- Top of page
- Materials and methods
The solution structure of the protein disulfide oxidoreductase Mj0307 in the reduced form has been solved by nuclear magnetic resonance. The secondary and tertiary structure of this protein from the archaebacterium Methanococcus jannaschii is similar to the structures that have been solved for the glutaredoxin proteins from Escherichia coli, although Mj0307 also shows features that are characteristic of thioredoxin proteins. Some aspects of Mj0307's unique behavior can be explained by comparing structure-based sequence alignments with mesophilic bacterial and eukaryotic glutaredoxin and thioredoxin proteins. It is proposed that Mj0307, and similar archaebacterial proteins, may be most closely related to the mesophilic bacterial NrdH proteins. Together these proteins may form a unique subgroup within the family of protein disulfide oxidoreductases.
The regulation of protein disulfide bond formation and reduction is critical not only to protein structure, but it is also an effective electron transport mechanism for cellular processes ranging from ribonucleotide reduction to coordination of light and dark photosynthetic reactions (Holmgren 1989; Schurmann 1995; Prinz et al. 1997; Gilbert 1998). This wide range of protein disulfide bond activity is accomplished primarily by the protein disulfide oxidoreductases. All the members of this large family of proteins share a common structural fold, which contains an active site tetrapeptide sequence CXXC (Martin 1995). The active site cystine residues are capable of reversibly forming disulfide bonds that provide the redox character of these proteins. The common structural fold is composed of a βαβαββα secondary structure pattern, which has been termed the “thioredoxin fold.”
Thioredoxin (Trx) was the first member of the protein disulfide oxidoreductase family to be identified (Laurent et al. 1964). Glutaredoxins (Grx) were identified a decade later in Trx-deficient Escherichia coli (Holmgren 1976). Together, the biochemical and structural properties of Trx and Grx proteins have been extensively studied. Functionally, these proteins are distinguished by their reducing agents and substrates. Grx proteins are reduced by glutathione and glutaredoxin reductase, whereas Trx proteins are reduced by thioredoxin reductase (Trr). Furthermore, Trx proteins act more as general protein disulfide reductases, and Grx proteins display a reductive preference for glutathione mixed disulfide substrates (Berardi et al. 1998). Structurally, the bacterial and T4 Grx proteins contain only the minimal βαβαββα secondary structural elements of the thioredoxin fold, whereas several eukaryotic Grx proteins possess additional helices at the N and C termini (i.e., a αβαβαββαα secondary structure pattern). Trx proteins, however, have a βαβαβαββα secondary structure pattern that has been conserved in both bacterial and eukaryotic organisms.
Other well-studied members of the protein disulfide oxidoreductases include the bacterial Dsb proteins and eukaryotic protein disulfide isomerases (PDI). These proteins are quite different from the Trx and Grx proteins. Although they incorporate a thioredoxin fold domain, they have much longer amino acid sequences and possess additional structural domains (Martin et al. 1993; Kemmink et al. 1997; McCarthy et al. 2000). The Dsb and PDI proteins appear to be essential for promoting proper protein folding through disulfide bond shuffling (Bardwell et al. 1993; Gilbert 1998).
There are several other proteins with disulfide oxidoreductase activity that do not fit into any of the well-established subgroups. An example of these unclassified disulfide oxidoreductases is the NrdH group of proteins. These were first isolated from Lactococcus lactis and nearly identical homologs have been identified in E. coli and Salmonella typhimurium (Jordan et al. 1996, 1997). Despite the similarity to Grx proteins in both size and predicted secondary structure, the NrdH proteins are incapable of binding glutathione but interact well with Trr. This blend of Trx and Grx attributes, in addition to a limited knowledge of in vivo function, has made it difficult to classify or define the position of NrdH proteins in the disulfide oxidoreductase family.
The advent of whole genome sequencing data has facilitated the search for novel protein disulfide oxidoreductases. Far less is known about these proteins in archaebacteria than in the prokaryotes and eukaryotes. McFarlan and co-workers identified a small disulfide oxidoreductase protein (Mt0807) from Methanobacterium thermoautotrophicum that was similar to Grx in its sequence but displayed behavior inconsistent with either the Grx or the Trx proteins (McFarlan et al. 1992). Open reading frames from the archaebacterial genomes of Methanococcus jannaschii (gene Mj0307) and Archaeoglobus fulgidus (gene Af2145) contain proteins that are homologous with Mt0807 (51% and 34% sequence identity for Mj0307 and Af2145, respectively; see Fig. 1 for a sequence comparison of these proteins and other protein disulfide oxidoreductases). Lee and co-workers have reported that Mj0307 also does not display glutathione-dependent behavior but does weakly interact with E. coli Trr (Lee et al. 2000).
Given the importance of protein disulfide oxidoreductases in all organisms, it is of considerable interest to better understand the structure and function of these novel archaebacterial protein disulfide oxidoreductases. To this end, we report the solution structure of Mj0307 in its reduced state. Using these structural data, in combination with sequence alignments, we are able to explain several functional features of Mj0307. The data also suggest that these archaebacterial proteins and the mesophilic bacterial NrdH proteins may belong to the same subgroup within the protein disulfide oxidoreductase family.
- Top of page
- Materials and methods
The hyperthermophilic nature of M. jannaschii (optimum growth temperature of 85°C) was exploited in the resonance assignment process of Mj0307 by collecting the experiments relevant for assignments at 60°C. The elevated temperature significantly improves the quality of experiments that exclusively use J coupling for information transfer. The amide backbone (except for proline residues) and side chain assignments were established for all residues except M0, S1, and P11 through P17. There are unassigned 15N-1H correlations in the HSQC spectrum that could possibly belong to some of the unassigned residues, but assignments could not be made unambiguously. The pH of our nuclear magnetic resonance (NMR) samples was 6.1, and although biologically relevant, the amide proton exchange rate can be sufficient to lead to solvent exchange. Furthermore, the unassigned residues are located in solvent exposed and inherently flexible regions. It is possible that the combination of these inherent characteristics and sample properties may account for the inability to assign some of these residues. The assigned 15N-1H HSQC spectrum is shown in Figure 2, and a strip plot of slices from the 15N-resolved NOESY is shown in Figure 3.
Experiments that rely on the nuclear Overhauser effect (NOE) for resonance information transfer were of significantly lower quality at 60°C compared with those collected at 25°C or 35°C. This observation can be explained by considering the relationship between the intensity of the NOE and the rotational correlation time. Equation 1(1) displays the critical situation when the NOE vanishes, irrespective of the mixing time:
where ω0 is the Lamour frequency and τc is the rotational correlation time (Ernst et al. 1997). At values of 600 and 500 MHz, the critical rotational correlation times are 1.7 nsec and 2.2 nsec, respectively. As discussed below, Mj0307 is structurally similar to the E. coli Grx-1 protein, which has an estimated τc of 6.4 nsec at 25°C (Kelly et al. 1997). Assuming that the τc of Mj0307 is the same, then the τc at 60°C can be estimated by determining the ratio of τc25°C/τc60°C by using equation 2(2):
where r is the radius of the protein, η is the solvent viscosity, kB is the Boltzmann constant, and T is the temperature. Using this approach, one obtains an approximate τc60°C of 2.9ns. Thus, the lower quality of the NOE data sets collected at 60°C, especially at 600 MHz, is almost certainly the result of approaching the critical τc value.
Even though the initial resonance assignments were made at 60°C, only modest changes in the chemical shifts were observed at lower temperatures. The observed changes in the backbone amide cross-peak positions in the 15N-1H HSQC spectra are shown in Figure 4. The changes between 15N-1H HSQC cross-peaks positions at 35°C and 25°C are also shown in Figure 4 because the 15N edited [1H-1H]NOESY-HSQC, and the 3D and 4D 13C edited [1H-1H]NOESY experiments were optimized at those temperatures, respectively. Although the observed changes in the 15N-1H HSQC cross-peak positions are modest, it can be seen from Figure 4 that the most temperature-sensitive changes between 60°C and the lower temperatures occur in the loop regions and at the edges of secondary structure. The most pronounced changes are at the start of the α2 helix and junction between β4 and α3. The changes between 35°C and 25°C are trivial throughout the entire protein.
The short- and medium-range backbone amide proton NOEs, backbone 3-bond HαHN coupling constants, 1Hα and 13Cα chemical shift indices, and the backbone amide proton/deuteron exchange protection pattern for Mj0307 are presented in Figure 5. From these data, it can be seen that Mj0307 adopts the minimal βαβαββα secondary structure pattern of the thioredoxin fold. Using these constraints, as well as long-range NOEs, we determined the solution structure of Mj0307. The structure calculations were constrained by 505 NOEs, 52 backbone ϕ torsion angle, and 30 hydrogen bonds inferred from proton/deuteron exchange protection. Having started from 60 random conformers, we refined the 20 best calculated structures by using DYANA (Güntert et al. 1997) by energy minimization using the program OPAL (Luginbühl et al. 1996). The refined structure ensemble possesses 0.1 ± 0.3 NOE distance violations ≥0.1 Å; 0.1 ± 0.2 residual dihedral angle violations ≥2.5°; total AMBER energies of −3031 ± 146 kcal/mole (with −235 ± 13 kcal/mole and 3522 ± 163 kcal/mole contributions from van der Waals and electrostatic terms, respectively; see also Table 1). The refined structures were also analyzed with PROCHECK (Laskowski et al. 1993; Rullman 1996), which indicated that 97% of the residues fall into allowed or generously allowed regions of the Ramachandran map. A stereoview of the N, Cα, and C′ backbone atoms from all 20 refined structures (PDB 1FO5) are presented in Figure 6A, with the local root-mean-square deviation (RMSD) shown in Figure 6B. A representative conformer of the final 20 structures with Kabsch-Sander secondary structure rendering is presented in Figure 7A (Kabsch and Sander 1983).
The tertiary structure of Mj0307 is similar to the oxidized and reduced E. coli Grx-1 structures (Fig. 7B). Differences between these structures include the α2 helix, which is considerably shorter in Mj0307. The backbone conformation of the residues in the α2 helix are not as well defined as their counterparts in α1 and α3 helices. Using the Kabsch-Sander secondary structure rendering (Kabsch and Sander 1983), we found that residues Q46 to E50 adopt an α-helical conformation in some of the final 20 conformers, whereas in the other conformers these residues adopt a bend conformation. In the conformation displayed in Figure 7, A and B, these residues are in the helical conformation. Another difference between Mj0307 and E. coli Grx1 is in the connection between β4 strand and α3 helix. In Mj0307, this connection is composed of the six residues I68 to K73, whereas this junction in Grx-1 (and all mesophilic prokaryotic or eukaryotic Grx structures) is composed of a single glycine residue.
The calculated structures also show that almost all of the residues that display the greatest changes with temperature, shown in Figure 4, are confined to one of two regions. Residues A18, A19, V36, and Y38 form one cluster, and the second is formed by V65, I68, T72, A75, and L76. The structures show that the side chains of these clustered residues pack against each other, and whatever changes that occur between the 60°C and the lower temperatures are confined to these two regions.
Although the structural data reveal that Mj0307 is structurally similar to the mesophilic Grx proteins, its biological function in vivo is not as clear. In agreement with previous findings (Lee et al. 2000), we found that Mj0307 does function as a general disulfide oxidoreductase as shown by its ability to precipitate insulin in the presence of DTT (data not shown). Neither Mj0307 nor its archael homolog, Mt0807, functions as a Grx-like protein (McFarlan et al. 1992; Lee et al. 2000). Furthermore, Mt0807 was shown not to interact with Trr from either E. coli or Corynebacterium nephridii, but Mj0307 displayed a weak interaction with E. coli Trr. These results suggest a weak functional similarity to Trx proteins. The active site tetrapeptide sequence in Mj0307 (Mj0307 residues 13 through 16) is the same as the mesophilic bacterial DsbA protein, which is essential for disulfide bond rearrangement and has been shown to catalyze the refolding of scrambled RNaseA (Yu et al. 1993). Despite this similarity to DsbA, we were not able to observe Mj0307 catalysis of scrambled RnaseA refolding (results not shown).
In an attempt to gain a better insight into the possible biological function of Mj0307, we examined amino acid sequence alignments with several known disulfide oxidoreductase proteins. Sequence alignments with either mesophilic bacterial DsbA or eukaryotic PDI proteins are quite poor (∼12%–13% and 6%–9% sequence identities for DsbA and PDI proteins, respectively), but alignments with the NrdH proteins are significantly better (21%–23% sequence identity). Sequence alignments using Grx and Trx proteins from mesophilic bacteria and eukaryotes show the best sequence homology (19%–30% and 16%–30% for Trx and Grx, respectively).
To further examine the relationship between Mj0307 and known Grx, Trx, and NrdH proteins, we examined structure-based sequence alignments (Fig. 8). Several conserved structural features were used to generate the alignment in Figure 8, among which included the conserved phenylalanine (or tyrosine) residue at position 8 (unless otherwise stated, position numbers refer to the numbering of Fig. 8), the CXXC active site tetrapeptide, the cis-proline residue at position 68, and the glycine residue at position 83.
Several interesting structure/function-related features are apparent in Figure 8. The residues that have been shown to make several important contacts with bound glutathione in Homo sapiens and E. coli Grx (residue positions 83–85) form the N terminus of a helix. In Mj0307, however, these residues are part of a loop structure, which is similar to the Trx and NrdH proteins. Thus, not only are the residue identities of positions 83–85 between Mj0307 and Grx proteins not conserved, but the secondary structure is also not conserved either. These sequence and structural differences most likely account for the inability of Mj0307, and Mt0807 as well, to behave as Grx proteins. In Trx proteins, the residues of this loop region at positions 83–87 have been shown to be essential for the interaction with Trr (Eklund et al. 1991). In Mj0307, these residues are similar to those found in E. coli Trx. This similarity, absent in Mt0807, may be enough to allow for an interaction with Trr. Although the loop region of positions 83–87 of Mj0307 resembles Trx proteins, the β-strand/loop/β-strand motif of positions 69–81 is more like the Grx proteins rather than the Trx proteins. The NrdH proteins also share a Grx resemblance in this region.
The alignment in Figure 8 also reveals that the residues that are conserved between Mj0307, Mt0807, and Af2145 cluster in the same regions as the residues that have been shown to be essential for either glutathione binding in Grx or protein–protein interactions in Trx. This observation would imply that the binding surface has been conserved among the Grx, Trx, and the Mj0307, Mt0807, and Af2145 proteins. The identity of the residues conserved between Mj0307, Mt0807, and Af2145 have both Grx and Trx characteristics. Although the conserved residues in positions 83–87 have some resemblance to E. coli Trx, the residues flanking the conserved cis-proline residue (position 68) have a Grx likeness. This likeness is highlighted by the presence of a valine residue at position 67, which is highly conserved in Grx proteins. The conserved residues in the active site and the surrounding region are also interesting because they do not have much resemblance to either a Grx or a Trx. The active site CXXC has been found to be highly conserved between Grx and Trx proteins. Although Mt0807 and Af2145 have an active site that is the same as the Grx proteins, the Mj0307 active site is identical to DsbA. The residue that immediately follows CXXC motif is conserved in all Trx and eukaryotic Grx proteins as an arginine or lysine, but this position in mesophilic bacteria is conserved as a valine and conserved as a proline in Mj0307, Mt0807, and Af2145. The consensus sequence of the five residues preceding the active site is FXKXX and F(W/S)AXW for Grx and Trx proteins, respectively. In the archael proteins, these residues form the pentapeptide FTSP(T/M), which is different from the Grx and Trx proteins as well as the NrdH proteins.
- Top of page
- Materials and methods
We have determined the solution structure of the Mj0307 protein from the archaebacteria M. jannaschii in its reduced form. The solution structure shows that the βαβαββα secondary structure in this protein adopts the minimal thioredoxin fold and incorporates many structural features characteristic of this type of fold (Martin 1995). The secondary structure pattern in Mj0307 also has been identified by Lee and co-workers (Lee et al. 2000). Although the secondary structure assignments from our work coincide well with their results, there are two areas in which there are differences. One area is the starting point of the α2 helix. Our data indicate that this helix begins at residue A19, but Lee and co-workers suggest that it begins at H15. Because of the difficulty in obtaining resonance assignment information for the residues between P11 and P17, we do not necessarily disagree with this possibility. The observed JHαHN coupling constant for A18 in our data, however, does suggest that it is not in a helical conformation. The second discrepancy is the initiation of the β3 strand. Our data indicate that this strand stretches from T58 to I61, instead of M54 to I61 as Lee and co-workers suggest. Unlike the beginning of α2, our assignments and data are complete in this region. The JHαHN coupling constant for V56 and the presence of dαiNi + 1 NOEs at M54 and V56 are consistent with an extended strand conformation, but there is no amide backbone proton/deuteron exchange protection for any of the residues between I53 to V56. This would suggest that these residues do not adopt a well-formed β strand as do residues T58–-I61, and this observation is verified in the calculated structures.
Although the experiments for the structural data were collected at temperatures (25°C and 35°C) that are much lower than the native conditions of Mj0307 (upward of 85°C), the 1H-15N HSQC cross-peak positions display only a slight temperature dependence between 60°C and 25°C. The residues that display the most sensitivity to temperature are those that have side chains that pack together at either the N terminus of the α2 helix or in the loop region between the β4 strand and α3 helix. The structural location and modest nature of temperature-dependent shifts suggest that there may only be slight temperature-dependent changes in local secondary structure and not gross changes in tertiary structure.
The structural similarity of Mj0307 to the mesophilic bacterial Grx proteins might suggest that Mj0307 also functions as a Grx protein. There is ample evidence, however, to argue against such a suggestion. It has been reported that Mj0307 and Mt0807 do not show Grx behavior in functional assays (McFarlan et al. 1992; Lee et al. 2000), and this functional behavior can be explained by the structure-based sequence alignment (Fig. 8) that shows that the residues necessary for glutathione binding have not been conserved. Furthermore, it is likely that M. jannaschii, like M. thermoautotrophicum and other achaebacteria, exist in the absence of glutathione, which prevents any protein in these organisms from being classified as a Grx protein (Newton and Javor 1985; McFarlan et al. 1992).
The active site tetrapeptide sequence is highly conserved among the different classes of protein disulfide oxidoreductases, and the Mj0307 active site is identical to the mesophilic bacterial DsbA proteins. It is most unlikely, however, that Mj0307 has a similar biological function as DsbA because of the very poor overall sequence similarity to DsbA, the large differences in secondary and tertiary structure, and the inability of Mj0307 to promote refolding of scrambled RnaseA.
Lee and co-workers have reported a standard state redox potential of −277 mV for Mj0307, and this value makes Mj0307 the most reductive protein disulfide oxidoreductase known (Lee et al. 2000). This value is in stark contrast to DsbA, which has an identical active site and the most oxidative redox standard state measured for protein disulfide oxidoreductases (Wunderlich and Glockshuber 1993; Zapun et al. 1993). It has been well documented that the redox potential of protein disulfide oxidoreductases is modulated by more than just the two residues in the center of the CXXC motif. Residues outside the active site and helix dipoles can significantly affect the active site redox potential, and slight changes in the three-dimensional structure also can influence how these factors affect the redox potential (Chivers et al. 1997; Rossmann et al. 1997; Guddat et al. 1998; Krimm et al. 1998). The temperature-dependent changes in 1H-15N HSQC cross-peak positions suggest that there are slight structural changes in the vicinity of the active site that are induced by temperature changes. It is plausible that these changes affect the redox potential of Mj0307, and thus the redox potentials measured under mesophilic temperature and pressure conditions may not accurately reflect the redox potential under the conditions at which M. jannaschii normally lives. The impact of these experimental conditions have been documented in studies with a rubredoxin protein from Pyrococcus furiosus, which showed dramatic differences in redox potentials as a function of temperature and pressure (de Pelichy and Smith 1999).
Lee and co-workers also have reported that Mj0307 displays a weak interaction with Trr from E. coli (Lee et al. 2000). This weak interaction may be directed by the loop region connecting the β4 strand and the α3 helix, which shows a resemblance to the same region in E. coli Trx (Fig. 8). The equivalent loop regions in Mt0807 and Af2145 do not have this resemblance to E. coli Trx, and, relatedly, Mt0807 has been shown not to interact with E. coli Trr (McFarlan et al. 1992). The residues outside this loop region that are conserved between the Mj0307, Mt0807, and Af2145 proteins also do not show much similarity to those residues known to be important for protein–protein interactions in the known mesophilic bacterial and eukaryotic Trx proteins.
The in vivo function of Mj0307, and also Mt0807 and Af2145, may be most similar to the mesophilic bacterial NrdH proteins. Despite the modest sequence identity and predicted phylogenetic distance (Jordan et al. 1997), the NrdH proteins and the three archael proteins share a common size and (predicted) secondary structure. At least one member of both sets of proteins have displayed an ability to interact with E. coli Trr. E. coli NrdH serves as a good substrate, whereas the interaction is modest with Mj0307 and nonexistent for Mt0807. The poor interaction between the archael proteins and E. coli Trr, however, may be a consequence of the low sequence identity between the mesophilic and archael proteins, and the necessary residues required for a strong interaction have not been conserved. Trr homologs have been identified in M. jannaschii, M. thermoautotrophicum, and A. fulgidus with moderate sequence identities (32%–36%) to E. coli Trr. If Mj0307, Mt0807, and Af2145 can be reduced by these Trr homologs, then the similarity to the NrdH proteins becomes quite strong. We are engaged currently in efforts to probe the potential interaction between the M. jannaschii proteins.
If the archaebacterial proteins Mj0307, Mt0807, and Af2145 and the NrdH are shown subsequently to be functionally similar, then these proteins may constitute a novel subgroup within the family of protein disulfide oxidoreductases. This subgroup would be characterized by its Grx-like secondary and tertiary structure as well as its ability to interact with an appropriate Trr and not glutathione. This potential new subgroup of protein disulfide oxidoreductases may provide an important bridge in understanding conserved and divergent pathways in archaebacteria, prokaryotes, and eukaryotes that are dependent on protein disulfide oxidation and reduction.