| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Departments of Chemistry (D.D., W.M.S.) and Biological Sciences (D.D.), and the Center for Biomolecular Dynamics (D.D., W.M.S.), Bowling Green State University, Bowling Green, Ohio 43403; and Department of Chemistry (R.C.P.), Ohio Northern University, Ada, Ohio 45810
Address all correspondence and requests for reprints to: W. M. Scovell, Bowling Green State University, Overman Hall, Bowling Green, Ohio 43403. E-mail: wscovel{at}bgnet.bgsu.edu.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSNote Added in ProofREFERENCES |
|---|
(ER) is a ligand-dependent transcription factor that regulates the expression of estrogen-responsive genes. A key step in the activation process is the initial binding of the ER dimer to the estrogen response element (ERE). We examined the effect of the coactivator proteins, HMGB1 and HMGB2, in enhancing ER binding affinity to single and tandem EREs. Using EMSAs, both HMGB proteins are shown to enhance ER binding and induce cooperative ER binding on tandem ERE elements. We demonstrate that HMGB proteins facilitate strong ER binding to ERE consensus half-sites, exhibiting binding affinities comparable with ER binding to consensus ERE in the absence of HMGB proteins. These findings reveal that although HMGB proteins enhance binding affinity, they also relax ER binding specificity. Deoxyribonuclease I footprinting demonstrates that ER binds very differently to consensus ERE and ERE consensus half-sites, whereas both deoxyribonuclease I and exonuclease III digestions show that the presence of HMGB1/2 does not alter the DNA protection in ER/ERE complexes. Protease digestions of the complexes support this conclusion and show that a global conformation change occurs in ER when bound to the different ER binding sites. Models for these interactions are discussed, together with a hit-and-run mechanism that HMGB proteins may utilize to produce these effects. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSNote Added in ProofREFERENCES |
|---|
). With the cooperation of coactivators, this regulatory complex communicates with factors of the preinitiation complex (PIC) at the promoter and leads to activation and expression of estrogen-responsive genes. In the case of many natural EREs, ER binding affinity correlates with transcriptional activity (3).
|
Alternative mechanisms must also be considered in light of the increasing number of E2-responsive genes that contain one or more ERE half-sites. This has been puzzling because ER does not bind efficiently to ERE-half-sites in vitro (7, 8), and this does not fit into the correlation of binding affinity with transcriptional activation (3). Specific genes that have regulatory ERE half-sites and are responsive to E2 include human progesterone receptor, rat prolactin, oxytocin, ovalbumin, and HMGB1 (9, 10, 11, 12). In the case of the ERE half-site in the ovalbumin promoter, a point mutation abolishes estrogen responsiveness, demonstrating that it is a functional site (11). In addition, a genetic screen in yeast identified sequences from a random oligonucleotide library that exhibited E2-dependent activation, with the majority of the clones containing only a single consensus ERE half-site (cEREm) (13). These findings suggest that auxiliary cofactors are present in vivo that can facilitate ER binding to ERE half-sites and/or single or tandem nonconsensus ERE and actively participate in the regulation of estrogen-responsive gene expression (14).
HMGB1 and HMGB2 are abundant, ubiquitous, architectural proteins that bind nonspecifically in the minor groove of DNA. They have been implicated in bending or increasing the flexibility of DNA that can facilitate the assembly of nucleoprotein complexes and nucleosome remodeling (15, 16, 17, 18, 19). HMGB proteins have been shown to bind to a number of basal and regulatory transcription factors and affect the level of transcription in a context-dependent manner. HMGB1 can serve as a reversible repressor by binding to TATA-binding protein (20, 21), whereas in contrast, it can function as a coactivator with a subset of HMGB-sensitive regulatory factors. HMGB proteins increase the binding affinity of this subset of factors that include steroid hormone receptors and their DBDs (22, 23, 24, 25, 26), p53 (27), HOX homeodomain proteins (28), Oct proteins (29, 30) and the Rel family of proteins (31). A common feature in both regulatory roles is the ability of HMGB1 to mediate a stronger binding affinity of the transcription factor for its cognate site, which in many cases correlates with increased transcriptional activation. HMGB proteins exhibit selectivity in their interactions with elements of the nuclear hormone receptor (NHR) superfamily in that they enhance class I receptor binding to their cognate response element, but exert no effect on class II receptors (32). It was proposed that the basis for this lies in the different roles that residues on the immediate carboxyl side of the DBD—referred to as the C-terminal extension (CTE)—play in the two different classes of receptors. Although residues in the CTE in class II receptors interact in the minor groove and are directly involved in HR/HRE binding (33), there is no structural evidence for CTE involvement in binding interactions for the steroid hormone receptors (34, 35, 36). The findings from CTE domain swap experiments between the class I and class II receptors were consistent with the idea that HMGB proteins target the CTE for class I NHRs and can facilitate an HMGB-dependent interaction between CTE residues with the minor groove, but not for the class II NHRs (35).
We demonstrate that the presence of HMGB1 and HMGB2 proteins exert a dramatic effect on ER binding to tandem EREs, in addition to their binding to simple consensus and nonconsensus EREs. HMGB proteins further facilitate the binding of ER to ERE half-sites and the nature of the ER/ERE binding interactions in the half-site differs in the extent of DNA that is protected from nuclease digestions and in the conformation of ER in the ER/ERE complex. The collective data provide evidence that HMGB proteins relax ER binding selectivity and significantly broaden the diversity of mechanisms by which ER can bind strongly to EREs by nonclassical means. This opens up new possibilities for understanding not only the action of ER on gene expression, but also for the interrelationship between HMGB1 and ER in estrogen-sensitive cancers.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSNote Added in ProofREFERENCES |
|---|
), we first examined the effect of HMGB1 and HMGB2 on a single core cERE site. Before these experiments, the influence of a number of factors was determined (data not shown). Optimum binding of ER to cERE occurs at 0.4 mM dithiothreitol (DTT), whereas levels above 4 mM actually inhibit binding. Preliminary experiments showed that the effect of either HMGB1 or HMGB2 protein on ER binding was maximized at 300–400 nM, a level at which there is no HMGB nonspecific binding to DNA. In contrast to previous reports (23), we find no difference in the effect that HMGB1 or HMGB2 exerts on ER binding when any of the various order-of-addition protocols are used. As a result, all experiments were carried out at 0.4 mM DTT, with 400 nM HMGB proteins preincubated with ER before reaction with DNA.
Figure 2A shows the binding of ER to cERE in the presence and absence of HMGB1 or HMGB2. Both HMGB proteins enhance ER binding as shown in the comparison of lanes with the same levels of ER (e.g. lanes 4, 10, and 16). These binding profiles are quantitatively expressed in Fig. 2B, indicating that HMGB1 increases the ER binding affinity approximately 3-fold, whereas HMGB2 increases the binding affinity by approximately 2-fold. A dissociation constant (Kd) value of 11 nM was obtained for ER/cERE, which is comparable to 10 nM previously reported (26, 37). The Kd value is decreased to 6 nM and 4 nM, in the presence of HMGB2 and HMGB1, respectively, and agrees with previous findings (26). Attempts to produce a supershift with anti-HMGB antibodies (from three different sources) were unsuccessful (data not shown), indicating that HMGB1 is not a stable component of the EMSA complex.
|
HMGB1/2 Proteins Enhance ER Binding to Two Tandem EREs in vit B1 ERU and Induce Cooperative Binding at ERE1
The vit B1 ERU contains the two imperfect EREs (ERE2/ERE1) separated by 20 bp, center-to-center (39, 40). Because the sequences in the individual half-sites for both ERE2 and ERE1 differ, ER binding at these sites will not be equivalent. Figure 3A shows the effect of HMGB proteins on the binding profile. At low levels of ER and in the absence of HMGB proteins, the single band is indicative of binding at one ERE (C1). The band intensity increases (up to 10 nM ER, lane 4), then decreases with increasing levels of ER, with the concomitant increase in the intensity of a second band that represents ER binding at both EREs (C2). The C2 band is observed at approximately 10 nM ER and then becomes the prominent band at progressively higher ER levels.
|
shows the quantitative profiles for total DNA binding (C1 + C2) in the presence and absence of HMGB1 and HMGB2. The K50 value, representing the point at which 50% of the DNA is in complex [ERE in complex = (C1+ C2)/[(C1 + C2 + ERE)] indicates that the HMGB proteins increase the overall binding by 3-fold.
Figure 3C shows the strikingly sharp changes in the levels of the individual C1 and C2 complexes as a function of ER and HMGB1 levels. For example, at 12 nM ER, the C1 and C2 levels are approximately equal at 20%, whereas in the presence of HMGB1, virtually all the complex is in the form of C2, whereas C2 represents 50% of complexed ER at 8 nM ER. A cooperative behavior is further supported by the profile for C1, which reaches a maximum at about 6 nM and then drops precipitously. This shows that the 3-fold increase in the ER binding affinity to the vit B1 ERU in the presence of HMGB proteins is due primarily to the increase in the driving force to form the C2 complex. The ER binding profile in the presence of HMGB 2 (data not shown) is comparable.
HMGB1/2 Proteins Facilitate ER Binding to cEREm
Virtually all the naturally occurring bipartite EREs contain imperfect palindromes (6). Although ER binds with only slightly lower affinity to these sites than to cERE (3), the in vitro ER binding affinity to ERE half-sites drops precipitously (8, 9, 11, 41). Because HMGB proteins increased ER binding affinity to consensus and imperfect EREs, it was of interest to determine whether they were able to increase the weak binding affinity of ER to ERE half-sites. To investigate this, we changed each base pair in one of the half-sites of cERE in an attempt to eliminate or significantly reduce ER interaction with this (nonspecific) ERE half-site, cEREm. The base pair changes were made in accord with results from binding studies on base analog and binding interference experiments (3, 42, 43, 44) that designated the importance of individual ER interactions at each base pair and its influence on the binding affinity in the ER/ERE interaction. Because a single mutation in the ERE can reduce the binding affinity by as much as 80% (45), it was anticipated that changes in all the base pairs in a half-site would markedly reduce ER binding. The sequence for cEREm is 5'-TGATGCCTCCGGTCActgGTTGGCAACCCAA-3' (Fig. 1B).
The effect of HMGB1 and HMGB2 on the binding of ER to cERE and to cEREm is shown in Fig. 4A. Although ER binds strongly to cERE in the absence of HMGB proteins (lanes 2–5), ER does not bind to cEREm (lanes 7–10). This indicates that the base pair changes in the 3'-half-site did significantly reduce ER binding affinity. We find that ER does bind weakly to cEREm, exhibiting a Kd value of about 80 nM. However, the presence of HMGB1 or HMGB2 strikingly changes the ER binding profiles, with complex formation evident at ER levels as low as 12 nM. The apparent Kd value for ER/EREm complex in the presence of HMGB protein is approximately 15 nM, which is comparable to the Kd for ER binding to cERE in the absence of HMGB proteins. This indicates that HMGB proteins exert a greater effect on ER binding to ERE half-sites (
5- to 6-fold) than on cERE or imperfect palindromic sites. The bar graph in Fig. 4B clearly shows that at the higher ER levels (above 12.5 nM), ER binding increases sharply, with the percent of complex being as great with cEREm in the presence of either HMGB protein, as found for ER binding to cERE in the absence of HMGB proteins.
|
compare the profile for ER binding to ERE2/ERE1 and ERE2/ERE1m. ER binds to ERE2 at the lowest levels of ER, with binding also occurring at ERE1 at higher ER levels. Although ER binds to both sites in the vit B1 ERU, the base pair changes in the ERE1 half-site abrogates ER binding to the ERE1m site, with binding limited to the ERE2 site (C1) (confirmed by DNase I footprint; data not shown). In the presence of either HMGB1 (lanes 12–15) or HMGB2 (lanes 17–20), ER binding occurs at both ERE2 and ERE1m sites. In addition, ER exhibits a greater binding affinity to ERE2/ERE1m in the presence of HMGB proteins than that observed for ER binding to ERE2/ERE1 in the absence of HMGB proteins (lanes 5, 15, and 20). Figure 5B shows a bar graph that compares the levels of ER binding and also includes the relative levels of one-site (C1) and two-site (C2) binding. In the presence of HMGB protein, C2 clearly becomes the predominant complex with ERE2/ERE1m at ER levels above 12 nM. With either HMGB protein, the percentage of C2 complex remarkably exceeds that observed when ER binds to ERE2/ERE1 in the absence of HMGB proteins. These data show that HMGB proteins also facilitate ER binding to an imperfect ERE half-site.
|
The Conformation of ER Differs in ER/cERE and ER/cEREm Complexes
ER binding to ERE half-sites is weak and exhibits nearly a 10-fold higher Kd value than that observed for binding to cERE. To determine whether the overall conformation of ER changed significantly when bound to cERE and cEREm, ER/ERE complexes were formed at concentrations at about their respective Kd values and digested with trypsin, chymotrypsin, or proteinase K. Because each protease exhibits a different specificity, the three digestion profiles should each be capable of revealing differences in the overall conformation of ER. The comparative digestion profiles for ER/cERE and ER/cEREm are shown in Fig. 6. Distinctive differences in the profiles for the ER complexes are observed for all three proteases, with some unique bands observed for the cERE or cEREm complexes. Interestingly, although the digestion profiles for each protease differed for ER binding to cERE and cEREm, the presence of HMGB1 had no additional effect (data not shown). In addition, the extent of digestion of ER in the complex appears more extensive in all cases for ER/cEREm, suggesting that the ER conformation is less rigid and/or more accessible to the proteases. This is consistent with the notion that, although both ER subunits in the ERE/cERE complex are bound firmly to both consensus half-sites, this does not occur in the ER/cEREm complex. In the latter complex, one ER subunit binds to the nonspecific sequence and permits a greater conformational flexibility that makes it more accessible to the proteases.
|
Figure 7, A and C, shows the FPs for ER bound to the cERE in the vit A2 ERE oligonucleotide, as described in Materials and Methods. The FP on the nontemplate strand (A) indicates strong protection of the symmetrical ERE extending beyond the cERE and covering 23–25 bp, similar to previous reports (38, 39). There is also at least one DNase I hypersensitive site (HS) immediately 3' to the ERE. Interestingly, the presence of HMGB1 does not change either the size of the FP or the position of the HSs. However, DNA protection becomes apparent at lower levels of ER (2- to 3-fold), in agreement with the EMSA findings. The FP for ER binding to the A2 EREm oligonucleotide (Fig. 7, B and D) is distinctly different from that for A2 ERE. The protection observed on the consensus half-site is comparable and extends over the same region as that observed in A2 ERE. However, the protection at the nonspecific DNA in the altered half-site is significantly reduced and does not quite extend through the 5–6 bp that are considered a putative ER binding site. This is observed on both the template and nontemplate strands. The A2 ERE FP extends to the very apparent AA dinucleotide 3' to the ERE on the nontemplate strand, which is clearly not protected in the cEREm. The HS site observed at the 3' end in the nontemplate strand of A2 ERE (Fig. 7A) is also not present in A2 EREm (Fig. 7B).
|
Exonuclease (Exo) III Digestion Mapping of the 3'-Boundaries in the ER/cERE and ER/cEREm Complexes Differ and Support the DNase I Footprints
Exo III nuclease digestion can provide additional evidence about the nature of ER protection of ERE sequences. Because Exo III progressively digests double-stranded DNA from the 3' ends, the extent of digestion on the template and nontemplate strands with A2 EREm may be expected to be distinctively asymmetric. Exo III digestion on the nontemplate and template strands of the ER/A2ERE complex (Fig. 8, A and C, respectively) produces a primary stop point on either strand at 8–9 bp outside the A2 ERE. The multiple stop points that were observed within the cERE in the absence of ER, now converge to a single major site outside the cERE, for the ER/cERE complex (cf. lanes 3 and 4). Similar to that found for DNase I FPs, this symmetrical profile is unaffected by the presence of HMGB1. This suggests that the effect of HMGB1 on interactions that enhance ER/cERE complex formation does not implicate ER-ERE interactions outside of the region already directly involved, in the absence of HMGB1.
|
) shows a dominant stop site about 4 bp 3' to the consensus half-site. In the presence of HMGB1, this stop point occurs at the same position, but additional sites are observed about 6 bp further 3' to the consensus half-site. Interestingly, these sites line up with the stop sites observed on the ER/cERE and restore the protection to the level found in the cERE. When the 3' digestion is directed into the nonspecific half-site (Fig. 8B), a number of sites that were apparent at low levels of ER binding (lane 4, 25 nM ER, about 30% of Kd) are not observed at higher ER levels and a more prominent band within the half-site is observed at 100 nM ER (slightly above Kd). The presence of HMGB1 clearly increases the protection at the nonspecific half-site, with a single cutting site observed at 50 nM ER, whereas the cutting sites that were more in the interior of the EREm are not detected. These data are consistent with HMGB1 significantly enhancing the binding affinity for ER for cEREm.
Figure 9 summarizes the DNase I and Exo III digestion profile data for ER complexed with cERE and cEREm in the vit A2 oligonucleotide. For both nuclease digestions, DNA protection by ER is independent of the presence of HMGB1 and is consistent with a specific and strong binding at the cEREm. The protection on ER/A2ERE is symmetric and exhibits a more extended DNA protection than in the ER/A2EREm complex. The nonspecific half-site in cEREm is only partially protected, consistent with the notion that this ER subunit is less firmly bound than that in the consensus half-site.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSNote Added in ProofREFERENCES |
|---|
Although ER binds very weakly to half-sites, HMGB proteins increase the binding affinity of ER for ERE half-sites approximately 6-fold. ER binding to the ERE1m half-site in the vit B1 ERU further demonstrates that HMGB1 facilitates strong ER binding to an imperfect ERE half-site. Not withstanding this, the DNase I and Exo III protection at a nonspecific half-site is reduced compared with that observed for consensus EREs and is similar to that observed for nonconsensus EREs (our unpublished data; and Ref.38). In addition, the mobility of the EMSA bands and the protease digestion profiles are virtually unaffected by the presence of HMGB proteins. This collection of data suggests that the target for the HMGB effect is primarily localized within or in the immediate vicinity of the ER DBD. Furthermore, the results of the nuclease digestions and the inability to observe a supershift with anti-HMGB1 in EMSA studies collectively support the conclusion that the HMGB protein is not a stable component in the ER/ERE complex, both in solution and in the EMSA conditions.
Plasticity Inherent in ER/ERE Complex Formation
The binding of ER to a spectrum of EREs noted here and elsewhere (3) points to an inherent plasticity of the ER/ERE interactions. The conformational dynamics of ER and DNA can lead to an ensemble of conformations that are more or less energetically accessible for any particular EREs (37, 47, 48). This demands that a broad range of ER-ERE accessible interactions can be used to achieve a stable complex (34, 36, 49, 50). As a result, either one or both binding partners rearrange or change conformation to optimize sterically sensitive interactions within the limits of energetically accessible states (51). The finding of more than 40 different natural EREs (3, 6) illustrates this and is tacit support to the notion that ER and/or the EREs are especially compliant to interactions between suboptimal surfaces, leading to a broad range of stable complexes.
The interaction of HMG proteins with ER/ERE complexes extends the limits of energetically accessible conformations, as reflected in the relaxation of ER binding specificity to include ERE half-sites. HMGB proteins appear to facilitate an admixture of at least two effects that help to override inherent energetic constraints in ER/ERE complexes. First, HMGB proteins bind to and bend DNA nonspecifically and produce a flexure in the DNA that makes the DNA much more pliant (17). In this way, HMGB can reduce the energetic cost of achieving alternate conformations and permit ER to sample a broader spectrum of permissible conformations. Second, HMGB proteins appear to target the CTE in steroid hormone receptors to provide an additional binding interface between these receptors and the minor groove in their cognate steroid response element (32, 35). This may permit steroid hormone receptors to use the CTE in a manner similar to that observed with class II and orphan receptors (4, 52).
A number of structural studies have begun to provide insight into the basis for this plasticity. For example, the crystal structures for ER DBD binding to cERE and to the nonconsensus ERE2 show specific residues undergo different binding interactions to yield stable complexes in both cases (49). Two structures for GR/glucocorticoid response element (GRE) binding also support facile binding rearrangement to suboptimum interfaces. The first structure involves the GR dimer binding to a GRE with two consensus half-sites, separated by the normal 3-bp spacing, whereas the second contains the same two consensus half-sites, but separated by 4 bp (36). However, because the constraint of the 3-bp spacer is thought to be necessary to maintain a strong dimer interface, the second GR subunit cannot bind to the second consensus half-site. In this latter structure, referred to as GRE4s, one GR subunit binds strongly to the consensus half-site, whereas the second GR subunit displays a greater tolerance in its GR/DNA interactions, and as a result, exhibits a relaxed selectivity as it binds to DNA nonspecifically (36). These findings are analogous to ER binding to an isolated cERE half-site and help to build a working model for the ER/cEREm complex. However, HMGB1 facilitates structural changes to increase the binding affinity, leading to a stable complex. The DNase I footprint that was observed for ER/cEREm is consistent with a weaker, more dynamic interaction, in that it is less extended over the nonspecific DNA in the cEREm.
The presence of HMGB proteins must invoke additional interactions to account for the increased ER binding affinity. Although some extra binding energy may be derived from enhancing the ER interactions with individual consensus half-sites in cERE, our Kd data suggest that they would lead to relatively small gains. Previous findings suggest the involvement of the CTE. In a seminal paper on ER DBD binding to EREs (53), the 12 residues immediately C-terminal to the core DBD (66 residues) were required to bind to cERE and that additional residues in the CTE were absolutely required for binding to nonconsensus EREs. It was also noted that residues in this region greatly increased the binding affinity of the 84 residue ER DBD to ERE. Together, these findings indicate that the energetics of CTE interactions with DNA are very sensitive to experimental conditions. However, Edwards has shown that HMGB proteins exert their effect on class I hormone receptors, but not class II, nonsteroid nuclear hormone receptors (32, 33). Experimental support for HMGB proteins targeting the CTE in SHRs was obtained from CTE domain swap experiments with PR and TR, in which HMGB2 enhanced the binding of the TR(PR-CTE) hybrid to TRE, but not for PR(TR-CTE) binding to GRE (35).
Hit-and-Run Mechanism
The collective evidence indicates that HMGB1 is a transient and unstable component of the ER/ERE complex. We propose that HMGB proteins enhance the binding interactions in the ER/ERE complexes at two interfaces by improving the orientation of interactions in the major groove and/or help facilitate CTE in gaining better access to and interacting more favorably in the minor groove. HMGB proteins are architectural proteins that have been demonstrated to induce DNA flexure and lead to a much less rigid DNA (17). In this model, HMGB protein binds in the minor groove in the vicinity of the ERE to pry open the groove. This permits the residues in the CTE to rearrange, gain greater access to and interact in the minor groove, which in the absence of HMGB, is energetically less accessible. Because both ER and HMGB proteins are undergoing a dynamic association and dissociation with DNA, ER can continue to maintain this binding affinity as long as HMGB remains available. In this way, HMGB proteins can dissociate from the complex during or after the ER/ERE interactions. In the complexation, with and without HMGB proteins, an ER/ERE complex is formed, but the thermodynamic constants differ, Ka' > Ka. This is a case of two equilibria (equations 1 and 2) in which the composition of the two complexes is identical, but the interactions between the components are not. As a result, the binding constants differ. In the absence of HMGB
 | (1) |
 | (2) |
A more general hit-and-run mechanism for HMGB with steroid hormone receptors has been proposed previously (15). The two alternative mechanisms propose (32, 35, 55) either that HMGB1 is a stable component of the steroid receptor/steroid response element complex and takes part in the interactions or that the HMGB protein is a transient component. Our data show that, in the case of ER and under our conditions, the latter scenario applies.
This hit-and-run mechanism is also consistent with what is known about protein mobility for HMGB1 and for one class of nuclear hormone receptor, the GR. Using FRAP, it was shown that GR was mobile both in vivo and in vitro, with fluorescence recovery in approximately 2 sec (56). HMGB1 displays a similar behavior, but exhibits a residence times much less than 1 sec, giving it the current distinction of the most mobile protein in the nucleus (57).
ER Binding Affinity Versus Binding Selectivity
HMGB proteins increase ER binding affinity to ERE half-sites and to nonconsensus and consensus EREs, with the order being: half-sites > nonconsensus > consensus. Interestingly, if both half-sites in cERE are changed in this same manner to make effectively a nonspecific DNA, HMGB increases ER binding to this nonspecific DNA by approximately three times (data not shown). The underlying implication of this trend is that because there is a gain in binding affinity to all sites, there is little or no gain in binding selectivity (KERE/Knonspecific), especially for the classical cEREs (3). These results beg the question as to what actually drives ER binding selectivity to functionally active sites? It seems reasonable that additional factors are essential to gain the high level of binding selectivity required. They may be complexed with ER before, or immediately after, DNA binding to help recognize and filter functional recognition sites from those that serve no direct functional purpose. In addition, HMGB proteins may also modulate the interaction of the ER/ERE complex with coactivators and corepressors. Certainly, one may expect that an equilibrium mixture of ER complexes will be competing for the functional sites within the genome.
Although many DNA-binding proteins exhibit a significant binding selectivity on their own, many do not. This latter family of transcription factors is extensive, including cAMP receptor protein (58), lac repressor (59), 434 repressor (60, 61), homeodomain proteins (62), and the nuclear hormone receptors (34, 36, 49, 50). In addition, restriction endonucleases, EcoRV, and Taq I, whereas exhibiting exquisite functional specificity, strikingly lack binding selectivity in the absence of divalent cation cofactors (63). It is found, however, that the structures of the EcoRV/DNA complexes are dependent on whether the EcoRV is binding to its recognition sequence and to nonspecific DNA. As a result, the subsequent and selective binding of the Mg ion cofactor to EcoRV bound to its recognition site leads to a functional complex, whereas complexes at other sites do not. Although there are unquestionably elements in ER and ERE that contribute significantly to ER binding selectivity (3), the binding affinity to most EREs does not differ by more than 100, with most differing by 10 or less (Ref.3 and references therein). This parallels the selectivity behavior for other steroid hormone receptors (64). The functional selectivity for ER and EcoRV may have a similar basis, in which a significant component of the binding discrimination does not appear until a more complete array of cognate cofactors (a combinatorial control) are assembled (65).
Possible Relevance to the Biological Activity of ER
The ubiquitous HMGB proteins facilitate ER binding to ERE half-sites and enhance the binding of steroid hormone receptors to their cognate response elements (22, 23, 24, 32). Considering the genetic screen that identified the majority of functional in vivo ERE sites as half-sites (13) and the increasing number of genes that have ERE half-sites (9, 10, 11, 12), our data suggest that HMGB proteins may play a critical role in the regulation of many estrogen-responsive genes.
The consensus half-sites in the response elements for estrogen and thyroid hormone receptors are identical, and it has been shown that there is a cross talk between ER and TR that produces differential regulation of specific genes (66, 67). In addition, the recognition sequence targeted by a number of orphan receptors contains the consensus ERE half-site that is preceded by a short AT-rich sequence (52). The plasticity of ER, together with the ability of HMGB1 to further relax ER binding specificity for EREs, represents a potential avenue by which HMGB proteins could significantly influence activities and the regulatory interplay between many of the receptors in the nuclear receptor superfamily (68, 69). Furthermore, because HMGB proteins increase the binding affinity for all steroid hormone receptors, they are logical candidates to serve as common mediators in the binding and functional cross talk between class I hormone receptors.
In addition, there may be a direct regulatory link between the expression of HMGB1 and ER. Regulatory ERE half-sites are found in the promoter and in intron 1 of the HMGB1 gene (12). In this regard, the levels of HMGB1 increase significantly on treatment of MCF-7 human breast cancer cells with estrogen (70, 71). It has also been reported that HMGB1 exhibits antiapoptotic activity that may play a role in mammary gland apoptosis (72). These collective findings suggest that the interrelationship between estrogen receptors and HMGB1 and their effect on estrogen-responsive genes may play a vital role in estrogen-sensitive cancers.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSNote Added in ProofREFERENCES |
|---|
was obtained from PanVera Corp. (Madison, WI) and stored in 50 mM Tris-HCl (pH 8.0), 500 mM KCl, 2 mM DTT, 1 mM EDTA, 1 mM sodium orthovanadate, and 10% glycerol. The ER binding activity was determined to be essentially 100% by standard EMSA titrations with cold oligonucleotides (74).
DNA Oligonucleotides
All oligonucleotides were obtained from Integrated DNA Technologies (Coralville, IA). Oligonucleotides (31 bp) that contained the cERE- (-GGTCActgTGACC-) or an ERE half-site (cEREm), with the latter having all 5 bp in the 3'-core consensus half-site changed (-GGTCActgGTTGG-), were used in the EMSA studies. Oligonucleotides for the Xenopus laevis vit B1 gene vit B1 ERU that contain tandem ERE2 and ERE1 sites were synthesized for a 53-bp fragment. The 3'-half-site in ERE1 was also changed, as noted above, to produce ERE2-ERE1m. DNase I and Exo III digestion studies were performed using an 83-bp oligonucleotide that contained the X. laevis vit A2 ERE from –366 to –284 (75). This vit A2 sequence, however, contains not only the A2 consensus ERE at –332 to –317, but also a previously unreported imperfect palindromic ERE upstream at –311 to –298. The two half-sites in the latter sequence (5'-AGTTAatgTAACC-3') were changed to (5'-CCAACatgGTTGG-3') to abrogate ER binding. In addition, the 3'-half-site of the A2 consensus ERE was changed similarly, as noted above for the cERE.
The 31- and 53-bp oligonucleotides used are shown below, with the ERE underlined in bold type and the 3-bp spacer in lower case letters.
cERE /cEREm
5'-TGATGCCTCCGGTCActg-T/G-G/T-A/T-C/G-C/G-CAACCCAA-3'
vit B1 ERU: ERE2-(ERE1/EREm)
5'-TGATGCCTCCAGTCActgTGACCCAACCCAAGTTAtca-T/G-G/T-A/T-C/G-C/G-TCTTA GTTGG-3'
EMSAs
EMSAs were performed as generally outlined previously (21). The DNA fragments (consensus ERE or cEREm) were 32P-end-labeled and incubated with ER alone or both ER and HMGB proteins in ER binding buffer [80 mM KCl, 10% glycerol, 15 mM Tris HCl (pH 7.9), 0.2 mM EDTA, 0.4 mM DTT, 100 ng/µl BSA] with final concentration of poly(deoxyinosinic-deoxycytidylic) acid (dI-dC)]·poly (dI-dC) at 0.4 ng/µl for the 31-bp ERE probe and at 0.7 ng/µl for the 53 bp ERE oligonucleotide. The proteins were incubated with buffer for 10 min at 4 C, and then the DNA reaction mixture was added, followed by a 15-min incubation at 4 C. All samples were electrophoresed in 0.35x TBE buffer at 200 V for 90 min in 4% or 5% nondenaturing polyacrylamide gels at 4 C. After electrophoresis, the gels were dried and exposed to x-ray film at –80 C.
The Kd values were obtained by titration of 100 pM DNA over a range of ER concentrations, with equilibrium established after 15 min at 4 C. Experiments in which HMGB proteins were present were carried out with 400 nM HMGB protein. The band intensities for the complex and free DNA were measured by exposing the dried gel to a PhosphorImager screen, which was scanned using the Molecular Dynamics (Sunnyvale, CA) PhosphorImager system. The ImageQuant software program (Molecular Dynamics) was used to measure the band intensities. The Kd value for the ER/ERE complex is equal to the free ER concentration at which there are equal concentrations of [DNA] and [ER/DNA] (i.e. Kd = [ER][DNA]/[[ER/DNA]). The percent of complex was plotted vs. the concentration of ER to generate the binding curves. The best fit of the data was derived (using Sigma Plot for PC) using over 40 data points from four independent determinations.
The polyclonal antibodies to HMGB1 that were used in the supershift experiments were obtained from R. Roeder, Santa Cruz Biotechnology (Santa Cruz, CA), and Upstate Biotechnology (Lake Placid, NY), with anti-ER (H222) obtained from G. Green. For supershift experiments, antibodies were added for an additional 10 min at 4 C after the formation of the ER/ERE.
Protease Digestion Profiles
32P-end-labeled DNA fragments (31 bp containing either the consensus ERE or the mutant ERE) were combined with ER (6.3 or 63 nM, respectively) in ER binding buffer. After a 20-min incubation on ice, trypsin, chymotrypsin, or proteinase K was added to levels indicated in the figure legend. The samples were incubated for an additional 10 min on ice and run as outlined above.
DNase I Digestions
DNase I footprints were obtained on DNA fragments (85 mers) containing the modified X. laevis vit A2 gene labeled at the 5' end of either the template or nontemplate strand. The labeled oligonucleotide was annealed with its complementary strand, after which the DNA was incubated at room temperature for 20 min with varying amounts of ER and, where indicated, 400 nM HMGB-1, in ER binding buffer containing 1.1 ng/µl poly(dI-dC). DNase I, Mg2+, and Ca 2+ were added to final concentrations of 0.008 U/µl, 1.1 mM, and 0.04 mM, respectively. The samples were cleaved for 1 min at room temperature after which the digestion was terminated by addition of DNase I stop solution to give final concentrations of 5 mM EDTA, 0.2% sodium dodecyl sulfate, and 10 ng/µl tRNA. The DNA was precipitated twice with ethanol, dried under vacuum, resuspended in loading buffer, heated at 90 C for 10 min, and electrophoresed on a DNA sequencing gel.
Exo III Digestions
DNA fragments labeled at the 5' end of either the template or nontemplate strand were annealed with the complementary strand and then reacted with increasing levels of ER and, where indicated, 400 nM HMGB-1 as described for DNase I footprints. Exo III and Mg2+ were added to final concentration of 1 U/µl and 5 mM, respectively. The samples were cleaved for 1 min at room temperature after which the digestion was terminated by adjusting to 15 mM EDTA, 0.6% sodium dodecyl sulfate, and 31 ng/µl tRNA. The DNA samples were processed as described for DNase I footprints.
|
Note Added in Proof |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSNote Added in ProofREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Abbreviations: AR, Androgen receptor; cERE, consensus ERE; cEREm, ERE consensus half-sites; CTE, C-terminal extension; DBD, DNA-binding domain; dI-dC, deoxyinosinic-deoxycytidilic acid; DNase, deoxyribonuclease; DTT, dithiothreitol; E2, estrogen; ER, estrogen receptor ; ERE, estrogen response element; ERU, estrogen response unit; Exo, exonuclease; FP, DNase I footprint; FRAP, fluorescence recovery after photobleaching; GR, glucocorticoid receptor; GRE, glucocorticoid response element; HMGB, high mobility group B; HRE, hormone response element; HS, hypersensitive site; Kd, dissociation constant; NHR, nuclear hormone receptors; PIC, (transcriptional) preinitiation complex; PR, progesterone receptor; SHR, steroid hormone receptor; vit, vitellogenin.
Received for publication March 24, 2004. Accepted for publication July 9, 2004.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSNote Added in ProofREFERENCES |
|---|
) and the presence of HMGB1 (O) or HMGB2 (
). The data represent the average of points from four independent titrations and were fit by Sigma Plot.
) in the presence of HMGB1. The percent of the DNA fragment in each complex, as denoted by 
was obtained from EMSA profiles and plotted as a function of ER concentration. The data represent the average of points from four independent titrations and were fitted to a smooth curve by Sigma Plot.
