Applied Microbiology and Biotechnology

, Volume 68, Issue 6, pp 779–785

Creation of Rhizopus oryzae lipase having a unique oxyanion hole by combinatorial mutagenesis in the lid domain

Authors

  • Seizaburo Shiraga
    • Division of Applied Life Sciences, Graduate School of AgricultureKyoto University
  • Masaji Ishiguro
    • Suntory Institute for Bioorganic Research
  • Harukazu Fukami
    • Suntory Institute for Fundamental Research
  • Masahiro Nakao
    • Suntory Institute for Fundamental Research
    • Division of Applied Life Sciences, Graduate School of AgricultureKyoto University
Applied Genetics and Molecular Biotechnology

DOI: 10.1007/s00253-005-1935-0

Cite this article as:
Shiraga, S., Ishiguro, M., Fukami, H. et al. Appl Microbiol Biotechnol (2005) 68: 779. doi:10.1007/s00253-005-1935-0

Abstract

Combinatorial libraries of the lid domain of Rhizopus oryzae lipase (ROL; Phe88Xaa, Ala91Xaa, Ile92Xaa) were displayed on the yeast cell surface using yeast cell-surface engineering. Among the 40,000 transformants in which ROL mutants were displayed on the yeast cell surface, ten clones showed clear halos on soybean oil-containing plates. Among these, some clones exhibited high activities toward fatty acid esters of fluorescein and contained non-polar amino acid residues in the mutated positions. Computer modeling of the mutants revealed that hydrophobic interactions between the substrates and amino acid residues in the open form of the lid might be critical for ROL activity. Based on these results, Thr93 and Asp94 were further combinatorially mutated. Among 6,000 transformants, the Thr93Thr, Asp94Ser and Thr93Ser, Asp94Ser transformants exhibited a significant shift in substrate specificity toward a short-chain substrate. Computer modeling of these mutants suggested that a unique oxyanion hole, which is composed of Thr85 Oγ and Ser94 Oγ, was formed and thus the substrate specificity was changed. Therefore, coupling combinatorial mutagenesis with the cell surface display of ROL could lead to the production of a unique ROL mutant.

Introduction

Lipases (EC 3.1.1.3) catalyze the hydrolysis of triacylglycerol and other fatty acid esters at a lipid–water interface. The physiological properties and biochemical features of several microbial, fungal and mammalian lipases have been investigated extensively (Svendsen 2000). The three-dimensional structures of these lipases have also been determined (Fischer and Pleiss 2003). All lipases have a catalytic triad composed of three amino acids (Ser-His-Asp/Glu), similar to serine proteases, and show an α–β hydrolase-fold. In addition, almost all lipases have a lid or flap domain that covers its catalytic triad. The lid undergoes a conformational change at the lipid–water interface, called “interfacial activation”, creating a large hydrophobic patch around the catalytic triad and resulting in activation of the lipase (Derewenda et al. 1992; Berg et al. 1998; Cajal et al. 2000). The substrate specificity of a fungal lipase, Rhizopus oryzae lipase (ROL), is for middle- or long-chain substrates (Ueda et al. 2002). Its three-dimensional structure can be easily deduced from that of R. niveus lipase, which has a 99% amino acid identity to ROL (Kohno et al. 1996).

In previous studies, we reported on the construction of a combinatorial library of six amino acids (Phe88-Arg89-Ser90-Ala91-Ile92-Thr93) comprising the lid domain of ROL, using yeast cell-surface engineering (Shiraga et al. 2002). Display systems of protein libraries on some biological resources have become useful methods for investigating the structure–function relationship of proteins (Benhar 2001). Among some display systems, a yeast display system can express many functional fungal proteins that are necessary for post-translational modifications to be made to their mature forms (Murai et al. 1997; Washida et al. 2001) and permits the combinatorial mutagenesis of multi-point amino acids to be carried out, because of the feasibility for the assay of mutants (Shiraga et al. 2004). Previous results obtained from screening soybean oil- or tributyrin-containing plates revealed that the sequential alignment of (basic amino acid)–(polar amino acid)–(non-polar amino acid) might be important in the function of the lid domain.

In this study, we first constructed a combinatorial library of the three amino acids of the lid domain and obtained some mutants that exhibited a higher activity and different specificity from that of the wild type. Based on the computer modeling of these mutants, we further constructed a Thr93- and Asp94-combinatorial library and obtained two mutants that exhibited significant shifts in substrate specificity toward short-chain substrates. Computer modeling of these mutants revealed that a unique oxyanion hole had been created.

Materials and methods

Chemicals

Tributyrin and soybean oil were obtained from Wako Pure Chemical Industries (Osaka, Japan). Fluorescein dibutyrate and fluorescein dilaurate were obtained from Lambda (Graz, Austria). Fluorescein dioctanoate and fluorescein dioleate were donated from Suntory (Kyoto, Japan).

Strains and media

Escherichia coli DH5α [F, endAl, hsdR17(rK,mK+), supE44, thi-1, λ, recAl, gyrA96, ΔlacU196 (ϕ80dlacZΔM15)] was used as a host for recombinant DNA manipulation. Saccharomyces cerevisiae strain MT8-1 (MATa, ade, his3, leu2, trp1, ura3; Tajima et al. 1985) was used as the host for the protein display. E. coli was grown in LB medium (1% tryptone, 0.5% yeast extract, 0.5% sodium chloride) containing 50 μg ml−1 ampicillin. Yeast was cultivated in SD-W medium [2% glucose, 0.7% yeast nitrogen base without amino acids (Difco, Detroit, Mich., USA), 0.002% adenine sulfate, 0.002% l-histidine-HCl, 0.003% l-leusine, 0.002% uracil] containing 2% casamino acids.

Construction of plasmids for the display of combinatorial mutated lipases

The plasmid for the display of the ROL on the cell surface of S. cerevisiae (pWRSL17S) and the negative-control plasmid (pMW1) were constructed as described by Shiraga et al. (2002).

The plasmids for the display expression of two combinatorial ROL libraries (Phe88Xaa, Ala91Xaa, Ile92Xaa; Thr93Xaa, Asp94Xaa) were constructed as follows. Two oligonucleotides were amplified by PCR: (1) m-Lid-2 for Phe88Xaa, Ala91Xaa, Ile92Xaa (5′-TGTTTTCCGCGGTACCAACTCCNNKAGAAGTNNKNNKACTGATATCGTCTTCAACTTTTCTGACTACAAGCCTGTCAAGGCGCCAAAGTTCATG-3′, where N is a mixture of A, T, G, C; K is a mixture of G and T; italics indicate sites for SacII, EheI) and (2) m-Lid-D,T for Thr93Xaa, Asp94Xaa (5′-TGTTTTCCGCGGTACCAACTCCTTCAGAAGTGCCATCNNKNNKATCGTCTTCAACTTTTCTGACTACAAGCCTGTCAAGGGCGCCAAAGTTCATG-3′, where italics indicate SacII, EheI). The respective primers used, with opposite directions, were (5′-TGTTTTCCGCGGTACCAACTCC-3′) and (5′-CATGAACTTTGGCGCCCTTGACAGGCTTGTAG-3′). The amplified fragments were digested by SacII and EheI and introduced into the larger isolated fragment of pWRSL17S after digestion with SacII and EheI. The resulting display plasmids for mutated ROLs were used to transform the yeast.

Screening by halo assay

Transformants displaying mutated ROLs on their cell surface were spread on SD-W + 2% casamino acids agar plates containing 0.2% (v/v) tributyrin or 0.2% (v/v) soybean oil as substrate, together with 1% (w/v) gall powder as an emulsifier. Colonies able to hydrolyze tributyrin or soybean oil were identified as clear halos, as described by Shiraga et al. (2004).

Lipase activity assay using fluorescein substrates

Fluorescein dibutyrate (C=4), fluorescein dioctanoate (C=8), fluorescein dilaurate (C=12) and fluorescein dioleate (C=18), each at 5.0×10−5 M, were independently dissolved in methyl cellosolve/0.1 M Tris-HCl buffer (pH 8.0) at 1: 19 (Kramer and Guilbault 1963). Yeasts were cultivated in 10 ml SD-W medium at 30°C to an optical density (OD600) of 2.0 and then precipitated by centrifugation at 800 g for 10 min. After washing the cells (OD600 0.2), they were suspended in 100 μl Tris-HCl buffer (pH 8.0) and the reaction was started by the addition of 100 μl substrate solution. On incubation at 37°C, the fluorescence intensities of the solutions were measured by means of a Fluoroscan Ascent fluorometer (Labsystems, Helsinki, Finland) on tissue-culture plates (353047 Multiwell 24-well; Becton Dickinson Labware, N.J., USA) at 15-min intervals. Filters with excitation at 485 nm and emission at 527 nm were employed.

DNA sequencing

About 200-bp DNA fragments encoding the combinatorially mutated portions were amplified from the yeast colonies by PCR (Sambrook and Russel 2001). Amplified fragments were purified by Quantum Prep PCR Kleen spin columns (Bio-Rad, Richmond, Calif., USA) and were then sequenced using a Dye terminator cycle sequencing ready reaction kit and an ABI Prism 373A DNA sequencer (Perkin–Elmer/Applied Biosystems, Foster City, Calif., USA).

Modeling of substrate–enzyme complex structures

The crystal structure of ROL has not been determined at this time, so the crystal structure of R. niveus lipase (PDB ID: 1LGY), which has a 99% amino acid identity to ROL, was used for the model building of the mutants. The mutations were carried out with a Homology module installed in Insight II ver. 2000 (Molecular Simulations, San Diego, Calif., USA). The initial structure of the enzyme was optimized by molecular mechanics calculations using Discover 3 ver. 98.0 (Molecular Simulations). The minimized structure was then optimized by molecular dynamics calculations at 298 K with the cell multipole method, a distance-dependent dielectric constant and a time step of 1 fs for 100 ps, by sampling conformations at every 1 ps. The hundred conformations were minimized until the final root-mean-square deviation (rmsd) became less than 0.1 kcal mol−1 Å−1 and the lowest energy conformation was selected for the substrate-docking study.

The substrate was roughly docked into the ligand-binding cleft with the guidance of a hydrogen bond of the ester carbonyl oxygen with the backbone amide proton of Leu148. The initial complex model was minimized and the substrate-binding site was then covered by water molecules (20 Å sphere). The structure, consisting of the substrate and residues within 10 Å from the substrate, was energy-optimized in the presence of the water molecules by the molecular dynamics/minimization procedure, as described above. The lowest-energy structure was selected as an energy-refined complex model.

Results

Determination of mutation points

The lid of the lipase is a very unique and interesting domain. The dynamics of the movement of the lid at the lipid–water interface, called interfacial activation, have been extensively studied kinetically; and mutations or substitutions of the lid have also been reported (Secundo et al. 2004; Brocca et al. 2003). These reports suggest that the amino acid sequences of the lid largely affect the activity and substrate specificity of the lipase.

The lid domain of ROL consists of six amino acids (Phe88-Arg89-Ser90-Ala91-Ile92-Thr93). Combinatorial libraries of three amino acids containing the lid domain were screened. A combinatorial library of three amino acids is composed of 8.0×103 mutants. We initially screened five types of combinatorial library of three amino acids: library 1 Phe88Xaa, Ala91Xaa, Ile92Xaa, library 2 Ala91Xaa, Ile92Xaa, Thr93Xaa, library 3 Phe88Xaa, Ile92Xaa, Thr93Xaa, library 4 Arg89Xaa, Ser90Xaa, Ala91Xaa, library 5 Arg89Xaa, Ser90Xaa, Thr93Xaa. Among the five combinatorial libraries described above, only clones obtained from library 1 (Phe88Xaa, Ala91Xaa, Ile92Xaa) made halos on plates containing soybean oil.

Evaluation of clones from combinatorial library 1 (Phe88Xaa, Ala91Xaa, Ile92Xaa) using fluorescein substrates

The plasmids for displaying combinatorially mutated ROLs (Phe88Xaa, Ala91Xaa, Ile92Xaa) on the yeast cell surface were constructed as described in Materials and methods. A total of about 4×104 clones were further obtained on plates containing soybean oil. Forty-four clones were randomly selected and the randomness of N (=A, T, G, C equimolar mixture) was checked. The results demonstrated the absence of any significant bias (A=13.8%, T=25.0%, G=36.4%, C=24.7%). Among ten clones that produced halos, four produced halos that were larger than that of wild-type ROL yeasts. The relative hydrolysis rates and substrate specificity of these four clones were measured, using four types of fatty acid esters of fluorescein (C=4, 8, 12, 18; Fig. 1). Clone 2-3 exhibited 1.3 times higher relative activities toward C=4 than the wild type and each of the clones exhibited unique substrate specificity. These sequences of clones revealed that the clones containing large hydrophobic amino acids residues (Phe, Val, Ile, Leu) at positions 88 and 91 exhibited high activities. Non-hydrophilic amino acids (Gly, Cys) were frequently found in the sequences of six clones that produced small halos and in 14 that produced no halos (Table 1). In the substitutions that showed small or no halos, the secondary structure of the α-helix of the lid appeared to have been destroyed.
https://static-content.springer.com/image/art%3A10.1007%2Fs00253-005-1935-0/MediaObjects/253_2005_1935_Fig1_HTML.gif
Fig. 1

Relative hydrolysis activities toward four kinds of fluorescein substrates of wild-type and combinatorially mutated ROL yeasts. Data are means based on five independent measurements. Deviations are all less than 10%

Table 1

Amino acid sequences of Phe88, Ala91, Ile92 combinatorial library

Wild type

Phe88, Ala91, Ile92

Clones formed, larger halos

Clones formed, no halos

2-1 Leu Thr Leu

 2-30 Phe Ala Asp

2-2 Val Val Val

 2-31 Leu Ala Ser

2-3 Leu Leu Ile

 2-32 Arg Arg Val

2-4 Phe Ser Ser

 2-33 Phe Asp Asp

 

 2-34 Ala Gly Val

Clones formed, halos

 2-35 Trp Ser Phe

2-14 Cys Phe Val

 2-36 Cys Ser Cys

2-20 Ala Val Ala

 2-37 Phe Gly Leu

2-21 Val Ala Leu

 2-38 Gly Val Leu

2-25 Arg Leu Gly

 2-39 Arg Asp Cys

2-26 Cys Gly Val

 2-40 Are Met Met

2-28 Ala Val Leu

 2-41 Ala Ile Arg

 

 2-42 Phe Asp His

 

 2-43 Gly Ala Arg

Molecular modeling of ROL mutant 2-3 with fluorescein dibutyrate

As shown in Fig. 1, among the clones obtained from the combinatorial library Phe88Xaa, Ala91Xaa, Ile92Xaa, clone 2-3 exhibited the highest activity toward fluorescein dibutyrate. The structure of the ROL mutant 2-3–fluorescein dibutyrate complex was investigated (Fig. 2). Phe88Leu and Ala91Leu comprised a larger hydrophobic patch for interaction with the hydrophobic tail of substrates than that of the wild type. In particular, the amino acid side-chain of Ala91Leu, which is located near the hydrophobic tail of the substrates, resulted in enhanced ROL activities. Considering the fact that clones containing small hydrophobic amino acid residues, for example 2-20, 2-21, 2-28, produced small halos and mutants with non-polar amino acid residues produced no halos at all (Table 1), these results suggest that the activity of ROL could be enhanced if the hydrophobic patch formed by the lid and adjacent amino acids were larger.
https://static-content.springer.com/image/art%3A10.1007%2Fs00253-005-1935-0/MediaObjects/253_2005_1935_Fig2_HTML.gif
Fig. 2

Model of ROL mutant 2-3–fluorescein dibutyrate complex. Substitutions of Leu88 (Phe in wild type) and Leu91 (Ala in wild type) are represented. Hydrophobic amino acid residues of the lid and adjacent amino acids contribute to the hydrophobic patch which interacts with fluorescein dibutyrate (light blue). Thr93 and Asp94 are located near the hydrophobic patch, but have hydrophilic amino acid residues

Evaluation of the lipase activity of mutants from the combinatorial library Thr93, Asp94, using fluorescein substrates

Thr93 and Asp94 were located near the hydrophobic patch and the substitution of Thr93 and Asp94 by hydrophobic amino acids appeared to lead to the formation of a larger hydrophobic patch (Fig. 2). Although it has been suggested that hydrogen bonds between Asp94 and Thr85 or other amino acids are important for the stability of the lid-open form (Herrgard et al. 2000), we prepared another combinatorial library of Thr93Xaa, Asp94Xaa. The Thr93 and Asp94 combinatorial library of wild-type ROL was constructed and screened on plates containing soybean oil, as before. A total of about 6.0×103 clones were obtained on the plates. All of the 400 possible combinations of the two positions were comprehensively screened. Some of the clones produced halos with the following sequences: Thr93, Asp94 (wild type), Thr93, Ser94 (clone 3-7) and Ser93, Ser94 (clones 3-9). Because all the other clones among the 400 possible combinations for positions 93 and 94 produced no halos, only these three types could exhibit hydrolytic activity toward triacylglycerol. Clones 3-7 and 3-9 exhibited a much higher substrate specificity toward short-chain fluorescence substrates (C=4). Clone 3-9 exhibited a relative activity toward C=4 that was two times higher than that of the wild type (Fig. 1).

Molecular modeling of ROL mutants 3-7 and 3-9 with fluorescein dibutyrate

The fact that the substitution of Asp94 with other amino acids, except for Ser, caused a loss of activity suggests that hydrogen bonds between Asp94 and other amino acids might be necessary for its function. We therefore carried out modeling of the structure of ROL mutant 3-7–fluorescein dibutyrate and 3-9–fluorescein dibutyrate complexes, in an attempt to determine why Asp94Ser exhibited hydrolytic activity toward short-chain substrates (Figs. 3, 4). In the wild-type ROL, Asp94 interacts tightly with Thr85 (at a distance of 2.86 Å), and this interaction serves to stabilize the lid-open form. Moreover, Thr85N forms an oxyanion hole (distance from the carbonylic oxygen of the substrate is 2.82 Å; Fig.  3a), which is necessary for serine protease family enzymes to stabilize the tetrahedral intermediate produced in the hydrolytic reaction caused by the catalytic triad (Cygler and Schrag 1997). Although the oxyanion hole usually consists of main-chain NH groups, it has been reported that the side-chain OH of Ser contributes to the oxyanion hole (Carlos et al. 2000). Ser94 in mutant 3-7 also interacts with Thr85 (at a distance of 2.98 Å) and both Ser94 Oγ and Thr85 Oγ form a new oxyanion hole (distances from the carbonylic oxygen of the substrate are 3.03 Å from Thr85 Oγ, 3.09 Å from Ser94 Oγ; Fig. 3b). Only mutants 3-7 and 3-9 are able to create a unique oxyanion hole using the hydroxyl groups of Ser94 and Thr85; and these mutants also exhibit hydrolytic activities.
https://static-content.springer.com/image/art%3A10.1007%2Fs00253-005-1935-0/MediaObjects/253_2005_1935_Fig3_HTML.gif
Fig. 3

Models of oxyanion holes of wild-type ROL (a) and mutant 3-7 ROL (b). Hydrogen bonds of Thr85 and Asp94 (wild type, a) and Thr85 and Ser94 (mutant 3-7, b) are represented by black lines. Potential hydrogen bonds between donors [Thr85N in wild type (a), Thr85 Oγ, Ser94 Oγ in mutant 3-7 (b)] and the carbonyl oxygen of substrates are represented by red dashed lines

In the case of mutant 3-7, the hydrophobic patch formed by rearrangement of the lid is narrow, compared with the wild type, because the lid cannot open as fully as the wild type (Fig. 4a). The hydrophobic patch formed by the not completely opened lid is able to interact with short-chain substrates more tightly than the wild type, but long-chain substrates are prevented from encountering the catalytic triad because of hindrance by the narrow hydrophobic patch. As a result, clone 3-7 exhibits a short-chain substrate specificity. In mutant 3-9, a water molecule occupies the space created by the substitution of Thr93Ser and makes the lid opening more narrow, resulting in mutant 3-9 having a greater short-chain substrate specificity than mutant 3-7 (Fig. 4b).
https://static-content.springer.com/image/art%3A10.1007%2Fs00253-005-1935-0/MediaObjects/253_2005_1935_Fig4_HTML.gif
Fig. 4

Comparison of the lid-openness of the wild type and ROL mutant 3-7 (a) and the lid-openness of mutants 3-7 and 3-9 (b). Amino acid backbones are represented by ribbon models

Although substitutions of Thr93 and Asp94 with hydrophobic amino acids could not be obtained in this screening, the combinatorial library of multi-point amino acids created mutants having an entirely new oxyanion hole and these mutants exhibited unique substrate specificities. These results cannot be predicted by computer modeling, and the findings suggest that combinatorial mutagenesis coupled with cell surface display represents an effective approach to protein engineering.

Discussion

Mutated lipases exhibiting a high substrate specificity toward short-chain substrates and having a unique oxyanion hole were obtained from a combinatorial library of the lid domain of ROL through a combination of a yeast display system, combinatorial mutagenesis and computer modeling. ROL were expressed functionally in S. cerevisiae (Takahashi et al. 1998) or E. coli (Beer et al. 1998) and also displayed functionally on the surface of S. cerevisiae (Washida et al. 2001). Although some mutagenesis studies on ROL were reported using an E. coli expression system (Beer et al. 1996) or a yeast display system (Shibamoto et al. 2004), they used point mutagenesis or error-prone PCR, not combinatorial mutagenesis. In combinatorial mutagenesis, several amino acids or domains that appear to be important for the function or stability of the protein can be comprehensively and simultaneously mutated. Rapid screening for the function of the combinatorial library using the yeast display system is capable of supplying a large data set relating to how the structure of an objective domain affects protein function; and computer modeling (in silico) based on combinatorial mutants fundamentally results in a closer understanding of this structure–function relationship. In this study, practically, we created mutant lipases with higher substrate specificity to C=4 substrates. These mutants could be useful in the production of flavors containing short-chain fatty acid esters in various industrial products (cosmetics, food et al.). Moreover, lipases with a high substrate specificity to middle- or long-chain fatty acid esters could be prepared with similar strategies.

Copyright information

© Springer-Verlag 2005