The TM1459 gene of Thermotoga maritima encodes a conserved hypothetical protein with a molecular weight of 12,977 Da (residues 1–114) and a calculated isoelectric point of 5.6. Currently, no functional annotation has been made for this protein, made fold recognition methods such as Fold and Function Assignment System (FFAS)1 have recognized significant sequence similarity to the family of cupins.2 Here, we report the crystal structure of TM1459 with an endogenous manganese ligand that was determined using the semiautomated high-throughput pipeline of the Joint Center for Structural Genomics (JCSG).3
The structure of TM1459 [Fig. 1(a)] was determined to 1.65 Å resolution using the molecular replacement (MR) method using a search model constructed from oxalate oxidase (PDB: 1FI2),4 despite a very low sequence identity (18%). Data collection, modeling and refinement statistics are summarized in Table I. The final model includes two protein molecules (residues 1–114), 2 manganese ions, and 252 water molecules. The Matthews' coefficient (Vm) for TM1459 is 2.71 Å3/Da, and the estimated solvent content is 54.2%. The Ramachandran plot, produced by Procheck 3.4,5 shows that 92.7% of the residues are in the most favored regions, and 7.3% are in additional allowed regions.
|Unit cell parameters||a = b = 52.55 Å, c = 96.27 Å, α = β = 90 °, γ = 120 °|
|Resolution range (Å)||48.13–1.65|
|Number of observations||124,395|
|Number of unique reflections||35,029|
|Completeness (%)||97.8 (87.6)a|
|Mean I/σ(I)||15.6 (1.9)a|
|Rmeas on I||0.067 (0.577)a|
|Highest resolution shell (Å)||1.74–1.65|
|Model and refinement statistics|
|Resolution range (Å)||33.07–1.65||Data set used in refinement||λ0|
|No. of reflections (total)||34,991||Cutoff criteria|||F| > 0|
|No. of reflections (test)||1,729||Rcrystd||0.208|
|Completeness (% total)||97.7||Rfreee||0.249|
|Restraints (RMS observed)|
|Bond length||0.018 Å|
|Average isotropic B-value||24.6 Å2|
|ESU-based on free R value||0.10 Å|
The final model of the TM1459 monomer is composed of 11 β-strands (β1–β11) and two short 310-helices (H1, H2). The total β-strand content is 43.6%. The TM1459 structure is characterized by two antiparallel β-sheets (A and B) that form a jellyroll β-sandwich with a topology that is characteristic of the cupin-barrel fold3, 4 [Fig. 1(a)]. The seven-stranded β-sheet A (β1′, β2–β4, β6, β9, β11) has a 2347561 topology, where β-strand β1′ is contributed from the neighboring subunit of the dimer by domain-swapping (Fig. 1). The four-stranded β-sheet B (β5, β7, β8, β10) has 1423 topology [Fig. 1(a)]. Each of the 11 β-strands runs approximately perpendicular to the barrel axis. TM1459 forms a dimer linked by a pair of crossovers between the adjacent edge β-strands β1 and β9 from different subunits in the dimer [Fig. 1(b)]. The dimer interface corresponds to interactions between the two A β-sheets with a buried surface area of 863 Å2 per monomer.6
Each TM1459 domain has a metal-binding site in the mouth of the β-barrel [Fig. 2(b)]. The metal ion has octahedral coordination in which four ligands are contributed by conserved histidine side chains that define a new sub-family of cupins. The metal coordination is different from the His, His, Glu, His metal-coordination typically found in cupins and is the second example of a metal coordinated by four histidine residues similar to that observed for manganese in the photosynthetic reaction center of Rhodobacter sphaeroides (PDB: 1YST).7 The metal-binding residues are His52, His54, His58, and His92 with metal-to-atom distances of 2.19, 2.19, 2.25 and 2.13 Å, respectively. The remaining coordination sites are occupied by two water molecules (W1, W2) at distances of 2.06 and 2.24 Å [Fig. 2(b)]. The metal has been assigned as manganese because it gave the best refined B-factor agreement with surrounding atoms (tested for a series of metals) and also based on the identity (Mn) and similar octahedral coordination of the metal in the structural homologue oxalate oxidase.4 Sigma-A-weighted OMIT maps show density continuous with the sulfhydryl group of Cys106, which in subunit A extends as a continuous tube past the Mn-coordinated water molecules. Although the density could not be identified and was modeled as an unknown ligand (UNL) H-bonded to Cys106 [Fig. 2(b)], it does suggest a catalytic role for Cys106.
A structural similarity search, performed with the coordinates of TM1459 using the DALI server,8 indicated that the closest structural homologue is oxalate oxidase (germin) from Hordeum vulgare (PDB: 1FI2), which was used here as a MR search model.4 The RMSD between TM1459 and oxalate oxidase is 1.4 Å over 74 aligned residues with 20% sequence identity. Another structural homologue is TM1287, a putative oxalate decarboxylase whose structure was determined recently (PDB: 1O4T).9 The RMSD between TM1459 and TM1287 is 1.7 Å over 80 aligned residues with 18% sequence identity.
According to FFAS,1 TM1459 has at least five distant homologues in the Thermotoga proteome: TM1287 (18% sequence identity), TM1010 (16%), TM0656 (13%), TM1112 (10%), and TM0736 (8%). Sequence similarity searches with the TM1459 sequence against the non-redundant protein sequence database revealed more than 1000 homologues in the three kingdoms of life, with about 100 homologues from this new sub-family that contain metals coordinated by four histidines. This subfamily comprises single-domain proteins like TM1459, as well as multi-domain proteins like the family of mannose phosphorylases. Models for TM1459 homologues can be accessed at http://www1.jcsg.org/cgi-bin/models/get_mor.pl?key=TM1459.
The crystal structure reported here is the first representation of a novel subfamily of cupins that contains a metal site coordinated by four histidines. The information reported here, in combination with further biochemical and biophysical studies, will yield valuable insights into the functional determinants of this protein family and the thermostability of these organisms.