Article
Reversal of the DNA-Binding-Induced Loop L1 Conformational Switch in an Engineered Human p53 Protein Soheila Emamzadah 1 , Laurence Tropia 1 , Ilena Vincenti 2 , Benoît Falquet 2 and Thanos D. Halazonetis 1 1 - Department of Molecular Biology, University of Geneva, 1205 Geneva, Switzerland 2 - Department of Biochemistry, University of Geneva, 1205 Geneva, Switzerland
Correspondence to Thanos D. Halazonetis: [email protected] http://dx.doi.org/10.1016/j.jmb.2013.12.020 Edited by C. Kalodimos
Abstract The gene encoding the p53 tumor suppressor protein, a sequence-specific DNA binding transcription factor, is the most frequently mutated gene in human cancer. Crystal structures of homo-oligomerizing p53 polypeptides with specific DNA suggest that DNA binding is associated with a conformational switch. Specifically, in the absence of DNA, loop L1 of the p53 DNA binding domain adopts an extended conformation, whereas two p53 subunits switch to a recessed loop L1 conformation when bound to DNA as a tetramer. We previously designed a p53 protein, p53FG, with amino substitutions S121F and V122G targeting loop L1. These two substitutions enhanced the affinity of p53 for specific DNA yet, counterintuitively, decreased the residency time of p53 on DNA. Here, we confirmed these DNA binding properties of p53FG using a different method. We also determined by crystallography the structure of p53FG in its free state and bound to DNA as a tetramer. In the free state, loop L1 adopted a recessed conformation, whereas upon DNA binding, two subunits switched to the extended loop L1 conformation, resulting in a final structure that was very similar to that of wild-type p53 bound to DNA. Thus, altering the apo structure of p53 changed its DNA binding properties, even though the DNA-bound structure was not altered. © 2014 The Authors. Published by Elsevier Ltd. All rights reserved.
Introduction The gene encoding the tumor suppressor p53 protein is the most frequently mutated gene in human cancer [1,2]. The p53 protein is activated in response to DNA damage, and once activated, it enhances transcription of genes that induce cell cycle arrest, apoptosis, or senescence [3,4]. Since oncogenes induce DNA damage, it is not surprising that p53 is the most frequently mutated gene in human cancer [5]. The p53 protein contains two independently folding domains: a sequence-specific DNA binding domain at its center (also known as the core domain) and a homo-tetramerization domain toward the C-terminus [6–8]. These two domains are flanked by three flexible, unstructured regions: an N-terminal transactivation region, a linker region between the DNA binding and tetramerization domains, and a C-terminal basic region
[9]. Most of the p53 mutations in human cancer target the sequence-specific DNA binding domain [1,2]. Several three-dimensional structures of p53 have been determined [10–18]. Most of these encompass only the core domain, either by itself or in complex with DNA [10–15]. Recently, we solved structures of a p53 polypeptide containing both the DNA binding and homo-oligomerization domains in its free form and bound to specific DNA sites [16,17]. Consistent with wild-type p53 (p53wt) binding DNA as tetramer, we observed four p53 subunits bound to the specific DNA site. The specific DNA site encompasses four tandem pentanucleotide repeats, and each repeat is recognized by one p53 DNA binding domain. Unlike the structures containing only the core domain of p53 bound to DNA, we observed a conformational switch within loop L1 upon sequence-specific DNA binding. The conformational change of loop L1 involves only two of the four p53 subunits bound to DNA, those
0022-2836/$ - see front matter © 2014 The Authors. Published by Elsevier Ltd. All rights reserved. J. Mol. Biol. (2014) 426, 936–944
937
Reversal of the DNA-Binding-Induced Loop L1 Confor
a
b
DNA: CDKN1A
v/[p53]
DNA: CONS
v/[p53] 0.015
0.020 0.015
p53FG (KD) 6.93+2.48 nM
0.012
p53FG (KD) 11.0+1.0 nM
0.009
0.010 0.006
p53wt (KD) 14.5+1.7 nM
0.005
p53wt (KD) 9.21+0.23 nM
0.003 0 0.8
0 0.7
0.8
0.9
1.0 v
c
0.85
0.95
0.9
1.0 v
d DNA: CDKN1A p53FG (koff) 0.16+0.01 s−1 0.16+0.01 s−1 0.16+0.02 s−1
ln(RUmax/RU) 1.0 0.8
1.5
p53FG (koff) 0.33+0.02 s−1 0.35+0.04 s−1 0.34+0.04 s−1
1.2
p53wt (koff) 0.077+0.003 s−1
0.6
DNA: CONS
ln(RUmax/RU)
0.9
0.4
0.6
0.2
0.3
0
p53wt (koff) 0.105+0.006 s−1 0.121+0.005 s−1 0.124+0.010 s−1
0 0
2
4
6
8
10
12 t(s)
0
2
4
6
8
10 t(s)
Fig. 1. Amino acid substitutions within loop L1 decrease the half-life of p53–DNA complexes despite increasing the affinity for specific DNA. (a and b) Scatchard plots for binding of human p53wt and p53FG (residues 79–393) to the native p53 response element present in the CDKN1A gene (a) or to a consensus p53 response element (b). The KD values (mean ± 1 SD) shown correspond to − 1/slope of the fraction of bound DNA divided by the concentration of free p53 (v/[p53]) as a function of the fraction of bound DNA (v). (c and d) Plot of the natural logarithm of RUmax divided by RU [ln(RUmax/RU)] over time (t) in seconds (s) for binding of p53 and p53FG to CDKN1A (c) and consensus (d) p53 response elements. The koff values shown (mean ± 1 SD) correspond to the slope of the plot. The kinetic experiments were performed with p53 protein concentrations of 118 nM (blue symbols), 178 nM (red symbols), and 267 nM (green symbols).
subunits recognizing the most 5-prime and most 3-prime of the four tandem pentanucleotide repeats [16]. We refer to these two repeats as the outer repeats. On the basis of the conformational switch in loop L1, we proposed that p53 recognizes sequencespecific DNA via a two-step, induced-fit mechanism [16,17,19]. The first step involves an initial rigid-body interaction between p53 and DNA, followed (in the second step) by conformational changes that allow a better fit between p53 and the DNA. To test this model, we analyzed p53 mutants containing amino acid substitutions within loop L1 [16]. The design of these mutants was guided by the free and DNAbound conformations of p53. The most interesting mutant, hereafter referred to as p53FG, contained two substitutions within loop L1: Ser121 with Phe (S121F) and Val122 with Gly (V122G). These two substitutions were predicted to favor the recessed DNA-bound conformation of loop L1 and also to impart conformational flexibility to the loop, thereby
decreasing the kinetic barrier between the free and DNA-bound conformations. Fluorescence anisotropy experiments performed with p53wt and p53FG revealed that p53FG bound to DNA with higher affinity than p53wt [16]. One would have expected that the higher affinity for DNA would have translated into longer half-lives for the p53–DNA complexes. However, the half-lives of the p53FG–DNA complexes were shorter than those of the p53wt–DNA complexes [16]. This unexpected observation prompted us to propose that the conformational switch provides a kinetic barrier to p53 DNA binding and, thereby, regulates the residency time of p53 on DNA. The design of p53FG was guided by the structure of p53wt bound to DNA [16]. While we had predicted that the S121F and V122G substitutions would facilitate the recessed loop L1 conformation, this was never demonstrated experimentally. Further, how Phe121 would be accommodated in the p53– DNA structure was unclear, since we had not solved
938
Reversal of the DNA-Binding-Induced Loop L1 Confor
a
b
p53FG/L1 p53wt
p53FG/L1 p53wt Cep−1 S238 S121 F121
F121
L1
S238
S121
Fig. 2. Three-dimensional structure of the p53FG DNA binding domain in its apo form. (a) Overall structure of p53FG (colored green; except for loop L1, colored blue) superimposed on the structure of the DNA binding domain of p53wt (gray). (b) Superimposition of the loop L1 structures and of the beta-strand C-terminal to loop L1 of p53FG in its apo form [colored as in (a)], p53wt (gray), and Cep-1 (yellow). The side chains of Phe121 (p53FG), Ser121 (p53wt), and Ser238 (Cep-1; corresponding to Ser121 of p53wt) are indicated. The orientation of the top view is the same as in (a); the bottom view is obtained by a 90° rotation along the indicated axis. The r.m.s. difference between the positions of the alpha-carbon atoms of loops L1 of p53wt and p53FG (residues 113–123), in their apo form, is 3.5 Å.
the structure of p53FG in complex with DNA. Here, we address these questions by determining the three-dimensional structures of p53FG in its apo form and in complex with specific DNA. In addition, we use surface plasmon resonance (SPR), as a different biophysical method than the one used in our previous study [16], to characterize the DNA binding properties of p53wt and p53FG. The new results further support our model that conformational changes in loop L1 allow p53 DNA binding affinities and kinetics to be regulated independently of each other. Thus, p53 could serve as a paradigm toward understanding how transcription factors occupy their specific sites in vivo. Specifically, large differences in half-lives of protein–DNA complexes, made possible by conformational changes, rather than large differences in affinities may allow discrimination of specific from non-specific DNA sites.
Results Amino acid substitutions within loop L1 decrease the half-life of p53–DNA complexes despite increasing DNA binding affinity We previously used fluorescence anisotropy to measure binding affinities and off-rates of various p53 proteins for sequence-specific DNA [16]. We found that p53FG bound specific DNA with higher affinity than p53wt. Even though the higher affinity of p53FG for DNA should translate to longer residency times of p53 on DNA, quite unexpectedly, the half-life of the p53FG–DNA complexes was shorter than the half-life of the p53wt–DNA complexes. To determine whether these results could be repro-
duced with a different method, we employed SPR, which also allows affinity and kinetic binding constants to be determined [20,21]. We examined binding of p53 to two different DNA sequences: one containing the natural p53 response element in the CDKN1A (p21) gene and one containing a consensus p53 binding site (Fig. 1). Oligonucleotides containing these two DNA sequences were synthesized with a biotin tag to allow them to be immobilized on streptavidin-coated SPR sensor chips. Binding measurements were performed using purified, recombinant p53 proteins corresponding to residues 79–393 of the full-length human p53 protein. These proteins lacked only the transactivation domain, which is not thought to be involved in DNA binding. One protein had a wild-type loop L1 sequence (p53wt), while the other (p53FG) contained two amino acid substitutions, S121F and V122G, in loop L1. Both p53 proteins also contained amino acid substitutions that enhance the thermal stability and solubility of the DNA binding domain. These substitutions do not affect p53 function but render p53 amenable to biophysical analysis [16]. The first series of SPR experiments was used to determine DNA binding affinities (expressed by the dissociation constant KD). The results were visualized using Scatchard plots showing the fraction of bound DNA divided by the concentration of free p53 (v/[p53]) as a function of the fraction of bound DNA (v). Wild-type p53 bound the CDKN1A response element with a dissociation constant of 14.5 ± 1.7 nM (mean ± 1 SD), whereas p53FG had a dissociation constant of 11.0 ± 1.0 nM (P b 0.05 for the difference in KD values; Fig. 1a). For the consensus binding site, the calculated dissociation constants were 9.21 ± 0.23 and 6.93 ± 2.48 nM for p53wt and p53FG, respectively
939
Reversal of the DNA-Binding-Induced Loop L1 Confor
a
b
3’
5’
A
B p53FG (B) p53FG (A) p53FG (apo) p53wt (apo)
C D
5’
3’
c p53wt
p53FG
Fig. 3. Three-dimensional structure of a multidomain p53FG oligomer bound to DNA. (a) Overall structure of two p53FG dimers bound to DNA. The p53 subunits are labeled A–D. (b) Superimposition of loops L1 of subunits A (green) and B (yellow) of p53FG bound to DNA to loops L1 of p53wt (gray) and p53FG (blue) in their apo forms. The r.m.s. difference between the positions of the alpha-carbon atoms of loop L1 of p53FG bound to the outer DNA repeat (subunit A) relative to loops L1 of p53wt and p53FG in their apo forms is 3.9 and 3.4 Å, respectively. The r.m.s. difference between the positions of the alpha-carbon atoms of loop L1 of p53FG bound to the inner DNA repeat (subunit B) relative to loops L1 of p53wt and p53FG in their apo forms is 0.8 and 3.2 Å, respectively. (c) Cartoon depicting the conformational switching of loops L1 of p53wt and p53FG upon DNA binding.
(Fig. 1b). The difference in affinities of wild-type p53 and p53FG for specific DNA is not as high as the one we previously observed by fluorescence anisotropy (7.75 ± 0.52 versus 1.96 ± 0.20 nM for p53wt and p53FG, respectively [16]); however, the SPR measurements still showed a higher DNA binding affinity for p53FG than for p53wt. In order to calculate off-rate constants, we relied on the SPR response units (RUs) measured during dissociation of p53 from specific DNA over time, using the method of O'Shannessy et al. [20], which posits that the slope of the plot of ln(RUmax/RU) over time is equal to the off-rate constant (koff). The koff values for the complexes of p53wt and p53FG with the
CDKN1A response element were 0.077 ± 0.003 and 0.16 ± 0.01 s −1, respectively (P b 0.005 for the difference in koff values; Fig. 1c). For the consensus binding site, the off-rates were 0.121 ± 0.005 and 0.35 ± 0.04 s − 1 for p53wt and p53FG, respectively (P b 0.005 for the difference in koff values; Fig. 1d). These latter values were consistent with those previously obtained by fluorescence anisotropy (0.068 ± 0.008 versus 0.34 ± 0.03 s −1 for p53wt and p53FG, respectively [16]). Thus, the off-rate values obtained by two different techniques reveal that the half-lives of the p53FG–DNA complexes were shorter than the halflives of the p53wt–DNA complexes despite p53FG having higher affinity for specific DNA than p53wt.
940
Reversal of the DNA-Binding-Induced Loop L1 Confor
a
b H2 H2
G2a
G2b
R280
R280 A276
A276
K120
L1
G122
C277 F121
C3a
T3b R283 G4’b
F121
C277
G5’a
A5’b
B
K120
A
G122
L1
c p53FG (B) p53FG (A) p53wt (B) p53wt (A) p53cr
Fig. 4. Sequence-specific contacts of p53FG with DNA. (a) Interface of subunit B of p53FG with DNA. (b) Interface of subunit A of p53FG with DNA. The subunits are labeled and are shown in the same orientation as subunit B in Fig. 3a. (c) Superimposition of loops L1 of subunits A (green) and B (yellow) of p53FG bound to DNA to loops L1 of subunits A (blue) and B (gray) of p53wt bound to DNA. The conformation of loop L1 of mouse p53 cross-linked (p53cr) to DNA (purple [14]) is also shown. The r.m.s. difference between the positions of the alpha-carbon atoms of loop L1 of p53FG and p53wt bound to the outer DNA repeats (subunits A) is 2.1 Å; the r.m.s. difference between the positions of the alpha-carbon atoms of loop L1 of p53FG and p53wt bound to the inner DNA repeats (subunits B) is 1.6 Å.
Structure of p53FG in its apo (free) form The structure of the DNA binding domain of p53FG encompassing the two substitutions within loop L1 was determined at a resolution of 1.25 Å (Table S1). The overall structure was similar to that of wild-type p53, except for the conformation of loop L1, which in p53FG, remarkably, adopted a recessed conformation (Fig. 2a). So far, all structures of p53 and of its family member p73 have been shown to adopt an extended loop L1 conformation in their apo form [10,22,23] with only one exception, that of Caenorhabditis elegans p53 [24]. Superimposition of the L1 loops of p53wt, p53FG, and C. elegans p53 (Cep-1) revealed that loop L1 of p53FG is as recessed as that of C. elegans p53 (Fig. 2b). Formally, the recessed conformation of loop L1 in p53FG could be attributed to crystal packing contacts, since the side chain of Phe121 is within 3.5–4.8 Å of the side chains of Arg174, Pro177, and Glu180 of a symmetry-related p53 molecule in the crystal lattice (Fig. S1). However,
most of the crystal packing contacts between these two subunits are not mediated by loop L1, but rather by helix H2 (Fig. S1). Further, changes in the conformation of loop L1 as a result of crystal packing contacts have not been observed in any of the other p53 crystal structures determined to date. Thus, we argue that the recessed conformation of loop L1 in the p53FG crystal structure is representative of its apo conformation in solution. Structure of a p53FG oligomer bound to sequencespecific DNA The structure of a p53FG polypeptide encompassing both the core and homo-oligomerization domains in complex with a consensus-specific DNA site (CONS26) was determined at a resolution of 2.9 Å (Fig. 3a and Table S1). The overall structure of this homo-oligomerizing p53FG polypeptide bound to DNA was very similar to the previously determined structures of homo-oligomerizing p53wt in complex with DNA (Fig. S2). Specifically, the two subunits of
Reversal of the DNA-Binding-Induced Loop L1 Confor
the p53FG tetramer recognizing the central two pentanucleotide repeats of the p53 DNA site (inner repeats) adopted the extended loop L1 conformation, whereas the two subunits recognizing the outer repeats adopted the recessed loop L1 conformation (Fig. 3b). Given that, in the apo form, loop L1 of p53FG is recessed, the structure reveals a conformational switch upon DNA binding. However, whereas in p53wt, the conformational switch involves the loops of the subunits binding the outer pentanucleotide repeats (from an extended to a recessed conformation); in p53FG, the switch affected the loops of the subunits binding the inner repeats (from a recessed to an extended conformation) (Fig. 3c). In regard to the subunits binding the inner pentanucleotide repeats, we observed few differences in the contacts of p53FG with DNA, relative to the contacts of p53wt with DNA [16,17]. Specifically, whereas in p53wt, the side chain of Lys120 contacts atoms N7 and O6 of the guanine at position G4′b [17]; in p53FG, the side chain of Lys120, due to steric interference with the side chain of Phe121, was pushed almost outside the major groove of the DNA and contacted the DNA sugar-phosphate backbone (Fig. 4a and Fig. S3). Further, the side chain of Ser121 of p53wt interacts with the sugar-phosphate backbone of the cytosine at position C5c [17], but in p53FG, the side chain of Phe121 was close to oxygen O6 of the guanine at position G4′b and did not interact with the DNA sugar-phosphate backbone (Fig. 4a). In regard to the subunits binding the outer pentanucleotide repeats, the only difference in the contacts of p53FG with DNA, compared to the contacts of p53wt with DNA, involved residue 121. In the structure of p53wt bound to the native site present in the CDKN1A gene, Ser121 contacts the DNA backbone [17]. The corresponding residue, Phe121, in the p53FG–DNA structure was oriented toward the interior of the protein. In this conformation, due to its large hydrophobic side chain, Phe121 could help stabilize the recessed conformation of loop L1 (Fig. 4b). The overall similar contacts of p53wt and p53FG to DNA are further reflected in the similar DNAbound conformations of loops L1 of these two proteins. Thus, the subunits of p53wt and p53FG recognizing the inner repeats have similar extended loop L1 conformations, while the subunits recognizing the outer repeats have similar recessed loop L1 conformations (Fig. 4c).
Discussion We previously described the structures of p53 tetramers bound to a consensus-specific DNA site (CONS26) or to the native CDKN1A p53 response
941 element [16,17]. These structures showed a conformational switch involving loops L1 of the DNA binding domains of p53. The conformational switch involved only two of the four subunits, those subunits bound to the outer pentanucleotide repeats of the p53 response element. In these two subunits, the loop L1 conformation switched from extended (in the apo form) to recessed (in the DNA-bound form). In our previous study, we predicted that two substitutions within loop L1, S121F and V122G, would facilitate the conformational switch from the extended to the recessed conformation on the rationale that the side chain of Phe121 could potentially form hydrophobic interactions stabilizing the recessed conformation of loop L1 and that Gly122 lacks a hydrophobic side chain to stabilize the extended loop L1 conformation [16]. The structure of p53FG in its apo form, determined here, supports these predictions since, in the absence of DNA, loop L1 of p53FG adopted a recessed conformation. Upon DNA binding, p53FG underwent a conformational switch, such that, in two subunits, loop L1 adopted an extended conformation. Thus, even though the structures of p53wt and p53FG differ in their apo form, upon binding to specific DNA, these two p53 proteins adopt essentially identical structures. Cep-1, the p53 homolog in C. elegans, adopts a recessed loop L1 conformation in the apo form, reminiscent of that observed in the apo form of p53FG [24]. We envision that Cep-1, like p53FG, will also switch conformation upon DNA binding, thereby explaining the paradox as to how p53wt and Cep-1 recognize identical DNA sequences despite having different loop L1 structures in their apo forms [24]. Most previously determined structures of p53 DNA binding domains in complex with DNA have not revealed any conformational switch in p53 upon DNA binding [10–15]. The only exception was the structure of a p53 tetramer–DNA complex, in which p53 had been cross-linked to the DNA [14]. In this case, all four p53 subunits adopted a recessed loop L1 conformation. More recently, subtle conformational changes have been observed in loop L1 of p73 bound to DNA [25]. However, the extent of these conformational changes is more limited than the changes described here. We speculate that the previous studies have failed to observe major loop L1 conformational changes upon DNA binding because they examined polypeptides corresponding only to the DNA binding domain of p53. The affinity for specific DNA of polypeptides containing only the core domain of p53 is orders of magnitudes lower than the affinity of polypeptides containing both the DNA binding and oligomerization domains [26]. Thus, only the latter polypeptides may recapitulate physiological DNA binding. Based on fluorescence anisotropy measurements, we had previously argued that the conformational
942 switch in loop L1 of p53 allows DNA binding affinities and off-rates to be regulated independently of each other [16]. Specifically, we had observed that the p53FG mutant has higher affinity for specific DNA than p53wt; however, its residency time on DNA is shorter than that of p53wt. We reproduced here these results employing SPR. Thus, two biophysical methods now support our premise that the conformational switch of loop L1 allows more precise control of the DNA binding properties of p53. The DNA binding properties of p53FG can be explained from the perspectives of either the induced-fit model or the conformational selection model [27]. In both cases, we speculate that loop L1 is conformationally more flexible in p53FG than in p53wt, a hypothesis supported by the crystallographic B-factor values (Fig. S4). According to the induced-fit model, higher loop L1 flexibility would result in less energy being required for loop L1 in p53FG to adopt the DNA-bound conformation. In turn, this would lead not only to higher affinity but also to higher on-rates and higher off-rates, since the off-rate is equal to the on-rate multiplied by the dissociation constant [16]. From the perspective of the conformational selection model, the conformational flexibility would result in a higher fraction of molecules adopting the DNA-bound conformation, leading not only to higher on-rates for specific DNA binding but also to higher off-rates, given that the increase in affinity is of a smaller magnitude than the increase in on-rates. Central to the importance of DNA binding kinetics on in vivo function, p53FG is a weaker transcriptional activator in cells than p53wt [16]. Analysis of other transcription factors suggests that our findings with p53 may be broadly relevant [28]. For example, single molecule imaging in living Escherichia coli shows that the lac repressor slides several times over its specific DNA site before actually binding to it [29]. This behavior is consistent with a low on-rate, and accordingly a low off-rate, for sequence-specific DNA binding, as predicted by our model [16] and also as demonstrated by molecular dynamics simulations of a large number of DNA binding proteins [30]. Of course, further studies will be required to establish the generality of the importance of conformational switches for regulating residency times of DNA binding proteins on DNA and for discriminating specific from non-specific sites.
Experimental Procedures
Reversal of the DNA-Binding-Induced Loop L1 Confor
This protein had 13 amino acid substitutions within the DNA binding domain that enhance thermal stability [16]. To crystallize p53FG–DNA complexes, we used a p53FG protein that included both the DNA binding and tetramerization domains (residues 94– 355) but that also had a deletion of 29 amino acids (residues 293–321) corresponding to the unstructured linker between these two domains [16]. This p53 polypeptide also had 13 amino acid substitutions within the DNA binding domain that enhance thermal stability, as well as two amino acid substitutions in the oligomerization domain that change its stoichiometry from tetramer to dimer [16]. For biophysical measurements, we used p53wt and p53FG proteins corresponding to residues 79–393 of full-length human p53. These proteins also had the 13 amino acid substitutions in the DNA binding domain that enhance thermal stability. All p53 polypeptides were expressed in E. coli. The cells were lysed in buffer consisting of 25 mM BTP (bis-tris propane) (pH 6.8), 50 mM NaCl, 5 mM DTT, and protease inhibitors, and the polypeptides were purified by cation exchange (Sepharose SP column; Pharmacia Biotech) chromatography followed by gel-filtration chromatography (Superdex 200 column; Pharmacia Biotech). Oligonucleotides p53FG was crystallized in complex with DNA containing a consensus p53 binding site (CONS26; Fig. S4). The double-stranded DNA was generated by annealing oligonucleotides with the following sequences: upper strand: [AC-GGGCA-TGTCTGGGCA-TGTCT-CAAA] and lower strand: [TTTGCCCGT-ACAGA-CCCGT-ACAGA-GT] (the specific pentanucleotides are indicated in italics). After annealing, we produce two-nucleotide sticky ends on either end, facilitating the assembly of individual molecules into a longer pseudo-continuous DNA molecule spanning the length of the crystal. For SPR, we used the same specific consensus DNA sequence as for our previous anisotropy fluorescence experiments: upper strand: [biotin-TTAGATGTAGATGATGAT-GAGCA-TGTTCGAGCA-TGTTC-ATATGATGAC] and lower strand: [GTCATCATAT-GAACA-TGCTC-GAACA-TGCTCATCATCATCTACATCTA]. We also used DNA containing the p53 response element present in the CDKN1A (p21) gene: upper strand: [biotin-TTAGATGTAGATGATGAT-GAACA-TGTCC-CAACATGTTG-ATATGATGAC] and lower strand: [GTCATCATAT-CAACA-TGTTG-GGACA-TGTTCATCATCATCTACATCTA].
Protein sample preparation and purification Surface plasmon resonance To crystallize p53FG in its apo form, we used a recombinant protein corresponding only to the human DNA binding domain (residues 94–294).
Biotinylated DNAs were resuspended at a final concentration of 200 nM in buffer consisting of
943
Reversal of the DNA-Binding-Induced Loop L1 Confor
150 mM NaCl, 5 mM DTT, and 25 mM BTP (pH 6.8). The biotinylated DNAs were then immobilized on streptavidin-coated sensor chips (Sensor chip SA, Biacore X100; GE Healthcare) according to the instructions of the manufacturer. The sensor chips contained two flow cells; DNA was immobilized on one flow cell, while the other flow cell served as reference. To determine p53 DNA binding off-rates, we delivered p53 protein solutions (118–267 nM protein concentration, as indicated) using a flow rate of 30 μl/min and a contact time with the DNA sensor of 180 s; this was followed by a dissociation time of 60 s. Between experiments, the sensor chips were injected with 1 M NaCl at 30 μl/min to remove any traces of p53 protein remaining on the chip. The SPR RUs measured over time during dissociation of p53 from the DNA sensor chip were used to calculate off-rate constants, according to the method of O'Shannessy et al. [20], which posits that the slope of the curve of ln(RUmax/RU) over time is equal to the off-rate constant (koff). To determine p53 DNA binding affinity constants, we delivered p53 protein solutions using a flow rate of 10 μl/min for the p53 protein solutions and a contact time with the DNA sensor of 240 s, followed by a dissociation time of 240 s. For each p53 protein, the affinity constants were determined using protein concentrations ranging from 80 to 200 nM. Between experiments, the sensor chips were injected with 1 M NaCl at 30 μl/min to remove any traces of p53 protein remaining on the chip. The fraction of bound DNA was determined as the ratio of the observed difference of RUs upon p53 binding (ΔRUp53) divided by the RUs corresponding to the theoretical maximum difference (ΔRUmaxθ). The latter is equal to ΔRUDNA × (MWp53/MWDNA) × K, where ΔRUDNA is the observed difference of RUs upon binding of DNA to the chip; MWDNA and MWp53 are the molecular weights of the DNA duplex and the p53 tetramer, respectively; and K represents the fraction of DNA bound on the chip that is capable of binding to p53. Experiments with increasing amounts of p53 revealed that K was equal to 0.5. Crystallization, data collection, and structure refinement The purified p53FG polypeptide corresponding only to the DNA binding domain was concentrated to about 8 mg/ml and crystallized at 4 °C under hanging-drop vapor diffusion conditions in 48-well plates (Hampton Research). The precipitant solution used for crystallization consisted of 0.2 M ammonium fluoride and 20% polyethylene glycol 3350. The p53FG–DNA complex was formed by incubating protein and DNA at a 1:1.1 molar ratio at 4 °C for 1 h in 25 mM BTP (pH 6.8), 150 mM NaCl, and 5 mM DTT buffer. The complex was then purified by gel-
filtration chromatography (Superdex 200 column; Pharmacia Biotech), concentrated to about 5 mg/ml protein concentration, and crystallized at 4 °C under hanging-drop vapor diffusion conditions in 48-well plates (Hampton Research). The precipitant solution consisted of 0.2 M potassium sodium tartrate tetrahydrate (pH 7.0) and 20% polyethylene glycol 3350. X-ray diffraction data sets were collected at the ID14-1 (p53FG apo form) and ID23-2 (p53FG–DNA complex) beamlines of the European Synchrotron Radiation Facility (Grenoble, France). Reflection data were indexed, integrated, and scaled using the programs MOSFLM and SCALE of the CCP4 software package [31]. Initial solutions were determined by molecular replacement and further refined with the programs CNS and O [32,33] using noncrystallographic symmetry restraints for the protein chains of the p53FG–DNA complex. Figures were prepared using the programs MOLSCRIPT, BOBSCRIPT, and RASTER3D [34–36]. Accession numbers Coordinates and structure factors have been deposited in the Protein Data Bank with accession numbers 4MZI and 4MZR.
Acknowledgements We thank the staff of the European Synchrotron Radiation Facility, Grenoble, France, for help in collecting the X-ray diffraction data. This work was supported by a Swiss National Science Foundation grant to T.D.H.
Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.jmb.2013.12. 020.
Received 2 September 2013; Received in revised form 17 December 2013; Accepted 18 December 2013 Available online 27 December 2013
Keywords: p53; DNA binding; crystallography; induced fit; conformational selection
944 This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial-No Derivative Works License, which permits non-commercial use, distribution, and reproduction in any medium, provided the original author and source are credited. Abbreviations used: SPR, surface plasmon resonance; RU, response unit.
References [1] Hollstein M, Sidransky D, Vogelstein B, Harris CC. p53 mutations in human cancers. Science 1991;253:49–53. [2] Hainaut P, Soussi T, Shomer B, Hollstein M, Greenblatt M, Hovig E, et al. Database of p53 gene somatic mutations in human tumors and cell lines: updated compilation and future prospects. Nucleic Acids Res 1997;25:151–7. [3] Kastan MB, Onyekwere O, Sidransky D, Vogelstein B, Craig RW. Participation of p53 protein in the cellular response to DNA damage. Cancer Res 1991;51:6304–11. [4] Itahana K, Dimri G, Campisi J. Regulation of cellular senescence by p53. Eur J Biochem 2001;268:2784–91. [5] Halazonetis TD, Gorgoulis VG, Bartek J. An oncogeneinduced DNA damage model for cancer development. Science 2008;319:1352–5. [6] Farmer G, Bargonetti J, Zhu H, Friedman P, Prywes R, Prives C. Wild-type p53 activates transcription in vitro. Nature 1992;358:83–6. [7] Vogelstein B, Lane D, Levine AJ. Surfing the p53 network. Nature 2000;408:307–10. [8] Vousden KH, Prives C. Blinded by the light: the growing complexity of p53. Cell 2009;137:413–31. [9] Joerger AC, Fersht AR. Structural biology of the tumor suppressor p53. Annu Rev Biochem 2008;77:557–82. [10] Cho Y, Gorina S, Jeffrey PD, Pavletich NP. Crystal structure of a p53 tumor suppressor-DNA complex: understanding tumorigenic mutations. Science 1994;265:346–55. [11] Kitayner M, Rozenberg H, Kessler N, Rabinovich D, Shaulov L, Haran TE, et al. Structural basis of DNA recognition by p53 tetramers. Mol Cell 2006;22:741–53. [12] Kitayner M, Rozenberg H, Rohs R, Suad O, Rabinovich D, Honig B, et al. Diversity in DNA recognition by p53 revealed by crystal structures with Hoogsteen base pairs. Nat Struct Mol Biol 2010;17:423–9. [13] Ho WC, Fitzgerald MX, Marmorstein R. Structure of the p53 core domain dimer bound to DNA. J Biol Chem 2006;281:20494–502. [14] Malecka KA, Ho WC, Marmorstein R. Crystal structure of a p53 core tetramer bound to DNA. Oncogene 2009;28:325–33. [15] Chen Y, Dey R, Chen L. Crystal structure of the p53 core domain bound to a full consensus site as a self-assembled tetramer. Structure 2010;18:246–56. [16] Petty TJ, Emamzadah S, Costantino L, Petkova I, Stavridi ES, Saven JG, et al. An induced fit mechanism regulates p53 DNA binding kinetics to confer sequence specificity. EMBO J 2011;30:2167–76. [17] Emamzadah S, Tropia L, Halazonetis TD. Crystal structure of a multidomain human p53 tetramer bound to the natural CDKN1A (p21) p53-response element. Mol Cancer Res 2011;9:1493–9.
Reversal of the DNA-Binding-Induced Loop L1 Confor
[18] Pham N, Lucumi A, Cheung N, Viadiu H. The tetramer of p53 in the absence of DNA forms a relaxed quaternary state. Biochemistry 2012;51:8053–5. [19] Koshland DE. Application of a theory of enzyme specificity to protein synthesis. Proc Natl Acad Sci U S A 1958;44:98–104. [20] O'Shannessy DJ, Brigham-Burke M, Soneson KK, Hensley P, Brooks I. Determination of rate and equilibrium binding constants for macromolecular interactions using surface plasmon resonance: use of nonlinear least squares analysis methods. Anal Biochem 1993;212:457–68. [21] Cooper MA. Optical biosensors in drug discovery. Nat Rev Drug Discov 2002;1:515–28. [22] Zhao K, Chai X, Johnston K, Clements A, Marmorstein R. Crystal structure of the mouse p53 core DNA-binding domain at 2.7 Å resolution. J Biol Chem 2001;276:12120–7. [23] Canning P, Von Delft F, Bullock AN. Structural basis for ASPP2 recognition by the tumor suppressor p73. J Mol Biol 2012;423:515–27. [24] Huyen Y, Jeffrey PD, Derry WB, Rothman JH, Pavletich NP, Stavridi ES, et al. Structural differences in the DNA binding domains of human p53 and its C. elegans ortholog Cep-1. Structure 2004;12:1237–43. [25] Ethayathulla AS, Tse PW, Monti P, Nguyen S, Inga A, Fronza G, et al. Structure of p73 DNA-binding domain tetramer modulates p73 transactivation. Proc Natl Acad Sci U S A 2012;109:6066–71. [26] Weinberg RL, Veprintsev DB, Fersht AR. Cooperative binding of tetrameric p53 to DNA. J Mol Biol 2004;341:1145–59. [27] Hammes GG, Chang YC, Oas TG. Conformational selection or induced fit: a flux description of reaction mechanism. Proc Natl Acad Sci U S A 2009;106:13737–41. [28] Tubbs JL, Tainer JA. p53 conformational switching for selectivity may reveal a general solution for specific DNA binding. EMBO J 2011;30:2099–100. [29] Hammar P, Leroy P, Mahmutovic A, Marklund EG, Berg OG, Elf J. The lac repressor displays facilitated diffusion in living cells. Science 2012;336:1595–8. [30] Marcovitz A, Levy Y. Frustration in protein–DNA binding influences conformational switching and target search kinetics. Proc Natl Acad Sci U S A 2011;108:17957–62. [31] Collaborative Computational Project, Number 4. The CCP4 suite: programs for protein crystallography. Acta Crystallogr Sect D Biol Crystallogr 1994;50:760–3. [32] Brünger AT, Adams PD, Clore GM, DeLano WL, Gros P, Grosse-Kunstleve RW, et al. Crystallography & NMR system: a new software suite for macromolecular structure determination. Acta Crystallogr Sect D Biol Crystallogr 1998;54:905–21. [33] Jones TA, Zou JY, Cowan SW, Kjeldgaard M. Improved methods for building protein models in electron density maps and the location of errors in these models. Acta Crystallogr Sect A Found Crystallogr 1991;47:110–9. [34] Kraulis PJ. MOLSCRIPT: a program to produce both detailed and schematic plots of protein structures. J Appl Crystallogr 1991;24:946–50. [35] Esnouf RM. An extensively modified version of MolScript that includes greatly enhanced coloring capabilities. J Mol Graphics Modell 1997;15:132–4. [36] Merritt EA, Bacon DJ. Raster3D: photorealistic molecular graphics. Methods Enzymol 1997;277:505–24.
Reversal of the DNA-Binding-Induced Loop L1 Conformational Switch in an Engineered Human p53 Protein Soheila Emamzadah 1 , Laurence Tropia 1 , Ilena Vincenti 2 , Benoît Falquet 2 and Thanos D. Halazonetis 1 1 - Department of Molecular Biology, University of Geneva, 1205 Geneva, Switzerland 2 - Department of Biochemistry, University of Geneva, 1205 Geneva, Switzerland
Correspondence to Thanos D. Halazonetis: [email protected] http://dx.doi.org/10.1016/j.jmb.2013.12.020 Edited by C. Kalodimos
Abstract The gene encoding the p53 tumor suppressor protein, a sequence-specific DNA binding transcription factor, is the most frequently mutated gene in human cancer. Crystal structures of homo-oligomerizing p53 polypeptides with specific DNA suggest that DNA binding is associated with a conformational switch. Specifically, in the absence of DNA, loop L1 of the p53 DNA binding domain adopts an extended conformation, whereas two p53 subunits switch to a recessed loop L1 conformation when bound to DNA as a tetramer. We previously designed a p53 protein, p53FG, with amino substitutions S121F and V122G targeting loop L1. These two substitutions enhanced the affinity of p53 for specific DNA yet, counterintuitively, decreased the residency time of p53 on DNA. Here, we confirmed these DNA binding properties of p53FG using a different method. We also determined by crystallography the structure of p53FG in its free state and bound to DNA as a tetramer. In the free state, loop L1 adopted a recessed conformation, whereas upon DNA binding, two subunits switched to the extended loop L1 conformation, resulting in a final structure that was very similar to that of wild-type p53 bound to DNA. Thus, altering the apo structure of p53 changed its DNA binding properties, even though the DNA-bound structure was not altered. © 2014 The Authors. Published by Elsevier Ltd. All rights reserved.
Introduction The gene encoding the tumor suppressor p53 protein is the most frequently mutated gene in human cancer [1,2]. The p53 protein is activated in response to DNA damage, and once activated, it enhances transcription of genes that induce cell cycle arrest, apoptosis, or senescence [3,4]. Since oncogenes induce DNA damage, it is not surprising that p53 is the most frequently mutated gene in human cancer [5]. The p53 protein contains two independently folding domains: a sequence-specific DNA binding domain at its center (also known as the core domain) and a homo-tetramerization domain toward the C-terminus [6–8]. These two domains are flanked by three flexible, unstructured regions: an N-terminal transactivation region, a linker region between the DNA binding and tetramerization domains, and a C-terminal basic region
[9]. Most of the p53 mutations in human cancer target the sequence-specific DNA binding domain [1,2]. Several three-dimensional structures of p53 have been determined [10–18]. Most of these encompass only the core domain, either by itself or in complex with DNA [10–15]. Recently, we solved structures of a p53 polypeptide containing both the DNA binding and homo-oligomerization domains in its free form and bound to specific DNA sites [16,17]. Consistent with wild-type p53 (p53wt) binding DNA as tetramer, we observed four p53 subunits bound to the specific DNA site. The specific DNA site encompasses four tandem pentanucleotide repeats, and each repeat is recognized by one p53 DNA binding domain. Unlike the structures containing only the core domain of p53 bound to DNA, we observed a conformational switch within loop L1 upon sequence-specific DNA binding. The conformational change of loop L1 involves only two of the four p53 subunits bound to DNA, those
0022-2836/$ - see front matter © 2014 The Authors. Published by Elsevier Ltd. All rights reserved. J. Mol. Biol. (2014) 426, 936–944
937
Reversal of the DNA-Binding-Induced Loop L1 Confor
a
b
DNA: CDKN1A
v/[p53]
DNA: CONS
v/[p53] 0.015
0.020 0.015
p53FG (KD) 6.93+2.48 nM
0.012
p53FG (KD) 11.0+1.0 nM
0.009
0.010 0.006
p53wt (KD) 14.5+1.7 nM
0.005
p53wt (KD) 9.21+0.23 nM
0.003 0 0.8
0 0.7
0.8
0.9
1.0 v
c
0.85
0.95
0.9
1.0 v
d DNA: CDKN1A p53FG (koff) 0.16+0.01 s−1 0.16+0.01 s−1 0.16+0.02 s−1
ln(RUmax/RU) 1.0 0.8
1.5
p53FG (koff) 0.33+0.02 s−1 0.35+0.04 s−1 0.34+0.04 s−1
1.2
p53wt (koff) 0.077+0.003 s−1
0.6
DNA: CONS
ln(RUmax/RU)
0.9
0.4
0.6
0.2
0.3
0
p53wt (koff) 0.105+0.006 s−1 0.121+0.005 s−1 0.124+0.010 s−1
0 0
2
4
6
8
10
12 t(s)
0
2
4
6
8
10 t(s)
Fig. 1. Amino acid substitutions within loop L1 decrease the half-life of p53–DNA complexes despite increasing the affinity for specific DNA. (a and b) Scatchard plots for binding of human p53wt and p53FG (residues 79–393) to the native p53 response element present in the CDKN1A gene (a) or to a consensus p53 response element (b). The KD values (mean ± 1 SD) shown correspond to − 1/slope of the fraction of bound DNA divided by the concentration of free p53 (v/[p53]) as a function of the fraction of bound DNA (v). (c and d) Plot of the natural logarithm of RUmax divided by RU [ln(RUmax/RU)] over time (t) in seconds (s) for binding of p53 and p53FG to CDKN1A (c) and consensus (d) p53 response elements. The koff values shown (mean ± 1 SD) correspond to the slope of the plot. The kinetic experiments were performed with p53 protein concentrations of 118 nM (blue symbols), 178 nM (red symbols), and 267 nM (green symbols).
subunits recognizing the most 5-prime and most 3-prime of the four tandem pentanucleotide repeats [16]. We refer to these two repeats as the outer repeats. On the basis of the conformational switch in loop L1, we proposed that p53 recognizes sequencespecific DNA via a two-step, induced-fit mechanism [16,17,19]. The first step involves an initial rigid-body interaction between p53 and DNA, followed (in the second step) by conformational changes that allow a better fit between p53 and the DNA. To test this model, we analyzed p53 mutants containing amino acid substitutions within loop L1 [16]. The design of these mutants was guided by the free and DNAbound conformations of p53. The most interesting mutant, hereafter referred to as p53FG, contained two substitutions within loop L1: Ser121 with Phe (S121F) and Val122 with Gly (V122G). These two substitutions were predicted to favor the recessed DNA-bound conformation of loop L1 and also to impart conformational flexibility to the loop, thereby
decreasing the kinetic barrier between the free and DNA-bound conformations. Fluorescence anisotropy experiments performed with p53wt and p53FG revealed that p53FG bound to DNA with higher affinity than p53wt [16]. One would have expected that the higher affinity for DNA would have translated into longer half-lives for the p53–DNA complexes. However, the half-lives of the p53FG–DNA complexes were shorter than those of the p53wt–DNA complexes [16]. This unexpected observation prompted us to propose that the conformational switch provides a kinetic barrier to p53 DNA binding and, thereby, regulates the residency time of p53 on DNA. The design of p53FG was guided by the structure of p53wt bound to DNA [16]. While we had predicted that the S121F and V122G substitutions would facilitate the recessed loop L1 conformation, this was never demonstrated experimentally. Further, how Phe121 would be accommodated in the p53– DNA structure was unclear, since we had not solved
938
Reversal of the DNA-Binding-Induced Loop L1 Confor
a
b
p53FG/L1 p53wt
p53FG/L1 p53wt Cep−1 S238 S121 F121
F121
L1
S238
S121
Fig. 2. Three-dimensional structure of the p53FG DNA binding domain in its apo form. (a) Overall structure of p53FG (colored green; except for loop L1, colored blue) superimposed on the structure of the DNA binding domain of p53wt (gray). (b) Superimposition of the loop L1 structures and of the beta-strand C-terminal to loop L1 of p53FG in its apo form [colored as in (a)], p53wt (gray), and Cep-1 (yellow). The side chains of Phe121 (p53FG), Ser121 (p53wt), and Ser238 (Cep-1; corresponding to Ser121 of p53wt) are indicated. The orientation of the top view is the same as in (a); the bottom view is obtained by a 90° rotation along the indicated axis. The r.m.s. difference between the positions of the alpha-carbon atoms of loops L1 of p53wt and p53FG (residues 113–123), in their apo form, is 3.5 Å.
the structure of p53FG in complex with DNA. Here, we address these questions by determining the three-dimensional structures of p53FG in its apo form and in complex with specific DNA. In addition, we use surface plasmon resonance (SPR), as a different biophysical method than the one used in our previous study [16], to characterize the DNA binding properties of p53wt and p53FG. The new results further support our model that conformational changes in loop L1 allow p53 DNA binding affinities and kinetics to be regulated independently of each other. Thus, p53 could serve as a paradigm toward understanding how transcription factors occupy their specific sites in vivo. Specifically, large differences in half-lives of protein–DNA complexes, made possible by conformational changes, rather than large differences in affinities may allow discrimination of specific from non-specific DNA sites.
Results Amino acid substitutions within loop L1 decrease the half-life of p53–DNA complexes despite increasing DNA binding affinity We previously used fluorescence anisotropy to measure binding affinities and off-rates of various p53 proteins for sequence-specific DNA [16]. We found that p53FG bound specific DNA with higher affinity than p53wt. Even though the higher affinity of p53FG for DNA should translate to longer residency times of p53 on DNA, quite unexpectedly, the half-life of the p53FG–DNA complexes was shorter than the half-life of the p53wt–DNA complexes. To determine whether these results could be repro-
duced with a different method, we employed SPR, which also allows affinity and kinetic binding constants to be determined [20,21]. We examined binding of p53 to two different DNA sequences: one containing the natural p53 response element in the CDKN1A (p21) gene and one containing a consensus p53 binding site (Fig. 1). Oligonucleotides containing these two DNA sequences were synthesized with a biotin tag to allow them to be immobilized on streptavidin-coated SPR sensor chips. Binding measurements were performed using purified, recombinant p53 proteins corresponding to residues 79–393 of the full-length human p53 protein. These proteins lacked only the transactivation domain, which is not thought to be involved in DNA binding. One protein had a wild-type loop L1 sequence (p53wt), while the other (p53FG) contained two amino acid substitutions, S121F and V122G, in loop L1. Both p53 proteins also contained amino acid substitutions that enhance the thermal stability and solubility of the DNA binding domain. These substitutions do not affect p53 function but render p53 amenable to biophysical analysis [16]. The first series of SPR experiments was used to determine DNA binding affinities (expressed by the dissociation constant KD). The results were visualized using Scatchard plots showing the fraction of bound DNA divided by the concentration of free p53 (v/[p53]) as a function of the fraction of bound DNA (v). Wild-type p53 bound the CDKN1A response element with a dissociation constant of 14.5 ± 1.7 nM (mean ± 1 SD), whereas p53FG had a dissociation constant of 11.0 ± 1.0 nM (P b 0.05 for the difference in KD values; Fig. 1a). For the consensus binding site, the calculated dissociation constants were 9.21 ± 0.23 and 6.93 ± 2.48 nM for p53wt and p53FG, respectively
939
Reversal of the DNA-Binding-Induced Loop L1 Confor
a
b
3’
5’
A
B p53FG (B) p53FG (A) p53FG (apo) p53wt (apo)
C D
5’
3’
c p53wt
p53FG
Fig. 3. Three-dimensional structure of a multidomain p53FG oligomer bound to DNA. (a) Overall structure of two p53FG dimers bound to DNA. The p53 subunits are labeled A–D. (b) Superimposition of loops L1 of subunits A (green) and B (yellow) of p53FG bound to DNA to loops L1 of p53wt (gray) and p53FG (blue) in their apo forms. The r.m.s. difference between the positions of the alpha-carbon atoms of loop L1 of p53FG bound to the outer DNA repeat (subunit A) relative to loops L1 of p53wt and p53FG in their apo forms is 3.9 and 3.4 Å, respectively. The r.m.s. difference between the positions of the alpha-carbon atoms of loop L1 of p53FG bound to the inner DNA repeat (subunit B) relative to loops L1 of p53wt and p53FG in their apo forms is 0.8 and 3.2 Å, respectively. (c) Cartoon depicting the conformational switching of loops L1 of p53wt and p53FG upon DNA binding.
(Fig. 1b). The difference in affinities of wild-type p53 and p53FG for specific DNA is not as high as the one we previously observed by fluorescence anisotropy (7.75 ± 0.52 versus 1.96 ± 0.20 nM for p53wt and p53FG, respectively [16]); however, the SPR measurements still showed a higher DNA binding affinity for p53FG than for p53wt. In order to calculate off-rate constants, we relied on the SPR response units (RUs) measured during dissociation of p53 from specific DNA over time, using the method of O'Shannessy et al. [20], which posits that the slope of the plot of ln(RUmax/RU) over time is equal to the off-rate constant (koff). The koff values for the complexes of p53wt and p53FG with the
CDKN1A response element were 0.077 ± 0.003 and 0.16 ± 0.01 s −1, respectively (P b 0.005 for the difference in koff values; Fig. 1c). For the consensus binding site, the off-rates were 0.121 ± 0.005 and 0.35 ± 0.04 s − 1 for p53wt and p53FG, respectively (P b 0.005 for the difference in koff values; Fig. 1d). These latter values were consistent with those previously obtained by fluorescence anisotropy (0.068 ± 0.008 versus 0.34 ± 0.03 s −1 for p53wt and p53FG, respectively [16]). Thus, the off-rate values obtained by two different techniques reveal that the half-lives of the p53FG–DNA complexes were shorter than the halflives of the p53wt–DNA complexes despite p53FG having higher affinity for specific DNA than p53wt.
940
Reversal of the DNA-Binding-Induced Loop L1 Confor
a
b H2 H2
G2a
G2b
R280
R280 A276
A276
K120
L1
G122
C277 F121
C3a
T3b R283 G4’b
F121
C277
G5’a
A5’b
B
K120
A
G122
L1
c p53FG (B) p53FG (A) p53wt (B) p53wt (A) p53cr
Fig. 4. Sequence-specific contacts of p53FG with DNA. (a) Interface of subunit B of p53FG with DNA. (b) Interface of subunit A of p53FG with DNA. The subunits are labeled and are shown in the same orientation as subunit B in Fig. 3a. (c) Superimposition of loops L1 of subunits A (green) and B (yellow) of p53FG bound to DNA to loops L1 of subunits A (blue) and B (gray) of p53wt bound to DNA. The conformation of loop L1 of mouse p53 cross-linked (p53cr) to DNA (purple [14]) is also shown. The r.m.s. difference between the positions of the alpha-carbon atoms of loop L1 of p53FG and p53wt bound to the outer DNA repeats (subunits A) is 2.1 Å; the r.m.s. difference between the positions of the alpha-carbon atoms of loop L1 of p53FG and p53wt bound to the inner DNA repeats (subunits B) is 1.6 Å.
Structure of p53FG in its apo (free) form The structure of the DNA binding domain of p53FG encompassing the two substitutions within loop L1 was determined at a resolution of 1.25 Å (Table S1). The overall structure was similar to that of wild-type p53, except for the conformation of loop L1, which in p53FG, remarkably, adopted a recessed conformation (Fig. 2a). So far, all structures of p53 and of its family member p73 have been shown to adopt an extended loop L1 conformation in their apo form [10,22,23] with only one exception, that of Caenorhabditis elegans p53 [24]. Superimposition of the L1 loops of p53wt, p53FG, and C. elegans p53 (Cep-1) revealed that loop L1 of p53FG is as recessed as that of C. elegans p53 (Fig. 2b). Formally, the recessed conformation of loop L1 in p53FG could be attributed to crystal packing contacts, since the side chain of Phe121 is within 3.5–4.8 Å of the side chains of Arg174, Pro177, and Glu180 of a symmetry-related p53 molecule in the crystal lattice (Fig. S1). However,
most of the crystal packing contacts between these two subunits are not mediated by loop L1, but rather by helix H2 (Fig. S1). Further, changes in the conformation of loop L1 as a result of crystal packing contacts have not been observed in any of the other p53 crystal structures determined to date. Thus, we argue that the recessed conformation of loop L1 in the p53FG crystal structure is representative of its apo conformation in solution. Structure of a p53FG oligomer bound to sequencespecific DNA The structure of a p53FG polypeptide encompassing both the core and homo-oligomerization domains in complex with a consensus-specific DNA site (CONS26) was determined at a resolution of 2.9 Å (Fig. 3a and Table S1). The overall structure of this homo-oligomerizing p53FG polypeptide bound to DNA was very similar to the previously determined structures of homo-oligomerizing p53wt in complex with DNA (Fig. S2). Specifically, the two subunits of
Reversal of the DNA-Binding-Induced Loop L1 Confor
the p53FG tetramer recognizing the central two pentanucleotide repeats of the p53 DNA site (inner repeats) adopted the extended loop L1 conformation, whereas the two subunits recognizing the outer repeats adopted the recessed loop L1 conformation (Fig. 3b). Given that, in the apo form, loop L1 of p53FG is recessed, the structure reveals a conformational switch upon DNA binding. However, whereas in p53wt, the conformational switch involves the loops of the subunits binding the outer pentanucleotide repeats (from an extended to a recessed conformation); in p53FG, the switch affected the loops of the subunits binding the inner repeats (from a recessed to an extended conformation) (Fig. 3c). In regard to the subunits binding the inner pentanucleotide repeats, we observed few differences in the contacts of p53FG with DNA, relative to the contacts of p53wt with DNA [16,17]. Specifically, whereas in p53wt, the side chain of Lys120 contacts atoms N7 and O6 of the guanine at position G4′b [17]; in p53FG, the side chain of Lys120, due to steric interference with the side chain of Phe121, was pushed almost outside the major groove of the DNA and contacted the DNA sugar-phosphate backbone (Fig. 4a and Fig. S3). Further, the side chain of Ser121 of p53wt interacts with the sugar-phosphate backbone of the cytosine at position C5c [17], but in p53FG, the side chain of Phe121 was close to oxygen O6 of the guanine at position G4′b and did not interact with the DNA sugar-phosphate backbone (Fig. 4a). In regard to the subunits binding the outer pentanucleotide repeats, the only difference in the contacts of p53FG with DNA, compared to the contacts of p53wt with DNA, involved residue 121. In the structure of p53wt bound to the native site present in the CDKN1A gene, Ser121 contacts the DNA backbone [17]. The corresponding residue, Phe121, in the p53FG–DNA structure was oriented toward the interior of the protein. In this conformation, due to its large hydrophobic side chain, Phe121 could help stabilize the recessed conformation of loop L1 (Fig. 4b). The overall similar contacts of p53wt and p53FG to DNA are further reflected in the similar DNAbound conformations of loops L1 of these two proteins. Thus, the subunits of p53wt and p53FG recognizing the inner repeats have similar extended loop L1 conformations, while the subunits recognizing the outer repeats have similar recessed loop L1 conformations (Fig. 4c).
Discussion We previously described the structures of p53 tetramers bound to a consensus-specific DNA site (CONS26) or to the native CDKN1A p53 response
941 element [16,17]. These structures showed a conformational switch involving loops L1 of the DNA binding domains of p53. The conformational switch involved only two of the four subunits, those subunits bound to the outer pentanucleotide repeats of the p53 response element. In these two subunits, the loop L1 conformation switched from extended (in the apo form) to recessed (in the DNA-bound form). In our previous study, we predicted that two substitutions within loop L1, S121F and V122G, would facilitate the conformational switch from the extended to the recessed conformation on the rationale that the side chain of Phe121 could potentially form hydrophobic interactions stabilizing the recessed conformation of loop L1 and that Gly122 lacks a hydrophobic side chain to stabilize the extended loop L1 conformation [16]. The structure of p53FG in its apo form, determined here, supports these predictions since, in the absence of DNA, loop L1 of p53FG adopted a recessed conformation. Upon DNA binding, p53FG underwent a conformational switch, such that, in two subunits, loop L1 adopted an extended conformation. Thus, even though the structures of p53wt and p53FG differ in their apo form, upon binding to specific DNA, these two p53 proteins adopt essentially identical structures. Cep-1, the p53 homolog in C. elegans, adopts a recessed loop L1 conformation in the apo form, reminiscent of that observed in the apo form of p53FG [24]. We envision that Cep-1, like p53FG, will also switch conformation upon DNA binding, thereby explaining the paradox as to how p53wt and Cep-1 recognize identical DNA sequences despite having different loop L1 structures in their apo forms [24]. Most previously determined structures of p53 DNA binding domains in complex with DNA have not revealed any conformational switch in p53 upon DNA binding [10–15]. The only exception was the structure of a p53 tetramer–DNA complex, in which p53 had been cross-linked to the DNA [14]. In this case, all four p53 subunits adopted a recessed loop L1 conformation. More recently, subtle conformational changes have been observed in loop L1 of p73 bound to DNA [25]. However, the extent of these conformational changes is more limited than the changes described here. We speculate that the previous studies have failed to observe major loop L1 conformational changes upon DNA binding because they examined polypeptides corresponding only to the DNA binding domain of p53. The affinity for specific DNA of polypeptides containing only the core domain of p53 is orders of magnitudes lower than the affinity of polypeptides containing both the DNA binding and oligomerization domains [26]. Thus, only the latter polypeptides may recapitulate physiological DNA binding. Based on fluorescence anisotropy measurements, we had previously argued that the conformational
942 switch in loop L1 of p53 allows DNA binding affinities and off-rates to be regulated independently of each other [16]. Specifically, we had observed that the p53FG mutant has higher affinity for specific DNA than p53wt; however, its residency time on DNA is shorter than that of p53wt. We reproduced here these results employing SPR. Thus, two biophysical methods now support our premise that the conformational switch of loop L1 allows more precise control of the DNA binding properties of p53. The DNA binding properties of p53FG can be explained from the perspectives of either the induced-fit model or the conformational selection model [27]. In both cases, we speculate that loop L1 is conformationally more flexible in p53FG than in p53wt, a hypothesis supported by the crystallographic B-factor values (Fig. S4). According to the induced-fit model, higher loop L1 flexibility would result in less energy being required for loop L1 in p53FG to adopt the DNA-bound conformation. In turn, this would lead not only to higher affinity but also to higher on-rates and higher off-rates, since the off-rate is equal to the on-rate multiplied by the dissociation constant [16]. From the perspective of the conformational selection model, the conformational flexibility would result in a higher fraction of molecules adopting the DNA-bound conformation, leading not only to higher on-rates for specific DNA binding but also to higher off-rates, given that the increase in affinity is of a smaller magnitude than the increase in on-rates. Central to the importance of DNA binding kinetics on in vivo function, p53FG is a weaker transcriptional activator in cells than p53wt [16]. Analysis of other transcription factors suggests that our findings with p53 may be broadly relevant [28]. For example, single molecule imaging in living Escherichia coli shows that the lac repressor slides several times over its specific DNA site before actually binding to it [29]. This behavior is consistent with a low on-rate, and accordingly a low off-rate, for sequence-specific DNA binding, as predicted by our model [16] and also as demonstrated by molecular dynamics simulations of a large number of DNA binding proteins [30]. Of course, further studies will be required to establish the generality of the importance of conformational switches for regulating residency times of DNA binding proteins on DNA and for discriminating specific from non-specific sites.
Experimental Procedures
Reversal of the DNA-Binding-Induced Loop L1 Confor
This protein had 13 amino acid substitutions within the DNA binding domain that enhance thermal stability [16]. To crystallize p53FG–DNA complexes, we used a p53FG protein that included both the DNA binding and tetramerization domains (residues 94– 355) but that also had a deletion of 29 amino acids (residues 293–321) corresponding to the unstructured linker between these two domains [16]. This p53 polypeptide also had 13 amino acid substitutions within the DNA binding domain that enhance thermal stability, as well as two amino acid substitutions in the oligomerization domain that change its stoichiometry from tetramer to dimer [16]. For biophysical measurements, we used p53wt and p53FG proteins corresponding to residues 79–393 of full-length human p53. These proteins also had the 13 amino acid substitutions in the DNA binding domain that enhance thermal stability. All p53 polypeptides were expressed in E. coli. The cells were lysed in buffer consisting of 25 mM BTP (bis-tris propane) (pH 6.8), 50 mM NaCl, 5 mM DTT, and protease inhibitors, and the polypeptides were purified by cation exchange (Sepharose SP column; Pharmacia Biotech) chromatography followed by gel-filtration chromatography (Superdex 200 column; Pharmacia Biotech). Oligonucleotides p53FG was crystallized in complex with DNA containing a consensus p53 binding site (CONS26; Fig. S4). The double-stranded DNA was generated by annealing oligonucleotides with the following sequences: upper strand: [AC-GGGCA-TGTCTGGGCA-TGTCT-CAAA] and lower strand: [TTTGCCCGT-ACAGA-CCCGT-ACAGA-GT] (the specific pentanucleotides are indicated in italics). After annealing, we produce two-nucleotide sticky ends on either end, facilitating the assembly of individual molecules into a longer pseudo-continuous DNA molecule spanning the length of the crystal. For SPR, we used the same specific consensus DNA sequence as for our previous anisotropy fluorescence experiments: upper strand: [biotin-TTAGATGTAGATGATGAT-GAGCA-TGTTCGAGCA-TGTTC-ATATGATGAC] and lower strand: [GTCATCATAT-GAACA-TGCTC-GAACA-TGCTCATCATCATCTACATCTA]. We also used DNA containing the p53 response element present in the CDKN1A (p21) gene: upper strand: [biotin-TTAGATGTAGATGATGAT-GAACA-TGTCC-CAACATGTTG-ATATGATGAC] and lower strand: [GTCATCATAT-CAACA-TGTTG-GGACA-TGTTCATCATCATCTACATCTA].
Protein sample preparation and purification Surface plasmon resonance To crystallize p53FG in its apo form, we used a recombinant protein corresponding only to the human DNA binding domain (residues 94–294).
Biotinylated DNAs were resuspended at a final concentration of 200 nM in buffer consisting of
943
Reversal of the DNA-Binding-Induced Loop L1 Confor
150 mM NaCl, 5 mM DTT, and 25 mM BTP (pH 6.8). The biotinylated DNAs were then immobilized on streptavidin-coated sensor chips (Sensor chip SA, Biacore X100; GE Healthcare) according to the instructions of the manufacturer. The sensor chips contained two flow cells; DNA was immobilized on one flow cell, while the other flow cell served as reference. To determine p53 DNA binding off-rates, we delivered p53 protein solutions (118–267 nM protein concentration, as indicated) using a flow rate of 30 μl/min and a contact time with the DNA sensor of 180 s; this was followed by a dissociation time of 60 s. Between experiments, the sensor chips were injected with 1 M NaCl at 30 μl/min to remove any traces of p53 protein remaining on the chip. The SPR RUs measured over time during dissociation of p53 from the DNA sensor chip were used to calculate off-rate constants, according to the method of O'Shannessy et al. [20], which posits that the slope of the curve of ln(RUmax/RU) over time is equal to the off-rate constant (koff). To determine p53 DNA binding affinity constants, we delivered p53 protein solutions using a flow rate of 10 μl/min for the p53 protein solutions and a contact time with the DNA sensor of 240 s, followed by a dissociation time of 240 s. For each p53 protein, the affinity constants were determined using protein concentrations ranging from 80 to 200 nM. Between experiments, the sensor chips were injected with 1 M NaCl at 30 μl/min to remove any traces of p53 protein remaining on the chip. The fraction of bound DNA was determined as the ratio of the observed difference of RUs upon p53 binding (ΔRUp53) divided by the RUs corresponding to the theoretical maximum difference (ΔRUmaxθ). The latter is equal to ΔRUDNA × (MWp53/MWDNA) × K, where ΔRUDNA is the observed difference of RUs upon binding of DNA to the chip; MWDNA and MWp53 are the molecular weights of the DNA duplex and the p53 tetramer, respectively; and K represents the fraction of DNA bound on the chip that is capable of binding to p53. Experiments with increasing amounts of p53 revealed that K was equal to 0.5. Crystallization, data collection, and structure refinement The purified p53FG polypeptide corresponding only to the DNA binding domain was concentrated to about 8 mg/ml and crystallized at 4 °C under hanging-drop vapor diffusion conditions in 48-well plates (Hampton Research). The precipitant solution used for crystallization consisted of 0.2 M ammonium fluoride and 20% polyethylene glycol 3350. The p53FG–DNA complex was formed by incubating protein and DNA at a 1:1.1 molar ratio at 4 °C for 1 h in 25 mM BTP (pH 6.8), 150 mM NaCl, and 5 mM DTT buffer. The complex was then purified by gel-
filtration chromatography (Superdex 200 column; Pharmacia Biotech), concentrated to about 5 mg/ml protein concentration, and crystallized at 4 °C under hanging-drop vapor diffusion conditions in 48-well plates (Hampton Research). The precipitant solution consisted of 0.2 M potassium sodium tartrate tetrahydrate (pH 7.0) and 20% polyethylene glycol 3350. X-ray diffraction data sets were collected at the ID14-1 (p53FG apo form) and ID23-2 (p53FG–DNA complex) beamlines of the European Synchrotron Radiation Facility (Grenoble, France). Reflection data were indexed, integrated, and scaled using the programs MOSFLM and SCALE of the CCP4 software package [31]. Initial solutions were determined by molecular replacement and further refined with the programs CNS and O [32,33] using noncrystallographic symmetry restraints for the protein chains of the p53FG–DNA complex. Figures were prepared using the programs MOLSCRIPT, BOBSCRIPT, and RASTER3D [34–36]. Accession numbers Coordinates and structure factors have been deposited in the Protein Data Bank with accession numbers 4MZI and 4MZR.
Acknowledgements We thank the staff of the European Synchrotron Radiation Facility, Grenoble, France, for help in collecting the X-ray diffraction data. This work was supported by a Swiss National Science Foundation grant to T.D.H.
Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.jmb.2013.12. 020.
Received 2 September 2013; Received in revised form 17 December 2013; Accepted 18 December 2013 Available online 27 December 2013
Keywords: p53; DNA binding; crystallography; induced fit; conformational selection
944 This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial-No Derivative Works License, which permits non-commercial use, distribution, and reproduction in any medium, provided the original author and source are credited. Abbreviations used: SPR, surface plasmon resonance; RU, response unit.
References [1] Hollstein M, Sidransky D, Vogelstein B, Harris CC. p53 mutations in human cancers. Science 1991;253:49–53. [2] Hainaut P, Soussi T, Shomer B, Hollstein M, Greenblatt M, Hovig E, et al. Database of p53 gene somatic mutations in human tumors and cell lines: updated compilation and future prospects. Nucleic Acids Res 1997;25:151–7. [3] Kastan MB, Onyekwere O, Sidransky D, Vogelstein B, Craig RW. Participation of p53 protein in the cellular response to DNA damage. Cancer Res 1991;51:6304–11. [4] Itahana K, Dimri G, Campisi J. Regulation of cellular senescence by p53. Eur J Biochem 2001;268:2784–91. [5] Halazonetis TD, Gorgoulis VG, Bartek J. An oncogeneinduced DNA damage model for cancer development. Science 2008;319:1352–5. [6] Farmer G, Bargonetti J, Zhu H, Friedman P, Prywes R, Prives C. Wild-type p53 activates transcription in vitro. Nature 1992;358:83–6. [7] Vogelstein B, Lane D, Levine AJ. Surfing the p53 network. Nature 2000;408:307–10. [8] Vousden KH, Prives C. Blinded by the light: the growing complexity of p53. Cell 2009;137:413–31. [9] Joerger AC, Fersht AR. Structural biology of the tumor suppressor p53. Annu Rev Biochem 2008;77:557–82. [10] Cho Y, Gorina S, Jeffrey PD, Pavletich NP. Crystal structure of a p53 tumor suppressor-DNA complex: understanding tumorigenic mutations. Science 1994;265:346–55. [11] Kitayner M, Rozenberg H, Kessler N, Rabinovich D, Shaulov L, Haran TE, et al. Structural basis of DNA recognition by p53 tetramers. Mol Cell 2006;22:741–53. [12] Kitayner M, Rozenberg H, Rohs R, Suad O, Rabinovich D, Honig B, et al. Diversity in DNA recognition by p53 revealed by crystal structures with Hoogsteen base pairs. Nat Struct Mol Biol 2010;17:423–9. [13] Ho WC, Fitzgerald MX, Marmorstein R. Structure of the p53 core domain dimer bound to DNA. J Biol Chem 2006;281:20494–502. [14] Malecka KA, Ho WC, Marmorstein R. Crystal structure of a p53 core tetramer bound to DNA. Oncogene 2009;28:325–33. [15] Chen Y, Dey R, Chen L. Crystal structure of the p53 core domain bound to a full consensus site as a self-assembled tetramer. Structure 2010;18:246–56. [16] Petty TJ, Emamzadah S, Costantino L, Petkova I, Stavridi ES, Saven JG, et al. An induced fit mechanism regulates p53 DNA binding kinetics to confer sequence specificity. EMBO J 2011;30:2167–76. [17] Emamzadah S, Tropia L, Halazonetis TD. Crystal structure of a multidomain human p53 tetramer bound to the natural CDKN1A (p21) p53-response element. Mol Cancer Res 2011;9:1493–9.
Reversal of the DNA-Binding-Induced Loop L1 Confor
[18] Pham N, Lucumi A, Cheung N, Viadiu H. The tetramer of p53 in the absence of DNA forms a relaxed quaternary state. Biochemistry 2012;51:8053–5. [19] Koshland DE. Application of a theory of enzyme specificity to protein synthesis. Proc Natl Acad Sci U S A 1958;44:98–104. [20] O'Shannessy DJ, Brigham-Burke M, Soneson KK, Hensley P, Brooks I. Determination of rate and equilibrium binding constants for macromolecular interactions using surface plasmon resonance: use of nonlinear least squares analysis methods. Anal Biochem 1993;212:457–68. [21] Cooper MA. Optical biosensors in drug discovery. Nat Rev Drug Discov 2002;1:515–28. [22] Zhao K, Chai X, Johnston K, Clements A, Marmorstein R. Crystal structure of the mouse p53 core DNA-binding domain at 2.7 Å resolution. J Biol Chem 2001;276:12120–7. [23] Canning P, Von Delft F, Bullock AN. Structural basis for ASPP2 recognition by the tumor suppressor p73. J Mol Biol 2012;423:515–27. [24] Huyen Y, Jeffrey PD, Derry WB, Rothman JH, Pavletich NP, Stavridi ES, et al. Structural differences in the DNA binding domains of human p53 and its C. elegans ortholog Cep-1. Structure 2004;12:1237–43. [25] Ethayathulla AS, Tse PW, Monti P, Nguyen S, Inga A, Fronza G, et al. Structure of p73 DNA-binding domain tetramer modulates p73 transactivation. Proc Natl Acad Sci U S A 2012;109:6066–71. [26] Weinberg RL, Veprintsev DB, Fersht AR. Cooperative binding of tetrameric p53 to DNA. J Mol Biol 2004;341:1145–59. [27] Hammes GG, Chang YC, Oas TG. Conformational selection or induced fit: a flux description of reaction mechanism. Proc Natl Acad Sci U S A 2009;106:13737–41. [28] Tubbs JL, Tainer JA. p53 conformational switching for selectivity may reveal a general solution for specific DNA binding. EMBO J 2011;30:2099–100. [29] Hammar P, Leroy P, Mahmutovic A, Marklund EG, Berg OG, Elf J. The lac repressor displays facilitated diffusion in living cells. Science 2012;336:1595–8. [30] Marcovitz A, Levy Y. Frustration in protein–DNA binding influences conformational switching and target search kinetics. Proc Natl Acad Sci U S A 2011;108:17957–62. [31] Collaborative Computational Project, Number 4. The CCP4 suite: programs for protein crystallography. Acta Crystallogr Sect D Biol Crystallogr 1994;50:760–3. [32] Brünger AT, Adams PD, Clore GM, DeLano WL, Gros P, Grosse-Kunstleve RW, et al. Crystallography & NMR system: a new software suite for macromolecular structure determination. Acta Crystallogr Sect D Biol Crystallogr 1998;54:905–21. [33] Jones TA, Zou JY, Cowan SW, Kjeldgaard M. Improved methods for building protein models in electron density maps and the location of errors in these models. Acta Crystallogr Sect A Found Crystallogr 1991;47:110–9. [34] Kraulis PJ. MOLSCRIPT: a program to produce both detailed and schematic plots of protein structures. J Appl Crystallogr 1991;24:946–50. [35] Esnouf RM. An extensively modified version of MolScript that includes greatly enhanced coloring capabilities. J Mol Graphics Modell 1997;15:132–4. [36] Merritt EA, Bacon DJ. Raster3D: photorealistic molecular graphics. Methods Enzymol 1997;277:505–24.
