High precision NMR structure and function of the RING-H2 finger domain of EL5, a rice protein whose expression is increased upon exposure to pathogen-derived oligosaccharides.

EL5, a RING-H2 finger protein, is rapidly induced by N-acetylchitooligosaccharides in rice cell. We expressed the EL5 RING-H2 finger domain in Escherichia coli and determined its structure in solution by NMR spectroscopy. The EL5 RING-H2 finger domain consists of two-stranded beta-sheets (beta1, Ala(147)-Phe(149); beta2, Gly(156)-His(158)), one alpha-helix (Cys(161)-Leu(166)), and two large N- and C-terminal loops. It is stabilized by two tetrahedrally coordinated zinc ions. This structure is similar to that of other RING finger domains of proteins of known function. From structural analogies, we inferred that the EL5 RING-H2 finger is a binding domain for ubiquitin-conjugating enzyme (E2). The binding site is probably formed by solvent-exposed hydrophobic residues of the N- and C-terminal loops and the alpha-helix. We demonstrated that the fusion protein with EL5-(96-181) and maltose-binding protein (MBP) was polyubiquitinated by incubation with ubiquitin, ubiquitin-activating enzyme (E1), and a rice E2 protein, OsUBC5b. This supported the idea that the EL5 RING finger domain is essential for ubiquitin-ligase activity of EL5. By NMR titration experiments, we identified residues that are critical for the interaction between the EL5 RING-H2 finger and OsUBC5b. We conclude that the RING-H2 finger domain of EL5 is the E2 binding site of EL5.

Upon sensing the invasion of microorganisms, plants evoke a variety of defense reactions, including the synthesis of antimicrobial compounds (phytoalexins) and proteins. Many of these biochemical reactions are based on the activation of defenserelated genes. In some cases, the level of protein accumulation and the rapidity of gene induction in the host plant are correlated to the degree of its disease resistance. Therefore, it might be possible to control disease resistance by modifying the regulatory factors for the expression of defense-related genes.
Such regulatory factors could be elements of signal transduction pathways leading from the recognition of invading pathogens to the activation of defense-related genes. Most of the defense responses are reproducible in suspension-cultured cells treated with specific substances called elicitor (1). Chitin fragments (N-acetylchitooligosaccharides) can act as elicitors (2), which induce the transient expression of several "early responsive" genes, such as EL5 (3). EL5 is a RING finger protein, which is structurally related to proteins of the Arabidopsis ATL family. These proteins are characterized by a transmembrane domain (domain I), basic domain (domain II), conserved domain (domain III), and RING-H2 finger domain (domain IV) followed by the C-terminal region with highly diverse amino acid sequences (4). Although some ATL family genes resemble EL5 in being induced in early stages of the defense responses (5), their biochemical function is obscure. Recently, it was shown that the fusion protein of EL5 with maltose-binding protein (MBP) 1 was polyubiquitinated by incubation with ubiquitin-activating enzyme (E1) and ubiquitin-conjugating enzyme (E2). Apparently, EL5 acted as a ubiquitin ligase (E3) and catalyzed the transfer of ubiquitin to the MBP moiety. Further, it was shown that a rice E2, OsUBC5b was induced by elicitor (6). Although these results strongly suggest that EL5 and OsUBC5b have roles in plant defense response through the turnover of protein(s) via the ubiquitin/proteasome system, their molecular function has not been clarified at the atomic level.
The RING finger motif, which is known from many functionally distinct proteins, was first identified in the product of the human gene RING1 (Really Interesting New Gene 1) that is located proximal to the major histocompatibility region on chromosome 6 (7). The RING finger motif is defined by the consensus sequence Cys-X 2 -Cys-X 9 -39 -Cys-X 1-3 -His-X 2-3 -(Cys/ His)-X 2 -Cys-X 4 -48 -Cys-X 2 -Cys, in which X can be any amino acid. It binds two zinc atoms with its Cys and His residues in a unique "cross-brace" arrangement. The invariable spacing between the second and third pair of Cys/His residues indicates conservation of the distance between the two zinc-binding sites (8). The RING finger motif is widely distributed among proteins that play major roles in cell growth and differentiation (9). Certain types of RING finger domains seem to be required for multimerization, while others are elements of proteins involved in ubiquitination (10). Ubiqutination usually results in the formation of a bond between the C terminus of ubiquitin (Gly 76 ) and the ⑀-amino group of a substrate Lys residue. This reaction requires the sequential action of three enzymes: (i) an activating enzyme (E1) that forms a thiol ester with the carbonyl group of Gly 76 in ubiquitin, (ii) a conjugating enzyme (E2) that transiently carries the ubiquitin as a thiol ester, and (iii) a ligase (E3) that transfers the activated ubiquitin from the E2 to the substrate Lys residue. The efficiency and high selectivity of ubiquitination reactions depend on the accuracy of E3 action. All known E3s utilized either of catalytic domains, the RING finger domain or the HECT domain (11). The structures of RING finger domains have been determined by NMR (12,13) and x-ray (14,15). However, the relationship between molecular function and high order structure has not yet been established.
To elucidate the properties of EL5 at the atomic level, we determined the three-dimensional structure of its RING-H2 finger domain in solution by NMR spectroscopy. Furthermore, we characterized its functional properties by an ubiquitination assay in vitro and by NMR titration experiments. Our results illuminate the function of not only EL5 but also the ATL family proteins in general.
All of the clones were transformed into Escherichia coli BL21(DE3) cells. The bacteria were grown at 37°C in Luria Bertani for non-labeled protein and in M9 minimal medium for uniformly 15 N-and 15 N/ 13 Clabeled protein, with 50 g/ml ampicillin. Protein expression was induced by addition of isopropyl-␤-D-thiogalactopyranoside (IPTG) at a concentration of 1 mM. For EL5(129 -181) and EL5(96 -181), ZnSO 4 (final concentration 100 M) was added at the same time. After 3 h the bacteria were harvested by centrifugation, and the pellets were frozen.
For purification of Trx-OsUBC5b, frozen pellets were thawed on ice, resuspended in 20 mM phosphate buffer (pH 7.4) with 500 mM NaCl, and lysed by sonication. Insoluble material remaining after lysis was removed by centrifugation at 27,000 ϫ g for 30 min. The supernatant was loaded onto a 5-ml Ni-chelating column. After washing the column with buffer (20 mM phosphate buffer, pH 7.4, 500 mM NaCl), protein was eluted in buffer using a linear gradient of imidazole (0 -500 mM). Peak fractions were concentrated and applied to a HiLoad Superdex 75 pg 26/60 column equilibrated in buffer (20 mM phosphate buffer, pH 7.4, 100 mM NaCl). Trx tag was removed by cleavage with 28 units of thrombin (Novagen) per about 13 mg of protein at 37°C for 12 h. The solution was applied to a Ni-chelating column. Remaining tag was removed by cleavage with 130 units of enterokinase per about 5 mg of protein at 37°C for 16 h. Then the solution was applied to a HiLoad Superdex 75pg 26/60 column. Peak fractions were dialyzed against NMR buffer (100 mM NaCl, 20 mM Tris-HCl buffer, pH 7.0, 5 mM dithiothreitol) and concentrated using a Centricon spin dialysis tube (30-kDa cutoff; Amicon, Inc.).
NMR Spectroscopy-All NMR spectra were recorded at 35°C on a Bruker DMX750 spectrometer equipped with a 5-mm inverse triple resonance probe head with three-axis gradient coils. 1 H, 13 C, and 15 N sequential resonance assignments were obtained using two-dimensional double resonance and three-dimensional double and triple resonance through-bond correlation experiments (16 -18): two-dimensional 1 H-15 N HSQC, two-dimensional 1 H-13 C CT-HSQC optimized for observation of either aliphatic or aromatic signals, three-dimensional 15 N-separated HOHAHA-HSQC, three-dimensional HNHA, threedimensional HNHB, three-dimensional HNCO, three-dimensional CBCA(CO)HN, three-dimensional HBHA(CBCACO)NH, three-dimensional HNCACB, three-dimensional C(CO)NH, three-dimensional HCABGCO, and three-dimensional HCCH-TOCSY. Stereospecific assignments of ␣and ␤-protons and the methyl side chains of Val and Leu residues were achieved by a combination of quantitative J measurements and NOESY data. 3 J couplings were measured using quantitative two-dimensional and three-dimensional J correlation spectroscopy (18). Interproton distance restraints were derived from multidimensional NOE spectra (16 -18): three-dimensional 15 N-separated NOESY-HSQC spectrum with a mixing time of 100 ms, threedimensional 13 C/ 15 N-separated NOESY-HSQC spectrum with a mixing time of 100 ms, and four-dimensional 13 C/ 13 C-separated HMQC-NOESY-HMQC spectrum with a mixing time of 100 ms. Amide-proton exchange rates were determined by recording a series of two-dimensional 1 H-15 N HSQC spectra at different time points immediately after the H 2 O buffer was changed to D 2 O buffer. These spectra were processed using NMRPipe software (19) and were analyzed using Capp/ Pipp/Stapp software (20). 1 H, 13 C, and 15 N chemical shifts were referenced to HDO (4.68 ppm at 35°C), indirectly to TSP ( 13 C) (21), and to liquid ammonia ( 15 N), respectively (22).
Structure Calculations-NOE-derived interproton distance restraints were classified into four ranges: 1.8 -2.7, 1.8 -3.3, 1.8 -4.3, and 1.8 -5.0 Å, corresponding to strong, medium, weak, and very weak NOEs. The upper limit was corrected for constraints involving methyl protons and methylene protons that were not assigned stereospecifically (23). Hydrogen bond distance restraints were applied to N and O atoms (2.8 -3.3 Å) and to HN and O atoms (1.8 -2.3 Å), in regular secondary structures that had small amide exchange rates. Torsion angle restraints on and were derived from 3 J HNH␣ coupling constants (24), short-range NOEs (␣H i to NH iϩ1 and ␤H i to NH iϩ1 ), and a data base analysis of backbone ( 13 C␣, 13 C␤, 13 CЈ, 1 H␣, 15 N) chemical shifts using the program TALOS (25). Side-chain 1 angle restraints were derived from HOHAHA connectivities and the distances between ␣H and ␤1H and between ␣H and ␤2H, which were estimated from NOEs and ROEs. Values for 3 J H␣H␤ and 3 J NH␤ coupling constants were also taken into consideration. The structures of the RING-H2 finger domain of EL5 were calculated using the hybrid distance geometrydynamical simulated annealing method (26), as contained in X-PLOR 3.1 (27). For structure calculations, we used 755 interproton distance restraints (comprising 290 intraresidue, 185 sequential (i Ϫ j ϭ 1), 73 medium range (1 Ͻ i Ϫ j Ͻ 5), and 207 long range (i Ϫ j Ն 5) restraints) obtained from heteronuclear three-and four-dimensional NOE spectra. In addition to the NOE-derived distance restraints, 18 distance restraints for 9 hydrogen bonds and 140 dihedral angle restraints (46 , 47 , 38 1 , and 9 2 ) were included in the calculation. A final set of 15 lowest energy structures was selected from 100 calculations. None of them had NOE and dihedral angle violations of Ͼ0.5 Å and Ͼ5°, respectively. Structural statistics calculated for the final 20 structures are summarized in Table I. Statistics did not change significantly when the 40 lowest energy structures were used for calculation. The average coordinate of the ensembles of the final 15 structures was subjected to 500 cycles of Powell restrained energy minimization to improve stereochemistry and non-bonded contacts. The structural statistics for the restrained energy minimized average structure are also summarized in Table I 325 and Gly 96 -Val 181 were amplified by PCR using 5Ј-GAATTCGGAGGAGGGGTCGACCCG-3Ј and 5Ј-GGAATTCT-CAATCCGGACATGCGC-3Ј primers or 5Ј-CGAATTCGGAGGAGGGG-TCGACCCG-3Ј and 5Ј-GGAATTCTCAATCCGGACATGCGC-3Ј primers, respectively. The PCR products were digested with EcoRI and inserted into the EcoRI site of pMAL-p2 (New England BioLabs, Beverly, MA) to express MBP-EL5(96 -325) and MBP-EL5(96 -181). To yield MBP for negative control experiments, pMAL-p2 was digested with EcoRI, blunted, and self-ligated. All PCR products were verified by DNA sequencing. The plasmids were introduced into the E. coli strain JM109 to produce the recombinant proteins and were purified by amylose affinity chromatography according to the manufacturer's instructions (New England BioLabs).
In Vitro Ubiquitination Experiments-The ubiquitination reaction was performed in 150 l of ubiquitination buffer containing 300 ng/l bovine ubiquitin (Sigma), 50 ng of mouse E1, 10 ng OsUBC5b as E2, and 250 ng of MBP fusion protein. The reaction buffer was incubated at 35°C for 1 h, and the reaction was stopped by the addition of SDS sample buffer. After boiling for 5 min, the samples were separated by 7.5% SDS-PAGE and subjected to immunoblotting using anti-MBP antibody (New England BioLabs).

Determination of NMR Experimental
Conditions-EL5 consists of a transmembrane domain (domain I), a basic domain (domain II), a conserved domain (domain III), and the RING-H2 finger domain (domain IV) followed by the C terminus (Fig. 1a). For NMR experiments, two protein fragments were prepared: EL5 (129 -181, domain IV), containing the RING-H2 finger domain only, and EL5 (96 -181, domain III ϩ IV), which additionally contained the conserved domain. The 1 H-15 N HSQC spectrum of EL5(129 -181) obtained at 35°C showed dispersed, intense signals, indicating a well-defined molecular structure (Fig. 1b). In the spectrum of EL5(96 -181), the strengths of peaks differed considerably (Fig. 1c). The weak signals corresponded to those of EL5(129 -181), that is, to the RING-H2 finger domain. The intense signals were concentrated in the center of the spectrum, indicating that domain III is a highly flexible random coil structure.
To investigate the role of Zn 2ϩ in the folding of the EL5 RING-H2 domain, Zn 2ϩ was chelated by EDTA. Removal of zinc from the purified domain by addition of EDTA led to a loss of the chemical shift dispersion characteristics of the folded proteins (Fig. 1d). The protein was reversibly refolded upon addition of excess Zn 2ϩ , as indicated by the reappearance of a spectrum that was identical to that in Fig. 1b. Therefore we decided to carry out the purification of recombinant EL5(129 -181) and its NMR analysis in the presence Zn 2ϩ . Side-chain 1 H and 13 C signals in aromatic regions were assigned to aromatic amino acids according to cross-peak patterns observed in 1 H-13 C CT-HSQC and CT-HSQC-relay spectra that were optimized for aromatic side chains. The aromatic H␦ and C␦ signals were then correlated to sequentially assigned backbone signals through cross-peaks observed in three-dimensional 15 N-separated NOESY-HSQC and 13 C/ 15 Nseparated NOESY-HSQC spectra and a four-dimensional 13 C/ 13 C-separated NOESY spectrum.
In order to obtain a high-resolution NMR structure, the resonance assignments were refined by stereospecific assignments of numerous methylene and methyl protons of Val and Leu residues. Stereospecific assignments of Gly ␣-protons were derived from 3 J HNH␣ coupling constants (24), intraresidue NOEs between HN and ␣-protons, and sequential NOEs between Gly ␣-protons (i) and HN protons (iϩ1). ␤-protons were stereospecifically assigned using J couplings obtained from HNHB (29) and HOHAHA-HSQC spectra, and by intraresidue ␣-␤ distances estimated from NOE/ROE spectra, as described by Clore and Gronenborn (17). In some cases, information from local secondary structure and short-range NOEs was used to facilitate stereospecific assignments. Signals from all Val ␥-methyl groups were stereospecifically assigned from estimates of 3 J NC␥ (30), 3 J NH␤ (29), and 3 J HNH␣ coupling constants and from NOEs between Val ␣-␤ and ␣-␥ protons. Leu ␦-methyl group signals were stereospecifically assigned from NOE patterns observed in three-dimensional NOESY, ROESY, and four-dimensional NOESY spectra.
On the basis of essentially complete signal assignments, interproton distance constraints were derived from short range NOE connectives obtained from three-dimensional 15 N-separated NOESY-HSQC, 13 C/ 15 N-separated NOESY-HSQC, and four-dimensional 13 C/ 13 C-separated HSQC-NOESY-HSQC spectra. Dihedral constraints for angles were determined from 3 J HNH␣ coupling constants derived from H␣/HN intensity ratios measured in HNHA experiments. Slowly exchanging amide protons were assigned in 1 H-15 N HSQC hydrogen exchange experiment and were identified as protons involved in interresidue hydrogen bonds. The ␤-strand and helical domains indicated by the NOE and J coupling data were corroborated by secondary structural elements predicted by the observed displacements of C␣ and C␤ chemical shifts from their random-coil values (31). In summary, the analysis indicated that the RING-H2 finger domain of EL5 contained two ␤-strands and an ␣-helix in a ␤␤␣ arrangement.
Tertiary Structure of RING-H2 Finger Domain of EL5-The three-dimensional structure of the RING-H2 finger domain of EL5 was determined by a hybrid distance geometry/dynamicsimulated annealing approach (26) based upon 913 experimental restraints derived from NMR spectroscopy. Structural statistics for the EL5 RING-H2 finger domain are shown in Table  I. The superposed backbone N, C␣, and CЈ coordinates of the final 15 structures (Fig. 2a) were well aligned, except for resi- ͳSAʹ is the restrained regularized mean structure obtained by averaging the backbone coordinates of the individual SA structures best-fitted to each other (residues 132-178), followed by restrained regularization of the mean structure.
b These values were estimated using X-PLOR3.1. The final values of the force constants used for the calculations are as follows: 1000 kcal mol Ϫ1 Å Ϫ2 for bond lengths, 500 kcal mol Ϫ1 rad Ϫ2 for bond angles and improper torsions, 4 kcal mol Ϫ3 Å Ϫ4 for the quartic van der Waals term with van der Waals radii set to 0.75 times the values used in the CHARMM empirical energy function, 50 kcal mol Ϫ1 Å Ϫ2 for NOE-derived restraints and for hydrogen-bonding distance restraints with the ceiling value of 1000 kcal mol Ϫ1 , and 200 kcal mol Ϫ1 rad Ϫ2 for dihedral angle restraints.
c None of the structures exhibited interproton distance violations Ͼ0.5 Å and dihedral angle violations Ͼ5°. The distance restraints comprise 290 intraresidue, 185 sequential (͉i Ϫ j͉ ϭ 1), 74 medium-range (1 Ͻ ͉i Ϫ j͉ Ͻ 5) and 206 long-range (͉i Ϫ j͉ Ն 5) NOEs, as well as 54 hydrogen-bonding restraints for 9 hydrogen bonds. The dihedral angle restraints involve 71 , 73 , 37 1  The three-dimensional structures of a few RING finger domains have been determined. The molecular function of two of them, RAG1 and c-Cbl, is known. The V(D)J recombinationactivating protein, RAG1, is a dimer. Its dimerization domain consists of a zinc finger and a RING finger (14). It also contains a unique binuclear zinc cluster instead of the mononuclear zinc site in the RING finger. The determination of the crystal structure of c-Cbl bound to a specific ubiquitin-conjugating enzyme (E2), UbcH7, has revealed that the RING finger domain of c-Cbl recruits the E2 (15). The RING finger domain of EL5 is structurally similar to those of RAG1 and c-Cbl (Fig. 3a). The backbones of the secondary structural elements of the proteins are almost superimposable. The backbone (N, C␣, CЈ, O) atomic r.m.s.d. values for the ␤-strandand ␣-helix-forming residues between the EL5 RING finger domain and RAG-1 or c-Cbl are 0.73 and 1.30 Å, respectively.
Although the overall three-dimensional structures of the RING finger domains of RAG1 and c-Cbl are similar, their molecular functions are different. To obtain information on the molecular function of the RING-H2 finger domain of EL5, we compared the RING finger domains of the three proteins at the atomic level. The RAG1 RING finger domain has one additional ␣-helix each at its N-and C terminus (Fig. 3a). The RAG1 dimer interface is stabilized by an extensive hydrophobic core containing two clusters of three Phe residues in these additional ␣-helices (14). In accordance with the observation that c-Cbl is a monomer, its RING finger domain lacks the dimerstabilizing N-and C-terminal ␣-helices. As the EL5 RING-H2 finger domain resembles c-Cbl in this respect, it is not expected to form dimers. The c-Cbl RING finger domain binds E2 along a hydrophobic groove (indicated by arrows in Fig. 3b) that is formed by the ␣-helix of the ␤␤␣ structure and the two zincchelating N-and C-terminal loops (Ref. 15, compare electron potential map in Fig. 3b). On the contrary, the RING finger domain of RAG1, which is a DNA-binding protein without ubiquitin ligase (E3) activity (32), does not possess this groove. Its N-and C-terminal loops are closer together with the remaining space occupied by the side chain of Arg 57 (Fig. 3b). Thus, the groove that enables the hydrophobic interaction between c-Cbl and E2 is occupied by a basic group (Arg 57 ) in RAG1, preventing interactions between RAG1 and E2. The residues in c-Cbl that form the hydrophobic contact with E2 (Ile 383 , Cys 404 , Ser 407 , Trp 408 , Ser 411 , and Pro 417 ) (15) are mostly conserved in the RING-H2 finger domain of EL5 (Val 136 , Cys 161 , Trp 165 , Ser 168 , and Pro 173 ) as shown in Fig. 3b. The spatial arrangements of these hydrophobic residues is quite similar in the two molecules (Fig. 3b), suggesting that EL5 should bind E2 similarly as c-Cbl does.
The EL5 RING-H2 Finger Domain Catalyzes Auto-ubiquitination in Vitro-Takai et al. (6) showed that the fusion protein of EL5(96 -325) and maltose-binding protein (MBP-EL5(96 -325)) was polyubiquitinated by incubation with ubiquitin, ubiquitin-activating enzyme (E1), and UbcH5a or OsUBC5a/b, a rice E2. Thus, the EL5 domains III, IV, and the C-terminal region are sufficient to catalyze ubiquitin transfer to the MBP moiety in cooperation with E2. It was also demonstrated that replacement of Cys 153 by Ser abolished the E3 activity. Our structural analysis suggested that the RING-H2 finger domain of EL5 binds E2. We expressed the EL5 RING-H2 finger domain as a fusion protein with MBP (MBP-EL5(96 -181) and MBP-EL5(129 -181), respectively) and determined its E3 activity. When MBP-EL5(129 -181) was incubated with ubiquitin, E1 and OsUBC5b in an in vitro ubiquitination assay, only one ubiquitinated derivative of the fusion protein was detected. When MBP-EL5(96 -181) was used in the same experiment, several ubiquitinated derivatives of the fusion protein were observed (Fig. 4, lanes 3 and 4). The latter result was similar to that obtained with MBP-EL5(96 -325) (Fig. 4, lanes 1 and 2). These findings showed that only the RING-H2 finger domain was sufficient to bind E2, but that the polyubiqutination chain reaction was disturbed in the small construct, MBP-EL5(129 -181), probably because the MBP moiety sterically hindered the development of a polyubiquitin chain. Therefore we decided to carry out the in vitro ubiquitination assay using MBP-EL5(96 -181). Since Zn 2ϩ is needed for the correct folding of the EL5 RING finger domain, we tested its effect in the ubiquitination assay. No ubiquitination was detectable in the presence of EDTA (Fig. 4, lanes 5 and 6), but the activity was restored by addition of excess ZnSO 4 (Fig. 4, lanes 7 and 8). Thus, Zn 2ϩ facilitates effective EL5-mediated ubiquitination by structurally stabilizing the EL5 RING-H2 finger domain.
Identification of Residues in the EL5 RING-H2 Finger Domain That Interact with OsUBC5b-To identify EL5 residues that interact with E2, we recorded amide 15 N and 1 H chemical shifts of the EL5 RING-H2 finger domain as a function of the concentration of OsUBC5b. Residues located close to OsUBC5b in the E3/E2 complex were anticipated to exhibit large changes in their chemical shifts, because the shift depends on the residue's magnetic environment. The HSQC signal of EL5(96 -181) had indicated that the conserved domain remained unchanged (i.e. unstructured) in the presence of OsUBC5b, implying that it was not involved in the EL5/OsUBC5b interaction. Therefore we used EL5(129 -181) in the NMR titration experiments.
In these experiments, the extreme broadening of the signals observed in several residues indicated intermediate exchange on the NMR time scale. From the titration curve that was plotted on the basis of the chemical shift variations observed in the RING-H2 finger domain, the stoichiometry between RING-H2 finger domain and OsUBC5b (1:1) and the dissociation constant (K d ϳ1 ϫ 10 Ϫ5 M) were calculated. Seven residues (Val 136 , Cys 137 , Ala 147 , Arg 148 , Glu 160 , Thr 171 , and Leu 174 ) displayed significant chemical shift perturbations upon complex formation with OsUBC5b (Fig. 5a). The amide signals of 5 residues (Leu 138 , Val 162 , Asp 163 , Met 164 , and Trp 165 ) were not detectable due to extreme broadening of the signal. We ascribe this phenomenon to an exchange between the free and bound forms on the chemical shift time scale. The residues with the chemical shifts most sensitive (chemical shift perturbations, ⌬␦ ϭ ⌬␦ 1HN ϩ 0.1 ⌬␦ 15N greater than 0.15 ppm) to complex formation were mapped on the free form structure of the EL5 RING-H2 finger domain (Fig. 5b). These residues are all localized on one side of the molecule where they form the E2 binding surface. The location of this binding site is in good agreement with that of the binding site of the c-Cbl RING finger domain for E2 (15). DISCUSSION Intense interest in RING finger proteins has arisen because of their role in human disease and their widespread occurrence.  4) and MBP (lanes 9 and 10) were incubated at 35°C with ATP, ubiquitin, E1, and OsUBC5b (E2) for 1 h, and were then subjected to SDS-PAGE followed by immunoblotting with an anti-MBP antibody. MBP-EL5(96 -181) was incubated with EDTA before starting the assay (lanes [5][6][7][8], in the absence (lanes 5 and 6) or in the presence of an excess amount of ZnSO 4 (lanes 7 and 8). Ladders of bands at higher molecular weights (lanes 2, 4, and 8) indicate the occurrence of ubiquitination.
FIG. 5. a, NMR chemical shift perturbation of the EL5 RING-H2 finger domain upon binding to OsUBC5b. Changes in the NMR chemical shifts of RING-H2 finger domain (⌬␦) as induced by complex formation with OsUBC5b, were calculated by the function ⌬␦ ϭ ⌬␦ HN (pink) ϩ 0.10 ⌬␦ 15N (magenta). The light blue bars indicate resonances broadened beyond recognition. b, two views of the surface of the EL5 RING-H2 finger domain. Residues that showed highly sensitive backbone amide chemical shift (⌬␦ Ͼ 0.15 ppm) are colored magenta. Residues marked in red were not detectable due to extreme broadening of the signal after binding of OsUBC5b. The left view is in the same orientation as the models in Fig. 3, while the right view is from the opposite side.
In higher plants, numerous genes for RING finger proteins have been identified. For example, 387 RING finger proteins have been predicted from a search of the Arabidopsis genome data base (33). Although RING finger proteins play important roles in plants, there is no information on their structure and/or the relationship between the structure and function.
The EL5 RING-H2 Finger Domain Interacts with UbcH4/5a-Type Ubiquitin-conjugating Enzymes-Our structural-based analysis suggested that the RING-H2 finger domain of EL5 could bind E2. NMR titration experiments and an in vitro ubiquitination assay showed that the RING-H2 finger domain of EL5 could bind the E2, OsUBC5b, along a groove formed by a cluster of hydrophobic residues. The primary sequence of the EL5 RING-H2 finger domain shows high similarity with other RING finger domains that interact with UbcH4/5a (Fig. 6a). In fact, the fusion protein of EL5 with MBP was polyubiquitinated by incubation with E1 and UbcH5a. OsUBC5b, the E2 used in the present study, is highly similar to UbcH4/5a (6). Thus, available evidence suggests that the EL5 RING-H2 finger domain belongs to the group of UbcH4/5a-type E2 binding domains. Consequently, EL5 is a UbcH4/5a-type E2 binding E3.
Identification of Residues in the RING Finger Domain That Are Critical for E2 Interaction-The primary sequences of various RING finger domains are compared in Fig. 6a. Four groups can be distinguished on the basis of RING finger domain interaction with other proteins. The first group consists of ubiquitin ligases that cooperate with UbcH4/5a, but not with UbcH7/8; it includes EL5 and AO7 (34). Members of the second group, including HHARI (35), interact with UbcH7/8, but not with UbcH1 and UbcH5. c-Cbl forms a group on its own; it is E3interacting both with UbcH7 (36) and UbcH4 (37). The fourth group includes KAP-1 (38) and RAG1 (14). Its members appear to function in the formation of macromolecular assemblages.
The result of the EL5-E2 NMR titration experiment detects an altered chemical environment for amide groups on EL5. The observed chemical shift changes reveal direct contacts as well FIG. 6. a, alignment of amino acid sequences of RING finger domains. The Cys and His residues responsible for zinc binding are marked yellow. Residues that are conserved or conservatively exchanged with respect to the EL5 RING-H2 finger domain are marked blue and light blue, respectively. Asterisks indicate residues that are crucial for the binding of E2 by EL5, c-Cbl and CNOT4, respectively. b, EL5 RING finger domain. Residues that exhibited a high sensitive amide chemical shift in NMR titration experiments are highlighted. Residues that are well and poorly conserved between E2 binding RING finger domains (Group 1, 2, and 3) are shown in magenta and gray, respectively. Pro 173 , which cannot be detected in a 15 N HSQC spectrum, is shown in yellow. c, c-Cbl RING finger domain. Residues that interact with UbcH7 are colored magenta (14). d, CNOT4 RING finger domain. Residues that showed a high sensitive amide chemical shift in NMR titration experiment, and those that are critical for UbcH5b interaction are colored magenta (39).