research papers\(\def\hfill{\hskip 5em}\def\hfil{\hskip 3em}\def\eqno#1{\hfil {#1}}\)

Journal logoSTRUCTURAL
BIOLOGY
ISSN: 2059-7983

The folate-binding module of Thermus thermophilus cobalamin-dependent me­thionine synthase displays a distinct variation of the classical TIM barrel: a TIM barrel with a `twist'

aDepartment of Chemistry, University of Michigan, Ann Arbor, Michigan, USA, and bProgram in Biophysics, University of Michigan, Ann Arbor, Michigan, USA
*Correspondence e-mail: mkoutmos@umich.edu

Edited by G. J. Kleywegt, EMBL-EBI, Hinxton, England (Received 6 October 2017; accepted 21 December 2017)

Methyl transfer between methyltetrahydrofolate and corrinoid molecules is a key reaction in biology that is catalyzed by a number of enzymes in many prokaryotic and eukaryotic organisms. One classic example of such an enzyme is cobalamin-dependent methionine synthase (MS). MS is a large modular protein that utilizes an SN2-type mechanism to catalyze the chemically challenging methyl transfer from the tertiary amine (N5) of methyltetrahydrofolate to homocysteine in order to form methionine. Despite over half a century of study, many questions remain about how folate-dependent methyltransferases, and MS in particular, function. Here, the structure of the folate-binding (Fol) domain of MS from Thermus thermophilus is reported in the presence and absence of methyltetrahydrofolate. It is found that the methyltetrahydrofolate-binding environment is similar to those of previously described methyltransferases, highlighting the conserved role of this domain in binding, and perhaps activating, the methyltetrahydrofolate substrate. These structural studies further reveal a new distinct and uncharacterized topology in the C-terminal region of MS Fol domains. Furthermore, it is found that in contrast to the canonical TIM-barrel β8α8 fold found in all other folate-binding domains, MS Fol domains exhibit a unique β8α7 fold. It is posited that these structural differences are important for MS function.

1. Introduction

Cobalamin (Cbl)-dependent methionine synthase (MS) is an essential enzyme that is present in many organisms. The role of MS in cells is to catalyze the formation of tetrahydrofolate (H4folate) and methionine from methyltetrahydrofolate (CH3-H4folate) and homocysteine (Hcy) substrates (Fig. 1[link]a). Given its central role in metabolism, it is unsurprising that dysfunctions in MS are linked to human disease, including Cbl neuropathy (Metz, 1992[Metz, J. (1992). Annu. Rev. Nutr. 12, 59-79.]), megaloblastic anemia and homocysteinemia (Watkins et al., 2002[Watkins, D. et al. (2002). Am. J. Hum. Genet. 71, 143-153.]). In mammals, MS is a monomeric modular protein comprised of four modules arranged as follows (Fig. 1[link]b): an N-terminal homocysteine-binding (Hcy) domain, a folate-binding (Fol) domain, a Cbl-binding (Cob) domain and a C-terminal S-adenosyl­methionine-binding (AdoMet) domain. As their names imply, each module specifically binds either Hcy, CH3-H4folate, Cbl or AdoMet.

[Figure 1]
Figure 1
(a) The catalytic cycle of methionine synthase. AdoHcy, adenosylhomocysteine; AdoMet, adenosylmethionine; H4folate, tetrahydrofolate. (b) The domain structure of cobalamin-dependent methionine synthase.

The MS Cbl cofactor is representative of a class of bio­logical cofactors called corrinoids. Corrinoids are organo­metallic compounds characterized by a tetrapyrrole ring that harbors a Co-atom center. In MS, the Co atom is found in three different oxidation states: cobalt(I), cobalt(II) and cobalt(III) (Matthews et al., 2008[Matthews, R., Koutmos, M. & Datta, S. (2008). Curr. Opin. Struct. Biol. 18, 658-666.]). Cobalt(I) species are among nature's most potent nucleophiles (Schrauzer et al., 1968[Schrauzer, G. N., Deutsch, E. & Windgassen, R. J. (1968). J. Am. Chem. Soc. 90, 2441-2442.]; Banerjee, 1999[Banerjee, R. (1999). Editor. Chemistry and Biochemistry of B12. New York: Wiley.]) and are generally utilized to perform a variety of `difficult' chemistries including reductive dehalogenations (Payne et al., 2014[Payne, K. A., Quezada, C. P., Fisher, K., Dunstan, M. S., Collins, F. A., Sjuts, H., Levy, C., Hay, S., Rigby, S. E. & Leys, D. (2014). Nature (London), 517, 513-516.]; Bommer et al., 2014[Bommer, M., Kunze, C., Fesseler, J., Schubert, T., Diekert, G. & Dobbek, H. (2014). Science, 346, 455-458.]). Corrinoid-dependent enzymes using cobalt(I) promote a wide range of chemistries; however, the most common use of cobalt(I) corrinoids is as cofactors in methyl-transfer reactions (Matthews et al., 2008[Matthews, R., Koutmos, M. & Datta, S. (2008). Curr. Opin. Struct. Biol. 18, 658-666.]). For example, in addition to MS, cobalt(I) corrinoid species are used by the methyltransferase subunit (MeTr) of acyl-CoA synthase and CO dehydrogenase enzymes in the CO/CO2-fixation pathway of anaerobic microbes, and by the MeTr and CFeSP proteins to synthesize methane from acetate in Moorella thermoacetica (Ragsdale & Pierce, 2008[Ragsdale, S. W. & Pierce, E. (2008). Biochim. Biophys. Acta, 1784, 1873-1898.]; Kung et al., 2012[Kung, Y., Ando, N., Doukov, T. I., Blasiak, L. C., Bender, G., Seravalli, J., Ragsdale, S. W. & Drennan, C. L. (2012). Nature (London), 484, 265-269.]).

In contrast to MeTr and CSFeP, which are discrete proteins carrying the folate substrates and corrinoid cofactor, all MS enzymes carry these molecules in modules next to each other in the same polypeptide chain. While MS enzymes typically exist as four modules, there are exceptions to this module architecture (Fig. 1[link]b). In Thermotoga maritima, for example, MS is encoded as a two-component protein, with the Hcy, Fol and Cob modules encoded in one gene and the AdoMet module encoded in a separate open reading frame (Huang et al., 2007[Huang, S., Romanchuk, G., Pattridge, K., Lesley, S., Wilson, I., Matthews, R. & Ludwig, M. (2007). Protein Sci. 16, 1588-1595.]; Fig. 1[link]b). Despite the differences in subunit/module composition, most of the available structures of Fol modules from these corrinoid methyltransferases appear to exhibit the same classical β8α8 fold (Matthews et al., 2008[Matthews, R., Koutmos, M. & Datta, S. (2008). Curr. Opin. Struct. Biol. 18, 658-666.]) of the triosephosphate isomerase (TIM) barrel (Goldman et al., 2016[Goldman, A. D., Beatty, J. T. & Landweber, L. F. (2016). J. Mol. Evol. 82, 17-26.]). The available structures include folate-binding modules from several members of the MeTr (methyltransferase) subfamily [cd00740 in the NCBI Conserved Domain Database (CDD); Marchler-Bauer et al., 2015[Marchler-Bauer, A. et al. (2015). Nucleic Acids Res. 43, D222-D226.]] of the pterin-binding superfamily (cd00423 in the CCD; PDB entry 2e7f; Doukov et al., 2000[Doukov, T., Seravalli, J., Stezowski, J. J. & Ragsdale, S. W. (2000). Structure, 8, 817-830.]) and dihydropteroate synthase [DHPS; cd00739 subfamily; PDB entries 1ajz (Achari et al., 1997[Achari, A., Somers, D. O., Champness, J. N., Bryant, P. K., Rosemond, J. & Stammers, D. K. (1997). Nature Struct. Biol. 4, 490-497.]) and 1ad4 (Hampele et al., 1997[Hampele, I. C., D'Arcy, A., Dale, G. E., Kostrewa, D., Nielsen, J., Oefner, C., Page, M. G. P., Schönfeld, H.-J., Stüber, D. & Then, R. L. (1997). J. Mol. Biol. 268, 21-30.])].

Similar to Homo sapiens MS, the thermophilic bacterium Thermus thermophilus possesses a four-module Cbl-dependent MS. We selected T. thermophilus MS as a model for studying the structural and functional properties of MS because the enhanced stability of thermophiles can facilitate the crystallization of challenging pro­teins. Here, we describe the structure of the Fol module of T. thermophilus MS in the presence and absence of the CH3-H4folate substrate. We find that the CH3-H4folate-binding environment is similar to those of previously described folate-dependent methyltransferases, highlighting the conserved role of this module in binding and perhaps activating the CH3-H4folate substrate. Additionally, our structures reveal a significant difference in the C-terminal region of most MS Fol modules compared with other folate-binding or pterin-binding proteins. Our studies demonstrate that the structures of most MS Fol modules display a `twist' on the classical β8α8 fold, and are instead best described as having a β8α7 barrel. Therefore, two distinct variations of the TIM-barrel fold are found in pterin-binding proteins.

2. Materials and methods

2.1. Macromolecule production

A gene encoding full-length T. thermophilus MS was PCR-amplified from a T. thermophilus expression vector (Yokoyama et al., 2000[Yokoyama, S., Hirota, H., Kigawa, T., Yabuki, T., Shirouzu, M., Terada, T., Ito, Y., Matsuo, Y., Kuroda, Y., Nishimura, Y., Kyogoku, Y., Miki, K., Masui, R. & Kuramitsu, S. (2000). Nature Struct. Biol. 7, 943-945.]) purchased from RIKEN. The amplified fragment was subsequently cloned into a pMCSG7 plasmid using ligation-independent cloning (Stols et al., 2002[Stols, L., Gu, M., Dieckman, L., Raffen, R., Collart, F. R. & Donnelly, M. I. (2002). Protein Expr. Purif. 25, 8-15.]) to generate the pMCSG7(tMSwt) plasmid. An expression vector for a His761Lys full-length MS mutant (His761 is found in the Cob module) was generated by site-directed mutagenesis using a QuikChange Mutagenesis Kit (Agilent), and is named pMCSG7(tMSH761K). The coding region of the Fol module (Gln364–His648) was amplified from pMCSG7(tMSwt) and inserted into an empty pMCSG7 plasmid as described above [pMCSG7(tMSFol)]. All of our constructs contain N-terminal hexahistidine-fusion tags (His6 tags) followed by a cleavable Tobacco etch virus (TEV) protease site.

Proteins were expressed in Escherichia coli BL21 Star(DE3) cells (Novagen). For expression of the full-length MS protein, the cells were grown to an optical density of 0.6–0.8 in LB medium at 37°C and were then induced by the addition of isopropyl β-D-1-thiogalactopyranoside (final concentration of 0.1 mM; Promega) followed by overnight growth at 25°C. For expression of the single Fol module, E. coli BL21 Star(DE3)/pMCSG7(tMSFol) cells were inoculated in auto-induction medium (Studier, 2005[Studier, F. W. (2005). Protein Expr. Purif. 41, 207-234.]) and cultured at 30°C for 22–24 h. The cells were collected by centrifugation and stored at −80°C.

Purification of full-length T. thermophilus MS without Cbl (apo T. thermophilus MSH761K) was performed as described below. E. coli BL21 Star(DE3)/pMCSG7(tMSH761K) cells were resuspended in buffer [50 mM potassium phosphate buffer (KPB) pH 7.2, 1 mM phenylmethylsulfonyl fluoride, 0.5 µg ml−1 lysozyme], lysed by sonication and pelleted by centrifugation (10 000g for 20 min at 4°C). The resulting supernatant was subjected to a heat step by incubation at 70°C for 20 min. The heat-treated suspension was subsequently centrifuged (10 000g for 20 min) and filtered using a Millex Filter Unit (0.45 µm; Millipore, Darmstadt, Germany). The filtrate was loaded onto a 5 ml nickel-affinity column (HiTrap Chelating; GE Healthcare Life Sciences) at room temperature (in buffer A: 50 mM KPB pH 7.2, 50 mM imidazole). The column was washed with at least five column volumes of buffer B (buffer A + 0.3 M sodium chloride). The bound protein was eluted with ∼4 column volumes of 50 mM KPB pH 7.2, 125 mM imidazole. Eluates containing His6-tagged protein were pooled together and were dialyzed overnight at 4°C against 50 mM KPB pH 7.2. The resulting dialyzed protein was concentrated and stored at 4°C.

To express and purify the Fol module of T. thermophilus MS, pMCSG7(tMSFol) was transformed into E. coli BL21 Star(DE3) cells. Cell lysis followed by heat treatment was performed as described above for the full-length MS preparation. After adding imidazole to a final concentration of 20 mM, the crude protein solution was loaded onto a nickel-affinity column. The column was washed with buffer B and the bound protein was eluted with buffer C (buffer B + 200 mM imidazole). The eluate was dia­lyzed overnight against 50 mM Tris–HCl pH 7.4 in the presence of His6-tagged TEV protease at 4°C overnight, after which the dialysate was incubated at 70°C for 15 min to remove TEV protease. Purified MS Fol module was concentrated and stored at 4°C.

2.2. Crystallization

For crystallization purposes, full-length apo T. thermophilus MSH761K was run through a size-exclusion column (S200; GE Healthcare) equilibrated with 25 mM Tris–HCl pH 7.5, 0.1 M NaCl. The full-length T. thermophilus MS and Fol module samples were concentrated to ∼15 and ∼25 mg ml−1, respectively. All protein samples were filtered through a Spin-X centrifuge filter (Corning Costar) prior to crystallization screening.

Crystals of full-length T. thermophilus MS were obtained by the sitting-drop vapor-diffusion method by mixing 0.3 µl protein solution (14 mg ml−1) with 0.3 µl reservoir solution. Full-length apo T. thermophilus MS was mixed with 0.8 M succinic acid pH 7.0 at 20°C and afforded crystals of CH3-H4folate-free MS Fol module (1). Crystals of 1 were cryoprotected for 5 min before being flash-cooled in liquid nitrogen by transfer to a solution consisting of 20% glycerol, 0.8 M succinic acid pH 7.0.

For the crystallization of single Fol module MS constructs, the hanging-drop vapor-diffusion method was used. The Fol module of T. thermophilus MS (15 mg ml−1) was mixed with reservoir solution (0.1 M sodium citrate pH 4.5, 18% PEG 3350, 0.2 M trisodium citrate) in a 1:1 ratio at 4°C. Crystals of the Fol module with native CH3-H4folate present (2) were cryoprotected for 5 min before being flash-cooled in liquid nitrogen by transfer to a solution consisting of 0.1 M sodium citrate pH 4.5, 18% PEG 3350, 0.2 M trisodium citrate, 20% glycerol.

Crystals of the Fol module grown in the presence of externally provided CH3-H4folate (3) were transferred to a solution consisting of 0.1 M sodium citrate pH 4.5, 22% PEG 3350, 0.2 M trisodium citrate, 20% glycerol, ∼1.5 mM (6S)-5-CH3-H4folate (Eprova), in which they were stored overnight before they being flash-cooled in liquid nitrogen.

2.3. Data collection and processing

Diffraction data for all crystals were collected at 100 K on the GM/CAT-CAT 23-ID-B beamline at Advanced Photon Source (APS), Argonne National Laboratory using a MAR300 detector for 1 and using a DECTRIS EIGER X 16M detector for 2 and 3. All data sets were processed with XDS (Kabsch, 2010[Kabsch, W. (2010). Acta Cryst. D66, 125-132.]): to 2.1 Å resolution for 1, 2.4 Å resolution for 2 and 2.1 Å resolution for 3. Data-collection and processing statistics are summarized in Table 1[link]. The crystals of 1 belonged to space group P43212 (unit-cell parameters a = b = 55.2, c = 197.1 Å) with one molecule in the asymmetric unit (Matthews coefficient VM = 2.3 Å3 Da−1, 45.9% solvent content) and the crystals of 3 belonged to space group P6122 (unit-cell parameters a = b = 188.5, c = 247.1 Å) with six molecules in the asymmetric unit (Matthews coefficient VM = 3.2 Å3 Da−1, 61.5% solvent content), whereas the crystals of 2 belonged to space group P6122 (unit-cell parameters a = b = 111.5, c = 242.7 Å) with two molecules in the asymmetric unit (Matthews coefficient VM = 3.5 Å3 Da−1, 64.5% solvent content).

Table 1
X-ray crystallographic data-collection and refinement statistics

  MS Fol (1) MS Fol + CH3-H4folate (2) MS Fol + CH3-H4folate (3)
Data collection
 Beamline GM/CA 23-ID-B, APS GM/CA 23-ID-B, APS GM/CA 23-ID-B, APS
 Wavelength (Å) 1.0332 1.0332 1.0332
 Temperature (K) 100 100 100
 Resolution (Å) 50.0–2.10 (2.18–2.10) 50.0–2.10 (2.16–2.10) 50.0–2.40 (2.54–2.40)
 Crystal-to-detector distance (mm) 300 300 350
 Rotation range per image (°) 0.5 0.2 0.2
 Total rotation range (°) 130 110 40
 Exposure time per image (s) 1 0.2 0.2
 Space group P43212 P6122 P6122
 Unit-cell parameters (Å, °) a = 55.2, b = 55.2, c = 197.1, α = β = γ = 90 a = 111.5, b = 111.5, c = 242.7, α = β = 90, γ = 120 a = 188.5, b = 188.5, c = 247.1, α = β = 90, γ = 120
 Observed reflections 123053 561666 453517
 Unique reflections 18533 52825 187437
Rmeas (%) 5.9 (54.0) 9.5 (79.1) 10.3 (73.1)
Rmerge (%) 5.1 (41.6) 8.7 (71.6) 8.1 (56.3)
Rp.i.m. (%) 2.9 (33.4) 3.8 (32.9) 6.1 (45.6)
 〈I/σ(I)〉 17.0 (2.1) 13.1 (1.9) 11.3 (2.03)
 CC1/2 0.999 (0.783) 0.999 (0.909) 0.996 (0.692)
 Multiplicity 5.8 (3.7) 10.6 (10.2) 4.5 (4.4)
 Completeness (%) 98.0 (93.6) 99.9 (99.9) 99.4 (98.4)
 Overall B (Wilson plot) (Å2) 32.6 42.8 39.6
Refinement
 Resolution range 53.14–2.10 96.55–2.10 163.2–2.40
 No. of reflections
  Work set 17499 50064 95477
  Test set 955 2667 5056
Rwork/Rfree (%) 20.2/25.9 19.4/22.2 19.0/23.0
 No. of non-H atoms
  Protein 2217 4434 13273
  Water 67 128 322
  C2F 0 66 198
B factors (Å2)
  Protein 50.4 61.7 50.5
  Water 52.0 57.1 43.1
  C2F n/a 63.5 35.3
 R.m.s. deviations      
  Bond lengths (Å) 0.008 0.009 0.009
  Bond angles (°) 1.300 1.420 1.363
 Ramachandran plot (%)      
  Favored 96.8 97.4 97.0
  Allowed 2.5 2.4 2.9
  Outliers 0.7 0.2 0.1
MolProbity score 1.10 [100th percentile] 1.00 [100th percentile] 1.18 [100th percentile]
 PDB code 5von 5vop 5voo
†6S-5-Methyltetrahydrofolate.

The crystals of 1 were obtained using full-length T. thermophilus MS; however, the unit cell was far too small to correspond to the full-length polypeptide based on the Matthews coefficient. This was confirmed by SDS–PAGE analysis of the crystals, which showed a molecular weight of ∼31 kDa which corresponds to a single module. Therefore, the crystals corresponding to a single module were formed after proteolytic cleavage, possibly by trace contamination with protease during crystallization. Since we had not ascertained the nature of this single module, we used all four MS modules as search models for a molecular-replacement solution. Initial phases were obtained with Phaser (McCoy et al., 2007[McCoy, A. J., Grosse-Kunstleve, R. W., Adams, P. D., Winn, M. D., Storoni, L. C. & Read, R. J. (2007). J. Appl. Cryst. 40, 658-674.]) only when the Fol module of T. maritima MS (PDB entry 3bof; Koutmos et al., 2008[Koutmos, M., Pejchal, R., Bomer, T. M., Matthews, R. G., Smith, J. L. & Ludwig, M. L. (2008). Proc. Natl Acad. Sci. USA, 105, 3286-3291.]) was used as a search model. Initial phases for 2 and 3 were obtained using the resulting structure of 1 as a search model.

2.4. Structure solution and refinement

All three structures were initially refined with PHENIX (Adams et al., 2002[Adams, P. D., Grosse-Kunstleve, R. W., Hung, L.-W., Ioerger, T. R., McCoy, A. J., Moriarty, N. W., Read, R. J., Sacchettini, J. C., Sauter, N. K. & Terwilliger, T. C. (2002). Acta Cryst. D58, 1948-1954.]). Refinement included simulated annealing in torsional and Cartesian space, coordination minimization and restrained individual B-factor adjustment with maximum-likelihood targets. REFMAC5 (Murshudov et al., 2011[Murshudov, G. N., Skubák, P., Lebedev, A. A., Pannu, N. S., Steiner, R. A., Nicholls, R. A., Winn, M. D., Long, F. & Vagin, A. A. (2011). Acta Cryst. D67, 355-367.]) was then employed for restrained refinement using isotropic individual B factors with maximum-likelihood targets. In all three refinements the Babinet model for bulk-solvent scaling was utilized. Refinement was followed by model building and modification with Coot (Emsley et al., 2010[Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. (2010). Acta Cryst. D66, 486-501.]). Several iterative rounds of refinement followed by model building and modification were performed. In the early rounds of refinement, no ligands were fitted for structures 2 and 3.

After the protein model was complete and water molecules had been assigned, clear electron density was fitted in the active sites of 2 and 3 and subsequently refined as CH3-H4folate (C2F) ligands. For 2, since no external folate was provided the electron density corresponds to native folate substrate. We considered both the CH3-H4folate (C2F) substrate and the H4folate (THG) product as potential ligands that could occupy the electron density in the active site. When H4folate was modeled there was strong residual electron density near the N5 position, clearly corresponding to the expected position of the methyl group in CH3-H4folate. CH3-H4folate fits well in the composite OMIT electron density, as shown in Supplementary Fig. S1. Therefore, we concluded that the native ligand corresponds to CH3-H4folate. It must be noted that there is residual but weak electron density extending from the glutamate tail, indicating that the native ligand has a polyglutamate moiety. Owing to the poor electron density in this region we did not attempt to model such a polyglutamate tail, and only a single glutamate is present in our model.

The final model was used to create an unbiased refined composite 2FoFc OMIT map with PHENIX (Supplementary Fig. S1). Crystallographic information as well as refinement statistics are provided in Table 1[link]. The geometric quality of the model and its agreement with the structure factors were assessed with MolProbity (Chen et al., 2010[Chen, V. B., Arendall, W. B., Headd, J. J., Keedy, D. A., Immormino, R. M., Kapral, G. J., Murray, L. W., Richardson, J. S. & Richardson, D. C. (2010). Acta Cryst. D66, 12-21.]). For 1, MolProbity reported a clashscore and a MolProbity score of 1.11 (100th percentile) and 1.10 (100th percentile) for 1, while 96.8% of the residues were in the favored Ramachandran plot regions, with 0.7% residues in outlier regions. For 2, the reported values were 1.00 (100th percentile) and 1.97 (100th percentile), respectively, while 97.9% of the residues were in the favored Ramachandran plot regions, with 0.4% residues in outlier regions. For 3, the reported values were 1.18 (100th percentile) and 2.24 (100th percentile), respectively, while 97.0% of the residues were in the favored Ramachandran plot regions, with 0.1% of residues in outlier regions. Figures displaying crystal structures were generated with PyMOL (v.1.8; Schrödinger; http://www.pymol.org).

3. Results

3.1. Crystal structure of the Fol module of T. thermophilus MS

The structure of the T. thermophilus MS Fol module is shown in Fig. 2[link](a). The overall structure is in good agreement with previously published structures of folate-binding modules from other MeTr family members. We aligned the sequences of selected MeTr members, including the Fol module of T. thermophilus MS, as shown in Fig. 2[link](b), to highlight invariant and highly conserved residues among various folate-binding modules. To assess the potential differences between MS Fol modules from different species, we aligned our structure of the MS Fol module from T. thermophilus with that of the previously described MS module from T. maritima (PDB entry 1q8j; Evans et al., 2004[Evans, J., Huddler, D., Hilgers, M., Romanchuk, G., Matthews, R. & Ludwig, M. (2004). Proc. Natl Acad. Sci. USA, 101, 3729-3736.]; Fig. 3[link]a). The Fol modules from T. thermophilus and T. maritima MS share 25% identity (based on 292 amino acids). Moreover, their structural superposition was indicative of the structural similarity between these two modules, with an r.m.s.d. for Cα atoms of 1.98 Å (222 residues).

[Figure 2]
Figure 2
Structure of the T. thermophilus MS folate-binding module. (a) The crystal structure of the T. thermophilus MS Fol module is shown in cartoon mode. Helices and sheets are colored cyan and red, respectively. (b) Amino-acid alignment of the folate-binding modules of MS and MeTr. Thermus, Thermus thermophilus MS (GenBank accession No. NC_006461); E.coli, Escherichia coli MS (J04975); human, H. sapiens MS (U73338); B.thetaiotaomicron, Bacteroides thetaiotaomicron VPI-5482 MS (NC_004663); Thermotoga, Thermotoga maritima MS (NC_000853); D.hafniense, Desulfitobacterium hafniense DCB-2 methyltetrahydrofolate-corrinoid/iron–sulfur protein methyltransferase (Dhaf_0722; ACL18786); M.thermoacetica, Moorella thermoacetica 5-methyltetrahydrofolate-corrinoid/iron–sulfur protein co-methyltransferase (AcsE; Q46389). Protein portions forming helices and sheets are colored cyan and magenta, respectively. Amino-acid chains that comprise structurally undetermined regions are shown in gray. Residues conserved only in MS are highlighted in yellow and those conserved across all methyltransferases are highlighted in green. The last helix of the Fol module in T. thermotoga MS (α8F) is underlined in blue and the C-terminal region including helices α8aF, α8bF, α8cF and α8dF of T. thermophilus MS is underlined in red. The red vertical lines indicate the C-terminal ends of the expressed and crystallized protein constructs.
[Figure 3]
Figure 3
Demonstration of the two topology types observed in the C-terminal region of folate-binding modules. (a) Superposition of the Fol modules from T. thermophilus and T. thermotoga MS. The Fol module of T. thermophilus MS is illustrated in cartoon mode and shown in green, except for the C-terminal four helices, which are shown in red. T. thermotoga MS (PDB entry 1q8j) is shown in cyan, except for the last helix (α8F), which is colored blue. (b) Structural alignment of the Fol modules of MS. The Fol modules of T. thermophilus MS [colored green and red as in (a)], H. sapiens (pale blue; PDB entry 4ccz) and B. thetaiotaomicron (yellow; PDB entry 3k13) are drawn in cartoon mode. (c) Structural alignment of pterin-binding superfamily members. The Fol module of T. thermotoga MS is shown in cyan and blue as in (a). The folate-binding domains of MeTr from M. thermoacetica (PDB entry 2e7f) and MT2DH from D. hafniense (PDB entry 4o1e) are displayed in pink and gray, respectively, and the last nonconserved helices are transparent. A monomer of DHPS from S. aureus (PDB entry 1ad4) is colored orange.

In addition to the anticipated similarities, we also observed a region with significant differences. While the eight β-strands in the center of the TIM barrel and the first seven α-helices that flank the central barrel are comparable between the T. maritima and T. thermophilus MS Fol modules, the region corresponding to the eighth α-helix at the C-terminus of the TIM barrels is different in T. thermophilus MS. In typical TIM barrels, such as the T. maritima MS Fol module (Figs. 2[link]b and 3[link]a), the eighth β-strand is followed by a short four-residue loop (538–541) and the eighth α-helix (14 residues in length). This creates a characteristic antiparallel strand–turn–helix motif that places the eighth helix almost parallel to the barrel. In this case, the eighth α-helix forms an α/α interface and interacts with both the surrounding α1 and α7 helices (Fig. 3[link]a). In contrast to this, the Fol module of T. thermophilus MS does not have the antiparallel helix–turn–helix motif present at the C-terminus. Instead, we observe a unique arrangement that is distinct from other folate-binding modules.

In T. thermophilus MS Fol the eighth β-strand is followed by an 11-residue loop containing two 310-helices (α8aF and α8bF) that extend away from the barrel and bend to wrap around the solvent-exposed side of the α7 helix. This loop is followed by two antiparallel helices, α8cF and α8dF, that stack against the outer sides of barrel helices α6 and α7 (Figs. 2[link]a and 3[link]a). The resulting arrangement resembles a helical bundle that displays an extensive interface with multiple electrostatic and hydrophobic interactions and a total buried area of 1000 Å2 as calculated by the PISA server (Krissinel & Henrick, 2007[Krissinel, E. & Henrick, K. (2007). J. Mol. Biol. 372, 774-797.]). Consequently, whereas the T. maritima MS Fol module displays a typical and canonical β8α8 barrel, the Fol module of T. thermophilus MS is best described as a β8α7 barrel. Thus, the T. thermophilus MS Fol module represents the first described example of a noncanonical subclass of TIM barrels.

A DALI server search (Holm & Rosenström, 2010[Holm, L. & Rosenström, P. (2010). Nucleic Acids Res. 38, W545-W549.]) using the Fol module of T. thermophilus MS as a search model identified the Fol module from H. sapiens MS (PDB entry 4ccz; Structural Genomics Consortium, unpublished work) as the closest structural match (Z-score = 37.1, r.m.s.d. = 3.6 Å). The second closest structural homolog was that of the MS Fol module from Bacteroides thetaiotaomicron (PDB entry 3k13; Midwest Center for Structural Genomics, unpublished work; Z-score = 27.2, r.m.s.d. = 2.0 Å). Both of these structures have been deposited in the PDB, but are not discussed in published manuscripts. A structural superposition of the Fol modules of T. thermophilus, H. sapiens and B. thetaiotaomicron MS is graphically illustrated in Fig. 3[link](b). It is evident that these three Fol modules share the same topology, including an identical β8α7 TIM-barrel. These differ from the Fol module of T. maritima MS, which exhibits the more common β8α8 fold.

3.2. Comparison of the Fol module of T. thermophilus MS with folate-binding and pterin-binding modules from various species

In addition to the findings discussed above, a DALI search for MS Fol modules identified other distinct structural homologs. Most of these also belong to the pterin-binding superfamily (MeTr subfamily; cd00423 in the CDD). For the purposes of comparison, the following three structurally related proteins were selected: the MeTr subunit of acyl-CoA synthase/CO dehydrogenase from Moorella thermoacetica (PDB entry 2e7f; Doukov et al., 2007[Doukov, T., Hemmi, H., Drennan, C. & Ragsdale, S. (2007). J. Biol. Chem. 282, 6609-6618.]), the methyltransferase component, the MT2DH product of Dhaf0722, from Desulfitobacterium hafniense DCB-2 (PDB entry 4o1e; Sjuts et al., 2015[Sjuts, H., Dunstan, M. S., Fisher, K. & Leys, D. (2015). Acta Cryst. D71, 1900-1908.]) and DHPS from Staphylococcus aureus (PDB entry 1ad4; Hampele et al., 1997[Hampele, I. C., D'Arcy, A., Dale, G. E., Kostrewa, D., Nielsen, J., Oefner, C., Page, M. G. P., Schönfeld, H.-J., Stüber, D. & Then, R. L. (1997). J. Mol. Biol. 268, 21-30.]). As shown in Fig. 3[link](c), all three of these protein structures share the hallmark β8α8-barrel motif found ubiquitously throughout this family. The C-terminal topology of these three structures is identical to that of the Fol module of T. maritima MS, albeit distinct from that of the Fol module of all other MS enzymes.

3.3. Comparison of folate-binding modes

To determine how the T. thermophilus MS Fol module binds its folate substrate, the isolated Fol module was expressed, purified and crystallized in the presence and absence of CH3-H4folate. We observed electron density corresponding to CH3-H4folate even in crystals that had not been soaked or co-crystallized with CH3-H4folate (Fig. 4[link]), suggesting that the Fol module co-purifies in complex with its substrate. Additionally, we solved the structure of a T. thermophilus MS Fol module that was proteolytically derived from the full-length MS. In contrast to the isolated module, the crystallized proteolytic fragment did not show any electron density associated with CH3-H4folate. This allowed us to assess the potential impact on the Fol module structure when CH3-H4folate binds. Comparison of the substrate-free structure with the substrate-bound structure (Supplementary Fig. S2) showed no significant differences, with r.m.s.d.s of 1.293 and 0.637 Å based on all atoms (2217) and all backbone atoms (1128), respectively. The folate-bound structure is slightly more open relative to the folate-free structure. When the CH3-H4folate ligand is present it presses against helix α1aF and the loop connecting it to β1F. This causes helix α1aF, the loop connecting it to β1F and helix α1bF to move away from the ligand. This is the only ligand-induced change that results in the observed slight opening of the barrel. There is little change in the residues that line the substrate-binding pocket (Supplementary Fig. S2b); the majority of these residues are in the same position in both structures, reflecting a lock-and-key type of fit between the ligand and its binding module.

[Figure 4]
Figure 4
Stereoview of the CH3-H4folate-binding site. The Fol module is rendered in cartoon mode, while the active-site side chains and the folate substrate are shown as sticks. The weighted and refined 2mFoDFc map (blue) shown for the folate substrate is contoured at 1.5σ.

In T. thermophilus MS the pterin ring of folate is recognized through residues that are conserved among all Fol modules. The amino acids comprising the folate-binding site are displayed as sticks in Fig. 5[link] (Asp411, Asp443, Asn464, Asp531, Asn573 and Thr534); these are all conserved except for Thr534 (Fig. 2[link]b). The binding-pocket residues form an extensive hydrogen-bonding network that is nearly identical in all Fol modules. The only differences that are observed in the folate-binding pocket between the various Fol module structures are found near the N3 and O4 folate atoms. For example, in the structure of T. thermophilus MS a single water molecule (W1) is present near folate atoms N3 and O4, while two water molecules (W1 and W2) are observed in the other Fol modules (Fig. 5[link]). W1 is present in all available folate-bound structures and mediates a contact between an invariant Asp carboxylate (Asp531 in T. thermophilus MS) and the carbonyl of the pterin ring (O4). In the three structures containing a second water molecule (W2; Figs. 5[link]b, 5[link]c and 5[link]d), W2 forms interactions with the carbonyl O atom of the peptide backbone (Val or Ile), the amino group of an invariant Asn and W1. In contrast, Thr534 of T. thermophilus MS interacts directly with W1 (Fig. 5[link]a).

[Figure 5]
Figure 5
Schematic illustration of the interaction between amino-acid residues and CH3-H4folate in T. thermophilus MS (a), T. thermotoga MS (b), M. thermoacetica MeTr (c) and D. hafniense MT2DH (d).

Additionally, there are differences in a conserved Ser residue near the p-aminobenzoate moiety (Ser572 in T. thermophilus MS). The equivalent Ser198 in D. hafniense MT2DH (Fig. 5[link]d) is proposed to play an important role in substrate activation based on the recent structures of MT2DH–CH3-H4folate complexes (Sjuts et al., 2015[Sjuts, H., Dunstan, M. S., Fisher, K. & Leys, D. (2015). Acta Cryst. D71, 1900-1908.]). In all other Fol modules this Ser residue forms a hydrogen bond to an invariant Glu (Glu373 in T. thermophilus MS; Figs. 4[link] and 6[link]), but in D. hafniense MT2DH it drastically relocates and forms a hydrogen bond to a water molecule (W3) that interacts with the pterin N5 and O4 atoms. This water is not observed in other protein–CH3-H4folate binary structures (W3 in Fig. 5[link]d). The unique Ser arrangement in D. hafniense MT2DH may be important for the distinct functionality of MT2DH, which supports the reverse reaction compared with other folate-dependent methyltransferases: methylation of the folate substrate (Sjuts et al., 2015[Sjuts, H., Dunstan, M. S., Fisher, K. & Leys, D. (2015). Acta Cryst. D71, 1900-1908.]). In the other available structures of Fol modules with a substrate bound, including those of MS enzymes, instead of W3 an invariant Asn [Asn199 in MeTr (Fig. 5[link]c), and Asn573 and Asn508 in T. thermophilus MS and H. sapiens MS, respectively (Figs. 5[link]a and 5[link]b)] directly interacts with the O4 and N5 atoms of the folate substrate. It has been proposed that this invariant Asn is essential for catalytic activity, but not for substrate binding, in all folate-dependent methytransferases (Doukov et al., 2007[Doukov, T., Hemmi, H., Drennan, C. & Ragsdale, S. (2007). J. Biol. Chem. 282, 6609-6618.]). Lastly, three invariant Asp residues (Asp411, Asp443 and Asp531) are expected to be necessary for substrate activation based on studies of E. coli MS (Smith & Matthews, 2000[Smith, A. E. & Matthews, R. G. (2000). Biochemistry, 39, 13880-13890.]).

[Figure 6]
Figure 6
Comparison of the C-terminal regions of the T. thermophilus Fol module monomer and the T. thermotoga Fol module dimer. The Fol module of T. thermophilus MS (green and red) is superimposed on one of the T. thermotoga MS Fol modules (cyan and blue) as shown in Fig. 3[link](a). The adjacent Fol module of the T. thermotoga MS dimer is partially drawn and shown in yellow.

4. Discussion

Our studies of the Fol module of T. thermophilus MS revealed a unique variation of the prototypical β8α8 TIM-barrel fold that is present in all other pterin-binding proteins (Marchler-Bauer et al., 2015[Marchler-Bauer, A. et al. (2015). Nucleic Acids Res. 43, D222-D226.]). The pterin-binding superfamily (cd00423) is divided into two subfamilies: the DHPS (cd00739) and MeTr (cd00740) subfamilies. Based on our results, we propose that the MeTr subfamily should be further divided into two subgroups: one exhibiting the classical β8α8 fold and a second displaying the distinct β8α7 fold described here and only found thus far in the Fol modules of Cbl-dependent methionine synthases.

Prior to this work, the only previously published structure of a Fol module was from T. maritima. Here, we compare the Fol domain of MS from T. maritima with our structure from T. thermophilus, and also with structures deposited in the PDB (but not published) from H. sapiens and B. thetaiotaomicron. We find that with the exception of T. maritima MS, all MS Fol modules appear to be members of the new MeTr subfamily containing a distinct β8α7 TIM barrel. We first speculated that this may be because in contrast to the majority of species, which encode their MS proteins in a single gene, T. maritima MS encodes the first three modules (Hcy/Fol/Cob) in a single gene and the last activation module (AdoMet) separately (Huang et al., 2007[Huang, S., Romanchuk, G., Pattridge, K., Lesley, S., Wilson, I., Matthews, R. & Ludwig, M. (2007). Protein Sci. 16, 1588-1595.]). However, this seems unlikely because B. thetaiotaomicron MS, like T. maritima MS, is also encoded in a two-component system (NCBI locus tags BT_0180 for the Hcy, Fol and Cob modules and BT_0249 for the putative AdoMet module) yet its Fol module still exhibits a β8α7-barrel fold. Another explanation for this difference may arise from the observation that among the crystallized MS Fol modules only T. maritima MS Fol is involved in homodimer formation both in crystallo and in solution. Close examination of the T. maritima MS dimerization interface (Fig. 6[link]) reveals that it is formed mainly through interactions of helices α7 and α8 with the same helices (α7′ and α8′) from an adjacent molecule. The total buried surface area in the interface is ∼1044 Å2, which is nearly identical to that found in the equivalent C-terminal helical bundle present in the other MS Fol modules. Interestingly, superposition of the Fol modules of T. thermophilus and T. maritima MS revealed that in T. maritima the MS Fol homodimer interface helix α7′ occupies the exact same position as helix α8cF in the T. thermophilus MS Fol module (Fig. 6[link]). Furthermore, T. maritima MS was crystallized with a truncated C-terminus lacking residues in the linker between the Fol and Cob modules (Fig. 2[link]b). The truncated T. maritima construct lacked amino acids corresponding to the entire helix α8dF, in contrast to the three MS Fol modules with a β8α7 fold. It is therefore possible that the classical β8α8 fold found in the T. maritima MS Fol module may be an artifact of the crystallization construct owing to the absence of the linker region residues that are present in all of the other determined MS folate structures. This would allow the truncated T. maritima MS Fol module to fold differently, inhibiting the ability of this C-terminal region to fold into the noncanonical TIM-barrel (β8α7) structure seen in the structures of MSs from all other sources. However, an alternative justification for this variation may lie in the differences found both in the C-terminal region of the Fol module and in the linker region connecting the Fol module to the subsequent Cob module. In T. maritima MS there are 29 amino acids between the eighth strand of the folate domain (β8F) and the first helix of the N-terminal four-helical bundle of the Cob module. In the other MSs the linker regions are substantially longer, ranging between 59 and 63 residues. Given the differences in the linker region in T. maritima MS, we believe that the latter is most likely to represent an outlier and has a distinct method of connecting the Fol module to the Cob module. Moreover, close examination of the C-terminal region, and especially helices α7 and β8 of the Fol modules, reveal significant variations in the amino-acid sequences that might explain the different conformations. Specifically, we find four major differences between Fol modules with a classical fold (e.g. T. maritima) and a `twist' fold (e.g. T. thermophilus) (Fig. 7[link] and Supplementary Fig. S3).

  • (i) In the Fol modules with a classical fold, three outward-facing hydrophobic residues (in positions −12, −11 and −8; Fig. 7[link]a) in helix α7 which form intramolecular contacts with the α7′ and α8′ helices of an adjacent monomer are all substituted by charged residues in the Fol modules with a β8α7 fold (Fig. 7[link]b).

  • (ii) Similarly, in the β8α8 modules three inward-facing hydrophobic residues in helix α8 that form part of the interface with the adjacent α7 and α1 helices (in positions 7, 10 and 13; Fig. 7[link]a) and two outward-facing residues (in positions 16 and 17; Fig. 7[link]a) which form intramolecular contacts with the α7′ helix of an adjacent monomer are mainly replaced by charged residues in β8α7 modules (Fig. 7[link]b).

  • (iii) In β8α8 modules, there is a cluster of mainly charged residues downstream of the α8 helix (in positions 19, 22 and 23; Fig. 7[link]a) that are all substituted with hydrophobic residues in β8α7 modules. The latter residues that reside on the α8cF helix are part of an extensive hydrophobic interface formed between helices α8cF, α8dF and α7 (Fig. 7[link]b).

  • (iv) In Fol modules with a β8α7 fold helices α6 and α7 are longer than in modules with a β8α8 fold (Fig. 2[link]b), and these extensions form part of the extended interface with helices α8cF and α8dF (Supplementary Fig. S3).

[Figure 7]
Figure 7
The top panels show close-ups of the two distinct interfaces formed in Fol modules with a β8α8 and a β8α7 fold in cartoon mode. Important residues that form part of the interfaces are shown as sticks. The bottom panels display the graphical sequence variance between the C-terminal regions in the classical (β8α8) and `twist' (β8α7) folds, as generated with WebLogo (Crooks et al., 2004[Crooks, G. E., Hon, G., Chandonia, J.-M. & Brenner, S. E. (2004). Genome Res. 14, 1188-1190.]). More than 300 sequences (with representative sequences exhibiting between 90 and 15% sequence identity) for each of the different folds were used in WebLogo. The numbering at the bottom corresponds to the T. maritima (left) and T. thermophilus (right) sequences, whereas the numbering at the top is used as a reference and corresponds to the positions upstream (negative) or downstream (positive) of the last residue in the β8 strand. In (a) helix α8 (green) is involved in intermolecular interactions with helices α7 and α1 (green) and mainly hydrophobic intramolecular interactions with helices α7′ and α8′ (red) from an adjacent monomer. In (b) helix α8cF (teal) forms extensive intramolecular hydrophobic interactions with helices α8dF (teal) and α7 (green). The solvent-exposed polar residues Asp592, Glu593 and Lys596 of helix α7 [positions −12, −11 and −8; bottom panel in (b)] in T. thermophilus MS are replaced by the hydrophobic residues Val525, Leu526 and Ser529 [positions −12, −11 and −8; bottom panel in (a)] that form part of the intramolecular interface in T. maritima MS. The solvent-exposed Pro611, Gln614 and Glu617 residues [positions 7, 10 and 13; bottom panel in (b)] in T. thermophilus MS are replaced by the hydrophobic residues Leu544, Thr547 and Leu552 [positions 7, 10 and 13; bottom panel in (a)] that reside in helix α8 and form part of its intermolecular interface with α7 and α1 in T. maritima MS. In T. thermophilus MS, the hydrophobic residues Ala623 and Leu626 residing in α8cF [positions 19 and 22; bottom panel in (b)] are replaced by the charged residues Glu556 and Glu559 in T. maritima MS.

A phylogenetic analysis (Supplementary Fig. S4) demonstrating that the T. maritima Fol module groups with other members of the MeTr family and not with the MS Fol modules may lend support to the notion that T. maritima MS is an outlier with a distinct method of connecting the Fol module to the Cob module. More structures of these modules are required to ascertain the significance of the observed variance. It is tempting to speculate that the variation in the TIM-barrel fold and the unique C-terminal arrangement in the majority of the Fol modules in MSs may have functional implications. Given the ability of the linker between the Fol and Cob modules to propagate structural changes, we hypothesize that this linker plays a significant role in the structural rearrangements that are required to accommodate the multitude of conformations observed during catalysis by and reactivation of MS (Fig. 1[link]a).

Despite the clear structural differences in the C-terminus of the various MS Fol modules, the rest of the Fol module (including the folate-binding site) is essentially indistinguishable across all structures (Figs. 3[link] and 5[link]; Evans et al., 2004[Evans, J., Huddler, D., Hilgers, M., Romanchuk, G., Matthews, R. & Ludwig, M. (2004). Proc. Natl Acad. Sci. USA, 101, 3729-3736.]; Doukov et al., 2007[Doukov, T., Hemmi, H., Drennan, C. & Ragsdale, S. (2007). J. Biol. Chem. 282, 6609-6618.]; Sjuts et al., 2015[Sjuts, H., Dunstan, M. S., Fisher, K. & Leys, D. (2015). Acta Cryst. D71, 1900-1908.]). In general, the interactions between the folate substrate and active-site residues are mostly invariable in all folate-corrinoid methyltransferases. However, small differences can be found, as illustrated by the comparison of the CH3-H4folate-binding site for selected representative methyltransferases shown in Fig. 5[link].

While the structures of the excised Fol modules do not reveal any substantial CH3-H4folate-induced structural changes (Supplementary Fig. S2), it cannot be ruled out that such changes may occur upon the formation of a ternary complex. Structural changes could be induced upon the addition of the Cbl cofactor and its binding module in the context of the full-length enzyme. This ternary catalytically active conformation would support methyl transfer from the CH3-H4folate substrate to the cob(I)alamin cofactor. However, comparison of our T. thermophilus MS structures with those of other corrinoid-binding methyltransferases suggest that during catalysis the folate substrate is most likely to be activated by subtle changes in the active site. Moreover, very limited motion of the folate substrate is expected upon its interaction with the corrinoid cofactor during catalysis. While there is an emerging picture of the binary structures of protein–CH3-H4folate complexes, these might solely represent a binding (or recognition) mode and not an activation, catalysis-ready mode. The differences between the binding and the activation modes during the catalytic steps warrant future investigation.

5. Conclusions

We determined the crystal structure of the MS Fol module from T. thermophilus in the presence and absence of the CH3-H4folate substrate. We find that the active-site arrangement and the mode of folate recognition and binding are indistinguishable from those of previously determined folate-dependent methyltransferases, with the exception of MT2DH. Despite the similarities, our structures of the Fol module from T. thermophilus MS reveal a unique variation in the C-terminal region distinct from other folate-binding proteins with a classical β8α8 fold. This unique topology is present in other MS Fol modules for which structures have been deposited. The fold of these MS Fol modules is a modified version of the classical TIM-barrel fold and can be best described as a β8α7 barrel. Critical for the formation of this TIM barrel with a `twist' is the presence on its C-terminus of residues that are part of a linker region connecting the Fol module to the Cob module in MS enzymes. We propose that the presence of this unique β8α7 fold in the Fol module of MS proteins distinguishes them from other members of the MeTr family.

Acknowledgements

The authors acknowledge GM/CA at the Advanced Photon Source for beam time and thank Clay Brown (LSI, University of Michigan) for providing the pMCSG7 plasmid. KY carried out molecular biology, protein expression and crystallography; KY and MK analysed the data. Both authors contributed to the manuscript. The authors declare no competing financial interests.

Funding information

This work was supported by the American Heart Association (grant 13SDG14560056 to MK).

References

First citationAchari, A., Somers, D. O., Champness, J. N., Bryant, P. K., Rosemond, J. & Stammers, D. K. (1997). Nature Struct. Biol. 4, 490–497.  CrossRef CAS PubMed Google Scholar
First citationAdams, P. D., Grosse-Kunstleve, R. W., Hung, L.-W., Ioerger, T. R., McCoy, A. J., Moriarty, N. W., Read, R. J., Sacchettini, J. C., Sauter, N. K. & Terwilliger, T. C. (2002). Acta Cryst. D58, 1948–1954.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationBanerjee, R. (1999). Editor. Chemistry and Biochemistry of B12. New York: Wiley.  Google Scholar
First citationBommer, M., Kunze, C., Fesseler, J., Schubert, T., Diekert, G. & Dobbek, H. (2014). Science, 346, 455–458.  CrossRef CAS PubMed Google Scholar
First citationChen, V. B., Arendall, W. B., Headd, J. J., Keedy, D. A., Immormino, R. M., Kapral, G. J., Murray, L. W., Richardson, J. S. & Richardson, D. C. (2010). Acta Cryst. D66, 12–21.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationCrooks, G. E., Hon, G., Chandonia, J.-M. & Brenner, S. E. (2004). Genome Res. 14, 1188–1190.  Web of Science CrossRef PubMed CAS Google Scholar
First citationDoukov, T., Hemmi, H., Drennan, C. & Ragsdale, S. (2007). J. Biol. Chem. 282, 6609–6618.  CrossRef PubMed CAS Google Scholar
First citationDoukov, T., Seravalli, J., Stezowski, J. J. & Ragsdale, S. W. (2000). Structure, 8, 817–830.  CrossRef PubMed CAS Google Scholar
First citationEmsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. (2010). Acta Cryst. D66, 486–501.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationEvans, J., Huddler, D., Hilgers, M., Romanchuk, G., Matthews, R. & Ludwig, M. (2004). Proc. Natl Acad. Sci. USA, 101, 3729–3736.  CrossRef PubMed CAS Google Scholar
First citationGoldman, A. D., Beatty, J. T. & Landweber, L. F. (2016). J. Mol. Evol. 82, 17–26.  CrossRef CAS PubMed Google Scholar
First citationHampele, I. C., D'Arcy, A., Dale, G. E., Kostrewa, D., Nielsen, J., Oefner, C., Page, M. G. P., Schönfeld, H.-J., Stüber, D. & Then, R. L. (1997). J. Mol. Biol. 268, 21–30.  CrossRef CAS PubMed Google Scholar
First citationHolm, L. & Rosenström, P. (2010). Nucleic Acids Res. 38, W545–W549.  Web of Science CrossRef CAS PubMed Google Scholar
First citationHuang, S., Romanchuk, G., Pattridge, K., Lesley, S., Wilson, I., Matthews, R. & Ludwig, M. (2007). Protein Sci. 16, 1588–1595.  CrossRef PubMed CAS Google Scholar
First citationKabsch, W. (2010). Acta Cryst. D66, 125–132.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationKoutmos, M., Pejchal, R., Bomer, T. M., Matthews, R. G., Smith, J. L. & Ludwig, M. L. (2008). Proc. Natl Acad. Sci. USA, 105, 3286–3291.  CrossRef PubMed CAS Google Scholar
First citationKrissinel, E. & Henrick, K. (2007). J. Mol. Biol. 372, 774–797.  Web of Science CrossRef PubMed CAS Google Scholar
First citationKung, Y., Ando, N., Doukov, T. I., Blasiak, L. C., Bender, G., Seravalli, J., Ragsdale, S. W. & Drennan, C. L. (2012). Nature (London), 484, 265–269.  Web of Science CrossRef CAS PubMed Google Scholar
First citationMarchler-Bauer, A. et al. (2015). Nucleic Acids Res. 43, D222–D226.  Web of Science CAS PubMed Google Scholar
First citationMatthews, R., Koutmos, M. & Datta, S. (2008). Curr. Opin. Struct. Biol. 18, 658–666.  CrossRef PubMed CAS Google Scholar
First citationMcCoy, A. J., Grosse-Kunstleve, R. W., Adams, P. D., Winn, M. D., Storoni, L. C. & Read, R. J. (2007). J. Appl. Cryst. 40, 658–674.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationMetz, J. (1992). Annu. Rev. Nutr. 12, 59–79.  CrossRef PubMed CAS Google Scholar
First citationMurshudov, G. N., Skubák, P., Lebedev, A. A., Pannu, N. S., Steiner, R. A., Nicholls, R. A., Winn, M. D., Long, F. & Vagin, A. A. (2011). Acta Cryst. D67, 355–367.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationPayne, K. A., Quezada, C. P., Fisher, K., Dunstan, M. S., Collins, F. A., Sjuts, H., Levy, C., Hay, S., Rigby, S. E. & Leys, D. (2014). Nature (London), 517, 513–516.  CrossRef PubMed Google Scholar
First citationRagsdale, S. W. & Pierce, E. (2008). Biochim. Biophys. Acta, 1784, 1873–1898.  CrossRef PubMed CAS Google Scholar
First citationSchrauzer, G. N., Deutsch, E. & Windgassen, R. J. (1968). J. Am. Chem. Soc. 90, 2441–2442.  CrossRef CAS PubMed Google Scholar
First citationSjuts, H., Dunstan, M. S., Fisher, K. & Leys, D. (2015). Acta Cryst. D71, 1900–1908.  CrossRef IUCr Journals Google Scholar
First citationSmith, A. E. & Matthews, R. G. (2000). Biochemistry, 39, 13880–13890.  CrossRef PubMed CAS Google Scholar
First citationStols, L., Gu, M., Dieckman, L., Raffen, R., Collart, F. R. & Donnelly, M. I. (2002). Protein Expr. Purif. 25, 8–15.  Web of Science CrossRef PubMed CAS Google Scholar
First citationStudier, F. W. (2005). Protein Expr. Purif. 41, 207–234.  Web of Science CrossRef PubMed CAS Google Scholar
First citationWatkins, D. et al. (2002). Am. J. Hum. Genet. 71, 143–153.  CrossRef PubMed CAS Google Scholar
First citationYokoyama, S., Hirota, H., Kigawa, T., Yabuki, T., Shirouzu, M., Terada, T., Ito, Y., Matsuo, Y., Kuroda, Y., Nishimura, Y., Kyogoku, Y., Miki, K., Masui, R. & Kuramitsu, S. (2000). Nature Struct. Biol. 7, 943–945.  Web of Science CrossRef PubMed CAS Google Scholar

© International Union of Crystallography. Prior permission is not required to reproduce short quotations, tables and figures from this article, provided the original authors and source are cited. For more information, click here.

Journal logoSTRUCTURAL
BIOLOGY
ISSN: 2059-7983
Follow Acta Cryst. D
Sign up for e-alerts
Follow Acta Cryst. on Twitter
Follow us on facebook
Sign up for RSS feeds