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


  • Clement Angkawidjaja,

    1. Department of Material and Life Science, Graduate School of Engineering, Osaka University, Japan
    2. International College, Osaka University, Japan
    Search for more papers by this author
  • Yuichi Koga,

    1. Department of Material and Life Science, Graduate School of Engineering, Osaka University, Japan
    Search for more papers by this author
  • Kazufumi Takano,

    1. Department of Material and Life Science, Graduate School of Engineering, Osaka University, Japan
    2. CREST, JST, Osaka, Japan
    Search for more papers by this author
    • Present address
      Graduate School of Life and Environmental Sciences, Kyoto Prefectural University, Kyoto, Japan

  • Shigenori Kanaya

    1. Department of Material and Life Science, Graduate School of Engineering, Osaka University, Japan
    Search for more papers by this author

C. Angkawidjaja, Department of Material and Life Science, Graduate School of Engineering, Osaka University, 2-1 Yamadaoka, Suita, Osaka 565-0871, Japan
Fax: +81 6 6879 4517
Tel: +81 6 6879 4517
S. Kanaya, Department of Material and Life Science, Graduate School of Engineering, Osaka University, 2-1 Yamadaoka, Suita, Osaka 565-0871, Japan
Fax: +81 6 6879 7938
Tel: +81 6 6879 7938


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.


The atomic coordinates and structural factors have been deposited in the Protein Data Bank with accession numbers of PDB: 3AIK for noninhibited Sto-Est, PDB: 3AIL for DEP-bound, PDB: 3AIM for R267E, PDB: 3AIN for R267G, and PDB: 3AIO for R267K

Structured digital abstract


center of mass




hormone-sensitive lipase


main chain


molecular dynamics




side chain


Sulfolobus tokodaii


Sulfolobus tokodaii esterase


Carboxylesterase (EC3.1.1.1) is a class of enzymes that catalyzes hydrolysis of the carboxyl ester bond. It prefers short to medium chains of monoesters as a substrate, although it can also hydrolyze short-chain triglycerides. Together with lipase, carboxylesterases have broad substrate tolerance, high regio- and stereoselectivity, and stability in organic solvent, rendering them useful as biocatalysts for the synthesis of important materials [1]. Based on their amino acid sequence similarities and the presence of several conserved motifs, carboxylesterases are classified into four blocks, C, L, H and X, as described on the ESTHER database [2]. Block H includes the plant carboxylesterase and hormone-sensitive lipase (HSL) families. The HSL family consists of carboxylesterases and lipases that are distributed in all kingdoms of life with sequence similarities to mammalian HSL [3]. Bacterial members of the HSL family correspond to family IV of the bacterial lipolytic enzyme classification [4]. There has been increasing interest in microbial HSL-like carboxylesterases, especially those from thermophilic sources, because they are promising for use in biocatalysis [1,5]. Several crystal structures of archaeal HSL have been reported, such as those from Archaeoglobus fulgidus [6] and Pyrobaculum calidifontis [7].

Sulfolobus tokodaii strain 7 (Sto) is an aerobic thermoacidophilic archaeon with an optimum growth temperature and pH of 80 °C and 2.5, respectively [8]. The Sto genome sequence was reported in 2001 [9] and contains one gene (ST0071; accession no. BAB65028.1) that encodes a 303-amino acid HSL-like carboxylesterase (Sto-Est). Sto-Est shares amino acid sequence similarities with experimentally identified esterases from Sulfolobus shibatae (accession no. BAD82944.1, 70% identity), Archaeoglobus fulgidus (accession no. NP_070544.1, 44% identity), Bacillus megaterium (accession no. AAQ08176.1, 43% identity), Alicyclobacillus acidocaldaricus (accession no. YP_003186268.1, 42% identity) and Oleomonas sagaranensis (accession no. BAA82510.1, 41% identity). The amino acid sequence of Sto-Est is compared with its homolog proteins in Fig. 1. It has been reported that Sto-Est can be overproduced in Escherichia coli and purified as an active protein [10]. The enzyme was shown to exhibit high thermostability and to retain activity in the presence of organic solvents.

Figure 1.

 Amino acid sequence alignment of HSL-like carboxylesterases from Sulfolobus tokodaii (Sto), Sulfolobus shibatae (Ssh), Archaeoglobus fulgidus (Afu), Bacillus megaterium (Bme), Alicyclobacillus acidocaldaricus (Aac) and Oleomonas sagaranensis (Osa). The signature HSL-like sequence is shown in square brackets. Amino acids residues that form the catalytic triad and oxyanion hole are indicated by asterisks and open circles, respectively. The cysteine residue conserved in Sulfolobus genus is indicated by an arrow.

Here, we describe the high-resolution crystal structures of Sto-Est in the free form and in complex with diethylphosphate (DEP). All crystals contain four molecules per asymmetric unit that assemble into two biological dimers. Residues forming the oxyanion hole were identified from the DEP-bound structure. We identified R267, which forms a basic pocket together with R246 and is located close to the dimer interface, as being unique to thermophilic archaeal members of this enzyme family. Mutations of R267 to Glu, Gly and Lys were performed and the effects of each mutation on the stability of the dimer and the biochemical properties of the protein were examined. Furthermore, the crystal structures of all R267 mutants were determined. The monomer or dimer integrity of Sto-Est and R267 mutants under high temperature was examined by molecular dynamics (MD) simulation. In addition, the dimer or monomer integrity of the proteins under chemical denaturation was examined by urea/PAGE and activity measurement in the presence of urea.


Overproduction, purification, crystallization and Sto-Est structure in the noninhibited form

The gene encoding Sto-Est was cloned and from the soluble fraction of E. coli. SDS/PAGE shows that the molecular mass of the protein is 37.8 kDa, in accordance with the predicted molecular mass of 36.0 kDa (with the N-terminal His-tag included). The molecular mass of Sto-Est was estimated to be ∼ 80 kDa by gel-filtration chromatography, indicating that it is dimeric in solution. Approximately 50 mg of pure protein could be obtained from 1 L of culture. All protein samples were purified to > 95% purity.

Sto-Est structure (1.95 Å resolution) is shown in Fig. 2A. The four molecules are indicated as chains A, B, C and D, all of which share relatively the same structure with an rmsd of 282 Cα atoms ranging from 0.18 to 0.33 Å. Amino acid residues 1–20 and the uncleaved N-terminal His-tag are not visible in the electron-density map, probably because they are disordered in the crystal.

Figure 2.

 Sto-Est structures. (A) The four molecules of Sto-Est present in the asymmetric unit. The chain IDs are indicated and active-site residues are shown in the stick model. (B) The monomeric structure of Sto-Est, shown with the hypothetical substrate entrance/product exit tunnel and active-site residues. (C) Stereoview of the interface of the Sto-Est biological dimer formed by molecules B and C. (D) The structure around the active site of DEP-bound Sto-Est is shown with the corresponding 2Fo–Fc electron-density map at sigma level of 1.0 and a resolution cut-off of 1.91 Å. The residues forming the active site and oxyanion hole are indicated. The distances between O3 atom of DEP and the nitrogen atoms forming the oxyanion hole are shown.

The active site consists of S150, D243 and H273. The Oγ atoms of nucleophilic S150 in all four molecules are solvent-exposed with an average accessible surface area of 13.1 Å2. The structure contains 10 molecules of 2-methyl-2,4-pentanediol (MPD; present in crystallization buffer) that are mostly located at the substrate-binding pocket. One to three MPD molecules are present per protein molecule. The most significant differences in the four molecules are the position and number of MPD molecules. Molecule B is of particular interest because it contains three MPD molecules lining its substrate-binding pocket. Using the program caver [11], a substrate entrance/product exit tunnel could be identified, which extends from underneath a region consisting of 310-helix1–α-helix5–310-helix2 to the active-site S150 and then outward to the solvent area (Fig. 2B).

Dimer arrangement

The four molecules in the asymmetric unit form two distinct intermolecular interfaces. One is between chains A–D or B–C, and the other is between chains A–B or C–D. Analyses of intermolecular contacts using ccp4-based pisa and web-based pic showed that, although both interfaces have similar contact area, the A–D or B–C interface contains more H-bond and salt bridges than the A–B or C–D interface, which is mostly mediated by hydrophobic interactions (Table 1). Furthermore, there is one MPD molecule each in the A–B or C–D interface, suggesting that these assemblies are formed during crystallization. The values of Gibbs’ free energy of dissociation (ΔGdiss) of A–D and B–C dimers are 11.0 and 11.2 kJ·mol−1, respectively, indicating that the interactions are stable. These data suggest that the biological dimer is composed of the A–D or B–C molecules.

Table 1.   Intermolecular interactions.
ChainsInterface area (Å2)No. of H-bondsNo of salt bridgesNo. of hydrophobic interaction

The H-bond network in the biological dimer interface is formed by a combination of side chain (SC) or main chain (MC) atoms. According to pisa analysis, there are 16 H-bonds (Table 1), among which four are formed between MC–MC atoms. These MC–MC H-bonds are located at the β8–β8 interface (Fig. 2C), forming an intermolecular antiparallel β-sheet structure. The residue pairs that form this bond are S265–N269, R267–R267 and N269–S265, with the first residue of the pair corresponding to one molecule and the second to the other.

Crystallization and Sto-Est structure in the DEP-bound form (DEP–Sto-Est)

DEP–Sto-Est crystals were prepared by soaking Sto-Est crystals in the mother liquor with the addition of diethyl-p-nitrophenyl phosphatewith 20 m in excess concentration compared with that of the protein. The structure of DEP–Sto-Est was solved to 1.91 Å resolution and is highly similar to that of the noninhibited form (rmsd of Cα atoms = 0.177 Å), except that the active-site S150 of each monomer is diethylphosphorylated (Fig. 2D). The O3 atom of the DEP group of DEP–S150 is located within hydrogen bond distance (2.8–3.0 Å) from the main-chain nitrogen atoms of G79, G80 and A151, indicating that these residues form the oxyanion hole. Similar to the noninhibited structure, amino acid residues 1–20 are not visible.

Comparison with other esterases

DALI search [12] resulted in several structures with significant structural homologies. Pyrobaculum calidifontis (PestE), AFEST and Alicyclobacillus acidocaldaricus esterase (EST2) (PDB: 2YH2, 1JJI and 1EVQ, respectively) have high structural homology with Sto-Est (Z-score = 45.0, 42.6 and 41.9, rmsd of Cα atoms = 1.3, 1.4 and 1.7 Å, respectively). HSL-like carboxylesterases from environmental samples Est1 (PDB: 2C7B), Est5 (PDB: 3FAK) and Est7 (PDB: 3DNM) are also structurally homologous with Sto-Est (Z-score = 44.7, 39.1 and 38.8, respectively; rmsd of Cα atoms = 1.5, 1.7 and 1.6 Å, respectively). It is worthy of note that Est5 and Est7 are mesophilic enzymes [13,14]. Other mesophilic enzyme structures that are homologous to Sto-Est are those of Bacillus subtilis brefeldin A HSL-like carboxylesterase (PDB: 1JKM), Rhodococcus sp. heroin esterase (PDB: 1LZK), Salmonella typhimurium acetyl esterase (PDB: 3GA7) and plant carboxylesterase AeCXE1 from Actinidia eriantha (PDB: 2O7R), with values for the Z-score and rmsd of Cα atoms ranging from 30.9 to 39.5 and 1.6 to 2.1 Å, respectively. A psychrophilic enzyme from Oleispira antartica (PDB: 3I6Y) was found to be a distinct homolog of Sto-Est (Z-score = 19.9, rmsd of Cα atoms = 3.0 Å).

Basic pocket at the dimer interface of archaeal enzymes

DALI structural alignment of Sto-Est with its thermophilic, mesophilic and psychrophilic structural homologs (Fig. 3A) revealed that the arginine and glutamate residues corresponding to R267 and E250, respectively, of Sto-Est are unique to the archaeal enzymes, which are PestE, AFEST and Est1. It is worthy of note that Est1 was isolated from an environmental sample, but was proposed to originate from archaeal source [15]. As shown in Fig. 3B, the positions of arginine side chains in Sto-Est, Est1 AFEST and PestE are highly superimposable, suggesting that the side chain positioning is important.

Figure 3.

 Structural alignment. (A) DALI structural alignment at the region consisting of β7–α6–β8 of esterases from Sulfolobus tokodaii (Sto), Pyrobaculum calidifontis (PstE), environmental metagenomic samples (Est1, Est5, Est7), Archaeoglobus fulgidus (Afu), Alicyclobacillus acidocaldaricus (Aac), Salmonella typhimurium (Sty), Rhodococcus sp. (Rho), Bacillus subtilis (Bsu), Actinidia eriantha (Aer) and Oleispira antartica (Oan). The alignment order from top to bottom is based on the structural similarity of each protein to Sto-Est. Residues corresponding to R246, E250 and R267 in Sto-Est are indicated by numbers. X represents selenomethionine. (B) Superposition of R267 in Sto-Est on the structures indicated in the amino acid sequence alignment of Fig. 3A.

The structure of Sto-Est revealed that R267 is located at the dimer interface. Together with R246, R267 from two molecules form a basic pocket (Fig. 4), and have their guanidino nitrogen atoms located within 3.9–4.5 Å of one another. The positive charges of these guanidino groups are neutralized by the negatively charged carboxylic oxygen atoms of E250 and D247 that form salt bridges with R267 and R246, respectively. It is worthy of note that R246 and D247 are highly conserved in the other enzymes, even in those not from archaea, whereas E250 is also conserved in a psychrophilic enzyme from Oleispira antartica (Fig. 3A). R246 and D247 are part of a conserved motif in which the active-site aspartate residue (D243) is located.

Figure 4.

 The biological dimer of Sto-Est, as suggested by pisa analysis. The basic and acidic residues forming the hydrogen bond network close to the dimer interface are shown as stick models.

Construction and purification of R267 mutants

In order to examine the importance of R267 in conferring thermostability to the HSL-like archaeal esterases, R267 mutants of Sto-Est were constructed by overlap PCR and were overproduced and purified with the same method as that of Sto-Est. Because it is likely that R267 contributes to the stabilization of Sto-Est by forming an electrostatic interaction with E250, we constructed three R267 mutants in which R267 is replaced with Asp (acidic), Gly (neutral) or Lys (basic). The R267 mutants were overproduced and purified with the same method as that of Sto-Est. The overproduction level and purification yield of the R267 mutants are similar to those of Sto-Est. Gel filtration also shows that, like Sto-Est, R267 mutants form dimers in solution.

Urea-induced dimer dissociation

Urea-induced dimer dissociation was performed to analyze the difference in the stability of dimer assembly among the proteins. The protein samples were incubated with 0–6 m urea for 2.5 h, followed by urea PAGE. The relative amount of the protein existing in the dimer, monomer and denatured forms was calculated based on the protein-band intensity in the gel. As shown in Fig. 5A, Sto-Est and R267K have stronger dimer integrity than R267G and R267E.

Figure 5.

 (A) Dimer-to-monomer ratio of Sto-Est, R267K, R267E and R267G as examined by urea-induced denaturation. (B) Relative activity of Sto-Est and R267 mutants at 30 °C under different urea concentrations. (C) Relative activity of the Sto-Est and R267 mutants at 60 °C under different urea concentrations. (D) Optimum temperature for esterase activity of Sto-Est and R267 mutants. In (B) and (C), the activity of each enzyme without urea is defined as 100%. The point representations are as indicated in (A). In all cases, the values shown are averages of triplicate experiments and error bars represent the standard deviation.

Biochemical characterization

The enzymatic properties of Sto-Est and R267 mutants were examined. The optimum pH for activity of Sto-Est and R267 mutants is 6.0. The acyl-length selectivity of the proteins against p-nitrophenyl ester substrates is also similar, with the following order: C8 > C6 > C4 > C3 > C2. The optimum temperatures of R267E and R267G mutants are reduced relative to that of Sto-Est, whereas that of R267K is similar to that of Sto-Est (Fig. 5D).

We also examined the activity of Sto-Est and R267 mutants in the presence of 1–6 m urea against p-nitrophenyl butyrate at low (30 °C) or high (60 °C) temperatures (Fig. 5B,C). The activities are not severely affected by the addition of urea for all the samples examined, indicating that the protein exhibits activity even in the monomeric form.


The far-UV CD spectra and thermal denaturation curve of Sto-Est at different pH values were examined. The far-UV CD spectra at 25 °C are the same at different pH values, indicating that there are no significant structural differences (Fig. 6A). Sto-Est stability depends on pH and is at its maximum at neutral to alkaline pH (7–9) (Fig. 6B). Table 2 summarizes the fitted parameters of Sto-Est thermal denaturation. Dithiothreitol (1 mm) did not affect the thermostability at any pH.

Figure 6.

 (A) Far-UV CD spectra of Sto-Est at different pH values and in the presence or absence of 1 mm dithiothreitol. (B). Heat denaturation profile of Sto-Est. The colors used are as the same as Fig. 7A.

Table 2.   Apparent melting temperatures (Tm app.) of Sto-Est and its mutant proteins.
pH T m app./ΔTm app. (°C)

The far-UV CD spectra of R267 mutants are similar to that of Sto-Est at 25 °C at any pH, indicating that there are no significant structural differences (not shown). However, there are significant differences in stability, as indicated in the apparent Tm values (Table 2). At acidic pH (3 and 4), all R267 mutants have similar apparent Tm with that of Sto-Est. At higher pH, more significant changes are observed. At pH 7, the apparent Tm values for R267E and R267G are similar, whereas that of R267K is higher than those of the other R267 mutants, but lower than that of Sto-Est. All samples experienced irreversible heat-induced denaturation so that the reported apparent Tm should only be taken as a qualitative estimate of the thermostability of the protein.

Crystallization and structures of R267 mutants

R267 mutations cause significant changes in Sto-Est thermostability but do not change the overall structure of the protein. In order to analyze whether small local changes that are not detectable by CD spectroscopy occur by the mutations, we crystallized R267 mutants and determined their structures. R267 mutants were crystallized readily at the same condition as that of Sto-Est, and their structures were determined at 1.65–2.3 Å resolution (Table 3).

Table 3.   Data collection and refinement statistics. Numbers given in parentheses are for the highest resolution shell.
  1. R merge = ∑hkli |I(hkl) − <I(hkl)>|/∑hkliIi(hkl) where Ii(hkl) is the ith measurement of reflection hkl and <I(hkl)> is the weighted mean of all measurements of reflection hkl. Free R-value was calculated using 5% of reflections omitted from the refinement.

Data collection
 DetectorDIP-6040Quantum 210Quantum 210Quantum 210Quantum 210
 Space group P 21 P 21 P 21 P 21 P 21
Unit cell dimensions
 a (Å)76.3376.3276.2176.3776.34
 b (Å)114.75114.37114.84114.95114.61
 c (Å)102.21101.89102.22102.06102.20
 β (o)108.84108.44109.74109.55109.55
 Resolution range (Å)50.00–1.95 (1.98–1.95)50.00–1.91 (1.98–1.91)50.00–2.30 (2.38–2.30)50.00–1.65 (1.71–1.65)50.00–1.70 (1.76–1.70)
 Total observations1 503 408698 102971 1831 536 9101 678 681
 Unique reflections120 550123 52171 172198 205182 172
 Redundancy6.1 (5.9)3.1 (3.0)3.7 (3.7)3.8 (3.6)3.8 (3.8)
 Completeness98.9 (97.7)96.1 (96.9)98.9 (98.3)99.7 (98.9)99.9 (100.0)
 I/σ22.7 (3.1)23.2 (3.6)9.7 (2.8)25.7 (3.3)18.8 (2.2)
 Rmerge (%)8.5 (48.9)4.9 (31.6)15.7 (39.2)5.2 (38.1)7.4 (53.0)
Refinement statistics
 Resolution range (Å)44.28–1.95 (2.00–1.95)28.10–1.91 (1.96–1.91)32.46–2.30 (2.36–2.30)25.35–1.65 (1.69–1.65)26.62–1.70 (1.74–1.70)
 No. reflections113 863117 12767 472188 002172 304
 Rfactor (%)15.916.221.017.917.7
 Rfree (%)19.419.124.920.120.7
rmsd from ideal
 Bond lengths (Å)0.0110.0120.0070.0110.009
 Bond angles (°)
Ramachandran plot
 Most favored (%)94.895.693.294.694.5
 Allowed (%)
 Disallowed (%)00000
Heteroatoms10 MPD, 5 PO43−4 MPD, 4 PO43−10 MPD, 5 PO43−12 MPD, 5 PO43−15 MPD, 5 PO43−
No. of water molecules631664158755610
PDB accession code3AIK3AIL3AIM3AIN3AIO

The structures of R267 mutants are highly similar to that of Sto-Est, except at the side-chain residues of the mutation sites (Fig. 7). Like the noninhibited and DEP-bound structures, amino acid residues 1–20 are not visible in the structures of all R267 mutants. The main-chain rmsd values between the Sto-Est and R267 mutants are 0.38 for R267E, 0.34 for R267G and 0.35 Å for R267K. The crystallographic B-factor distributions of R267 mutants are also similar to that of Sto-Est (not shown). Furthermore, there are no significant differences in the Φ/Ψ values between the structures of the R267 mutants and Sto-Est, as shown by the Kleywegt plots (not shown) [16]. The ΔGdiss values of the biological dimers of the R267 mutants are highly similar to that of Sto-Est, ranging from 11.0 to 11.6 kJ·mol−1. Overall, there are no significant structural changes that could be observed between Sto-Est and R267 mutants.

Figure 7.

 The structures around the mutation sites of R267E (A), R267G (B) and R267K (C). The electron-density maps of the mutated residues are shown.

Looking at the structures around the mutation sites, we observed that in the R267E structure, E267 is located very close to E250, with the distances between Oε2 of E267 and Oε1 or Oε2 of E250 between 3.5 and 3.6 Å or 4.3 and 4.7 Å, respectively, in the four molecules (Fig. 8A). The R267G structure shows that the amino acid residues around the mutation site in each monomer assume similar conformations to those in Sto-Est, even in the absence of the arginine side chain (Fig. 8B). The R267K structure shows that K267 forms long-range electrostatic interaction with E250 (Fig. 7C), with an average distance between the Nζ atom of K267 and Oε1 or Oε2 of E250 of 5.4 Å, as expected from the thermostability of R267K which is higher than those of R267E and R267G (Table 2).

Figure 8.

 The rmsd of atomic coordinates. (A) The rmsd of the Cα positions of Sto-Est, R267E, R267G and R267K during 10-ns simulation at 500 K (thick line) or 310 K (thin line), following a least square fit against the Cα atoms as a reference. (B) The rmsd of the monomer atoms of each chain (chains B and C of the crystal structure) during 10-ns simulation at 500 K (thick line) or 310 K (thin line) of Sto-Est, R267E, R267G and R267K.

MD simulation

MD simulation was performed in order to gain further understanding of the molecular mechanism behind the thermal destabilization of the dimer integrity by R267 mutations. Chains B and C (the biological dimer) of the crystal structures of Sto-Est (noninhibited form), R267E, R267G and R267K were used as the starting structures. MD simulation was done at high (500 K) and normal (310 K) temperatures [17,18].

The rmsd values of the entire Cα atoms (dimeric structures) during simulation at 500 K suggest that R267E and R267G fluctuated significantly higher than R267K or Sto-Est (Fig. 8A). The fluctuation is less significant at the whole-atom rmsd of the monomers alone (Fig. 8B). We then measured the distance between the center of mass (CoM) of the dimer interface (residue 263–270 of each monomer) and found that the monomers of R267E and R267G are more separated than those of Sto-Est and R267K (Table 4). The aforementioned phenomena are much less intense at normal temperature (310 K). Altogether, the data show that the dimer stabilities of R267E and R267G are lower than those of Sto-Est and R267K at high temperature.

Table 4.   Distances between atoms or groups of atom during simulation.
Residue pair (atom or group of atoms)Initial distance (Å)Average distance during simulation (Å)
310 K500 K
Dimer interface (CoM of residues no. 263–270 in chains B and C) of:
 Sto-Est4.24.6 ± 0.25.5 ± 0.8
 R267E4.24.4 ± 0.215.0 ± 3.1
 R267G4.24.3 ± 0.213.0 ± 4.2
 R267K4.24.4 ± 0.25.7 ± 0.6
R267 chain B–R267 chain C (CoM guanidines) in Sto-Est4.16.0 ± 0.97.9 ± 1.4
E267 chain B–E267 chain C (CoM Oε atoms) in R267E5.56.7 ± 0.814.1 ± 3.4
K267 chain B–K267 chain C (Nζ atoms) in R267K4.76.4 ± 1.37.1 ± 1.6
R267–E250 (CoM guanidine and CoM OE atoms) in Sto-Est
 Chain B3.74.1 ± 0.37.1 ± 2.9
 Chain C3.53.7 ± 0.213.2 ± 6.7
E267–E250 (CoM Oε atoms) in R267E
 Chain B5.46.5 ± 1.112.0 ± 4.0
 Chain C5.46.4 ± 0.921.9 ± 7.0
K267–E250 (Nζ atom and CoM OE atoms) in R267K
 Chain B5.45.8 ± 1.77.6 ± 3.4
 Chain C5.45.7 ± 1.912.3 ± 5.8

We also measured the distances between the CoM of guanidine groups of R267, CoM of Oε atoms of E267, and Nζ of K267 in Sto-Est, R267E and R267K proteins, respectively. As shown in Table 4, the measured distances equilibrated similarly at 310-K simulation (around 6.0–6.7 Å). These distances are higher than the initial crystallographic distances (4.1–5.5 Å). At 500-K simulation, the CoM distance of Oε atoms of E267 is significantly higher than that of the CoM of guanidine groups of R267 or and Nζ of K267 in Sto-Est (14.1 Å compared with 7.9 or 7.1 Å, respectively).

The distances between the ionizable atoms of residues 267 and 250 that form salt bridges in Sto-Est were also measured. During 310-K simulation, the distances remained relatively similar with the crystallographic distances. This is not the case with 500-K simulation, in which the distances varied considerably, with higher values in one of the monomers.


Here, we report the structure of a thermostable esterase from the thermoacidophilic archaeon, Sulfolobus tokodaii (Sto-Est). The structure shows a typical α/β hydrolase fold, with the core consisting of a α/β sheet of eight β sheets connected by α helices [17–19]. The amino acid sequence of Sto-Est reveals the presence of HSL-like motif ([L/I/V/M/F]-[L/I/V/M/F]-x-[L/I/V/M/F]-H-G-G-[S/A/G]-[F/Y/W]-x-x-x-[S/T/D/N]-x-x-[S/T]-H, indicating that it is an HSL-like carboxylesterase [20]. Sto-Est is dimeric in solution, as indicated by its gel-filtration profile. A homologous esterase, Est1 that was isolated from a metagenomic library, was reported to be dimeric in solution, with its dimer assembly mode experimentally proven and similar to that of Sto-Est [15]. Similarly, a homologous archaeal esterase PestE was also reported to be dimeric in solution [7]. By contrast, other homologous thermostable carboxylesterases AFEST (archaeal esterase) and EST2 (bacterial esterase), characterized by the same group, were reported to be monomeric in solution, as examined by gel filtration [23–25]. Similarly, homologous mesophilic brefeldin A esterase from mesophilic B. subtilis (Bs-Est) was also reported to be monomeric in solution [26,27]. Although different in solution oligomeric states, the crystal structures of all enzymes are highly similar. It is worthy of note that the optimum temperature for the activity of Sto-Est (75 °C) is similar to those of AFEST (80 °C) and EST2 (70 °C) and lower than that of EstE1 (95 °C) [13,14,23,24].

There have been attempts to understand the stabilization mechanism of thermostable HSL-like carboxylesterases. Structural comparison between AFEST and EST2 suggested that shortening of the loops in AFEST contributes significantly to its higher stability compared with of EST2 [6]. This conclusion, however, was drawn only from structure comparison without verification by mutagenesis. The hyperthermostable EstE1 has also been one of the targets for such study. From mutational and biophysical studies, it was proposed that the main factor that contributes to EstE1 hyperthermostability is the strong hydrophobic interaction in the dimer interface, formed by V274, F276 and L299 from each monomer [14]. These residues are conserved in Sto-Est, corresponding to V266, F268 and L291 and are not conserved in AFEST or EST2, corresponding to V278, Y280 and Q303. It was shown that the V274/F276A double-mutant protein of EstE1 exists as a monomer and has its apparent Tm reduced by 20 °C, indicating that the dimerization is important for its stability. Because these residues are conserved in Sto-Est, it is highly likely that these residues also maintain a stable Sto-Est dimer and contribute to the stabilization of Sto-Est.

We also determined the structure of Sto-Est in complex with DEP and determined the residues forming the oxyanion hole (Fig. 2D). These residues (G79, G80 and A151) are shared by the other homologous esterases, determined in complex with amphipiles that resemble the similar tetrahedral intermediate structure, such as EST2 (G82, G84 and A156) [24] and AFEST (G88, G89 and Ala151) [6]. Amino acid sequence alignment shows that the first two residues are highly conserved in almost all of the HSL-like carboxylesterases and are located in the HSL signature motif, whereas the third residue is also conserved, located several residues after the nucleophilic serine (Fig. 1), indicating that the same catalytic mechanism is shared by these esterases.

Amino acid and DALI structure alignments of several homologous HSL-like carboxylesterases from archaeal and bacterial sources show that R267 and E250 are unique to archaeal enzymes. These residues might contribute to the thermal stabilization of archaeal enzymes or at least of Sto-Est. Interestingly, R267 of each monomer is located close to the dimer interface in the Sto-Est structure (Fig. 4). R267 is also located close to R246. Such close proximity of positively charged amino acid residues would be highly disadvantageous for protein structural integrity if they are not neutralized by electrostatic interactions with negatively charged amino acids. In the case of Sto-Est, E250 neutralizes R267 and D247 neutralizes R246 (Fig. 4). Our study suggests that R267 is important for the dimer integrity of Sto-Est that contributes to its overall stability. This seems not to be the case for AFEST, which exists as a monomer in solution, possesses an R267 equivalent and is more stable than Sto-Est [6,23]. Further study is required to understand the role of this residue on the other archaeal esterases.

R267 mutants, R267E, R267G and R267K, were generated to examine the role of R267 in the stabilization of archaeal enzymes. The R267 mutants have similar or lower optimum temperature and apparent Tm values compared with Sto-Est, especially at pH 5–9. At lower pH values (3 and 4), the differences in apparent Tm are lower, emphasizing the fact that the reduced apparent Tm values are due to disruption of the H-bond or ionic interaction. Among R267 mutants, R267E is the most unstable, followed by R267G. R267K has similar thermal and chemical stability to Sto-Est.

The crystal structures of the R267 mutants are similar to that of Sto-Est, even around the position of the mutation site. The R267E structure shows that although the closely located negatively charged Oε atoms of E267 in molecule B and C or A and D should generate strong electrostatic repulsion forces at neutral to basic pH in which the carboxyl oxygen atoms are deprotonated, they are still located close to each other. It is likely that the four intermolecular MC–MC H-bond between β8 strands of the dimer is strong enough to overcome the electrostatic repulsion of the deprotonated carboxyl oxygen atoms. The structure of R267G shows that the glutamate residues around the dimer interface are at the lowest energy conformations because they are not changed even with the extra space provided by the absence of the side chain of residue 267. The structure of R267K shows that the Nζ atoms of K267 in every protein molecule form long-distance ionic interactions with the Oε atoms of E250 in the same molecules and thus may contribute to the forces that stabilize the protein, although with lower efficiency than R267 of Sto-Est which forms two H-bonds with E250.

A urea-induced dimer-dissociation experiment showed that R267 mutation did not result in significant destabilization of the monomer structures. The activities of the proteins are also not affected by the addition of 1–6 m urea, although R267E and R267G are mostly in the monomeric forms in the presence of 4–6 m urea. This result indicates that the activity of Sto-Est is not affected by its oligomerization state. Dimer destabilization was observed in the mutation of R267 with Glu and Gly (R267E an R267G, respectively). MD simulation results also suggest that R267 does not cause stabilization of the monomeric structure, because there are no significant differences in the monomer rmsd of Sto-Est and R267 mutants during 500-K simulation. This may be behind the observation that the effects of R267G and R267E mutations on the optimum temperature and apparent Tm of the mutant proteins are less severe than the effects on dimer integrity. Taken together, the data suggest that the crucial role of R267 is to stabilize the dimer integrity of Sto-Est at high temperature and, to a certain extent, the dimer integrity plays a role in the overal stability of the protein.

Materials and methods

Gene cloning

The gene encoding Sto-Est was amplified by PCR using primer 1 (5′-GGAATTCATATGATAGACCCTAAAATTAAA-3′) and primer 2 (5′-GGTCGACTTATTTTCCGTAAAATACTTTTC-3′), the EcoRI site for primer 1 and SalI site for primer 2 underlined and the NdeI site for primer 1 is given in italics. The genomic DNA of S. tokodaii 7 (provided by A. Yamagishi, Tokyo University of Pharmacy and Life Science, Japan), which was prepared from a sarkosyl lysate as described previously [28] was used as a template. PCR was performed in 25 cycles using a thermal cycler (Gene Amp PCR System 2400; Perkin–Elmer Foster City, CA, USA) and KOD Plus DNA polymerase (Toyobo, Osaka, Japan). The resulting DNA fragment was digested with EcoRI and SalI and ligated to the EcoRI–SalI sites of plasmid pBlueScript II SK(+) (Toyobo) for DNA sequencing. For protein overproduction under T7 promoter, the pBlueScript II SK(+)–Sto-Est was digested with NdeI and SalI and the Sto-Est gene was ligated to the corresponding restriction sites of pET 28a+ (Novagen, Billerica, MA, USA). The nucleotide sequence was confirmed with an ABI PRISM 310 Genetic Analyzer (Perkin–Elmer) and there was no difference found between the cloned sequence and the sequence in the database (Genbank: BAB65028.1). All DNA oligomers for PCR were synthesized by Hokkaido System Science (Sapporo, Japan).


The genes encoding R267E, R267G and R267K were constructed by overlap PCR using pET28a+–Sto-Est as a template. The mutations were confirmed by DNA sequencing. Overproduction and purification of R267 mutants were carried out using a similar protocol to that used for Sto-Est.

Overproduction and purification

Overproduction of Sto-Est was carried out by growing E. coli BL21 (DE3) CodonPlus, previously transformed with pET28a+–Sto-Est, in NZCYM medium containing 50 mg·L−1 kanamycin and 30 mg·L−1 chloramphenicol at 37 °C until A660 reached 0.5. Induction was done by adding 1 mm isopropyl thio-β-d-galactoside, followed by overnight incubation at 25 °C. E. coli cells were collected by centrifugation, resuspended in 20 mm Tris/HCl pH 9.0 containing 5 mm imidazole and 150 mm NaCl, lysed by French press and the soluble fraction was collected and applied to HiTrap Chelating column (GE Healthcare, Piscataway, NJ, USA). Elution was done by linear gradient of imidazole to 500 mm and the protein was eluted at imidazole concentration of 150–250 mm. The collected fractions were then applied to gel filtration, using HiLoad 16/60 Superdex 200 column (GE Healthcare), equilibrated with 5 mm Tris/HCl pH 9.0 containing 50 mm NaCl. The fractions containing Sto-Est were collected and kept at −30 °C until used.

Activity assay

Esterase activity was examined as previously described [29] using p-nitrophenyl butyrate (0.5 mm) as the substrate in a reaction mixture containing 10 mm buffer pH 6.0 and 10% acetonitrile. The amount of enzyme used was 70 ng. The reaction was terminated by the addition of 1% SDS. The amount of product (p-nitrophenol) was measured using a spectrophotometer at 348 nm and a molar absorbance of 5400 m−1·cm−1. One unit of activity is defined as the amount of enzyme that catalyzes the production of 1 μmol product per minute at given conditions. The activity assay under denaturating condition was performed in the presence of 1–6 m urea.

The optimum pH for activity was determined using sodium acetate (pH 4.0, 4.5, 5.0 and 5.5), sodium phosphate (pH 5.5, 6.0, 6.5 and 7.0), Tris/HCl (pH 7.0, 7.5, 8.0, 8.5) and Gly–NaOH (pH 8.5, 9.0, 9.5). Sto-Est exhibited the highest activity at pH 6.0. The optimum temperature for activity was determined by incubating the reaction mixture at various temperatures. Substrate selectivity was examined using various acyl lengths of p-nitrophenyl esters.

CD spectroscopy

CD spectroscopy was carried out using a J-725 automatic spectropolarimeter (Japan Spectroscopic, Tokyo, Japan) with a 2-mm cuvette at 25 °C. The protein concentration used was 0.12 mg·mL−1. Mean residue ellipticity (θ), with units of deg cm2·dmol−1, was calculated using an average amino acid relative molecular mass of 110. Thermal denaturation was performed by monitoring the CD value at 222 nm during temperature ramping from 25 to 95 °C with a 1 °C·min−1 scanning rate. The buffers (10 mm) used were Gly–HCl pH 3.0, sodium acetate pH 4.0 and 5.0, sodium phosphate pH 6.0 and 7.0, and Gly–NaOH buffer pH 9.0. Dithiothreitol (1 mm) was added when indicated.

Crystallization and data collection

Initial crystallization screening was done using Crystal Screen kits (Hampton Research, Bedford, MA, USA) by sitting-drop vapor diffusion technique in CombiClover plates (Emerald Biosystems). Sto-Est crystals were found in Crystal Screen II No. 43, composed of 0.1 m Tris/HCl pH 8.5 containing 0.2 m ammonium phosphate monobasic and 50% MPD. The crystals were used without further optimization. X-Ray diffraction datasets were collected at the BL44XU or BL38B1 beamline in SPring-8 (Hyogo, Japan). Crystals were mounted directly on a cryoloop (Hampton Research, Aliso Viejo, CA, USA) without any additional cryo protection. Data collection was performed at 100 K under a nitrogen stream using X-rays with wavelength of 1.0 Å. The detector used was a DIP-6040 image plate detector (Bruker-AXS, Madison, WI, USA) or Quantum 210 CCD detector (ADSC, Poway, CA, USA). Data indexing, integration and scaling were peformed using denzo and scalepack programs as standalones or in the hkl2000 suite [30]. Data collection and refinement statistics are shown in Table 3.

Structure determination and analysis

The Sto-Est crystals belong to space group P21 with four molecules per asymmetric unit. The Sto-Est structure was solved by molecular replacement method using Molrep [31] in the ccp4 crystallographic suite [32] using a monomer of the crystal structure of AFEST (PDB: 1JJI) as template. The resultant Sto-Est structure was then used to solve the other structures. Several cycles of refinement were carried out using refmac [33] and coot [34]. Progress in the structural refinement was evaluated by the free R-factor and by inspection of stereochemical parameters using rampage [35]. The structural figures were prepared using pymol (

The substrate entrance/product exit tunnel was identified using the program caver [10]. Analyses of the molecular interfaces were performed using pisa [36] and pic [37]. The rmsd values between the structures were calculated using lsqkab [32].

MD simulation

MD simulation was done using gromacs v. 3.3.3 [38] utilizing the GROMOS G53a6 force field on a DELL DPeR900 server (HPC Technologies) running on CentOS4. The crystal structures consisting the biological dimer of noninhibited Sto-Est, R267E, R267G and R267K were used as starting structures. The protein molecule is placed in a cubic simulation box with the distance between the protein outer layer and simulation box set at 1.0 nm. SPC water molecules were added together with 6–12 Na+ molecules to neutralize the system. The simulation systems consisted of ∼ 97 000 atoms each.

Energy minimization was carried out using steepest descents method such that the maximum of force on any atom (Fmax) was < 1000 kJ, followed by position restrained simulation at 310 or 500 K for 100 ps to equilibrate water molecules. MD simulations were conducted under constant pressure and temperature at 310 and 500K (temperature-coupling constant 0.1 ps) and 1 bar with the isotropic Parrinello–Rahman pressure-coupling scheme [39] with pressure-coupling time constant of 0.5 ps. A 2-fs time step was used throughout, and the van der Waal’s radius and Coulomb interactions were computed within cut-offs of 14 and 10 Å, respectively. MD simulation was performed for 10 ns. The simulation trajectories were analyzed using several auxiliary programs provided with the gromacs package, using g_rms and g_dist utilities to compute the rmsd of atomic positions and atom–atom distances, respectively. The rmsd calculation was performed by superposing the Cα atoms of the dimer or the entire atoms of the monomer.

Urea-induced dimer dissociation

Urea-induced dimer dissociation was carried out by incubating the proteins in 10 mm Tris/HCl pH 8.0 containing 0, 1, 2, 3, 4, 5, 6, 7 or 8 m urea for 2.5 h, followed by native PAGE containing the corresponding concentrations of urea. The gel was then stained with Coomassie Brilliant Blue and the intensities of the bands representing the dimeric, monomeric and denatured states of the samples were estimated using imagej [40]. The dimer ratio was calculated with the following equation:



The synchrotron radiation experiments were performed at the BL44XU and BL38B1 in SPring-8 with the approval of the Japan Synchrotron Radiation Research Institute (Proposal No. 2009B6915, 2010A1158 and 2009B1159). We thank Dr A. Yamagishi for providing S. tokodaii 7 genomic DNA. This work was supported in part by a grant (21380065) from the Ministry of Education, Culture, Sports, Science, and Technology of Japan and by an Industrial Technology Research Grant Program from the New Energy and Industrial Technology Development Organization of Japan.