The crystal structure of hypothetical protein MTH1491 from Methanobacterium thermoautotrophicum

Abstract

As part of our structural proteomics initiative, we have determined the crystal structure of MTH1491, a previously uncharacterized hypothetical protein from Methanobacterium thermoautotrophicum. MTH1491 is one of numerous structural genomics targets selected in a genome-wide survey of uncharacterized proteins. It belongs to a family of proteins whose biological function is not known. The crystal structure of MTH1491, the first structure for this family of proteins, consists of an overall five-stranded parallel β-sheet with strand order 51234 and flanking helices. The oligomeric form of this molecule is a trimer as seen from both crystal contacts and gel filtration studies. Analysis revealed that the structure of MTH1491 is similar to that of dehydrogenases, amidohydrolases, and oxidoreductases. Using a combination of sequence and structural analyses, we showed that MTH1491 does not belong to either the dehydrogenase or the amidohydrolase superfamilies of proteins.

The genome sequencing efforts have now provided biologists with coding information for thousands of new proteins, most of which have no known function that can be predicted using sequence-based methods. One of the aims of structural proteomics is to determine the three-dimensional structures of these newly discovered proteins (Christendat et al. 2000a). It is anticipated that, in some instances, these structures will lead directly to a hypothesis about function, because structure and function are often conserved in the absence of sequence homology. In other cases, the connection between structure and function will not yet be clear. Consequently in the near future, we can expect a considerable increase in the number of new structures of proteins with no known function.

Our group has set out to determine the three-dimensional structures of nonmembrane proteins for which sequence-based methods have been unable to predict a function. Membrane proteins were excluded because it is technologically challenging to produce them on a high throughput scale. This approach is designed to produce a subset of protein structures that may be useful in elucidating protein function and that will contribute to our understanding of the sequence/structure relationship.

The proteins for our analysis were selected from Methanobacterium thermoautotrophicum (MTH) whose genome comprises about 1885 open reading frames (ORFs). Fewer than 50% of the MTH gene products can be assigned a function based on their amino acid sequence analysis. Here, we report the three-dimensional structure of an MTH protein whose biological function is not known and has thus been annotated as unknown in the GenBank database. This protein (MTH1491) has several sequence homologs, all of which are also currently annotated as either unknown or hypothetical, indicating that their biological function has yet to be determined. A combination of structural, sequence, and co-crystallization data suggests a putative role for this family of proteins in sulfur oxidation or metabolism. The structural data provided here for MTH1491 will now enable more detailed functional analysis of this entire family of unknown proteins.

Results and Discussion

Amino acid sequence analysis

A PSI-BLAST analysis of the MTH1491 sequence against the nonredundant protein database identified a total of eight proteins from Archaeoglobus fulgidus (gi11498518), Sulfolobus tokodaii (gi15920325), Thermoplasma volcanium (gi14324794), Ralstonia solanacearum (gi17430280), Clostridium perfringens (gi18143755), Deinococcus radiodurans (gi15805095), Thermoplasma acidophilus (gi10640324), and Aquifex aeolicus (gi15606383) with sequence similarity ranging from 81% to 62%. These proteins are annotated as either conserved, hypothetical, or unknown, indicating that their biological function is yet to be determined (Altschul et al. 1997). Amino acid residues that are conserved within the nine organisms include Asp/Glu12, Asn27, Cys72, N74, Tyr110, and Arg/Lys 112. None of these sequence homologs have been functionally characterized; however, sequence analysis predicted that these proteins contain a "conserved cysteine-containing domain," which is proposed to function as a disulfide bond redox regulator. Proteins of this nature are involved in biochemical reactions such as sulfite reduction and regulation of the redox states of proteins in cells. These proteins all have a highly conserved cysteine that aligns with Cys72 of MTH1491. One protein with a conserved cysteine-containing domain is DsrF, a small soluble protein in the metabolic pathway for the oxidation of sulfur in phototrophic bacteria (Pott and Dahl 1998).

Structure determination

The crystal structure of MTH1491 was determined to 2.3 Å resolution by multiwavelength anomalous dispersion (MAD) phasing from a selenium-containing protein crystal. Initial phases were obtained using crystallography and NMR system (CNS), which produced high-quality electron density for most of the molecule except for the region between residues 74–81 for which the density is discontinuous. However, after several rounds of refinement with CNS, the electron density in this region became visible to a level sufficient for tracing of the main chain atoms. Residues 74–81 of the protein, found in the middle of helix 3, are solvent exposed and are not involved in any crystal contacts with other molecules, which results in their being disordered. In addition, the two amino-terminal residues, whose density is also missing, were omitted from the final model. The final model consists of 111 amino acid residues, from 3 to 113, with proline 113 in cis conformation. CNS maximum likelihood refinement to 2.3 Å resolution resulted in an R_cryst of 20.2 and an R_free of 22.6. According to PROCHECK (Laskowski et al. 1993), 96.7% of all residues in this model are in the most favored regions and 3.3% are in the additional allowed regions of the Ramachandran plot.

Overall fold

MTH1491 is a single domain molecule of 113 residues, with an overall topologic arrangement of alternating β-strands and α-helices with strand order 51234 (Fig. 1A ▶). This arrangement produces a five-stranded twisted parallel β-sheet flanked by helices on both faces and a disordered helix (α-helix 3) that lies perpendicular to the flanking helices. This structural arrangement, which is analogous to the Rossmann fold, is shared between diverse families of proteins and includes dehydrogenases, amidases, nucleotidyltransferases, and a number of flavin-binding proteins such as flavin reductase and ferredoxin.

Fig. 1.

(A) Overview of the MTH1491 structure. A schematic ribbon diagram of the MTH1491 subunit is shown with helices colored in yellow and strands colored in blue. The overall fold of the molecule consists of a five stranded β-sheet with strand order 51234 flanked by two α-helices on both ends. The major secondary structural elements and chain termini are labeled. (B) Overview of the trimeric structure of MTH1491. Each subunit of the trimer is colored differently. The threefold axis of symmetry is in the plane. This plane (the A face of the molecule) of the molecule is labeled for illustrative and discussion purposes. The arrows indicate the positions of the proposed active sites, which are located between two subunits producing a total of three active sites in the trimer. (C) Electrostatic surface representation of the trimeric form of MTH1491 with blue and red surfaces representing positive and negative potential, respectively. The six arginine residues at the threefold axis contributed to form a positively charged region at the core of the trimer, which is compensated with the remaining surface being negatively charged.

Subunit interactions

Gel filtration analysis yielded an approximate molecular mass of 36 kD for MTH1491, which corresponded with the protein being a trimer in solution. Subunit contacts and arrangement observed from the crystal structure also revealed a trimer (Fig. 1B ▶). The total surface of the oligomeric form of MTH1491 is 15,270 Ų, of which 4629 Ų is buried at the subunit interfaces. The subunit interactions are stabilized by a combination of hydrophobic, hydrogen-bonding, and ionic interactions, but none of these interactions predominate. The hydrophobic contacts are formed by contributions from residues Leu28, Leu32, Ile11 from one subunit with Val97, Ile100 and Val101 from another. The hydrogen-bonding network at the interface is produced by a number of side chain and main chain interactions. These interactions include the side chain hydroxyl of Tyr110 with the backbone amide of Arg112, the backbone amide of Tyr110 with side chain carbonyl of Gln104, the main chain carbonyl of Arg112 with main chain amide of Arg112 from the two interacting molecules. Hydrogen bonding is contributed from the carboxy-terminal proline of one subunit to the guanidinium group of Arg112 from another subunit. In addition, there is a prominent ionic interaction at the interface that involves the guanidinium group of Arg17 from one subunit and the carboxylate of Glu12 with the other subunit (Fig. 2A ▶). The threefold axis on the A face of the trimer is dominated by an extensive charged network that involves Asp11, Glu12, Arg17, Arg112, and the carboxylate of the carboxy-terminal Pro from each of the three subunits (Fig. 2A ▶). These interactions produce a large pocket that is lined by highly charged groups, but the overall charge is compensated by the opposing interactions as shown by the electrostatic surface representation (Fig. 1C ▶).

Fig. 2.

(A) A detailed view of the trimeric interface of MTH1491. A number of charged residues cluster at the threefold axis. Two arginines, an aspartate, and a glutamate are contributed from each subunit to form this highly polar interface. One of the two arginines, Arg112 is conserved as either a Lys or Arg in all nine homologs and the glutamate is also conserved as either an aspartate or glutamate in all of the homologs. (B) A detailed view of the proposed active site of MTH1491. Residues from two subunits contributed to from a complete active site and the conserved Cys residue (Cys72) is also located in this pocket. The Asn residues (Asn27 and Asn74), which are absolutely conserved between the nine homologs, are also found in this pocket.

Location of the active site

The proposed active site of oligomeric MTH1491 was located by searching the refined model for large cavities that harbored the highly conserved residues identified from the sequence alignment. We focused in particular on Cys72, which is conserved in this family of functionally unknown proteins as well as all of the proteins containing a conserved cysteine-containing domain and is thought to play a key role in sulfur oxidation. Cys72 as well as Asn27 and Asn74, three residues that are absolutely conserved in all nine homologs were found in a large cavity at the subunit interface of the trimer (Fig. 2B ▶). The conserved residues are shared between two subunits to form a complete site for a total of three active sites in the trimer. This proposed active site is composed mainly of uncharged polar residues, such as serine, threonine, cysteine, and asparagine as well as a number of well-ordered water molecules (Fig. 2B ▶).

Structure comparison

Proteins with structures similar to MTH1491 were identified using DALI (Holm and Sander 1997). The most structurally conserved protein is glutaminase–asparaginase (DALI Z-score = 6.3, r.m.s.d. = 3.7 Å over 89 C_α residues) and belongs to the family of amidohydrolases (Jakob et al. 1997). Amidohydrolases catalyze the hydrolysis of Asn and Gln residues to the corresponding acidic residues Asp and Glu. The second family of structurally similar proteins are dehydrogenases (with a range of DALI Z-scores between 5 and 3); however, this is not surprising as MTH1491 shares the β/α topology that is common to dehydrogenases. Flavin reductase is also structurally similar (DALI Z-score = 5.9, r.m.s.d. = 3.5 Å over 90 C_α residues) and belongs to the oxidoreductase family (Ingelman et al. 1999).

We initiated a more detailed structural and sequence analysis of MTH1491 with the above mentioned protein families identified by DALI. Dehydrogenases share several general structural features, including a conserved sequence motif GxGXXG, and a conserved aspartate residue at the carboxyl end of β-strand 2 (Lesk 1995). Conserved glycine residues are important for the proper packing of the helices against the β-strands and the conserved aspartate is involved in a hydrogen-bonding interaction with the 2`-hydroxyl group of the adenylyl ribose moiety. MTH1491 did not reveal any of these conserved features; therefore, it is not likely that MTH1491 is a dehydrogenase.

The structures of amidohydrolases are now available from Pseudomonas (Jakob et al. 1997), Escherichia coli (Swain et al. 1993), and Erwinia chrysanthemi (Miller et al. 1993). The DALI search identified Pseudomonas glutaminase–asparaginase (PGA) as the closest structural relative to MTH1491 with the highest Z-score (6.3) compared to all other identified proteins. Although the carboxy-terminal domain of PGA aligns well with the overall structure of MTH1491, the proposed active site residues in MTH1491 are not conserved in this region of PGA. The active site of PGA is found between the amino- and the carboxy-terminal domains; active site groups that are involved in catalysis are mainly located within the amino-terminal domain and substrate-binding groups are found in the carboxy-terminal domain. This leaves the class of oxidoreductases as possible structural and functional homologs of MTH1491. This possibility could neither be confirmed nor ruled out based on structural analysis.

In an effort to seek further evidence for a redox-related role, we analyzed the genomic organization of MTH1491 to identify a possible regulatory operon. Most of the neighboring ORF gene products are annotated as either hypothetical or unknown with the exception of the adjacent upstream gene of MTH1491, which codes for a 5` nucleotidase. Sequence analysis of this predicted nucleotidase with PSI-BLAST revealed that it shares a high degree of homology to the SoxB protein from Paracoccus denitrificans, A. aeolicus, and a number of other organisms. The thiosulfate oxidizing enzyme system of Paracoccus pantotrophus is located in the periplasm and four proteins SoxXA, SoxYZ, SoxB, and SoxCD are required for the complete oxidation of thiosulfate (Friedrich et al. 2000). Therefore, it is likely that this protein annotated as 5` nucleotidase is involved in sulfur oxidation. Because MTH1491 is adjacent to SoxB on the genome, and it has a conserved cysteine-containing domain, it is likely that MTH1491 may also be involved in sulfur oxidation.

To functionally understand how MTH1491 fits in the overall framework of this family of sulfur oxidizing proteins, we co-crystallized it with a variety of sulfate salts. Interestingly, we obtained a different crystal form with 0.05 M ammonium sulfate added to the original crystallization condition, suggesting an interaction with sulfate. Taken together, our structural and genomic analysis point to a role for MTH1491 in sulfur metabolism. Keeping in mind that the aim of structural genomics projects is to provide structural data to the general public for functional studies, further analysis to determine the function of MTH1491 was not undertaken. Nevertheless, these observations do indicate that MTH1491 may be involved in sulfate binding and may play a role in sulfur oxidation. However, further functional studies are required to confirm these findings.

Materials and methods

Cloning, protein expression, purification, and crystallization

The MTH1491 gene (GenBank accession number G69065) was cloned from genomic MTH DNA and its gene product was recombinantly expressed, selenomethionine labeled, and purified as described elsewhere for other MTH proteins (Christendat et al. 2000b). Screening for crystallization conditions was also performed as described elsewhere for other MTH proteins (Christendat et al. 2000b). The final crystallization condition consists of methyl-pentanediol (MPD) as precipitant. The crystals chosen for X-ray data collection were flash-frozen in this buffer, which also acted as cryoprotectant.

Crystallographic studies

The single crystals grow as rods with maximum dimensions of 0.30 mm by 0.10 mm by 0.4 mm. Crystals selected for MAD data collection were grown in 30% MPD and 100 mM HEPES at pH 7.5 at 20°C. The crystals belonged to the hexagonal space group P6₃, with the following unit cell parameters: a = b = 82.4, and c = 37.0 Å. The Matthew's coefficient, V_M, was determined to be 2.9 ų Da⁻¹ implying a solvent content of 57.3% with a single molecule of MTH1491 in each asymmetric unit (Matthews 1968; Westbrook 1985). These values are within the range normally found for protein crystals. The diffraction data from the remote wavelength have an R_sym of 7.1% and a completeness of 99.4% to 2.3 Å. Data collection statistics are summarized in Table 1.

Table 1.

Summary of data collection and refinement statistics

X-ray data	Peak	Edge	Remote	Refinement data
Space group	P63	P63	P63	Rcryste	20.2
Unit cell (Å3)	82.5 × 82.5 × 37.1	82.5 × 82.5 × 37.1	82.5 × 82.5 × 37.1	Rfreef	22.6
Resolution (Å)	50–2.30	50–2.30	50–2.30	protein atoms (no.)	862
Wavelength (Å)	0.97948	0.97964	0.93927	water molecules (no.)	19
No. of Se sites	3	3	3	r.m.s.d. bond lengths (Å)	0.007
No. of observations	77,390	77,281	77,199	r.m.s.d. bond angles (°)	1.1
No. of unique reflections	6492	6500	6566	r.m.s.d. dihedrals (°)	23.2
Intensity (I/σ≤I>)	34 (12.3)a	38 (11.1)	32 (7.4)	average B factor (°)	26.4
Completeness (%)	98.4 (99.8)	97.9 (88.6)	99.4 (99.8)
Rsymb	0.071 (0.183)	0.064 (0.192)	0.071 (0.313)
FOMMADc	0.77	FOMSFd	0.87

^a Numbers in parentheses represent values in the highest resolution shell 2.38–2.30 Å.

^b R_sym = ∑|I-〈I〉|/∑I, where I is the observed integrated intensity, 〈I〉 is the average integrated intensity obtained from multiple measurements, and the summation is over all observed reflections.

^c FOM_MAD = Figure of merit after MAD phasing.

^d FOM_SF = Figure of merit after solvent flipping.

^e R_cryst = |Fobs Fcalc|/|Fobs|.

^f R_free was calculated using randomly selected reflections (5%).

X-ray diffraction and structure determination

Making use of the anomalous scattering of Se atoms, three wavelength MAD data were collected at 100K at the 19ID beamline of the Structural Biology Center at the Advanced Photon Source, Argonne National Laboratory. Diffraction data were collected on the SBC-2, a 3 by 3 CCD detector built at APS, to 2.3 Å resolution from a single crystal containing SeMet-labeled protein at three different X-ray wavelengths near the Se edge (Table 1). The MAD data were processed using the HKL2000 suite of programs (Otwinowski and Minor 1997). Data collection statistics are presented in Table 1. The MAD phasing component of CNS was used to locate the three selenium sites, calculate the phases, and modify the density (Brunger et al. 1998). Electron density visualization and model building were done with O (Jones et al. 1991). Rigid body and simulated annealing torsion angle refinement were followed by individual B-factor refinement and performed using CNS 1.0 (Brunger et al. 1998). Several rounds of refinement were combined with model rebuilding in O after inspection of both 2F_o–F_c and F_o–F_c maps. Water molecules were initially picked using CNS and then manually verified in O using the following criteria: a peak of at least 2.5 σ in an F_o–F_c map, a peak of at least 1.0 σ in a 2F_o–F_c map, and reasonable intermolecular interactions. Refinement statistics are found in Table 1. The programs MOLSCRIPT (Kraulis 1991), RASTER 3D (Merrit and Murphy 1991), and SPOCK (Christopher 1998) were used in the production of the figures.

Accession number

Atomic coordinates have been deposited into the Protein Data Bank as PDB ID 1L1S.