Solution structure of the C-terminal DUF1000 domain of the human thioredoxin-like 1 protein

Abstract

The thioredoxin (TRX) system, comprising TRX, TRX reductase, and NADPH, is ubiquitous in all living cells.1 Different forms of TRX proteins have been found in all organisms, from prokaryotes to humans.1-12 TRX proteins are a group of small (∼12 kDa) redox-active proteins, characterized by the sequence Cys-Gly-Pro-Cys at the active site, which is localized in a protrusion of the three-dimensional structure. The formation of an intramolecular disulfide bridge between the cysteine residues in the active site can supply reducing equivalents to, typically, a disulfide substrate. The oxidized domain is then reduced back by the NADPH-dependent flavoprotein TRX reductase.1 There have been few studies devoted to the characterization of the TXL-1 protein, also known as TRP32.5-12 This 32 kDa protein is composed of two domains: a catalytic N-terminal TXLN domain (resembling one of the common TRX-domains),7 and a C-terminal domain of unknown function, DUF1000, with the sequence displayed in Figure 1(F). There are no putative TXL-1 homologues in either the E. coli or yeast genome sequence, suggesting that TXL-1 may be evolutionarily conserved in multicellular eukaryotes. In contrast to human TRX,1 the direct ability to reduce H2O2 has not been demonstrated for TXL-1. Human TXL-1 exhibits four-fold lower protein disulfide reduction activity than TRX, and its activity to reduce insulin disulfide bonds in the presence of NADPH and calf thymus TRX reductase is delayed by over 45 min.9 This can be attributed to the glycine residue that substitutes for the tryptophan in TRX, immediately preceding the active site of the catalytic N-terminal TXLN domain. Recently, it was discovered that the C-terminal DUF1000 domain of TXL-1 was necessary and sufficient to bind to Rpn11, a metalloprotease subunit of the 19S regulatory particle of the 25S proteasome lid, which couples the deubiquitination and degradation of proteasome substrates.10 Neither oxidative nor reductive stress significantly altered TXL-1 mRNA levels.10 TXL-1 is an abundant protein, and like other proteasome components, is expressed at roughly similar levels in all tested tissues and at approximately equimolar levels with the subunits of the 26S proteasome.10 These exciting discoveries highlight the pressing need for a DUF1000 structure. In this study, we determined its solution structure using NMR spectroscopy. Previous crystallization attempts were unsuccessful.7 NMR structures of our DUF1000-construct [Fig. 1(A–D)] and of the Arabidopsis thaliana At3g04780.1-des15 ortholog (PDB id., 1xoy) [Fig. 1(E)] with a structure-based sequence alignment [Fig. 1(F)]. A: Superposition of 20 model structures (backbone). B: Ribbon diagram of a representative model. C: Mesh presentation, generated using PyMol25 with chainbow coloring. That is, residues are colored as a rainbow that begins with blue and ends with red. The putative binding pocket was calculated using CASTp26 with a solvent probe of radius 1.4 Å; it has a volume of 159.4 Å3. The atoms bordering this pocket are shown as spheres: I16 CG2; C21 C, O, CB, SG; E22 N, CA, C, O; C23 N, CB, SG; F32 C, O, CB; D33 CG, OD1, OD2; L36 CG, CD1, CD2; I53 CG2; T54 O; V55 CG1, CG2; F149 CD1, CE1. In the water-refined structures (not shown), the atom F68CE1 also borders this pocket. D: Close-up of the putative binding pocket, highlighting the sidechains of the bordering residues. Interestingly, the solvent-exposed C23HG is among them. E: The orientation of this figure relative to Figure 1(B) was obtained by mean fitting of the backbones of residues 12–39, 50–86, and 114–155 of the DUF1000-construct structure with residues 16–43, 58–94, and 120–161 of the ortholog structure. The resulting RMS deviation is 2.1 Å. F: NOE data supports the conclusion that the residues marked with * likely form helices, in a similar manner as the corresponding residues in the ortholog structure. The DNA encoding the C-terminal DUF1000 domain of human TXL-1 (residues 122–279 of 289 total) was subcloned by PCR from the human Sugano cDNA library,13 with the accession number SPL08258 (Genbank accession: BP340183). This DNA fragment was cloned into the expression vector pCR2.1 (Invitrogen), as a fusion with an N-terminal 6-His affinity tag and a TEV protease cleavage site. The DUF1000-construct (see Fig. 1) has linker sequences (13 residues in total) at the N-terminus (GSSGSSG) and the C-terminus (SGPSSG), which are both derived from the expression vector. The 13C/15N-labeled fusion protein was synthesized by the cell-free protein expression system14, 15 and was purified as follows.16 First the solution was adsorbed to a HiTrap Chelating column (GE Healthcare), which was washed with buffer A (50 mM Tris-HCl buffer [pH 8.0], containing 500 mM sodium chloride and 10 mM imidazole) and eluted with buffer B (50 mM Tris-HCl buffer [pH 8.0], containing 500 mM sodium chloride and 500 mM imidazole). To remove the His-tag, the eluted protein was incubated at 30°C for 1 h with TEV protease. After dialysis against buffer A without imidazole, the dialysate was mixed with imidazole to a final concentration of 10 mM, and then was applied to a HiTrap Chelating column, which was washed with buffer A. The flow-through fraction was loaded onto a HiTrap Desalting column (GE Healthcare) with buffer C (20 mM Tris-HCl buffer [pH 8.0]). The DUF1000-containing fractions were applied to a HiTrap Q column (GE Healthcare), using a concentration gradient of buffer C and buffer D (20 mM Tris-HCl buffer [pH 8.0], containing 1M sodium chloride). The DUF1000-containing fractions were collected, and a protease inhibitor cocktail (Complete [EDTA-free], Roche Applied Science) and DTT (final concentration, 1 mM) were added. For NMR measurements, the purified protein was concentrated to 1.02 mM in 1H2O/2H2O (9:1) 20 mM Tris d11-HCl buffer (pH 7.0), containing 100 mM NaCl, 1 mM 1,4-DL-dithiothreitol-d10 (d-DTT), and 0.02% NaN3. All NMR measurements were performed at 25 or 27°C on Bruker AVANCE 600 and AVANCE 900 spectrometers. Sequence-specific backbone assignments were made with the 13C/15N-labeled sample, using standard triple-resonance experiments17: HNCO, HNCA, HN(CA)CO, CBCA(CO)NH, HN(CA)CB, and HN(CO)CA. Assignments of aliphatic side chains were obtained from (all 3D) HBHA(CO)NH, H(CC)(CO)NH (also called H(CCCO)NH-TOCSY), CC(CO)NH (also called (H)CC(CO)NH-TOCSY), HCCH-TOCSY, and CCH-TOCSY spectra, aromatic assignments from 3D CCH-COSY (aromatic). 3D 15N- and 13C-edited NOESY-HSQC spectra with 80 ms mixing times (collected on a Bruker AVANCE 900 spectrometer) were used to determine distance restraints. The spectra were processed with the program NMRPipe18 and were analyzed with the programs Kujira19 and NMRView.20 The NOESY spectra provided more NOE peaks (8472) than usual for proteins of comparable size. Automated NOE cross-peak assignments and structure calculations with torsion angle dynamics were performed using CYANA.21 Dihedral angle restraints (ϕ, ψ) were derived using the program TALOS.22 No hydrogen bond constraints were used. A total of 100 structures were independently calculated, and the 20 best ones (according to the target function) were selected for further analysis. The structures were validated using PROCHECK-NMR.23 MOLMOL,24 PyMol,25 and CASTp26 were used to analyze the calculated structures and to prepare drawings of the structures. Structural statistics are summarized in Table I. The backbone and side chain assignments were almost complete, except for residues in the N-terminal tag region. There are few unassigned peaks. The target function of cycle 1 is 105.96 Å2. The atomic coordinates and NMR restraint data are available from the protein data bank (PDB), (http://www.rcsb.org), accession code 1WWY. Further NMR data (chemical shift assignments and time-domain NMR data) have been deposited in the BioMagResBank database (http://www.bmrb.wisc.edu), BMRB accession number 10336. Twenty water-refined structure models were obtained from the 20 model structures from CYANA by energy minimization, using the program CHARMM.27, 28 Langevin dynamics were used with the PARAM22 all-hydrogen parameter set.27, 28 The average solvent effect was characterized by a generalized Born implicit solvent model using molecular volume (GBMV), incorporating solvent accessible surface area (SASA) calculations.27, 28 The fold is dominated by a jelly roll β-sandwich structure [Fig. 1(A, B)]. The β-sandwich is formed by face-to-face packing of two anti-parallel β-sheets, with the first composed of five strands (β1: residues 22–24, β3: 52–60, β5: 79–84, β6: 104–105, and β8: 127–134), and the second composed of three strands (β4: 61–68, β7: 117–118, and β10: 149–155), see Figure 1(F). Another β-sheet is formed by the strands: β2: 43–44 and β9: 145–146. There is only one comparable published protein structure, the solution NMR structure of At3g04780.1-des15, a slightly modified product of the At3g04780.1 gene.29 This Arabidopsis thaliana (At) ortholog (PDB id., 1xoy) shares 39% sequence identity and 52% sequence similarity with DUF1000 [Fig. 1(F)], but unlike TXL-1, it is a single-domain protein and thus lacks a TRX-like domain. Therefore, its function may be different from that of TXL-1. The global folds of both structures are similar [cf. Fig. 1(B,E)], with some relatively minor differences in the loop regions. Their RMS deviation over 107 residues is 2.1 Å (see Fig. 1E legend for details). The jelly roll β-sandwich is formed by a sheet composed of five strands (β2: 26–27, β4: 60–70, β6: 85–92, β7: 109–112, and β9: 133–143) and a sheet composed of four strands (β1: 14–15, β5: 71–78, β8: 123–124, and β11: 154–160). Again, there is one additional β-sheet, this time formed by the strands: β3: 52–53 and β10: 151–152. The two additional helices α2: 36–41 and α4: 116–120 are likely to have analogues in DUF-1000 as well (supporting NOE data exists), at the positions labeled with * in Fig. 1(F), residues 32–37 and 109–113. However, the DUF1000 structure has a cavity [Fig. 1(C,D)] exposing the sulfhydryl (thiol) group of C23, which is absent in the At ortholog structure. This cavity is quite large (159.4 Å3) and the quality of the structure in this area is high (e.g., many assigned NOEs, high precision). To verify the existence of this cavity, we calculated the structures in implicit solvent. The structures before and after water refinement closely overlap (RMS deviation between them: 0.38 Å). Some formal structural parameters improved, whereas others deteriorated (see Table I). The water-refined structures show a helix at residues 32–34. These are residues labeled with * in Figure 1(F), as mentioned above. The cavity is also present in the water-refined structures, and it is 15% larger. Among the cavity bordering residues listed in the legend to Figure 1, the conformations of F133 and F149 changed slightly, whereas residue F68 was a new member. Thus, the probability of the existence of this cavity in the native protein may be judged as high. We propose that the cavity identified in the previous section [Fig. 1(C,D)] with the exposed sulfhydryl(thiol) group of C23, and the other residues bordering it (see Fig. 1 legend) constitutes a binding site. The existence of a binding site was suggested by the finding that the C-terminal DUF1000 domain of TXL-1 was necessary and sufficient to bind to Rpn11.10 For confirmation, a structure of the DUF1000-Rpn11 complex is necessary. This is likely to present a formidable challenge, considering its size (around 53kDa), and noting the sample preparation problems associated with DUF1000.7 Binding partners other than Rnp11 are certainly conceivable, but at this point their existence is purely speculative. At least 85% of the cellular TXL-1 is proteasome-associated.10 It has been demonstrated that TXL-1 overexpression protects against glucose deprivation-induced toxicity, but not against hydrogen peroxide cytotoxicity.9 TXL-1 is also arsenite-regulated.11 Though, these effects can be explained by the association of TXL-1 with the proteasome is presently an open question, but an alternate explanation may be that a metabolite in the signaling pathway, induced by the effects of glucose deprivation or arsenite-induced toxicity, but not by the effects of hydrogen peroxide induced toxicity, might bind to DUF1000. The authors would like to thank the following persons: Satoru Watanabe, Takushi Harada, Yukiko Fujikura, Masaagki Aoki, Kazuharu Hanada, Yasuko Tomo, Takayoshi Matsuda, Masaomi Ikari, Eiko Seki, Natsuko Matsuda, Yoko Motoda, and Naohiro Kobayashi. This work was performed at RIKEN Systems and Structural Biology Center, 1-7-22 Suehiro-cho, Tsurumi-ku, Yokohama, Kanagawa, Japan.