“Similarity Trap” in protein-protein interactions could be carcinogenic: simulations of p53 core domain complexed with 53BP1 and BRCA1 BRCT domains

doi:10.1016/j.str.2006.10.009

Structure. Author manuscript; available in PMC 2008 June 14.

Published in final edited form as:

Structure. 2006 December; 14(12): 1811–1821.

doi: 10.1016/j.str.2006.10.009

PMCID: PMC2429992

NIHMSID: NIHMS14891

“Similarity Trap” in protein-protein interactions could be carcinogenic: simulations of p53 core domain complexed with 53BP1 and BRCA1 BRCT domains

Jin Liu,¹ Yongping Pan,¹ Buyong Ma,¹^* and Ruth Nussinov^1,²^*

¹ Basic Research Program, SAIC-Frederick, Inc. Center for Cancer Research Nanobiology Program, NCI-Frederick, Frederick, MD 21702

² Sackler Institute of Molecular Medicine, Department of Human Genetics and Molecular Medicine, Sackler School of Medicine, Tel Aviv University, Tel Aviv 69978, Israel

*Correspondence: Correspondence can be addressed to Buyong Ma, Email: mab/at/ncifcrf.gov, Ruth Nussinov, Email: ruthn/at/ncifcrf.gov

The publisher's final edited version of this article is available at Structure

See other articles in PMC that cite the published article.

Publisher's Disclaimer

Summary

Similar binding sites often imply similar protein-protein interactions and similar functions; however, similar binding sites may also constitute traps for nonfunctional associations. How are similar sites distinguished to prevent misassociations? BRCT domain from breast cancer-susceptibility gene product BRCA1 is structurally similar to 53BP1-BRCT domain, yet with different binding behavior with p53 core domain. 53BP1-BRCT domain forms a stable complex with p53. In contrast, BRCA1-p53 interaction is weak or other mechanisms operate. To delineate the difference, we designed thirteen BRCA1-BRCT mutants, and computationally investigated the structural and stability changes compared to the experimental p53-53BP1 structure. Interestingly, of the thirteen, the two mutations which are cancerous and involve non-conserved residues are those that enforced p53 core domain binding with BRCA1-BRCT in a way similar to p53-53BP1 binding. Hence, falling into the “similarity trap” may disrupt normal BRCA1 and p53 functions. Our results illustrate how this trap is avoided in the native state.

Keywords: protein-protein interactions, BRCT domain, 53BP1 BRCT-p53 interactions, BRCA1-p53 interaction, protein binding sites

Introduction

The use of genomic sequences to infer and detect protein function and protein-protein interactions has been widely accepted as an effective approach to untangle complex biological phenomena (Marcotte, et al.,1999). In particular, protein-protein interactions are also used to infer domain-domain interactions (Deng, et al.,2002). Structural conservation in protein-protein interactions (Ma, et al.,2003) distinguishes a protein binding site from the exposed protein surface; and often, structural similarity may lead to a similar protein-protein interaction and function (Aloy, et al.,2004, Aloy, et al.,2003, Inbar, et al.,2003). These observations are easily understandable based on the evolution and the physico-chemical nature of protein-protein interactions. Nevertheless, given these similarities, the question arises how does nature tell apart subtle differences and what happens if nature’s choice fails.

One example is a domain (BRCT) from the breast cancer-susceptibility gene product BRCA1. BRCA1 relates to 45% of the families with inherited breast cancers and 90% of the families with inherited breast and ovarian cancers (Futreal, et al.,1994, Miki, et al.,1994). BRCA1 encodes a large protein of 1863 amino acids, with a zinc-finger RING domain N-terminal and tandem BRCT (BRCA1 C-terminal) domains. BRCT was first identified in BRCA1 as ~95 amino acid tandem repeats (Koonin, et al.,1996) and has been found in many proteins such as p53-binding protein 53BP1(Derbyshire, et al.,2002, Joo, et al.,2002), the base excision response scaffold protein XRCC1 and DNA ligase IV (Sibanda, et al.,2001), many of which appear to participate in cell cycle checkpoints or DNA repair in many species (Glover, et al.,2004). BRCT domain of BRCA1 is the target for a number of cancer-related mutations, and there is evidence that loss of BRCA1 BRCT domains leads to tumor formation in mice (Ludwig, et al.,2001).

There are data that BRCA1 stimulates p53 transcriptional activity (Chai, et al.,1999, Hartman and Ford,2003, MacLachlan, et al.,2002, Navaraj, et al.,2005, Zhang, et al.,1998). p53 is protein which is critical to genome stability, and missense mutations in p53 are involved in 50% of all human cancers (Hainaut and Hollstein,2000). BRCA1 has been reported to be able to physically associate with p53 with two interaction domains: the central disordered region of BRCA1 interacting with C-terminal domain of p53 (Mark, et al.,2005) and the BRCT domain of BRCA1 binding with the core domain of p53 (Chai, et al.,1999).

53BP1 – p53 interactions have been supported by biological experiments and observed directly in the resolved x-ray structure of the 53BP1-p53 complex. The 53BP1-p53 binding site partially overlaps the p53 DNA binding site thus inhibits the DNA binding activities of p53 (Iwabuchi, et al.,1994). Although in vitro studies have shown that p53 cannot bind 53BP1 and DNA simultaneously (Iwabuchi, et al.,1994), there is increasing evidence that 53BP1 enhances the transcription function of p53 (Iwabuchi, et al.,1998), (Glover, et al.,2004).

Both 53BP1 and human BRCA1 have two BRCT repeats, with high structural similarities even though the sequence identity is only 19%. Each repeat consists of four β strands and four α helices with the exception that one of the α helices is disordered in the C-terminal repeat of BRCA1, as shown in Figure 1(a). The BRCT region of 53BP1 was taken from the crystal structure of 53BP1-p53 complex (PDB code 1kzy) and was superimposed (by SwissView) on the crystal structure of BRCA1 BRCT (PDB code 1jnx), with a root mean squared deviation (RMSD) of 1.44 Å for 133 out of 211 BRCA1 Cα atoms, including all eight β strands and seven of eight α helices. The N-terminal repeat (repeat 1) of 53BP1 and BRCA1 can be superimposed with RMSD of 1.38 Å (for 69 out of 88 Cα atoms), and the C-terminal repeat (repeat 2) superimposition has an RMSD of 1.25 Å (for 60 out of 94 Cα atoms). The sequence identities of repeat 1 and repeat 2 are 24% and 17%, respectively. The least conserved region is the linker between repeat 1 and repeat 2 with a low 10% identity. Except for the linker, the region involved in 53BP1 bound to p53, including α3A through α4A, has a striking structural conservation with the corresponding region of BRCA1, with an RMSD 0.58 Å in 23 out of 23 Cα atoms. The sequence identity of this region (26%) is also higher than that in the other regions.

Figure 1

(a) Structures of the BRCT domains of 53BP1 and BRCA1. The regions bound to p53 are shown in yellow. The linker region and LA loop are shown in red. (b) Superimposition of the BRCT domain of 53BP1 (green) and BRCA1 (red) to build a model of BRCA1 interacting (more ...)

Despite the structural conservations, p53 core domain interacts with the BRCT domains of 53BP1 and BRCA1 proteins to different extents. Using biophysical methods, including isothermal titration calorimetry, analytical ultracentrifugation, and analytical size-exclusion chromatography, Ekblad et al confirmed the p53 core domain interactions with the BRCT domain of 53BP1 protein, but not with BRCA1 BRCT domain (Ekblad, et al.,2004). While it is possible that these biophysical methods are not sensitive enough, it does imply that if there is an interaction between BRCA1 BRCT domain and p53 core domain it is very weak or other mechanisms operate.

In order to delineate the difference between the p53 core domain interactions with the BRCT domains of 53BP1 and BRCA1 proteins, we designed thirteen BRCA1-BRCT domain mutants, and computationally investigated the stability changes of the simulated complexes, in comparison with the experimentally known p53-53BP1 structure and the structural changes that take place. Four of the mutations we chose are known to be carcinogenic. Our results revealed that the p53 core domain interactions with the BRCT domains of 53BP1 and BRCA1 proteins have different charge-charge interaction and hydrophobic patterns, leading to decreased binding abilities of the BRCA1-BRCT domain, despite the conservation of the BRCT structure in 53BP1 and BRCA1. Most importantly, it is the two non-conserved carcinogenic mutants that enforced p53 core domain binding with BRCA1-BRCT domain the most in a way similar to that of the 53BP1 protein. Therefore, we propose that the known binding pattern in p53-53BP1 interaction is a “similarity trap” that nature has to avoid in the p53-BRCA1 interactions. Falling into the “similarity trap” may disrupt the normal functions of both BRCA1 and p53, and hence may be carcinogenic.

Computational methods

(1) Modeling

Currently, there is no resolved structure to provide any atomic details of the interaction of BRCA1 with p53. However, the complex structure of p53 with 53BP1, whose BRCT domain is strikingly similar to BRCA1 BRCT domain, was solved (Derbyshire, et al.,2002, Joo, et al.,2002). Here, we build a model of BRCA1-p53 complex based on the crystal structure of 53BP1 BRCT domain bound to p53 core domain (PDB code: 1kzy) and the crystal structure of BRCA1 BRCT domain (PDB code: 1jnx) as well as the NMR structure of the C-terminal repeat of BRCA1 BRCT domain (PDB code: 1oqa). Specifically, we performed structural alignment of 53BP1 BRCT domain and BRCA1 BRCT domain, followed by replacing 53BP1 in the 53BP1-p53 complex by BRCA1. Because four residues are missing in the BRCA1 crystal structure, but three of these are available in the NMR structure of the C-terminal repeat of BRCA1 BRCT domain, the BRCA1 crystal structure was aligned with the NMR structure of its C-terminal repeat to fill the coordinates of missing residues. The fourth missing residue was modeled using CHARMM. The four missing residues underwent 20 ps dynamics after energy minimization for 500 steps with the rest of the protein fixed. The starting structures of the BRCA1-p53 complex with mutations were built from the wild type model, with all of the backbone atoms superimposed on the corresponding atoms of the wild type structure. The side chains were generated using CHARMM and underwent 500 steps of energy minimization with the rest of protein fixed to remove any steric conflicts before the MD simulations were performed.

(2) MD simulations

Molecular dynamics (MD) simulations have been performed with CHARMM 22 (MacKerell,1998) force field using CHARMM (Brooks,1983). The starting structure of the 53BP1 BRCT domain bound to p53 core domain was constructed from the crystal structure solved by Joo et al (Joo, et al.,2002) (PDB code 1kzy). The homology structure of the complex of BRCA1 BRCT region with p53 was constructed, based on the crystal structure of 53BP1-p53 and the crystal and NMR structure of the BRCT region of BRCA1 (Gaiser, et al.,2004, Williams, et al.,2001) (PDB code 1jnx and 1oqa). The 53BP1 monomer structure is extracted from the 53BP1-p53 complex. The structure of 27 residues are missing in 53BP1. Missing residues were added as random coils and 20 ps dynamics were performed after energy minimization for 500 steps with the rest of the protein fixed. All models are solvated in a TIP3P water box with a minimum distance of 10 Å from the edge of the box to any protein atom. The charges of the system were neutralized by adding chloride or sodium ions. To eliminate residual unfavorable interactions between the solvent and the protein, the solvated systems were first minimized for 500 steps with the protein restrained followed by another 500 steps of minimization for the whole system using the steepest decent algorithm. After 20 ps equilibration with the NVT ensemble, the production simulations were performed for 10 ns with the NPT ensemble at a temperature of 300K. During the production simulation, the time step was 2 fs, with a SHAKE constraint on all bonds containing hydrogen atoms, and the non-bonded cutoff was 12Å. Structures were saved every 2 ps. For p53 involved simulations, the distances between the zinc atom and its coordinate residues, including three Cys and one His residues were constrained within 0.2 Å of the crystal distance using the nuclear overhauser enhancement (NOE) module implemented in CHARMM. The simulation results were analyzed with CHARMM. The contact residue pairs of 53BP1-p53 are defined as residues pairs with minimum atom distance less that 4 Å. For the BRCA1-p53 interaction, the corresponding pairs to 53BP1-p53 were analyzed as well as residue pairs with minimum atom distance less than 4 Å.

(3) Free energy calculations

The interaction between BRCA1 and p53 were calculated as follows:

(1)

where E_MM is the molecular mechanical energy of the system consisting of all components in the CHARMM potential energy function, as shown equation (2),

(2)

and G_solv could be written as two terms, as shown in equation (3).

(3)

The electrostatic contribution to the solvation energy G_elec were calculated with the Generalized Born using the Molecular Volume (GBMV) method (Lee, et al.,2003). The most accurate grid-based module was used. The non-electronic term G_ne was calculated through solvent accessible surface area (SASA) calculations.

Energy calculations [ΔG(BRCA1-p53)] were performed on the 500 structures extracted from the simulation trajectories at 20 ps time interval during the 10 ns simulations. Statistical analysis was performed by obtaining the overall averages from these 500 structures.

The contribution of mutations to the binding energy of BRCA1 and p53 were evaluated using the following equation,

(4)

Results

In the biophysical analysis of BRCT domain of 53BP1 and BRCA1, Ekblad and coworkers reported that the BRCT domain of 53BP1 is slightly more stable than that of BRCA1 (Ekblad, et al.,2004). Molecular dynamics simulations were performed to investigate the stability of these two proteins. The RMSD of the Cα atoms of these two proteins over time with respect to the starting minimized structures are very close to each other, as shown in Figure 2(a), with BRCA1 slightly larger than 53BP1, which is consistent with the experimental report. The conserved subunit repeat 1 has the closest stability to 53BP1 and BRCA1 (Figure 2(b)). The more disordered α2B of BRCA1 contributes to the slightly more flexible repeat 2 vs the corresponding region of 53BP1, as shown in Figure 2(d). The inter-repeat linker is not as conserved as other regions. 53BP1 with the β-hairpin-like structure is more stable than BRCA1 with loop LA followed by the α-helix structure at the linker position, as shown in Figure 2(c). Note that the linker region is one of the binding regions of 53BP1 with p53. Structural diversity and flexibility at this region may contribute to a weaker binding of BRCA1 to p53 than 53BP1.

Figure 2

Cα RMSDs for the BRCT domain and its sub-domain of 53BP1 (blue) and BRCA1 (red) as well as 53BP1-p53 (blue) and BRCA1-p53 (red) complexes. (a) Cα RMSDs of 53BP1 are slightly lower than of BRCA1 for the whole BRCT domain including repeat (more ...)

53BP1-p53 and BRCA1-p53 complexes

The interaction regions of 53BP1 with p53 include part of the p53 DNA binding region and 53BP1 N-terminal repeat as well as the linker region. Specifically, loops L2, L3 and H1 helix of p53 interact with the linker, α4A and α3A of 53BP1, respectively. One of the DNA binding features, H2, is not involved in the 53bp1-p53 interaction. A model was built based on the 53BP1-p53 crystal structure and BRCA1 NMR structure, as shown in Figure 1(b). Molecular dynamics simulations were performed for both 53BP1-p53 and BRCA1-p53 complexes. In spite of the relatively small contact area with p53, the interface of 53BP1-p53 complex is rather stable. The RMSD value of the interaction region vs the initial structure is around 2Å (blue in Figure 2(e)). The BRCT structures are similar in BRCA1 and 53BP1, however, the simulations suggest that the interaction region of the BRCA1-p53 complex is much less stable (Figure 2(e), in red).

The interface sequence alignment of BRCA1 and 53BP1 is shown in Figure 1(d). 53BP1 has ten residues that contact with p53, including three in α3A, four in α4A, and three in the linker region. BRCA1 retains eight of these ten residues to interact with p53, but the two contact positions in the linker region are different because of the distinct linker structures of 53BP1 and BRCA1. Among those eight same-position residues, the least conserved residue is P1849/R1737. 53BP1 Proline 1849 interacts with Arginine 248 of p53 by favorable stacking interaction; however, there is a charge conflict between the corresponding R248-R1737 destabilizing the interaction of BRCA1 and 53BP1. BRCA1 K1724 in α3A also has a charge conflict with p53 R181.

In addition to the sequence distinction at the interface, there are two structural differences that may also affect the interaction of BRCA1 with p53. The structure between β3A and α2A is one of them, with a 6 residue loop in 53BP1 but an 11 residue loop in BRCA1. The longer loop of BRCA1 reaches p53, providing BRCA1 an additional possible contact with p53 as compared to 53BP1, including interactions of Arg248-Glu1694 and Met243-Phe1695.

The linker region structure is also different between 53BP1 and BRCA1. In 53BP1 the linker is composed of two β strands followed by a loop, whereas BRCA1 has a loop followed by α helix. Due to the structural difference, corresponding residues in the BRCA1 linker region cannot interact with p53 as in 53BP1. Instead, in the BRCA1-p53 complex two other residues, R1744 and K1750, contact with R248 and Q167 of p53, respectively. The R1744-R248 charge conflict destabilizes the BRCA1-p53 interface.

In the stable 53BP1-p53 complex most interacting residues retain their interaction over time, as shown in Figure 3(a). Three 53BP1 domains bind to p53. These interface residue pair distances suggest that the most stable interaction is α3A and α4A of 53BP1 with the H1 helix and L3 loop of p53, respectively. The only exception is that the distance of residue pair R248-L1847 increases after 9ns. The least stable interaction is at p53 L2 with the 53bp1 linker region. There are two contacts in this region, Gln165-D1861 and Gln167-Q1863. The minimum distance of both contacts fluctuated and the contact is lost after 8 ns simulation.

Figure 3

Variations of distances between interface residue pairs from (a) 53BP1-p53 and (b) BRCA1-p53 complexes. The top plots are residue pairs between p53 and 53BP1/BRCA1 α4 region residues. The bottom plots are residues pairs between p53 and 53BP1/BRCA1 (more ...)

The interacting residues distances of BRCA1-p53 complex along time are shown in Figure 3(b). All residue pairs in α3A fall apart from p53 after 8ns, if not earlier, as shown at the top. The interaction of BRCA1 α3A with p53 H1 helix is lost after 2ns, due to the K1724-R181 charge repulsion. The BRCA1 linker-p53 L2 loop contact region fluctuated during the 10 ns due to the flexibility of the linker. The most stable residue pair, M243-F1695 in p53 L3 loop and BRCA1 LA loop, does not exist in 53BP1. In short, all three contact regions between 53BP1 and p53 lose contact with p53 in BRCA1 after 10 ns simulation, and the only region that retains contact is the LA loop region. By comparing snapshots at 0 ns and 10 ns (Figure 4, top), it appears that the linker region has dramatically moved away from p53 L3 loop toward p53 H2 helix. There are two possible reasons: the charge repulsion between BRCA1 Arg1744 in the linker region with p53 Arg248 in the L3 loop and the charge attraction between BRCA1 Arg1744 with p53 Glu285. The distance between these two residue pairs is shown in Figure 3(b). The Arg248-Arg1744 distance increased sharply after 5 ns, while the Glu285-Arg1744 distance started to decrease after 5 ns. It is conceivable therefore that the charge repulsion of p53 L3 loop drives the flexible linker loop toward the charge attraction of the H2 helix.

Figure 4

Snapshots of the BRCA1-p53 interface from the wild type, F1695L and D1733G at 0 ns and 10 ns. The top panel is the wild type BRCA1-p53 interface, where the R1744 at the linker region drifts away from crowded LA loop region from 0 ns to 10 ns. The middle (more ...)

The other possible reason for the loss of the contact of BRCA1 linker with the p53 L2 loop can be attributed to the bulky LA loop. The longer LA loop, including a phenylalanine residue, of BRCA1 makes the contact region more crowded blocking the rotation of the charged residue in the p53 L3 loop to avoid the positive charge rich region of the BRCA1 linker, driving away the flexible linker region.

BRCA1-p53 Mutations

As discussed in the previous section, we attribute the weaker interaction of BRCA1 with p53 to the electrostatic repulsion between α3A, α4A as well as the linker region of BRCA1 and p53, the steric hindrance of BRCA1 LA loop, and the flexible loop of the linker of tandem BRCT. To investigate whether these electrostatic and steric factors can be modulated allowing the formation of a stable complex, thirteen simulations of mutants in α3A, α4A, LA and in the linker region have been performed. Among these thirteen mutants, mutations of 1724, 1733, 1737 and 1744 were designed to investigate the effect of electrostatic repulsion. Mutations of 1694 and 1695 are on the LA loop, which provides a unique contact region of BRCA1 with p53 comparing to 53BP1, are designed for probing the steric hindrance of the BRCA1 LA loop. A mutant with two mutations at 1737 and at 1744 was also designed to compare with the single point mutations at each of these sites. Four of these thirteen mutants, F1695L, N1730S, D1733G, and F1734S are carcinogenic; these are the only known carcinogenic mutations of BRCA1 which are in contact with residues with p53 in our modeled complex. Interestingly, F1695 and D1733 are not conserved between BRCA1 and 53BP1, whilst N1730 and F1734 are conserved. The results of simulations are shown in Figures 5 and and66.

Figure 5

Cα RMSDs of interface for BRCA1 mutants interacting with p53. The top plots are mutants of the LA loop (left) and α3 region (right). The middle plots are mutants of α4 region. The bottom plots are mutants of linker region.

Figure 6

Cα RMSFs relative to the average structures of the last 5 ns of the trajectories for the BRCA1 mutants. The top plots are mutants of LA loop (left) and α3 region (right). The middle plots are mutants of α4 region. The bottom plots (more ...)

In α3A, there is an electrostatic repulsion between positively charged residues K1724 and R181 of p53. Two mutants, K1724E and K1724H, were simulated. K1724E is expected to stabilize the interface. Since the corresponding residue in 53BP1 is histidine, K1724H mutant was also simulated. Figure 5 (top, right) shows that K1724E stabilizes the interface, whilst K1724H has a slightly smaller RMSD value along time than the wild type BRCA1 in the first 5 ns, but after 5 ns the RMSD of K1724 increases and has no significant difference from that of wild type BRCA1. The RMSF plot (Figure 6) shows that the residue fluctuation at the interface of K1724E significantly decreases, especially in the linker region. This again is not the case for K1724H. This result is consistent with our expectation that eliminating positive charge repulsion in α3A region can stabilize the BRCA1-p53 interface.

At the α4A region, the complexes of D1733R and D1733G mutants with wild type p53 were simulated. Unsurprisingly, both mutants stabilize the interface, as shown in Figure 5 (middle, left). The negatively charged Asp1733 not only attracts p53 R249, it also attracts BRCA1 R1737 from the linker region. Thus, the charge repulsion of R249–R1737 is one of the causes that induce the fluctuation at the linker region. Replacing the negative aspartic acid by neutral glycine or positively charged arginine reduces the fluctuation of the linker region, as shown in Figure 6.

Simulations of mutants N1730S and F1734S were also performed in the α4A region. Both N1730S and F1734S are carcinogenic mutations; however, N1730 and F1734 are conserved with 53BP1. F1734 interacts with p53 M243; thus, unsurprisingly F1734S does not stabilize the interface. N1730S has slightly smaller RMSD value along time than the wild type BRCA1.

Mutants R1737P and R1737E replace arginine by neutral or negatively charged residues to avoid charge conflict with p53 Arg248. R1744A and R1744E also mutate arginine on the linker region to avoid interaction with E1694 on the LA loop. Interestingly, Figure 5 (bottom, right) shows that R1737P and R1737E have smaller RMSD values than the wild type BRCA1 but the RMSD value of R1744A is close to the wild type. In addition, the RMSF plot suggested that both R1737E and R1737P decrease the fluctuation at the linker region but R1744A increases it, and that R1744E decreases the fluctuation at the linker region but increases the fluctuation in the LA loop. This is not surprising and could be explained by the elimination of charge repulsion of R1737 of BRCA1 with p53 R248, thus stabilizing the linker region, however, losing the electrostatic attraction of R1744 and E1694 destabilizes the system even more. The electrostatic repulsion destabilizes the LA loop. In order to further probe the mutational effects, we performed simulations for the double point mutant R1737E_R1744E and compared the results with the simulations of the corresponding single point mutations. Both RMSD and RMSF results suggested that the difference between single point mutations and double point mutations is not significant.

The contact of the LA loop with p53 is unique to BRCA1 whereas the corresponding loop region of 53BP1 is not long enough for interaction. The BRCA1 LA loop forms the Arg248-Glu1694 and Met243-Phe1695 interactions with the L3 loop of p53. However, the existence of these new contacts destabilizes the contact region of BRCA1 with p53 because Glu1694 on the LA loop not only forms a contact with Arg248 on the p53 L3 loop; it also attracts Arg1744 at the beginning of the BRCT linker. The repulsion between the two arginine residues causes a large conformational change at the interface, including twisting the LA loop and pushing the linker loop toward the direction of p53 H2 helix. The direct consequence of the conformational change is the loss of the contact between the BRCA1 linker with the p53 L2 loop, thus the interaction of p53 and BRCA1 BRCT is weakened.

To test whether the LA loop is responsible for the flexibility of BRCA1-p53 interface, two mutations E1694K and F1695L, have been performed in the LA loop. As shown in Figure 5, the E1694K avoided the two arginine residue situation, p53 R248 and BRCA1 R1744, stabilizing the linker loop during the first 6 ns, however, the repulsion of the lysine and linker region arginine R1744 finally causes structural fluctuation of the linker region, as shown in the RMSF plot (Figure 6), increasing the RMSD of E1694K significantly after 6 ns. Although F1695L is not directly connected to the interaction at the linker region, leucine is less likely to cause steric hindrance than the bulky phenylalanine. Arg248 can rotate to form Arg-Asn hydrogen bond instead of interacting with Glu1694. Therefore, repulsion of Arg248-Arg1744 is again avoided and the linker is stabilized.

The binding energy differences, and average RMSD of the complex interface were calculated for each mutant, and are listed in Table 1. The more negative the ΔΔG, the stronger the BRCA1-p53 binding it implies; and, smaller values of the RMSD of the complex interface suggest more stable interface. Comparing to wild type BRCA1, mutants F1695L and D1733G have the most negative ΔΔG, suggesting stronger binding with p53 than the wild type BRCA1 in this orientation. In addition, these two mutants have the smallest average RMSD values. Interestingly, these two mutants are the only cancer-related mutants among all the mutants simulated according to the National Human Genome Research Institute (NHGRI) breast cancer mutation database. Considering that the binding energy of F1695L mutants with p53 is more negative and the average RMSD value is smaller than D1733G, we predict that among these thirteen mutants, F1695L could be the mutant candidate to most stabilize the interface of BRCA1-p53 complex in the conformation similar to the 53BP1-p53 complex.

Table 1

Comparison of the binding energy and interface average Cα RMSD for BRCA1 mutants bound to p53. The mutants with the most negative binding energy are bolded.

Discussion

BRCA1 has been observed to associate with p53 in various cell functions. However, the interaction between the BRCA1 BRCT domain and the p53 core domain has been suggested to be weak and the structure of the complex has yet to be solved. On the other hand, the complex structure of 53BP1, which has remarkable similarity to the BRCT tandem repeat in BRCA1, has been solved in complex with p53. Given the structural conservation, it is plausible that should the BRCA1 interact with p53, the interaction would mimic that of 53BP1. However, our simulations suggest that an interaction between BRCA1 BRCT domain and the p53 core domain in a binding mode similar to that of the 53BP1 is not favorable for BRCA1. A potential reason why this interface is disfavored by nature is that this binding mode blocks the interaction of p53 with DNA. Alternatively, the interaction between BRCA1 and p53 may take place in a different binding mode. One of the possible binding modes of p53 to BRCA1 is to the second BRCT domain (Chai, et al.,1999). In the literature, there are two experiments with different results related to p53 binding. The first observed p53 binding using biological gel shift with the second repeat only (Chai, et al.,1999); on the other hand, the second biophysical analysis observed no binding using two repeats (Ekblad, et al.,2004). Thus it is possible that functional BRCA1 has a large conformational change to expose the second repeat binding site. On the p53 side, it is also possible that binding occurs on the phosphorylated site, rather than in the DNA binding region. We are investigating these potential alternative binding modes.

If the BRCA1-BRCT domain interacts with p53 in a different mode, it has to be substantially stronger than the 53BP1-p53 binding mode to shift the equilibrium toward this mode and avoid a “similarity trap”. However, we can not exclude the possibility that BRCA1 BRCT domain has no interaction with p53 core domain at all. In both cases, the non-conserved cancer-related mutations appear to lead BRCA1 and p53 to bind in a 53BP1-p53 binding mode, blocking the binding of p53 to the DNA and potentially causing cancer development. Remarkably, N1730S and F1734S are also carcinogenic mutants; however, unlike F1695 and D1733, both N1730 and F1734 are residues conserved between BRCA1 and 53BP1. Interestingly, our results indicate that the N1730S and F1734S mutations do not shift the equilibrium toward a 53BP1-p53 binding mode. This observation is in agreement with statistical observations of sequence variation in ligand binding site (Magliery and Regan,2005). Sequence analysis of four protein families has indicated that it is the variant rather than the conserved residues which may determine the binding sites. Thus, a “similarity trap” mechanism could be carcinogenic; but not every carcinogenic mutant necessarily follows such a dysfunctional mechanism.

To our knowledge, 53BP1 is not reported to interact with p53 in other domains. On the other hand, the BRCA1 central region has been reported to interact with p53 C-terminal domain (Mark, et al.,2005), which could hinder the 53BP1-like binding of the BRCT with carcinogenic mutations, competing with it. However, it is not clear how strong such a binding of the BRCA1 central region with the p53 C-terminal domain is.

F1695L is the mutant that stabilizes the complex the most among these thirteen mutants probably since phenylalanine prevents the movement of Arg248. If phenylalanine is replaced by a hydrophobic yet more flexible residue, it allows the Arg248 side chain to rotate away from Arg1744 thus preventing the drifting of the linker residues, retaining the contact of the linker with the L3 loop. F1695L is breast cancer related mutation. Thus the prediction that the F1695L mutation stabilizes the BRCA1-p53 complex may provide some insight into the association and its role in breast cancer. D1733G, another cancer associated mutant, also stabilizes the complex interface to some degree because it avoids attracting a positively charged residue from the BRCA1 loop region thus preventing accumulation of positive charge around the p53 L3 loop. Taken together, by stabilizing the unwanted interactions, the p53 binds BRCA1 in a way similar to the p53-53bp1 binding pattern, thus falling into a “similarity trap”. That is, the two carcinogenic mutants strongly disturbed the native functions.

Apart from the proposed mechanism, the F1695L mutation may also disrupt another important binding with BACH1 (Joo, et al.,2002). The BRCA1 BRCT domain has recently been identified as phospho-protein binding domain with evidence of its interaction with phosphorylated DNA helicase BACH1 as well as CtBP interacting protein (CtIP) (Clapperton, et al.,2004, Manke, et al.,2003, Varma, et al.,2005, Williams, et al.,2004, Yu, et al.,2003). The interaction between the BRCA1 BRCT domain and BACH1 has been implicated to be cell cycle regulated and is required for G2/M cell-cycle checkpoint control in response to DNA damage (Manke, et al.,2003, Yu, et al.,2003).

The BRCA1 BRCT domain interacts with the phosphorylated DNA helicase BACH1 (Manke, et al.,2003, Yu, et al.,2003). The interaction sites are different from the interaction sites we explored for BRCA1-p53 interaction. It has been found that the F1695L mutation can reduce the interaction between BRCA1 BRCT with BACH1. Based on this discovery, Joo et al. initially hypothesized that the LA loop is one of the interaction regions with BACH1 (Joo, et al.,2002), but the solved structure of BRCA1-BACH1 complex shows that the LA loop is not involved in this interaction, lending support to our proposed mechanism. Our results imply a possible explanation to the observation that BRCT-p53 interaction may regulate the BRCT-BACH1 interactions. The BRCT domain can not bind to both p53 core domain at this binding site and phosphoprotein at the same time. BRCA1-BACH1 interaction is only detected in S and G2/M phase whereas BRCA1-p53 interaction is suggested to take place in G1 phase. It is possible that BRCA1 initially participates in stabilization and accumulation of p53 protein by forming transient complex with p53, and then is free to interact with BACH1. With the cancer-related mutation F1695L, the unregulated “similarity trapped” BRCA1-p53 binding prevents the BRCA1 interaction with BACH1.

Conclusions

Our molecular dynamics simulations suggest that the interaction of BRCA1 BRCT domain with p53 core domain is weaker than that of 53BP1 in a similar binding mode despite their conserved structures. The weaker interaction of BRCA1-p53 is attributed to the linker region structure, the longer loop at the interface, as well as the electrostatic repulsion at the BRCA1-p53 interface. A total of thirteen BRCA1-BRCT domain mutants were designed and the stabilities of their interactions with p53 and structural effects were tested using MD simulation. Of the thirteen, four mutants are carcinogenic; these are the only known carcinogenic mutations of BRCA1 in contact with p53 when modeled in a 53BP1-like binding mode. Two of these four, N1730 and F1734 are conserved between BRCA1 and 53BP1, and two F1695 and D1733 are not conserved. Interestingly, it is the two non-conserved carcinogenic mutants, F1695L and D1733G, which enforced the binding of p53 core domain with BRCA1-BRCT domain in a way similar to that with 53BP1 protein. Therefore, we propose that the known binding pattern in p53-53BP1 interaction is a “similarity trap” that nature has to avoid in the p53-BRCA1 interactions. Falling into the “similarity trap” might disrupt normal functions of both BRCA1 and p53 and is carcinogenic. Here we have analyzed the respective interfaces to obtain an insight into such a trap that is avoided in the native functional protein state.

Acknowledgments

We thank Drs. C.-J. Tsai, K. Gunasekaran, H. Jang and J. Zheng for helpful discussions and suggestions. This project has been funded in whole or in part with Federal funds from the National Cancer Institute, National Institutes of Health, under contract number NO1-CO-12400. Computations performed in this project were conducted using Biowulf cluster at the National Institutes of Health, Bethesda, MD (http://biowulf.nih.gov). The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the US Government. This research was supported (in part) by the Intramural Research Program of the NIH, National Cancer Institute, Center for Cancer Research.

Footnotes

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

Aloy P, Ceulemans H, Stark A, Russell RB. The relationship between sequence and interaction divergence in proteins. J Mol Biol. 2003;332:989–998. [PubMed]
Aloy P, Bottcher B, Ceulemans H, Leutwein C, Mellwig C, Fischer S, Gavin AC, Bork P, Superti-Furga G, Serrano L, Russell RB. Structure-based assembly of protein complexes in yeast. Science. 2004;303:2026–2029. [PubMed]
Brooks BR, Bruccoleri RE, Olafson BD, States DJ, Swaminathan S, Karplus M. CHARMM: Aprogram for macromolecular energy, minimization, and dynamics calculations. J Comput Chem. 1983;4:187–217.
Chai YL, Cui J, Shao N, Shyam E, Reddy P, Rao VN. The second BRCT domain of BRCA1 proteins interacts with p53 and stimulates transcription from the p21WAF1/CIP1 promoter. Oncogene. 1999;18:263–268. [PubMed]
Clapperton JA, Manke IA, Lowery DM, Ho T, Haire LF, Yaffe MB, Smerdon SJ. Structure and mechanism of BRCA1 BRCT domain recognition of phosphorylated BACH1 with implications for cancer. Nat Struct Mol Biol. 2004;11:512–518. [PubMed]
Deng M, Mehta S, Sun F, Chen T. Inferring domain-domain interactions from protein-protein interactions. Genome Res. 2002;12:1540–1548. [PubMed]
Derbyshire DJ, Basu BP, Serpell LC, Joo WS, Date T, Iwabuchi K, Doherty AJ. Crystal structure of human 53BP1 BRCT domains bound to p53 tumour suppressor. Embo J. 2002;21:3863–3872. [PubMed]
Ekblad CM, Friedler A, Veprintsev D, Weinberg RL, Itzhaki LS. Comparison of BRCT domains of BRCA1 and 53BP1: a biophysical analysis. Protein Sci. 2004;13:617–625. [PubMed]
Futreal PA, Liu Q, Shattuck-Eidens D, Cochran C, Harshman K, Tavtigian S, Bennett LM, Haugen-Strano A, Swensen J, Miki Y, et al. BRCA1 mutations in primary breast and ovarian carcinomas. Science. 1994;266:120–122. [PubMed]
Gaiser OJ, Ball LJ, Schmieder P, Leitner D, Strauss H, Wahl M, Kuhne R, Oschkinat H, Heinemann U. Solution structure, backbone dynamics, and association behavior of the C-terminal BRCT domain from the breast cancer-associated protein BRCA1. Biochemistry. 2004;43:15983–15995. [PubMed]
Glover JN, Williams RS, Lee MS. Interactions between BRCT repeats and phosphoproteins: tangled up in two. Trends Biochem Sci. 2004;29:579–585. [PubMed]
Hainaut P, Hollstein M. p53 and human cancer: the first ten thousand mutations. Adv Cancer Res. 2000;77:81–137. [PubMed]
Hartman AR, Ford JM. BRCA1 and p53: compensatory roles in DNA repair. J Mol Med. 2003;81:700–707. [PubMed]
Inbar Y, Benyamini H, Nussinov R, Wolfson HJ. Protein structure prediction via combinatorial assembly of sub-structural units. Bioinformatics . 2003;19(Suppl 1):i158–168. [PubMed]
Iwabuchi K, Bartel PL, Li B, Marraccino R, Fields S. Two cellular proteins that bind to wild-type but not mutant p53. Proc Natl Acad Sci U S A. 1994;91:6098–6102. [PubMed]
Iwabuchi K, Li B, Massa HF, Trask BJ, Date T, Fields S. Stimulation of p53-mediated transcriptional activation by the p53-binding proteins, 53BP1 and 53BP2. J Biol Chem. 1998;273:26061–26068. [PubMed]
Joo WS, Jeffrey PD, Cantor SB, Finnin MS, Livingston DM, Pavletich NP. Structure of the 53BP1 BRCT region bound to p53 and its comparison to the Brca1 BRCT structure. Genes Dev. 2002;16:583–593. [PubMed]
Koonin EV, Altschul SF, Bork P. BRCA1 protein products … Functional motifs. Nat Genet. 1996;13:266–268. [PubMed]
Lee MS, Feig M, Salsbury FR, Jr, Brooks CL., 3rd New analytic approximation to the standard molecular volume definition and its application to generalized Born calculations. J Comput Chem. 2003;24:1348–1356. [PubMed]
Ludwig T, Fisher P, Ganesan S, Efstratiadis A. Tumorigenesis in mice carrying a truncating Brca1 mutation. Genes Dev. 2001;15:1188–1193. [PubMed]
Ma B, Elkayam T, Wolfson H, Nussinov R. Protein-protein interactions: structurally conserved residues distinguish between binding sites and exposed protein surfaces. Proc Natl Acad Sci U S A. 2003;100:5772–5777. [PubMed]
MacKerell AD, Jr, Bashford D, Bellott RL, Dunbrack RL, Jr, Evanseck JD, Field MJ, Fischer S, Gao J, Guo H, Ha S, Joseph-McCarthy D, Kuchnir L, Kuczera K, Lau FTK, Mattos C, Michnick S, Ngo T, Nguyen DT, Prodhom B, Reiher WE, III, Roux B, Schlenkrich M, Smith JC, Stote R, Straub J, Watanabe M, Wiorkiewicz-Kuczera J, Yin D, Karplus M. All-Atom Empirical Potential for Molecular Modeling and Dynamics Studies of Proteins. J Phys Chem. 1998;B102:3586–3616.
MacLachlan TK, Takimoto R, El-Deiry WS. BRCA1 directs a selective p53-dependent transcriptional response towards growth arrest and DNA repair targets. Mol Cell Biol. 2002;22:4280–4292. [PMC free article] [PubMed]
Magliery TJ, Regan L. Sequence variation in ligand binding sites in proteins. BMC Bioinformatics. 2005;6:240. [PMC free article] [PubMed]
Manke IA, Lowery DM, Nguyen A, Yaffe MB. BRCT repeats as phosphopeptide-binding modules involved in protein targeting. Science. 2003;302:636–639. [PubMed]
Marcotte EM, Pellegrini M, Ng HL, Rice DW, Yeates TO, Eisenberg D. Detecting protein function and protein-protein interactions from genome sequences. Science. 1999;285:751–753. [PubMed]
Mark WY, Liao JC, Lu Y, Ayed A, Laister R, Szymczyna B, Chakrabartty A, Arrowsmith CH. Characterization of segments from the central region of BRCA1: an intrinsically disordered scaffold for multiple protein-protein and protein-DNA interactions? J Mol Biol. 2005;345:275–287. [PubMed]
Miki Y, Swensen J, Shattuck-Eidens D, Futreal PA, Harshman K, Tavtigian S, Liu Q, Cochran C, Bennett LM, Ding W, et al. A strong candidate for the breast and ovarian cancer susceptibility gene BRCA1. Science. 1994;266:66–71. [PubMed]
Navaraj A, Mori T, El-Deiry WS. Cooperation between BRCA1 and p53 in repair of cyclobutane pyrimidine dimers. Cancer Biol Ther. 2005;4:1409–1414. [PubMed]
Sibanda BL, Critchlow SE, Begun J, Pei XY, Jackson SP, Blundell TL, Pellegrini L. Crystal structure of an Xrcc4-DNA ligase IV complex. Nat Struct Biol. 2001;8:1015–1019. [PubMed]
Varma AK, Brown RS, Birrane G, Ladias JA. Structural basis for cell cycle checkpoint control by the BRCA1-CtIP complex. Biochemistry. 2005;44:10941–10946. [PubMed]
Williams RS, Green R, Glover JN. Crystal structure of the BRCT repeat region from the breast cancer-associated protein BRCA1. Nat Struct Biol. 2001;8:838–842. [PubMed]
Williams RS, Lee MS, Hau DD, Glover JN. Structural basis of phosphopeptide recognition by the BRCT domain of BRCA1. Nat Struct Mol Biol. 2004;11:519–525. [PubMed]
Yu X, Chini CC, He M, Mer G, Chen J. The BRCT domain is a phospho-protein binding domain. Science. 2003;302:639–642. [PubMed]
Zhang H, Somasundaram K, Peng Y, Tian H, Zhang H, Bi D, Weber BL, El-Deiry WS. BRCA1 physically associates with p53 and stimulates its transcriptional activity. Oncogene. 1998;16:1713–1721. [PubMed]