Conversion of a Rhizopus chinensis lipase into an esterase by lid swapping.

In an effort to explore the feasibility of converting a lipase into an esterase by modifying the lid region, we designed and characterized two novel Rhizopus chinensis lipase variants by lid swapping. The substrate specificity of an R. chinensis lipase was successfully modified toward water-soluble substrates, that is, turned into an esterase, by replacing the hydrophobic lid with a hydrophilic lid from ferulic acid esterase from Aspergillus niger Meanwhile, as a comparison, the lid of R. chinensis lipase was replaced by a hydrophobic lid from Rhizomucor miehei lipase, which did not alter its substrate specificity but led to a 5.4-fold higher catalytic efficiency (k*cat/K*m) toward p-nitrophenyl laurate. Based on the analysis of structure-function relationships, it suggests that the amphipathic nature of the lid is very important for the substrate specificity. This study provides new insight into the structural basis of lipase specificities and a way to tune the substrate preference of lipases.

Transformation of P. pastoris and selection of His + recombinants P. pastoris GS115 was transformed with 1 g of Sal I-linearized expression vector by electroporation. The His + transformants were obtained on minimal dextrose medium plate. The resistant genetic transformants were screened on YPD-G418 plate. The insertion and methanol metabolization were checked by PCR. The PCR amplifi cations were carried out according to Invitrogen's recommendations with genomic DNA and primers complementary to the 5 ′ and 3 ′ regions of the AOX1 gene.

Expression and purifi cation of lipases
The P. pastoris His + transformants were cultured in 25 ml of buffered glycerol-complex medium shaken at 30°C and 250 rpm in 250 ml glass fl asks. When cultures reached an optical density at 600 nm (OD 600 ) of ‫ف‬ 6.0, the cells were centrifuged and resuspended in 100 ml buffered methanol-complex medium to an OD 600 of 1.0, shaken at 28°C and 250 rpm in 500 ml glass fl asks for 120 h. The cultures were supplemented with 5 g/l methanol every 12 h to induce the expression of lipase.
After induction, the lipases from the culture supernatant were harvested by centrifugation at 8,000 g at 4°C for 20 min and then concentrated and interchanged with 10 mM phosphate buffer (pH 8.0) by ultrafi ltration through a 10 kDa membrane (Millipore). Then, the concentrated solutions were loaded onto a HiTrap SP FF column (Pharmacia, 5 × 5 ml) equilibrated with 20 mM acetic acid-sodium acetate buffer (pH 5.0), and eluted with 0-0.5 M NaCl in the same buffer. Fractions containing lipase activities were pooled, concentrated, and chromatographed on a HiTrap Phenyl HP (Pharmacia, 5 × 5 ml) equilibrated in 50 mM sodium phosphate buffer (pH 7.5) containing 1.6 M ammonium sulfate. Lipases were then eluted in an ammonium sulfate concentration gradient ranging from 1.6 to 0 M in 50 mM sodium phosphate buffer (pH 7.5). Protein fractions with lipase activities were collected and dialyzed against phosphate buffer (pH 8.0). The purifi ed enzymes were stored at 4°C. Protein fractions collected from the previous steps were analyzed by SDS-PAGE . Protein concentrations were determined using the Bradford assay. Bovine serum albumin was used as a standard.

Infl uence of pH and temperature on activity and stability
The enzymatic activities for hydrolyses of p -nitrophenyl esters were measured according to Kordel et al. ( 24 ). Optimal pH was determined by examining the activity of the enzyme using p-nitrophenyl palmitate ( p NPP) as substrate at 40°C in the following buffers: 50 mM citrate (pH 5.0-5.5), 50 mM phosphate (pH 6.0-8.0), 50 mM Tris-HCl (pH 8.5-9.0), and 50 mM carbonate (pH 9.5-10.0). The pH stability was determined by incubating 0.2 U lipase solution in buffers (50 mM, pH 5.0-10.0) for 1 h at 25°C and analyzing the residual activity. Optimal temperature was determined by measuring the enzyme activity using p NPP as hydrolysis of milk fat and other endogenous esters is believed to be the key reaction to produce special products with characteristic fl avor profi les, which is based on the different substrate specifi cities of lipases (17)(18)(19). Rhizomucor miehei lipase (RML) exhibits the substrate selectivity toward p -nitrophenyl decanoate (C10) and myristate (C14) ( 20 ), and Rhizopus oryzae lipase (ROL) shows the highest specifi city toward tricaprylin (C8) and triolein (C18) ( 13 ). The thermostable variant S4-3 ( 21 ) from R. chinensis CCTCC M201021 ( 22 ) shows the chain-length specifi city toward tricaprylin (C8) and trilaurin (C12). However, the enzyme specifi city with regard to the chain length in the substrate is not necessarily optimal for the manufacture of a particular product. Therefore, the molecular manipulation of lipases to alter chain-length specifi city is obviously of industrial and theoretical interest.
In this research, in an effort to explore the feasibility of converting a lipase into an esterase by modifying the lid region, we designed and characterized two novel R. chinensis lipase variants by lid swapping. The R. chinensis lipase S4-3 was successfully converted into an esterase by replacing the hydrophobic lid with a hydrophilic lid from ferulic acid esterase from Aspergillus niger (AnFaeA). Meanwhile, as a comparison, the lid of S4-3 replaced by a hydrophobic lid from RML was also studied.

Mutagenesis and construction of the lipase expression vector
The plasmids for the expression of genes of an S4-3 lipase variant generated from the lid of S4-3 swapped with that from RML (S4-3M) and an S4-3 lipase variant generated from the lid of S4-3 swapped with that from AnFaeA (S4-3N) were constructed as follows: The lid replacement was carried out by the overlap extension PCR strategy ( 23 ). A pair of end-primers, Pro-F (5 ′ -TCAAG-ATC CCTAGG GTTCCTGTTGCTGGTCATAAAGGTTC-3 ′ , italics indicate Avr II) and Pro-R (5 ′ -AATTCCAGT GCGGCCGC TTACAA-ACAGCTTCCTTCG-3 ′ , italics indicate Not I), the primers RMLF for S4-3N using various concentrations of p -nitrophenyl laurate ( p NPL), p -nitrophenyl acetate, or p -nitrophenyl propionate as substrate according to the method described by Burdette et al. ( 26 ).

3D structure simulation
A 3D model of R. chinensis lipase in a closed conformation was built by SWISS-MODEL protein automated modeling program ( 27 ) on the basis of the crystal structure of lLGY (crystal structure of lipase II from Rhizopus niveus solved with a resolution of 2.20 Å) ( 28 ). The structure of R. chinensis lipase in an open form was modeled by SWISS-MODEL based on the 3D crystallographic structure of the inhibited R. miehei lipase (PDB id: 4TGL) ( 29 ). The structures were energy minimized by molecular dynamics for 10 ps at 298 K.

Sequence alignment and lid swapping
Homology modeling of the R. chinensis lipase S4-3 shows an identity of 81.1% compared with the crystal structure of lLGY. The secondary structure of R. chinensis lipase consists of nine ␣ -helices and eight ␤ -strands, and the catalytic triad of Ser145-His257-Asp204 is shielded by a mobile lid in the "closed" form. The lid in R. chinensis lipase S4-3 consists of 6 amino acids (underlined, 83 TNS FRSAIT DMVFT 96 ) plus 3-5 amino acids as two hinges around the lid. The superimposed apo -structures of S4-3, RML (PDB id: 3TGL), and AnFaeA (PDB id: 1USW) ( Fig. 1 ) show the close structural relations between these three structures, except the big difference in the lid region between AnFaeA and S4-3/RML. The lids of S4-3 and RML in their apo -structures are in a closed form, which shields the active site and the substrate binding region. However, the lid of AnFaeA is kept in an open form, and AnFaeA also does not demonstrate interfacial activation ( 30 ). Figure 2 shows the amino acid sequence alignment of S4-3, RML, and AnFaeA, and the box indicates the lid region. The identity of amino acids is 55.93% between S4-3 and RML and 28.21% between S4-3 and AnFaeA. Because the lid, the important functional motif, is one of the most variable regions in terms of sequences, it is an interesting target for investigation of its impact on substrate at optimal pH under various temperatures (20°C-70°C). Thermostability was determined by preincubating 0.2 U lipase for various intervals up to 1 h at temperatures ranging from 20°C to 70°C in optimal pH buffer and analyzing the residual activity.

Lipase substrate specifi city
Chain-length specifi city of the purifi ed enzyme was analyzed by using various fatty acid esters of triglycerides as substrates. The enzymatic activities for hydrolyses of triglycerides were determined by the spectrophotometric assay at 430 nm using the formation of copper soaps for the detection of free fatty acids ( 25 ). One unit (1 U) of lipase activity was defi ned as the amount of enzyme needed to liberate 1 mol of free fatty acids per minute under defi ned reaction conditions.

Interfacial kinetics
The values of interfacial kinetic parameters K * m and k* cat were determined in a heterogeneous medium reacted under the condition of pH 8.0 and 40°C for S4-3 and S4-3M and of pH 8.0 and 45°C   while for S4-3 only in the range of pH 8.0-8.5, and a sharp decrease below or above pH 8 was observed for S4-3M. S4-3M and S4-3N were stable over a broader pH range, which preserved 80% residual activity at the pH range of 6.5-8.5, whereas S4-3 only maintained 80% residual activity at pH 7.5-8.5 ( Fig. 4B ). The effects of temperature on lipase activities were examined at temperatures between 20°C and 70°C ( Fig. 4C ). S4-3N was most active at 45°C, which was 5°C higher than the optimum temperatures of S4-3M and S4-3. In addition, S4-3N maintained higher activity at temperatures above 45°C compared with S4-3M and S4-3. S4-3M and S4-3N were both less thermostable than S4-3 ( Fig. 4D ). After incubation at 40°C for 1 h, S4-3 retained ‫ف‬ 100% residue activity, but the residual activities of S4-3M and S4-3N were only 45% and 38%, respectively.

Substrate specifi city
The lid in lipase plays an important role in modulating not only stability and activity but also substrate specifi city. A comparative study of fatty acid chain-length specifi city was carried out by measuring the initial rate of hydrolysis of various fatty acid esters of triglycerides by S4-3, S4-3M, and S4-3N. As shown in Table 1 , specifi c activities were obtained for the purifi ed lipases toward triacetin, tributyrin, tricaproin, tricaprylin, trilaurin, tripalmitin, and tristearin (C2-C18). The relative activities (activity toward trilaurin equals 100%) are shown in Fig. 5 . For the lid swapped with the lid from AnFaeA, the specifi c activity of S4-3N toward the substrates with fatty acid chain longer than C6 reduced signifi cantly by 6-to 410-fold. On the contrary, the specifi c activity of S4-3N toward short-chain substrates increased the catalytic properties of the lipase. The lid of S4-3 was swapped with that from AnFaeA or RML to generate the variant S4-3N or S4-3M, respectively.

Enzymatic properties of S4-3, S4-3M, and S4-3N
S4-3, S4-3M, and S4-3N were constructed and extracellularly expressed under the control of the methanolinducible alcohol oxides promoter in P. pastoris GS115. The parent S4-3 and the variants S4-3M and S4-3N were purifi ed to electrophoretically homogeneous states at the fi nal step as shown in Fig. 3 . Purifi ed lipases were used for determination of their properties ( Fig. 4 ). The activities and stabilities of S4-3, S4-3M, and S4-3N were measured over a pH range of 5.0-10.0 at 40°C. They all showed the optimum activities at pH 8.0 ( Fig. 4A ). However, S4-3N was more active below pH 6.0 compared with S4-3 and S4-3M, and the activity of S4-3N was >85% in the range of pH 7.5-9.0  p -nitrophenyl propionate (C3)], the k * cat value of S4-3N was about 3 times higher than those of the parent and S4-3M, and according to the low K * m value of S4-3N, the affi nity between S4-3N and short-chain fatty acid esters was much stronger than those of the parent and S4-3M. The k * cat / K * m value of S4-3N toward p -nitrophenyl acetate (C2) or p -nitrophenyl propionate (C3) was one magnitude higher than that toward p NPL (C12).

DISCUSSION
Most lipases, including S4-3, have a lid domain that covers the catalytic triad, and the movement of an ␣ -helical lid at the lipid-water interface created a large hydrophobic patch around the catalytic triad, resulting in activation of lipase. In this paper, the hydrophobic lid of S4-3 was replaced with that of homologs to explore the feasibility of converting a lipase into an esterase and to evaluate the role of the lid in the modulation of catalytic properties, including pH and temperature properties, kinetics, and substrate specifi city. In order to minimize structure perturbation, we chose two closely structure-related enzymes, AnFaeA and RML, with a lid in the same size for lid swapping. AnFaeA is an esterase with a hydrophilic lid kept in an open form, which hydrolyzes the ester bonds between hydroxycinnamic acids and polysaccharide or lignin. From a different genus, RML is the closest related lipase with a hydrophobic lid, which showed similar substrate selectivity toward p -nitrophenyl decanoate (C10) and myristate (C14) compared with S4-3 ( 20 ), but the reported catalytic effi ciency of RML was much higher than that of lipases from Rhizopus sp. ( 31 ). Insertion of the lid from AnFaeA or RML into the S4-3 backbone did not change much of the optimum pH, the pH stability, and the optimum temperature. However, the lid swapping did negatively affect the thermostability of the variants. The results suggested that the lid swapping destabilized the structure of the variants; however, the subtle change of the structures could not be judged from the far-UV circular dichroism spectra of the enzymes (data not shown). A site-directed mutagenesis study also demonstrated that mutations in the lid region destabilized the protein ( 32 ). The most apparent changes by lid swapping were that the replacement of the S4-3 lid with that of AnFaeA shifted the specifi city toward short-chain substrates (C2-C6) compared with that of the parent (C12), while the replacement of the S4-3 lid with that of RML enhanced the catalytic effi ciency toward long-chain fatty acid substrate. by 7.2-fold (C3) and 38.0-fold (C2), respectively. Accordingly, S4-3N displayed an altered substrate preference toward short-chain fatty acid esters, which exhibited a 4,440-fold higher relative activity toward the triacetin (C2)/trilaurin (C12) activity ratio compared with the parent. For the variant S4-3M, we found that replacement of the lid with that from RML caused a 1.5-to 3.3-fold increase in the specifi c activity toward those substrates (C2, C6, C8, C12, and C16) and a 40% reduction toward tristearin (C18) compared with the corresponding activity of the parent. Comparing the relative activity toward various substrates tested ( Fig. 5 ), S4-3M and S4-3 both exhibited higher substrate specifi cities toward tricaprylin (C8) and trilaurin (C12), while S4-3N mostly preferred short-chain substrates (C2-C6).

Interfacial kinetic parameters
Interfacial kinetic parameters K * m and k * cat were used to determine the intrinsic substrate fatty acyl specifi city of lipase catalysis. By measuring the interfacial kinetic parameters toward short-and long-chain fatty acid esters, we attempted to obtain more insight into the impact by lid swapping. As shown in Table 2 , when the long-chain fatty acid ester p NPL (C12) was used as the substrate, the K * m value of S4-3M increased subtly, while the K * m value of S4-3N was 4.8 times higher than that of the parent, which indicated that the affi nity between S4-3N and the substrate p NPL decreased signifi cantly. The k * cat value of S4-3M increased by 6.2 times, and the value for S4-3N decreased by 6.0 times compared with that of the parent, which suggested that the introduced new lid interfered with the interaction between the catalytic active site and the long-chain fatty acid ester. On the other hand, in agreement with the shift of the substrate specifi city of S4-3N toward shortchain fatty acid esters [ p -nitrophenyl acetate (C2) and The reaction conditions were pH 8.0 and 40°C for S4-3 and S4-3M, and pH 8.0 and 45°C for S4-3N, respectively. residue in the substrate binding groove, which probably partially contributed to the high C2/C12 activity ratio and the high catalytic effi ciency toward short-chain fatty acid esters. In addition, the substitution of positively charged Arg87 with noncharged Thr disrupted the original salt bridge between Arg87 and Asp61, which might destabilize the conformation of the variant.
The lid swapped with the hydrophobic lid from RML did not change the substrate specifi city of the parent S4-3. As shown in Fig. 6 , the exposed open lid showed the same hydrophobic pattern, which is crucial for the interaction with the water-insoluble substrates. However, the lid swapping resulted in the enhancement of the catalytic efficiency toward p NPL (C12), which was another interesting issue of the impact of the lid. The k * cat for S4-3M toward p NPL increased considerably by 6.2 times with a small increase of K * m , accounting for an increase in k * cat / K * m . It was very interesting because the k * cat value of the chimera S4-3M with a lid from RML was comparable with that of RML toward p -nitrophenyl caprylate ( 31 ). The different amino acid composition of the RML lid might be one of the contributors to the high catalytic effi ciency. As shown in Fig. 6 , the lid residues Thr83, Ala89, Asp92, Met93, and Phe95 in the lid-open form of S4-3 were predicted to reside in the substrate-binding groove. The corresponding residues in RML were Ser82, Trp88, Asp91, Leu92, and Phe94, among which Asp91 and Phe94 were conserved. After replacement with the lid from RML, one important difference in the substrate-binding region of S4-3M was the substitution from Thr83 (S4-3) to Ser82 (RML). This residue was proposed to act in the formation of the oxyanion hole, which stabilized a negatively charged oxygen in the tetrahedral intermediates of the covalent bound substrate ( 36,37 ). From molecular dynamics, it was evident The available X-ray structures of free ROL, RML coupled with substrate-like inhibitor, and AnFaeA give a hint of the molecular basis of substrate-enzyme interactions ( 28,29,33 ). AnFaeA is an esterase that has high substrate specifi city toward hydrophilic feruloyl esters ( 34 ). Moreover, the lid in AnFaeA is always held in an open form without showing interfacial activity ( 34 ). The overall structure between S4-3 and AnFaeA is very similar with a backbone RMSD of 2.9 Å, while the deviation of the lid is about 7.5 Å. As shown in Fig. 2 , the lid of AnFaeA was more hydrophilic with 85.7% polar residues compared with 57.1% polar residues in the lid of S4-3. Figure 6 shows that upon activation in S4-3N, 80% polar residues of the open lid were exposed on the protein surface, which constituted one side of the substrate binding region, while in the parent S4-3 only 40% polar residues was observed. The variant S4-3N swapped with the AnFaeA lid exhibited substrate specifi city toward short-chain substrates (C2-C3) ( Table 2 ). It suggested that the variant S4-3N with a hydrophilic lid favored binding of relatively hydrophilic substrates such as p -nitrophenyl acetate and p -nitrophenyl propionate. Besides the signifi cant increase in the hydrophilicity of the lid, replacement of key residues appears worthy of discussion. Residues Thr83, Arg87, Asp92, and Phe95 in the lid of S4-3 were important residues involved in the catalytic process. Thr83 and Asp92 were conserved in S4-3 and AnFaeA, while Arg87 and Phe95 in S4-3 corresponded to Thr and Tyr in AnFaeA, respectively. According to the literature, in ROL upon activation Arg87 approached Asp61 to form a salt bridge that stabilized the open conformation ( 35 ), and Phe95 located at the end of the hydrophobic substrate-binding groove is important for the substrate specifi city of ROL ( 13 ). After lid swapping, the mutation Phe95Tyr in S4-3N introduced a polar  3 consists of the amino acids T83, A89, D92, M93, F95, F112, L146, P178, I205, V206, V209,  P210, P211, F214, and I254, which were colored in green, purple, and yellow. The nonpolar residues in the lid regions were colored in purple, and the polar residues in the lid regions were colored in yellow. The marked number of the residues in the lid regions was according to the number in the original amino acid sequences. The corresponding amino acids in S4-3M and S4-3N were also colored in the same way. that Thr has only a weak hydrogen bonding interaction (Thr83 O ␥ -p NPL O2 = 2.98 Å) with the substrate p NPL in S4-3, while Ser showed a strong interaction (Ser83 O ␥ -p NPL O2 = 2.67 Å) with the substrate in S4-3M. S4-3M showed a 6.2-fold higher k * cat value compared with the parent, which was probably due to the stronger interaction by the mutation Thr83Ser. On the contrary, Joerger and Haas ( 12 ) reported that the single mutation Thr83Ser in ROL caused 2-to 7-fold reductions in the activity toward triolein, tricaprylin, and tributyrin. The single mutation might cause geometry disturbance of the substrate binding region, resulting in a large decrease in the lipase activity. However, in our case the replacement of the whole lid including the mutation Thr83Ser favored the interaction of the oxygen hole residue Ser with the substrate. Another important mutation was from Ala89 (S4-3) to Trp88 (RML); it was postulated that this Trp might play a role in the interfacial activation process in RML ( 38 ). We observed a slightly broad substrate scope for the variant S4-3M ( Fig. 5 and Table 1 ), which was probably attributable to the mutation of Ala89 to a more hydrophobic Trp in the substrate binding groove.
In order to verify that the hydrophobicity of the lid is involved in the catalytic process, a proof-of-concept sitedirected mutagenesis was carried out to mutate a hydrophobic residue Phe94 (using RML numbering) into a polar residue Tyr in the S4-3M variant. Tyr was the corresponding residue in the lid of S4-3N. Thus, the mutation Phe94Tyr in the lid of S4-3M may contribute to an increased effi ciency toward short-chain fatty acid esters. As expected, after mutation the variant exhibited an 8-fold higher relative activity toward the triacetin (C2)/trilaurin (C12) activity ratio. From this result, it clearly appeared that hydrophobic pattern of the lipase lid had drastic effects on its substrate specifi city.
Lipases and esterases with diverse chain-length specifi city are the most widely used classes of enzymes in industry. In the present study, for the fi rst time a lipase has been converted into an esterase by modifying the lid region. The substrate specifi city of the R. chinensis lipase was successfully modifi ed toward water-soluble substrates by replacing the hydrophobic lid with a hydrophilic lid from AnFaeA. Homology-based modeling and site-directed mutagenesis reveals that not only the amphipathic nature of the lid but also its specifi c amino acid sequences are vitally important for the enzymatic properties of lipases. Our fi ndings may provide an opportunity for better understanding of the structural basis of the substrate specifi city of lipases and guide the design of novel lipases for application in industry.