Structural Basis of Ubiquitin Recognition by the Ubiquitin-associated (UBA) Domain of the Ubiquitin Ligase EDD*

EDD (or HYD) is an E3 ubiquitin ligase in the family of HECT (homologous to E6-AP C terminus) ligases. EDD contains an N-terminal ubiquitin-associated (UBA) domain, which is present in a variety of proteins involved in ubiquitin-mediated processes. Here, we use isothermal titration calorimetry (ITC), NMR titrations, and pull-down assays to show that the EDD UBA domain binds ubiquitin. The 1.85Å crystal structure of the complex with ubiquitin reveals the structural basis of ubiquitin recognition by UBA helices α1 and α3. The structure shows a larger number of intermolecular hydrogen bonds than observed in previous UBA/ubiquitin complexes. Two of these involve ordered water molecules. The functional importance of residues at the UBA/ubiquitin interface was confirmed using site-directed mutagenesis. Surface plasmon resonance (SPR) measurements show that the EDD UBA domain does not have a strong preference for polyubiquitin chains over monoubiquitin. This suggests that EDD binds to monoubiquitinated proteins, which is consistent with its involvement in DNA damage repair pathways.

The E3 isolated by differential display (EDD) 4 protein is the human ortholog of the Drosophila melanogaster tumor suppressor hyperplastic discs protein (HYD). Both belong to the family of HECT (homologous to E6-AP C terminus) E3 ubiq-uitin ligases, which target specific proteins for ubiquitin-mediated proteolysis. The highly conserved ubiquitin/proteasome pathway controls the degradation of many critical regulatory proteins. Proteins are targeted by conjugation of a 76-residue ubiquitin moiety to lysine residues via an isopeptide bond. This occurs through a combined set of reactions involving activating (E1), conjugating (E2), and ligating (E3) enzymes. E3 enzymes physically interact with their substrates and are thus critical determinants of the specificity of ubiquitination. Two main groups of E3 ubiquitin ligases include RING finger (1) and HECT domain ligases (2), the latter being found in EDD.
In Drosophila, EDD is required for regulation of cell proliferation during development (3). It is inferred that EDD functions in signal transduction downstream of the receptors to initiate and/or maintain proliferation as well as signals to terminate proliferation (3). Cells with mutations in EDD fail to properly terminate proliferation leading to a tumorous phenotype. The mutations also result in developmental abnormalities such as adult sterility due to germ cell defects (4). EDD is frequently overexpressed in breast and ovarian cancer supporting a potential role in cancer development (5,6). EDD is also involved in DNA damage signaling where TopB1, a target for ubiquitinylation by EDD (7), co-localizes with BRCA1 at stalled replication forks (7,8). Recently, EDD was shown to activate the DNA damage checkpoint kinase CHK2 (9). EDD also interacts with thecalciumandintegrin-bindingprotein(CIB)inaDNAdamagedependent manner (10). Finally, EDD is an in vivo substrate for the extracellular signal-regulated kinases (ERK) 1 and 2 (11). As a ubiquitin ligase, EDD was shown to control the levels of poly(A)-binding protein interacting protein 2 (Paip2) (12).
Structurally, EDD contains a ubiquitin-associated (UBA) domain at its N terminus (10), two nuclear localization signals, a zinc finger-like UBR domain involved in recognition of type 1 N-degrons (13), a domain homologous to the C-terminal domain of poly(A)-binding protein (PABC), and a HECT domain at its far C terminus. UBA domains were originally identified in a variety of proteins involved in ubiquitin-mediated processes (14,15). The three-dimensional structure of a UBA domain shows a bundle of three ␣-helices (16). UBA domains generally contain surface patches of hydrophobic residues, which are involved in protein-protein interactions. Most UBA domains bind ubiquitin and ubiquitin-like (UBL) domains (17). Two structures of a UBA domain in complex with monoubiquitin determined by NMR spectroscopy show that the first * This work was supported by CIHR Operating Grants MOP-11545 (to M. P.) and MOP-14219 (to K. G.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. loop and the C-terminal part of helix ␣3 of the UBA domain bind to the ␤-sheet of ubiquitin centered on Ile44 (18,19). In particular, the methionine residue from the ␣1-␣2 loop, conserved in many UBA domains, makes key hydrophobic contacts with ubiquitin. Despite the existing structural information, the mechanism of ubiquitin recognition is still not understood for a number of UBA domains. For instance, some UBA domains do not contain the critical methionine or even a hydrophobic residue in the corresponding position and still bind to ubiquitin efficiently (20). The low sequence homology between UBA domains makes predicting and understanding their function difficult. For example, recent high-resolution crystal structures of UBA domains from Cbl ubiquitin ligases revealed that, unlike previously studied UBA domains, the Cbl-b UBA domain binds ubiquitin via UBA helix ␣1 (21) while the c-Cbl UBA domain dimerizes (22).
Here, we used pull-down assays, SPR, ITC, and NMR spectroscopy to show that UBA from EDD ligase binds to monoubiquitin and polyubiquitin chains. To understand the structural basis of this binding, we performed x-ray crystallography and mutational analysis of this UBA domain. We report the crystal structure of EDD UBA in complex with ubiquitin. The structure improves our understanding of ubiquitin recognition by UBA domains and provides a basis for further functional studies of the ubiquitin ligase EDD.

EXPERIMENTAL PROCEDURES
Protein Expression, Preparation, and Purification-The UBA domain from human EDD ligase (residues 180 -230) was cloned into a pGEX-4T-1 vector (Amersham Biosciences) and expressed in Escherichia coli BL21 (DE3) in rich (LB) medium as a fusion with N-terminal GST tag. To aid quantification, an I180Y mutation was introduced into the constructs used for NMR and ITC. The wild-type sequence was used for crystallization, SPR, pull-down assays, and site-directed mutagenesis studies. GST fusion proteins were purified by affinity chromatography on glutathione-Sepharose resin, and the tag was removed by cleavage with thrombin, leaving a two-residue Gly-Ser N-terminal extension. The cleaved protein was additionally purified using size-exclusion chromatography (Superdex-75) and buffer exchanged using Centricon 5000 concentrators. For NMR experiments, the protein was labeled by growth in M9 minimal medium with [ 15 N]ammonium sulfate or [ 15 N]ammonium sulfate/[ 13 C]glucose as the sole sources of nitrogen and carbon. Bovine ubiquitin was purchased from Sigma-Aldrich and used without further purification. Site-directed mutagenesis was performed using QuikChange TM (Stratagene) and confirmed by DNA sequencing.
Isothermal Titration Calorimetry (ITC) Measurements-Experiments were carried out on a MicroCal VP-ITC titration calorimeter (MicroCal Inc., Northampton, MA) using the VPViewer software for instrument control and data acquisition. The buffer used for ITC experiments contained 25 mM phosphate buffer (pH 7.0), 70 mM NaCl. During a titration experiment, 50 M sample of UBA from EDD was thermostatted at 293 K in a stirred (260 rpm) reaction cell of 1.4 ml. Thirty-seven injections, each of 8-l volume and 10-s duration with a 5-min interval between injections, were carried out using a 296-l syringe filled with 1.0 mM ubiquitin solution. The calorimetric data were processed, and the binding affinity determined using the software ORIGIN (version 5.0) provided by the manufacturer.
Surface Plasmon Resonance (SPR)-based Biosensor Analysis -The interaction of UBA from EDD with the synthetic ubiquitin was analyzed by SPR (23) using a BIACORE 3000 optical biosensor (Biacore AB, Uppsala, Sweden). Protein dilutions were performed in running buffer, which contained 10 mM HEPES at pH 7.4, 150 mM NaCl, 3.4 mM EDTA, and 0.005% Tween 20. The concentration of EDD UBA was determined by hydrolysis and amino acid analysis. In a first series of experiments, monoubiquitin, K48-linked, K63-linked di-and tetraubiquitin were immobilized onto CM5 biosensor chip flow cells using amine coupling chemistry as described previously (23), leading to 2000 RUs of monoubiquitin, 3200 and 6300 RU of K63-and K48-diubiquitin, 11500 and 11000 RU of K48-and K63-tetraubiquitin. Biosensor experiments were repeated a minimum of two times and corrected for binding to a separate control flow cell (activated and blocked). The experiments were performed at 25°C using a flow rate of 50 l/min. For each experiment, at least 5 different concentrations of EDD UBA were injected over each experimental and control flow cell (1 min injections, extraclean procedure). In a second series of experiments, EDD UBA was biotinylated by addition and 90 min of incubation of 2 mg of EZ-Link® NHS-Chromogenic-Biotin (Pierce Biotechnology) to 0.6 mol of EDD UBA in 1 ml buffer (10 mM HEPES pH 7.4, 150 mM NaCl) plus 200 l Me 2 SO. Biotinylated protein was purified by gel filtration and quantified by optical spectroscopy (354 nm). The biotinylated EDD UBA (85 RU) was captured onto a streptavidin-coated (SA) sensorchip flow cell previously prepared by three successive pulses of 1 M NaCl in 50 mM NaOH, according to the manufacturer's recommendations. Injections of monoubiquitin, K48-and N-to-C-linked diubiquitin were performed in duplicate over EDD UBA and control (streptavidin only) surfaces.
Thermodynamic dissociation constants (K d ) were determined using control-corrected plateau values; the K d was derived from fitting of the experimental values. For analysis of ubiquitin chains binding to the UBA-derivatized surface, a global analysis was used in which the number of binding sites on the surface was constrained to be equal in all experiments.
Crystallization-Conditions were identified utilizing hanging drop vapor diffusion using the JCSG crystallization suite (Qiagen). The best crystals of the complex were obtained by equilibrating a 0.6-l drop of an equimolar protein mixture (15 mg/ml total) in 20 mM Tris-HCl (pH 8.0), mixed with 0.6 l of reservoir solution containing 0.1 M citric acid (pH 5.0) and 20% (w/v) PEG 6000 and suspended over 0.6 ml of reservoir solution. Crystals grew in 3-7 days at 20°C. For data collection, crystals were picked up in a nylon loop and flash-cooled in a N 2 cold stream. The solution for cryoprotection contained the reservoir solution with addition of 20% (v/v) of glycerol. The crystals contained four UBA and four ubiquitin molecules in the asymmetric unit corresponding to V m ϭ 2.2 Å 3 Da Ϫ1 and a solvent content of 44% (24).
Structure Solution and Refinement-Diffraction data from a single crystal of the UBA-ubiquitin complex were collected on an ADSC Quantum-315 CCD detector (Area Detector Systems Corp.) at beamline X29 at the NSLS, Brookhaven National Laboratory (Table 1). Data processing and scaling were performed with HKL2000 (25). The structure was determined by molecular replacement with Phaser (26), using the coordinates of ubiquitin (PDB entry 1UBQ) and the UBA domain from Cbl-b (PDB entry 2OOA). The initial model obtained from Phaser was completed and adjusted with the program Xfit (27) and was improved by several cycles of refinement, using the program CNS v.1.1 (28) and model refitting. The refinement statistics are given in Table 1. Out of 53 residues in the UBA construct, the current model does not include residues Arg 226 -Asp 230 in chain B, Glu 229 -Asp 230 in chains D and F, and Asp 228 -Asp 230 in chain H. The three C-terminal residues of ubiquitin are also disordered in the chains A and G, while chains C and E are complete due to crystal packing. The final model has good ste-reochemistry (  15 N-labeled 0.5-1.0 mM wild-type UBA or its mutants to the final molar ratio of 3 to 1 or 4 to 1, respectively. All NMR experiments were performed at 303 K using Varian 500 MHz and Bruker 600 MHz spectrometers. NMR spectra were processed using NMRPipe (30) and analyzed with XEASY (31).

EDD UBA Binds to Ubiquitin and Polyubiquitin Chains-To
study the previously uncharacterized UBA domain, residues Ile 180 to Asp 230 of EDD were cloned for expression in E. coli. We tested its binding to ubiquitin using pull-down assays with ubiquitin agarose (Fig. 1A). EDD UBA bound to monoubiquitin with an affinity similar to that of the UBA domain from the Cbl-b ubiquitin ligase (21). Because many UBA domains have higher affinity for ubiquitin chains than for monoubiquitin, we also performed in vitro binding assays with K48-and K63linked ubiquitin chains. While EDD UBA bound both forms of polyubiquitin, it displayed relatively higher affinity for K63linked chains compared with the UBA domain from Cbl-b (Fig. 1B).
To further characterize the binding, we applied isothermal titration calorimetry (ITC) to determine binding constant and thermodynamic parameters of interactions between the EDD UBA domain and ubiquitin (Fig. 1C). The affinity (K d ) of the EDD UBA domain for monoubiquitin was measured to be 60 Ϯ 15 M, which is higher than the affinity of most characterized UBA domains (32) but again very similar to the affinity of Cbl-b (21).
EDD UBA Does Not Have Specificity for Ubiquitin Chains-We used surface plasmon resonance (SPR) experiments to verify the pull-down results and quantify the affinity of the EDD UBA domain for different ubiquitin chains (Fig. 2). Five different SPR surfaces were prepared and EDD UBA domain was injected over them ( Fig. 2A). At the highest UBA concentration (0.55 mM), the binding curves did not reach saturation, which limited the extent of analysis possible. In addition, the surfaces appeared to suffer from heterogeneity resulting from multiple modes of ubiquitin and polyubiquitin coupling to the surface. This led to large overestimates of the K d 's but still allowed qualitative comparisons between surfaces. As a control, the UBA domain of Mud1 was used (33).
K d values of EDD UBA for the polyubiquitin surfaces were roughly half the K d for monoubiquitin (Fig. 2A). The UBA domain showed slightly better affinity for K63-diubiquitin than for K48-diubiquitin but there was no difference between the two tetraubiquitin surfaces. In contrast, the Mud1 UBA showed a 5-fold preference for K48-linked polyubiquitin with K d values of 58 M for the K48-Ub 4 surface and 283 M for the K63-Ub 4 surface (supplemental Fig. S1). Our conclusion is that unlike Mud1, the UBA domain from EDD did not show markedly higher affinity for the K48-or K63-linked polyubiquitin surfaces.
We also carried out reciprocal SPR experiments injecting ubiquitin chains over a surface on which the EDD UBA domain had been captured (Fig. 2B). Although we were limited by the amounts and types of ubiquitin chains that could be used, this second approach had the advantage that a single surface could be used to compare affinities. The lower loading of the surface also significantly improved the agreement between affinity of monoubiquitin measured by SPR (127 M) and ITC (60 M). On the UBA surface, K48-linked and N-to-C diubiquitin both bound with higher affinity than monoubiquitin but also with detectable negative cooperativity (Fig. 2B). The K d for N-to-C diubiquitin was half that of K48-diubiquitin, consistent with the more open conformation of this type of chain in solution (34).
The results from the SPR experiments suggest that the EDD UBA domain binds to polyubiquitin relatively poorly compared with other UBA domains (20,33). The higher apparent affinity for polyubiquitin results from the larger number monoubiquitin units available for binding (avidity effects) and is not due to specific recognition of polyubiquitin chains. K63-linked and N-to-C chains showed better binding due to the more open conformation of these types of chains (34). In contrast, K48linked chains bound somewhat less well due to their relatively closed conformation where competition exists between ubiquitin-ubiquitin interactions and UBA binding (35).
Structure of EDD UBA-To understand the structural basis of the ubiquitin-binding properties of EDD UBA, we co-crystallized ubiquitin with the EDD UBA domain and solved the structure of the complex at 1.85-Å resolution ( Table 1). The crystals contained four UBA-ubiquitin complexes in the asymmetric unit. All four complexes were nearly identical; their backbone atoms could be overlaid with an r.m.s.d. of 0.37 Å. The UBA structures themselves could be overlaid with an r.m.s.d. of 0.29 Å.
The EDD fragment shows the canonical UBA fold consisting of a three-helical bundle (␣1-␣3) (Fig. 3). The domain has a well-defined hydrophobic core that includes Ile 190 , Ala 193 , Leu 197 , Ile 205 , Leu 209 , Ala 219 , and Leu 223 (Fig. 3A). Among the  notable features, the N-terminal portion before the UBA domain contains a short helical turn that packs against helices ␣1 and ␣2. This ␣0-helix covers the hydrophobic patch consisting of Ile 190 , Ile 206 , and Leu 214 that would be otherwise solventexposed (Fig. 3B). Residues Ile 180 -Pro 186 adopted an essentially identical conformation in all four copies in asymmetric unit, which suggests that the ␣0-helix is structurally relevant. In addition, mutagenesis of Leu 189 , Ile 190 , and Ile 206 destabilized the protein (see below). On the other hand, NMR experiments showed that the helix is only partially populated. Heteronuclear NOEs revealed that the Ile 180 -Ser 183 region is relatively mobile in solution (supplemental Fig. S2) and 15 N NOESY NMR data did not show the medium range NOEs characteristic of a helical turn (data not shown). Superposition of the EDD UBA structure with related structures reveals that EDD UBA is structurally similar to CUE (coupling of ubiquitin conjugation to endoplasmic reticulum degradation) domains. CUE domains are a conserved monou-biquitin-binding fold that resembles UBA domains at the level of amino acid sequence, size, tertiary fold, and ubiquitin binding (18,36) but are distinguished by a longer ␣1-␣2 loop containing an MFP signature motif. The EDD UBA ␣1-␣2 loop adopts a conformation identical to that of CUE domain structures with residues Val 196 -Leu 197 -Gln 198 superimposable on the MFP motif (Fig. 3D). Leu 197 from the ␣1-␣2 loop is part of the hydrophobic core and stacks against Leu 223 . The loop is relatively rigid in solution as evidenced by heteronuclear NOEs (supplemental Fig. S2).
EDD UBA-Ubiquitin Recognition Elements-The UBA/ubiquitin intermolecular interface consists of the ␤-sheet surface of ubiquitin and helix ␣3 and the C terminus of helix ␣1 of EDD UBA (Fig. 4A). The majority of interactions are hydrophobic. The side chain of Val 196 fits into a hydrophobic pocket formed by the side chains of Leu 8 , His 68 , and Val 70 of ubiquitin (Fig. 4B). Half a turn earlier, the aliphatic atoms of Gln 192 interact with the side chain of Leu 8 . The side chain of Gln 198 in the ␣1-␣2 loop contacts the backbone of Ala 46 . The full length of helix ␣3 is involved in ubiquitin binding starting with the side chain of Val 216 that contacts ubiquitin Leu 8 . Leu 224 is another crucial determinant of ubiquitin recognition as its side chain binds in a groove formed by Val 70 , Ile 44 , Gln 49 , and Arg 42 of ubiquitin (Fig. 4C).
The hydrophobic interactions between the EDD UBA domain and ubiquitin are strengthened by a variety of hydrogen bonds. The side chain of Asn 221 makes hydrogen bonds with the backbone carbonyl of Leu 71 and the side chain of Arg 42 , which also hydrogen bonds with the side chain of Ser 225 (Fig.  4D). The carbonyl of Leu 224 forms a hydrogen bond with the side chain of Gln 49 . The structure also reveals intermolecular hydrogen bonds involving ordered water molecules. Thus, carbonyl of Val 196 hydrogen bonds with amide of Gly 47 via a bound water molecule (Fig. 4E). Similarly, another water molecule bridges the side chain of Asn 217 and the amide of Leu 71 (Fig. 4F). Notably, the above-mentioned ionic interactions were observed in all four copies of the EDD UBA/ubiquitin structure and, thus, unlikely to be affected by crystal packing.
NMR Confirms the Ubiquitin-binding Site in EDD UBA-To verify the mechanism of ubiquitin recognition in solution, EDD UBA was expressed in media supplemented with 15 NH 4 Cl. The The residues forming hydrophobic core are shown as sticks and labeled. Helices are labeled from ␣0 to ␣3. B, hydrophobic residues of the ␣1-␣2 surface contact hydrophobic residues in the ␣0 helix. C, backbone overlay of the EDD UBA structure (yellow) with Dsk2 (green), Ede1 (dark blue), and Cue2 (light blue) structures. D, residues in the ␣1-␣2 loop of EDD and Cue2. The loop conformation is strikingly similar in the UBA and CUE domain structures despite the different amino acids sequence. E, sequence alignment of the UBA domains from EDD, Dsk2, Ede1, and Ubx5 and CUE domain from Cue2. The ␣-helices and locations of amino acid substitutions in EDD that destabilize the protein (horizontal bars), that block ubiquitin binding (red X), or that have no effect (green circles) are shown. 15 N-1 H correlation spectra of the unliganded domain showed the good dispersion of signals characteristic of a folded protein but with heterogeneity in peak heights suggesting the existence of conformational exchange or aggregation (supplemental Fig.  S3). Addition of unlabeled ubiquitin caused chemical shift changes for a limited number of UBA residues and, notably, sharpened the broad signals in the NMR spectra, making them amenable to assignment using standard heteronuclear experiments with double-labeled protein. As a result, all signals of the ubiquitin-bound form were assigned with the exception of missing signals from the N-terminal Gly-Ser cloning linker (Fig.  5A). These assignments and NMR titration spectra were used to assign the signals in ligand-free EDD UBA, which were then verified using an 15 N NOESY experiment.
Site-directed Mutagenesis-We prepared seven point mutants of EDD UBA and tested their binding to ubiquitin in in vitro pull-down assays (Fig. 6). Mutants V196K, L214N, L218K, and L224K were confirmed to be correctly folded as verified using two-dimensional HSQC NMR spectra; the three remaining mutants L189K, I190K, and I206K were partially unfolded (data not shown and supplemental Fig. S3). The destabilized mutations are located at the interface with helix ␣0 (Fig. 3B), which is consistent with the importance of this helix in stabilizing the structure of the isolated EDD UBA domain.
Among the folded mutants, the V196K and L224K substitutions completely abrogated binding to ubiquitin-agarose (Fig.  6B) and pull-down of K48-and K63-linked polyubiquitin (Fig.  6, C and D). The lack of ubiquitin binding was confirmed by NMR experiments: a 4-fold excess of ubiquitin did not produce observable chemical shift changes in the HSQC spectra (data not shown). The mutants I206K (which could fold in the presence of ubiquitin), L214N, and L218K showed binding to both monoubiquitin and polyubiquitin chains (Fig. 6). Significantly, we found no evidence of mutations that would differentially affect K48-or K63-linked polyubiquitin binding.

DISCUSSION
This work reports the structure of a previously uncharacterized UBA domain from EDD. We show that the domain binds to ubiquitin and present a comprehensive molecular characterization of the EDD UBA/ubiquitin interactions. EDD was first classified as a ubiquitin ligase based on the presence of catalytic HECT domain. Ubiquitin binding by the UBA domain provides another link for the involvement of EDD in the ubiquitin-mediated processes.
Previous UBA-ubiquitin complex structures exhibit two distinct binding modes. While most UBA domains recognize ubiquitin and ubiquitin-like domains via the ␣1-␣2 loop and helix ␣3 (18,19,37,39,(41)(42)(43), the UBA domain from Cbl-b ubiquitin ligase binds to ubiquitin using helix ␣1 (21). The EDD UBA-ubiquitin structure is largely similar to the former group which also includes CUE-ubiquitin complexes (38,44). The distinction between UBA and CUE domains is blurred by the EDD UBA domain, which contains elements of both families.
The structure also underlines the challenges of sequence alignment of UBA domains, which results from their small size and low sequence identity. The initial motivation for structural studies of EDD UBA was to explain ubiquitin binding by EDD, which lacks the conserved methionine residue found in the ␣1-␣2 loop of other UBA domains. The structure revealed that the EDD possesses two extra residues in the ␣1-␣2 loop, which positions a hydrophobic residue, valine, in place of the conserved methionine. The larger ␣1-␣2 loop is a characteristic of CUE domains but may occur in other UBA domains, such as the domain from the Ubx5 protein (also called YDR330w) (45). The Ubx5 UBA domain aligns well with EDD UBA and contains isoleucine at the position corresponding to Val 196 of EDD UBA (Fig. 3E).
The EDD UBA-ubiquitin structure explains how a smaller hydrophobic residue, valine, can replace the larger methionine residue. Superposition of five complexes shows that the EDD UBA is slightly shifted relative to ubiquitin compared with other UBA/CUE domains (Fig. 7). In helix ␣1, this moves the side chain of Val 196 so that it occupies the same position as the methionine residue in other complexes with ubiquitin. The shift also leads to a larger distance between carbonyl of Val 196 and amide of Gly 47 of ubiquitin, which are unable to form the direct hydrogen bond seen in other UBA-ubiquitin complexes. Interestingly, this intermolecular interaction is preserved in the EDD UBA/ubiquitin   (L214N, L218K), and residues that destabilized the UBA domain when mutated are colored gray (L189K, I190K, L206K). structure via a bound water molecule. The EDD UBA-ubiquitin structure reveals novel intermolecular ionic interactions, most of them involving Arg 42 of ubiquitin that forms a network of intra-and intermolecular hydrogen bonds (Fig. 4D). Perhaps, due to more extensive polar intermolecular interactions, the buried surface (725 Å 2 ) of the EDD UBA/ubiquitin complex is larger than that of ubiquitin bound to the UBA domains of Ede1 (19) (370 Å 2 ) and Dsk2p (18) (480 Å 2 ).
Our results show that the EDD UBA domain does not bind polyubiquitin chains significantly more tightly than monoubiquitin. Monoubiquitination is a common post-translational modification in DNA damage repair pathways. Monoubiquitinated proliferating cell nuclear antigen (PCNA) activates translesion DNA synthesis by damage-tolerant polymerases in yeast leading to DNA damage-induced mutagenesis (46). In another example, FANCD2, a protein involved in Fanconi Anemia, is monoubiquitinated upon DNA damage and recruited to chromatin-associated nuclear foci, where it interacts with BRCA1 and other DNA repair enzymes (47). Interestingly, EDD is also involved in DNA damage signaling via ubiquitinylation of TopB1 (7), which co-localizes with BRCA1 at stalled replication forks (7,8). Understanding of ubiquitin binding properties of EDD ubiquitin ligase provides a basis for functional studies of its importance for the EDD family of proteins.