Introduction

Since its discovery in 1979, the role of p53 within cells has been intensely investigated by many research groups. Its importance in cancer biology is highlighted by the fact that more than 50% of tumors have mutations in p53, which makes it one of the most important proteins with respect to human disease. p53 is a tetrameric transcription factor that suppresses tumor formation by activating a set of genes that induce either cell cycle arrest or apoptosis (Kastan et al, 1992; el-Deiry et al, 1993; Symonds et al, 1994). Which one of the two pathways is chosen depends on the nature of the cellular stress that triggers the p53 response, such as DNA damage by UV or -irradiation, hypoxia, or oncogene activation (Levine et al, 2006). However, the question how p53 distinguishes these diverse cellular stress signals and determines which set of genes to be turned on is still not well understood, although recent reports on the role of ASPP (Samuels-Lev et al, 2001) and acetylation of K120 in the DNA binding domain (DBD) of p53 by Tip60 suggest a mechanism in which the apoptotic pathway is preferred over the cell cycle arrest pathway (Sykes et al, 2006; Tang et al, 2006).

During the past 10 years, p53 has expanded into a protein family with complexity and diversity, exemplified by the discovery of different splice variants of p53 and two mammalian paralogues, p63 and p73 (Yang and McKeon, 2000; Lu and Abrams, 2006). Sequence analysis of the two mammalian paralogues has revealed major differences to p53. Structurally, p63 and p73 have additional domains that are not found in p53, and exist in various isoforms created by a combination of C-terminal splicing and different N-terminal promoters (Kaghad et al, 1997; De Laurenzi et al, 1998; Yang et al, 1998). While p53 has an N-terminal transactivation domain (TA) for recruitment of core transcriptional factors, a central DBD for recognition of promoter sequences, an oligomerization domain (OD) for tetramerization, and a short basic stretch of 30 amino acids for regulation of transcriptional activity, the C-terminus of p63 and p73 contains (depending on the splice form) a sterile alpha motif (SAM) domain and a transcriptional inhibitory domain (TID). SAM domains are small protein–protein interaction modules that are found in a wide variety of different proteins, ranging from kinases and transcriptional regulators to cell surface receptors (Schultz et al, 1997). The TID, an unstructured region C-terminal to the SAM domain, was shown to inhibit the transcriptional activity of p63 by interacting with the TA domain (Serber et al, 2002). Functional analyses through knockout mice studies have shown that both proteins do not seem to be tumor suppressors like p53, instead they play important roles in development and maintenance of epithelial tissue (p63) or maintenance of certain neurons (p73) (Mills et al, 1999; Yang et al, 1999, 2000). The identification of mammalian p53, p63, and p73 allowed rapid classification of p53 homologues in other species. Vertebrate species like Xenopus, zebrafish, and chicken possess all three paralogues, while invertebrate species such as squids, clams, and molluscs have p53 homologues that are more closely related to the mammalian p63 and p73 than to mammalian p53. Interestingly, sequences from other invertebrate species like nematodes, fruit flies, and beetles cannot be classified as either p53 or p63/p73 like, and show similarity only within the DBD.

According to the current hypothesis on the evolution of the p53 protein family, p63/p73 resemble the ancestral form, while p53 evolved later (Lu and Abrams, 2006). The identification of many new p53 protein family members in invertebrates could further elucidate the evolutionary development of this important protein family. The structural differences between p53 and p63/p73 also reflect their different functions in surveillance of the genetic integrity of a cell and in tissue development, respectively. Since some of the invertebrate p53 forms show significant differences to both p53 and p63/p73, studying these family members might shed some new light on the evolution of function of p53. Particularly, the recent discovery of p53 homologues in the two important model organisms Drosophila melanogaster (Dmp53) and Caenorhabditis elegans (CEP-1) provides new genetic tools for the investigation of their function (Brodsky et al, 2000; Ollmann et al, 2000; Derry et al, 2001). Both proteins show only a low degree of sequence homology to p53 in the DBD, yet they exhibit very similar DNA binding specificity (Brodsky et al, 2000; Huyen et al, 2004). Studies have shown that both proteins are functionally distinct from mammalian p53. While the hallmark of p53 is its dual function in inducing either cell cycle arrest or apoptosis, both CEP-1 and Dmp53 can only induce apoptosis and not cell cycle arrest upon irradiation. These observations led to the idea that induction of cell cycle arrest was a later evolutionary development, since Drosophila and C. elegans have long diverged from the vertebrates.

The lack of any recognizable domain C-terminal to the DBD, including the highly conserved OD and the distinct functions of CEP-1 and Dmp53, prompted us to investigate the domain organization of the C-terminus in both proteins. We have determined the NMR structure of the C-terminus of CEP-1 and Dmp53. The CEP-1 C-terminus is composed of two sub-domains, an OD followed by a SAM domain that closely interacts with the OD. Surprisingly, the OD domain of CEP-1 forms a dimer instead of the usual tetramer observed in human p53. Equally surprising, the Dmp53 C-terminus reveals a unique oligomerization fold not found in other p53 proteins, in which an extra -strand and an extra helix complement the canonical p53 oligomerization fold. Furthermore, deletion studies and NMR data show that protein domains C-terminal to the OD are necessary to maintain the tertiary fold. The findings from this study suggest that the ancestral p53 form had a similar domain organization as p63/p73, and that the OD of vertebrate p53 that is conformationally stable without additional structural elements evolved later.

Results

Identification of a conserved domain in the C-terminus of CEP-1

Based on the protein sequence length, CEP-1 with 644 amino acids seems more closely related to p63 and p73; thus, we suspected the existence of additional domains C-terminal to the DBD in CEP-1. However, sequence analysis failed to reveal significant homology to domains frequently found in p63/p73, such as the SAM domain or even the OD identified so far in all members of the p53 family (Figure 1A).

Sequence comparison of CEP-1 from two Nematode species, C. elegans and Caenorhabditis briggsae, identified a conserved region C-terminal to the DBD (after residues 510, C. elegans numbering) (Supplementary Figure S1). After several rounds of optimization, we identified a core domain between residues 528–644 in the C-terminus of CEP-1, with a molecular weight of 13.6 kDa. A melting curve monitored by circular dichroism spectroscopy revealed a sharp transition point at 41°C, an indicator of a folded domain with a two-state unfolding transition (Supplementary Figure S2). Since this identified domain is C-terminal to the DBD, it should contain the OD responsible for tetramerization, a hallmark of the p53 protein family (Lee et al, 1994). In contrast to p53, however, the protein elutes off a size-exclusion column as a dimer (Figure 2A). This dimeric state was further confirmed by velocity sedimentation measurements, in which the data fit the expected molecular weight of a dimer (27.2 kDa) (Table I). Since neither a classical OD nor any other domains could be identified based on sequence comparison, we used standard heteronuclear NMR spectroscopy to investigate the structure of the C-terminus.

Overall structure of the C-terminus of CEP-1

The three-dimensional structure of the C-terminus of CEP-1 reveals that it consists of two interacting domains. Despite the lack of a significant sequence homology, one of these two domains is an OD, as found in all members of the p53 protein family, and the second domain is a SAM domain (Figures 1B and C). The dimerization interface of the OD consists mostly of interactions between the -strand and the -helix of one OD packing against the other OD in an antiparallel manner, as observed in the human p53 OD structure (Lee et al, 1994). The superimposed structures of the OD in human p53 dimer and CEP-1 have an r.m.s.d. of 2.0 Å, with the -helix in CEP-1 being two turns shorter than the corresponding helix in human p53 (Supplementary Figure S3). The SAM domain is composed of five helices that adopt the same topology as other SAM domain structures, for example, in EphB2 and p73 (Chi et al, 1999; Thanos et al, 1999). However, the SAM domain in CEP-1 resembles the EphB2 structure more closely than the p73 SAM domain, as indicated by its smaller r.m.s.d. values: 1.45 Å between EphB2 and CEP-1, and 1.83 Å between p73 and CEP-1 (Supplementary Figure S3). The OD and the SAM domain are connected by a 16-amino-acid linker, of which some resonances are either missing or could not be assigned due to significant overlap in the spectra. This linker region contains a short helix in its C-terminus that makes contacts with the first helix of the SAM domain. Direct interactions between the OD and the SAM domain are observed between the -helix of the OD and the last -helix of the SAM domain within the same monomer, and the sheet of the OD from the other monomer. The relative angle between the helix of the OD and the last helix of the SAM domain is 125°.

The OD of CEP-1

The OD is a well-conserved domain among vertebrate p53 forms, due to its necessity in tetramer formation (Chene, 2001). In contrast, the OD of CEP-1 shows a poor sequence homology, including key residues that are strictly conserved in all vertebrate p53 sequences (Figure 3A). The signature glycine residue that permits a sharp turn between the -strand and the -helix in the structure of the OD of p53 is replaced by a serine, threonine dipeptide in CEP-1 (Figures 3A and B). This substitution forms the angle between the -helix and the -strand within a monomer to widen from 26°, as seen in the p53 OD, to 50° in CEP-1. Further substitutions involve the dimer interface. In p53, the dimer interface is stabilized by hydrophobic interactions between F338 from the -helix of one monomer with F328 of the -strand of the other monomer. In CEP-1, this dimer interface is maintained by interaction between Y542 and the aliphatic part of the side chain of R533, confirmed by NOEs between the , , and protons of R533 and the and protons of Y542.

The most significant difference between CEP-1 and other members of the p53 protein family is its oligomerization property. While all other p53 protein family members form a tetramer through the OD, CEP-1 forms a dimer (Figure 2A; Table I). The typical OD of p53 family members is a dimer of dimers, involving two separate and distinct interfaces (Jeffrey et al, 1995) (Figure 4A). The first dimerization interface consists of an antiparallel -sheet formed by the -strand of each monomer, with additional contacts to the -helix. This dimerization interface is, as described above, conserved in CEP-1. The second dimerization interface in p53 is formed by interaction of the two helices of each dimer with the corresponding helices of the second dimer, thus creating a four-helix bundle. Key residues of the hydrophobic core of this tetramerization interface in p53 are M340, L344, A347, L348, and L350 from each helix (Chene, 2001) (Figure 4A). A structural alignment of the CEP-1 OD with other members of the p53 protein family shows that M340, A347, and L348 are replaced by K544, R551, and E552 in CEP-1 (Figures 3A and B). This substitution forms a ring of charged residues that encircles F548 from each monomer in the center of the ring (Figure 4B). If the OD of CEP-1 formed a tetramer with similar packing arrangement as the human p53 OD, then this electrostatic charged ring would form unfavorable interactions in the tetrameric interface due to charge repulsion. Thus, the lack of charge complementarity for residues K544, R551, and E552 explains the inability of the OD in CEP-1 to form a tetramer.

To test this hypothesis, we conducted two series of mutations toward converting the OD of CEP-1 from a dimer into a tetramer. Upon mutation of R551 and E552 to leucine, we observed an equilibrium of two oligomeric forms, with roughly 33% of CEP-1 in a tetrameric and 67% in a dimeric form (Figure 2A). Additional mutation of K544 to methionine transforms the OD of CEP-1 completely into a tetrameric state, as indicated by gel filtration and confirmed by analytical ultracentrifugation (Figure 2A; Table I). It is also worth noting that our constructs contained the whole C-terminus of CEP-1, which includes the SAM domain. Thus, it excludes the possibility that the SAM domain could contribute to tetramerization, since tetramerization can be achieved solely by mutations in the OD of CEP-1.

The SAM domain of CEP-1

Another significant difference between the OD of CEP-1 and that of other p53 family members is its close interaction with the C-terminal SAM domain. SAM domains are found in many different proteins, including ephrin receptors, transcriptional repressors, as well as the p53 homologues p63 and p73. They are composed of four short helices followed by a long helix that completes its hydrophobic core. SAM domains are protein–protein interaction modules that have the ability to homo- and hetero-dimerize, as well as oligomerize. In the ephrin receptor, the SAM domain forms a dimer in the crystal structure involving the last helix and loop 3 (between helices 3 and 4) (Thanos et al, 1999). In CEP-1, the equivalent loop 3 region is exposed to the solvent, and the C-terminus of the last helix interacts with the helix in the OD domain; thus, it is unlikely that this site constitutes a dimerization surface with another SAM domain. In the p63/p73 protein family, the SAM domain is a monomer in isolation and, to date, no homo-dimerization tendency has been identified (Chi et al, 1999). Similarly, the SAM domain of CEP-1 exists as a monomer in isolation (Table I), and no contacts are made between the SAM domains in the full-length construct of the C-terminus of CEP-1. Recently, SAM domains in the Smaug family, a translational repressor, have also been found as part of RNA binding modules that recognizes RNA hairpins with the loop sequence CUGGC (Aviv et al, 2003; Green et al, 2003). The RNA interaction surface on the SAM domain is composed of the N-terminus of helix 5 and loop 1 (between helices 1 and 2), which has a positively charged surface. In CEP-1, this region does not have a highly positive charged surface, so it is unlikely that the SAM domain in CEP-1 interacts with RNA as in Smaug.

Characterization of C-terminus of Dmp53

In addition to CEP-1, the Drosophila homologue of p53, Dmp53, shows low sequence homology to other p53 protein family members, again mainly confined to the DBD. The protein sequence length suggests it is more related to p53. Dmp53 induces apoptosis through the transcription of genes, such as hid, reaper, and sickle (Brodsky et al, 2004). The functional similarity of Dmp53 and CEP-1 and their low sequence homology to other p53 protein family members prompted us to investigate the oligomeric state of Dmp53 as well. We have purified the OD of Dmp53 (315–361, Dmp53 numbering in Ollmann et al, 2000), and shown by gel filtration as well as analytical ultracentrifugation that it forms a dimer (Figure 2B; Table I). NMR investigations of this dimeric form, however, have revealed that it exhibits conformational heterogeneity characterized by broad peaks in the NMR spectrum (Figure 5). Similar to human p53, Dmp53 also contains a stretch of highly basic residues C-terminal to the OD. Others have shown that this basic region in human p53 serves as a regulatory tail that controls DNA binding affinity (Ahn and Prives, 2001). To test whether this region has any structural function in Dmp53, and in particular, if it stabilizes the dimer, we purified a construct that includes the OD and this C-terminal region (additional 24 residues). Surprisingly, this extended OD forms a tetramer and its NMR spectra indicate the existence of a well folded single conformation (Figure 5; Table I). Structure determination by NMR spectroscopy shows that the C-terminal 24 amino acids form an -helix that packs against the -helix that forms the canonical OD (Figures 1B and D). In addition, the N-terminus is extended relative to the p53 OD and contains an additional -strand that forms together with the canonical -strand and the corresponding elements of another monomer an antiparallel four-stranded -sheet. Thus the Dmp53 OD domain contains both an N-terminal extension (the 1-strand) as well as a C-terminal extension (the H2 -helix) relative to human p53 and can best be described as a four-helical hairpin bundle sandwiched between two four-stranded -sheets. In Dmp53, the conserved glycine residue in the canonical OD is replaced by proline and asparagine (Figures 3A and B). The tetramerization interface is built by the inner four helices that pack against each other similar to the four helices of human p53 (Figure 4C). In contrast to human p53, the tetramerization interface of Dmp53 contains a central cluster of charged residues made up of lysine 352 (K352) and glutamic acid 353 (E353) from each helix, which can form four salt bridges that constitute the core of the tetramerization interface. Mutating E353 of this cluster to lysine disrupts the electrostatic pairings and destroys the tetramerization interface, thus forming a stable dimer (Figure 2B). The core of the tetramerization interface is shifted by one helical turn toward the C-terminus compared to human p53 (Figure 3A). In addition, the inner helix of each monomer (H1:C) also packs across the tetrameric interface to the C-terminal helix (H2:B) of another monomer, with contacts between L359 from the inner helix and A371 and L375 of the C-terminal helix (Figure 4C).

Further contacts of the C-terminal helix (H2:B) that contribute to the dimer interface involve a surface formed by the inner helix (H1:B) of the same monomer and the N-terminal -strand (1:A) of another monomer within a dimeric unit. In addition, this structural arrangement of the dimeric unit is capped by interaction of H370 (H2:B) with two tryptophans from the -strand (W321, 1:A) and the inner helix (W342, H1:A) of the other monomer (Figure 5). The involvement of the C-terminal helix in both the dimeric as well as the tetrameric interface explains why its deletion destabilizes the structure and results in a conformationally unstable dimer. It also predicts that removing the first -strand should have a similar destabilizing effect as deleting the C-terminal helix. Indeed, the loss of strand 1 forms a dimer that shows broad peaks in NMR spectra as well as a non-sigmoidal transition in CD-melting experiments, suggesting that it is conformationally unstable (Figures 2B and 5; Supplementary Figure S4). These results demonstrate that the minimal OD, consisting of one -strand followed by one helix, as identified in human p53, does not form a stable tetrameric structure in Dmp53. The additional N-terminal (-strand) and C-terminal (-helix) extensions play a key role in stabilizing Dmp53.

The SAM domain stabilizes the OD in CEP-1

Based on these observations with Dmp53, we asked if the SAM domain in CEP-1 might have a similar stabilizing function as the C-terminal helix in Dmp53. The absence of the SAM domain in the C-terminus of CEP-1 (528–555) indeed forms a less stable structure, as indicated by the lack of a sharp sigmoidal transition in the temperature denaturation curve as well as the absence of an alpha helical pattern (two minima at 208 and 222 nm) in the wavelength scan, as observed by CD spectroscopy (Supplementary Figure S2). Since the expression level of the isolated OD decreased significantly relative to the full-length C-terminus of CEP-1, we were not able to record an NMR spectrum.

Evolution of the p53 protein family

The structural analysis of the C-termini of CEP-1 and Dmp53 provides an opportunity to trace the evolution of the p53 protein family, based on the presence of the SAM and structure of the OD domains in the C-terminus. While the positioning of nematodes in a phylogenetic tree remains controversial (ecdysozoa versus coelomata) (Aguinaldo et al, 1997), we placed the currently available sequences of the p53 protein family members from the NCBI library into a phylogenetic tree constructed from the ecdysozoa perspective, which is the currently most favored view. As shown in Figure 6, the SAM domain has been integrated early into the p53 protein family, since it can be identified in the protostome branch represented by nematodes and molluscs, as well as in the early deuterostomes such as echinoderm and urochordate. In phylum Urochordata, a gene duplication seems to have occurred that results in a p53 form without a SAM domain, which resembles vertebrate p53 in addition to a p63/p73-like form. The loss of the SAM domain occurred also within the protostome branch in the Arthropoda phylum represented by Dmp53. However, in arthropods, a C-terminal helix replaces the SAM domain and serves a similar functional role in stabilizing the OD. While the functional role of the SAM domain in human p63 and p73 remains elusive, the existence of the SAM domain in p53-like molecules in the protostome phylum underscores its important role within the p53 protein family, such as its contribution in stabilizing the OD. The current study provides additional evidence that the vertebrate p53 is a recent evolutionary development, and p63 and p73 are the more ancestral family members.

However, this ecdysozoa-based phylogenetic tree does not permit a direct correlation of the oligomeric state and the evolutionary development of p53. Instead this tree would either suggest that the dimeric CEP-1 has de-evolved from an ancestral tetrameric p53, or that tetramerization has evolved twice independently. Given the high sequence identity between the oligomierzation domains of vertebrates and molluscs, this last hypothesis seems less likely. Alternatively, a phylogenetic tree that is based on a coelomata view places nematodes in an earlier phylogenetic branch than both Drosophila and vertebrates (Figure 6B). This hypothesis supports the view that p53 evolved from a dimeric molecule into a tetrameric form with further divergence between molluscs and arthropods. A variation of this phylogenetic tree places the Arthropoda phylum before the divergence of molluscs and vertebrates, and would be most consistent with our structural data (Sidow and Thomas, 1994; Hedges, 2002).

Discussion

The low homology of the C-terminus of CEP-1 and Dmp53 with other p53/p63/p73 proteins suggests that significant differences in both structure and function might exist in comparison with other p53 protein family members (Lu and Abrams, 2006). Indeed, this study shows that the C-terminus of CEP-1 is a dimeric molecule, making the C. elegans form of p53 the first native dimer of the entire protein family. The existence of a natural dimeric form of p53 reaffirms previous identification of residues that are essential for tetramerization (Mateu and Fersht, 1998, 1999), since the C-terminus of CEP-1 can be converted into a tetrameric form by mutations of residues at the tetrameric interface (K544, R551, E552) into hydrophobic residues.

The structure of the OD of CEP-1 illustrates a conserved strategy in the p53 protein family to utilize the electrostatic nature of the tetrameric interface of the OD to control the oligomeric state of the protein. McCoy et al (1997) had reported that mutating three key residues of the tetrameric interface of the OD in human p53 (M340K, F341I, L344Y) results in a dimeric molecule, with the orientation of the helices switched from antiparallel to parallel. Interestingly, in CEP-1, the corresponding residues (K544, V545, F548) show greater similarity to the p53 mutant sequence than to the p53 wild-type sequence, yet the structure of the CEP-1 OD resembles the structure of the p53 wild-type OD more closely, by adopting an antiparallel packing of helices (angle between the helices of one dimer unit: p53 156°; CEP-1 125°; mutant 78°). Mapping CEP-1 residues onto the structure of the p53 mutant shows that CEP-1 retains the antiparallel orientation due to unfavorable charge repulsion by the two lysines (K554) located at the C-terminus of the helix. In the mutant p53 molecule, these lysines are replaced by two leucines (L350) that form favorable hydrophobic packing, thus enabling a parallel orientation of the helices. The structural and mutational analyses of CEP-1 show that controlling the electrostatic nature of the interface allows the formation of both dimers as well as tetramers, without changing the antiparallel topology of the OD.

The presence of the SAM domain at the C-terminus of CEP-1 is a surprise discovery, since sequence alignment does not reveal its existence. The SAM domain of CEP-1 is C-terminal to the OD and makes contacts with the OD through the last helix, but not to the SAM domain from the other monomer. In other SAM domain containing proteins, the SAM domain functions as a homo- or hetero-dimerization domain, for example, in the ephrin receptor B2 and the SAM domains of Ste11 and Ste50 in Saccharomyces cerevisiae (Thanos et al, 1999; Kwan et al, 2006). Based on our structure, however, the SAM domain of CEP-1 does not seem to play a direct role in oligomerization. This result also has implications for p63 and p73, which both contain SAM domains at their C-terminus. The structure of the C-terminus of CEP-1 as well as the incapability of the isolated CEP-1/p63/p73 SAM domains to oligomerize suggests that SAM domains in the p53 protein family also do not form homo dimers or oligomers in the context of the full-length protein.

Despite the observation that CEP-1 is a dimer in solution, the possibility exists that two CEP-1 dimers can form a functional tetramer through cooperative binding to their promoter sites. Analyzing the promoter sequences of known targets of CEP-1 and Dmp53 should reflect the difference in the oligomeric state of both proteins in vivo. The promoter region of egl-1 (Hofmann et al, 2002), a CEP-1 inducible gene, contains only a half site (AAACAAGCTT), which satisfied the p53 consensus sequence motif for a dimeric p53 molecule (CEP-1) (el-Deiry et al, 1992). In contrast, the Drosophila p53 responsive gene, reaper, has two half sites (TGACATGTTT/GAACAAGTCG) (Brodsky et al, 2000), which allows binding of a tetrameric p53 molecule (Dmp53).

The hypothesis that a dimeric p53 is the ancestral form is further supported by a study on the formation of human p53 in rabbit reticulate lysate (Nicholls et al, 2002). This study showed that human p53 first forms a dimer co-translationally, and that tetramers are only formed at a later stage post-translationally. It was further shown that when mutant and wild-type p53 are coexpressed, there is only one form of heterotetramer, which consists of a dimer of wild type and a dimer of mutants. Thus, even after millions of years of evolution, the mammalian p53 still retains a dimeric building block as its basic unit.

Dmp53 and perhaps the entire arthropod phylum have adopted a different tetramerization mode by utilizing an additional helix C-terminal to the canonical mammalian p53 OD, and an additional -strand before the OD. By sequence alignment with CEP-1, one would conclude that Dmp53 would be a dimer due to the presence of the charged resides K352 and E353 in Dmp53. However, the helix and the additional -strand transform Dmp53 into a tetramer. The additional -strand provides a stable dimeric unit, which positions helix 2 of Dmp53 in the correct orientation to interact with the helix 1 of another dimeric unit, and concomitantly allows attractive electrostatic interactions along the tetrameric interface between the two charged residues. The composition of this unique mode is necessary for tetramerization, since a deletion of either additional element results in incorrect topology of helices, thus reverting to the dimeric state as in CEP-1.

It is interesting to note that only the C-terminus of the p53 protein family has undergone significant evolutionary changes (dimer or tetramer, presence or absence of the SAM domain), while the DNA specificity and the structure of the DBD have remained basically unchanged. The X-ray crystal structure of the DBD of CEP-1 was recently solved (Huyen et al, 2004). It demonstrates that despite a low sequence homology of only 15%, the core structure is very similar, and the DNA binding sequence specificity of CEP-1 is virtually identical to human p53, even though loop L1, a part of the DNA binding interface in the DBD of human p53, adopts a different conformation in CEP-1. In Dmp53, the DBD is 25% identical to human p53, which is higher than in the case of CEP-1. This higher sequence identity in Dmp53 and the high conservation of the structure and function of the DBD of CEP-1 predict that the structure of the Dmp53 DBD and its DNA binding specificity are also highly conserved. Despite this high conservation in the DBD, the biological function of the individual members of the p53 protein family is distinct, and sequences C-terminal to the DBD show very significant divergence, thus suggesting that these C-termini play an important role in specifying the biological function of the individual family members.

The availability of structural and biochemical data for CEP-1, Dmp53, and human p53 in combination with p53 sequence data from many species sheds some light on the ancestral form of p53 and its evolutionary development. It was estimated that C. elegans and Drosophila have diverged from vertebrates 550 millions years ago, thus CEP-1 and Dmp53 could resemble the ancestral form of p53. Despite the lack of a SAM domain in Dmp53, the additional structural elements (the first -strand and the last -helix) serve a similar stabilizing function as the SAM domain in CEP-1. Oligomerization in the ancestral p53 forms most likely required either the fusion of two domains (like in CEP-1) or additional structural elements (like in Dmp53). The minimal OD found in vertebrate p53 that can form stable tetramers by itself is probably a later evolutionary result, while the C-terminal tail that is a necessary structural element in Dmp53 became an important regulatory region including many sites for posttranslational modifications.

In summary, we have determined the structure of the C-terminal domain of CEP-1 and Dmp53, and show nature's multifaceted means to achieve oligomerization in p53, besides the canonical form found in mammalian p53. Furthermore, we have shown that additional structural elements identified in CEP-1 and Dmp53 that are not present in human p53 are necessary for the integrity of the OD. A loss of these elements results in conformationally unstable structures, and in Dmp53, leads to a change in the oligomeric property. The structural investigations described here suggest an evolutionary path from an ancestral dimeric form over tetrameric forms that need additional stabilization elements to the minimal tetramerization domain known from mammalian p53.

Materials and methods

Protein expression and molecular clonings

All CEP-1 and Dmp53 constructs used for structural studies were cloned into plasmid pGEX-6P-2 (Amersham Bioscience) or pBH4 (gift from Wendell Lim laboratory) using BamHI and XhoI sites. Mutagenesis constructs of both CEP-1 and Dmp53 were prepared by the QuickChange protocol from Stratagene. BL21 cells were grown to an OD₆₀₀ of 0.8 and induced with 500 M IPTG at 25°C for 8 h. The proteins were purified as described in the manufacturer's protocol, cleaved by precision protease for the pGEX plasmid or by TEV protease for the pBH4 plasmid, and further purified on a Superdex-75 gel filtration column. Protein samples were stored in a buffer containing 20 mM sodium phosphate (pH 7.0), 100 mM sodium chloride, and 0.03% sodium azide. For the expression of ¹⁵N- and ¹⁵N/¹³C-labeled proteins, bacteria were first grown in LB media to an OD₆₀₀ of 0.8, then transferred to M9 minimal media with the appropriate isotopic components, and induced under the same condition as described above. Protein samples used to obtain inter-monomer NOEs consisted of an equal ratio of ¹⁵N- and ¹³C-labeled proteins. For CEP-1, 2 weeks of equilibration time was needed to obtain signals in the experiment. For Dmp53, equal molar ratio of proteins were mixed, denatured in 6 M guanidium hydrochloride, and then refolded in the buffer described above.

NMR experiments and structure calculations

Backbone residues of CEP-1 and Dmp53 were assigned using the TROSY version of HNCA and HNCOCA. For CEP-1, specific labeling of lysine, tyrosine, and leucine were used to confirm assignments. Distance constraints were derived from ¹⁵N-NOESY-HSQC, and ¹³C-NOESY-HSQC. Aromatic protons were assigned based on 2D-D₂O-NOESY, 2D-D₂O-TOCSY, and non-constant time 3D-¹³C-NOESY. Inter-monomer NOEs were obtained through a 4D constant time J-Resolved NOESY (Melacini, 2000), measured with a 1:1 mixture of ¹²C- and ¹³C-labeled proteins in both cases. In the case of Dmp53, the protein had to be denatured first with guanidium hydrochloride before mixing of the ¹²C- and ¹³C-labeled proteins, and subsequent refolding by dialysis. For CEP-1, 44 unambiguous inter-monomer distance constraints, all located in the -sheet, were identified from the 4D J-Resolved NOESY. Overall, 114 inter-monomer NOEs were assigned. In the case of Dmp53, 184 NOEs obtained from the 4D J-Resolved NOESY and an overall of 240 inter-monomeric NOEs were used. Of these inter-monomeric NOEs, 46 are located in the -sheet, 130 between a -strand of one monomer and an -helix of the other monomer within one dimeric unit, and 31 between helices of different monomers within a dimer were observed. In addition, 33 NOEs across the tetrameric interface were identified. Dihedral angle constraints were derived from TALOS based on chemical shifts of N, CA, HA, and CB (Cornilescu et al, 1999). Hydrogen bond constraints of secondary structure elements were based on TALOS calculations and confirmed by characteristic NOE patterns for -helices and -sheets, as well as deuterium hydrogen exchange measurements. Structure calculations were carried out with Aria 1.2, with modified protocols that imposed a C₂ symmetry for CEP-1 and a D₂ symmetry for Dmp53 throughout every stage of calculation (Linge et al, 2001). For the structure calculation of CEP-1, unambiguous inter-monomer constraints obtained from the 4D J-Resolved NOESY were included in every iteration of the calculation. For Dmp53, unambiguous NOEs between the 2-strands of two monomers established the existence of an antiparallel inter-monomeric -sheet. Using these unambiguous NOEs as a starting point, we first calculated dimer structures. Based on these dimer structures, the other distance constraints obtained from the 4D J-Resolved NOESY were evaluated for their consistency with the dimer structure. Several distance constraints that could not be satisfied within the dimer were assumed to be constraints across the tetrameric interface, and were used as those in the following structure calculations. Twenty structures were calculated in seven iterations, and 100 structures were calculated in the last iteration. Initial assigned peaks were separated into unambiguous and ambiguous peaks, and ambiguous peaks with multiple assignments and violated restraints were manually inspected. The new ambiguous peak list included peaks with multiple assignments and inter-monomer assignments (not obtained from the 4D constant time J-Resolved NOESY). After multiple iterations of peak inspections and structure calculations, 100 structures were calculated and the best 20 structures were used for water refinements and analysis. The structural statistics of CEP-1 and Dmp53 are listed in Supplementary Tables 2 and 3. Images were prepared with Pymol (Delano, 2002). MOLMOL was used for structural alignment of NMR-derived models and generation of the electrostatic map of the molecule (Koradi et al, 1996). LSQMAN was used for structural alignment of different proteins (Kleywegt, 1996) and the structure validation program of the PDB server for further structural analysis.

Circular dichroism experiments

Temperature scans (20–90°C for CEP-1, 20–100°C for Dmp53) were measured in a Jasco 810 CD spectrometer. For CEP-1 OD alone construct (528–555), it contains additional his-tags and TEV protease cleavage site. Due to the unusual circular dichroism spectra in Dmp53, the observed wavelength for temperature scan was chosen based on signal intensity. Thus, for wild-type Dmp53, CD was observed at 234 nm, Dmp53 (326–385) at 228 nm, and Dmp53 (315–361) at 220 nm. For temperature denaturation scans, ellipticity of each protein sample was converted into fractional ellipticity with respect to the signal at 100% denatured state, in order to normalize the data for all samples. Protein concentration of each construct ranged from 50 to 300 M, in 20 mM sodium phosphate (pH 7.0) and 100 mM NaCl buffer.

Analytical ultracentrifugation experiments

Analytical ultracentrifugation runs were conducted on an Optima XL-A centrifuge (Beckman Coulter Instruments, CA). The data were collected at a wavelength of 280 nm.

Sedimentation velocity Sedimentation velocity (SV) experiments were conducted with 200–300 l samples in 20 mM sodium phosphate (pH 7.0), 100 mM NaCl at protein concentrations of 0.5–1 mg/ml. Absorbance data were acquired at rotor speeds of 35 000–40 000 r.p.m. and at a temperature of 20°C. The buffer density of 1.005 g/ml and viscosity of 1.031 cPoise and the protein partial-specific volumes were calculated using the software SEDNTERP, kindly provided by Dr J Philo. Data were analyzed using the c(s) continuous distribution of Lamm equation solutions with the software SEDFIT (Schuck, 2000; Schuck et al, 2002).

Sedimentation equilibrium Sedimentation equilibrium experiments were conducted at 20°C at rotor speeds of 15 000 r.p.m. at an optical density of 0.283. Global non-linear regression of the experimental absorbance profiles was performed using the software SEDPHAT, kindly provided by Dr P Schuck.

Supplementary data

Supplementary data are available at The EMBO Journal Online (http://www.embojournal.org).

Acknowledgements

A plasmid with a partial CEP-1 sequence was a gift from Brent Derry. We thank Felician Dancea, Wesley McGinn-Straub, Zach Serber, Michael Reese, Florian Durst, Meichen Shi for discussions. This work was supported by the Centre for Biomolecular Magnetic Resonance at the University Frankfurt (BMRZ), the DFG (DO 545/2-1), EU-Grant EPISTEM (LSHB-CT-019067) and Philip Morris USA Inc., and by Philip Morris International.

References

Aguinaldo AM, Turbeville JM, Linford LS, Rivera MC, Garey JR, Raff RA, Lake JA (1997) Evidence for a clade of nematodes, arthropods and other moulting animals. Nature 387: 489–493 | Article | PubMed | ISI | ChemPort |

Ahn J, Prives C (2001) The C-terminus of p53: the more you learn the less you know. Nat Struct Biol 8: 730–732 | Article | PubMed | ISI | ChemPort |

Aviv T, Lin Z, Lau S, Rendl LM, Sicheri F, Smibert CA (2003) The RNA-binding SAM domain of Smaug defines a new family of post-transcriptional regulators. Nat Struct Biol 10: 614–621 | Article | PubMed | ISI | ChemPort |

Brodsky MH, Nordstrom W, Tsang G, Kwan E, Rubin GM, Abrams JM (2000) Drosophila p53 binds a damage response element at the reaper locus. Cell 101: 103–113 | Article | PubMed | ISI | ChemPort |

Brodsky MH, Weinert BT, Tsang G, Rong YS, McGinnis NM, Golic KG, Rio DC, Rubin GM (2004) Drosophila melanogaster MNK/Chk2 and p53 regulate multiple DNA repair and apoptotic pathways following DNA damage. Mol Cell Biol 24: 1219–1231 | Article | PubMed | ISI | ChemPort |

Chene P (2001) The role of tetramerization in p53 function. Oncogene 20: 2611–2617 | Article | PubMed | ISI | ChemPort |

Chi SW, Ayed A, Arrowsmith CH (1999) Solution structure of a conserved C-terminal domain of p73 with structural homology to the SAM domain. EMBO J 18: 4438–4445 | Article | PubMed | ISI | ChemPort |

Cornilescu G, Delaglio F, Bax A (1999) Protein backbone angle restraints from searching a database for chemical shift and sequence homology. J Biomol NMR 13: 289–302 | Article | PubMed | ISI | ChemPort |

De Laurenzi V, Costanzo A, Barcaroli D, Terrinoni A, Falco M, Annicchiarico-Petruzzelli M, Levrero M, Melino G (1998) Two new p73 splice variants, gamma and delta, with different transcriptional activity. J Exp Med 188: 1763–1768 | Article | PubMed | ISI | ChemPort |

DeLano WL (2002) The PyMOL Molecular Graphics System

Derry WB, Putzke AP, Rothman JH (2001) Caenorhabditis elegans p53: role in apoptosis, meiosis, and stress resistance. Science 294: 591–595 | Article | PubMed | ISI | ChemPort |

el-Deiry WS, Kern SE, Pietenpol JA, Kinzler KW, Vogelstein B (1992) Definition of a consensus binding site for p53. Nat Genet 1: 45–49 | Article | PubMed | ChemPort |

el-Deiry WS, Tokino T, Velculescu VE, Levy DB, Parsons R, Trent JM, Lin D, Mercer WE, Kinzler KW, Vogelstein B (1993) WAF1, a potential mediator of p53 tumor suppression. Cell 75: 817–825 | Article | PubMed | ChemPort |

Green JB, Gardner CD, Wharton RP, Aggarwal AK (2003) RNA recognition via the SAM domain of Smaug. Mol Cell 11: 1537–1548 | Article | PubMed | ISI | ChemPort |

Hedges SB (2002) The origin and evolution of model organisms. Nat Rev Genet 3: 838–849 | Article | PubMed | ISI | ChemPort |

Hofmann ER, Milstein S, Boulton SJ, Ye M, Hofmann JJ, Stergiou L, Gartner A, Vidal M, Hengartner MO (2002) Caenorhabditis elegans HUS-1 is a DNA damage checkpoint protein required for genome stability and EGL-1-mediated apoptosis. Curr Biol 12: 1908–1918 | Article | PubMed | ISI | ChemPort |

Huyen Y, Jeffrey PD, Derry WB, Rothman JH, Pavletich NP, Stavridi ES, Halazonetis TD (2004) Structural differences in the DNA binding domains of human p53 and its C. elegans ortholog Cep-1. Structure (Camb) 12: 1237–1243 | ChemPort |

Jeffrey PD, Gorina S, Pavletich NP (1995) Crystal structure of the tetramerization domain of the p53 tumor suppressor at 1.7 angstroms. Science 267: 1498–1502 | Article | PubMed | ISI | ChemPort |

Kaghad M, Bonnet H, Yang A, Creancier L, Biscan JC, Valent A, Minty A, Chalon P, Lelias JM, Dumont X, Ferrara P, McKeon F, Caput D (1997) Monoallelically expressed gene related to p53 at 1p36, a region frequently deleted in neuroblastoma and other human cancers. Cell 90: 809–819 | Article | PubMed | ISI | ChemPort |

Kastan MB, Zhan Q, el-Deiry WS, Carrier F, Jacks T, Walsh WV, Plunkett BS, Vogelstein B, Fornace Jr AJ (1992) A mammalian cell cycle checkpoint pathway utilizing p53 and GADD45 is defective in ataxia-telangiectasia. Cell 71: 587–597 | Article | PubMed | ISI | ChemPort |

Kleywegt GJ (1996) Use of non-crystallographic symmetry in protein structure refinement. Acta Cryst D 52: 842–857 | Article | ISI | ChemPort |

Koradi R, Billeter M, Wuthrich K (1996) MOLMOL: a program for display and analysis of macromolecular structures. J Mol Graph 14: 51–55, 29–32 | Article | PubMed | ISI | ChemPort |

Kwan JJ, Warner N, Maini J, Chan Tung KW, Zakaria H, Pawson T, Donaldson LW (2006) Saccharomyces cerevisiae Ste50 binds the MAPKKK Ste11 through a head-to-tail SAM domain interaction. J Mol Biol 356: 142–154 | Article | PubMed | ISI | ChemPort |

Lee W, Harvey TS, Yin Y, Yau P, Litchfield D, Arrowsmith CH (1994) Solution structure of the tetrameric minimum transforming domain of p53. Nat Struct Biol 1: 877–890 | Article | PubMed | ISI | ChemPort |

Levine AJ, Hu W, Feng Z (2006) The P53 pathway: what questions remain to be explored? Cell Death Differ 13: 1027–1036 | Article | PubMed | ISI | ChemPort |

Linge JP, O'Donoghue SI, Nilges M (2001) Automated assignment of ambiguous nuclear overhauser effects with ARIA. Methods Enzymol 339: 71–90 | PubMed | ISI | ChemPort |

Lu WJ, Abrams JM (2006) Lessons from p53 in non-mammalian models. Cell Death Differ 13: 909–912 | Article | PubMed | ISI | ChemPort |

Mateu MG, Fersht AR (1998) Nine hydrophobic side chains are key determinants of the thermodynamic stability and oligomerization status of tumour suppressor p53 tetramerization domain. EMBO J 17: 2748–2758 | Article | PubMed | ISI | ChemPort |

Mateu MG, Fersht AR (1999) Mutually compensatory mutations during evolution of the tetramerization domain of tumor suppressor p53 lead to impaired hetero-oligomerization. Proc Natl Acad Sci USA 96: 3595–3599 | Article | PubMed | ChemPort |

McCoy M, Stavridi ES, Waterman JL, Wieczorek AM, Opella SJ, Halazonetis TD (1997) Hydrophobic side-chain size is a determinant of the three-dimensional structure of the p53 oligomerization domain. EMBO J 16: 6230–6236 | Article | PubMed | ISI | ChemPort |

Melacini G (2000) Separation of intra- and intermolecular NOEs through simultanerous editing and J-Compensated filtering: A 4D quadrature-free constant-time J-Resolved approach. J Am Chem Soc 122: 9735–9738 | Article | ISI | ChemPort |

Mills AA, Zheng B, Wang XJ, Vogel H, Roop DR, Bradley A (1999) p63 is a p53 homologue required for limb and epidermal morphogenesis. Nature 398: 708–713 | Article | PubMed | ISI | ChemPort |

Nicholls CD, McLure KG, Shields MA, Lee PW (2002) Biogenesis of p53 involves cotranslational dimerization of monomers and posttranslational dimerization of dimers. Implications on the dominant negative effect. J Biol Chem 277: 12937–12945 | Article | PubMed | ISI | ChemPort |

Ollmann M, Young LM, Di Como CJ, Karim F, Belvin M, Robertson S, Whittaker K, Demsky M, Fisher WW, Buchman A, Duyk G, Friedman L, Prives C, Kopczynski C (2000) Drosophila p53 is a structural and functional homolog of the tumor suppressor p53. Cell 101: 91–101 | Article | PubMed | ISI | ChemPort |

Samuels-Lev Y, O'Connor DJ, Bergamaschi D, Trigiante G, Hsieh JK, Zhong S, Campargue I, Naumovski L, Crook T, Lu X (2001) ASPP proteins specifically stimulate the apoptotic function of p53. Mol Cell 8: 781–794 | Article | PubMed | ISI | ChemPort |

Schuck P (2000) Size-distribution analysis of macromolecules by sedimentation velocity ultracentrifugation and lamm equation modeling. Biophys J 78: 1606–1619 | PubMed | ISI | ChemPort |

Schuck P, Perugini MA, Gonzales NR, Howlett GJ, Schubert D (2002) Size-distribution analysis of proteins by analytical ultracentrifugation: strategies and application to model systems. Biophys J 82: 1096–1111 | PubMed | ISI | ChemPort |

Schultz J, Ponting CP, Hofmann K, Bork P (1997) SAM as a protein interaction domain involved in developmental regulation. Protein Sci 6: 249–253 | PubMed | ISI | ChemPort |

Serber Z, Lai HC, Yang A, Ou HD, Sigal MS, Kelly AE, Darimont BD, Duijf PH, Van Bokhoven H, McKeon F, Dotsch V (2002) A C-terminal inhibitory domain controls the activity of p63 by an intramolecular mechanism. Mol Cell Biol 22: 8601–8611 | Article | PubMed | ISI | ChemPort |

Sidow A, Thomas WK (1994) A molecular evolutionary framework for eukaryotic model organisms. Curr Biol 4: 596–603 | Article | PubMed | ISI | ChemPort |

Sykes SM, Mellert HS, Holbert MA, Li K, Marmorstein R, Lane WS, McMahon SB (2006) Acetylation of the p53 DNA-binding domain regulates apoptosis induction. Mol Cell 24: 841–851 | Article | PubMed | ISI | ChemPort |

Symonds H, Krall L, Remington L, Saenz-Robles M, Lowe S, Jacks T, Van Dyke T (1994) p53-dependent apoptosis suppresses tumor growth and progression in vivo. Cell 78: 703–711 | Article | PubMed | ISI | ChemPort |

Tang Y, Luo J, Zhang W, Gu W (2006) Tip60-dependent acetylation of p53 modulates the decision between cell-cycle arrest and apoptosis. Mol Cell 24: 827–839 | Article | PubMed | ISI | ChemPort |

Thanos CD, Goodwill KE, Bowie JU (1999) Oligomeric structure of the human EphB2 receptor SAM domain. Science 283: 833–836 | Article | PubMed | ISI | ChemPort |

Yang A, Kaghad M, Wang Y, Gillett E, Fleming MD, Dotsch V, Andrews NC, Caput D, McKeon F (1998) p63, a p53 homolog at 3q27-29, encodes multiple products with transactivating, death-inducing, and dominant-negative activities. Mol Cell 2: 305–316 | Article | PubMed | ISI | ChemPort |

Yang A, McKeon F (2000) P63 and P73: P53 mimics, menaces and more. Nat Rev Mol Cell Biol 1: 199–207 | Article | PubMed | ISI | ChemPort |

Yang A, Schweitzer R, Sun D, Kaghad M, Walker N, Bronson RT, Tabin C, Sharpe A, Caput D, Crum C, McKeon F (1999) p63 is essential for regenerative proliferation in limb, craniofacial and epithelial development. Nature 398: 714–718 | Article | PubMed | ISI | ChemPort |

Yang A, Walker N, Bronson R, Kaghad M, Oosterwegel M, Bonnin J, Vagner C, Bonnet H, Dikkes P, Sharpe A, McKeon F, Caput D (2000) p73-deficient mice have neurological, pheromonal and inflammatory defects but lack spontaneous tumours. Nature 404: 99–103 | Article | PubMed | ISI | ChemPort |