Báo cáo khoa học: Solution structure and backbone dynamics of the XPC-binding domain of the human DNA repair protein hHR23B docx

Solution structure and backbone dynamics of the XPC-binding domain of the human DNA repair protein hHR23B Byoungkook Kim 1, *, Kyoung-Seok Ryu 2, *, Hyun-Jin Kim 1 , Sung-Jae Cho 1 and Byong-Seok Choi 1 1 Department of Chemistry, and National Creative Research Initiative Center for the Repair System of Damaged DNA, Korea Advanced Institute of Science and Technology, South Korea 2 Korea Basic Science Institute, Daejon, South Korea Nucleotide excision repair (NER) is an important pathway for the removal of DNA lesions caused by diverse environmental factors, such as UV irradiation and chemical modifications [1,2]. There are two human homologs (A and B) of the yeast Rad23 protein (hHR23A and hHR23B), both of which can form a complex with the xeroderma pigmentosum group C protein (XPC) [1,3]. Recent in vitro and in vivo studies point to a role for the XPC–hHR23B complex as the initiator of global genomic NER [1,4]. Although the precise functions performed by hHR23A and hHR23B alone in human NER have not yet been determined, Keywords hHR23B; nucleotide excision repair; stress- inducible; structure; xeroderma pigmentosum group C protein Correspondence B S. Choi, Department of Chemistry and National Creative Research Initiative Center for Repair System of Damaged DNA, Korea Advanced Institute of Science and Technology, Yusong-Gu, Gusong-Dong 373-1, Daejon 305-701, South Korea Fax: +82 42 869 2810 Tel: +82 42 869 2868 E-mail: byongseok.choi@kaist.ac.kr *Byoungkook Kim and Kyoung-Seok Ryu contributed equally to this work Note The atomic coordinates of the bundle of 20 conformers have been deposited in the RCSB Protein Data Bank with entry code 1PVE (Received 9 November 2004, revised 4 March 2005, accepted 17 March 2005) doi:10.1111/j.1742-4658.2005.04667.x Human cells contain two homologs of the yeast RAD23 protein, hHR23A and hHR23B, which participate in the DNA repair process. hHR23B hou- ses a domain (residues 277–332, called XPCB) that binds specifically and directly to the xeroderma pigmentosum group C protein (XPC) to initiate nucleotide excision repair (NER). This domain shares sequence homology with a heat shock chaperonin-binding motif that is also found in the stress- inducible yeast phosphoprotein STI1. We determined the solution structure of a protein fragment containing amino acids 275–342 of hHR23B (termed XPCB–hHR23B) and compared it with the previously reported solution structures of the corresponding domain of hHR23A. The periodic position- ing of proline residues in XPCB–hHR23B produced kinked a helices and assisted in the formation of a compact domain. Although the overall struc- ture of the XPCB domain was similar in both XPCB–hHR23B and XPCB–hHR23A, the N-terminal part (residues 275–283) of XPCB– hHR23B was more flexible than the corresponding part of hHR23A. We tried to infer the characteristics of this flexibility through 15 N-relaxation studies. The hydrophobic surface of XPCB–hHR23B, which results from the diverse distribution of N-terminal region, might give rise to the func- tional pleiotropy observed in vivo for hHR23B, but not for hHR23A. Abbreviations hHR23B, human homolog B of yeast Rad23; NER, nucleotide excision repair; RMSD, root mean square deviation; STI1, stress-inducible, heat shock chaperonin-binding motif; UBA, ubiquitin-associated domains; UbL, ubiquitin-like domain; XPC, xeroderma pigmentosum group C protein; XPCB, XPC binding. FEBS Journal 272 (2005) 2467–2476 ª 2005 FEBS 2467 many reports suggest that these proteins stabilize the XPC protein by protecting it from 26S proteasome- dependent protein degradation [5–7]. Another bio- chemical analysis of the damage-recognition process in NER revealed that hHR23B is necessary for XPA ⁄ replication protein A-mediated displacement of the XPC–hHR23B complex from damaged DNA [8]. Both hHR23A and hHR23B have four well-defined functional domains (Fig. 1A). These include an N-ter- minal ubiquitin-like (UbL) domain, the XPC-binding domain, and two ubiquitin-associated domains (UBA1 and UBA2) [9,10]. The UbL domain has a high bind- ing affinity for polyubiquitin binding site 2 of the human S5a protein, which is supposed to serve as a shuttle delivering polyubiquitinated, degradable protein substrates to the proteasome [10]. The UBA domains occur in many enzymes involved in the ubiquitination pathway and in cell-cycle check points [5,11]. It also has been shown recently that the intramolecular inter- action between the UbL and UBA domains of hHR23B may regulate NER by modulating the pro- teolysis of XPC [12,13]. The XPC-binding domain of hHR23B (residues 277–332, referred to herein as XPCB–hHR23B) houses an XPC-stimulation activity that functions in a manner similar to that of full-length hHR23B; this activity par- ticipates in DNA damage discrimination in vitro and in the enhancement of cell survival in vivo [9]. From the amino acid sequence, Masutani et al. [9] showed that the XPC-binding domain of hHR23B has a partly repetitive character [that is, it contains various versions of the sequence (P)QLLQQ(I)] and a highly amphi- pathic nature, which is evident when the domain is represented as a helical wheel [9]. XPC-binding domain-like sequences are also found in protein linking IAP with cytoskeleton (PLIC), which also has both UbL and UBA domains [14]. Consensus motifs from proteins with XPC-binding domain-like sequen- ces are shown in Fig. 1B. It is interesting that the XPC-binding domain has been classified as a heat shock chaperonin-binding motif, which is also found in the stress-inducible phosphoprotein STI1 [15,16]. The presence of sequence similarity between the XPCB domain and STI1 is not surprising, because XPC is also induced by a kind of cellular stress (i.e. DNA damage). Studies have also suggested that hHR23B has more diverse in vivo functions than hHR23A. For example, only hHR23B was codetected with XPC protein during the affinity fractionation of mammalian crude extract, using an immobilized glutathione-S-transferase (GST)– S5a fusion protein [17]. S5a can bind to both polyubi- quitinated proteins and the N-terminal ubqiuitin-like domain of hHR23A ⁄ B [18]. This might occur because only a small fraction of cellular hHR23A exists in a complex with XPC [7]. Experiments with knockout mice that carry a homozygous deficiency in either the mHR23A or mHR23B gene showed that these two pro- teins are functionally redundant in terms of response to DNA damage by UV light. The XPC protein was not detected in the double knockout cell line, but could be detected after treatment with a proteasome inhibitor. It is interesting that only the mHR23B knockout mouse showed defects in postnatal growth, suggesting that mHR23B may have functions beyond those related to XPC and DNA repair [19]. Here, we report the three-dimensional solution struc- ture of the XPCB–hHR23B (275–342) fragment (XPCB–hHR23B), which contains the XPC interaction domain, and compare this with the recently reported structure of XPCB–hHR23A [13,20]. Both overall A B Fig. 1. Proteins with STI1-homologous domain and the sequence alignment of the XPCB domain with its homologs. (A) Domain presenta- tions of the STI1-homologous proteins, showing the UbL (ubiquitin-like), UBA (ubiquitin associated), STI1 (stress inducible) and TPR (tetratrico peptide repeat) domains. (B) Multiple sequence alignments of the XPCB domains of hHR23B and hHR23A with other STI1-homologous domains from yeast STI1 and DSK2, and from human PLIC2 were obtained from a simple modular architecture research tool ( SMART) [16]. The Pro residues are indicated in bold. Dynamic structure of XPC-binding domain of hHR23B B. Kim et al. 2468 FEBS Journal 272 (2005) 2467–2476 ª 2005 FEBS structures are similar, but the N-terminal part (275– 283) of XPCB–hHR23B is more flexible than that of hHR23A. 15 N-Heteronuclear relaxation analyses were performed with XPCB–hHR23B to gain precise infor- mation concerning the flexibility of the NH vectors along the peptide chain. We analyzed the diverse hydrophobic surfaces of XPCB–hHR23B, which result from the flexible N-terminal region, and attempt to determine the structure of the XPC-binding surface. Results and Discussion Although the minimal domain of hHR23B (277–332) binding to the XPC protein (XPCB–hHR23B) has been reported previously [9], the solubility of this frag- ment was too low for NMR experiments (< 0.2 mm, data not shown). To increase the solubility of XPCB– hHR23B for NMR analysis, we added two more N-terminal amino acids, P275 and L276, which were selected according to the sequence alignment of STI1 homologs (Fig. 1); as such, XPCB–hHR23B could be concentrated up to 1 mm. Still, the peak of the 15 N-HSQC spectrum was broad, and its intensity was not uniform with increasing concentrations of protein (data not shown). The line-broadening observed at higher protein concentrations seemed to result from nonspecific hydrophobic interactions between XPCB– hHR23B subunits, which were due to the higher con- tent of hydrophobic residues in the XPCB. To reduce these intermolecular interactions, we either included additional amino acids from the C-terminal part of XPCB–hHR23B (residues 333–342, QEAGGQGG GGG) or solubilized the protein fragment in 10 mm CHAPS buffer. The combination of these two procedures markedly improved the quality of the 15 N- HSQC spectra for XPCB–hHR23B (Supplementary material, Fig. S1). The presence of CHAPS had a negligible effect on the chemical shift values in the 15 N-HSQC spectra (data not shown). Although the peak regions of Hb and Hc in the 15 N-HSQC spectra were complicated because of the high content of Gln, Leu and Glu, we were able to accomplish complete side-chain assignment with the aid of additional HCCH-COSY spectrum. After the automatic NOE assignment and structure calculation using cyana [21], we obtained the 1242 assigned dis- tance restraints from the 1948 NOE cross-peaks. For the energy-minimized final structure, we used the amber7 program after pseudo-atom correction for the obtained distance restraints. A stereoview of the calcu- lated XPCB–hHR23B structures is shown in Fig. 2A, and the statistics of structure calculation are summar- ized in Table 1. XPCB forms a very compact, roughly five-helix bundle: (a) helix 1 consists of residues E277 to L279 or R280 and assumes the geometry of a 3 10 helix; (b) helix 2 consists of residues F285 to I292, of which the front boundary was slightly variable; (c) helix 3 spans residues P296 to E309, which has a 3 10 helix that contains residues P296 to L298; (d) helix 4 is formed by residues P311 to S318; and (e) helix 5 consists of residues Q321 to L328. The C-terminal Gly-rich region is a flexible random coil, as was predicted by the chemical shift index (Fig. 3A). The Pro residues are likely to be the cornerstones for the boundaries of the helices, and their periodical presence made the XPCB domain fold in a compact manner by introducing helical breaks and turns or kinks (Figs 1 and 2A). The N-terminal part of XPCB–hHR23B ABC Fig. 2. NMR structure of the XPCB domain of hHR23B. (A) Stereoview of the 12 superimposed structures of XPCB–hHR23B. All Pro resi- dues conserved among the STI1 homologs are marked in blue, and the Pro residues not conserved among the STI1 homologs are marked in cyan. (B) Ribbon presentation of the XPCB–hHR23B three-dimensional structure. (C) Seven structures of XPCB–hHR23B (yellow) and of XPCB–hHR23A (green, from Walters et al. [13]) are superimposed. The red and blue residues are amino acids that differ between XPCB– hHR23B and XPCB–hHR23A. The side chains of residues with orientations that differ between hHR23B and hHR23A are shown. Also, b-N and b-C are the N- and C-termini of XPCB–hHR23B; a-N and a-C are the N- and C-termini of hHR23A. B. Kim et al. Dynamic structure of XPC-binding domain of hHR23B FEBS Journal 272 (2005) 2467–2476 ª 2005 FEBS 2469 (amino acids 275–283) was not well converged, sug- gesting that this part of the molecule was more flexible than the rest of the polypeptide. Indeed, the relatively negligible long-range NOE cross-peaks in this segment fit well with this hypothesis (data not shown). For the rest of XPCB–hHR23B (amino acids 284–332), the backbone regions were well converged, and the root mean square deviation (RMSD) was 0.63 A ˚ (Table 1). Although, the overall three-dimensional structure of XPCB–hHR23B was very similar to that of XPCB– hHR23A (Fig. 2C), it was difficult to identify the flexi- bility in the N-terminal region of the two previously reported NMR structures of XPCB–hHR23A [13,20]. The sequence homology between the XPCB regions of hHR23A and hHR23B is very high (88%), and the nine amino acids that differ [namely, N281(A) to D237 (B), Q287 (A) to N243 (B), I291 (A) to V247 (B), S297 (A) to A253 (B), I306 (A) to L262 (B), R308(A) to Q264 (B), Q319 (A) to R275 (B), H323 (A) to Q279 (B) and V332(A) to P288 (B)], simply increased the flexibility of the backbone segment, including helix 1 (Fig. 2C). Although our overall structure for XPCB–hHR23B was more similar to the structure of XPCB–hHR23A determined by Waters et al. (RMSD,  2.12 A ˚ ) [13] than that by Kamionka and Feigon (RMSD,  3.43 A ˚ ) [20], this trend is reversed when Table 1. Statistics of structure calculation. RMSD, root mean square deviation. Parameter Value Total NOE distance restraints (#) 717 i, i 241 i, i + 1 33 i, i + 2 121 i, i + 3 41 i, i + 4 89 Long-range NOE (|i – j| > 4) 1242 Total angle restraints (#) 70 / (TALOS + experimental 3 J HNHa )42 w (TALOS) 28 20-structures from AMBER (kCalÆmol )1 ) Total energy )3087.9 ± 12.3 E (NOE violation) 15.1 ± 1.6 E (Angle violation) 0.0 ± 0.0 AMBER FF99 force field E (Non-restraint) a )3103.1 ± 11.7 20-structures PROCHECK analysis (%) Most favored regions 78.5 Additionally allowed regions 20.2 Generously allowed regions 1.3 Disallowed regions 0.0 RMSD (A ˚ ) Backbone 0.83 ± 0.306 Residues 275–330 All atoms 2.00 ± 0.350 RMSD (A ˚ ) Backbone 0.90 ± 0.291 Residues 275–283 All atoms 2.43 ± 0.466 RMSD (A ˚ ) Backbone 0.45 ± 0.114 Residues 284–330 All atoms 1.55 ± 0.204 a Summation of energies defined by AMBER force field. A B C D E F G H Fig. 3. Relaxation studies of XPCB–hHR23B at 500 MHz field. (A) The chemical shift index (CSI) clearly shows the well-defined five helical regions. R 1 (B), R 2 (C), and 15 N- 1 H heteronuclear NOE (D) values were used to obtain the ordered parameters (E), the internal correlation times (F), the exchange rates (G), and the model types (H) from the TENSOR2 analysis. The values marked by asterisks (*) in (C) and (G) are 24.0 ± 0.94 (s )1 ) and 14.8 ± 1.62 (s )1 ), respect- ively. Dynamic structure of XPC-binding domain of hHR23B B. Kim et al. 2470 FEBS Journal 272 (2005) 2467–2476 ª 2005 FEBS considering the local structure. For example, the struc- ture of hHR23B I306 differed significant from that of the corresponding residue (L262) of hHR23A. I306 was well ordered and was part of a hydrophobic core, whereas L262 is somewhat flexible and extrudes out from the molecule core (Fig. 2C). However, the confi- guration of L262 from the XPCB–hHR23A structure of Kamionka and Feigon shows solvent exposure sim- ilar to that of I306 in XPCB–hHR23B (Fig. 5). The C-terminus of XPCB–hHR23B extends in a different direction than those of both XPCB–hHR23As. This may result from the altered amino acid sequence (P331–V332 in hHR23B vs. P287–P288 in hHR23A), because the consecutive prolines of XPCB–hHR23A restrict the direction of the C-terminus in a different way to the Pro–Val sequence in hHR23B. We next performed 15 N-relaxation experiments and tensor2 analysis [22] to examine in more detail the flexible characteristics of the 275–284 segment of hHR23B. Because of the high quality of our 15 N-HSQC spectra, 54 of the 58 protonated backbone nitrogen atoms were available for relaxation measure- ments. The values of 15 NR 1 ,R 2 and 15 N- 1 H hetero- nuclear NOEs are shown in Fig. 3. The results obtained for residues 336–342 (GGQGGGG) were omitted from Fig. 3, as the NOE values of these resi- dues were in the far negative and their chemical shift index values were assigned to 0, indicating that this region is very flexible (Fig. 3A). Excepting the C-ter- minal segment from E330 to G342, the heteronuclear NOE and R 2 measurements showed a uniform distri- bution over most of the amino acid sequence, demon- strating values typical for a globular protein. The molecular size of XPCB–hHR23B (275–342) is 8.14 kDa, and the initial R 2 ⁄ R 1 ratio values obtained from 600 and 500 MHz NMR machines corresponded to an overall correlation times (s init c ) of 5.7 and 6.9 ns, respectively. The higher correlation time at 500 MHz may be caused by slight experimental differences, including a slight difference in buffer conditions and the lower temperature used for the experiment at 500 MHz (25 instead of 27 °C), which gives better HSQC spectra. Because of the known dependence of the overall correlation time s c on the molecular size of various proteins [23], both s c values of the XPCB are well matched to those of spherical molecules of similar size with a smooth surface moving in an ideal liquid. By increasing the error ranges from the fitting to sin- gle exponential decay (1.75 and 2.0 times for the relax- ation data at 500 and 600 MHz, respectively), it was possible to find a proper diffusion tensor model. Fol- lowing the determination and assignment of appropri- ate spectral density function models for each residue, the overall correlation time was again optimized using tensor2. Residue-specific models were selected to minimize the overall v 2 with respect to s c . The analyzed results of both relaxation data at 600 and 500 MHz were quite similar (Supplementary Fig. S2; and Fig. 3), but the values of the order parameters at 600 MHz were low at the residues 278, 281, 322 and 323. Inspection of (S 2 , s i ) parametric space for these residues (assigned to model type 2, 4, and 5) showed a very diverse distribution in the Monte Carlo simula- tion using tensor2. Similar inspection of the residues 278 and 281 in the 500 MHz data showed less diverse distribution in the (S 2 , s i ) parametric space and the residues, 322 and 323 were assigned to model type 1. It is possible that the quality of relaxation data at 500 MHz is better than at 600 MHz or that the field- dependent motion results in a different model type for the specific residues. With respect to the regions of the XPC-binding domain that have well-defined secondary structure (Fig. 3A), the order parameter (S 2 ) was higher than 0.85, whereas smaller values were usually obtained for the less ordered C-terminal part of XPCB–hHR23B (Fig. 3E). Although a model-free ana- lysis of is not based on the exact physical motions of the molecule, it can provide important information regarding the dimensionless characteristics of the mole- cule’s backbone. Most residues are well fitted to model type 1–4, which can be described by the combination of three terms; an order parameter (S 2 ), an internal correlation time (s i ), and a conformational exchange term (R ex ). From the relaxation studies, we identified that N-terminal segment of XPCB–hHR23B (275–283) has a distinctive exchange process (Fig. 3). Interest- ingly, the heteronuclear NOE values of this segment were almost similar to other well-refined regions, in spite of the presence of internal motion (s i ) and a remarkable exchange process (R ex ). The NOE values combined with the presence of internal motion and the exchange terms in N-terminal segment show that the N-H vectors of each residue of this region are correla- ted in motion on the axis of helix 1 (i.e. this segment is not independently flexible). The XPC–hHR23B complex was reported to be sta- ble even in 0.3 m salt, and the binding mode is driven mainly by hydrophobic interactions [9]. Our results show that XPCB–hHR23B has a more diverse hydro- phobic surface than the XPCB–hHR23A, because of the heterogeneous distribution of N-terminal segment (Fig. 4). Two major hydrophobic trails (HTs) (HT1 and HT2) were identified in XPCB–hHR23B, one between helix 2 and helix 3, and one between helix 3 and helix 4–5. These two trails were linked by a hydro- phobic linker patch between helix 2 and helix 3 (P296, B. Kim et al. Dynamic structure of XPC-binding domain of hHR23B FEBS Journal 272 (2005) 2467–2476 ª 2005 FEBS 2471 L298, L299, and P300) and thus form a U-shaped hydrophobic surface (Fig. 4A,B). The boundary and size of the hydrophobic linker patch and HT2 were well conserved in all calculated structures, because these hydrophobic surfaces were formed by the well- converged parts of XPCB–hHR23B. However, because of secondary effects from the divergent positioning of the flexible N-terminal region (275–283), the boundary and size of HT1 were variable; in contrast, this vari- ability of the HT1 region was not detected for XPCB– hHR23A in the Kamionka and Feigon study [20]. Moreover, the HT1 region was not observed in an ear- lier structure of XPCB–hHR23A reported by Waters et al. [13]. The effect of the motion of the N-terminal region on the hydrophobic surface area is more obvi- ous in Fig. 4C. The hydrophobic interior appeared to be relatively well covered by helix 1 in Structures 1 and 2, but its larger part was exposed to the outside in Structure 3, because of reduced shielding by helix 1. We calculated the solvent-accessible areas of XPCB– hHR23A and XPCB–hHR23B for each residue. It is clearly shown that the variation in the solvent-access- ible area of the N-terminal part of XPCB–hHR23B is markedly higher than that of XPCB–hHR23A (Fig. 5). The total surface area of all these structures is very similar; XPCB–hHR23B,  41 nm 2 , XPCB–hHR23As from Waters et al. and Kamionka and Feigon,  41 and  43 nm 2 , respectively. The diverse hydrophobic surface, which resulted from the heterogeneous distri- bution of the N-terminal segment of XPCB–hHR23B, was not observed in XPCB–hHR23A and could explain the inferior solubility of XPCB–hHR23B com- pared with that of XPCB–hHR23A [20]. We tried to express the entire hHR23B-binding domain of human XPC (amino acids, 496–734) in Escherichia coli so as to identify the precise XPC con- tact sites in the XPCB–hHR23B. However, this domain was expressed in an insoluble form with var- ious expression vectors (N- and C-terminal His-tag, GST-tag, and thioredoxin-tag) and in a number of E. coli strains. It is possible that, if we expressed por- tions of the hHR23B-binding domain of human XPC, the peptide segments we selected would be more amen- able to purification in a soluble form. Therefore, we Fig. 4. Surface presentations of the three representative structures of XPCB–hHR23B. The hydrophobic surface is presented in yellow, and the polar and charged surfaces are shown in white. HT1 and HT2 are hydrophobic trail 1 and 2, respectively. HL denotes the hydrophobic linker patch. Some structures from the AMBER7 calcula- tion showed a short distance between the side chains of Q284 and E309 (marked by the asterisk). This is an artifact from the electro- static force field of AMBER7, because no NOE cross-peak between these side chains was observed. Fig. 5. Solvent-accessible surface areas of XPCB–hHR23A and B. The solvent-accessible surface areas and their deviations for each residue are shown for two XPCB–hHR23A structures; Waters et al. [13] (A1), Kamionka and Feigon [20] (A2), and XPCB–hHR23B (B). Dynamic structure of XPC-binding domain of hHR23B B. Kim et al. 2472 FEBS Journal 272 (2005) 2467–2476 ª 2005 FEBS sought to determine which XPC peptide segments within the hHR23B-binding domain were relevant to the domain’s function. It was reported that a segment of the human Stch (hStch) protein can bind to the STI1-homologous domain (Fig. 1B) of the Chap1 (hPLIC-1) protein [24]. Therefore, we performed sequence alignment with the hHR23B-binding domain of XPC and hStch (because the XPCB domain is sus- pected to shares sequence similarity with STI1), and selected two segments with the highest degree of simi- larity, although their values are low. We then construc- ted two GST-tagged expression vectors that contained DNA sequences that corresponded to the two selected segments (which encoded amino acids 566–608 and 613–661 of the hHR23B-binding domain of XPC). However, we not able to detect binding of either XPCB–hHR23B or the complete hHR23B protein in GST pull-down assays using these two constructs (data not shown). This inability to detect XPCB–hHR23B binding in these pull-down assays suggests that either XPCB–hHR23B recognizes XPC regions not present in the two selected segments or binding of the two pro- teins requires specifically folded motifs not present in the protein fragments. The latter hypothesis is more likely, because the hydrophobic surface of XPCB is delocalized in two ways, and hHR23B has been repor- ted to stabilize XPC from heat denaturation [25]. The slight variability of the hydrophobic surface of XPCB– hHR23B resulting from its innate flexibility could be another advantage of adapting to the larger binding counter part. The difference in flexibility of the N-terminal regions of XPCB–hHR23B and XPCB–hHR23A may be one way to explain why these domains, which share high sequence homology, have such different solubilities, even though they have similarly folded structures. The presence of the divergent hydrophobic surface, which results from the more flexible N-terminal part of XPCB–hHR23B compared with hHR23A, may explain why the former has more diverse in vivo functions [7,18,19]. In order to further our knowledge with respect to the mechanisms of action of these proteins, it is crucial that we define the precise boundaries of the STI1-homologous domain and compare the struc- tures of these domains from various proteins. Such analysis should help to determine the minimal unit for proper folding to yield a functional domain. A number of other cellular proteins exist that, like hHR23B, con- tain the STI1, UBL and UBA domains connected by relatively flexible linkers. It is likely that the STI1 domains of these proteins have evolved to specify and modulate target proteins through a common mechan- ism related to proteolysis. Experimental procedures Cloning and purification of the XPCB–hHR23B domain The cDNAs encoding the hHR23B were generously provi- ded by F. Hanaoka (Osaka, Japan) [9]. A cDNA fragment containing the XPC-binding motif of hHR23B (277–332), plus two extra N-terminal residues (Pro and Leu) and 10 more C-terminal residues (333–342), was subcloned into the pET15b vector at the NdeI and BamHI sites (Novagen, Madison, WI, USA). The protein was expressed in the E. coli BL21 (DE3) pLysS strain by using isopropyl thio- b-d-galactoside induction at 37 °C. The N-terminal His- tagged form of the XPCB–hHR23B protein was purified by using a Ni-NTA column (Qiagen, Valencia, CA, USA), and the terminal His-tag was removed by the thrombin diges- tion. An additional purification step of gel permeation chromatography was performed on a Superdex 75 column (Amersham Biosciences). Uniformly 15 N-labeled and 13 C, 15 N-labeled XPCB–hHR23B (275–342) were obtained by growing the bacteria in M9 minimal medium supplemen- ted by [ 15 N]ammonium chloride and [ 13 C]glucose. Acquisition and processing of NMR data NMR samples were prepared in buffers (pH 7.0, 40 mm sodium phosphate and 160 mm sodium chloride) with or without 10 mm Chaps. All NMR spectra were recorded at 27 °C using a 600 MHz, Varian INOVA spectrometer (Varian Associates Inc., Palo Alto, CA, USA). For the backbone and side-chain assignments of XPCB–hHR23B, we used the following general triple-resonance experiments: HNCACB [26], CBCA(CO)NH [27], HNCO [28], C(CC- TOCSY-CO)N-NH [29], H(CC-TOCSY-CO)N-NH [29], HCCH-TOCSY [30], HCCH-COSY [30], and TOCSY- N15-HSQC (mixing time, 100 ms). For structure determin- ation, we extracted the NOE-distance restraints from NOESY-N15-HSQC (mixing times, 80 ms and 150 ms) and NOESY-C13-HSQC (mixing time, 100 ms). R 1 (1 ⁄ T 1 ) values of 15 N were measured from spectra recorded with nine different delays of T ¼ 10, 50, 100, 200, 400, 600, 800, 1000, and 1200 ms with relaxation delays of 1.5 s. R 2 (1 ⁄ T 2 ) values were determined from spectra recor- ded with duration delays of 10, 30, 50, 70, 130, 190, and 250 ms with relaxation delays of 1 s. Steady-state 15 N- 1 H NOEs were measured following the method described in Farrow et al. [31] using proton saturation periods of 3 and 5 s, and then the average 15 N- 1 H NOE was obtained. Addi- tional R 1 values (20, 40, 80, 140, 240, 400, 800, and 1200 ms with a delay of 3 s) and R 2 values (16.8, 33.5, 50.3, 67.0, 100.5, 134.0, 184.3, and 234.6 ms with a delay of 1 s) and two independent set heteronuclear NOEs (satura- tion period of 3 s) were obtained using a cryoprobe- installed 500 MHz, Bruker Avance at the Korea Basic B. Kim et al. Dynamic structure of XPC-binding domain of hHR23B FEBS Journal 272 (2005) 2467–2476 ª 2005 FEBS 2473 Science Institute (Daejon, South Korea). The buffer condi- tion was slightly different to that used in the 600 MHz NMR machine (pH 7.2, 50 mm Hepes and 200 mm NaCl), because the presence of highly charged and small ion, such as phosphate, increases the 90 degree pulse length in the cryoprobe. All NMR data were processed using nmrpipe [32] and analyzed using sparky [33]. The errors of R 1 and R 2 were estimated from the errors in the single exponential decay fitting, and those of the 15 N- 1 H NOEs were obtained from the difference of two independent experiments. The values of error were adjusted for the proper tensor2 analy- sis, by increasing of the errors from the fitting (1.75 and 2.0 times for the relaxation data of 500 and 600 MHz, respect- ively), in which the maximum and minimum errors were fixed to 7.5 and 3.0%, respectively. Spectral density func- tions assuming an isotropic rotational diffusion tensor were calculated in 1000-step Monte Carlo simulations using the tensor2 program [22]. Structure calculation and analysis In total, 28 sets of dihedral angle restraints (/, u) with good prediction scores were gathered from TALOS chem- ical shift analysis [34], and an additional 14 angles restraints (/) were obtained from the intensity-modulated 15 N-HSQC experiment [35]. The automatic NOE assign- ment and structure calculations were performed using the cyana program [21] and the 1948 NOE cross-peaks (1165 from 13 C-NOESY-HSQC and 783 from 15 N-NOESY- HSQC). The 50 conformers with the lowest final target function values with pseudo-atom correction were the input for structure refinement with the amber7 program using an Amber FF99 force field [36]. The 50 conformers were simu- lated, annealed, and energy-minimized for 15 ps. During this calculation, the generalized Born model was applied for a better simulation of an electrostatic interaction in a vacuum. We deleted a few distance restraints that were con- sistently violated during the structure calculation and increased the upper boundary of some distance restraints slightly by comparing with the NOESY spectra. However, we tried to retain the original values obtained from the auto-assignment and the structure calculation using cyana when the distance violations were relatively small. Among the 30 structures with the lowest total energies, we selected 20 structures with the lowest NMR restraint violations for further analysis. There were no angle restraint violations for the final 20 structures and there were 33–50 distance restraint violations (0.11 ± 0.04 A ˚ ) among 1242 total dis- tance restraints for each structure. The quality of the struc- tures was assessed using the refined energy terms and the procheck program [37]. The surface electrostatic potential distribution of the best overall model of XPCB–hHR23B was calculated with delphi [38]. The solvent-accessible surface area for XPCB domains were calculated using gromacs (v3.2.1) [39] and chimera [40] was used to analyze the structures and to prepare drawings of the struc- tures. Acknowledgements Financial support was obtained from the National Creative Research Initiatives of the Korean Ministry of Science and Technology. Molecular graphics images were produced using the Chimera package from the Computer Graphics Laboratory of the University of California, San Francisco (supported by NIH P41 RR- 01081). References 1 Batty D, Rapic¢-Otrin V, Levine AS & Wood RD (2000) Stable binding of human XPC complex to irra- diated DNA confers strong discrimination for damaged sites. J Mol Biol 300, 275–290. 2 Davies AA, Friedberg EC, Tomkinson AE, Wood RD & West SC (1995) Role of the Rad1 and Rad10 proteins in nucleotide excision repair and recombination. J Biol Chem 270, 24638–24641. 3 Masutani C, Sugasawa K, Yanagisawa J, Sonoyama T, Ui M, Enomoto T, Takio K, Tanaka K, van der Spek PJ, Bootsma D et al. (1994) Purification and cloning of a nucleotide excision repair complex involving the xero- derma pigmentosum group C protein and a human homologue of yeast RAD23. EMBO J 13, 1831–1843. 4 Sugasawa K, Ng JM, Masutani C, Iwai S, van der Spek PJ, Eker AP, Hanaoka F, Bootsma D & Hoeijmakers JH (1998) Xeroderma pigmentosum group C protein complex is the initiator of global genome nucleotide excision repair. Mol Cell 2, 223–232. 5 Clarke DJ, Mondesert G, Segal M, Bertolaet BL, Jensen S, Wolff M, Henze M & Reed SI (2001) Dosage sup- pressors of pds1 implicate ubiquitin-associated domains in checkpoint control. Mol Cell Biol 21, 1997–2007. 6 Raasi S & Pickart CM (2003) Rad23 ubiquitin-asso- ciated domains (UBA) inhibit 26S proteasome-catalyzed proteolysis by sequestering lysine 48-linked polyubiqui- tin chains. J Biol Chem 278, 8951–8959. 7 van der Spek PJ, Visser CE, Hanaoka F, Smit B, Hage- meijer A, Bootsma D & Hoeijmakers JH (1996) Clon- ing, comparative mapping, and RNA expression of the mouse homologues of the Saccharomyces cerevisiae nucleotide excision repair gene RAD23. Genomics 31, 20–27. 8 You JS, Wang M & Lee SH (2003) Biochemical analysis of the damage recognition process in nucleotide excision repair. J Biol Chem 278, 7476–7485. 9 Masutani C, Araki M, Sugasawa K, van der Spek PJ, Yamada A, Uchida A, Maekawa T, Bootsma D, Hoeij- makers JH & Hanaoka F (1997) Identification and char- Dynamic structure of XPC-binding domain of hHR23B B. Kim et al. 2474 FEBS Journal 272 (2005) 2467–2476 ª 2005 FEBS acterization of XPC-binding domain of hHR23B. Mol Cell Biol 17, 6915–6923. 10 Hiyama H, Yokoi M, Masutani C, Sugasawa K, Mae- kawa T, Tanaka K, Hoeijmakers JH & Hanaoka F (1999) Interaction of hHR23 with S5a. The ubiquitin-like domain of hHR23 mediates interaction with S5a subunit of 26S proteasome. J Biol Chem 274, 28019–28025. 11 Bertolaet BL, Clarke DJ, Wolff M, Watson MH, Henze M, Divita G & Reed SI (2001) UBA domains of DNA damage-inducible proteins interact with ubiquitin. Nat Struct Biol 8, 417–422. 12 Ryu KS, Lee KJ, Bae SH, Kim BK, Kim KA & Choi BS (2003) Binding surface mapping of intra- and inter- domain interactions among hHR23B, ubiquitin, and polyubiquitin binding site 2 of S5a. J Biol Chem 278, 36621–36627. 13 Walters KJ, Lech PJ, Goh AM, Wang Q & Howley PM (2003) DNA-repair protein hHR23a alters its protein structure upon binding proteasomal subunit S5a. Proc Natl Acad Sci USA 100, 12694–12699. 14 Kleijnen MF, Shih AH, Zhou P, Kumar S, Soccio RE, Kedersha NL, Gill G & Howley PM (2000) The hPLIC proteins may provide a link between the ubiquitination machinery and the proteasome. Mol Cell 6, 409–419. 15 Lassle M, Blatch GL, Kundra V, Takatori T & Zetter BR (1997) Stress-inducible, murine protein mSTI1. Characterization of binding domains for heat shock proteins and in vitro phosphorylation by different kinases. J Biol Chem 272, 1876–1884. 16 Schultz J, Milpetz F, Bork P & Ponting CP (1998) SMART, a simple modular architecture research tool: identification of signaling domains. Proc Natl Acad Sci USA 95, 5857–5864. 17 Li L, Lu X, Peterson C & Legerski R (1997) XPC inter- acts with both HHR23B and HHR23A in vivo. Mutat Res 383, 197–203. 18 Layfield R, Tooth D, Landon M, Dawson S, Mayer J & Alban A (2001) Purification of poly-ubiquitinated proteins by S5a-affinity chromatography. Proteomics 1, 773–777. 19 Ng JM, Vrieling H, Sugasawa K et al. (2002) Develop- mental defects and male sterility in mice lacking the ubiquitin-like DNA repair gene mHR23B. Mol Cell Biol 22, 1233–1245. 20 Kamionka M & Feigon J (2004) Structure of the XPC binding domain of hHR23A reveals hydrophobic patches for protein interaction. Protein Sci 13, 2370– 2377. 21 Herrmann T, Guntert P & Wuthrich K (2002) Protein NMR structure determination with automated NOE assignment using the new software candid and the tor- sion angle dynamics algorithm dyana. J Mol Biol 319, 209–227. 22 Dosset P, Hus JC, Blackledge M & Marion D (2000) Efficient analysis of macromolecular rotational diffusion from heteronuclear relaxation data. J Biomol NMR 16, 23–28. 23 Engelke J & Ruterjans H (1999) Recent developments in studying the dynamics of protein structures from 15 N and 13 C relaxation time mesurements. In Biological Magnetic Resonance: Structure Computation and Dynamics in Protein NMR (Krishna NR & Berliner L, eds), pp. 357–418. Plenum Press, New York. 24 Kaye FJ, Modi S, Ivanovska I, Koonin EV, Thress K, Kubo A, Kornbluth S & Rose MD (2000) A family of ubiquitin-like proteins binds the ATPase domain of Hsp70-like Stch. FEBS Lett 467, 348–355. 25 Araki M, Masutani C, Takemura M, Uchida A, Suga- sawa K, Kondoh J, Ohkuma Y & Hanaoka F (2001) Centrosome protein centrin 2 ⁄ caltractin 1 is part of the xeroderma pigmentosum group C complex that initiates global genome nucleotide excision repair. J Biol Chem 276, 18665–18672. 26 Wittekind MML (1993) HNCACB, a high-sensitivity 3D NMR experiment to correlate amide-proton and nitrogen resonances with the alpha- and beta-car- bon resonances in proteins. J Magn Reson B101, 201– 205. 27 Grzesiek SBA (1992) Correlating backbone amide and side chain resonances in larger proteins by multiple relayed triple resonance NMR. J Am Chem Soc 114, 6291–6293. 28 Ikura M, Kay LE & Bax A (1990) A novel approach for sequential assignment of 1 H, 13 C, and 15 N spectra of proteins: heteronuclear triple-resonance three-dimen- sional NMR spectroscopy. Application to calmodulin. Biochemistry 29, 4659–4667. 29 Grzesiek S, Anglister J & Bax A (1993) Correlation of backbone amide and aliphatic side-chain resonances in 13 C ⁄ 15 N-enriched proteins by isotropic mixing of 13 C magnetization. J Magn Reson B101, 114–119. 30 Bax A, Clore GM, Driscoll PC, Gronenborn AM, Ikura M & Kay LE (1990) Practical aspects of proton-carbon- carbon-proton three-dimensional correlation spectro- scopy of 13 C-labeled proteins. J Magn Reson 87, 620– 627. 31 Farrow NA, Zhang O, Forman-Kay JD & Kay LE (1994) A heteronuclear correlation experiment for simul- taneous determination of 15 N longitudinal decay and chemical exchange rates of systems in slow equilibrium. J Biomol NMR 4, 727–734. 32 Delaglio F, Grzesiek S, Vuister GW, Zhu G, Pfeifer J & Bax A (1995) NMRPipe: a multidimensional spectral processing system based on UNIX pipes. J Biomol NMR 6, 277–293. 33 Goddard TD & Kneller DG. SPARKY 3. University of California, San Francisco. 34 Cornilescu G, Delaglio F & Bax A (1999) Protein back- bone angle restraints from searching a database for che- B. Kim et al. Dynamic structure of XPC-binding domain of hHR23B FEBS Journal 272 (2005) 2467–2476 ª 2005 FEBS 2475 mical shift and sequence homology. J Biomol NMR 13, 289–302. 35 Permi P, Kilpelainen I, Annila A & Heikkinen S (2000) Intensity modulated HSQC and HMQC: two simple methods to measure 3 J HNHa in proteins. J Biomol NMR 16, 29–37. 36 Case DA, Pearlman DA, Caldwell JW, Cheatham TE III, Wang J, Ross WS, Simmerling CL, Darden TA, Merz KM, Stanton RV et al. (2002) AMBER In. University of California, San Francisco. 37 Laskowski RA, Rullmannn JA, MacArthur MW, Kaptein R & Thornton JM (1996) AQUA and PROCHECK-NMR: programs for checking the quality of protein structures solved by NMR. J Biomol NMR 8, 477–486. 38 Honig B & Nicholls A (1995) Classical electrostatics in biology and chemistry. Science 268, 1144–1149. 39 Lindahl E, Hess B & van der Spoel D (2001) gromacs 3.0: a package for molecular simulation and trajectory analysis. J Mol Mod 7, 306–317. 40 Huang CC, Couch GS, Pettersen EF & Ferrin TE (1996) Chimera: an extensible molecular modeling appli- cation constructed using standard components. Pacific Symp Biocomput 1, 724. Supplementary material The following material is available from http://www. blackwellpublishing.com/products/journals/suppmat/EJB/ EJB4667/EJB4667sm.htm. Fig. S1. 15N-HSQC spectrum of XPCB–hHR23B. Fig. S2. Relaxation studies of XPCB–hHR23B at 600 MHz field. Dynamic structure of XPC-binding domain of hHR23B B. Kim et al. 2476 FEBS Journal 272 (2005) 2467–2476 ª 2005 FEBS . Solution structure and backbone dynamics of the XPC-binding domain of the human DNA repair protein hHR23B Byoungkook Kim 1, *,. breaks and turns or kinks (Figs 1 and 2A). The N-terminal part of XPCB hHR23B ABC Fig. 2. NMR structure of the XPCB domain of hHR23B. (A) Stereoview of the