Structure, biochemical characterization and analysis of the pleomorphism of carboxylesterase Cest-2923 from Lactobacillus plantarum WCFS1


  • Rocío Benavente,

    1. Department of Crystallography and Structural Biology, Institute of Physical Chemistry Rocasolano, CSIC, Madrid, Spain
    Search for more papers by this author
  • María Esteban-Torres,

    1. Laboratory of Bacterial Biotechnology, Institute of Food Science and Technology and Nutrition (ICTAN), CSIC, Madrid, Spain
    Search for more papers by this author
  • Iván Acebrón,

    1. Department of Crystallography and Structural Biology, Institute of Physical Chemistry Rocasolano, CSIC, Madrid, Spain
    Search for more papers by this author
  • Blanca de las Rivas,

    1. Laboratory of Bacterial Biotechnology, Institute of Food Science and Technology and Nutrition (ICTAN), CSIC, Madrid, Spain
    Search for more papers by this author
  • Rosario Muñoz,

    1. Laboratory of Bacterial Biotechnology, Institute of Food Science and Technology and Nutrition (ICTAN), CSIC, Madrid, Spain
    Search for more papers by this author
  • Yanaisis Álvarez,

    1. Department of Crystallography and Structural Biology, Institute of Physical Chemistry Rocasolano, CSIC, Madrid, Spain
    Search for more papers by this author
  • José M. Mancheño

    Corresponding author
    1. Department of Crystallography and Structural Biology, Institute of Physical Chemistry Rocasolano, CSIC, Madrid, Spain
    • Correspondence

      J. M. Mancheño, Department of Crystallography and Structural Biology, Institute of Physical Chemistry Rocasolano, CSIC, Serrano 119, E-28006 Madrid, Spain

      Fax: +34 564 2431

      Tel: +34 917 459 547


    Search for more papers by this author


The hydrolase fold is one of the most versatile structures in the protein realm according to the diversity of sequences adopting such a three-dimensional architecture. In the present study, we clarified the crystal structure of the carboxylesterase Cest-2923 from the lactic acid bacterium Lactobacillus plantarum WCFS1 refined to 2.1 Å resolution, determined its main biochemical characteristics and also carried out an analysis of its associative behaviour in solution. We found that the versatility of a canonical α/β hydrolase fold, the basic framework of the crystal structure of Cest-2923, also extends to its oligomeric behaviour in solution. Thus, we discovered that Cest-2923 exhibits a pH-dependent pleomorphic behaviour in solution involving monomers, canonical dimers and tetramers. Although, at neutral pH, the system is mainly shifted to dimeric species, under acidic conditions, tetrameric species predominate. Despite these tetramers resulting from the association of canonical dimers, as is commonly found in many other carboxylesterases from the hormone-sensitive lipase family, they can be defined as ‘noncanonical’ because they represent a different association mode. We identified this same type of tetramer in the closest relative of Cest-2923 that has been structurally characterized: the sugar hydrolase YeeB from Lactococcus lactis. The observed associative behaviour is consistent with the different crystallographic results for Cest-2923 from structural genomics consortia. Finally, the presence of sulfate or acetate molecules (depending on the crystal form analysed) in the close vicinity of the nucleophile Ser116 allows us to identify interactions with the putative oxyanion hole and deduce the existence of hydrolytic activity within Cest-2923 crystals.

Structured digital abstract

Cest-2923 and Cest-2923 bind by x-ray crystallography (1, 2)

Cest-2923 and Cest-2923 bind by cosedimentation in solution (1, 2)


The atomic coordinates and structure factors have been deposited in the Protein Data Bank with accession numbers: 4BZW for Cest-2923 from native crystals not soaked with substrates (P6322 space group); 4C01 for Cest-2923 from crystals soaked with phenyl acetate (C2 space group); 4BZZ for Cest-2923 from crystals soaked with isopropenyl acetate (P622 space group).


carboxylesterase from Lactobacillus plantarum coded by gene lp_2923


European Synchrotron Radiation Facility


hormone-sensitive lipase


Joint Center for Structural Genomics


lactic acid bacteria


North East Structural Genomics Consortium


Protein Data Bank


p-nitrophenyl acetate


p-nitrophenyl benzoate


Carboxylesterases (EC comprise widely distributed enzymes in animals, plants and microorganisms [1] that hydrolyse (synthesize) carboxylic esters. Specifically, the term carboxylesterase is adopted for enzymes acting on soluble, monomeric (e.g. below their critical micellar concentration) substrates, in contrast to lipases (EC, which display maximal activity against water-insoluble substrates [2, 3]. Based on their amino acid sequences and particularly in the presence of specific sequence motifs, esterases and lipases are classified into four blocks: C, H, L and X [4]. Block H includes the plant carboxylesterase and hormone-sensitive lipase (HSL) families. The members of this latter family are distributed over all kingdoms of life and share sequence similarity to mammalian HSL [5]. Family IV of a previous classification of bacterial lipolytic enzymes [3] corresponds to the bacterial members of the HSL family.

Structurally, carboxylesterases belong to the α/β hydrolase superfamily of enzymes [6-8]. The crystal structures currently determined for members of this superfamily [524 Protein Data Bank (PDB) entries using the term α/β hydrolase fold from the Structural Classification of Proteins;] reveal a highly conserved protein architecture, the α/β hydrolase fold, a central, mostly parallel β-sheet, surrounded on both sides by α-helices. In all cases, the catalytic machinery is based on a highly conserved catalytic triad made up of a nucleophile (serine), an acid (Asp/Glu) and a histidine. The nucleophile, which is located in the so-called nucleophile elbow [8], as well as the ‘oxyanion hole’ involved in the stabilization of the negatively-charged tetrahedral intermediates, comprise the hallmarks of the α/β hydrolases.

One remarkable feature of the α/β hydrolase fold is its structural versatility: on the one hand, sequences with a very low level of similarity (< 20%) essentially adopt the same structural core and, on the other hand, it means that large domains can be added to this fold, which remains as an easily identifiable, conserved core. Usually, these new domains are inserted between the canonical strands β6 and β7 and are assumed to modulate specific properties of the enzyme [7, 8].

Although lactic acid bacteria (LAB) play an important role in the production of fermented food products [9], little information exists concerning LAB as a source of enzymes [10]. In this regard, the recently reported genome of Lactobacillus plantarum WCFS1 [11] reveals the existence of numerous ORFs coding putative esterases, among them the gene coding for the putative carboxylesterase Cest-2923 (lp_2923). The enzyme was recently crystallized by two independent structural genomics consortia [Joint Center for Structural Genomics (JCSG) and North East Structural Genomics Consortium (NESG)] reporting incomplete models for the enzyme (PDB code: 3bxp for JCSG and 3d3n for NESG), which as revealed in the present study, show features in conflict with the presence of the canonical catalytic triad.

In the present study, we report the complete crystal structure of Cest-2923, which, as expected from sequence comparisons and the structural data from the above structural genomics consortia, exhibits a canonical α/β hydrolase fold. In addition, our biophysical studies reveal that the enzyme shows complex pH-dependent pleomorphic behaviour in solution. Finally, we determined the main biochemical features of Cest-2923, which demonstrate that the enzyme is a true carboxylesterase, as expected from the genome annotation.

Results and Discussion

General comments about search models and description of the monomer structure

N-terminally His-tagged Cest-2923, when overexpressed using the pURI3-TEV vector as in our previous structural studies [12], was revealed to be marginally stable in solution and therefore not suitable for crystallization experiments. By contrast, the C-terminally His-tagged variant behaved well in solution and was suitable for crystallization. The enzyme crystallized in the hexagonal space group P6322 (unit-cell dimensions a = b = 141.65 Å, c = 165.74 Å, α = β = 90º, γ = 120º) with two independent protein molecules per asymmetric unit. Diffraction data up to 2.1 Å resolution were measured at beamline ID29 at the European Synchrotron Radiation Facility (ESRF) (Grenoble, France) on a Pilatus 6M detector (Dectris Ltd, Baden, Switzerland) and the structure of the esterase was solved by molecular replacement. At that time, there were two sets of atomic coordinates for Cest-2923 in the PDB, one deposited by the NESG (PDB code: 3d3n) and the other one by the JCSG (PDB code: 3bxp). Previous inspection of these models revealed two aspects: first, the two independent molecules of the corresponding asymmetric units were incomplete, lacking a catalytically relevant region around residues Gly231 to Tyr250 and, second, most remarkably, they exhibited structural features incompatible with the presence of the catalytic triad typical of an α/β hydrolase fold, which is the basic architecture of this enzyme. These latter anomalous features are considered in detail below. Only the shortest deposited model for Cest-2923 (molecule A from the PDB entry 3d3n) is devoid of these anomalous structural features and therefore was used as search model. This model lacks the 25-residue sequence stretch between residues Gly231 and Ala255, in conflict with the classic definition of the catalytic triad because it contains the catalytic residue His233 (see below), plus a three-residue loop (Phe28-Thr30).

In this regard, the residues forming the catalytic triad of Cest-2923 (Ser116, His233 and Asp201) were predicted in a straightforward manner from a simple blast search against the PDB, which in turn revealed the putative sugar hydrolase YeeB from Lactococcus lactis (PDB entry 3hxk) and the carboxylesterase lp_1002 (PDB entry 3bjr) as the closest homologues (33% and 32% sequence identity, respectively) to be structurally characterized (Fig. 1).

Figure 1.

Amino acid sequence alignment of Cest-2923 with its closest homologuess. Cest-2923 shares 33% and 32% sequence identity with the putative sugar hydrolase YeeB from L. lactis and the carboxylesterase lp_1002 from L. plantarum, respectively. Colour code: blue, conserved amino acid residues; yellow, similar residues; red, residues forming the catalytic triad in YeeB and lp_1002.

The two Cest-2923 independent molecules making up the asymmetric unit of our hexagonal crystal are essentially identical (0.36 Å rmsd for 275 Cα atoms). We arbitrarily chose chain A as reference for description of the protein. As indicated above, the structure of the Cest-2923 subunit is based on a canonical α/β hydrolase fold [6-8] consisting of a central eight-stranded β-sheet, with strand β2 being the only antiparallel strand (Fig. 2). The β-sheet shows a marked left-handed twist with an approximate angle of 90º between strands β1 and β8. This core is surrounded by five helices, with α1 (Ala54-Met61) and α7 (Trp258-Glu268) lying on one side of the sheet and α2 (Trp84-His103), α3 (Ala117-Val128) and α6 (Ile207-Gln218) on the opposite side. Although Cest-2923 is currently classified as member of the HSL subfamily of α/βhydrolases [13] by the ESTHER database, the enzyme does not exhibit a distinct cap domain over the active site. Curiously, the loop between strand β8 (Thr223-Phe228) and helix α7 is highly flexible as deduced from the above mentioned models from the JCSG and NESG consortia for the enzyme where this region cannot be modelled, and may perhaps function as a pseudo-flap.

Figure 2.

Overall fold of the Cest-2923 subunit from L. plantarum WCFS1. The tertiary structure of the Cest-2923 subunit fits the canonical α/β hydrolase fold where a central eight-stranded β-sheet (shown in green) is surrounded by five α-helices (orange). Two orthogonal views of the protein subunit are shown.

Comparison with other esterases

A dali search [14] revealed numerous structures with high structural similarity, as otherwise expected as a result of the high level of conservation of the α/β hydrolase fold [6-8]. Sugar hydrolase YeeB from L. lactis (PDB entry: 3hxk) and carboxylesterase lp_1002 from L. plantarum WCFS1 (PDB entry: 3bjr) have the highest structural similarity (Z-score = 34.7 and 28.8; rmsd = 1.5 and 2.0 Å for 250 and 244 Cα atoms, respectively). Other esterases from the HSL subfamily such as Est1 from environmental samples [15], Sto-Est from Sulfolobus tokodaii [16], PestE from Pyrobaculum calidifontis [17] and acetyl esterase from Salmonella typhimurium (PDB entry: 3ga7) are also structurally similar (Z-score = 25.4, 24.8, 24.6 and 24.5; rmsd = 2.5, 2.6, 2.5 and 2.6 Å for 236, 236, 234 and 240 Cα atoms, respectively). In these latter cases, the structural similarity is limited to the α/β hydrolase fold (core β-sheet plus the two surrounding layers of α-helices) because, as indicated above, Cest-2923 lacks a cap domain characteristic of some HSL enzymes.

Definition of the active site of Cest-2923: the catalytic triad

As a member of the α/β hydrolase superfamily, the catalytic machinery of Cest-2923 is based on a catalytic triad formed by Ser, His and Asp/Glu residues (Ser116, His233 and Asp201), which are located at canonical sites along the protein sequence. The nucleophile Ser116 of Cest-2923 is situated in the so-called nucleophilic elbow, within a highly conserved sequence (Gly-X-Ser-X-Gly) (with X denoting any amino acid) [13], with its backbone angles residing in an unfavourable region of the Ramachandran plot (φ = 52º, ψ = −118º). This constrained conformation is stabilized by a dense network of hydrogen bonds affecting both the polypeptide chain (two 2.9 Å hydrogen bonds between the carbonyl oxygen atom and the amide nitrogen atoms of Gly119 and Val156, respectively, and a 3.0 Å hydrogen bond between the amide nitrogen atom of Ser116 and the carbonyl oxygen of Gly153) and the hydroxyl side chain (a hydrogen bond (2.8 Å) between the Oγ atom and the Nε2 atom of His233) (Fig. 3A).

Figure 3.

Catalytic triad machinery of the Cest-2923 subunit from L. plantarum WCFS1. (A) Stereo view of the catalytic triad environment of Cest-2923. Residues of the catalytic triad (Ser116, His233, Asp201) are shown as orange sticks and those participating in hydrogen bonding interactions as grey sticks; potential H-bonds are indicated as black, broken lines. A conserved water molecule that interacts with Asp201, which is conserved within the HSL family of enzymes, is shown as a blue sphere. (B) Stereo view of the surroundings of the sulfate molecule found in the active site of Cest-2923. The electron density map (2Fo – Fc in blue; contoured at 1-σ level) for the sulfate is shown. Residues of the catalytic triad are shown as light yellow sticks. Distances are given in Å.

The proton carrier residues His233 and Asp201 of the charge-relay system locate downstream the strands β8 and β7, respectively. Particularly, the amino acid His233 is located within the 29-residue long loop connecting strand β8 to helix α7. The Nδ1 atom of His233 is hydrogen-bonded to the Oδ1 atom (2.7 Å) and to the Oδ2 atom (3.1 Å) of Asp201. The conformation of this last side chain is further stabilized by a hydrogen bond between the Oδ2 atom and the NH of Val204 (2.9 Å) and is also at hydrogen bond distance to the CO of this latter residue (3.1 Å). In addition, the Oδ1 atom is hydrogen-bonded to the Nε2 atom of Gln197 (3.0 Å) and also to a highly ordered water molecule (2.7 Å). This solvent molecule together with its close interacting neighbour (2.9 Å distance between them) are present in all structural homologues of Cest-2923 found with dali and have been proposed to play a structural role in this family of enzymes [18]. In particular, in Cest-2923 these solvent molecules may contribute to maintain the proper conformation of this part of the active site by bridging the loops where His233 and Asp201 are situated. Thus, molecule HOH45 hydrogen bonds to the NH atom (3.0 Å) and to the Oγ1 atom (2.8 Å) of Thr198 from the loop connecting strand β7 to helix α6, and HOH4 makes hydrogen bonds to the NH atom (3.1 Å) of Leu235 and to the CO atom (2.8 Å) of Ile232 from the loop between strand β8 and helix α7.

Interestingly, a spherical blob of electron density appeared in close contact to the hydroxyl group of the nucleophile Ser116 of both independent molecules that has been assigned to a sulfate molecule coming from the crystallization solution (Fig. 3B). This molecule may be claimed to mimic the negatively-charged tetrahedral intermediates of the catalytic reaction, which are stabilized by the so-called oxyanion hole [7]. In this case, the sulfate molecule is stabilized by polar interactions with amide nitrogen atoms of Gly43, Gly44 and Ala117, which would suggest the existence of a tridentate oxyanion hole similar to those from heroin esterase [19], esterase EstE1 [15] or carboxylesterase EST2 from Alicyclobacillus acidocaldarius [18], in contrast to the most commonly observed bidentate counterparts [7].

As indicated above, the atomic models of Cest-2923 deposited by the JCSG and NESG consortia (PDB code: 3d3n and 3bxp, respectively), exhibited structural features that are incompatible with the presence of a catalytic triad (Fig. S1). Specifically, in molecule A from PDB entry 3d3n, which is the shortest model, the Asp201 side chain location overlaps with that of His233 (in our model), which, in this model, is not defined (Fig. S1A). Conversely, the trace of the loop that contains the catalytic residue His233 in molecule B places this residue at a distance of 15 Å from the nucleophile Ser116 (distance between Cα atoms), a geometry incompatible with the built-up of a catalytic triad (Fig. S1B). The conformation of the connecting loop between strand β7 and helix α6 almost coincides with our model, placing the catalytic Asp201 in a similar position. Similar anomalous structural features can be detected in molecules A and B from PDB entry 3bxp (Fig. S1B): in this case, the trace of the connecting loop between strand β7 and helix α6 from molecule A places Asp233 in a wrong position and orientation, preventing the catalytic His233 residue from being located in its proper, canonical position, close to the nucleophile Ser116 (Fig. S1C). Finally, similar features can be identified in molecule B (Fig. S1D).

Oligomeric states of Cest_Lp2923: a case of pH-dependent pleomorphism

The two molecules in the asymmetric unit of the Cest-2923 hexagonal crystal tightly associate, forming a dimer. The association mode is similar to the one observed for other dimeric esterases from the α/β hydrolase superfamily [16, 17] and involves the antiparallel association between β8 strands and interactions between helices preceding (α6) and following (α7) this latter β strand (Fig. 4a). We refer to these dimers as canonical ones. Analysis of the interfaces within the Cest-2923 crystal with pisa [20] and pic [21] web servers reveals that the contact region between monomers has an interface area of ~ 980 Å2. The intermolecular contacts identified in this area are 26 hydrogen bonds, 20 hydrophobic and aromatic–aromatic interactions and two salt bridges. Among the set of hydrogen bonds detected, there are four between main chain atoms, which are those directly involved in the association between β8 strands (particularly from residues Tyr225 and Leu227).

Figure 4.

Dimeric and tetrameric assemblies of Cest-2923 from L. plantarum WCFS1. (A) Canonical dimer of Cest-2923. The association of the subunits involves interactions between strands β8 and α-helices α6 and α7. The two subunits are shown in different colours. (B) Noncanonical tetramer of Cest-2923. The contacting region between canonical dimers is mainly formed by α-helix α1, the contacting loops between strands β2 and β3 and between strand β8 and α-helix α7 and the C-terminal end. The four subunits are shown in different colours. (C) Canonical tetramer of HSL esterases. As an example, it is shown the tetramer of the hyperthermophilic esterase EstE1 (PDB code: 2c7b) [15]. Further details on the comparison between canonical and noncanonical tetramers are provided in Fig. S3. (D) Three-dimensional superposition between the noncanonical tetramers of Cest-2923 (in transparent, grey cartoons) and those found for the putative sugar hydrolase YeeB from L. lactis (PDB code: 3hxk).

In addition, this same analysis revealed the existence of a second contact region between symmetry-related molecules (molecules A-A and B-B) involving an even larger interface area (1460 Å2). The assembly of Cest_Lp2923 monomers through the joint combination of these two contact regions renders a tetrameric structure made up of two canonical dimers, which is stable according to pisa (total buried area: 9780 Å2; Gibb′s free energy of dissociation, ΔGdiss: 11.5 kcal·mol−1) (Fig. 4B). In this regard, it is worth noting that canonical dimers of close relatives of Cest-2923 such as EstE1 [15], Sto-Est [16], PestE [17] and Brefeldin A esterase [22] also form tetramers within the crystals (‘canonical tetramers’), which, nevertheless, are not comparable to the one observed for Cest-2923 (Fig. 4C). It is obvious that these observations raise the question of whether the Cest-2923 tetramers are crystallographic (i.e. by-products of the crystallization process). In this regard, results from different experimental approaches reveal a complex behaviour of Cest-2923: first, the crystal forms analysed by the structural genomics consortia (PDB code: 3d3n and PDB 3bxp) reveal canonical dimers as the highest order oligomeric form within the crystals, which would support the crystallographic nature of the observed tetramers. By contrast, analyses of the oligomeric state of Cest-2923 in solution by analytical ultracentrifugation techniques (Fig. 5) reveals a complex scenario that is consistent with an associative system involving monomeric, dimeric and tetrameric species. Thus, sedimentation velocity studies carried out in Tris buffer (20 mm Tris, pH 8.0, with 0.1 m NaCl) show that Cest-2923 behaves as (at least) two molecular species in solution, with sedimentation coefficients of 3.2 ± 0.2 S (n = 3) and 6.6 ± 0.1 S (n = 3), and estimated mean molecular masses of 32 kDa and 96 kDa, respectively, which are values consistent with monomeric and trimeric Cest-2923 species (Fig. 5a). Considering the above crystallographic results, where dimers and tetramers have been observed for this protein, this scenario can be easily explained in terms of two association equilibria: a fast equilibrium (within the time scale of the sedimentation velocity experiment, namely 6 h) between dimers and tetramers, therefore explaining the mean molecular mass of the 6.8 S species (intermediate between dimers and tetramers), and a second, slow equilibrium between monomers and dimers, therefore explaining the peak assignable to the monomer.

Figure 5.

Analytical ultracentrifugation analysis of Cest-2923 at neutral and acidic conditions. (A) Sedimentation coefficient c(s) distributions for Cest-2923 (13 μm) in Tris buffer (20 mm Tris-HCl, pH 8.0, with 0.1 m NaCl). Raw sedimentation velocity profiles for this analysis were acquired using A280, 148 288 g, 20 °C and different times (not shown). Calculations were conducted using sedfit [33]. (B) Sedimentation equilibrium analysis of Cest-2923 (13 μm) in Tris buffer (20 mm Tris-HCl, pH 8.0, with 0.1 m NaCl) at 10 545 g (open circles) and 21 916 g (open squares). A280 is plotted against the radial position from the centre of the rotor. The fit to the data set (solid line curves) corresponds to a monomer–tetramer association equilibrium. Residuals from this fit are shown in the inset. (C) As in (B) but where the fit corresponds to an ideal species. The best-fit weight mean molecular mass is 69.8 ± 5.4 kDa. (D) Sedimentation equilibrium analysis of Cest-2923 (13 μm) in acetate buffer (20 nm sodium acetate, pH 5.5, with 0.1 m NaCl) at 13 149 g (open circles) and 27 845 g (open squares). Other experimental conditions are identical to those at neutral conditions. The fit to the data set (solid line curves) corresponds to a monomer–tetramer association equilibrium. Residuals from this fit are shown in the inset. (E) As shown in (D) but where the fit corresponds to an ideal species. The best-fit weight mean molecular mass is 98.6 ± 0.6 kDa.

As expected from these latter ultracentrifugation results, sedimentation equilibrium experiments carried out under identical experimental conditions not only fitted well to a monomer to tetramer association scheme, but also to an ideal model considering a unique species with an estimated mean molecular mass of 69.8 ± 5.4 kDa (Fig. 5B and 5C), which would indicate that the system is mainly shifted to the dimeric species under neutral conditions. In this regard, it is worth noting that the two crystal forms prepared by the structural genomics consortia, where only canonical dimers were observed, were both obtained at pH 7.5 (PDB code: 3bxp: 30% 1,2-propanediol, 20% poly(ethylene glycol) 400, 0.1 m Hepes, pH 7.5; PDB code: 3d3n: 5% poly(ethylene glycol) 8000, 0.1 m calcium acetate, 0.1 m HEPES, pH 7.5).

Conversely, because the Cest-2923 crystals described in the present study, which revealed the presence of tetramers, were obtained in 1.7 m ammonium sulfate and 0.15 m sodium acetate (pH 4.6), we also analysed the behaviour of the enzyme in solution at acidic conditions. Unexpectedly, we found that Cest-2923 precipitates in sodium acetate buffer (20 mm sodium acetate, pH 4.6 with 0.1 m NaCl) even at a low protein concentration (0.2 mg·mL−1), although it was stable at pH 5.5. In this regard, both the far-UV CD spectrum at 25 °C and the thermal denaturation curve measured in acetate buffer, pH 5.5, are indistinguishable from those registered at pH 8.0, indicating that there are no significant structural differences and changes in stability (Fig. S2). Under these acidic conditions, sedimentation velocity analyses revealed a profile similar to those observed at pH 8.0, although with a relative lower contribution of the monomeric species for samples with the same protein concentration in comparable experiments (not shown). As expected, sedimentation equilibrium experiments fitted well to a monomer to tetramer associative model, and also to an ideal one with a unique species with an mean molecular mass of 98.6 ± 0.6 kDa (Fig. 5D and 5E), indicating a displacement of the equilibria towards the tetrameric form relative to the neutral conditions.

Combining the results derived from these two distinct experimental approaches (crystallographic and analytical ultracentrifugation), we can deduce that Cest-2923 behaves in solution as an associative system well described by two equilibria; namely, a fast equilibrium between species with molecular masses consistent with monomers and dimers and another much faster one between dimeric and tetrameric species and, second, that the crystallization process is an active player in selecting a pre-existing oligomeric form: dimers at neutral conditions and tetramers at acidic ones. Hence, we claim that the oligomeric forms determined by protein crystallography are not a mere crystallographic by-product but reveal intrinsic, associative properties of Cest-2923. Nevertheless, determination of the structural basis of the pH-dependence of the complex associative behaviour of Cest-2923 is not straightforward. Analysis of the interactions between Cest-2923 subunits with pisa [20] and pic [21] revealed that the hydrophobic and hydrogen-bond interactions are the main driving force for protein association not only for dimer formation, but also for tetramer formation. If this is the case, a displacement of the association equilibria towards tetramer formation can be predicted both under acidic or basic conditions with respect to the Cest-2923 isoelectric point (theoretical value of 6.5). In this sense, we carried out analytical ultracentrifugation studies of Cest-2923 at pH 9.0 (20 mm Tris, pH 9.0, 0.1 m NaCl) essentially as described above (both sedimentation velocity and equilibrium assays). The results obtained (Fig. S3) are similar to those obtained under acidic conditions, in agreement with an association process mainly guided by hydrophobic interactions.

As indicated above, EstE1 [15], Sto-Est [16], PestE [17] and Brefeldin A esterase [22] form canonical tetramers within the crystals (Fig. 4C). These tetramers have not been considered as a crystallographic by-product because they were observed in different crystal lattices with crystals prepared in disparate conditions. It is worth noting that this conclusion was derived despite these proteins being dimers in solution at much lower concentrations [17]. It is obvious that this behaviour resembles the one described for Cest-2923 and therefore similar association equilibria for these proteins cannot be dismissed. From a structural viewpoint, Cest-2923 tetramers are not canonical ones, and are almost perfectly superimposable on those of the putative sugar hydrolase YeeB from L. lactis (PDB code: 3hxk), one of its closest structural relatives (Fig. 4D). When compared, it can be seen that both canonical and noncanonical tetramers result from a head-to-head association because the same region from both canonical dimers is the one involved in the association, although this region is different in both types of tetramers: thus, defining the cis face of the dimers as the one in which the C-terminal α-helix of the participating monomers is situated, noncanonical tetramers would result from a cis-to-cis association, whereas canonical ones would result from a trans-to-trans association. In both cases, the final oligomer exhibits a 222 point group, although it is evident that the relative orientation of the dimers within tetramers is also different: if one dimer of each tetramer is fixed as a reference and is equally oriented for comparison, the other dimers are rotated with respect to each other around 60º (Fig. S4).

In sum, these results indicate that Cest-2923 displays a pleomorphic behaviour because it can form different oligomeric assemblies depending on pH, protein concentration and probably other environmental conditions that can be derived from crystallization experiments. We consider that this behaviour reflects an intrinsic property of the enzyme, although, on the other hand, we have no experimental basis to claim that this property has functional consequences in vivo. Undoubtedly, this needs to be investigated further.

Biochemical characterization

The most relevant enzymatic properties of Cest-2923 have been examined (Fig. 6). The optimum pH for hydrolytic activity against p-nitrophenyl acetate (pNPA) (see below) is 7.0 (Fig. 6A), which is a value typically observed for esterases, in contrast to the higher pH values (~ 8.0) displayed by lipases [23]. With respect to temperature, the esterase presented the highest activity at ~ 30 °C, although, at 45 °C, it exhibited a high level of activity (~ 65%) (Fig. 6B). These values for optimum pH and temperature are commonly found in other esterases from Lactobacilli [24-26]. On the other hand, temperature stability measurements show a drastic reduction in Cest-2923 hydrolytic activity upon incubation of the esterase at 55 °C (Fig. 6C). This result agrees well with the analysis of protein thermostability carried out by far-UV CD spectroscopy, which revealed an apparent t1/2 value at ~ 60 °C at neutral and acidic pH values (Fig. S2).

Figure 6.

Biochemical characterization of Cest-2923 from L. plantarum WCFS1. (A) Dependence on pH of hydrolytic activity of Cest-2923 against pNPA. Open squares indicate measurements in Tris buffer (pH 7.0) and phosphate buffer (pH 8.0) carried out to discard buffer-specific side effects. (B) Dependence on temperature of hydrolytic activity of Cest-2923 against pNPA. The optimum temperature for esterase activity was ~ 30 °C. (C) Analysis of the temperature stability of Cest-2923. Recombinant esterase was incubated in 50 mm sodium phosphate buffer pH 7.0 at 20 (open circles), 30 °C (closed circles), 37 °C (open squares), 45 °C (closed squares), 55 °C (open triangles) and 65 °C (closed triangles) for 15 min, 30 min, and 1, 2, 3, 4, 6 and 20 h. In all cases, the values shown are the mean of three independent experiments. (D) Dependence of the esterase activity of Cest-2923 on the aliphatic chain length of p-nitrophenyl (p-NP): p-NP acetate (C2); p-NP butyrate (C4); p-NP caprylate (C8); p-NP laurate (C12); and p-NP myristate (C14).

The acyl-length selectivity against p-nitrophenyl ester substrates follows the order: C2 > C4 > C8 > C12 > C14, indicating a preference for short acyl-length esters (Fig. 6D). The kinetic parameters for C2 and C4 substrates were determined spectrophotometrically. In both cases, Cest-2923 exhibited a hyperbolic Michaelis–Menten kinetics (not shown). The kinetic parameters are shown in Table 1. From the values of these parameters, it can be deduced that the catalytic efficiency (kcat/Km) for pNPA hydrolysis is ~ 75-fold greater than that observed for p-nitrophenyl benzoate hydrolysis (pNPB). Also, substrate specificity has been analysed with the use of a library of esters, as previously described [27]. This study reveals maximum hydrolysis against phenyl acetate, which is considered as reference (100% activity), and also significant activity (> 10% of the activity observed for phenyl acetate) against methyl bromoacetate and isopropenyl acetate. Within the limitations of the ester library employed, this result suggests a broader specificity for the alcohol part of the substrate and a preference for small moieties for the acid part.

Table 1. Kinetic parameters for pNPA and pNPB hydrolysis by Cest-2923. Enzyme activities were determined at 30 °C in 50 mm sodium phosphate buffer (pH 7.0). Results are the mean ± SD of three independent experiments.
SubstrateVmax (μmol·min−1·mg−1)Km (mm)kcat (s−1)kcat/Vmax (s−1·mm−1)
pNPA660 ± 501.7 ± 0.2343.6 ± 26202 ± 28
pNPB40 ± 87.6 ± 1.220.8 ± 42.7 ± 0.7

These three substrates were subsequently used for crystallographic studies aiming to determine the structure of the corresponding complexes. Accordingly, native crystals were incubated for ~ 60 s with a cryosolution containing the substrate at a concentration of 10 mm. Diffraction data recorded at beamline ID23-1 from the ESRF (Grenoble, France) with two types of these crystals (those prepared with methyl bromoacetate rapidly cracked) allowed the opportunity to solve their structure, which unfortunately did not result in the structure of the complexes but unexpectedly provided evidence of catalysis occurring within the crystals. Thus, we observed that, upon incubation with the corresponding cryosolution, the space group of the native crystals (P6322) changed: those crystals incubated with phenyl acetate became monoclinic (C2 space group), whereas those incubated with isopropenyl acetate changed to the P622 hexagonal space group. Interestingly, in both cases, the hexagonal packing was preserved and, indeed, indexing of the diffraction data identified the native unit cell and 622 point group, which suggests that these changes in crystal symmetry result from the ordered incorporation of new molecules within the crystal lattice (Table 2). One interesting observation is the presence in both types of crystals of an acetate molecule in the vicinity of the nucleophile Ser116, instead of a sulfate molecule, which appeared in native crystals (Fig. S5). We consider this to be remarkable because acetate by itself cannot replace sulfate in the active site as deduced from the native structure. We interpret this result as indicating the substitution of sulfate by the corresponding substrate, which is further hydrolysed. After this step, the alcohol is released in contrast to the acetate molecule, which remains in the active site.

Table 2. Data collection and refinement statistics.
  1. a

     Values for the highest resolution shell are given in parenthesis.

PDB accession code 4BZW 4C01 4BZZ
Data collection
BeamlineID29 (ESRF)ID23-1 (ESRF)ID23-1 (ESRF)
Wavelength (Å)0.94000.96870.9687
Space groupP 6322C 2P 622
Unit-cell parameters (Å)

a = b = 141.65,

c = 165.74

α = β = 90º, γ = 120º

a = 244.9, b = 141.4,

c = 82.69 α = γ = 90º,

β = 90.0º

a = b = 141.72

c = 82.29

α = β = 90º, γ = 120º

Resolution range (Å)46.37–2.1549.20–2.3048.97–2.99
Number of measured reflections1051 757509 57574 713
Number of unique reflectionsa53 876 (5215)124 688 (12 482)9994 (1306)
Mean I/σ(I)a17.2 (6.2)143.2 (46.5)13.7 (4.1)
Completeness (%)a99.9 (98.8)99.9 (100)99.6 (98.9)
Redundancya19.5 (18.0)4.1 (4.3)7.5 (7.3)
Rmerge (%)a; Rpim (%)a13.3 (51.4); 3.1 (12.4)14.3 (66.1); 13.6 (60.9)11.9 (47.1); 4.6 (18.5)
Wilson B-factor (Å2)
Molecules/nonhydrogen atoms
Protein2/43326/12 8971/2144
Other molecules1/817/861/3
Mean B-factors (Å2); protein27.041.049.7
Mean B-factors (Å2); water39.138.745.2
Mean B-factors (Å2); other76.064.894.2
rmsd bond length (Å)0.0070.0080.008
rmsd angles (º)1.0001.1601.060
Ramachandran favoured (%)
Ramachandran outliers (%)

Materials and methods

Bacterial strains, plasmids and DNA manipulations

L. plantarum WCFS1 was grown in MRS medium at 30 °C without shaking [28]. Escherichia coli DH10B was used for all DNA manipulations and E. coli BL21 (DE3) was used for expression in pURI3-Cter vector [29]. E. coli strains were cultured in LB medium at 37 °C and 140 r.p.m.

The gene lp_2923 from L. plantarum WCFS1 coding for Cest-2923 was cloned and overexpressed as described previously [30]. Briefly, the gene was PCR-amplified with HS Prime Start DNA polymerase (Takara Bio Inc., Otsu, Japan) using the primers 856 (TAACTTTAAGAAGGAGA TATACATatgcaagttgaacagcgcaca) and 857 (GCTATTAA-TGAT-GATGATGATGATGataattacc agctaacaatc) (the nucleotides pairing the expression vector sequence are shown in uppercase characters, and the nucleotides pairing the lp_2923 gene sequence are shown in lowercase characters). The corresponding 831-bp purified PCR product was inserted into the pURI3-Cter vector using the restriction enzyme- and ligation-free cloning procedure to produce C-terminally His-tagged Cest-2923 essentially as described previously [30].

E. coli DH10B cells were transformed, recombinant plasmids were isolated, and those containing the correct insert were identified by restriction-enzyme analysis, verified by DNA sequencing, and then transformed into E. coli BL21 (DE3) cells.

Expression and purification of recombinant Cest-2923

Cells carrying the recombinant plasmid pURI3-Cter-lp_2923 were grown at 37 °C in LB media containing ampicillin (100 μg·mL−1) and induced by adding 0.4 mm isopropyl thio-β-d-galactoside. After induction, the cells were grown at 22 °C for 20 h and collected by centrifugation using a J-25 Avanti centrifuge (Beckman Coulter, Fullerton, CA, USA) (7500 g for 15 min at 4 °C). Cells were resuspended in 20 mL of 20 mm Tris-HCl, pH 8.0, containing 100 mm NaCl·L−1 cell culture. Crude extracts were prepared by French Press lysis of cell suspensions. The lysate was centrifuged at 17 400 g for 40 min at 4 °C using a J-25 Avanti centrifuge.

The supernatant was filtered through a 0.22 μm filter (Millipore, Billerica, MA, USA) and subsequently loaded onto a HisTrap-FF column (GE Healthcare, Uppsala, Sweden), previously equilibrated in binding buffer (20 mm Tris-HCl, pH 8.0, containing 100 mm NaCl and 10 mm imidazole). The recombinant His-tagged Cest-2923 was eluted with a linear gradient of imidazole (from 10 mm to 500 mm) with an ÄKTA Prime Plus (GE Healthcare). Fractions containing Cest-2923 were pooled and concentrated by ultrafiltration with a YM-10 membrane (Amicon Corp., Danvers, MA, USA). The protein (2 mL) was then loaded onto a HiLoad™ 16/60 Superdex 200 prep-grade (GE Healthcare) equilibrated in Tris buffer (20 mm Tris-HCl, pH 8.0 containing 100 mm NaCl). The purity of the enzyme was checked by SDS/PAGE. Pure protein was finally concentrated to 10 mg·mL−1 for crystallization trials by ultrafiltration and stored at −80 °C until use.

Enzyme activity assay

Esterase activity was determined spectrophotometrically using pNPA as substrate. The rate of hydrolysis of pNPA for 10 min at 30 °C was measured in 50 mm sodium phosphate buffer pH 7.0 at 348 nm in a spectrophotometer (UVmini-1240; Shimadzu Corp., Kyoto, Japan). The reaction was stopped by chilling on ice.

To carry out the reaction, a stock solution of 25 mm of pNPA was prepared in acetonitrile/isopropanol (1/4, v/v) [31] and mixed with 50 mm sodium phosphate buffer (pH 7.0) to obtain a substrate final concentration of 1 mm. Reaction was initiated by adding 10 μg of protein. Control reactions containing no enzyme were utilized to account for any spontaneous hydrolysis of the substrates tested. Enzyme assays were performed in triplicate. One unit of esterase activity was defined as the amount of enzyme required to release 1 μmol of p-nitrophenol per minute under the previously described conditions.

Substrate specificity

The substrate specificity of Cest-2923 was investigated by two different approaches: first, we studied the dependence of the hydrolytic activity of the enzyme on the aliphatic chain length of the substrates using p-nitrophenyl esters with various chain lengths: pNPA (C2); p-nitrophenyl butyrate (C4); p-nitrophenyl caprylate (C8); p-nitrophenyl laurate (C12) and p-nitrophenyl myristate (C14), essentially as described previously [10]. Briefly, a stock solution of each p-nitrophenyl ester was prepared in acetonitrile/isopropanol (1/4, v/v). Substrates were emulsified to a final concentration of 0.5 mm in 50 mm sodium phosphate buffer (pH 7.0), containing 1.1 mg·mL−1 Arabic gum and 4.4 mg·mL−1 Triton X-100. The reaction mix consisted of 990 μL of emulsified substrate and 10 μL of enzyme solution (10 μg of protein). Reactions were carried out at 30 °C in a spectrophotometer (UVmini-1240; Shimadzu Corp.), as described above.

Second, substrate specificity was also analysed using the ester library described previously by Liu et al. [9]. Briefly, the screening was performed in a 96-well plate Flat Bottom (Sarstedt, Nümbrecht, Germany) where final reaction volume was 200 μL and each well contained a different substrate (1 mm final concentration in 1% acetonitrile). A buffer/indicator solution containing 0.44 mm of p-nitrophenol was used as pH indicator in 1 mm sodium phosphate buffer (pH 7.2). Cest-2923 (10 μg in 20 μL of 1 mm sodium phosphate buffer, pH 7.2) was added to each well and reactions were followed by monitoring the decrease in A410 for 2 h at 30 °C in a Synergy HT microplate spectrophotometer (BioTek Inc., Winooski, VT, USA). Blanks without enzyme were carried out for each substrate and data were collected in triplicate.

The general ester library consisted of 50 commercially available esters. These were chosen not only to identify acyl chain length preferences of the hydrolases, but also for their ability to hydrolyse hindered or charged substrates. Simple alkyl esters, as well as activated esters (vinyl and phenyl esters, esters with electron-withdrawing substituents in the acyl portion), were included to test whether activated esters would react faster.

Biochemical characterization of Cest-2923

To investigate temperature effect, reactions were performed in 50 mm sodium phosphate buffer (pH 7.0) at 4, 20, 30, 37, 40, 45, 55 and 65 °C. The effect of pH was investigated by assaying esterase activity in a range of pH values from 5.5 to 9.0 at 30 °C. Buffers used were acetic acid-sodium acetate buffer for pH 5.5, sodium phosphate buffer for pH 6–7, Tris-HCl buffer for pH 8 and glycine-NaOH buffer for pH 9. A 100 mm concentration was used in all the buffers. Although initially activity measurements were carried out at lower pH values (between 3.0 and 5.0), these are not reported because protein aggregation was detected in acetate buffer, pH 4.6 or 5.0.

For temperature stability measurements, the recombinant esterase was incubated in 50 mm sodium phosphate buffer, pH 7.0, at 20, 30, 37, 45, 55 and 65 °C for 15 min, 30 min, and 1, 2, 3, 4, 6 and 20 h. After incubation, the residual activity was measured as described above.

Analytical ultracentrifugation

Equilibrium and sedimentation velocity ultracentrifugation experiments were performed at 8073 g at 20 °C, using a XL-A ultracentrifuge (Beckman Coulter) and standard double sector centerpiece cells. Solvent density (1.0029 mg·mL−1) and the partial specific volume of Cest-2923 (0.719) were calculated from the buffer composition (100 mm NaCl, 20 mm Tris-HCl or 100 mm NaCl, 20 mm sodium acetate) and from the predicted amino acid composition, respectively, with sednterp [32]. Data from sedimentation velocity and equilibrium experiments were analysed using sedfit [33] and heteroanalysis [34], respectively.

CD spectroscopy

Far-UV CD measurements were carried out on a Jasco J-715 spectropolarimeter (Jasco Inc., Easton, MD, USA) equipped with a thermostated cell holder and a Peltier temperature control accessory. The instrument was calibrated with (+)-10-camphorsulfonic acid. CD spectra were recorded in 0.1-cm path length quartz cell cuvettes from 250 to 205 nm, using a protein concentration of 6.4 μm (bandwidth, 1 nm; response, 4 s; scan speed, 20 nm·min−1). Each presented spectrum is the average accumulation of four scans. Baseline subtraction was performed in all cases. Results are expressed as mean residue ellipticity [θ]MRW, in units of degree cm2·dmol−1 of amino acid (Mr = 110 for this protein). Thermal transitions were also analysed by CD spectroscopy by monitoring the variation of the ellipticity at 218 nm as the temperature was increased from 20 to 90 °C at 50 °C·h−1. The normalized ellipticity value at each temperature was calculated as ([θ]t – [θ]25)/([θ]90 – [θ]25), where [θ]t is the ellipticity value at temperature t, and [θ]25 and [θ]90 are the ellipticity values at 25 °C and 90 °C, respectively. Three different samples were analysed, although the traces shown correspond to individual samples.

Crystallization and data collection

Initial crystallization conditions for Cest-2923 were determined by the sitting-drop vapour diffusion method with commercial screens from Hampton Research (Riverside, CA, USA) in Innovaplate SD-2 96-well plates set up using a Nanodrop Innovadyne robot (IDEX Health & Science LLC, Oak Harbor, WA, USA). Each drop contained 250 nL of protein (7 mg·mL−1) in Tris-HCl buffer (20 mm Tris-HCl, pH 8.0, containing 0.1 m NaCl) and 250 nL of reservoir solution. Drops were equilibrated against 65 μL of reservoir solution. Scaling up of the crystallization conditions using hanging drops in 24-well VDX plates (2 : 1 protein : precipitant volume ratio; total volume 3 μL) rendered high quality diffracting crystals in 1.7 m ammonium sulfate, 0.15 m sodium acetate (pH 4.6). Crystals suitable for X-ray analysis were transferred to an optimized cryoprotectant solution [reservoir solution plus 20% (v/v) 2-methyl-2,4-pentanediol] for ~ 10 s and then cryocooled at 100 K in the cold nitrogen-gas stream. Diffraction data were recorded on a Pilatus 6M detector (Area Detector Systems Corp., Poway, CA, USA) at beamline ID29 at the ESRF (Grenoble, France). Diffraction images were processed with xds [35] and the data scaled and analysed using scala [36] from the ccp4 software suite [37]. Cryosoaking experiments were carried out with native Cest-2923 crystals, with a cryosolution containing phenyl acetate or isopropenyl acetate (10 mm final concentration. These new crystals were measured at beamline ID23-1 at the ESRF. Diffraction data were recorded on a Pilatus 6M-F detector (Area Detector Systems Corp.). Data statistics are summarized in Table 2.

Structure determination and refinement

The crystal structure of Cest-2923 was determined by the molecular replacement method with phaser [38] using phenix [39]. The atomic coordinates of the enzyme deposited by the NSGC (PDB code: 3d3n) were used as search model. Importantly, the 25-residue long loop containing the catalytic His233 residue not defined in this model could be modelled with the diffraction data obtained from the new crystal form described in the present study. Conversely, the atomic model deposited by the JCSG for Cest_Lest-2923 (PDB code: 3bxp) is discussed above. Restraint refinement and automatic water molecule placement was conducted using phenix-refine [40]. Stereochemical validation was carried out using molprobity [41]. The refinement statistics are summarized in Table 2. The final model has an R-factor of 15.5% and an Rfree of 19.5%, and included 550 amino acid residues, nine sulfate molecules, three acetate molecules, one Tris and 520 solvent molecules. Details on the structures derived from crystals soaked with substrates phenyl acetate and isopropenyl acetate are provided in Table 2. The pisa [20] and pic [21] webservers were used to calculate values of buried interface areas. Ribbon diagrams were prepared using pymol [42].


We thank the ESRF (Grenoble, France) for provision of synchrotron radiation facilities (ID29 and ID23-1). J.M.M. thanks Margarita Menéndez for her helpful advice on ultracentrifugation analyses. Financial support received from the Ministerio de Economía y Competitividad (BFU2010-17929/BMC), as well as from Factoría de Cristalización (Consolider-Ingenio-2007) (J.M.M.), AGL2011-22745 (R.M.) and FUN-C-FOOD Consolider 25506 (R.M.), is greatly appreciated. Y.A., M.E.-T. and I.A. are recipients of fellowships: CSIC-CITMA, JAE Predoc (CSIC) and FPU (MEC), respectively.