Structural insights into the substrate specificity of two esterases from the thermophilic Rhizomucor miehei

Two hormone-sensitive lipase (HSL) family esterases (RmEstA and RmEstB) from the thermophilic fungus Rhizomucor miehei, exhibiting distinct substrate specificity, have been recently reported to show great potential in industrial applications. In this study, the crystal structures of RmEstA and RmEstB were determined at 2.15 Å and 2.43 Å resolutions, respectively. The structures of RmEstA and RmEstB showed two distinctive domains, a catalytic domain and a cap domain, with the classical α/β-hydrolase fold. Catalytic triads consisting of residues Ser161, Asp262, and His292 in RmEstA, and Ser164, Asp261, and His291 in RmEstB were found in the respective canonical positions. Structural comparison of RmEstA and RmEstB revealed that their distinct substrate specificity might be attributed to their different substrate-binding pockets. The aromatic amino acids Phe222 and Trp92, located in the center of the substrate-binding pocket of RmEstB, blocked this pocket, thus narrowing its catalytic range for substrates (C2–C8). Two mutants (F222A and W92F in RmEstB) showing higher catalytic activity toward long-chain substrates further confirmed the hypothesized interference. This is the first report of HSL family esterase structures from filamentous fungi.jlr The information on structure-function relationships could open important avenues of exploration for further industrial applications of esterases.

Esterases (EC 3.1.1.1) are a general class of carboxylic ester hydrolases, which catalyze the cleavage and formation of ester bonds ( 1 ). They exhibit maximum activity toward water-soluble or emulsifi ed esters of short-chain diluted enzyme solution was prepared in 400 l of 50 mM Tris-HCl buffer pH 7.5, and after preheating for 2 min, 50 l of 20 mM p NPA substrate (in pure isopropanol) was added. The mixture was incubated at 50°C for 10 min. The reaction was stopped by adding 500 l of 300 mM phosphate buffer pH 7.0 containing 5% (w/v) SDS. The released p -nitrophenol ( p NP) was quantifi ed by measuring the absorbance at 410 nm. One unit of enzyme activity was defi ned as the amount of enzyme required to liberate 1 mol p NP per minute under the above assay conditions. Protein concentration was measured by the Lowry method using BSA as the standard. Specifi c activity was expressed in units per milligram protein.

Protein crystallization and X-ray data collection
Proteins were crystallized by the sitting-drop vapor diffusion method at 293 K by mixing 1 µl protein solution with an equal volume of reservoir solution. Crystals of Rm EstA were obtained with a reservoir solution containing 25% (w/v) polyethylene glycol (PEG) 3350 and 0.2 M (NH 4 ) 2 SO 4 in 0.1 M MES buffer pH 6.0. The Rm EstA crystals were observed 2 days later. Crystals of Rm EstB were obtained with a reservoir solution containing 20% (w/v) PEG4000 and 10% (v/v) 2-propanol in 0.1 M HEPES buffer pH 7.5. The Rm EstB crystals were observed after 7 days.
For X-ray diffraction experiments, each crystal was fi shed from the crystallization drop using a nylon loop (Hampton Research), soaked briefl y in a cryoprotectant solution (the crystallization solution supplemented with 20% v/v glycerol), and then fl ashcooled in liquid nitrogen at 70 K. X-ray diffraction data of Rm EstA and Rm EstB were collected from single crystals at beamline 3W1A at the Beijing Synchrotron Radiation Facility (BSRF) and beamline BL-17U at the Shanghai Synchrotron Research Facility (SSRF), respectively. All diffraction data were indexed, integrated, and scaled using the program HKL-2000 ( 20 ).

Structure determination and refi nement
The structure of Rm EstB was determined by the molecular replacement (MR) method using the structure of PDB entry 1JJI ( Archaeoglobus fulgidus carboxylesterase) as the search model. The structure of Rm EstA was solved by MR using the refi ned structure of Rm EstB as the search model. The structural models were built and refi ned using the Phenix ( 21 ) and Coot ( 22 ) programs. R values for all data were reduced by several cycles of simulated annealing, minimization, and B -factor refi nement using Phenix.refi ne followed by manual model rebuilding. The final models were analyzed and validated with MolProbity ( 23 ). Structural homologs of esterases were identifi ed in the DALI server ( 24 ). Structural superpositions and RMSD (root-meansquare deviation) calculation were performed with the LSQMAN program ( 25 ). Secondary-structure elements were identifi ed by the DSSP ( Defi ne Secondary Structure of Proteins ) program ( 26 ). Figures were prepared with PyMOL ( 27 ). Sequence alignment was created by ClustalW ( 28 ). Data collection and refi nement statistics are given in Table 1 .

Substrate-specifi city analysis
Substrate specifi city of Rm EstB mutants were investigated according to standard enzyme assay in 50 mM Tris-HCl buffer pH 7.5 at 50°C using different p NP esters as the substrates, including p NPA, p -nitrophenyl butyrate ( p NPB), p -nitrophenyl hexanoate ( p NPH), p -nitrophenyl caprylate ( p NPC), p -nitrophenyl decanoate ( p NPD), p -nitrophenyl laurate ( p NPL), p -nitrophenyl myristate ( p NPM) and p -nitrophenyl palmitate ( p NPP). One unit of enzyme activity was defi ned as the amount of enzyme required to release 1.0 mol of p NP per minute under the above assay conditions. acidocaldarius (PDB: 1EVQ), Bs Est from Bacillus subtilis (PDB: 1JKM), and EstEl (PDB: 2C7B) and EstE7 (PDB: 3DNM) from metagenomic libraries. However, there have been no structural descriptions of HSL family esterase from a fi lamentous fungus.
HSL family esterases share a characteristic ␣ / ␤ -hydrolase fold, which is composed of a central ␤ -sheet surrounded on both sides by ␣ -helices, serving as a stable protein core. The amino acid substitutions, loop insertions, and deletions occurring in the central cores during evolution have led to enzymes with diverse catalytic functions ( 17 ). The catalytic mechanism of ␣ / ␤ hydrolase is based on a catalytic triad made up of a nucleophile (Ser), an acid (Asp or Glu), and a His. The central nucleophile is located within a conserved G-X-S-X-G motif in the "nucleophile elbow" ( 18 ). Though the catalytic mechanism is nearly identical in all ␣ / ␤ hydrolases, their substrate specifi cities are very different. The reasons for this phenomenon have not been elucidated at the structural level.
We recently characterized two novel HSL family esterases from the thermophilic fungus Rhizomucor miehei : Rm EstA and Rm EstB ( 15,16 ). The physiological substrates of Rm EstA and Rm EstB detected in our studies have been found to be short-chain triacylglycerol, linalyl acetate, and butyl butyrate. Both of these enzymes belong to the ␣ / ␤ hydrolases and exhibit distinct substrate specifi cities: Rm EstA shows highest activity toward longer-chain esters, whereas Rm EstB favors hydrolysis of shorter-chain esters. Here, to decipher the structural basis of their different substrate specifi city, we report the crystal structures of the two esterases and detail their structural differences. This work is a fi rst attempt to characterize HSL family esterases from a fi lamentous fungus at the structural level to gain insights into their structure-function relationships.

Enzyme assays and protein determination
Esterase activity was determined as described by Gutiérrez-Fern á ndez et al. ( 19 ) using p -nitrophenyl acetate ( p NPA) as the substrate with minor modifications. Briefly, 50 l of suitably The overall structure of the Rm EstA monomer was analogous to that of the Rm EstB monomer ( Fig. 2 ). The structures of Rm EstA and Rm EstB could be divided into two domains: a catalytic domain (residues 51-191 and 253-322 for Rm EstA, residues 51-191 and 253-322 for Rm EstB) and a cap domain (residues 4-51 and 207-247 for Rm EstA, residues 3-51 and 206-247 for Rm EstB). The catalytic domains had the canonical architecture of an ␣ / ␤ -hydrolase fold protein consisting of a central ␤ -sheet of eight mostly parallel strands surrounded by ␣ -helices ( Fig. 2 ). The core ␤ -sheets of each monomer were related by 2-fold symmetry to form an extended intermolecular 16-stranded ␤ -sheet. The central ␤ -sheet displayed a left-handed superhelical twist, with ␤ 1 and ␤ 8 strands crossing each other at an angle of ‫ف‬ 120° ( Fig. 2 ).

Structural comparison of Rm EstA and Rm EstB with other esterases
The analysis of structural similarity carried out with DALI search suggested high structural similarity of both Rm EstA and Rm EstB with other reported esterases. Rm EstA exhibited high structural similarity with the esterases from P. calidifontis ( Pc Est, PDB code: 3ZWQ), A. fulgidus ( Af Est, PDB code: 1JJI), Salmonella typhimurum ( St Est, PDB code: 3GA7), and A. acidocaldarius ( Aa Est, PDB code: 1EVQ), with Z-score and RMSD of C ␣ atom values of 39.7 and 1.9, 39 and 2.0, 38.8 and 2.1, and 38.2 and 2.2, respectively. Rm EstB displayed high structural similarity with the esterases from P. calidifontis ( Pc Est, PDB code: 3ZWQ), A. fulgidus ( Af Est, PDB code: 1JJI), A. acidocaldarius ( Aa Est, PDB code: 1EVQ), and metagenomic library (EstE1, PDB code:

Accession numbers
The atomic coordinates and structural factors for crystal structures of Rm EstA and Rm EstB were deposited in the PDB under accession numbers 4WY5 and 4WY8, respectively.

Overall structures of Rm EstA and Rm EstB
The crystal structures of Rm EstA and Rm EstB were determined at 2.41 Å and 2.27 Å resolutions, respectively. The crystallographic statistics for data collection and structure refi nement are summarized in Table 1 . The triclinic space group of Rm EstA was P 1 with two monomers in the asymmetric unit ( Fig. 1A ). Amino acid residues 1-3 of Rm EstA and the uncleaved C-terminal His-tag were not visible on the electron-density map. In the asymmetric unit, Rm EstA formed a dimer in complex with three sulfates, and two of the sulfates interacted with two monomers via Arg104 ( Fig. 1B ). Note that the third single sulfate was found in the crevice of two monomers, with four Arg residues from two monomers forming an arched area ( Fig.  1C ). The monoclinic space group of Rm EstB was C 121 with four molecules in the asymmetric unit. The four monomers were arranged as two canonical dimers to further form a tetramer via hydrogen bonding network ( Fig.  1D ). Four hydrogen bonds were involved in the formation of the tetramer: Ile39 and Asp40 of chain A were directly hydrogen bonded to Lys149 of chain C, Glu55 of chain A was directly hydrogen bonded to Asp75 of chain C, and Val51 of chain A was directly hydrogen bonded to Gln79 of chain C ( Fig. 1E ). Values in parenthesis are from the last resolution shell.
; 95% and 5% of refl ections were used for R work and R free , respectively .
esterases. The structural differences were found mainly in the loop regions.

The active sites of Rm EstA and Rm EstB
A classical catalytic triad consisting of Ser161 (Ser164 in Rm EstB) as the nucleophile, His292 (His291 in Rm EstB) as the proton acceptor/donor, and Asp262 (Asp261 in Rm EstB) as the residue stabilizing the His was identifi ed in 2C7B), with Z-score and RMSD of C ␣ atom values of 41.5 and 1.8, 40.5 and 1.9, 40.3 and 2.1, and 40.3 and 1.7, respectively. Note that the three-dimensional structures of Rm EstA and Rm EstB shared high similarity with those of the other members of the ␣ / ␤ -hydrolase fold family, though they showed low sequence identities (<40%). Superimposing Rm EstA/ Rm EstB on the structures of Pc Est and Af Est revealed similar overall folds of these HSL family pocket of Rm EstB extended ‫ف‬ 11 Å from the protein surface to the catalytic residue Ser164. This deep hydrophobic cleft was funnel-shaped and surrounded by four ␣ -helices ( ␣ 1, ␣ 2, ␣ 6, and ␣ 8) and the loop regions (His88-Gly91 and Ile290-Ala297). The substrate-binding pocket of Rm EstA was a channel running through the whole protein and the entrance to this channel was surrounded by fi ve ␣ -helices ( ␣ 1, ␣ 2, ␣ 6, ␣ 8, and ␣ 9), the 3 10 -helix G2 (His197-Lys199) and the loop regions (His87-Gly90 and Ile291-Ala298) ( Fig. 4B ).

Substrate specifi cities
The structural differences between the substrate-binding pockets of the two esterases ( Rm EstA and Rm EstB) might contribute to the difference in carbon chain lengths of the Rm EstA ( Rm EstB) ( Fig. 3 ). The key nucleophile Ser161 in Rm EstA (Ser164 in Rm EstB) was found within the conserved pentapeptide sequence Gly-X-Ser-X-Gly, which is located at the apex of the nucleophile elbow, a sharp turn connecting ␤ 5 and ␣ 6 ( 5 ). A hydrogen bond (2.7 Å in Rm EstA, 2.4 Å in Rm EstB) between the O ␥ atom of Ser161 (Ser164 in Rm EstB) and the N 2 atom of His292 (His291 in Rm EstB) stabilized the conformation of the nucleophile Ser161 in Rm EstA (Ser164 in Rm EstB). The side chains of His292 and Asp262 in Rm EstA were stabilized by a network of hydrogen bonds located at the carboxyl edge of ␤strands 7 and 8, respectively. However, this phenomenon was not found in Rm EstB. The His-Gly-Gly-Gly motif (residues 87-90 in Rm EstA and residues 88-91 in Rm EstB), which is usually conserved in the HSL family, was found upstream of the active sites. The oxyanion hole was created by residues Gly89, Gly90, and Ala162 in Rm EstA, and by Gly90, Gly91, and Ala165 in Rm EstB ( Fig. 3 ). The main-chain nitrogen atoms of the oxyanion hole donate hydrogen to the cleaved substrate ( 29 ), stabilizing the negative charges on the tetrahedral intermediates arising from the nucleophilic attack of Ser161 in Rm EstA or Ser164 in Rm EstB.

Comparison with Rm EstA and Rm EstB
Superposition analysis revealed high structural homogeneity between Rm EstA and Rm EstB ( Fig. 4A ). Superpositioning of Rm EstA onto Rm EstB exhibits an overall RMSD of 1.41 Å for 311 corresponding C ␣ atoms, though they shared only 46% sequence identity. The folding patterns of Rm EstA and Rm EstB presented a common core domain, where the assignment of the secondary-structure elements was almost the same. Residues 4-51 of Rm EstA and residues 3-51 of Rm EstB made up the cap domains upstream of their respective catalysis domains. It is interesting that the two cap domains shared no sequence similarity but formed similar tertiary structures. Superposition of the surface of Rm EstA and sticks of Rm EstB suggested that their most striking structural differences were localized in the substrate-binding pockets ( Fig. 4B ). The substrate-binding  toward C2 was slightly decreased. However, the specifi c activity of Rm EstB-F222A toward esters with relatively longer chains, such as C4 and C6, increased signifi cantly by 1.65and 1.4-fold, respectively ( Table 2 ). The mutant Rm EstB-W92F showed similar results: the specifi c activities of Rm EstB-W92F toward C4 and C6 increased by 1.33-and 1.11-fold, respectively, whereas that toward C2 remained almost unchanged ( Table 2 ). ester substrates ( Fig. 4 ). Two aromatic amino acids (Phe222 and Trp92), located in the center of the substrate-binding pocket of Rm EstB, might block this pocket and narrow the substrate specifi city of Rm EstB. To validate this speculation, two mutants ( Rm EstB-F222A and Rm EstB-W92F) were designed, and the substrate specifi cities of two mutants were determined. Compared with that of the wild-type enzyme, the specifi c activity of the mutant Rm EstB-F222A specifi city, regioselectivity, thermophilicity, and thermostability ( 29 ). The ␣ / ␤ -hydrolase fold family members have a highly conserved nucleophile-His-acid catalytic triad, with Ser as the nucleophile and Asp or Glu as the acid ( 17,35 ). The nucleophiles of Rm EstA and Rm EstB are Ser161 and Ser164, respectively, positioned in the conserved sequence Gly-X-Ser-X-Gly at a sharp turn connecting ␤ 5 and ␣ 6, similar to the esterase EstE1 ( 5 ).
Previous biochemical characterizations of Rm EstA and Rm EstB have indicated distinct substrate specifi cities for the two enzymes. Rm EstA can hydrolyze esters of longer carbon chain lengths (up to C16), with the highest activity observed for C6 ( 15 ), whereas Rm EstB favors the hydrolysis of esters with shorter carbon chain lengths, with highest activity observed for C2 ( 16 ). These properties differ from most other HSL family esterases, which show the highest activity for butyrate (C4) or caproate esters (C6) ( 36,37 ). A structural comparison of Rm EstA and Rm EstB suggested that the differences in the substrate-binding pocket might make a marked contribution to their distinct substrate specifi cities. The substrate-binding pocket of Rm EstB is funnel-shaped with ‫ف‬ 11 Å from the protein surface to the catalytic residue Ser164, a distance that just fi ts the acyl chains of substrates with carbon chain lengths shorter than C4. Esters with acyl chain lengths longer than C4 therefore hardly bind to the substrate-binding site due to steric hindrance ( Fig. 5 ). On the other hand, the substrate-binding pocket of Rm EstA is a curved tunnel which can accommodate ester substrates with long acyl chains ( Fig. 5 ).
The structural differences between the substrate-binding pockets of the two esterases suggest that several aromatic residues (Phe222 and Trp92) block the substrate-binding pocket in Rm EstB ( Fig. 4B ), which might contribute to the restriction in carbon chain lengths of the ester substrates. Phe222 and Trp92 are located in the center of the substrate-binding pocket, close together (4.1 Å). Aromatic residues in the substrate-binding pocket can transform the substrate specifi city of esterase, as confi rmed in the other previous study ( 18 ). To confi rm the function of aromatic residues Trp92 and Phe222 in Rm EstB's substrate specifi city, two mutants, Rm EstB-F222A and Rm EstB-W92F, were created. The substrate specifi cities of mutant Rm EstB-F222A for p NPB (C4) and p NPH (C6)

DISCUSSION
The ␣ / ␤ -hydrolase superfamily is one of the largest enzyme superfamilies recognized to date and ubiquitous from all kingdoms of life ( 3 ). Despite modest degrees of overall primary sequence homology, the basic structure fold of the ␣ / ␤ -hydrolases is extraordinarily conserved. However, diverse ␣ / ␤ -hydrolases have different substrate specifi cities, which contain a variety of enzymes, including esterases, lipases, proteases, dehalogenases, peroxidases, and epoxide hydrolases, and play important roles in life activities. Therefore, it is imperative to identify structural differences among various ␣ / ␤ -hydrolases. Esterases often show broad substrate spectrum and are widely used as biocatalysts for the synthesis of important materials in pharmaceutical and chemical industries ( 4 ). Although some esterase structures have been determined in recent years ( 5,19,(30)(31)(32)(33)(34), no HSL esterase structure from a fi lamentous fungus has ever been reported. Here, we describe the structures of two HSL esterases from R. miehei , Rm EstA and Rm EstB, and elucidate the mechanism governing their different substrate specifi cities. The crystal structures of Rm EstA and Rm EstB allow us to address the molecular details of substrate binding and catalysis of the shortchain esters being different from other ␣ / ␤ -hydrolase superfamily esterases/lipases. Site-directed mutagenesis and structure-based rational design experiments can then be performed to search for enzymes with the improved catalytic effi ciency and/or suitability for industrial applications.
The three-dimensional structures of both Rm EstA and Rm EstB exhibit the typical ␣ / ␤ -hydrolase fold with a core consisting of eight ␤ -sheets surrounded by ␣ -helices. The structures of both Rm EstA and Rm EstB are composed of two clearly distinguishable domains, a catalytic domain and a cap domain ( Fig. 2 ), which are similar to those of most other esterases from the HSL family. However, several differences were found in the cap domain, for which ␣ -helices ␣ 1 and ␣ 2 showed the highest B -factors (data not shown). The cap domain of HSL family esterases generally has a poorly conserved amino acid sequence but is structurally similar to other esterases/lipases. This domain makes an important contribution to several aspects of HSL enzyme function, including enzyme activity, substrate  ( Fig. 6 ).

CONCLUSION
The three-dimensional structures of two HSL esterases from R. miehei , Rm EstA and Rm EstB, were determined. Both of the esterases were composed of a core catalytic domain were enhanced by 1.65-and 1.4-fold, respectively ( Table 2 ). The mutant Rm EstB-W92F showed similar substrate-specifi city changes, with 1.33-and 1.11-fold enhanced specifi city for p NPB (C4) and p NPH (C6), respectively ( Table 2 ). The substrate specifi cities of the two Rm EstB mutants toward longchain substrates also increased signifi cantly compared with that of the wild-type Rm EstB ( Table 2 ). Structural comparison and mutagenesis analysis in the present study indicated that the side chains of residues in the substrate-binding pocket create a steric hindrance, thereby potentially altering substrate specifi city in the esterases; moreover, residue Phe222 played a vital role in Rm EstB's variation in substrate specifi city. The results of multiple sequence alignment analysis further confi rmed this interference. Residues with large side chains were found at positions corresponding to position 222 of Rm EstB in esterases showing a preference for the hydrolysis of p NPA (C2), such as those from Escherichia coli [PDB code: 4KRY; corresponding residue: glutamate ( 30 )],  The key residue Phe222 of Rm EstB is labeled with a red dot in the amino acid sequence alignment. The optimal substrates of these esterases are listed, taken from previous studies ( 5,(30)(31)(32)(33)(34)38 ). and a cap domain, exhibiting the typical ␣ / ␤ -hydrolase fold. The side chains of residues in the substrate-binding pocket may create steric hindrance, thereby altering the substrate specifi city of the esterases. The results in the present study may be helpful for the construction of new variants to improve substrate specifi city of esterases and for further exploration of biotechnological applications.