Structural Basis for the Serratia marcescens Lipase Secretion System: Crystal Structures of the Membrane Fusion Protein and Nucleotide-Binding Domain.

Serratia marcescens secretes a lipase, LipA, through a type I secretion system (T1SS). The T1SS for LipA, the Lip system, is composed of an inner membrane ABC transporter with its nucleotide-binding domains (NBD), LipB, a membrane fusion protein, LipC, and an outer membrane channel protein, LipD. Passenger protein secreted by this system has been functionally and structurally characterized well, but relatively little information about the transporter complex is available. Here, we report the crystallographic studies of LipC without the membrane anchor region, LipC-, and the NBD of LipB (LipB-NBD). LipC- crystallographic analysis has led to the determination of the structure of the long α-helical and lipoyl domains, but not the area where it interacts with LipB, suggesting that the region is flexible without LipB. The long α-helical domain has three α-helices, which interacts with LipD in the periplasm. LipB-NBD has the common overall architecture and ATP hydrolysis activity of ABC transporter NBDs. Using the predicted models of full-length LipB and LipD, the overall structural insight into the Lip system is discussed.

Structural and mechanistic insights into the kynurenine aminotransferase-mediated excretion of kynurenic acid.

Kynurenine aminotransferase (KAT) is a homodimeric pyridoxal protein that mediates the catalytic conversion of kynurenine (KYN) to kynurenic acid (KYA), an endogenous N-methyl-d-aspartate (NMDA) receptor antagonist. KAT is involved in the biosynthesis of glutamic and aspartic acid, functions as a neurotransmitter for the NMDA receptor in mammals, and is regulated by allosteric mechanisms. Its importance in various diseases such as schizophrenia makes KAT a highly attractive drug target. Here, we present the crystal structure of the Pyrococcus horikoshii KAT (PhKAT) in complex with pyridoxamine phosphates (PMP), KYN, and KYA. Surprisingly, the PMP was bound to the LYS-269 of phKAT by forming a covalent hydrazine bond. This crystal structure clearly shows that an amino group of KYN was transaminated to PLP, which forms a Schiff's base with the LYS-269 of the KYN. Thus, our structure confirms that the PMPs represent an intermediate state during the KAT reaction. Thus, PhKAT catalyzes the sequential conversion of KYN to KYA via the formation of an intermediate 4-(2-aminophenyl)-2,4-dioxobutanoate (4AD), which is spontaneously converted to KYA in the absence of an amino group acceptor. Furthermore, we identified the two entry and exit sites of the PhKAT homodimer for KYN and KYA, respectively. The structural data on PhKAT presented in this manuscript contributes to further the understanding of transaminase enzyme reaction mechanisms.

Structural basis for salt-dependent folding of ribonuclease H1 from halophilic archaeon Halobacterium sp. NRC-1.

RNase H1 from extreme halophilic archaeon Halobacterium sp. NRC-1 (Halo-RNase H1) requires ⩾2M NaCl, ⩾10mM MnCl2, or ⩾300mM MgCl2 for folding. To understand the structural basis for this salt-dependent folding of Halo-RNase H1, the crystal structure of Halo-RNase H1 was determined in the presence of 10mM MnCl2. The structure of Halo-RNase H1 highly resembles those of metagenome-derived LC11-RNase H1 and Sulfolobus tokodaii RNase H1 (Sto-RNase H1), except that it contains two Mn(2+) ions at the active site and has three bi-aspartate sites on its surface. To examine whether negative charge repulsion at these sites are responsible for low-salt denaturation of Halo-RNase H1, a series of the mutant proteins of Halo-RNase H1 at these sites were constructed. The far-UV CD spectra of these mutant proteins measured in the presence of various concentrations of NaCl suggest that these mutant proteins exist in an equilibrium between a partially folded state and a folded state. However, the fraction of the protein in a folded state is nearly 0% for the active site mutant, 40% for the bi-aspartate site mutant, and 70% for the mutant at both sites in the absence of salt. The active site mutant requires relatively low concentration (∼0.5M) of salt for folding. These results suggest that suppression of negative charge repulsion at both active and bi-aspartate sites by salt is necessary to yield a folded protein.

Formation of the high-affinity calcium binding site in pro-subtilisin E with the insertion sequence IS1 of pro-Tk-subtilisin.

Subtilisin E is activated from its inactive precursor Pro-subtilisin E by autoprocessing and degradation of the propeptide. Subtilisin E has two calcium binding sites, the high-affinity Ca1 site and the low-affinity Ca2 site. The Ca1 site is conserved in various subtilisin-like proteases and is important for stability. This site is not formed in Pro-subtilisin E, because the structural rearrangement of the N-terminal region of the subtilisin domain upon autoprocessing is necessary for the formation of this site. As a result, Pro-subtilisin E is not fully folded. In contrast, Pro-Tk-subtilisin from Thermococcus kodakarensis is fully folded, because it does not require the structural rearrangement upon autoprocessing for the formation of the Ca1 site due to the presence of the insertion sequence IS1 between the propeptide and subtilisin domains. To examine whether the Ca1 site is formed in Pro-subtilisin E by inserting IS1 between the propeptide and subtilisin domains, the Pro-subtilisin E mutant with this insertion, IS1-Pro-subtilisin E, and its active site mutants, IS1-Pro-S221A and IS1-Pro-S221C, were constructed and characterized. The crystal structure of IS1-Pro-S221A revealed that this protein is fully folded and the Ca1 site is formed. In this structure, IS1 serves as a linker that brings the N-terminus of the subtilisin domain near the Ca1 site. IS1-Pro-S221A in a calcium-bound form was more stable than that in a calcium-free form by 13.1 °C. IS1-Pro-S221C was more rapidly autoprocessed than Pro-S221C. These results suggest that IS1 facilitates the formation of the Ca1 site and the complete folding of Pro-subtilisin E and thereby accelerates its autoprocessing.

Crystallization and X-ray structure determination of a thermoalkalophilic lipase from Geobacillus SBS-4S.

A thermoalkalophilic lipase (LIPSBS) from the newly isolated Geobacillus strain SBS-4S which hydrolyzes a wide range of fatty acids has been characterized. In the present study, the crystallization of purified LIPSBS using the sitting-drop vapour-diffusion method and its X-ray diffraction studies are described. The crystals belonged to the orthorhombic space group P212121, with unit-cell parameters a = 55.13, b = 71.75, c = 126.26 Å. The structure was determined at 1.6 Å resolution by the molecular-replacement method using the lipase from G. stearothermophilus L1 as a model.

Crystal structure of stable protein CutA1 from psychrotrophic bacterium Shewanella sp. SIB1.

CutA1 is widely found in bacteria, plants and animals, including humans. The functions of CutA1, however, have not been well clarified. It is known that CutA1s from Pyrococcus horikoshii, Thermus thermophilus and Oryza sativa unfold at temperatures remarkably higher than the growth temperatures of the host organisms. In this work the crystal structure of CutA1 from the psychrotrophic bacterium Shewanella sp. SIB1 (SIB1-CutA1) in a trimeric form was determined at 2.7 Šresolution. This is the first crystal structure of a psychrotrophic CutA1. The overall structure of SIB1-CutA1 is similar to those of CutA1 from Homo sapiens, Escherichia coli, Pyrococcus horikoshii, Thermus thermophilus, Termotoga maritima, Oryza sativa and Rattus norvergicus. A peculiarity is observed in the ß2 strand. The ß2 strand is divided into two short ß strands, ß2a and ß2b, in SIB1-CutA1. A thermal denaturation experiment revealed that SIB1-CutA1 does not unfold completely at 363 K at pH 7.0, although Shewanella sp. SIB1 cannot grow at temperatures exceeding 303 K. These results indicate that the trimeric structural motif of CutA1 is the critical factor in its unusually high stability and suggest that CutA1 needs to maintain its high stability in order to function, even in psychrotrophs.

Affinity shift of ATP upon glycerol binding to a glycerol kinase from the hyperthermophilic archaeon Thermococcus kodakarensis KOD1.

Glycerol kinase (GK) is a key enzyme of glycerol metabolism. It participates in glycolysis and lipid membrane biosynthesis. A hexamer of GK from the hyperthermophilic archaeon Thermococcus kodakarensis KOD1(Tk-GK) was identified as a substrate-binding form of the enzyme. Here, the X-ray crystal structure analysis and the biochemical analysis was done and the relationships between its unique oligomer structure and substrate binding affinity were investigated. Wild type GK and mutant K271E GK, which disrupts the hexamer formation interface, were crystallized with and without their substrates and analyzed at 2.19-3.05 Å resolution. In the absence of glycerol, Tk-GK was a dimer in solution. In the presence of its glycerol substrate, however, it became a hexamer consisting of three symmetrical dimers about the threefold axis. Through glycerol binding, all Tk-GK molecules in the hexamer were in closed form as a result of domain-motion. The closed form of Tk-GK had tenfold higher ATP affinity than the open form of Tk-GK. The hexamer structure stabilized the closed conformation and enhanced ATP binding affinity when the GK was bound to glycerol. This molecular mechanism is quite simple activity regulation mechanism among known GKs.

Crystal structure of the Fab region of a neutralizing antibody against granulocyte-macrophage colony-stimulating factor.

An increased level of granulocyte-macrophage colony-stimulating factor has a potential role in the development of autoimmune diseases, and the neutralization of its activity by monoclonal antibodies is a promising therapy for some diseases. Here, the crystal structure of the Fab region of EV1007, a fully human antibody expressed in Chinese hamster ovary cells that was developed from human peripheral blood mononuclear cells, is described. The structure closely resembles that of MB007, which is the Fab region of the same antibody expressed in Escherichia coli [Blech et al. (2012), Biochem. J. 447, 205-215], except at the hinge regions between the immunoglobulin domains and the H3 loop region. This paper presents evidence for the flexibility of the hinge and H3 loop regions of the antibody based on the comparison of two independently solved crystal structures.

Crystal structure of a family I.3 lipase from Pseudomonas sp. MIS38 in a closed conformation.

The crystal structure of a family I.3 lipase from Pseudomonas sp. MIS38 in a closed conformation was determined at 1.5A resolution. This structure highly resembles that of Serratia marcescens LipA in an open conformation, except for the structures of two lids. Lid1 is anchored by a Ca2+ ion (Ca1) in an open conformation, but lacks this Ca1 site and greatly changes its structure and position in a closed conformation. Lid2 forms a helical hairpin in an open conformation, but does not form it and covers the active site in a closed conformation. Based on these results, we discuss on the lid-opening mechanism.

Extracellular overproduction and preliminary crystallographic analysis of a family I.3 lipase.

A family I.3 lipase from Pseudomonas sp. MIS38 was secreted from Escherichia coli cells to the external medium, purified and crystallized and preliminary crystallographic studies were performed. The crystal was grown at 277 K by the hanging-drop vapour-diffusion method. Native X-ray diffraction data were collected to 1.7 A resolution using synchrotron radiation at station BL38B1, SPring-8. The crystal belongs to space group P2(1), with unit-cell parameters a = 48.79, b = 84.06, c = 87.04 A. Assuming the presence of one molecule per asymmetric unit, the Matthews coefficient V(M) was calculated to be 2.73 A3 Da(-1) and the solvent content was 55%.

Importance of a repetitive nine-residue sequence motif for intracellular stability and functional structure of a family I.3 lipase.

PML5 is a functional derivative of a family I.3 lipase from Pseudomonas sp. MIS38 and contains five repeats of a nine-residue sequence motif. Two aspartate residues within the second and third repetitive sequences of PML5 were replaced by Ala. The secretion level, intracellular accumulation level, and stability of the resultant mutant protein were greatly reduced as compared to those of PML5. In addition, this mutant protein was inactive and did not bind Ca2+ ion. We propose that the repetitive sequences of PML5 form a beta-roll structure in the cells and thereby contribute to the intracellular stability and secretion efficiency of the protein.

Structural and biochemical characterization of a metagenome-derived esterase with a long N-terminal extension.

The genes encoding six novel esterolytic/lipolytic enzymes, termed LC-Est1∼6, were isolated from a fosmid library of a leaf-branch compost metagenome by functional screening using tributyrin agar plates. These enzymes greatly vary in size and amino acid sequence. The highest identity between the amino acid sequence of each enzyme and that available from the database varies from 44 to 73%. Of these metagenome-derived enzymes, LC-Est1 is characterized by the presence of a long N-terminal extension (LNTE, residues 26-283) between a putative signal peptide (residues 1-25) and a C-terminal esterase domain (residues 284-510). A putative esterase from Candidatus Solibacter usitatus (CSu-Est) is the only protein, which shows the significant amino acid sequence identity (46%) to the entire region of LC-Est1. To examine whether LC-Est1 exhibits activity and its LNTE is important for activity and stability of the esterase domain, LC-Est1 (residues 26-510), LC-Est1C (residues 284-510), and LC-Est1C* (residues 304-510) were overproduced in E. coli, purified, and characterized. LC-Est1C* was only used for structural analysis. The crystal structure of LC-Est1C* highly resembles that of the catalytic domain of Thermotoga maritima esterase, suggesting that LNTE is not required for folding of the esterase domain. The enzymatic activity of LC-Est1C was lower than that of LC-Est1 by 60%, although its substrate specificity was similar to that of LC-Est1. LC-Est1C was less stable than LC-Est1 by 3.3°C. These results suggest that LNTE of LC-Est1 rather exists as an independent domain but is required for maximal activity and stability of the esterase domain.

Structure, activity, and stability of metagenome-derived glycoside hydrolase family 9 endoglucanase with an N-terminal Ig-like domain.

A metagenome-derived glycoside hydrolase family 9 enzyme with an N-terminal immunoglobulin-like (Ig-like) domain, leaf-branch compost (LC)-CelG, was characterized and its crystal structure was determined. LC-CelG did not hydrolyze p-nitrophenyl cellobioside but hydrolyzed CM-cellulose, indicating that it is endoglucanase. LC-CelG exhibited the highest activity at 70°C and >80% of the maximal activity at a broad pH range of 5-9. Its denaturation temperature was 81.4°C, indicating that LC-CelG is a thermostable enzyme. The structure of LC-CelG resembles those of CelD from Clostridium thermocellum (CtCelD), Cel9A from Alicyclobacillus acidocaldarius (AaCel9A), and cellobiohydrolase CbhA from C. thermocellum (CtCbhA), which show relatively low (29-31%) amino acid sequence identities to LC-CelG. Three acidic active site residues are conserved as Asp194, Asp197, and Glu558 in LC-CelG. Ten of the thirteen residues that form the substrate binding pocket of AaCel9A are conserved in LC-CelG. Removal of the Ig-like domain reduced the activity and stability of LC-CelG by 100-fold and 6.3°C, respectively. Removal of the Gln40- and Asp99-mediated interactions between the Ig-like and catalytic domains destabilized LC-CelG by 5.0°C without significantly affecting its activity. These results suggest that the Ig-like domain contributes to the stabilization of LC-CelG mainly due to the Gln40- and Asp99-mediated interactions. Because the LC-CelG derivative lacking the Ig-like domain accumulated in Escherichia coli cells mostly in an insoluble form and this derivative accumulated in a soluble form exhibited very weak activity, the Ig-like domain may be required to make the conformation of the active site functional and prevent aggregation of the catalytic domain.

Structure and stability of metagenome-derived glycoside hydrolase family 12 cellulase (LC-CelA) a homolog of Cel12A from Rhodothermus marinus.

Ten genes encoding novel cellulases with putative signal peptides at the N-terminus, termed pre-LC-CelA-J, were isolated from a fosmid library of a leaf-branch compost metagenome by functional screening using agar plates containing carboxymethyl cellulose and trypan blue. All the cellulases except pre-LC-CelG have a 14-29 residue long flexible linker (FL) between the signal peptide and the catalytic domain. LC-CelA without a signal peptide (residues 20-261), which shows 76% amino acid sequence identity to Cel12A from Rhodothermus marinus (RmCel12A), was overproduced in Escherichia coli, purified and characterized. LC-CelA exhibited its highest activity across a broad pH range (pH 5-9) and at 90 °C, indicating that LC-CelA is a highly thermostable cellulase, like RmCel12A. The crystal structure of LC-CelA was determined at 1.85 Å resolution and is nearly identical to that of RmCel12A determined in a form without the FL. Both proteins contain two disulfide bonds. LC-CelA has a 16-residue FL (residues 20-35), most of which is not visible in the electron density map, probably due to structural disorder. However, Glu34 and Pro35 form hydrogen bonds with the central region of the protein. ΔFL-LC-CelA (residues 36-261) and E34A-LC-CelA with a single Glu34 â†’ Ala mutation were therefore constructed and characterized. ΔFL-LC-CelA and E34A-LC-CelA had lower melting temperatures (T m) than LC-CelA by 14.7 and 12.0 °C respectively. The T m of LC-CelA was also decreased by 28.0 °C in the presence of dithiothreitol. These results suggest that Glu34-mediated hydrogen bonds and the two disulfide bonds contribute to the stabilization of LC-CelA.

Rational design of a glycosynthase by the crystal structure of ß-galactosidase from Bacillus circulans (BgaC) and its use for the synthesis of N-acetyllactosamine type 1 glycan structures.

The crystal structure of ß-galactosidase from Bacillus circulans (BgaC) was determined at 1.8Å resolution. The overall structure of BgaC consists of three distinct domains, which are the catalytic domain with a TIM-barrel structure and two all-ß domains (ABDs). The main-chain fold and steric configurations of the acidic and aromatic residues at the active site were very similar to those of Streptococcus pneumoniae ß(1,3)-galactosidase BgaC in complex with galactose. The structure of BgaC was used for the rational design of a glycosynthase. BgaC belongs to the glycoside hydrolase family 35. The essential nucleophilic amino acid residue has been identified as glutamic acid at position 233 by site-directed mutagenesis. Construction of the active site mutant BgaC-Glu233Gly gave rise to a galactosynthase transferring the sugar moiety from α-d-galactopyranosyl fluoride (αGalF) to different ß-linked N-acetylglucosamine acceptor substrates in good yield (40-90%) with a remarkably stable product formation. Enzymatic syntheses with BgaC-Glu233Gly afforded the stereo- and regioselective synthesis of ß1-3-linked key galactosides like galacto-N-biose or lacto-N-biose.

Role of N-terminal extension of Bacillus stearothermophilus RNase H2 and C-terminal extension of Thermotoga maritima RNase H2.

Bacillus stearothermophilus RNase H2 (BstRNH2) and Thermotoga maritima RNase H2 (TmaRNH2) have N-terminal and C-terminal extensions, respectively, as compared with Aquifex aeolicus RNase H2 (AaeRNH2). To analyze the role of these extensions, BstRNH2 and TmaRNH2 without these extensions were constructed, and their biochemical properties were compared with those of their intact partners and AaeRNH2. The far-UV CD spectra of all proteins were similar, suggesting that the protein structure is not significantly altered by removal of these extensions. However, both the junction ribonuclease and RNase H activities of BstRNH2 and TmaRNH2, as well as their substrate-binding affinities, were considerably decreased by removal of these extensions. The stability of BstRNH2 and TmaRNH2 was also decreased by removal of these extensions. The activity, substrate binding affinity and stability of TmaRNH2 without the C-terminal 46 residues were partly restored by the attachment of the N-terminal extension of BstRNH2. These results suggest that the N-terminal extension of BstRNH2 functions as a substrate-binding domain and stabilizes the RNase H domain. Because the C-terminal extension of TmaRNH2 assumes a helix hairpin structure and does not make direct contact with the substrate, this extension is probably required to make the conformation of the substrate-binding site functional. AaeRNH2 showed comparable junction ribonuclease activity to those of BstRNH2 and TmaRNH2, and was more stable than these proteins, indicating that bacterial RNases H2 do not always require an N-terminal or C-terminal extension to increase activity, substrate-binding affinity, and/or stability.

Ca2+ regulates the Drosophila Stoned-A and Stoned-B proteins interaction with the C2B domain of Synaptotagmin-1.

The dicistronic Drosophila stoned gene is involved in exocytosis and/or endocytosis of synaptic vesicles. Mutations in either stonedA or stonedB cause a severe disruption of neurotransmission in fruit flies. Previous studies have shown that the coiled-coil domain of the Stoned-A and the µ-homology domain of the Stoned-B protein can interact with the C2B domain of Synaptotagmin-1. However, very little is known about the mechanism of interaction between the Stoned proteins and the C2B domain of Synaptotagmin-1. Here we report that these interactions are increased in the presence of Ca(2+). The Ca(2+)-dependent interaction between the µ-homology domain of Stoned-B and C2B domain of Synaptotagmin-1 is affected by phospholipids. The C-terminal region of the C2B domain, including the tryptophan-containing motif, and the Ca(2+) binding loop region that modulate the Ca(2+)-dependent oligomerization, regulates the binding of the Stoned-A and Stoned-B proteins to the C2B domain. Stoned-B, but not Stoned-A, interacts with the Ca(2+)-binding loop region of C2B domain. The results indicate that Ca(2+)-induced self-association of the C2B domain regulates the binding of both Stoned-A and Stoned-B proteins to Synaptotagmin-1. The Stoned proteins may regulate sustainable neurotransmission in vivo by binding to Ca(2+)-bound Synaptotagmin-1 associated synaptic vesicles.

Requirement of lid2 for interfacial activation of a family I.3 lipase with unique two lid structures.

A family I.3 lipase from Pseudomonas sp. MIS38 (PML) is characterized by the presence of two lids (lid1 and lid2) that greatly change conformation upon substrate binding. While lid1 represents the commonly known lid in lipases, lid2 is unique to PML and other family I.3 lipases. To clarify the role of lid2 in PML, a lid2 deletion mutant (ΔL2-PML) was constructed by deleting residues 35-64 of PML. ΔL2-PML requires calcium ions for both lipase and esterase activities as does PML, suggesting that it exhibits activity only when lid1 is fully open and anchored by the catalytically essential calcium ion, as does PML. However, when the enzymatic activity was determined using triacetin, the activity of PML exponentially increased as the substrate concentration reached and increased beyond the critical micellar concentration, while that of ΔL2-PML did not. These results indicate that PML undergoes interfacial activation, while ΔL2-PML does not. The activities of ΔL2-PML for long-chain triglycerides significantly decreased while its activity for fatty acid ethyl esters increased, compared with those of PML. Comparison of the tertiary models of ΔL2-PML in a closed and open conformation, which are optimized by molecular dynamics simulation, with the crystal structures of PML suggests that the hydrophobic surface area provided by lid1 and lid2 in an open conformation is considerably decreased by the deletion of lid2. We propose that the hydrophobic surface area provided by these lids is necessary to hold the micellar substrates firmly to the active site and therefore lid2 is required for interfacial activation of PML.

Role of polar and nonpolar residues at the active site for PPIase activity of FKBP22 from Shewanella sp. SIB1.

FKBP22 from the psychotropic bacterium Shewanella sp. SIB1 is a homodimeric protein with peptidyl prolyl cis-trans isomerase (PPIase) activity. According to a tertiary model, several nonpolar residues including Trp157 and Phe197 form a substrate-binding cavity, and Asp137 and Arg142, which form a salt bridge, are located at the edge of this cavity. To analyze the role of these residues, nine single (D137A, R142A, W157A/F/Y, F197A/L/Y/W) and one double (D137A/R142A) mutant protein of SIB1 FKBP22 were constructed. The far- and near-UV CD spectra of these mutant proteins suggest that the mutations at Asp137 and Arg142 do not seriously affect the protein structure, while those at Trp157 and Phe197 cause a local conformational change around the mutation site. Each mutation decreased the PPIase activities of SIB1 FKBP22 for peptide and protein substrates similarly without seriously affecting chaperone function. This result indicates that SIB1 FKBP22 does not require PPIase activity for chaperone function. The PPIase activities of R142A, D137A and D137A/R142A decreased in this order, suggesting that Asp137 and Arg142 play a principal and auxiliary role in catalytic function, respectively, but Arg142 can function as a substitute of Asp137. Because the PPIase activity of SIB1 FKBP22 was not fully lost by the removal of all polar residues around the active site, the desolvation effect may also contribute to the enzymatic activity. However, the mutations of Trp157 to Phe or Phe197 to Leu greatly decrease the enzymatic activity, suggesting that the shape of the substrate-binding cavity is also important for enzymatic activity.

Structure and stability of a thermostable carboxylesterase from the thermoacidophilic archaeon Sulfolobus tokodaii.

The hormone-sensitive lipase (HSL) family is comprised of carboxylesterases and lipases with similarity to mammalian HSL. Thermophilic enzymes of this family have a high potential for use in biocatalysis. We prepared and crystallized a carboxylesterase of the HSL family from Sulfolobus tokodaii (Sto-Est), and determined its structures in the presence and absence of an inhibitor. Sto-Est forms a dimer in solution and the crystal structure suggests the presence of a stable biological dimer. We identified a residue close to the dimer interface, R267, which is conserved in archaeal enzymes of HSL family and is in close proximity to the same residue from the other monomer. Mutations of R267 to Glu, Gly and Lys were conducted and the resultant R267 mutants were characterized and crystallized. The structures of R267E, R267G and R267K are highly similar to that of Sto-Est with only slight differences in atomic coordinates. The dimerized states of R267E and R267G are unstable under denaturing conditions or at high temperature, as shown by a urea-induced dimer dissociation experiment and molecular dynamics simulation. R267E is the most unstable mutant protein, followed by R267G and R267K, as shown by the thermal denaturation curve and optimum temperature for activity. From the data, we discuss the importance of R267 in maintaining the dimer integrity of Sto-Est.