Structure of the 53BP1 BRCT region and the interface between the tandemBRCTs

The individual BRCT repeats of 53BP1 have the same overall α/β fold observed in the structure of a single BRCT domain of XRCC1 (Zhang et al. 1998) but contain several sequence insertions as well as deviations in the positions of secondary structure elements (Fig. ​(Fig.1A).1A). The BRCT repeats of 53BP1 consist of a four-stranded parallel β-sheet (strands β1A to β4A and β1B to β4B; the suffixes A and B indicate the first and second 53BP1 repeats, respectively) flanked by two α-helices on one side (α1A, α1B and α3A, α3B) and one on the other (α2A, α2B; Fig. ​Fig.1A).1A). The C-terminal portion of the first BRCT repeat of 53BP1 contains a 3₁₀ helix near its C terminus (α4A in residues 1843–1847; Fig. ​Fig.11A,C).

The positions and orientations of individual helices in the 53BP1 BRCT repeats show considerable deviations from a canonical structure, as defined by the superposition of the BRCT structures from 53BP1, Brca1, and XRCC1 (Fig. ​(Fig.2A,C).2A,C). The most divergent structural elements are the α2A and α1B helices. Relative to the canonical structure, α2A is shifted by ~6 Å along its helical axis and tilted by ~30°, and α1B is tilted by ~11°. These helices pack in the inter-BRCT repeat interface (Figs. ​(Figs.1A,1A, ​A,2B)2B) and their deviation from the canonical structure is critical for the packing. The loops between the secondary structure elements also vary in size and precise structure (Figs. ​(Figs.1C,1C, ​C,2A).2A). The loop between β2 and β3, termed the c2 loop (Zhang et al. 1998), contains a five-residue helical insertion (α1‘A) in repeat 1 of 53BP1 and an eight-residue insertion in repeat 2 (Figs. ​(Figs.1C,1C, ​C,2A).2A).

Open in a separate window

Open in a separate window

Figure 2

The rat Brca1 BRCT domains have the same tandem arrangement as in 53BP1. (A) Superimposition of the two BRCT repeats (red and green) of 53BP1 and the single BRCT domain from XRCC1 (yellow). Divergent regions among the BRCT domains are located within loops (e.g., the insertion helix α1‘A in 53BP1) and also include the orientations and positions of the α1 and α2 helices. (B) Close-up view of the inter-BRCT repeat interface of 53BP1. Side chains from the first BRCT repeat are in light-red, those from the linker are in light-yellow, and those from the second are in light-green. The α2A helix of the first repeat (red) together with α1B and α3B helices of the second repeat (green) form a three-helix bundle that is stabilized by a β-hairpin linker (β1L and β2L, colored yellow) and an insertion helix α1‘A. White dotted lines indicate hydrogen bond interactions. (C) Superimposition of the tandem-BRCT repeats of 53BP1 (gray) and rat Brca1, with the first BRCT repeat colored magenta, the second BRCT repeat cyan, and the linker yellow. (D) Close-up view of the inter-BRCT repeat interface of rat Brca1. Side chains from the first repeat are in light-pink, those from the linker are in light-yellow, and those from the second repeat are in light-blue. White dotted lines indicate hydrogen bond interactions. The letter M indicates the tumor-derived missense mutations that occur at a high frequency (Met 1598, Arg 1645, Ala 1654, Met 1720, and Gly 1733 in rat).

The inter-BRCT repeat interface of 53BP1 contains both core BRCT motif elements and the inter-repeat linker. The interface buries a total of ~2400 Ų, with one third of it being contributed by the linker (Fig. ​(Fig.2B).2B). The central part of the interface forms through the packing of α2A of the first repeat with α1B and α3B of the second repeat, in a manner analogous to three-helix bundles (Fig. ​(Fig.2B).2B). The three α-helices contribute eight hydrophobic residues to the interface (Fig. ​(Fig.2B).2B). This three-helix bundle packing arrangement would not be possible if the positions and orientations of the α2A and α1B helices did not deviate from the canonical BRCT motif structure (Fig. ​(Fig.2A,B).2A,B). The hydrophobic interface contacts are flanked by charge-stabilized hydrogen bond networks near the ends of the helices (Fig. ​(Fig.22B).

The inter-repeat linker adopts a β-hairpin structure (residues 1851–1860; β1L and β2L, where L indicates a linker element) followed by an extended segment (residues 1861–1867), both of which pack extensively with both BRCT repeats at the inter-BRCT repeat interface (Figs. ​(Figs.1A,1A, ​A,2B).2B). One face of the hairpin contains hydrophobic residues that pack with residues from the α2A and α3B helices of the three-helix bundle (Fig. ​(Fig.22B).

A comparison of human and X. laevis 53BP1 sequences shows that many of the hydrophobic interface residues and all five amino acids that participate in salt bridges on the three-helix bundle and the linker β-hairpin are conserved (Figs. ​(Figs.1C,1C, ​C,2B),2B), consistent with a model that the two BRCT-repeat region is a single globular domain. In contrast, the single BRCT repeat structure of XRCC1 differs from the portions of the 53BP1 repeats that form the interface in that (1) it does not show any deviations from a canonical structure in its α1 and α2 helices, and (2) it has only one exposed hydrophobic amino acid (Met 62) in its α1, α2, and α3 helices.

Comparison of 53BP1 and rat Brca1 tandem-BRCTstructures

The entire tandem-BRCT structures of 53BP1 and rat Brca1 can be superimposed with a root mean square deviation (RMSD) of 1.71 Å in the positions of 138 out of 229 Brca1 Cα atoms (Fig. ​(Fig.2C),2C), including all eight BRCT core β strands and seven of the eight core helices (the Brca1 α2B helix is disordered in our crystals; Figs. ​Figs.1B,1B, ​B,2C).2C). The most striking structural conservation is at the inter-BRCT repeat interface (Fig. ​(Fig.2C).2C). The interface helices α2A of the first repeat and α1B and α3B of the second repeat have the same angular and positional deviations from the consensus structure as those of 53BP1, allowing them to form a three-helix bundle interface essentially identical to that of 53BP1 (Fig. ​(Fig.2C).2C). The central interactions in the three-helix bundle interface of rat Brca1 involve hydrophobic residues that occur at positions on the helices equivalent to those in 53BP1 (Fig. ​(Fig.2B,D).2B,D). Three of them are conserved between Brca1 and 53BP1 (Phe 1650, Leu 1651, and Ala 1654 of rat Brca1) and the rest are conservatively substituted (Figs. ​(Figs.1C,1C, ​C,2B,D).2B,D). Just as in the 53BP1 interface, the Brca1 three-helix bundle is flanked by charge-stabilized hydrogen bond interactions near the ends of the helices (Fig. ​(Fig.2B,D).2B,D). These hydrogen bond contacts (Arg 1645, Asp 1785, Arg 1727, and the backbone carbonyl of Ile 1626) occur at similar positions compared to those in 53BP1, but the detailed interactions differ somewhat due to insertions in the vicinity and differences in the linker region (Fig. ​(Fig.22B,D).

The 23-residue inter-repeat linker of rat Brca1 (residues 1682–1704) occupies the same position as that of 53BP1, packing on one side of the three-helix bundle and stabilizing the inter-BRCT repeat interface (Fig. ​(Fig.2B,D).2B,D). Like 53BP1, the Brca1 linker contains a β-hairpin-like structure that packs with the α2A and α3B helices, although only the first segment has β-sheet backbone phi–psi angles (β1L; Fig. ​Fig.2B,D).2B,D). The linker hairpin forms a continuous β-sheet-like structure with β3′A–β3^″ A (Fig. ​(Fig.2D),2D), which corresponds to a seven-residue insertion between β3A and α2A of Brca1 compared to 53BP1 (Fig. ​(Fig.1C).1C). The presence of the β3′A–β3"A insertion in Brca1 is associated with a rotation of the linker hairpin relative to that of 53BP1. Following the linker hairpin, Brca1 has a helix (α1L) compared to the extended segment of 53BP1 (Fig. ​(Fig.22B,D).

The structural conservation in the inter-BRCT repeat interfaces of 53BP1 and Brca1 suggests that their tandem-BRCT regions diverged from a common ancestral protein. Although it is possible that the two tandem-BRCT domains emerged separately through convergent evolution starting with a single ancestral BRCT domain, the similarities in the BRCT–BRCT contacts, such as those made by Arg 1645 of Brca1 and Arg 1811 of 53BP1 (Fig. ​(Fig.2B,D),2B,D), would make this less likely. The S. pombe Crb2 and S. cerevisiae Rad9 proteins have a pattern of hydrophobic and polar residues characteristic of the 53BP1 and Brca1 inter-BRCT repeat interfaces. Although the overall level of conservation is much lower compared to that between Brca1 and 53BP1, especially in the second BRCT repeats, the conservation in Crb2 and Rad9 of a few key interface residues, such as the arginines at positions corresponding to Arg 1645 of Brca1 and Arg 1811 of 53BP1, suggests that they may be evolutionarily related to Brca1 and 53BP1 and supports the hypothesis that they may share functional similarities.

53BP1–p53interactions

The 53BP1–p53 interactions are concentrated on the L3 loop of p53. The L3 loop has a rigid conformation that is stabilized by zinc binding, by backbone hydrogen bonds, and by side chain interactions with the rest of the p53 structure (Cho et al. 1994). The seven-residue tip (residues 243–249) of the L3 loop binds a contiguous region of 53BP1 consisting of the α3A and α4A helices at the C terminus of the first BRCT repeat and the inter-repeat linker (Fig. ​(Fig.3).3). The interactions rely primarily on networks of hydrogen bonds, involving both side chain and backbone groups, and to a lesser extent on van der Waals contacts (Fig. ​(Fig.3).3). The L3 loop is one of the two key structural elements that make up the DNA-binding surface of p53, containing an arginine residue (Arg 248) that binds in the minor groove of the DNA (Cho et al. 1994). The other structural element on the DNA-binding surface of p53 is the loop-sheet-helix motif (H2 helix), which is not involved in binding to 53BP1 (Figs. ​(Figs.1A,1A, ​A,3).3).

Open in a separate window

Figure 3

Stereo view of the 53BP1–p53 interface. The secondary structure of p53 is colored blue except for the L3 loop, which is cyan. The side chains of p53, first BRCT repeat, and the linker of 53BP1 are colored light-blue, light-red, and light-yellow, respectively. Hydrogen bonds are indicated by white dotted lines.

The 53BP1–p53 hydrogen bond networks are centered on Arg 248 and Arg 249 of p53. Arg 248 hydrogen bonds with the side chain of Asp 1861 from the β-hairpin of the inter-repeat linker and with the backbone carbonyl group of Leu 1847 at the start of the linker (Fig. ​(Fig.3).3). The Arg 248 side chain is stabilized by an intramolecular hydrogen bond with the OD1 group of Asn 247 (L3) whose ND2 group hydrogen bonds with the backbone carbonyl group of Leu 1847 from the α4A helix of 53BP1 (Fig. ​(Fig.3).3). The Arg 249-centered hydrogen bond network involves the OD1 group of Asn 1845 and the backbone carbonyl of Arg 1844, both from the α4A helix of 53BP1 (Fig. ​(Fig.3).3). In the structures of the p53–DNA complex and of the isolated p53, Arg 249 has an important role in stabilizing the structure of the L3 loop by hydrogen bonding with the backbone carbonyl groups of Gly 245 and Met 246 and with the side chain of Glu 171 from the L2 loop (Cho et al. 1994). In the 53BP1–p53 complex, the hydrogen bonds between Arg 249 and the backbone carbonyl groups are maintained, but the hydrogen bonds with Glu 171 are disrupted, with the NH1 and NH2 groups of Arg 249 now interacting with 53BP1 instead (Fig. ​(Fig.3).3). The Glu 171 side chain is displaced from its position in monomeric p53, and is disordered in the crystals.

Van der Waals contacts are made by the side chain of Met 243 and Cα of Gly 244, which wedge in between the α3A and α4A helices of 53BP1 and contact the side chains of Val 1829 and Tyr 1846 (Fig. ​(Fig.3).3). These p53 residues pack closely with 53BP1, allowing the backbone amide of Gly 244 of p53 to form a hydrogen bond with the backbone carbonyl group of Asn 1845 (Fig. ​(Fig.3).3). The L3 loop contacts are augmented by a few additional contacts from the adjacent L2 loop to α3A and the linker (Fig. ​(Fig.33).

The 53BP1 residues that contact p53 are either conserved (Val 1829, Tyr 1846, and Asn 1845) or are conservatively substituted (Asp 1861 to Glu) in X. laevis 53BP1 (Fig. ​(Fig.1C).1C). The residues that make up the α3A, α4A, and the β-hairpin part of the linker are also highly conserved (Fig. ​(Fig.1C),1C), indicating that these structural elements will adopt a very similar structure in X. laevis 53BP1. Relative to the overall ~68% identity between the X. laevis and human 53BP1 BRCT regions, the high conservation of these residues supports an important functional role for the p53-binding site of the 53BP1 structure.

Implications for the p53pathway

The L3 loop has a central role in p53’s tumor suppressor function because three of the six mutation hotspots (Gly 245, Arg 248, and Arg 249) and ~30% of the tumor-derived missense mutations map to this region (Cho et al. 1994). In addition, the L3 loop region corresponds to one of the five evolutionarily conserved regions of p53. Mutation of the Arg 248 hotspot, which makes key contacts to both DNA and 53BP1, disrupts both DNA binding and 53BP1 binding (Thukral et al. 1994). Similarly, mutations of Gly 245 and Arg 249, which have central roles in the structural integrity of the L3 loop (Cho et al. 1994), also disrupt both DNA and 53BP1 binding (Thukral et al. 1994). The involvement of these L3 loop residues in both DNA and 53BP1 binding makes it difficult to estimate the contribution, if any, that loss of 53BP1 binding may make to tumorigenesis. We note however, that Met 243, which is a key 53BP1 contact but has no role in DNA binding or stabilizing the structure of the L3 loop, is also found mutated in cancer although at a much lower frequency (six of 3393 missense mutations in the database), compared to the residues that contact DNA. Although another role for Met 243 cannot be ruled out, this observation suggests that loss of 53BP1 binding but not of DNA binding may contribute to the deregulation of the p53 DNA-damage response pathway in certain contexts.

The complex structure shows that 53BP1 binding would block the L3 loop from interacting with DNA (Figs. ​(Figs.1A,1A, ​A,3).3). This would interfere with sequence-specific DNA binding, due to the unavailability of Arg 248 for the recognition of the A/T-rich region of the consensus sequence (Cho et al. 1994). However, the H2 helix of p53 remains accessible in the 53BP1–p53 complex (Figs. ​(Figs.1A,1A, ​A,3).3). In principle, the 53BP1–p53 complex could still use the H2 helix to bind in the major groove of the DNA in a nonspecific manner if the DNA is bent away from the L3 loop or is a DNA end that terminates before the L3 loop region. In this hypothetical model of a 53BP1–p53–DNA complex, 53BP1 could provide additional contacts to the DNA.

Crystallization and datacollection

The 53BP1–p53 complex crystals were grown at 4°C using the hanging-drop vapor-diffusion method by mixing an equal volume of the purified protein complex and crystallization buffer containing 50 mM sodium citrate (pH 6.0), 125 mM ammonium acetate, 16% (w/v) polyethylene glycol (PEG) 4000, and 5 mM DTT. The best crystals were obtained in drops for which the reservoir solution lacked ammonium acetate. Diffraction data were collected at −170°C with crystals flash-frozen in the crystallization buffer containing 36% (w/v) PEG 4000 and lacking DTT. The crystals formed in space group P2₁2₁2₁ with a=73.1, b=95.0, c=133.7 Å and contained two complexes per asymmetric unit. Multi-wavelength anomalous diffraction (MAD) data were collected at the zinc edge at energies of 9662.5 electron Volts (eV), 9673.5 eV, and 9823.0 eV (corresponding to λ1, λ2, and λ3, respectively, in Table ​Table1)1) at the X4A beamline of the National Synchrotron Light Source at Brookhaven National Laboratory (NSLS). The rat Brca1 crystals were grown at 4°C using the hanging-drop vapor-diffusion method from a crystallization buffer containing 50 mM Tris-HCl (pH 8.5), 150 mM NaCl, 18% (w/v) PEG 8000, and 5 mM DTT. The crystals formed in space group C2 with a=84.5, b=58.7, c=65.4 Å, β=111.2° and contained one Brca1 molecule per asymmetric unit. For multiple isomorphous replacement (MIR) experiments, crystals were soaked in either 1 mM thimerosal or 2 mM K₂PtCl₄ for 2 h. Diffraction data were collected at −170°C with the crystals flash-frozen in the crystallization buffer containing 20% (w/v) PEG 8000 and 20% (w/v) PEG 400 and lacking DTT. The native data were collected at the A1 beamline of Cornell High Energy Synchrotron Source (CHESS) and the derivative data were collected at the X9A beamline of NSLS. All diffraction data were processed using the HKL suite (Otwinowski and Minor 1997).

Table 1

Summary of crystallographicanalysis

MAD data collection (53BP1–p53 complex)						MAD phasing statistics (15–2.7Å)
Wavelength (Å)	Resolution (Å)	Measured reflections	Unique reflections	Coverage (%)	Rsym (%)	Phasing power	Rcullis	Rcullis (ano)
λ1 (1.2832)	2.5(2.6–2.5)	169611	29872	89.4(56.5)	5.1(17.8)	—	—	0.84
λ2 (1.2817)	2.5(2.6–2.5)	170142	30000	89.4(56.4)	5.3(18.6)	0.71	0.88	0.79
λ3 (1.2622)	2.5(2.6–2.5)	172211	29847	89.5(55.9)	5.3(21.7)	0.88	0.84	0.83
Refinement statistics						RMSD
	Resolution range (Å)	Reflections (|F| > 0σ)	Atoms (total, water)	R-factor (%)	R-free (%)	bonds (Å)	angles (°)	B factor (Å2)
	15–2.5(2.6–2.5)	28192	6988,216	21.6(32.1)	25.6(39.5)	0.014	1.73	2.2
MIR data collection (rat Brca1)						MIR phasing statistics (15–3.4Å)
Crystal	Resolution (Å)	Measured reflections	Unique reflections	Coverage (%)	Rsym (%)	Phasing power	Rcullis
Native	2.1(2.18–2.10)	55198	17223	97.0(97.9)	8.2(45.7)	—	—
Thimerosal	3.7(3.87–3.70)	16590	3662	98.9(98.5)	14.8(22.8)	1.48	0.75
K2PtCl4	2.8(2.90–2.80)	24596	7387	99.1(99.0)	8.7(32.3)	0.93	0.85
Refinement statistics						RMSD
	Resolution range (Å)	Reflections (|F| > 0σ)	Atoms (total, water)	R-factor (%)	R-free (%)	bonds (Å)	angles (°)	B factor (Å2)
	15–2.3(2.4–2.3)	12529	1702,177	20.3(26.2)	25.8(34.1)	0.011	1.61	4.5

Open in a separate window

Coverage (%) indicates the percentage of total theoretically observable reflections measured. R_sym = Σ_hΣ_i |I_h,i − I_h|/Σ_hΣ_iI_h,i, where I_h is the mean intensity of the i observations of reflection h. Phasing power = [F_H(calc)²/(F_PH(obs) − F_PH(calc))²]^1/2, where F_PH(obs) and F_PH(calc) are the observed and calculated derivative structure factors, respectively. R_cullis is the mean residual lack-of-closure error divided by the dispersive or anomalous difference. Figure of Merit (FOM) = <Σ P(α) exp(iα)/ΣP(α)>, where P(α) is the probability distribution for the phase α. R = Σ |F_obs − F_calc|/ΣF_obs, where F_obs = F_P, and F_calc is the calculated protein structure factor from the atomic model. R-free is the R factor calculated using 5% of the reflection data chosen randomly and omitted from the refinement process, and R-factor is calculated with the remaining data in the refinement. The root means square deviation (RMSD) values in bond lengths and angles are deviations from ideal values, and the RMSD values in B factors are calculated between covalently bonded atoms. 

Structure determination andrefinement

For the 53BP1–p53 complex structure, initial phases were obtained from a molecular replacement solution of the p53 core domain structure using AMORE (CCP4 1994), and were used to locate the two zinc atoms in the asymmetric unit (one zinc per p53 molecule) using the MAD diffraction data. MAD phases, calculated with the program MLPHARE (CCP4), had a mean figure of merit (FOM) of 0.48 to 2.7 Å resolution, and were improved by density modification with DM (CCP4). Despite the presence of only one zinc atom per 474 amino acids, the experimental map was clearly interpretable for both p53 and 53BP1, including electron density for the majority of the side chains. The initial model of 53BP1 was built with the program O (Jones et al. 1991), and was refined by CNS (Brünger et al. 1998). The refinement statistics are shown in Table ​Table1.1. Nearly all residues (99.6%) are in the allowed regions of the Ramachandran plot. An N-terminal fragment (residues 1702–1713) and the region between β1A and α1A (residues 1741–1769) do not have interpretable electron density, and we presume are disordered. Residues 1741–1769 correspond to a 29 amino acid insertion that is not found in other BRCT repeats and is rich in polar and charged residues. For the BRCT region of rat Brca1, initial MIR phases were calculated with MLPHARE. They had a mean FOM of 0.42 to 3.4 Å, and were improved by solvent flattening and density modification using DM. The model was built with the program O and was refined by CNS. Five N-terminal residues (residues 1589–1593), a three-residue portion of the inter-repeat linker region (residues 1702–1704), and the C terminus (residues 1800–1817) do not have interpretable electron density, and we presume are disordered. In addition, residues 1757–1775, which form the α2B helix in the structure of the human BRCA1 BRCT repeats (Williams et al. 2001), are disordered in our crystals. The α2B helix is the least conserved structural element of the Brca1 BRCT repeats. The refinement statistics are shown in Table ​Table1.1. All residues are in the allowed regions of the Ramachandran plot.

We thank H. Erdjument-Bromage of the Sloan-Kettering Microchemistry Facility for N-terminal sequence and mass spectroscopic analyses, the members of the Pavletich lab and K. Lee for helpful discussions, C. Murray for administrative assistance, the staff of the National Synchrotron Light Source X4A and X9A beamlines and of the Cornell High Energy Synchrotron Source (CHESS) for help with data collection. We also thank M. Bennett and R. Wiseman for providing us with the cDNA for rat Brca1. This work was supported by the NIH, the Howard Hughes Medical Institute, the Dewitt Wallace Foundation, and the Samuel and May Rudin Foundation. Coordinates have been deposited in the RCSB Protein Data Bank. PDB accession numbers are 1KZY for 53BP1–p53 and 1L0B for rat Brac1.

The publication costs of this article were defrayed in part by payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 USC section 1734 solely to indicate this fact.