DNA and ATF-DNA complexes revealed the half-life of p50E4F complexes to be [ge]2 h, whereas the half-life of ATF complexes was [le]2 min (14). N-Terminally truncated GST-E4F fusion proteins were used in the same type of dissociation rate assay to determine if the high stability of p50E4F binding requires only the zinc finger domain. GST-E4F proteins were bound to a 32P-labeled E4F site probe and then challenged with a large excess of unlabeled probe to determine the relative dissociation rate of each protein (Fig. 6). Stable association with the labeled probe was seen with GST-WT but not with GST-N1 or any further truncated proteins. Therefore, residues within the first 83 amino acids of the E4F sequence, in addition to the zinc finger domain, are required for the high stability of a p50E4F-DNA complex.
Figure 6. p50E4F DNA binding stability is enhanced by residues near the N-terminus. DNA-protein complexes were formed using GST-WT, -N1, -N2, -N3, -N4 and -N5 proteins and 32P-labeled E4wt probe. Equilibrated binding reactions were challenged with a 500-fold molar excess of unlabeled E4wt competitor DNA and then loaded onto a running native polyacrylamide gel at the times indicated.
Phosphorylation within the p50E4F zinc finger domain appears to regulate an alternative structure required for DNA binding activity
One of the defining characteristics of the sequence-specific DNA binding activity of endogenous p50E4F and of cloned E4F derivatives expressed either in vitro or in transfected Cos cells is their phosphatase sensitivity (12-14). Alkaline phosphatase treatment also eliminated the DNA binding activities of GST-WT, GST-N3 and GST-N5 proteins (Fig. 7A), demonstrating that, similar to E4F proteins produced in mammalian systems, bacterially expressed E4F proteins also require phosphorylation for activity. Importantly, this result localized phosphatase sensitivity to the N5 protein (residues 189-262) and thus suggests that a phosphorylation site within the zinc finger domain is critical for DNA binding activity. In contrast, the DNA binding activity of the C193,6S mutant protein, in which the two conserved cysteine residues of the first consensus C2H2 zinc finger motif are mutated, was not affected by alkaline phosphatase treatment (Fig. 7B). Other substitution mutants that retained DNA binding activity were still phosphatase sensitive (data not shown). Thus the C193,6S mutation specifically abrogates the requirement for p50E4F phosphorylation. This suggests that disruption of the first zinc finger structure and phosphorylation at a critical site have similar effects on the structure of the zinc finger domain.
| Figure 7. Regulation of the p50E4F zinc finger region by phosphorylation. (A) Phosphorylation within the p50E4F zinc finger region is required for DNA binding activity. Gel shift analysis of GST-WT, GST-N3 and GST-N5 proteins that were untreated or treated with alkaline phosphatase (+ Alk. Phos.). (B) The C193,6S mutation eliminates the requirement for phosphorylation. Gel shift analysis of GST-N2 and GST-N2(C193,6S) proteins that were untreated or treated with alkaline phosphatase (+ Alk. Phos.). | 
 |
DISCUSSION
Two functional characteristics distinguish p50E4F from the myriad bZIP transcription factors that recognize the E1A-inducible elements in the adenovirus E4 promoter and are responsible for its role in regulating E4 transcription: the stability of p50E4F DNA binding is [sim]1-2 orders of magnitude greater than other competing factors and p50E4F DNA binding activity is stimulated by E1A in a phosphorylation-dependent manner. In this report two regions in p50E4F were shown to be responsible for these DNA binding properties. A region within the first 83 residues of the N-terminus was found by simple deletion to be required for the high stability of p50E4F-DNA complexes; identification of the specific residues that are required for the extra stabilization of p50E4F DNA binding awaits further analysis. More comprehensive analyses identified a zinc finger domain at the C-terminal end of the protein that mediates sequence-specific DNA binding and factor dimerization. Importantly, the phosphatase sensitivity of p50E4F DNA binding activity also mapped to this region, indicating that it contains functionally important phosphorylated residues. Moreover, the results from these studies suggest that phosphorylation within the zinc finger region regulates an alternative domain structure that is required for p50E4F DNA binding activity.
The p50E4F zinc finger domain and dimerization
The finding that p50E4F binds to DNA as a dimer was predictable given the palindromic nature of the E4F binding site and the fact that mutations in either half of the site abolish p50E4F binding (14). Experiments using the N5 mutant (Figs 2 and 5) localized both sequence-specific DNA binding and factor dimerization to residues 189-262, a region that contains only two C2H2 zinc finger motifs and the N-terminal half of a CCHC zinc finger motif. Analysis of additional mutations (Figs 3 and 4) showed that DNA binding activity required residues within the C-terminal half of the first C2H2 motif and the N-terminal half of the CCHC motif, but that the N-terminal half of the first C2H2 motif was dispensible. These results delineate a narrow region, containing only the central C2H2 zinc finger motif and [sim]20 residues flanking either side of it, that is sufficient for both DNA binding and, by implication, factor dimerization. How this region mediates both DNA contact and dimerization is not clear.
A single C2H2 or class I zinc finger domain consists of a 12 residue [alpha]-helix packed against a [beta]-hairpin, where the two conserved cysteines flanking the turn in the [beta]-hairpin and the two conserved histidines facing the inward side of the [alpha]-helix coordinate one Zn2+ ion and thereby position the [alpha]-helix in the major groove of the DNA to allow specific contact with three bases and adjacent phosphate residues (39-41,43-45). When bound to DNA this motif does not function as a dimerization interface. Thus proteins containing this structure typically have two or more zinc fingers concatenated in a linear array to enable stable and specific recognition of the larger cognate binding site and forgo a requirement for dimerization. In one case, the Drosophila GAGA protein, high affinity DNA binding has been shown to occur with only one C2H2 zinc finger (46). This interaction also requires two short basic regions within the 30 residues that precede the GAGA zinc finger, but does not involve dimerization; the protein binds as a monomer. Similarly, in class 4 zinc finger domains, a structurally distinct motif found in GATA transcription factors, a single C4 zinc finger followed by a C-terminal tail that wraps around into the minor groove mediates stable monomeric DNA binding (47). In contrast, the class 2 zinc finger domain found in nuclear hormone receptors, which contains two contiguous C4 zinc fingers, does form a structure that mediates both DNA contact and factor dimerization (48-52). However, there are no sequence or structural similarities between the class 2 domain and the p50E4F DNA binding domain.
Unfortunately, we have not been able to identify residues in this domain that specifically mediate dimerization per se. The heterodimerization assay used to analyze GST-E4F (Fig. 5) cannot distinguish mutations that affect dimerization from those that affect DNA contact. Other methods which are typically used to analyze protein dimerization, including co-immunoprecipitation, non-denaturing gel electrophoresis and glycerol gradient sedimentation, had very poor signal-to-noise ratios and could not reproducibly detect differences in heterodimerization caused by specific mutations. We suspect that one of the reasons for this is that the majority of E4F proteins produced in these systems are incorrectly folded and/or unphosphorylated and thus do not properly dimerize. Based on measurements of the amount of protein needed to bind 50% of the probe in gel shift assays we estimate that only 0.01-0.1% of bacterially expressed GST-E4F fusion proteins and a lower percentage of in vitro synthesized E4F proteins are functional and able to bind DNA. Moreover, the presence of multiple zinc coordinating structures in this domain may contribute to other indiscriminate interactions that obscure the readout of these assays.
We do note that simple mixing of GST-E4F or in vitro synthesized E4F proteins did not lead to formation of heterodimers capable of binding DNA, which instead required denaturation and renaturation or co-translation of the different sized proteins. This indicates that proper heterodimer formation requires conditions that prevent or break up pre-existing homodimers, suggesting that p50E4F dimers are extremely stable. Whether this is due to strong ionic interactions between specific residues, as occurs in STAT proteins (53), or other structural features is unclear.
Regulation of p50E4F DNA binding domain structure by phosphorylation
The phosphatase sensitivity of bacterially produced GST-E4F proteins (and in vitro synthesized E4F proteins; R.J.Rooney, unpublished data) points to an absolute requirement for phosphorylation for p50E4F function. Thus even though spurious kinase activities in these systems may be responsible for phosphorylating the critical E4F residues, the net effect is the same as when performed by the kinase stimulated by E1A in adenovirus-infected cells: stimulation of DNA binding activity. The phosphatase sensitivity of the GST-N5 protein localizes the critical site(s) of phosphorylation within the zinc finger domain (Fig. 7A). Interestingly, of all the mutant GST-E4F fusion proteins that can bind to DNA only the GST-N2(C193,6S) protein is phosphatase insensitive (Fig. 7B and data not shown). The C193,6S mutation eliminates two of the zinc coordinating residues in the first C2H2 motif and thus would prevent formation of this putative zinc finger structure. This mutation also increases DNA binding activity, implying that formation of a zinc finger structure by the first C2H2 motif is deleterious to p50E4F DNA binding activity (although it is possible that the C193,6S mutation makes the critical residues in this domain more accessible to activating kinases or less accessible to phosphatases).
We speculate that the function of critical phosphorylation is similar in effect to the C193,6S mutation in that it disrupts the [beta]-hairpin structure in the first C2H2 motif and thus allows participation of the conserved histidine and other C-terminal residues in an alternative secondary structure. Recent evidence indicates that phosphorylation within the N-terminal half of the second C2H2 zinc finger in WT1 inhibits DNA binding activity of the WT1 protein (54,55) and thus may provide another example in which phosphorylation disrupts a zinc finger structure. In the case of p50E4F, however, this would be positive for DNA binding activity, since the presence of the first zinc finger structure appears to inhibit E4F site binding.
Although the nature of the phosphorylation-induced alternative structure of the DNA binding domain is unknown, the simplest interpretation of our data suggests that either the [alpha]-helix from the first C2H2 motif and the [beta]-hairpin from the CCHC motif both participate in factor dimerization or that the CCHC [beta]-hairpins mediate dimerization and the two [alpha]-helices fit along the minor grooves on either side of the DNA recognition site. Either arrangement would place the two central C2H2 zinc finger structures of each dimer in opposite orientations and position them in adjacent major grooves for sequence-specific contact with both halves of the palindromic E4F site. Although clearly speculative, our findings suggest an unusual means for regulation of zinc finger function, which can ultimately be tested by mutational analysis of E1A-induced p50E4F phosphorylated residues and physical determination of the domain structure.
ACKNOWLEDGEMENTS
We thank Ruby Tharp for excellent technical assistance and Drs Paul Ney, Christopher Davies, Paul Brindle and Gerard Zambetti for valuable discussions and critical reading of the manuscript. This work was supported by NIH grant GM-51299 from the National Institute of General Medical Sciences (R.J.R.), NIH Cancer Center Support CORE grant P30 CA 21765 from the National Cancer Institute (St Jude Children's Research Hospital) and by the American Lebanese Syrian Associated Charities.
REFERENCES
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: www-admin{at}oup.co.uk
Last modification: 24 Mar 1998
Copyright© Oxford University Press, 1998.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  L. Le Cam, M. Lacroix, M. A. Ciemerych, C. Sardet, and P. Sicinski The E4F Protein Is Required for Mitotic Progression during Embryonic Cell Cycles Mol. Cell. Biol., July 15, 2004; 24(14): 6467 - 6475. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. Rizos, E. Diefenbach, P. Badhwar, S. Woodruff, T. M. Becker, R. J. Rooney, and R. F. Kefford Association of p14ARF with the p120E4F Transcriptional Repressor Enhances Cell Cycle Inhibition J. Biol. Chem., February 7, 2003; 278(7): 4981 - 4989. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Fajas, C. Paul, A. Vié, S. Estrach, R. Medema, J. M. Blanchard, C. Sardet, and M.-L. Vignais Cyclin A Is a Mediator of p120E4F-Dependent Cell Cycle Arrest in G1 Mol. Cell. Biol., April 15, 2001; 21(8): 2956 - 2966. [Abstract] [Full Text]
|
|
|
|
 |
|
 L. Fajas, C. Paul, O. Zugasti, L. Le Cam, J. Polanowska, E. Fabbrizio, R. Medema, M.-L. Vignais, and C. Sardet pRB binds to and modulates the transrepressing activity of the E1A-regulated transcription factor p120E4F PNAS, June 23, 2000; (2000) 130198397. [Abstract] [Full Text]
|
|
|
|
|
|
E. R. Fernandes and R. J. Rooney Suppression of E1A-Mediated Transformation by the p50E4F Transcription Factor Mol. Cell. Biol., July 1, 1999; 19(7): 4739 - 4749. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Fajas, C. Paul, O. Zugasti, L. Le Cam, J. Polanowska, E. Fabbrizio, R. Medema, M.-L. Vignais, and C. Sardet pRB binds to and modulates the transrepressing activity of the E1A-regulated transcription factor p120E4F PNAS, July 5, 2000; 97(14): 7738 - 7743. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||