Structure of the RAD9-RAD1-HUS1 checkpoint clamp bound to RHINO sheds light on the other side of the DNA clamp

DNA clamp, a highly conserved ring-shaped protein, binds dsDNA within its central pore. Also, DNA clamp interacts with various nuclear proteins on its front, thereby stimulating their enzymatic activities and biological functions. It has been assumed that the DNA clamp is a functionally single-faced ring from bacteria to humans. Here, we report the crystal structure of the heterotrimeric RAD9-RAD1-HUS1 (9-1-1) checkpoint clamp bound to a peptide of RHINO, a recently identified cancer-related protein that interacts with 9-1-1 and promotes activation of the DNA damage checkpoint. This is the first structure of 9-1-1 bound to its partner. The structure reveals that RHINO is unexpectedly bound to the edge and around the back of the 9-1-1 ring through specific interactions with the RAD1 subunit of 9-1-1. Our finding indicates that 9-1-1 is a functionally double-faced DNA clamp.

DNA clamp proteins are widely conserved from bacteria to humans and play a crucial role in DNA metabolism, such as DNA replication and repair (1). Each DNA clamp is a ringshaped protein that encircles dsDNA within its central pore, enabling it to slide along the dsDNA. In addition, the DNA clamp tethers various proteins involved in DNA metabolism and stimulates their enzymatic activities or biological functions. Bacteria has a single DNA clamp (␤-clamp) comprising a homodimeric ring (2), whereas eukaryotes have two kinds of DNA clamp. The first type is proliferating cell nuclear antigen (PCNA), 2 a homotrimeric protein (3). Most proteins that associate with PCNA have a PIP-box motif (4,5); this motif is present in hundreds of proteins, and its binding to PCNA is wellelucidated (6,7). The PIP-box is bound to the hydrophobic pocket in PCNA, underneath the so-called "interdomain connecting" loop (IDC-loop) that bridges two domains of a single PCNA subunit. The other type of DNA clamp is the heterotrimeric RAD9-RAD1-HUS1 complex (9-1-1). Each subunit of 9-1-1 comprises of a PCNA-like domain, whereas RAD9 additionally has an unstructured C-terminal tail. 9-1-1 functions as a DNA damage-specific clamp that activates the ATR-CHK1 checkpoint pathway (8,9). Similar to PCNA, the core structure of 9-1-1 lacking the C-terminal tail of RAD9 is a ring-shaped trimer (10 -12). Binding of proteins to 9-1-1 is thought to be analogous to binding to PCNA (10,12); as yet, however, no structure of 9-1-1 bound to its partners has been determined, and thus the underlying mechanism remains unclear.
Owing to their ring-shaped structures, DNA clamp proteins inherently have two faces, a front and a back. To date, all of the partners of ␤-clamp and PCNA have been found to bind to the same face of the DNA clamp, the so-called C-side; the C termini of PCNA subunits protrude from this side (6). In this study, the C-side is defined as the "front." Regardless of species, therefore, it has been assumed that clamp-binding proteins bind to the front of the DNA clamp, and thus the DNA clamp is a functionally single-faced ring from bacteria to humans. Indeed, previous reports have suggested that 9-1-1 also utilizes its front to interact with its partners (10,12,13). Here, we report the structure of 9-1-1 bound to a peptide derived from RHINO (RAD9, HUS1, RAD1-interacting nuclear orphan protein 1, RHNO1), a recently identified cancer-related protein that interacts with 9-1-1 and promotes activation of CHK1 in response to DNA damage (14 -16). This is the first structure of 9-1-1 bound to its partner. Our results provide not only a structural basis for the mechanism of interaction between RHINO and 9-1-1, but also a new finding about molecular interactions on DNA clamp.

Results and discussion
The RHINO peptide is bound to the edge and around the back of 9-1-1 RHINO is known to form a stoichiometric complex with 9-1-1 through interactions with RAD9 lacking the C-terminal tail and RAD1 (16). Residues 55 SWVSPDF 61 of RHINO are crucial for this interaction with 9-1-1 and activation of the checkpoint (15). Thus, to elucidate details of the interaction between 9-1-1 and its binding proteins, we determined the X-ray crystal structure of the 9-1-1 core ring bound to a RHINO peptide    Table 1). Electron density map shows that residues 55-64 are well-ordered, whereas the N-terminal region (residues 45-54) is disordered (Fig. 1B). In the structure, the RHINO peptide adopts an extended conformation and is bound to the RAD1 subunit. Unexpectedly, the peptide is bound to the edge and around the back of the 9-1-1 ring (Fig. 1A, left). This interaction is distinct from that of known clamp-peptide complexes, such as PCNA in complex with a p21 peptide, where the p21 peptide including the PIP-box is bound to a binding pocket underneath the IDC-loop on the front of PCNA (17) (Fig. 1C).

The RHINO peptide specifically interacts with the RAD1 subunit
The RHINO peptide is accommodated in the pocket on the edge and around the back of the RAD1 subunit of 9-1-1, and the chemical properties of the bound peptide are complementary to the charge distribution of the binding pocket ( Fig. 2A). The hydrophobic side chains of Trp 56 , Val 57 , Pro 59 , and Phe 61 of RHINO interact significantly with the pocket of RAD1 through van der Waals (vdW) contacts, as follows (Fig. 2B). Trp 56 of RHINO contacts with the side chain of Phe 64 of RAD1 by a face-to-edge interaction. Val 57 , Pro 59 , and Phe 61 of RHINO contact with the hydrophobic side chains of Phe 64 , Phe 266 , and Met 256 of RAD1, respectively, by CH-interactions. The methyl group of Thr 63 of RHINO is also accommodated into the hydrophobic pocket. In addition to these vdW interactions, the acidic side chain of Asp 60 of RHINO forms ion pairs with the basic side chains of Lys 155 and Arg 244 of RAD1 (Fig. 2B), which contribute to the basic patch on the surface ( Fig. 2A).  ACCELERATED COMMUNICATION: The other side of 9-1-1 DNA clamp Furthermore, hydrogen bonds between the peptide bond of Pro 59 -Asp 60 of RHINO and the side chain of Gln 254 of RAD1 are also observed. The bidentate hydrogen bond by Gln 254 retains the conformation of the peptide bond, thereby probably stabilizing the conformation for binding. Furthermore, the residues of RHINO involved in interaction with RAD1 are highly conserved (Fig. 2C).
For functional validation of the present structure, we performed interaction analysis using a GST pulldown assay with triple mutations (m1 (K155A/R244A/Q254A) or m2 (F64A/ M256A/F266A)) in the RAD1 subunit (Fig. 3). First, we confirmed binding of RHINO(45-64) to 9-1-1 (Fig. 3, lane 7). The triple mutations in Lys 155 , Arg 244 , and Gln 254 , which are engaged in electrostatic interactions with the RHINO peptide ( Fig. 2B), reduced the binding (Fig. 3, lane 8). Furthermore, the triple mutations in Phe 64 , Met 256 , and Phe 266 , which are involved in vdW contacts with the RHINO peptide (Fig. 2B), largely reduced the binding (Fig. 3, lane 9). These results are in good agreement with the crystal structure, and thus we conclude that the RHINO peptide is specifically bound to the pocket of RAD1. In addition, the pulldown assay might indicate that vdW contacts are more crucial for the binding.
In contrast to PCNA, 9-1-1 is a heterotrimer, and thus the molecular surface of each subunit possesses different physicochemical properties, whereas RAD9, HUS1, and PCNA also possess large concaves that seem to be available for a binding site on their backs and edges. Thus, by structural superimposition of RAD9, HUS1, and PCNA with RAD1 bound to the ACCELERATED COMMUNICATION: The other side of 9-1-1 DNA clamp RHINO peptide, we built hypothetical structures of the RHINO peptide bound to the backs and edges of RAD9, HUS1, and PCNA (Fig. 4). These models suggest that the RHINO peptide is unlikely to be bound to these surfaces because of steric and electrostatic conflicts. In fact, it has been reported that HUS1 and PCNA show no binding to RHINO (16). Furthermore, the mutation of 55 SWVSPDF 61 to 55 AAAAAA 61 in RHINO abolishes its interaction with 9-1-1 (15), and RHINO forms a stoichiometric complex with 9-1-1 (16). These results strongly suggest that the interactions between RHINO and RAD1 observed in this study are dominant in formation of the 9-1-1-RHINO complex, although RHINO might possess a minor binding region to interact with RAD9. Our present finding is consistent with a functional study of RHINO mutants (15), thereby indicating that the interactions of RHINO with the back and edge of RAD1 observed here is essential for activation of the ATR-CHK1 checkpoint pathway by RHINO.

9-1-1 is a functionally double-faced DNA clamp
DNA clamps are loaded on dsDNA by specific clamp loaders, which are pentameric protein complexes classified into the AAAϩ ATPase family (18). Eukaryotes have four clamp loaders: RFC, CTF18-RFC, ELG1-RFC, and RAD17-RFC (19). RFC, the primary clamp loader for PCNA, consists of one large and four small subunits: RFC1 and RFC2-5, respectively. The others consist of the RFC1 paralog as a large subunit and CTF18, ELG1, and RAD17, associated with universal small subunits, RFC2-5. The crystal structure of yeast homolog of the RFC-PCNA complex has revealed that RFC also binds to the front of PCNA for ring opening and loading of PCNA on dsDNA (20). In contrast to PCNA, 9-1-1 is loaded on dsDNA by RAD17-RFC (21). Based on sequence similarity and crystal structures, plausible models of interaction between RFC17-RFC and 9-1-1 have been proposed (10,12,13). These models suggest that RAD17-RFC also interacts with the front of 9-1-1. A preliminary result also implies that FEN1 nuclease might be bound to the front of RAD1 (12). Here, our structure reveals that the RHINO specifically interacts with the edge and back of 9-1-1 for activation of the checkpoint. Therefore, 9-1-1 could be a functionally double-faced DNA clamp and thereby might enable multiple protein-protein interactions in signaling of the DNA damage checkpoint. Although RHINO is unlikely to interact with the edge and back of PCNA (Fig. 4), we cannot exclude the possibility that PCNA utilizes its edge or back for protein-protein interactions. Further studies are needed to discuss that possibility.

Conclusion
In this study, we have determined the crystal structure of 9-1-1 in complex with the RHINO peptide. Most surprisingly, the structure reveals that 9-1-1 utilizes its edge and back to interact with RHINO. Our unexpected finding first sheds light on the other side of 9-1-1, thereby indicating that 9-1-1 is a functionally double-faced DNA clamp and fundamentally changing our view of DNA clamps. The structure also reveals detailed interactions between 9-1-1 and RHINO and thus could provide a new clue for developing chemical compounds or artificial peptides to down-regulate the ATR-CHK1 checkpoint pathway for cancer therapy.

ACCELERATED COMMUNICATION:
The other side of 9-1-1 DNA clamp carried out at 4°C. The supernatant was applied to a HiTrap Heparin column (GE Healthcare), and bound protein was eluted with a linear gradient of 250 mM to 1 M NaCl. The eluent was diluted with buffer III (50 mM Tris-HCl, pH 8.5, and 200 mM NaCl) and applied to an anion-exchange column, HiTrap Q column (GE Healthcare). The bound protein was eluted with a linear gradient of 200 mM to 1 M NaCl. The eluted protein was passed through a size-exclusion column, HiLoad 16/600 Super-dex200 (GE Healthcare), equilibrated with buffer IV (20 mM HEPES-NaOH, pH 7.4, and 100 mM NaCl), and then concentrated to 15-20 mg/ml using a Vivaspin concentrator (Sartorius). The purity of the RAD9 ⌬C -RAD1-HUS1 complex (9 ⌬C -1-1) was confirmed by SDS-PAGE with Coomassie Brilliant Blue staining. In the present study, RAD9 ⌬C and 9 ⌬C -1-1 are simply termed RAD9 and 9-1-1, respectively, unless otherwise noted. For interaction analysis, site-directed mutations in RAD1 (m1 (K155A/R244A/Q254A) or m2 (F64A/M256A/ F266A)) were introduced by a PCR-based mutagenesis. Recombinant 9 ⌬C -1-1 with RAD1 mutations was produced and purified by a procedure similar to that of the WT. All purified proteins were frozen in liquid N 2 and stored at Ϫ80°C until use.

Crystallization and structure determination
The RHINO peptide ( 45 SKPIDHSTITSWVSPDFDTA 64 ) was commercially synthesized (TORAY Research Center, Inc.) and dissolved in buffer IV. A 10-fold molar excess of peptide was incubated with 0.13 mM 9-1-1 and 8.8 mM DTT. Crystallization was performed by the vapor-diffusion method. Rod crystals were obtained with a reservoir solution of 0.1 M Bistris propane, pH 7.5, 0.1 M sodium citrate, and 16% PEG 3350 at 20°C. Crystals were transferred to a buffer containing the reservoir solu-tion and 25% sucrose. X-ray diffraction data were collected at Photon Factory beamline BL-17A (Tsukuba, Japan) with an Eiger X16M single-photon-counting detector (DECTRIS). Diffraction data were processed with the program XDS (22). The crystal structure of 9-1-1 bound to the RHINO peptide was determined by molecular replacement method using the program PHASER (23). Binding of the RHINO peptide to the RAD1 subunit was confirmed by difference Fourier map, and the structure of the peptide was built by the program Coot (24). The structure was manually improved and refined by the program PHENIX (25). The final electron density of the RHINO peptide is shown in Fig. 1B. Data collection and refinement statistics are given in Table 1. Atomic coordinates and structure factor amplitudes have been deposited in the Protein Data Bank (entry 6J8Y).

Antibodies
The rabbit anti-HUS1 and anti-RAD1 antibodies were gifts from Dr. Katsunori Sugimoto (State University of New Jersey). The rabbit anti-RAD9, mouse anti-GST, horseradish peroxidaseconjugated anti-rabbit IgG, and Alexa 647-conjugated antimouse IgG antibodies were purchased from LifeSpan Biosciences, Santa Cruz Biotechnology, Bio-Rad, and Jackson Immuno-Research Laboratories, respectively.