Search tips
Search criteria 


Biochem J. 2008 September 15; 414(Pt 3): 375–381.
Published online 2008 August 27. Prepublished online 2008 June 4. doi:  10.1042/BJ20080889
PMCID: PMC2552391

Structural and mechanistic insights into type II trypanosomatid tryparedoxin-dependent peroxidases


TbTDPX (Trypanosoma brucei tryparedoxin-dependent peroxidase) is a genetically validated drug target in the fight against African sleeping sickness. Despite its similarity to members of the GPX (glutathione peroxidase) family, TbTDPX2 is functional as a monomer, lacks a selenocysteine residue and relies instead on peroxidatic and resolving cysteine residues for catalysis and uses tryparedoxin rather than glutathione as electron donor. Kinetic studies indicate a saturable Ping Pong mechanism, unlike selenium-dependent GPXs, which display infinite Km and Vmax values. The structure of the reduced enzyme at 2.1 Å (0.21 nm) resolution reveals that the catalytic thiol groups are widely separated [19 Å (0.19 nm)] and thus unable to form a disulphide bond without a large conformational change in the secondary-structure architecture, as reported for certain plant GPXs. A model of the oxidized enzyme structure is presented and the implications for small-molecule inhibition are discussed.

Keywords: dithiol-dependent peroxidase, drug discovery, glutathione peroxidase, Leishmania, Trypanosoma, trypanothione
Abbreviations: GPX, glutathione peroxidase; His6, hexahistidine; Lm, Leishmania major; PEG, poly(ethylene glycol); Pt, Populus trichocarpaxdeltoides (hybrid poplar); r.m.s.d., root mean square deviation; Tb, Trypanosoma brucei; TDPX, tryparedoxin-dependent peroxidase; TryX, tryparedoxin


Human African trypanosomiasis (sleeping sickness) is responsible for an estimated 80000 new cases and 30000 deaths each year [1]. Current therapeutics against the protozoan parasite (Trypanosoma brucei sspp.) are inadequate and new drugs are urgently needed [2].

Much research has focussed on the unique thiol metabolism of trypanosomes, centred around the metabolite trypanothione [N1,N8-bis(glutathionyl)spermidine] [3]. The trypanothione peroxidase system prevents damage by oxidative stress by reduction of peroxides via NADPH, trypanothione reductase, trypanothione, TryX (tryparedoxin) and tryparedoxin peroxidase, a two-cysteine-residue-containing peroxiredoxin. All components of this cascade are essential for parasite survival [48].

Recently, a new member of this antioxidant defence system has been identified and characterized in T. brucei [911], T. cruzi [12] and Leishmania major [13]. These TDPXs (tryparedoxin-dependent peroxidases) show highest homology, among the mammalian proteins, with GPX4, the monomeric phospholipid GPX (glutathione peroxidase). However, glutathione is not a physiological substrate for these TDPXs, which use TryX, a thioredoxin-like protein, as an electron donor. Moreover, the active-site selenocysteine residue in GPXs is replaced by a cysteine residue for reduction of a broad range of hydroperoxides, including H2O2 and complex lipid hydroperoxides. A resolving cysteine residue missing in all human GPXs is essential for regeneration by TryX. The essentiality of TDPX in T. brucei, shown by RNA-interference studies [8], together with the different reaction mechanism from the mammalian GPX, suggest that TDPX could be a druggable target in these parasites.

There are three TDPXs in T. brucei, which are localized in the mitochondrion and the cytosol [10]. Apart from putative targeting sequences at the N- and C-termini, the core of these proteins is highly conserved. Previous studies have kinetically characterized a His6 (T. brucei TDPX3)(hexahistidine)-tagged TbTDPX3 that included the putative mitochondrial targeting sequence [9]. Here we have kinetically analysed the non-tagged and shorter recombinant TbTDPX2 and compared the results with those obtained with TbTDPX3. We report the first TDPX crystal structure from a trypanosomatid, the reduced form of TbTDPX2 at 2.1 Å (1 Å=0.1 nm), and present a model of the oxidized form that may assist in the future design of inhibitors against this target.


Cloning, expression, purification and characterization

TbTDPX2 from strain S427 was cloned into pET15b (Novagen), expressed in Escherichia coli BL21(DE3)pLysS and purified as the His6-tagged protein essentially as described for LmTDPX1 (Leishmania major TDPX1) [13].

The His6 tag was removed with thrombin and the enzyme purified to homogeneity by chromatography on nickel–Sepharose, benzamidine–Sepharose and gel filtration in 50 mM Tris/HCl buffer, pH 8.0. Assay conditions were essentially as described in [9]. Briefly, peroxidase activity was measured in a total volume of 250 μl at 25 °C containing 100 mM Tris/HCl, pH 7.6, 5 mM EDTA, 250 μM NADPH, 5 units of Tb trypanothione reductase, 100 μM trypanothione disulphide, 5 to 12.5 μM Tb TryX, 0.2 μM TbTDPX2 and 50–800 μM cumene hydroperoxide. Kinetic data were analysed as described in [13].

For crystallization, His6-tagged protein was reduced with 50 mM dithiothreitol, concentrated (Vivaspin 20 protein concentrator; Sartorius) and applied to a Superdex S75 26/60 (GE Healthcare) column equilibrated with 10 mM Tris/HCl, pH 8.0. TbTDPX2-containing fractions were combined, 10 mM dithiothreitol and 1 mM EDTA (final concns.) were added, and the protein was concentrated to 7.5 mg/ml before crystallization.

Crystallization, data collection, structure solution and refinement

Crystals of His6-tagged TbTDPX2 were grown by the hanging-drop method at 18 °C. Drops (1 μl) of protein (7.5 mg/ml) and reservoir solution {22% (w/v) PEG [poly(ethylene glycol)] 3350 and 0.1 M Tris/HCl, pH 8.0} were assembled, and monoclinic crystals (1.0 mm×0.05 mm×0.05 mm) were allowed to grow overnight. Crystals were cryoprotected in 35% PEG 3350 and 0.1 M Tris/HCl, pH 8.0, and flash-cooled in a stream of nitrogen at 103 K. Data were collected at the European Synchrotron Radiation Facility (Grenoble, France) on beamline ID14 EH4 using an ADSC QUANTUM Q315r CCD (charge-coupled device) detector and at a wavelength of 0.979 Å. A single crystal was used to collect 180° of data in seven batches, as it was translated along its length to avoid the effects of radiation damage. Data were processed and scaled using XDS (X-ray Detector Software) [14] and statistics are shown in Supplementary Table S1 at The crystals contained one molecule in the asymmetric unit, the space group was P21 with a Matthews coefficient of 2.2 Å3/Da, and the solvent content was 44%.

Potential structural homologue candidates were identified in the RSCB PDB (Research Collaboratory for Structural Bioinformatics Protein Data Bank). Chain A of structure 2P5Q, the reduced form of PtGPX5 [Populus trichocarpaxdeltoides (hybrid poplar) GPX5] [15], showing a sequence identity of 52%, was used for molecular replacement in MOLREP [16]. The solution from MOLREP underwent rigid body refinement followed by restrained refinement in REFMAC5 [17,18], giving an Rfactor and Rfree of 32 and 40% respectively. Residues were mutated from the PtGPX5 sequence to that of TbTDPX2 using COOT [19], and further cycles of restrained refinement interspersed with electron-density inspection, manual model building and manipulation and water addition were performed with REFMAC5 and COOT. All residues fall into the preferred and allowed regions of the Ramachandran plot. Surface-charge representations were prepared and visualized using PDB2SQR [20], APBS [21] and PyMOL (DeLano Scientific LLC). Refinement statistics can be found in Supplementary Table S1.

Sequence and structural alignments and modelling

Structural and sequence homologues were found using DALI [22] and BLAST searches of the PDB and Swiss-Prot databases. Only one structure, 1GP1 (bovine erythrocyte GPX1; [23]) was returned by DALI as a structural homologue, with an r.m.s.d. (root mean square deviation) of 1.8 Å over 159 Cα-atoms. However, the BLAST search of the PDB revealed several members of the GPX family, which when overlaid using LSQMAN [24,25] gave r.m.s.d. values of between 1.0 Å and 1.4 Å over more than 120 Cα-atoms and sequence identities of between 29 and 52%. SWISS-MODEL [25] was used to prepare a structural model of the oxidized form of TbTDPX2 on the basis of the structure of oxidized PtGPX5. The three-dimensional structures were visualized and analysed with COOT and PyMOL.


Kinetic analysis

Previous kinetic analyses of His6-tagged TbTDPX3 [9] showed no saturation kinetics, in contrast with that of non-tagged LmTDPX1 [13]. TbTDPX3 has a putative N-terminal mitochondrial targeting sequence, which, together with the His6 tag, might interfere with enzyme activity. Thus we chose to work with the shorter TbTDPX2 (Figure 1). Three independent PCRs were performed using genomic DNA from Tb strain 427 as template. All clones analysed contained two point mutations at the DNA level (C233A and T246C) in comparison with TbTDPX2 in the genome reference strain (gene Tb927.7.1130). C233A causes a mutation (T78N) at the amino acid level, and T246C is silent (Figure 1).

Figure 1
Sequence–structure alignment of TDPXs and homologues

Recombinant protein was expressed in E. coli and purified to homogeneity with and without the N-terminal His6 tag (Supplementary Figure S1A at SDS/PAGE analysis shows that non-tagged TbTDPX2, which has a predicted molecular mass of 19168 Da, runs as a monomer in both its reduced and oxidised forms (Supplementary Figure S1B). Additionally, analytical size-exclusion chromatography shows that both forms are eluted as single monomeric peaks with molecular masses of 19600 and 25400 Da for the reduced and oxidized proteins respectively (Supplementary Figure S1C). The minor size differences observed using both methods is likely to be a result of the conformational change predicted to occur between the reduced and oxidized states as outlined below.

Kinetic data for non-tagged TbTDPX2 were obtained by varying hydroperoxide concentrations at several fixed TryX concentrations. Results were analysed by fitting individual data sets to the Michaelis–Menten equation to obtain Vappmax for each concentration of TryX (Figure 2A). A double-reciprocal plot of 1/Vappmax versus 1/[TryX] gave finite values for Km and Vmax (Figure 2B). Since the double-reciprocal transformations of the original data set were parallel, the entire data set was globally fitted by non-linear regression to the equation describing a Ping Pong mechanism (Figure 2C). Similar results were found by analysis of progress curves using the integrated Dalziel equation analysed as described for TbTDPX3 [9] (Supplementary Figure S2 at The results of these analyses are summarized in Table 1 in comparison with published data for TbTDPX3 and LmTDPX1. The rate constants k1 and k2 for cumene hydroperoxide and TryX respectively are in reasonable agreement with those reported for TbTDPX3 [11], except that our results are compatible with finite Km values in the micromolar range rather than infinite ones. Nonetheless, the kinetic behaviour of TbTDPX2 and TbTDPX3 are broadly similar and thus the N-terminal His6 tag and N-terminal extension present in TbTDPX3 does not significantly perturb kinetic behaviour. Interestingly, LmTDPX1 and TbTDPX2 show similar catalytic-centre activity (kcat) and Km values for cumene hydroperoxide, but differ markedly in Km values for TryX, which is at least ten times lower for LmTDPX1.

Figure 2
Kinetic analysis of TDPX2 with TryX and cumene hydroperoxide as substrate
Table 1
Kinetic properties of TDPX with TryX as reducing agent and cumene hydroperoxide as substrate

Overall structure and comparisons

The crystal structure of TbTDPX2 has been solved using data to 2.1 Å and displays the well-characterized thioredoxin fold [26,27]. The structure (Figure 3) is constructed around a seven-stranded twisted β-sheet with parallel (β3–β4–β5–β1) and antiparallel (β3–β6–β7) sections sharing β3. The sheet is flanked by three α-helices on one side (α1, α2 and α4) and one on the other (α3). The secondary structure begins with a loop of two 310 helices (θ1 and θ2) separated by a β-hairpin (β1 and β2) followed by a β–α–β unit (β3–α1–β4). The model contains 163 residues from Ala3 to Thr166. The His6-affinity tag, N-terminal residues 1 and 2, and C-terminal residues 167 to 169 could not be modelled, owing to weak electron density, suggesting a high level of flexibility in these regions.

Figure 3
Ribbon diagram of TbTDPX2 with secondary-structure elements labelled

A BLAST search with the sequence for TbTDPX2 returned many homologues that can be grouped with those from other trypanosomes (~70% identity), plants (~50% identity) and animals (~30% identity) (Figure 1). Only eight structural homologues were found in the Protein Data Bank: bovine erythrocyte GPX1 (1GP1), human GPX1, GPX2, GPX3, GPX4, GPX5 and GPX7 (2F8A, 2HE3, 2R37, 2OBI [28], 2I3Y and 2P31), and the two PtGPX5 structures (2P5Q and 2P5R [15]). These structures were overlaid, revealing that the core subunit structures were well conserved, with r.m.s.d. values ranging from 1.0 Å over 155 Cα atoms for hGPX4, to 1.4 Å over 148 Cα atoms for hGPX3.

TbTDPX2 exhibits strong similarities to the related GPXs. Unlike the majority of GPX structures that exist as either homodimers or homotetramers in solution, TbTDPX2 exists as a monomer (Supplementary Figures S1B and S1C). Among the human GPXs, only hGPX4 is reported to be monomeric [28]. A minor deviation from the overlaid backbones is seen in loop α3–β6 of TbTDPX2 (Figure 4). Although the conformation is slightly different in GPX4, this loop is the same length in both structures and shorter than the long surface-exposed loop seen in the other GPX isozymes. It has been suggested that the existence of the extended loop in other GPXs limits accessibility to the active site [28], and the short loop in hGPX4 and TbTDPX2 may explain their broad substrate specificity. Two additional features of the TbTDPX2 structure, α2 and the short loop α2–β5, share the same position as their equivalents in GPX4 (Figure 4), but, as can be seen in the oligomeric GPXs, these structural elements are involved in intersubunit interfaces and are therefore determinants of the oligomerization state of this family. Interestingly, in the monomeric TbTDPX2, helix α2 contains the resolving Cys87 as does the dimeric PtGPX5. However, in PtGPX5 the dimer interface is different to those formed by the oligomeric selenocysteine-containing mammalian GPXs, and so the resolving cysteine residue of PtGPX5 is not constrained by the interface [15].

Figure 4
Cα backbone representations of TbTDPX2, human GPX4 and bovine GPX1

Bipartite active site

The active site of mammalian GPXs is comprised of a well-conserved cysteine, glutamine and tryptophan catalytic triad [23]. Mutational studies in TbTDPX3 suggests that Cys47 and Gln82 (equivalent to Cys39 and Gln74 in TbTDPX2) are essential for activity [11]. Mutation of Trp137 to glycine, however, resulted in an enzyme with low, but significant, activity, suggesting that it plays a structural, rather than a catalytic, role [11]. Overlaying TbTDPX2 with the other structurally described GPXs puts the triad components (Cys39, Gln74 and Trp129; Figure 5A) in similar positions, but in TbTDPX2 the distance between the Cys39 Sγ and Gln74 Oϵ1 atoms is on average 0.4 Å greater than with the other homologues, whereas the distance between Cys39 and the Trp129 Nϵ1 atom is on average 0.5 Å shorter.

Figure 5
Detailed stick representation of (A) the peroxidatic cysteine and (B) the resolving cysteine environments

One of the most unusual features of the reduced form of TbTDPX2 is that the peroxidatic Cys39 and resolving Cys87 are situated 19 Å away from each other (Figure 3) and reside in distinct environments. The peroxidatic Cys39 is found in the loop connecting β3 and α1 and, although this loop is solvent-accessible, the side chain of Cys39 points towards the molecule's interior, as seen in the reduced structure of PtGPX5 [15]. Cys39 is surrounded by Ala36, Tyr41, Gln74, Trp129, Asn130 and Phe131 that belong to loops β3–α1, β4–α2 and α3–β6 (Figure 5A). These residues are all conserved in the GPX family, except Tyr41, which is specific to TbTDPX2. The Cys39 Sγ atom accepts a hydrogen bond from the Tyr41 N atom (3.2 Å) and is 3.5 Å from the Trp129 Nϵ1 atom, 3.4 Å from the Asn130 Nδ2 atom and 3.9 Å from the Gln74 Oϵ1 atom. A water molecule sits 3.4 Å from the Cys39 Sγ atom on the surface of TbTDPX2. The Cys39 N atom hydrogen-bonds to the Thr42 OH group, which is conserved in TDPXs and GPXs. It is not essential for catalysis [11], but appears to stabilize the active-site structure. Both Ala36 and Tyr41 are situated on the same loop (β3–α1) as Cys39. Analysis of a surface-charge representation (Figure 6, left-hand panel) shows that the active-site region around Cys39 is not a well-defined pocket or cleft. However, it lies in a predominantly positively charged region of the protein that may influence its substrate specificity.

Figure 6
Surface electrostatic potential representation of TbTDPX2

The resolving Cys87 is situated at the C-terminus of helix α2. It lies on the surface of the protein, is solvent-accessible and is situated at one end of a distinct pocket which is 14 Å long and 10 Å wide (Figure 6, right-hand panel). The base of the pocket is lined by residues Tyr20, Pro95 and Ile96, with Val86 immediately below Cys87 (Figure 5B). Down one side are the charged side chains of Lys18, Asp16 and Lys83. The opposite side is lined by main-chain atoms of Ala92, Glu93 and Phe94, and a negatively charged patch of Asp78, Glu79, Glu80, Glu81 and Glu84 is seen on the electrostatic surface representation immediately above the charged side of the pocket. Cys87 does not interact directly with other residues in TbTDPX2 in the reduced state. This is particularly unusual, as it is situated in an α-helix, and although neighbouring residues follow the normal hydrogen-bonding pattern seen in such helices, Cys87 appears to remain non-bonded and flexible, suggesting its prerequisite for movement.

Possible mechanism and druggability

TDPXs are predicted to undergo significant conformational changes between their respective reduced and oxidized structures [11,13], as observed for PtGPX5 [15]. A model based on oxidized PtGPX5 (Figure 7) suggests that α2 must completely unravel, causing loop β4–α2 to bulge outwards and allowing Cys87 to travel some 12 Å to form a disulphide bond with Cys39, which has itself moved 10 Å on the flexible β3–α11 loop. The trigger for this conformational change, which must reside in conversion of the Cys39 thiolate group into a sulphenate group, merits further investigation. These complex structural changes could explain the finite Ping Pong mechanism found here by kinetic analysis (Figure 2 and Supplementary Figure S2) in contrast with the selenium type of glutathione peroxidases, which lack the resolving cysteine residue and display an infinite kcat.

Figure 7
Ribbon and surface potential representations of the model of oxidized TbTDPX2 based on PtGPX5

Although the peroxidative Cys39 occupies a rather flat featureless surface in the reduced form, the environment of Cys87 forms a defined pocket that could be exploited for drug design (Figure 6). Likewise the model of the oxidized form predicts a deep pocket immediately above the disulphide bond (Figure 7). Both forms therefore present potentially druggable sites for small-molecule inhibition. Structural studies on the oxidized form of TbTDPX2 are necessary to confirm this prediction.

Online data

Supplementary Table S1 and Supplementary Figures S1-S2:


We thank Dr Paul Fyfe and Dr Scott Cameron for crystallographic data collection at the European Synchrotron Radiation Facility. This work is supported by the Wellcome Trust (grant nos. WT 079838 and WT 083481).


1. Stuart K. D., Brun R., Croft S. L., Fairlamb A. H., Gurtler R. E., McKerrow J. H., Reed S., Tarleton R. L. Kinetoplastids: related protozoan pathogens, different diseases. J. Clin. Invest. 2008;118:1301–1310. [PMC free article] [PubMed]
2. Fairlamb A. H. Chemotherapy of human African trypanosomiasis: current and future prospects. Trends Parasitol. 2003;19:488–494. [PubMed]
3. Fairlamb A. H., Cerami A. Identification of a novel, thiol-containing co-factor essential for glutathione reductase enzyme activity in trypanosomatids. Mol. Biochem. Parasitol. 1985;14:187–198. [PubMed]
4. Krieger S., Schwarz W., Ariyanayagam M. R., Fairlamb A. H., Krauth-Siegel R. L., Clayton C. Trypanosomes lacking trypanothione reductase are avirulent and show increased sensitivity to oxidative stress. Mol. Microbiol. 2000;35:542–552. [PubMed]
5. Ariyanayagam M. R., Oza S. L., Guther M. L., Fairlamb A. H. Phenotypic analysis of trypanothione synthetase knockdown in the African trypanosome. Biochem. J. 2005;391:425–432. [PubMed]
6. Comini M. A., Guerrero S. A., Haile S., Menge U., Lunsdorf H., Flohé L. Validation of Trypanosoma brucei trypanothione synthetase as drug target. Free Radical Biol. Med. 2004;36:1289–1302. [PubMed]
7. Comini M. A., Krauth-Siegel R. L., Flohé L. Depletion of the thioredoxin homologue tryparedoxin impairs antioxidative defence in African trypanosomes. Biochem. J. 2007;402:43–49. [PubMed]
8. Wilkinson S. R., Horn D., Prathalingam S. R., Kelly J. M. RNA interference identifies two hydroperoxide metabolizing enzymes that are essential to the bloodstream form of the African trypanosome. J. Biol. Chem. 2003;278:31640–31646. [PubMed]
9. Hillebrand H., Schmidt A., Krauth-Siegel R. L. A second class of peroxidases linked to the trypanothione metabolism. J. Biol. Chem. 2003;278:6809–6815. [PubMed]
10. Schlecker T., Schmidt A., Dirdjaja N., Voncken F., Clayton C., Krauth-Siegel R. L. Substrate specificity, localization, and essential role of the glutathione peroxidase-type tryparedoxin peroxidases in Trypanosoma brucei. J. Biol. Chem. 2005;280:14385–14394. [PubMed]
11. Schlecker T., Comini M. A., Melchers J., Ruppert T., Krauth-Siegel R. L. Catalytic mechanism of the glutathione peroxidase-type tryparedoxin peroxidase of Trypanosoma brucei. Biochem. J. 2007;405:445–454. [PubMed]
12. Wilkinson S. R., Meyer D. J., Taylor M. C., Bromley E. V., Miles M. A., Kelly J. M. The Trypanosoma cruzi enzyme TcGPXI is a glycosomal peroxidase and can be linked to trypanothione reduction by glutathione or tryparedoxin. J. Biol. Chem. 2002;277:17062–17071. [PubMed]
13. König J., Fairlamb A. H. A comparative study of type I and type II tryparedoxin peroxidases in Leishmania major. FEBS J. 2007;274:5643–5658. [PubMed]
14. Kabsch W. Automatic processing of rotation diffraction data from crystals of initially unknown symmetry and cell constants. J. Appl. Crystallogr. 1993;26:795–800.
15. Koh C. S., Didierjean C., Navrot N., Panjikar S., Mulliert G., Rouhier N., Jacquot J. P., Aubry A., Shawkataly O., Corbier C. Crystal structures of a poplar thioredoxin peroxidase that exhibits the structure of glutathione peroxidases: Insights into redox-driven conformational changes. J. Mol. Biol. 2007;370:512–529. [PubMed]
16. Vagin A., Teplyakov A. MOLREP: an automated program for molecular replacement. J. Appl. Crystallogr. 1997;30:1022–1025.
17. Collaborative Computational Project, Number 4 The CCP4 suite: programs for protein crystallography. Acta Crystallogr. Sect. D Biol. Crystallogr. 1994;50:760–763. [PubMed]
18. Murshodov G. Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallogr. D. 1997;53:240–255. [PubMed]
19. Emsley P., Cowtan K. Coot: model-building tools for molecular graphics. Acta Crystallogr. D. 2004;60:2126–2132. [PubMed]
20. Dolinsky T. J., Nielsen J. E., McCammon J. A., Baker N. A. PDB2PQR: an automated pipeline for the setup of Poisson–Boltzmann electrostatics calculations. Nucleic Acids Res. 2004;32:W665–W667. [PMC free article] [PubMed]
21. Baker N. A., Sept D., Joseph S., Holst M. J., McCammon J. A. Electrostatics of nanosystems: application to microtubules and the ribosome. Proc. Natl. Acad Sci. U.S.A. 2001;98:10037–10041. [PubMed]
22. Holm L., Sander C. Mapping the protein universe. Science. 1996;273:595–603. [PubMed]
23. Epp O., Ladenstein R., Wendel A. The refined structure of the selenoenzyme glutathione peroxidase at 0.2 nm resolution. Eur. J. Biochem. 1983;133:51–69. [PubMed]
24. Kleywegt G. J., Jones T. A. Where freedom is given, liberties are taken. Structure. 1995;3:535–540. [PubMed]
25. Schwede T., Kopp J., Guex N., Peitsch M. C. SWISS-MODEL: an automated protein homology-modelling server. Nucleic Acids Res. 2003;31:3381–3385. [PMC free article] [PubMed]
26. Martin J. L. Thioredoxin – a fold for all reasons. Structure. 1995;3:245–250. [PubMed]
27. Qi Y., Grishin N. V. Structural classification of thioredoxin-like fold proteins. Proteins. 2005;58:376–388. [PubMed]
28. Scheerer P., Borchert A., Krauss N., Wessner H., Gerth C., Hohne W., Kuhn H. Structural basis for catalytic activity and enzyme polymerization of phospholipid hydroperoxide glutathione peroxidase-4 (GPx4) Biochemistry. 2007;46:9041–9049. [PubMed]

Articles from Portland Press Open Access are provided here courtesy of Portland Press Ltd