Crystal structure of the catalytic domain of a human thioredoxin-like protein Implications for substrate specificity and a novel regulation mechanism Jian Jin 1,2 , Xuehui Chen 1 , Yan Zhou 2 , Mark Bartlam 1 , Qing Guo 1 , Yiwei Liu 1 , Yixin Sun 1 , Yu Gao 1,2 , Sheng Ye 1 , Guangtao Li 2 , Zihe Rao 1 , Boqin Qiang 2 and Jiangang Yuan 2 1 Laboratory of Structural Biology and the MOE Laboratory of Protein Science, School of Life Science & Engineering, Tsinghua University, Beijing, China; 2 National Laboratory of Institute of Basic Medical Sciences, Peking Union Medical College and Chinese Academy of Medical Sciences, National Center of Human Genome Research, Beijing, China Thioredoxin is a ubiquitous dithiol oxidoreductase found in many organisms and involved in numerous biochemical processes. Human thioredoxin-like protein (hTRXL) is differentially expressed at different development stages of human fetal cerebrum and belongs to an expanding family of thioredoxins. We have solved the crystal structure of the recombinant N-terminal catalytic domain (hTRXL-N) of hTRXL in its oxidized form at 2.2-A ˚ resolution. Although this domain shares a similar three-dimensional structure with human thioredoxin (hTRX), a unique feature of hTRXL-N is the large number of positively charged residues distributed around the active site, which has been implicated in substrate specificity. Furthermore, the hTRXL-N crystal structure is monomeric while hTRX is dimeric in its four crystal structures (reduced, oxidized, C73S and C32S/C35S mutants) reported to date. As dimerization is the key regu- latory factor in hTRX, the positive charge and lack of dimer formation of hTRXL-N suggest that it could interact with the acidic amino-acid rich C-terminal region, thereby sug- gesting a novel regulation mechanism. Keywords: dithiol oxidoreductase; hTRXL; crystal structure; monomeric; N-terminal. Thioredoxin, a group of redox active proteins, is both ubiquitously present and evolutionarily conserved from prokaryotes to higher eukaryotes [1–3]. Thioredoxin was initially discovered in Escherichia coli as an electron donor for the essential enzyme ribonucleotide reductase [4] and, since then, many functions have been assigned to thio- redoxins not only associated with redox-mediated processes but also with structural roles. For example, they can also serve as a reducing agent in sulfate reduction [5,6] and methionine sulfoxide reduction in E. coli [7]. Moreover, E. coli thioredoxin-(SH) 2 can act as an essential subunit of T7 DNA polymerase [8] and is known to function in the maturation of filamentous bacteriophages M13 and f1 [9,10]. In eukaryotic cells, thioredoxin can facilitate refold- ing of disulfide-containing proteins [11] and modulate the activity of some transcription factors such as NF-kB and AP-1 [12,13]. Other functions include antioxidant action and the ability to reduce hydrogen peroxide [14], scavenging of free radicals [15], and protection of cells against oxidative stress [16]. A recent area of interest is the role of thioredoxin as a cell growth stimulator and an apoptosis inhibitor, both in vitro and in vivo. Recombinant human thioredoxin, when added to minimal culture medium in the absence of serum, stimulates the proliferation of a number of human solid tumor cell lines as measured over several days [17]. An adult T cell leukemia-derived factor, which augments the expression of interleukin-2 receptor and then stimu- lates T cell growth, was found to be identical to human thioredoxin [18]. WEHI7.2 cells stably transfected with human thioredoxin cDNA and displaying increased levels of cytoplasmic thioredoxin, showed increased growth and were resistant to drug-induced apoptosis both in vitro and in vivo [19]. In contrast, redox-inactive mutant thioredoxin reduces growth and enhances drug-induced apoptosis when transfected into WEHI7.2 cells [20]. Since the molecular studies have provided the proof-of-principle that the thioredoxin system is a rational target for anticancer drug development, the initial approach was to develop agents that might selectively inhibit the thioredoxin system and hence thioredoxin-dependent cell proliferation [21]. Members of the expanding thioredoxin family are char- acterized by an amino-acid sequence at the active site, -Cys- Gly-Pro-Cys-, conserved throughout evolution. Extensive structural characterization of thioredoxin has been carried out by both X-ray and NMR methods [22–26]. The globular Correspondence to Z. Rao, Laboratory of Structural Biology, School of Life Sciences and Engineering, Tsinghua University, Beijing, 100084, China. Fax: + 86 62773145, Tel.: + 86 62771493 E-mail: raozh@xtal.tsinghua.edu.cn Abbreviations: hTRXL, human thioredoxin-like protein; hTRXL-N, the N-terminal domain of human thioredoxin-like protein; hTRXL-C, the C-termianal region of human thioredoxin-like protein; hTRXL, gene of human thioredoxin-like protein; hTRX, human thioredoxin; EST, expressed sequence tags. Enzymes: thioredoxin (EC 1.8.4.8); flavoenzyme thioredoxin reductase (EC 1.6.4.5); thrombin (EC 3.4.21.5). Note: a website is available at http://www.xtal.tsinghua.edu.cn (Received 19 November 2001, revised 12 February 2002, accepted 21 February 2002) Eur. J. Biochem. 269, 2060–2068 (2002) Ó FEBS 2002 doi:10.1046/j.1432-1033.2002.02844.x structure consists of a central b sheet that is sandwiched by a helices. The active site of thioredoxin is localized in a protrusion of the protein surface [22], and the two cysteine residues provide the sulfhydryl groups involved in the thioredoxin-dependent reducing activity. The oxidized form (thioredoxin-S 2 ), where the two cysteine residues are linked by an intramolecular disulfide bond, is reduced by flavoen- zyme thioredoxin reductase and NADPH [2]. The reduced form [thioredoxin-(SH) 2 ] contains two thiol groups and can efficiently catalyze the reduction of many exposed disulfides. Therefore, thioredoxin can interact with a broad range of proteins either in electron transport for substrate reduction or in regulation of activity by a seemingly simple redox mechanism based on reversible oxidation of two cysteine thiol groups to a disulfide, accompanied by the transfer of two electrons and two protons. Human thioredoxin-like protein (hTRXL, m  32 kDa) can be regarded as a member of the mammalian thioredoxin family [27]. The other two members of this family, thioredoxin-1 (m  12 kDa) and mitochondrial thioredoxin-2 (m  18 kDa), are much smaller than hTRXL. Among the three types of thioredoxin proteins, least is known about hTRXL. Here, we report our work on the isolation of the gene, hTRXL, the functional identification of the gene product and the structure determination of its N-terminal catalytic domain. MATERIALS AND METHODS DDRT-PCR and full-length cDNA isolation Total RNA (2.5 lg) from 13- and 33-week-old human fetal cerebrum (Biochain) was reverse transcribed by Superscript II (Gibco-BRL) using a single-base anchored 3¢ primer (5¢-AAGCTTTTTTTTTTTN-3¢,N¼ C, G, A) each time. The cDNAs were subsequently amplified by PCR using the same 3¢ single-base anchored primer and 5¢ arbitrary primer. A detailed procedure for reverse transcription and differential display have been described previously [28,29]. The PCR products were electrophoresed on a 6% SDS/ PAGE gel (data not shown). cDNA fragments that showed differential display were recovered from the dried sequen- cing gel, reamplified and subcloned into PCRII using the TA cloning kit (Invitrogen, San Diego, CA, USA). In total, 90 selected expressed sequence tags (ESTs) were cloned and then sequenced. After database searching, HFBEST12 (GenBank accession no. U48630) was chosen to be used as a probe labeled with [a- 32 P]dCTP (Amersham) to screen the human fetal brain kDR2 cDNA library (ClonTech) for full-length cDNA. Three positive kDR2 phage clones were isolated and converted to pDR2 plasmid, as described in ClonTech’s manual. DNA sequencing was performed according to standard methods on an ABI 377 autosequ- encer (PerkinElmer). Northern blot analysis Total Poly(A) + RNA from 13- and 33-week-old human fetal cerebrum (Biochain) was electrophoresed in a 1% agarose gel containing 0.66 M formaldehyde and was blotted onto a Hybond-N + nylon membrane filter (Amer- sham). The blotted filter and the human adult multiple- tissue Northern blot membrane (ClonTech) were hybridized in accordance with manufacturer’s instructions. The probe is the differentially displayed EST HFBEST12 (GenBank accession no. U48630) isolated in DDRT-PCR, random- radiolabeled with [a- 32 P]dCTP. Cloning procedures, expression and purification The full-length hTRXL cDNA in pDR2 vector (ClonTech) was used as a template for PCR to create in-frame constructs for further cloning. Human thioredoxin full- length cDNA was also isolated by PCR-amplification using human fetal brain library (ClonTech) as a tem-plate. PET- 28 vector (Novagen) and PGEX-4T vector (Amersham Pharmacia Biotech) were used to create histidine-tagged and GST-fused proteins for bacterial expression. His-tagged proteins and the GST fusion proteins were expressed in E. coli strain BL21. His-tagged proteins were loaded onto a His-Trap column (Novagen) and eluted with a 5–300 m M imidazole gradient at pH 8.0 buffered with 20 m M Tris/HCl. GST fusion proteins were bound to glutathione–Sepharose beads (Amersham Pharmacia Bio- tech), and were cleaved by incubation with thrombin protease (Sigma) at 4 °C for 14 h. Insulin disulfide reduction assay E. coli thioredoxin (Sigma), His-tagged human thioredoxin (His-TRX), His-tagged hTRXL, His-tagged hTRXL-N (residues 1–122) and His-tagged hTRXL-C (residues 105– 289) were compared for the reducing activity of insulin disulfide bonds as described previously (30). The 600-lL reaction mixture contained 100 m M NaCl/P i (pH 7.0), 2m M EDTA, 0.13 m M bovine insulin (Sigma) and 5 m M proteins. A reaction was initiated by adding 1 m M dithio- threitol, and the A 650 was immediately recorded at room temperature. Measurements were performed using 1-min recordings and the nonenzymatic reduction of insulin by dithiothreitol was recorded in a control cuvette without thioredoxin. Crystallization and data collection The hTRXL-N crystals were grown by hanging-drop vapor-diffusion in ammonium sulfate system. Native data for TRXL-N was collected in house using a Rigaku rotating anode X-ray source and a MAR345 image plate to 2.22 A ˚ (31). Structure determination The crystals belong to space group C 2 with the unit cell dimensions of a ¼ 87.5 A ˚ , b ¼ 48.5 A ˚ , c ¼ 29.8 A ˚ , b ¼ 99.59°. The data were processed with DENZO / SCALE- PACK [32]. Data statistics are given in Table 1. The structure was solved by molecular replacement with CNS [33] using the structure of human thioredoxin reduced form (PDB code: 1ERT) as a search model, then refined smoothly in alternating steps of automatic adjustment with CNS and manual adjustment with the program O [34]. The final model has a final R-factor of 0.222 with a free R-factor of 0.253. Molecular graphics images were generated using a combination of BOBSCRIPT [35], GRASP [36], RASTER 3 D [37] and O [34]. Ó FEBS 2002 Crystal Structure of hTRXL-N (Eur. J. Biochem. 269) 2061 Data deposition Coordinates for TRXL-N have been deposited with the Protein Data Bank (PDB accession no., 1GH2, RCSB accession no., RCSB001506). RESULTS AND DISCUSSION hTRXL is a gene differentially expressed at different development stages mRNA extracted from human fetal brain tissues at different developmental stages (13- and 33-week-old cerebrum) was used for DDRT-PCR and the isolated EST (GenBank, accession no. U48630) with different expression patterns in these two stages was cloned into pBlue-Script vector and sequenced. cDNA library screening was performed using the EST obtained as a probe labeled by a- 32 Pandthe screening resulted in isolation of a novel, full length cDNA clone, hTRXL (human thioredoxin-like protein, GenBank accession no. AF051896). hTRXL is 1230-bp in length and contains an 867-bp ORF, which encodes for a protein with 289 amino acids and a calculated molecular mass of 32 kDa. A search of the nonredundant protein sequence database was performed using the BLAST program. Besides sharing the same sequence with Txl/TRP32 [38,39], the 105 residue N-terminal domain shared 42% identity and 55% similarity to human thioredoxin and contained the con- served active site sequence CGPC (Cys-Gly-Pro-Cys). The C-terminal 184 amino acids of hTRXL, which is rich in acidic amino acids, had no similarity to any proteins in the public databases. The full-length cDNA isolated and cloned by the method of DDRT-PCR and cDNA library screening is identical to the previously published Txl/TRP32 sequence [38,39]. Northern blot analysis using poly (A + ) RNA from the 13- and 33-week-old cerebrum demonstrated that the expression level of hTRXL in the former was distinctly higher than that in the latter (Fig. 1A). This confirmed that the results from the DDRT-PCR that hTRXL did have different expression levels in human cerebrum at different development stages. Northern blot analysis using mRNA from multiple adult human tissue showed that the hTRXL was a ubiquitously expressed gene (Fig. 1B). Both full-length hTRXL and its N-terminal domain have the thioredoxin-like reductase activity To investigate the thioredoxin-like reducing activity of hTRXL, we expressed recombinant hTRXL and human thioredoxin as His-tagged forms (His-hTRXL and His- TRX) in E. coli. Truncated hTRXLs corresponding to the N-terminal (His-hTRXL-N, residues 1–122) and C-terminal (His-hTRXL-C, residues 105–289) domains were also prepared, respectively. The expressed recombinant proteins were purified by His-Trap column chromatography. In contrast to previously published work on Txl/TRP32, in which the full-length proteins did not show any reducing activity, our experiments showed that both His-hTRXL and His-hTRXL-N possessed reducing activity for the insulin disulfide bonds. The former showed the kinetics faster than His-TRX but slower than E. coli thioredoxin (Sigma), while the latter showed similar reducing activity to His-TRX (Fig. 2). hTRXL and hTRXL-N (GST fusion expressed then cleaved) exhibited the same behavior in insulin disulfide Table 1. Summary of crystallographic data collection and refinement statistics. A. Data statistics Resolution (A ˚ ) 100–2.2 A ˚ Space group C 2 Unit cell (A ˚ , °) a ¼ 87.5 b ¼ 48.5 c ¼ 29.8 b ¼ 99.59 R merge (%) 0.089 (0.316) a No. of reflections 6710 (624) a Completeness (%) 99.8 (98.7) a I/I (I) 8.4 B. Refinement statistics Resolution (A ˚ ) 15–2.2 A ˚ R working (%) 22.2 (6026 reflections) R free (%) 25.3 (337 reflections) No. of nonhydrogen atoms Protein atoms 816 Solvent 44 Rmds deviation from ideal values Bond length (A ˚ ) 0.02 Bond angle (°) 1.97 Average B-factor (A ˚ 2 ) Protein atoms 24.0 Solvent molecules 37.3 a Numbers in parentheses are the corresponding numbers for the highest resolution shell (2.30–2.22 A ˚ ). Fig. 1. Expression pattern of the hTRXL transcript. Differential expression of hTRXL in human fetal cerebrum of 33- and 13-weeks- old. Human adult tissue Poly(A + ) RNA Northern blot (ClonTech). The 32 P-labeled probe is the EST obtained from DDRT-PCR (Gen- Bank accession no. U48630) and the control used is b-actin cDNA (ClonTech). 2062 J. Jin et al. (Eur. J. Biochem. 269) Ó FEBS 2002 reduction assay (data not shown). As expected, the His- hTRXL-C failed to reduce insulin, demonstrating that the N-terminal region is responsible for the dithio-reducing enzymatic activity and the C-terminal region has little direct effect on the activity of the enzyme. The function of this unique C-terminal domain remains unknown. Overall structure Crystals of the catalytic domain of hTRXL (hTRXL-N) were obtained from ammonium sulfate by hanging-drop vapor-diffusion method [31]. The crystals diffracted beyond 2.2-A ˚ resolution. The structure was determined by molecu- lar replacement with CNS [33] using the crystal structure of human thioredoxin in its reduced form as a search model (PDB ID: 1ERT). The structure was refined to a crystallo- graphic R-factor of 0.222 at 2.2-A ˚ resolution (Table 1). The overall structure is very similar to hTRX (rmsd 0.83 A ˚ )with the main difference being that hTRXL-N crystallized as a monomer while the hTRX crystallized as a disulfide-linked dimer. The N-terminal methionine and the C-terminus from Asn109 to Gly122 are not visible in the electron density map. The numbering convention used for hTRXL through- out starts from the N-terminal methionine, which is different from hTRX (1ERT), offsetting by 2. The hTRXL-N molecule contains a typical thioredoxin fold, consisting of two large folding units: one babab and another bba (Fig. 3A). Although the amino-acid sequence of hTRXL-N shows relatively low identity with that of thioredoxin from different species, the three-dimensional structure is similar (Table 2). Distinct differences occur primarily in the four peripheral ahelices of different molecules, while the hydrophobic core consisting of b sheets shows little difference with other TRX (Fig. 3B). Active site The location of the active site in all of the known thioredoxin structures is identical. It includes the end of b-2, two to three linking amino acids and the beginning of a-2 (Fig. 3A). It is evident that hTRXL-N (as well as other thioredoxins including hTRX) is distinct from most com- mon enzymes, whose active site is usually located in a deep cleft. This is because in TRX, the active site is located on a pronounced protrusion of the molecular surface (Fig. 3A,B), demonstrating that the thioredoxin family proteins are apt at interacting with larger molecules. This would agree with its role in various redox reactions with disulfide containing proteins in vivo and in vitro, as reported previously [4,12,13,40–42]. Despite low sequence identity, dissimilar crystal forms and dissimilar intermolecular contacts near the active site in the crystal, the conformation of the active site (-Cys-Gly-Pro- Cys-) of the hTRXL-N determined in the present study is very close to those of human and E. coli thioredoxin. In addition to the disulfide-bond between the two cysteine residues, three pairs of hydrogen bonds are formed in the active site of hTRXL-N (Fig. 4), accounting for the compactness and stability of the active site. The H-bond length between the carbonyl oxygen of Cys34 and the amide nitrogen of Leu38 is 2.99 A ˚ in this structure, as compared with 3.21 A ˚ in the oxidized form of hTRX (1ERU) and 3.49 A ˚ in the reduced form of hTRX (1ERT), respectively. Cys37 is stabilized by an S–O hydrogen bond with the hydroxyl of Thr76 (bond length 3.3 A ˚ ), which is not present in 1ERU and 1ERT due to a substitution of Thr76 for Met74. The carbonyl oxygen of Gly35 forms a well-aligned H-bond to the amide nitrogen of Arg39 with a length of 2.87 A ˚ , in comparison with the corresponding H-bond in 1ERU and 1ERT (both 3.00 A ˚ ), which suggests the a helix appears more compact in our structure. The conformational change between oxidized and reduced hTRXL-N would be very small and localized in the vicinity of the redox active cysteines, in agreement with the conclusions obtained from structural information on both human and E. coli thio- redoxins, based on both crystallography and NMR [22–26]. Nevertheless, the subtle structural differences between hTRXL-N and hTRX may be important for the different activities of thioredoxin involving a variety of target proteins. However, a remarkable feature of hTRXL-N protein is the large number of positively charged residues distributed around the active site. As shown in Fig. 5, in hTRXL, Lys28, Arg32, Arg39 and His62 replace residues Asp26, Thr30, Met37 and Asp60 that are highly conserved within the thioredoxins of mammals and chick. This suggests that the reaction site of the possible substrates may be rich in negatively charged residues. Another possibility is that the four positively charged residues might play an important role in the interaction with the C-terminal region since the latter carries a large number of acidic amino acids. Substrate specificity Alhough it is difficult to identify the true physiological partners of hTRXL, it was reported that this protein is not a substrate for thioredoxin reductase in the insulin assay, unlike human TRX and thioredoxins in other species [38,39]. The substituted residues around the active site may suggest different ligand specificity for hTRXL-N. A mul- tiple alignment of 83 samples of thioredoxins and thio- redoxin related proteins from archebacteria to human was performed (data shown only includes thioredoxins from mammals and chick). Ninety-six percent of residues are Fig. 2. Reductase activity of thioredoxin proteins. E. coli thioredoxin, His-hTRXL (full-length), His-hTRXL-N, His-hTRXL-C and His- hTRX (5 l M each) were assayed for their ability to reduce the disulfide bonds of insulin as described previously [42]. The incubation mixtures contained, in a final volume of 600 lL: 100 m M NaCl/P i (pH 7.0), 2m M EDTA, 0.13 m M bovine insulin (Sigma) and 1 m M dithiothrei- tol. Only dithiothreitol without thioredoxin served as control. The absorbance at 650 nm is plotted against time. Ó FEBS 2002 Crystal Structure of hTRXL-N (Eur. J. Biochem. 269) 2063 conserved within the thioredoxins of mammals and chick. In contrast, many of them are substituted in hTRXL-N (Fig. 5) and this may lead to divergence in substrate specificity. As expected, many residues in the four a helixes and loops on the molecular surface were found to be substituted while the residues in the five b sheets of the internal hydrophobic core are generally conserved. The most noticeable substitutions are Lys28, Met31, Arg32, Gly33, Leu38, Arg39, His62 and Thr76, which are highly conserved throughout evolution. Fig. 3. hTRXL-N structure. (A) Overall structure of N-terminal domain. Residues involved in the active site are depicted as ball and stick. Cysteines are coloured in yellow (Cys34 and Cys37); Near the active site, positively charged residues are coloured in blue (Lys28, Arg32, Arg39 and His62); other residues are coloured in grey (Met31, Gly33, Gly35 and Pro36); The disulfide bond between Cys34 and Cys37 of active site is coloured in orange. The figure was drawn using BOBSCRIPT (44). (B) Backbone superpositions of the six structures of hTRXL-N and other thioredoxin related proteins or domains. 1GH2 (crystal structure of hTRXL-N, 2–108), 2TRX (crystal structure of E. coli thioredoxin, 1–108), 1ERU and 1ERT (crystal structure of oxidized and reduced human thioredoxin, 1–105), 1DBY (NMR structure of thioredoxin in Chlamydomonas reinhardtii, 1–107), 1MEK (NMR structure of thioredoxin domain of protein-disulfide isomerase, 1–120) are coloured in cyan, blue, red, green, magenta, and yellow, respectively. For detailed rmsd values, see Table 2. Superposition calculation was performed using SHP program [49]. 2064 J. Jin et al. (Eur. J. Biochem. 269) Ó FEBS 2002 Instead of the large imidazole side chain of Trp31 in human TRX (Fig. 6B), which lies both in the active site and in the dimer interface of human TRX, Gly33 takes its place in hTRXL-N. This substitution may contribute to the inability of hTRXL-N to react with thioredoxin reductase. The role of this Trp residue in E. coli thioredoxin has been studied by site-directed mutagenesis: the apparent K m value of thioredoxin reductase with thioredoxin (TRX) as its substrate was increased twofold for the mutant TRX W31A, as compared with the wild type thioredoxin while K cat value remained the same. This results in a 50% reduction in catalytic efficiency (K cat /K m value) of the mutant [43]. So it can be deduced that a similar effect takes place when a Gly33 in hTRXL replaces the equivalent Trp31 in hTRX. Similarly, flexible long side chains of Lys28, Met31, Arg32, Leu38, Arg39 and His62 substituted in the areas in spatial proximity to the active site are also likely to contribute to substrate interaction, leading to divergence in substrate specificity. The NMR structures of human thioredoxin complexed with its target peptides from NFjB and Ref1, respectively, were reported several years ago [25,26]. The peptide Table 2. Sequence identity and rmsd deviations of five representative structures compared with hTRXL-N (1GH2). Protein structure PDB ID Sequence identity rmsd Reduced human thioredoxin, 1–105 1ERT 42% 0.80 A ˚ Oxidized human thioredoxin, 1–105 1ERU 42% 0.83 A ˚ E. coli thioredoxin, 1–108 2TRX 26% 0.97 A ˚ Thioredoxin in chlamydomonas reinhardtii, 1–107 1DBY 19% 1.10 A ˚ Thioredoxin domain of protein-disulfide isomerase, 1–120 1MEK 18% 1.33 A ˚ Fig. 4. Stereoviews of the 2F o ) F c map contoured at 1r at the hTRXL-N active site at 2.2 A ˚ resolution. Hydrogen bonds are represented as dotted lines. Fig. 5. Multiple alignments of thioredoxin homologs. Analignmentofthesheep (SWISSPROT accession no. P50413), Macaca mulatta (accession no. P29451), rat (accession no. P11232), mouse (accession no. P10639), chicken (accession no. P08629), human (Homo sapiens, accession No. P10599) and rabbit (accession no. P08628) thioredoxin is depicted. The14 residues conserved in the dimer inter- face are indicated with asterisks. Ó FEBS 2002 Crystal Structure of hTRXL-N (Eur. J. Biochem. 269) 2065 substrates in the hTRX–NFjB and hTRX–Ref1 complexes were wrapped around the protrusion of the reactive Cys32 in a crescent-shaped groove. However, the orientation of the Ref1 peptide is opposite to that of the target peptide from NFjB. The ability of hTRX to recognize peptides in opposite orientations indicates that this redox protein has succeeded in balancing specificity in substrate recognition with requirement for access to a variety of substrates. In this way, hTRX and perhaps thioredoxins from other species might have the potential to target a wide range of proteins within the cell. A comparison of corresponding hydropho- bic surfaces of hTRXL-N (1GH2) and the substrate-binding surface of hTRX (1CQH) reveals that a similar groove can also be found in the hTRXL surface. To deduce the molecular basis of the possible substrate specificity we compared the residues in the crescent-shaped groove in hTRX with those in the corresponding region in hTRXL-N. In contrast to the 42% sequence identity to hTRX in the whole of hTRXL-N, the assumed substrate-binding region ( 20 residues) shows a sequence identity of about 68.5%, and therefore suggests the similarity in the manner of binding. Despite this similarity, hTRXL is perhaps more inclined to bind proteins whose binding sites are negatively charged as there are four positively charged substitutions distributed around the active site as mentioned above. The monomeric structure of hTRXL-N hTRXL-N is monomeric in its crystal structure determined in the present work, while human thioredoxin (TRX) is dimeric in the four crystal structures reported to date (reduced, oxidized, C73S and C32S/C35S). The dimer interface of TRX consists of three components: an 1100 A ˚ 2 hydrophobic patch, five hydrogen bonds and the Cys73–Cys73 disulfide bond [23]. The substitution of these hydrogen bond forming residues in hTRXL-N may account for the formation of a monomer, instead of a dimer in the case of TRX. Furthermore, the loss of intermolecular disulfide-bonds and the disbandment of the hydrophobic patch may also obstruct the dimer formation (Fig. 6). The14 residues in the dimer interface are highly conserved among the thioredoxins of eight vertebrate species (Fig. 5), yet 10 of these 14 residues are substituted in hTRXL-N, suggesting that important structural and functional changes take place in this area. Human TRX is believed to function as a monomer in redox reactions, but the active site is largely blocked by dimer formation. Hence it has been proposed that dimer formation may play a role in regulating human thioredoxin [44,45]. The changes in the corresponding region of the dimer interface in hTRXL-N imply that different regulatory mechanisms may occur in hTRXL. It can be hypothesized that the unique hTRXL C-terminal region may have a similar role in regulation as in the dimer formation. Finally, we have not been able to crystallize the full-length protein. In order to gain some structural information and function clues of the unique C-terminal region, we used fold recognition and modeling to establish a model structure. Secondary structure prediction was performed using JPRED program [46,47]. The predicted secondary structure in the C-terminal region was shown to be relatively low, and this may partly explain why it was difficult to crystallize the hTRXL full-length protein and its C-terminal region. Fold recognition was performed using the FORESST program [48], Fig. 6. Space-filling model of the disabled dimer interface with H-bond formation residues substituted in hTRXL-N, and compared with its corresponding dimer formation surface in hTRX monomer (reduced) and coloured by residue type: aliphatic (white), positive (blue), negative (red), cysteine (yellow), polar (purple) and alcohol (cyan). Note that the numbering for hTRXL-N is larger by 2 than that in hTRX for the corresponding equivalent resi- dues. Fig. 7. Molecular surface comparison between hTRX and hTRXL-N. Molecular surface representations of hTRX (A) and hTRXL-N (B) around the active surface in the same orientation were produced using GRASP. Electrostatic surface potentials are contoured from )30(red)to30(blue)k b TÆe )1 .The ellipses highlight the position of active site in hTRX and hTRXL-N, respectively. 2066 J. Jin et al. (Eur. J. Biochem. 269) Ó FEBS 2002 the top solutions were classified from the SCOP database webserver and they all belonged to the Ôall-betaÕ family of proteins. Most of the top solutions share the immunoglo- bulin-like fold and the model was constructed according to the structure template of transthyretin (PDB code 1ETA) with the highest Z-score based on the sequence alignment from fold recognition. As shown in Fig. 7, the molecular surface around the active site of hTRXL-N (1GH2) is very different compared with that of hTRX (1ERU). The former is more positive (or much less negative) than the latter. As the C-terminal region is rich in acidic amino acids, if it does have some interaction with the N-terminal domain, the mechanism of regulating the catalytic activity may be similar to that of the dimer. In other word, the active site would be physically blocked and would have to ÔdissociateÕ to achieve active conformation by exposure of the active site. However, we can not exclude the possibility that this site functions as a recruiting factor or signal sequence leading the N-terminal thioredoxin-like domain to approach certain substrates. A study to find the possible substrates in the electron transport chain is currently underway. ACKNOWLEDGEMENTS Z. R. was supported by the following grants: NSFC no. 39870174 and no. 39970155; project Ô863Õ, No. 2001AA233011; project Ô973Õ, No. G1999075602, No. G1999011902 and no. 1998051105. J. Y. was supported by the National Natural Sciences Foundation of China (39830070), the National High Technology Research and Development Program (Z19-02-02-01) and the National Program for Key Basic Research Projects (G1998051002). REFERENCES 1. Holmgren, A. (1984) Enzymatic reduction-oxidation of protein disulfides by thioredoxin. Methods Enzymol. 107, 295–300. 2. Holmgren, A. (1985) Thioredoxin. Annu.Rev.Biochem.54, 237– 271. 3. Holmgren, A. (1989) Thioredoxin and glutaredoxin systems. J. Biol. Chem. 264, 13963–13966. 4. Laurent, T.C., Moore, E.C. & Reichard, P. (1964) Enzymatic synthesis of deoxyribonucleotides. J. Biol. Chem. 239, 3436–3444. 5. Black, S., Harte, E.M., Hudson, B. & Wartofsky, L. (1960) A specific enzymatic reduction of L(-) methionine sulfoxide and a related non-specific reduction of disulfides. J. Biol. Chem. 235, 2910–2916. 6. Wilson, L.G., Asahi, T. & Bandurski, R.S. (1961) Yeast sulfate- reducing systems. I. Reduction of sulfate to sulfite. J. Biol. Chem. 236, 1822–1829. 7. Porque, P.G., Baldesten, A. & Reichard, P. (1970) The involve- ment of the thioredoxin system in the reduction of methionine sulfoxide and sulfate. J. Biol. Chem. 245, 2371–2374. 8.Mark,D.F.&Richardson,C.C.(1976)Escherichia coli thio- redoxin: a subunit of bacteriophage T7 DNA polymerase. Proc. Natl Acad. Sci. USA 73, 780–784. 9. Russel, M. & Model, P. (1985) Thioredoxin is required for filamentous phage assembly. Proc.NatlAcad.Sci.USA82, 29–33. 10. Lim, C J., Haller, B. & Fuchs, J.A. (1985) Thioredoxin is the bacterial protein encoded by fip that is required for filamentous bacteriophage f1 assembly. J. Bacteriol. 161, 799–802. 11. Lundstrom, J. & Holmgren, A. (1990) Protein disulfide-isomerase is a substrate for thioredoxin reductase and has thioredoxin-like activity. J. Biol. Chem. 265, 9114–9120. 12. Meyer, M., Schreck, R. & Baeuerle, P.A. (1993) H2O2 and anti- oxidants have opposite effects on activation of NF-kappa B and AP-1 in intact cells: AP-1 as secondary antioxidant-responsive factor. EMBO J. 12, 2005–2015. 13. Schenk,H.,Klein,M.,Erdbrugger,W.,Droge,W.&Schulze, O.K. (1994) Distinct effects of thioredoxin and antioxidants on the activation of transcription factors NF-kappa B and AP-1. Proc. Natl Acad. Sci. USA 91, 1672–1676. 14. Spector, A., Yan, G.Z., Huang, R.R., McDermott, M.J., Gascoyne, P.R. & Pigiet, V. (1988) The effect of H2O2 upon thioredoxin-enriched lens epithelial cells. J. Biol. Chem. 263, 4984–4990. 15. Schallreuter, K.U. & Wood, J.M. (1986) The role of thioredoxin reductase in the reduction of free radicals at the surface of the epidermis. Biochem. Biophys. Res. Commun. 136, 630–637. 16. Nakamura,H.,Matsuda,M.,Furuke,K.,Kitaoka,Y.,Iwata,S., Toda, K., Inamoto, T., Yamaoka, Y., Ozawa, K. & Yodoi, J. (1994) Adult T cell leukemia-derived factor/human thioredoxin protects endothelial F-2 cell injury caused by activated neutrophils or hydrogen peroxide. Immunol. Lett. 42, 75–80. 17. Gasdaska, J.R., Berggren, M. & Powis, G. (1995) Cell growth stimulation by the redox protein thioredoxin occurs by a novel helper mechanism. Cell Growth Differ. 6, 1643–1650. 18. Tagaya, Y., Maeda, Y., Mitsui, A., Kondo, N., Matsui, H., Hamuro, J., Brown, N., Arai, K., Yokota, T., Wakasugi, H. & Yodoir, J. (1989) ATL-derived factor (ADF), an IL-2 receptor/ Tac inducer homologous to thioredoxin; possible involvement of dithiol-reduction in the IL-2 receptor induction. EMBO J. 8, 757–764. 19. Baker, A., Payne, C.M., Briehl, M.M. & Powis, G. (1997) Thioredoxin, a gene found overexpressed in human cancer, inhibitsapoptosisinvitroandinvivo.Cancer Res. 57, 5162–5167. 20. Freemerman, A.J. & Powis, G. (2000) A redox-inactive thio- redoxin reduces growth and enhances apoptosis in WEHI7.2 cells. Biochem. Biophys. Res. Commun. 274, 136–141. 21. Powis, G., Kirkpatrick, D.L., Angulo, M. & Baker, A. (1998) Thioredoxin redox control of cell growth and death and the effects of inhibitors. Chemico-Biol. Interacts 111–112, 23–34. 22. Jeng, M.F., Campbell, A.P., Begley, T., Holmgren, A., Case, D.A., Wright, P.E. & Dyson, H.J. (1994) High-resolution solution structures of oxidized and reduced Escherichia coli thioredoxin. Structure 2, 853–868. 23. Weichsel, A., Gasdaska, J.R., Powis, G. & Montfort, W.R. (1996) Crystal structures of reduced, oxidized, and mutated human thioredoxins: evidence for a regulatory homodimer. Structure 4, 735–751. 24. Qin, J., Clore, G.M. & Gronenborn, A.M. (1994) The high- resolution three-dimensional solution structures of the oxidized and reduced states of human thioredoxin. Structure 2, 503–522. 25. Qin, J., Clore, G.M., Kennedy, W.M., Huth, J.R. & Gronenborn, A.M. (1995) Solution structure of human thioredoxin in a mixed disulfide intermediate complex with its target peptide from the transcription factor NF kappa B. Structure 3, 289–297. 26. Qin, J., Clore, G.M., Kennedy, W.P., Kuszewski, J. & Gronenborn, A.M. (1996) The solution structure of human thioredoxin complexed with its target from Ref-1 reveals peptide chain reversal. Structure 4, 613–620. 27. Powis, G., Mustacich, D. & Coon, A. (2000) The role of the redox protein thioredoxin in cell growth and cancer. Free Radic. Bio. Med. 29, 312–322. 28. Liang, P. & Pardee, A.B. (1992) Differential display of eukaryotic messenger RNA by means of the polymerase chain reaction. Science 257, 967–971. 29. Bauer, D., Mu ¨ ller, H., Reich, J., Riedel, H., Ahrenkiel, V., Warthoe, P. & Strauss, M. (1993) Identification of differentially expressed mRNA species by an improved display technique (DDRT-PCR). Nucleic Acids Res. 21, 4272–4280. Ó FEBS 2002 Crystal Structure of hTRXL-N (Eur. J. Biochem. 269) 2067 30. Holmgren, A. (1979) Thioredoxin catalyzes the reduction of insulin disulfides by dithiothreitol and dihydrolipoamide. J. Biol. Chem. 254, 9627–9632. 31. Chen, X., Jin, J., Gao, Y., Guo, Q., Sun, Y., Tang, H., Yuan, J., Qiang, B. & Rao, Z. (2001) Crystallization and preliminary X-ray analysis of a Trx domain of human thioredoxin-like protein. Acta. Crystallogr. 57, 1712–1714. 32. Otwinowski, Z. & Minor, W. (1997) Processing of X-ray diffrac- tion data collected in oscillation mode. Methods Enzymol. 276, 307–326. 33. Bru ¨ nger, A.T., Adams, P.D., Clore, G.M., DeLano, W.L., Gros, P., Grosse-Kunstleve, R.W., Jiang, J.S., Kuszewski, J., Nilges, M., Pannu, N.S. et al. (1998) Crystallography & NMR system: a new software suite for macromolecular structure determination. Acta. Crystallogr. 54, 905–921. 34. Jones, T.A., Zou, J.Y., Cowan, S.W. & Kjeldgaard, M. (1991) Improved methods for building protein models in electron density maps and the location of errors in these models. Acta Crystallogr. 47, 110–119. 35. Esnouf, R.M. (1997) An extensively modified version of MolScript that includes greatly enhanced coloring capabilities. J. Mol. Graph. Model 15, 132–133. 36. Nicholls, A., Sharp, K.A. & Honig, B. (1991) Protein folding and association: insights from the interfacial and thermodynamic properties of hydrocarbons. Proteins 11, 281–296. 37. Merritt, E.A. & Murphy, M.E.P. (1994) Raster3d, Version 2.0. A program for photorealistic molecular graphics. Acta Crystallogr. 50, 869–873. 38. Antonio, M.V., Jan-A ˚ ke, G. & Giannis, S. (1998) Molecular cloning and expression of a cDNA encoding a human thioredoxin- like protein. Biochem. Biophys. Res. Commun. 243, 284–288. 39. Lee, K.K., Murakawa, M., Takahashi, S., Tsubuki, S., Kawashima, S., Sakamaki, K. & Yonehara, S. (1998) Purification, molecular cloning, and characterization of TRP32, a novel thioredoxin-related mammalian protein of 32 kDa. J. Biol. Chem. 273, 19160–19166. 40. Freedman, R.B., Hawkins, H.C., Murant, S.J. & Reid, L. (1988) Protein disulphide-isomerase: a homologue of thioredoxin implicated in the biosynthesis of secretory proteins. Biochem. Soc. Trans. 16, 96–99. 41. Ghosh, G., Duyne, G.V., Ghosh, S. & Sigier, P.B. (1995) Structure of NF-kappa B p50 homodimer bound to a kappa B site. Nature 373, 303–310. 42. Mu ¨ ller, G.W., Rey, F.A., Sodeoka, M., Verdine, G.L. & Harrison, S.C. (1995) Structure of the NF-kappa B p50 homodimer bound to DNA. Nature 373, 311–317. 43. Krause, G. & Holmgren, A. (1991) Substitution of the conserved tryptophan 31 in Escherichia coli thioredoxin by site-directed mutagenesis and structure-function analysis. J. Biol. Chem. 266, 4056–4066. 44. Ren, X., Bjornstedt, M., Shen, B., Ericson, M.L. & Holmgren, A. (1993) Mutagenesis of structural half-cystine residues in human thioredoxin and effects on the regulation of activity by seleno- diglutathione. Biochemistry 32, 9701–9708. 45. Holmgren, A. (1977) Bovine thioredoxin system. Purification of thioredoxin reductase from calf liver and thymus and studies of its function in disulfide reduction. J. Biol. Chem. 252, 4600– 4606. 46. Cuff, J.A., Clamp, M.E., Siddiqui, A.S., Finlay, M., Barton, G. & J. (1998) JPred: a consensus secondary structure prediction server. Bioinformatics 14, 892–893. 47. Cuff, J.A. & Barton, G.J. (1999) Evaluation and improvement of multiple sequence methods for protein secondary structure pre- diction. Proteins 34, 508–519. 48. Di Francesco, V., Munson, P.J. & Garnier, J. (1999) FORESST: fold recognition from secondary structure predictions of proteins. Bioinformatics 15, 131–140. 49. Stuart, D.I., Levine, M., Muirhead, H. & Stammers, D.K. (1979) Crystal structure of cat muscle pyruvate kinase at a resolution of 2.6 A ˚ . J. Mol. Biol. 134, 109–142. 2068 J. Jin et al. (Eur. J. Biochem. 269) Ó FEBS 2002 . Gao, Y. , Guo, Q., Sun, Y. , Tang, H., Yuan, J., Qiang, B. & Rao, Z. (2001) Crystallization and preliminary X-ray analysis of a Trx domain of human thioredoxin-like. Crystal structure of the catalytic domain of a human thioredoxin-like protein Implications for substrate specificity and a novel regulation mechanism Jian