New Efficient Recombinant Expression System To Engineer Candida antarctica Lipase B

Abstract

Here, we report the use of Yarrowia lipolytica as a versatile expression host for developing protein engineering approaches to modify the properties of Candida antarctica lipase B. A reliable screening protocol was defined and validated using a saturation mutagenesis library, yielding mutants displaying higher catalytic efficiencies than the wild-type enzyme.

Lipase B from Candida antarctica (CalB) is one of the most widely used enzymes in industrial biocatalysis (1, 4, 10). Modifying its properties by using protein engineering methods (including rational engineering, directed evolution, or the combination of both) (9, 20) is therefore a challenging goal for many research groups nowadays. Due to the facility of genetic manipulation, high transformation efficiency, and rapid growth rates, Escherichia coli remains to date the most popular expression host in such experimental strategies. However, expression of CalB as an active enzyme in E. coli is a challenge since most of the protein is insoluble and found in inclusion bodies (2, 11, 14, 16, 24). Eukaryotic expression using several host systems, such as Aspergillus oryzae (7, 28), Saccharomyces cerevisiae (27, 29), Pichia pastoris (11, 21) and Hansenula polymorpha (26), has also been described previously. However, to date, very few reports describe the engineering of CalB toward improved expression (26), increased thermostability (5, 27, 29), or modified enantioselectivity (18, 19, 25). Several problems, including production of insoluble proteins in E. coli, low secretion efficiency in S. cerevisiae, and poor transformation efficiency in P. pastoris, have limited the development of efficient screening protocols and thus the identification of CalB variants with interesting properties.

The yeast Yarrowia lipolytica has proven to be an efficient system for recombinant protein expression (15). Recently, a high-throughput screening procedure with the endogenous lipase Lip2 as a model protein was optimized using a strain (JMY1212) enabling single-copy cassette integration by homologous recombination into the genome at a zeta docking platform (derived from the Ylt1 retrotransposon) (23) with high transformation efficiency (8,000 transformants/μg of DNA) (3). Here, as a prerequisite for future protein engineering experiments, we used this approach to develop a new secretion system for CalB allowing the reliable screening of libraries as well as the rapid purification and characterization of interesting mutants (see detailed descriptions of experimental procedures in the supplemental material).

The DNA sequence encoding CalB without its natural N-terminal prepro region was PCR amplified from the pColdIII/CalB plasmid (14). This product was then fused by overlap extension PCR to the DNA sequence encoding the prepro region of Y. lipolytica Lip2 (Preprolip2), which targets the protein to the secretory pathway. The final fusion product was cloned into the JMP62-His₆Cter vector, yielding a plasmid with a cassette containing the Ura3 selection marker and the Preprolip2-CalB-His₆ fusion gene under the control of the oleic acid-inducible POX2 promoter (Fig. 1A). Upon NotI digestion, the JMY1212 strain was transformed with this expression cassette for single-copy integration into the genome (3).

FIG. 1.

Secretion of CalB-His₆ by Y. lipolytica and subsequent protein purification. (A) Schematic diagram of the expression cassette for CalB-His₆ released from the shuttle plasmid JMP62-CalB-His₆ upon NotI digestion. The cassette is flanked by zeta regions, allowing single-copy insertion into the genome of the Y. lipolytica JMY1212 strain. (B) SDS-PAGE analysis of CalB-His₆ secretion and affinity purification. Lanes: 1, Y. lipolytica culture supernatant; 2, flowthrough; 3 and 4, washes; 5, wash with 5 mM imidazole; 6 and 7, washes with 10 mM imidazole; 8 and 9, elution with 100 mM imidazole.

Four Ura-positive colonies were randomly chosen for inoculation into 20-ml flask cultures with oleic acid as an inducer. After 48 h of growth, the culture media were assayed for lipase activity toward para-nitrophenyl butyrate (p-NPB), yielding on average 513 ± 34 U/liter of culture (mean ± standard deviation). SDS-PAGE (Fig. 1B, lane 1) confirmed that CalB-His₆ was the major secreted protein. The production of the lipase was then scaled up in a 3-liter reactor by using an optimized culture procedure described previously (12). Under these conditions, CalB-His₆ was produced in amounts 10-fold larger than those in flask cultures (lipase activity, 5,090 ± 136 U/liter of culture; amount of protein, 190 ± 5 mg/liter of culture). An easy one-step procedure was also set up to purify the recombinant protein (to >95% purity) from the culture medium by using immobilized-metal affinity chromatography (Fig. 1B, lanes 8 and 9).

The culture conditions enabling the secretion of CalB-His₆ were miniaturized in the 96-well microplate format to set up a reliable high-throughput screening protocol (Fig. 2A; see also the detailed protocol in the supplemental material). Briefly, precultures in a 96-well microplate were grown for 24 h with glucose as the carbon source to allow homogeneous growth. The preculture microplate was then replicated to start a production culture in the presence of oleic acid. The protein production was optimized in a 96-deep-well format to obtain larger amounts of enzymes in one step for use in the second stage, assaying the activities of each variant on several substrates. After 30 h of growth, the supernatants containing the secreted recombinant enzymes were assayed for p-NPB hydrolysis. The procedure reproducibility was analyzed with experimental data collected using transformants expressing wild-type CalB-His₆ or the inactive S105A mutant (columns 2 to 7 or 9 to 11 in Fig. 2B, respectively). No growth due to cross-contamination was observed in uninoculated wells (columns 1, 8, and 12 in Fig. 2B). The mean wild-type activity calculated from results of repeated experiments was 199 ± 29 U/liter of culture (coefficient of variation [CV], 14.6%; n = 144). The enzyme production in deep-well microplates is thus 2- to 3-fold lower than that in flask cultures, which provide better aeration conditions. The probability of measuring wild-type activities at or higher than a defined threshold was calculated by assuming that the activity measurements follow a Gaussian distribution (22). This probability represents the expected frequency of false-positive transformants in screening for variants with improved activity (above the defined threshold). With a CV of 14.6%, selection of at least one false-positive variant would be a very rare event (approximately 1 in 1.5 × 10¹¹ screened variants) if our objective was to isolate mutants displaying 2-fold-increased activity. The miniaturized growth conditions described here thus resulted in satisfactorily reproducible activity levels for transformants expressing wild-type CalB, thereby strongly limiting the occurrence of false-positive variants during screening experiments.

FIG. 2.

High-throughput procedure to screen libraries of CalB variants. (A) Schematic flowchart of the screening protocol. The use of the p-NPB substrate is shown as an example. YTD medium, 10 g · liter⁻¹ yeast extract, 10 g · liter⁻¹ Bacto tryptone, and 10 g · liter⁻¹ glucose; YTO medium, 10 g · liter⁻¹ yeast extract, 20 g · liter⁻¹ Bacto tryptone, 20 g · liter⁻¹ oleic acid, and 50 mM Na/K phosphate buffer, pH 6.8. (B) Typical microplate profile of activities toward p-NPB obtained for wild-type CalB (columns 2 to 7) and the S105A inactive mutant (columns 9 to 11). Data were normalized with respect to the mean calculated for wild-type CalB activities and were plotted in order of well position (columns) or in descending order (squares). Columns 1, 8, and 12 represent samples containing culture medium but not inoculated with yeast. μ and σ represent the mean activity toward p-NPB for wild-type CalB and the standard deviation (n = 144), respectively.

To validate our screening system, we chose to create a small library of CalB variants by using saturation mutagenesis and screen the variants for p-NPB hydrolysis. Residue A281 was chosen based on findings from molecular docking studies that aimed at predicting the conformation of the substrate in the binding pocket and estimating the strength of the enzyme-substrate interaction (see Fig. S1A in the supplemental material). Following the construction of the library, 94 Ura-positive transformants were chosen randomly to cover at least 95% of the diversity (6) and screened for p-NPB hydrolysis. This experiment yielded 13 variants (positive hit rate, ∼15% of the library) with at least 3-fold-higher activity than the wild-type enzyme. The four best mutants were isolated for further characterization. Analysis of their sequences revealed that A281 was replaced by either D, N, F, or Y. Resulting kinetic constants for p-NPB hydrolysis are listed in Table 1 (see also Fig. S2 in the supplemental material). Overall, these variants displayed 4.2- to 5.8-fold-higher catalytic efficiency (k_cat/K_m ratios) than the parental lipase. A directed evolution experiment had previously led to the identification of double and triple mutants (A281E/V210I and A281E/V210I/V221D enzymes) displaying 4.4- and 3.2-fold-higher k_cat/K_m ratios for p-NPB hydrolysis than the wild-type enzyme, respectively (29). Here, we thus identified variants possessing comparable and greater enhancement through a single substitution at position 281. To gain some insight into the effects of the mutations, we analyzed the three-dimensional models of p-NPB covalently docked into the different isolated single mutants. Comparisons with the docking mode predicted for p-NPB with the parental enzyme (see Fig. S1A in the supplemental material) showed that the mutations introduced at position 281 (D, N, F, and Y) had clear effects on the orientation of p-NPB in the catalytic pocket and, more importantly, on the stabilization of the tetrahedral intermediate complex. Additional hydrogen bonding interactions between p-NPB and amino acid residues (I189 and V190) of the catalytic pocket of CalB variants that could assist in the departure of the leaving para-nitrophenyl group during hydrolysis (see, for example, data for A281Y in Fig. S1B in the supplemental material) were also observed. These interactions may result in a favorable enthalpic contribution to the free energy of the system and thus explain the enhancement of the catalytic efficiency of the variants compared to that of wild-type CalB.

TABLE 1.

Kinetic constants of wild-type CalB and selected variants for activity toward p-NPB^a

Lipase	Amino acid change	Km (μM) (% of wt value)	kcat (s−1) (% of wt value)	kcat/Km ratio (min−1·μM−1) (% of wt value)
wt CalB		583 (100)	110 (100)	11.3 (100)
A9	A281D	614 (105)	581 (528)	56.7 (502)
B5	A281F	444 (76)	463 (421)	62.6 (554)
F12	A281N	413 (71)	324 (295)	47.1 (416)
H3	A281Y	491 (84)	537 (488)	65.6 (580)

^awt, wild-type.

Our results show that CalB is secreted by Y. lipolytica in relatively large amounts (up to 200 mg/liter of culture) by using a single-copy integration strategy. This approach was chosen to allow reproducible expression levels compatible for the screening of variant libraries. Note that the production rate can be easily increased using multicopy vectors (17). High-yield CalB production (up to 1.3 g/liter of culture) in P. pastoris was achieved previously by using multicopy gene integration into the genome (8, 21). However, low transformation efficiencies (∼10² transformants/μg of DNA) and high expression variability among transformants make this host generally unsuitable for screening experiments. Expression of a FLAG-tagged CalB recombinant form in S. cerevisiae (yielding ∼80 mg/liter of culture) provides a more reliable screening system (29). A 7-day growth period, however, is needed to achieve enzyme production in microplates, whereas only 2 days were necessary in our protocol. Moreover, due to the expense of FLAG tag affinity chromatography (13), only small amounts of protein can be purified for characterization upon screening. In contrast to other prokaryotic and eukaryotic expression systems described previously (14, 21, 29), the use of Y. lipolytica enabled us to establish efficient procedures for both the construction and the screening of CalB variant libraries, as well as for the overproduction, secretion, and purification of the lipase necessary for the rapid characterization of interesting variants. The methods described here will now be used in-house to create and screen larger libraries of CalB mutants in order to identify new variants displaying original properties.