Portal de Periódicos da CAPES

Solution Structure of the Kaposi's Sarcoma-associated Herpesvirus K3 N-terminal Domain Reveals a Novel E2-binding C4HC3-type RING Domain

2004; Elsevier BV; Volume: 279; Issue: 51 Linguagem: Inglês

10.1074/jbc.m409662200

1083-351X

Roger B. Dodd, Mark D. Allen, Stephanie E. Brown, Christopher M. Sanderson, Lidia M. Duncan, Paul J. Lehner, Mark Bycroft, Randy J. Read,

Herpesvirus Infections and Treatments

RING domains are found in a large number of eukaryotic proteins. Most function as E3 ubiquitin-protein ligases, catalyzing the terminal step in the ubiquitination process. Structurally, these domains have been characterized as binding two zinc ions in a stable cross-brace motif. The tumorigenic human γ-herpesvirus Kaposi's sarcoma-associated herpesvirus encodes a ubiquitin-protein ligase termed K3, which functions as an immune evasion molecule by ubiquitinating major histocompatibility complex class I. K3 possesses at its N terminus a domain related to cellular RING domains but with an altered zinc ligand arrangement. This domain was initially characterized as a plant homeodomain, a structure not previously known to function as an E3. Here, it is conclusively demonstrated that the K3 N-terminal domain is a variant member of the RING domain family and not a plant homeodomain. The domain is found to interact with the cellular ubiquitin-conjugating enzymes UbcH5A to -C and UbcH13, which dock to the equivalent surface as on classical cellular RING domains. Interaction with UbcH13 suggests a possible role for K3 in catalyzing Lys63-linked ubiquitination. RING domains are found in a large number of eukaryotic proteins. Most function as E3 ubiquitin-protein ligases, catalyzing the terminal step in the ubiquitination process. Structurally, these domains have been characterized as binding two zinc ions in a stable cross-brace motif. The tumorigenic human γ-herpesvirus Kaposi's sarcoma-associated herpesvirus encodes a ubiquitin-protein ligase termed K3, which functions as an immune evasion molecule by ubiquitinating major histocompatibility complex class I. K3 possesses at its N terminus a domain related to cellular RING domains but with an altered zinc ligand arrangement. This domain was initially characterized as a plant homeodomain, a structure not previously known to function as an E3. Here, it is conclusively demonstrated that the K3 N-terminal domain is a variant member of the RING domain family and not a plant homeodomain. The domain is found to interact with the cellular ubiquitin-conjugating enzymes UbcH5A to -C and UbcH13, which dock to the equivalent surface as on classical cellular RING domains. Interaction with UbcH13 suggests a possible role for K3 in catalyzing Lys63-linked ubiquitination. Ubiquitination was originally characterized as a signal for protein destruction by the proteasome (1Tanaka K. Chiba T. Genes Cells. 1998; 3: 499-510Crossref PubMed Scopus (87) Google Scholar). However, recent developments have implicated ubiquitin in control of a multitude of cellular processes including signal transduction, apoptosis, DNA repair, protein-protein interaction, endocytosis, and protein trafficking (2Wilkinson C.R. Trends Cell Biol. 2002; 12: 545-546Abstract Full Text Full Text PDF PubMed Scopus (7) Google Scholar, 3Schnell J.D. Hicke L. J. Biol. Chem. 2003; 278: 35857-35860Abstract Full Text Full Text PDF PubMed Scopus (350) Google Scholar, 4Johnson E.S. Nat. Cell Biol. 2002; 4: E295-E298Crossref PubMed Scopus (41) Google Scholar). A host of proteins have been identified that can modify the ubiquitination process, bind ubiquitinated proteins, and deubiquitinate proteins. It has become apparent that the nature of the linkage in polyubiquitin chains dictates function, with Lys48-linked chains signaling proteasomal destruction and Lys63-linked chains controlling DNA repair and endocytic processes (5Varadan R. Assfalg M. Haririnia A. Raasi S. Pickart C. Fushman D. J. Biol. Chem. 2004; 279: 7055-7063Abstract Full Text Full Text PDF PubMed Scopus (280) Google Scholar). Given the importance of the ubiquitination system, it is not surprising that viruses have developed methods to exploit it. The cellular ubiquitination machinery consists of a catalytic cascade of three classes of enzyme acting in concert to transfer the small protein ubiquitin onto target proteins. The first enzyme (E1) 1The abbreviations used are: E1, ubiquitin-activating enzyme; E2, ubiquitin carrier protein; E3, ubiquitin-protein isopeptide ligase; HECT, homologous to the E6-AP carboxyl terminus; KSHV, Kaposi's sarcoma-associated herpesvirus, also termed human herpesvirus 8; MHC, major histocompatibility complex; ESCRT-I, endosomal sorting complex required for transport; MARCH, membrane-associated RING-CH; PHD, plant homeodomain; Ub, ubiquitin; Ubc, ubiquitin-conjugating enzyme (H suffix indicates human enzyme); NOE, nuclear Overhauser effect; NOESY, nuclear Overhauser effect spectroscopy; TOCSY, total correlation spectroscopy; DQF-COSY, double-quantum-filtered correlation spectroscopy; HSQC, heteronuclear single-quantum coherence; NTA, nitrilotriacetic acid; MOPS, 4-morpholinepropanesulfonic acid; r.m.s., root mean square. is termed a ubiquitin-activating enzyme and forms a thiol-ester bond between its active site cysteine and the carboxyl terminus of ubiquitin in an ATP-dependent reaction. The ubiquitin is then transferred to the cysteine residue of a ubiquitin-conjugating enzyme (E2). Finally, the E3 ubiquitin-protein ligase, which possesses a substrate recognition module, acts to bridge E2 and the target, allowing formation of isopeptide bonds between target lysine ϵ-amino groups and ubiquitin molecules. This modification can lead to monoubiquitination or the creation of polyubiquitin chains on the substrate. Ubiquitin-protein ligases possess one of three possible E2-interacting domains. The HECT domain is an enzymatically active domain, forming an E3-Ub covalent intermediate before transfer to substrate (6Schwarz S.E. Rosa J.L. Scheffner M. J. Biol. Chem. 1998; 273: 12148-12154Abstract Full Text Full Text PDF PubMed Scopus (142) Google Scholar). RING (7Lorick K.L. Jensen J.P. Fang S. Ong A.M. Hatakeyama S. Weissman A.M. Proc. Natl. Acad. Sci. U. S. A. 1999; 96: 11364-11369Crossref PubMed Scopus (947) Google Scholar, 8Joazeiro C.A. Weissman A.M. Cell. 2000; 102: 549-552Abstract Full Text Full Text PDF PubMed Scopus (1051) Google Scholar, 9Freemont P.S. Curr. Biol. 2000; 10: R84-R87Abstract Full Text Full Text PDF PubMed Google Scholar) and U box (10Hatakeyama S. Yada M. Matsumoto M. Ishida N. Nakayama K.I. J. Biol. Chem. 2001; 276: 33111-33120Abstract Full Text Full Text PDF PubMed Scopus (477) Google Scholar) domains are nonenzymatic and act as bridging proteins that bind to E2s. In addition to these domains, E3s possess a large variety of adaptor domains and motifs to allow binding to target proteins and accessory factors. The RING domain is a small (8 kDa), compact, and stable zinc-binding module. Two zinc ions are bound by a set of eight metal-ligating cysteine and histidine residues in what has been termed a “cross-brace” motif (11Borden K.L. Freemont P.S. Curr. Opin. Struct. Biol. 1996; 6: 395-401Crossref PubMed Scopus (418) Google Scholar, 12Barlow P.N. Luisi B. Milner A. Elliott M. Everett R. J. Mol. Biol. 1994; 237: 201-211Crossref PubMed Scopus (253) Google Scholar) (Fig. 1C). These residues have a characteristic spacing (Fig. 1, A and B) and are subclassified into RING-HC (C3HC4) and RING-H2 (C3H2C3) domains dependent on whether the fifth metal ligand is a cysteine or histidine, respectively. In addition to the residue spacing, RING domains possess two highly conserved hydrophobic residues, one immediately N-terminal to metal ligand 5 and the other C-terminal to ligand 6, that form part of the hydrophobic core of the structure (Fig. 1, A and B). An interesting development in the field of ubiquitination came with the investigation of two viral systems: murine γ-herpesvirus 68 and Kaposi's sarcoma-associated herpesvirus (KSHV). KSHV is a human herpesvirus and is implicated in the development of Kaposi's sarcoma in individuals afflicted by AIDS (13Chang Y. Cesarman E. Pessin M.S. Lee F. Culpepper J. Knowles D.M. Moore P.S. Science. 1994; 266: 1865-1869Crossref PubMed Scopus (5015) Google Scholar, 14Cesarman E. Knowles D.M. Semin. Cancer Biol. 1999; 9: 165-174Crossref PubMed Scopus (162) Google Scholar). Normally, MHC class I molecules present viral peptide antigens to cytotoxic T cells allowing recognition and destruction of infected cells. However, both murine γ-herpesvirus 68 and KSHV have developed an immune evasion strategy involving removal of cell surface MHC class I. This evasion is dependent on the expression of viral immediate early proteins: MK3 in the case of murine γ-herpesvirus 68 (15Stevenson P.G. Efstathiou S. Doherty P.C. Lehner P.J. Proc. Natl. Acad. Sci. U. S. A. 2000; 97: 8455-8460Crossref PubMed Scopus (193) Google Scholar) and K3 and K5 for KSHV (16Ishido S. Wang C. Lee B.S. Cohen G.B. Jung J.U. J. Virol. 2000; 74: 5300-5309Crossref PubMed Scopus (380) Google Scholar) (also referred to as MIR1 and -2, respectively). These proteins share a highly conserved N-terminal domain (MK3/K3, 40% identity; and K3/K5, 64% identity), two transmembrane domains, and a far less well conserved C-terminal region. Genes encoding homologous proteins have also been identified in other γ-herpesviruses, some poxviruses, and humans (17Fruh K. Bartee E. Gouveia K. Mansouri M. Virus Res. 2002; 88: 55-69Crossref PubMed Scopus (72) Google Scholar, 18Bartee E. Mansouri M. Hovey Nerenberg B.T. Gouveia K. Fruh K. J. Virol. 2004; 78: 1109-1120Crossref PubMed Scopus (250) Google Scholar). Experiments have demonstrated that MK3, K3, and K5 possess ubiquitin-protein ligase activity. All are able to catalyze the transfer of multiple ubiquitin molecules to MHC class I molecules in model systems. In the case of MK3, class I molecules are ubiquitinated on the endoplasmic reticulum membrane and destroyed by the proteasome (19Boname J.M. Stevenson P.G. Immunity. 2001; 15: 627-636Abstract Full Text Full Text PDF PubMed Scopus (175) Google Scholar). In contrast, K3 promotes ubiquitination in a post-endoplasmic reticulum compartment, leading to endocytosis of cell surface class I (20Coscoy L. Sanchez D.J. Ganem D. J. Cell Biol. 2001; 155: 1265-1273Crossref PubMed Scopus (249) Google Scholar, 21Hewitt E.W. Duncan L. Mufti D. Baker J. Stevenson P.G. Lehner P.J. EMBO J. 2002; 21: 2418-2429Crossref PubMed Scopus (162) Google Scholar, 22Haque M. Ueda K. Nakano K. Hirata Y. Parravicini C. Corbellino M. Yamanishi K. J. Gen. Virol. 2001; 82: 1175-1180Crossref PubMed Scopus (64) Google Scholar). Endocytosed class I molecules are then routed to the lysosome for destruction in a process dependent on Tsg101, a component of the mammalian ESCRT-I endosomal sorting complex (21Hewitt E.W. Duncan L. Mufti D. Baker J. Stevenson P.G. Lehner P.J. EMBO J. 2002; 21: 2418-2429Crossref PubMed Scopus (162) Google Scholar). Nine human, cellular homologs of K3, termed MARCH proteins, have been identified, and two (MARCH-IV and MARCH-IX) have a ubiquitin ligase activity similarly targeted against MHC class I (18Bartee E. Mansouri M. Hovey Nerenberg B.T. Gouveia K. Fruh K. J. Virol. 2004; 78: 1109-1120Crossref PubMed Scopus (250) Google Scholar). The domain at the N terminus of these proteins was initially identified as a plant homeodomain (PHD). This domain binds two zinc ions in an interleaved fashion similar to that employed by RING domains (23Capili A.D. Schultz D.C. Rauscher I.F. Borden K.L. EMBO J. 2001; 20: 165-177Crossref PubMed Scopus (164) Google Scholar). PHD and RING domains differ most obviously in the order of the metal ligands, with the middle two ligands swapped to give a C4HC3 arrangement in the PHD domain. However, recent reports have disputed the classification of these proteins' domains as PHD domains based only on the nature of the metal ligands (24Scheel H. Hofmann K. Trends Cell Biol. 2003; 13: 285-288Abstract Full Text Full Text PDF PubMed Scopus (45) Google Scholar, 25Coscoy L. Ganem D. Trends Cell Biol. 2003; 13: 7-12Abstract Full Text Full Text PDF PubMed Scopus (132) Google Scholar, 26Coscoy L. Ganem D. Trends Cell Biol. 2003; 13: 287-288Abstract Full Text Full Text PDF Google Scholar, 27Aravind L. Iyer L.M. Koonin E.V. Cell Cycle. 2003; 2: 123-126Crossref PubMed Scopus (83) Google Scholar). The domains lack a crucial tryptophan residue, 2 residues N-terminal to the seventh metal ligand, which is absolutely conserved in all PHD domains (Fig. 1, A and B). In PHD domains, this residue is involved in making contacts within the hydrophobic core, whereas in RING domains it is solvent exposed. This results in a substantial structural difference between PHD and RING domains in the region between the sixth and seventh metal ligands. In most RING domains of known structure, this stretch forms a helical segment containing a hydrophobic residue involved in E2 binding. In PHD domains, the chain diverges significantly from the path taken in all RING domains of known structure and approaches the second zinc-binding site from the opposite direction. Most PHD domains possess proline residues in this region, which are known to disrupt helix formation. Two recent reports have applied more rigorous techniques using hidden Markov models (24Scheel H. Hofmann K. Trends Cell Biol. 2003; 13: 285-288Abstract Full Text Full Text PDF PubMed Scopus (45) Google Scholar) and analysis of conserved structural features (27Aravind L. Iyer L.M. Koonin E.V. Cell Cycle. 2003; 2: 123-126Crossref PubMed Scopus (83) Google Scholar) to lend credence to the reclassification of the viral domains as variants of the classic RING domain. These variant domains are found within proteins from a wide variety of eukaryotic organisms including yeast, plants, Caenorhabditis elegans, and mammals. The question of classification extends beyond the issue of nomenclature. The suggestion that PHD domains in general could function as ubiquitin ligases would have huge implications, since many PHD-containing proteins are chromatin-associated and involved in genome maintenance and control. In this study, the solution structure of the N-terminal domain of the KSHV K3 protein has been determined. This unambiguously confirms its identification as a variant member of the RING domain family, resembling classic RINGs in many important respects. A yeast two-hybrid assay has been used to establish to which E2 ubiquitin-conjugating enzymes the K3 RING can bind and these findings have been confirmed in vitro. Using 15N-labeled protein, the E2-binding surface on the RING has been mapped by NMR chemical shift perturbations to a region equivalent to that on classic RING domains. Protein Preparation—The RING region of the KSHV K3 gene, corresponding to residues 1–60, was cloned into a modified pRSET(A) plasmid (Invitrogen) containing a lipoyl domain fusion tag and expressed by the E. coli strain C41(DE3) at 20 °C. For isotope labeling, K-MOPS minimal medium (28Neidhardt F.C. Bloch P.L. Smith D.F. J. Bacteriol. 1974; 119: 736-747Crossref PubMed Google Scholar) containing 15N-labeled NH4Cl and/or 13C-labeled glucose was used. Full-length sequences encoding UbcH5A to -C (UBE2D1 to -3) were cloned into the pETM-12 vector (EMBL), and UbcH13 (UBE2N) was cloned into pET-23d (Novagen). E2 proteins were expressed in Rosetta (DE3) pLysS (Novagen) at a temperature of 20 °C. All proteins were purified on an Ni2+-NTA superflow column (Qiagen) pre-equilibrated (50 mm sodium phosphate, 300 mm NaCl, 10 mm imidazole, 10 mm 2-mercaptoethanol, pH 8.0) at 4 °C and washed with 50 mm sodium phosphate, 300 mm NaCl, 20 mm imidazole, 10 mm 2-mercaptoethanol, pH 8.0. The proteins were eluted with 300 mm imidazole in this same buffer. Proteins were then dialyzed extensively against phosphate-buffered saline, pH 7.4, and in the case of the ubiquitin-conjugating enzymes immediately purified further on a Superdex 75 column (Amersham Biosciences). RING domain was treated with thrombin following dialysis and run again on Ni2+-NTA superflow to remove the lipoyl tag before application of the flow-through to a Superdex 75 column. Fractions containing RING were pooled and dialyzed against the appropriate NMR buffers. 1 mm RING domain was dialyzed into a 50 mm potassium phosphate buffer, pH 6.0 (containing 10% D2O). 0.3 mm RING-Ubc complexes were dialyzed into a 50 mm potassium phosphate buffer, pH 7.0 (containing 10% D2O). Yeast Two-hybrid Assay—Constructs encoding the first 80 residues of wild-type and W41A K3 (up to the start of the first transmembrane domain) were cloned into the prey vector pACTBD-B by in vivo gap repair cloning as described previously (29Estevez A.M. Lehner B. Sanderson C.M. Ruppert T. Clayton C. J. Biol. Chem. 2003; 278: 34943-34951Abstract Full Text Full Text PDF PubMed Scopus (78) Google Scholar). Full-length open reading frames for all of the ubiquitin-conjugating enzymes screened were cloned by in vivo gap repair into the bait vector pGBAD-B. The bait plasmids encoding the TRP1 marker were transformed by electroporation into the yeast strain PJ69-4A (MATa trp1–901 leu2–3, 112 ura3–52, his3–200 gal4Δ gal80 LYS2::GAL1-HIS3 GAL2-ADE2 met2::GAL7-lacZ) and tested for self-activation. The prey plasmids encoding the LEU2 marker were transformed into a mating type-switched, MATα derivative of PJ69–4A. Yeast containing the prey vectors were mated with MATa yeast containing the bait vectors on YPDA plates for 6 h at 30 °C before being replica-plated onto synthetic dropout (SD)-Trp/Leu plates to select for diploid yeast and (SD)-Trp/Leu/His and (SD)-Trp/Leu/Ade plates to assess the extent of reporter activation. Yeast strains were kindly provided by Dr. Philip James (Department of Biomolecular Chemistry, University of Wisconsin, Madison, WI), whereas bait and prey vectors were provided by Dr. David Markie (Molecular Genetics Laboratory, Pathology Department, Dunedin School of Medicine, Dunedin, New Zealand). NMR Spectroscopy—The NMR spectra were recorded on Bruker Avance-800, Avance-600, and DMX-500 spectrometers. Two-dimensional NOESY, TOCSY, DQF-COSY, 15N HSQC, constant time 13C HSQC, and HNCA HNCO and three-dimensional HNHB, 15N NOESY, and 15N TOCSY were recorded at 298 K. The mixing times chosen were 55 ms for TOCSY and 120 ms for NOESY. Spectra were referenced relative to external sodium 2,2-dimethyl-2-silapentane-5-sulfonate for signals of proton and carbon or liquid ammonia for that of nitrogen. The resonances were assigned using interresidue NOE connectivities by standard procedures (30Wuthrich K. NMR of Proteins and Nucleic Acids. John Wiley & Sons, Inc., New York1986Crossref Google Scholar). Approximately half the Hβ resonances were assigned stereospecifically using a combination of HNHB and DQF-COSY spectra. All of the Val Hγ and Leu Hδ resonances were stereospecifically assigned using a 10% 13C-labeled sample of the K3 RING domain (31Neri D. Szyperski T. Otting G. Senn H. Wuthrich K. Biochemistry. 1989; 28: 7510-7516Crossref PubMed Scopus (568) Google Scholar). Structure Determination—The distance constraints derived from the NOESY spectra were classified into four categories corresponding to interproton distance constraints of 1.8–2.8, 1.8–3.5, 1.8–4.75, and 1.8–6.0 Å, respectively. Hydrogen bonds were identified from an ensemble of structures calculated using only NOE and torsion angle restraints. Protein backbone NH groups whose signals were observed to slowly change with D2O at 283 K were assigned to acceptor groups and had hydrogen bonding restraints imposed. Subsequently, the structures were recalculated to check that the hydrogen bonds could be included without increasing the energy of the calculated structures and without altering the overall structure. By applying a simulated annealing protocol (32Nilges M. Clore G.M. Gronenborn A.M. FEBS Lett. 1988; 239: 129-136Crossref PubMed Scopus (523) Google Scholar) implemented in the XPLOR package (33Brünger A.T. X-PLOR: A System for X-ray Crystallography and NMR. Yale University Press, New Haven, CT1992Google Scholar) on 30 random structures with the 1260 experimental constraints, 28 structures were accepted where no distance violation was greater than 0.25 Å and no dihedral angle violation was larger than 5°. All of the spectroscopic characteristics suggested that the N-terminal residues up to Val6 were unfolded, and hence these residues were excluded from the statistical analysis. Chemical shift mapping was performed by acquiring 1H-15N HSQC spectra from solutions of K3 RING mixed with either UbcH5A or UbcH13 in a 1:1 molar ratio in 50 mm sodium phosphate, 100 mm NaCl, pH 7.4. Electrostatic Potential Calculations—The program GROMACS (34Lindahl E. Hess B. van der Spoel D. J. Mol. Model. 2001; 7: 306-317Crossref Google Scholar) was used to create a Protein Data Bank file containing the charge and radius for each atom. This file was used as an input for the potential program from the MEAD suite (35Bashford, D. (1997) in Scientific Computing in Object-oriented Parallel Environments: First International Conference, Iscope 97, Marina Del Rey, CA, December 8–11, 1997: Proceedings (Ishikawa, Y., Oldehoeft, R. R., Reynders, J. V. W., and Tholburn, M., eds) pp. 233-240, Springer-Verlag, BerlinGoogle Scholar). MEAD was used to solve the Poisson-Boltzmann equation to generate a three-dimensional map of the electrostatic potential field. Surface potential was visualized by coloring a molecular surface according to values in the MEAD-generated map. Solution Structure of the K3 N-terminal C4HC3 Domain— The solution structure of the K3 N-terminal domain was determined by heteronuclear NMR methods using 15N-labeled protein samples. A summary of the experimentally derived restraints used in structure calculations and the refinement statistics are presented in Table I. Backbone and all atom superpositions of 28 conformers are provided (Fig. 2, A and B). The structure, including side chains, is mostly very well defined except for a section of 5 residues outside the domain core at the extreme N terminus (root mean square (r.m.s.) deviation of all atoms from their mean positions is 0.88 Å excluding this region). This section has few interactions with the rest of the protein and may as a result be quite disordered. The main chain torsion angles fall primarily within the preferred regions of conformational space (36Laskowski R.A. Rullmannn J.A. MacArthur M.W. Kaptein R. Thornton J.M. J. Biomol. NMR. 1996; 8: 477-486Crossref PubMed Scopus (4474) Google Scholar).Table ISummary of conformational constraints and statistics for the accepted 28 structures of the K3 RING domainParameterValueStructural constraintsIntraresidue460Sequential283Medium range (2 ≤ |i – j| ≤ 4)148Long range (|i – j| > 4)334Dihedral angle constraints23TALOS constraints80Distance constraints for six hydrogen bonds12Zinc coordination constraints20Total1360Statistics for accepted structuresStatistics parameter ± S.D.r.m.s. deviation for distance constraints0.0110 ± 0.0013 År.m.s. deviation for dihedral constraints0.592 ± 0.081°Mean X-PLOR energy term (kcal mol-1) ± S.D.E (overall)118.72 ± 9.31E (van der Waals)7.62 ± 1.91E (distance constraints)2.24 ± 0.59E (dihedral and TALOS constraints)10.61 ± 0.76r.m.s. deviations from ideal geometry ± S.D.Bond lengths0.0021 ± 0.0002 ÅBond angles0.586 ± 0.016°Improper angles0.380 ± 0.014°Average atomic r.m.s. deviation from the meanstructure ± S.D.Residues 6-59 (N, Cα, C atoms)0.405 ± 0.068 ÅResidues 6-59 (all heavy atoms)0.881 ± 0.088 Å Open table in a new tab The overall topology of the fold is as expected with two interleaved zinc binding sites located at opposite sides of the domain, the first composed of three cysteine and one histidine side chains and the second composed completely of cysteines. Only a very limited amount of regular secondary structure is present (Fig. 3A). This secondary structure consists primarily of a 10-residue α-helix between the sixth and seventh metal ligands, although there is also a very small two-strand β-sheet toward the N terminus of the domain and a single turn of helix between the fourth and fifth ligands. The remainder of the domain is composed of ordered loops, restrained by the two zinc-binding sites, and a small hydrophobic core. The two tryptophan residues invariant among the subfamily of viral C4HC3 domains are not buried within the domain core; rather, they are located together on the surface, with the planes of the two side chains perpendicular to one another (Fig. 3A, purple side chains). This creates a hydrophobic patch on the surface of the molecule that is part of an extended hydrophobic region (Fig. 4A). Adjacent to these tryptophan residues is a small but pronounced hydrophobic pocket. Aside from this nonpolar region, the surface possesses some major charged regions (Fig. 4B). There is a highly negatively charged region that includes the surface abutting zinc site 1 and forms a shallow groove on the protein surface. This region is contiguous with the five N-terminal disordered residues, four of which are also acidic, negatively charged residues (sequence MEDED). The high negative charge in this section of the surface may allow interaction with a basic region elsewhere within the K3 molecule or within target molecules, although the precise mechanism of substrate recognition and ubiquitination is yet to be determined. On the opposite surface of the domain is a positively charged region composed predominantly of four arginine side chains. It is tempting to speculate that this region may interact with the C-terminal region of K3, which contains a high proportion of acidic residues. Structural Comparison with RING and PHD Domains—In order to determine the closest structural homologs of the K3 N-terminal domain, a search was performed against the Protein Data Bank data base using the EMBL DALI server (37Holm L. Sander C. J. Mol. Biol. 1993; 233: 123-138Crossref PubMed Scopus (3566) Google Scholar) (available on the World Wide Web at www.ebi.ac.uk/dali/index.html). Three significantly related structures were found, all of which were RING domains: EL5, a RING-H2 protein (r.m.s. deviation of Cα atoms 3.9 Å, 48 residues aligned), the C4C4 RING NOT4 transcriptional repressor (r.m.s. deviation 2.7 Å, 48 residues aligned), and RAG1 (r.m.s. deviation 3.8 Å, 51 residues aligned). The server found no significant similarity with PHD domains. A direct comparison of the K3 domain with a representative C3HC4 RING (RAG1) and a PHD domain (KAP-1) demonstrates that the K3 domain shares many more features with the C3HC4 RING (Fig. 3, A–C). The overall tertiary structure of the K3 domain is strikingly similar to that of the RAG1 RING domain, both sharing the α-helix conserved within members of the RING domain family. Although the K3 domain lacks the β-sheet located between the two zinc-binding sites in C3HC4-type domains (Fig. 3B), the overall topology of the chain in this region and throughout the core of the domain is very similar. The possession of an α-helix in place of this β-sheet may be a common feature of the viral domains based on their high sequence similarity. The second tryptophan in the K3 domain is also located at a position within the α-helix precisely equivalent to that of a conserved hydrophobic residue in RING domains. The PHD domain of KAP-1 follows a similar path to both the K3 domain and the RAG1 RING up to the sixth metal ligand prior to the RING α-helix. However, after this point, the KAP-1 chain diverges significantly (Fig. 3C, red section of chain), with no α-helix formed and a different main-chain conformation allowing burial of the additional PHD-specific tryptophan residue (23Capili A.D. Schultz D.C. Rauscher I.F. Borden K.L. EMBO J. 2001; 20: 165-177Crossref PubMed Scopus (164) Google Scholar). Therefore, these results establish unambiguously that the K3 C4HC3-type domain (and by inference those of the entire related viral and human families) is a variant type of RING domain. Identification of K3-interacting Ubiquitin-conjugating Enzymes—In order to function as an E3 ubiquitin-protein ligase, a protein must interact with at least one ubiquitin-conjugating enzyme. The KSHV genome itself does not encode any members of the Ubc family, suggesting that K3 exploits a cellular E2 to fulfill its function. In order to determine whether K3 could bind to any E2 proteins, a yeast two-hybrid assay was performed. In the assay, the K3 RING domain was used as bait against a selection of 24 human Ubc proteins. Only four of the possible Ubc proteins were identified as binding partners for K3: the core cellular E2s UbcH5A to -C and UbcH13 (Table II). Of these, only UbcH13 gave a signal on the more stringent Ade selection. A W41A mutant of the K3 RING, which lacks the conserved tryptophan residue important for c-Cbl interaction with UbcH7 (38Joazeiro C.A. Wing S.S. Huang H. Leverson J.D. Hunter T. Liu Y.C. Science. 1999; 286: 309-312Crossref PubMed Scopus (916) Google Scholar), was also included in the screen. Expression of K3 W41A in a cell culture model of K3 action results in an inactive molecule that, while able to bind MHC class I, is unable to promote ubiquitination (21Hewitt E.W. Duncan L. Mufti D. Baker J. Stevenson P.G. Lehner P.J. EMBO J. 2002; 21: 2418-2429Crossref PubMed Scopus (162) Google Scholar). This mutant was found to be incapable of interaction with any of the E2s screened. The C-terminal region of K3 (following the second transmembrane domain) was found not to interact with any E2 enzymes (data not shown).Table IIYeast two-hybrid screen of the K3 RING domain against ubiquitin-conjugating enzymesPrey geneK3 RINGK3 RING W41AAdeaSelection to test for expression of either ADE2 or HIS3HisaSelection to test for expression of either ADE2 or HIS3AdeHisUBE2Anbn, no signal; a, autoactivation; +, signal detectednnnUBE2BnnnnUBE2CnnnnUBE2D1cUBE2D1 to -3 encode UbcH5A to -C; UBE2N encodes UbcH13n+nnUBE2D2n+nnUBE2D3n+nnUBE2E1nnnnUBE2E3nnnnUBE2G1nnnnUBE2G2nanaUBE2HnnnnUBE2InnnnUBE2L3nnnnUBE2L6nnnnUBE2MnnnnUBE2NcUBE2D1 to -3 encode UbcH5A to -C; UBE2N encodes UbcH13++nnUBE2V1 transcript 1nnnnUBE