Combinatorial mutation on the β-glycosidase specific to 7-β-xylosyltaxanes and increasing the mutated enzyme production by engineering the recombinant yeast.

Taxol is a "blockbuster" antitumor drug produced by Taxus species with extremely low amount, while its analogue 7-β-xylosyl-10-deacetyltaxol is generally much higher in the plants. Both the fungal enzymes LXYL-P1-1 and LXYL-P1-2 can convert 7-β-xylosyl-10-deacetyltaxol into 10-deacetyltaxol for Taxol semi-synthesis. Of them, LXYL-P1-2 is twice more active than LXYL-P1-1, but there are only 11 significantly different amino acids in terms of the polarity and acidic-basic properties between them. In this study, single and multiple site-directed mutations at the 11 sites from LXYL-P1-1 to LXYL-P1-2 were performed to define the amino acids with upward bias in activities and to acquire variants with improved catalytic properties. Among all the 17 mutants, E12 (A72T/V91S) was the most active and even displayed 2.8- and 3-fold higher than LXYL-P1-2 on β-xylosidase and β-glucosidase activities. The possible mechanism for such improvement was proposed by homology modeling and molecular docking between E12 and 7-β-xylosyl-10-deacetyltaxol. The recombinant yeast GS115-P1E12-7 was constructed by introducing variant E12, the molecular chaperone gene pdi and the bacterial hemoglobin gene vhb. This engineered yeast rendered 4 times higher biomass enzyme activity than GS115-3.5K-P1-2 that had been used for demo-scale fermentation. Thus, GS115-P1E12-7 becomes a promising candidate to replace GS115-3.5K-P1-2 for industrial purpose.

Introduction

Glycoside hydrolases (GHs, EC 3.2.1), also known as glycosidases, are enzymes that can hydrolyze the glycosidic bonds in glycoconjugates1., 2., 3., 4.. GHs can be categorized into over 100 families based on the amino acid sequences of different species. Of them, β-xylosidases can mainly be classified into the 3, 30, 39, 43, 52 and 54 GH families. β-Xylosidases, the exogenous glycosidases that are widely distributed in nature, have been isolated from bacteria, actinomyces, fungi and some higher plants6., 7.. They can not only catalyze the hydrolysis of hemicellulosic xylans and remove the successive d-xylose residues from the non-reducing termini, but can also hydrolyze the glycosidic bonds between the xylose residue and aglycone of β-xylosyl-terpenes or steroids. Taxol (generic name: paclitaxel) refers to a diterpenoid compound with definite anti-cancer activity8., 9., 10.. At present, Taxol from nursery cultivated Taxus species accounts for one of the main sources for clinical supply. However, its natural content is extremely low, which ranges from 0 to 0.069%, and the highest content is found in the bark. In comparison, the content of its analogue 7-β-xylosyl-10-deacetyltaxol (XDT) is generally much higher than that of Taxol. Typically, if the xylosyl group of XDT is removed, the corresponding 10-deacetyltaxol (DT) can be converted into Taxol by acetylation at the C10 position. However, such analogue is generally discarded during Taxol extraction, leading to both resource waste and potential environmental pollution. In our previous study, two variants, LXYL-P1−1 and LXYL-P1−2 with dual GH3 β-xylosidase/β-glucosidase activities, were cloned from Lentinula edodes (strain M95.33) and characterized. Most importantly, such enzymes can specifically remove the xylosyl group from XDT to produce DT for the semi-synthesis of Taxol (Scheme 1)12., 13..

Scheme 1

Bioconversion of 7-β-xylosyl-10-deacetyltaxol to 10-deacetyltaxol for the semi-synthesis of Taxol.

Particularly, the activity of LXYL-P1−2 is twice higher than that of LXYL-P1−1. However, sequence analysis shows that both open reading frames (ORFs) have encoded 803 amino acids with over 97% sequence identity as well as 21 different amino acids. Moreover, all the 21 different amino acids are distributed in the first 386 positions of the two polypeptides. Typically, 11 of these 21 different amino acids have exhibited significant differences in terms of their polarity and acidic-basic properties (Fig. 1). For instance, the positions of 3, 69, 192, 310, 318, 322, 327 and 334 are occupied by hydrophilic polar or acidic amino acids in LXYL-P1−1, while the corresponding sites in LXYL-P1−2 are taken up by hydrophobic non-polar or basic amino acids. On the contrary, amino acids in the positions 72, 91 and 368 are hydrophobic and non-polar in LXYL-P1−1, whereas those are hydrophilic polar amino acids in LXYL-P1−2 (Fig. 1). Therefore, some of these amino acids may affect the enzyme activity of both variants, although they are not involved in the substrate binding and catalytic sites. Specifically, the site-directed mutations of these amino acids from the low-active LXYL-P1−1 to the high-active LXYL-P1−2 can not only contribute to defining the amino acids with upward bias in activity, but can also make for obtaining the β-glycosidase variants with higher activities.

Figure 1

Polarity and acidic-basic properties of the 21 differential amino acids between LXYL-P1−1 and LXYL-P1−2. The amino acids with different polarity and acidic—basic properties are indicated in red asterisk.

Pichia pastoris, the methylotrophic yeast, has been frequently used for the heterologous expression of a variety of industrially relevant biocatalysts and pharmaceutical proteins14., 15., 16.. As an aerobic microorganism, P. pastoris requires continuous oxygen supply and efficient utilization for the expression of recombinant proteins, which is particularly true at the fermentation stage of P. pastoris with high cell density. Vitreoscilla hemoglobin (VHB, coding gene: vhb) is produced by the gram-negative bacterium Vitreoscilla18., 19., 20., 21., 22.. Wu et al. reported that, the intracellular co-expression of vhb regulated by methanol-inducible promoter contributed to cell growth and β-galactosidase production in P. pastoris. In addition, Chen et al. also suggested that, the co-expression of vhb could promote cell growth and SAM production, which was achieved through enhancing the ATP synthetic rate in the recombinant P. pastoris. Therefore, the heterologous expression of vhb may alleviate oxygen limitation at the later fermentation stage, thus facilitating cell density and protein production. Protein disulfide isomerase (PDI), which belongs to the molecular chaperone, participates in protein folding as well as assembly and contributes to the rearrangement of incorrect disulfide pairings25., 26.. Typically, the initial eukaryotic transcription process is sophisticated and is achieved under the assistance of many protein factors. UPC (sterol uptake control protein), a zinc finger transcription factor, regulates the ergosterol homeostasis through binding to the promoter region of eukaryotic gene with cis-acting elements. UPC exerts no direct regulatory action on the heterologous expression of protein; nonetheless, it may affect protein production through interacting with other unknown target proteins. Therefore, the co-expression of the above factors with β-glycosidase gene may up-regulate the enzyme expression level. The β-glycosidases LXYL-P1−1 and LXYL-P1−2 have been successfully expressed in P. pastoris, and the recombinant yeast GS115-3.5K-P1−2 can be used as a biocatalyst to convert XDT into DT for the semi-synthesis of Taxol. On this account, it is of practical significance to further enhance the biomass activity to reduce the costs of bioconversion.

In this study, the 11 aforementioned amino acids were changed from LXYL-P1−1 to LXYL-P1−2 by single and multiple site-directed mutations, so as to identify the key amino acids with upward bias in enzyme activity and to obtain variants with improved catalytic properties. Besides, genes encoding PDI (P. pastoris origin), VHB (Vitreoscilla origin) and UPC (S. cereviase origin) that were driven by inducible or constitutive promoters were introduced into the recombinant Pichia pastoris that expressed the mutated highly-active enzyme. Subsequently, the effects of these factors on the biomass activity were explored. Finally, the recombinant strains with apparently improved biomass activities could be obtained.

Materials and methods

Plasmids and strains

All strains and plasmids used in this study are listed in Table 1 . Pichia pastoris GS115 strain and plasmids (pPIC3.5 K, pPICZA, and pPIC6A) were purchased from Invitrogen. The recombinant plasmids pPIC3.5K-LXYL-P1−1 and pPIC3.5K-LXYL-P1−2 were previously constructed in our laboratory. Pichia pastoris GS115-3.5K-P1−1 and GS115-3.5K-P1−2 were constructed by transforming the host strain P. Pastoris GS115 (Mut+) with the plasmids pPIC3.5K-LXYL-P1−1 and pPIC3.5K-LXYL-P1−2, and preserved at −80 °C prior to use.

Table 1

List of strains and plasmids used in this study.

Strain or plasmid	Description	Source
Srain
GS115	Mut+, His−	Invitrogen
GS115-3.5K-P1−1	GS115 integrated with pPIC3.5K-LXYL-P1−1, Mut+, His+, GeneticinR	Ref. 12
GS115-3.5K-P1−2	GS115 integrated with pPIC3.5K-LXYL-P1−2, Mut+, His+, GeneticinR	Ref. 12
GS115-P1E1~E17	GS115 integrated with pPIC3.5K-E1~E17, Mut+, His+, GeneticinR	This study
GS115-P1E12-1	GS115-P1E12 integrated with pPICZA-pdi, Mut+, His+, GeneticinR, ZeocinR	This study
GS115-P1E12-2	GS115-P1E12 integrated with pGAPZA-pdi, Mut+, His+, GeneticinR, ZeocinR	This study
GS115-P1E12-3	GS115-P1E12 integrated with pPICZA-vhb, Mut+, His+, GeneticinR, ZeocinR	This study
GS115-P1E12-4	GS115-P1E12 integrated with pGAPZA-vhb, Mut+, His+, GeneticinR, ZeocinR	This study
GS115-P1E12-5	GS115-P1E12 integrated with pPICZA-upc, Mut+, His+, GeneticinR, ZeocinR	This study
GS115-P1E12-6	GS115-P1E12 integrated with pGAPZA-upc, Mut+, His+, GeneticinR, ZeocinR	This study
GS115-P1E12-7	GS115-P1E12 integrated with pPICZA-vhb and pPIC6A-pdi, Mut+, His+, GeneticinR, ZeocinR, BlasticidinR	This study
GS115-P1E12-8	GS115-P1E12 integrated with pGAPZA-vhb and pPIC6A-pdi, Mut+, His+, GeneticinR, ZeocinR, BlasticidinR	This study
GS115-P1E12-9	GS115-P1E12 integrated with pGAPZA-pdi and pPIC6A-vhb, Mut+, His+, GeneticinR, ZeocinR, BlasticidinR	This study
Plasmid
pPIC3.5 K	PAOX1, TAOX1, pBR322 ori, His4, AmpR, GeneticinR	Invitrogen
pPICZA	PAOX1, TAOX1, pUC ori, ZeocinR	Invitrogen
pGAPZA	PGAP, TAOX1, pUC ori, ZeocinR	Invitrogen
pPIC6A	PAOX1, TAOX1, pUC ori, BlasticidinR	Invitrogen
pPIC3.5K-LXYL-P1−1	pPIC3.5 K with Lxyl-p1−1, GeneticinR, His+	Ref. 12
pPIC3.5K-LXYL-P1−2	pPIC3.5 K with Lxyl-p1−2, GeneticinR, His+	Ref. 12
pPIC3.5K-E1-E17	pPIC3.5 K with E1-E17 mutated from Lxyl-p1−1, GeneticinR, His+	This study
pPICZA-pdi	pPICZA with pdi (PAOX1), ZeocinR	This study
pPICZA-vhb	pPICZA with vhb (PAOX1), ZeocinR	This study
pPICZA-upc	pPICZA with upc (PAOX1), ZeocinR	This study
pGAPZA-pdi	pGAPZA with pdi (PGAP), ZeocinR	This study
pGAPZA-vhb	pGAPZA with vhb (PGAP), ZeocinR	This study
pGAPZA-upc	pGAPZA with upc (PGAP), ZeocinR	This study
pPIC6A-pdi	pPIC6A with pdi (PAOX1), BlasticidinR	This study
pPIC6A-vhb	pPIC6A with vhb (PAOX1), BlasticidinR	This study

Single and multiple site-directed mutagenesis of LXYL-P1−1

The primers with single or multiple site-directed mutations are listed in Supporting Information Table S1. The single site-directed mutations in E1—E11 were conducted through whole-plasmid amplification PCR. Meanwhile, the expression plasmid pPIC3.5K-LXYL-P1−1, which harbored the glycoside hydrolase gene Lxyl-p1−1, was used as a template in the PCR reactions. PCR was performed with Phusion DNA polymerase (NEB, Ipswitch, MA, USA) using the following pairs of primers: S3P-F1/S3P-R1, S69A-F2/S69A-R2, A72T-F3/A72T-R3, V91S-F4/V91S-R4, Q192L-F5/Q192L-R5, S310A-F6/S310A-R6, G318A-F7/G318A-R7, E322K-F8/E322K-R8, E327A-F9/E327A-R9, T334A-F10/T334A-R10 and I368T-F11/I368T-R11, respectively. The PCR conditions for amplification consisted of 98 °C for 30 s, 30 cycles of 10 s at 98 °C, 30 s at 52—72 °C (each cycle increased 0.7 °C), 6 min at 72 °C, and a final 10 min extension at 72 °C. The PCR products were purified and digested by Dpn I at 37 °C for 3 h and then transformed into DMT competent cells (Transgen, Beijing, China). After the sequences were confirmed, the corresponding vectors were referred as pPIC3.5K-E1–E11.

Additionally, the multiple site-directed mutations of LXYL-P1−1 were performed using the PCR-based overlap extension method28., 29.. To construct the double mutant E12, the two individual fragments were amplified by Phusion DNA polymerase using primers P1−1-F/V91S-R and V91S-F/P1−1-R, respectively, with the plasmid pPIC3.5K-E3 being used as a template. Similarly, the PCR conditions for amplification consisted of 98 °C for 30 s, 30 cycles of 10 s at 98 °C, 30 s at 60 °C, 1 min at 72 °C, and a final 10 min extension at 72 °C. The PCR products were separated by gel electrophoresis and were purified using a gel extraction kit (Transgen, Beijing, China). Later, the overlap extension was performed by mixing 100 ng of the two fragments in equimolar amounts with Phusion PCR buffer, dNTPs, and Phusion polymerase in a total volume of 25 μL. The PCR conditions for amplification were 98 °C for 30 s, 15 cycles of 10 s at 98 °C, 30 s at 60 °C, 72 °C for 30 s/kb, followed a 10 min incubation at 72 °C. Then, 2 μL of the unpurified PCR product was further used as a template for the second round PCR. Additionally, P1−1-F and P1−1-R, Phusion PCR buffer, dNTPs, and Phusion polymerase were added into the PCR mixture in a final volume of 50 μL. The amplification was performed identically to the PCR reaction of the individual fragments. The fusion PCR products were separated by gel electrophoresis and purified using a gel purification kit, as described above. Finally, the fragment E12 containing the A72T and V91S mutations was obtained. Then E12 was ligated into the expression vector pPIC3.5 K at the BamHI and NotI restriction sites to generate the expression plasmid pPIC3.5K-E12. Accordingly, the plasmid pPIC3.5K-E12 was used as a template, and the two individual fragments were amplified using primers P1−1-F/I368T-R and I368T-F/P1−1-R, respectively, and fused by overlap extension PCR to obtain E13. Then the E13 was digested and ligated into pPIC3.5 K, so as to obtain the expression plasmid pPIC3.5K-E13 by the same methods mentioned above. For the construction of E14, the plasmid pPIC3.5K-LXYL-P1−1 was used as a template, and the three separate fragments were amplified using primers S3P-F/S69A-A72T-R, S69A-A72T-F/Q192L-R and Q192L-F/P1−1-R, respectively. Afterwards, these fragments were fused by the overlap extension PCR to gain E14. For the construction of E15, the plasmid pPIC3.5K-E4 was used as a template, and the three individual fragments were amplified using primers S3P-F/S69A-A72T-R, S69A-A72T-F/Q192L-R, Q192L-F/S310A-G318A-R and S310A-G318A-F/P1−1-R, respectively, and the four independent fragments were fused by overlap extension PCR to gain E15. Accordingly, E14 and E15 were digested and ligated into pPIC3.5 K to obtain the expression plasmids pPIC3.5K-E14 and pPIC3.5K-E15, respectively. For constructing the multiple site-directed mutations in E16, the plasmid pPIC3.5K-E15 was used as a template, and the three fragments were amplified using primers S3P-F/S310A-G318A-E322K-R, E322K-T327A-F/I368T-R and I368T-F/P1−1-R, respectively. The three fragments were fused by overlap extension PCR to gain E16. Consistently, E17 was constructed in the same way using with the primers S3P-F/S310A-G318A-E322K-R, E322K-T327A-T334A-F/I368T-R and I368T-F/P1−1-R, respectively. Finally, E16 and E17 were digested and ligated into pPIC3.5 K to obtain the expression plasmids pPIC3.5K-E16 and pPIC3.5K-E17, respectively. Eventually, all the recombinant plasmids with multiple site-directed mutations were confirmed by nucleotide sequence analysis.

Expression of LXYL-P1−1 variants in P. pastoris

The engineered P. pastoris strains containing Lxyl-p1−1 variants were constructed as mentioned previously. Briefly, 10 μg recombinant vectors pPIC3.5K-E1—E17 were linearized with SacI and then introduced into P. pastoris GS115 via electroporation transformation according to the manufacturer׳s instructions (Invitrogen, USA). The transformants were initially selected on MD plates (containing 13.4 g/L yeast nitrogen base, 0.4 mg/L biotin, 20 g/L dextrose, and 15 g/L agar) and then were screened for recombinants with multiple gene integration on YPD plates (containing 10 g/L yeast extract, 20 g/L tryptone, 20 g/L d-glucose, and 15 g/L agar) supplemented with 4 mg/mL G418. The recombinant yeasts harboring the Lxyl-p1−1 variants were firstly inoculated in 500-mL shake flasks containing 100 mL buffered minimal glycerol complex medium (BMGY) medium (containing 10 g/L yeast extract, 20 g/L tryptone, 13.4 g/L YNB, 0.4 mg/L biotin, 10 g/L glycerol, 100 mmol/L potassium phosphate buffer, pH 6.0) at 30 °C with shaking at 200 rpm for 48–60 h. Then methanol was added every day to maintain 1% (v/v) for the induction of the gene expression. After 7 days of induction, the recombinant enzymes were isolated and purified according to the method described in our previous report.

β-Xylosidase and β-glucosidase activity analysis of LXYL-P1−1 mutants

The β-xylosidase and β-glucosidase activities of the purified LXYL-P1−1mutants were measured by detecting the amount of p-nitrophenol released from the substrate p-nitrophenyl-β-d-xylopyranoside (PNP-Xyl) or p-nitrophenyl-β-d-glucopyranoside (PNP-Glc) under the optimum reaction conditions. Concretely, the 60 μL reaction volume contained 50 μL of 5 mmol/L PNP-Xyl or PNP-Glc and 10 μL of 0.1 mg/mL enzyme in 50 mmol/L sodium acetate buffer with pH 5.0. The reaction was performed under 50 °C for 20 min. Reactions were terminated by adding 1 mL saturated Na2B4O7 solution. The enzymatic activity was assayed using spectrophotometry based on the absorbance at 405 nm. One unit of activity was defined as the amount of enzyme that catalyzed the formation of 1 nmol/L p-nitrophenol per minute.

Kinetics of LXYL-P1−1 mutants

The kinetic parameters of the recombinant LXYL-P1−1 mutant enzymes against XDT were determined at the XDT concentration ranging from 0.039—5.0 mmol/L as described previously. DT formation was analyzed through HPLC. The kinetic data on XDT were processed by a proportional weighted fit using a nonlinear regression analysis program based on Michaelis–Menten enzyme kinetics. All data were presented as means±SD of three independent repeats.

Homology modeling of LXYL-P1−1 three-dimensional structure and molecular docking with XDT

The three-dimensional (3-D) model of LXYL-P1−1 was predicted by Swiss Model Server (http://swissmodel.expasy.org/) using the available structure of the Aspergillus fumigatus (Protein Data Bank code 5fji), which has 44% sequence identity and 59% sequence similarity to LXYL-P1−1, as a template. The predicted 3D structure and substrate XDT were selected for the docking experiments with AutoDockTools.

Construction of VHB, PDI and UPC co-expression vectors

All primers used for constructing the co-expression plasmids are shown in Supporting Information Table S2. The pdi gene was amplified from the genomic DNA of P. pastoris GS115 based on the published sequence using primers PDI-F and PDI-R. The vhb gene (GenBank accession number: AF292694) was amplified from the vector pSKVHb using primers VHB-F and VHB-R. The intronless upc2.1 gene (GenBank accession number: CP020126.1) was amplified from the genomic DNA of S. cereviase INVSc1 using primers UPC-F and UPC-R. The purified PCR products were ligated into plasmids pPICZA, pGAPZA and pPIC6A, respectively, and transformed into E. coli Top 10 (Transgen, Beijing, China). The transformants were then selected on the low salt LB plates (containing 10 g/L peptone, 5 g/L yeast extract, 5 g/L NaCl, and 15 g/L agar). Eventually, the final recombinant plasmids were validated by DNA sequencing for correct sequence and direction, and were designated as pPICZA-pdi/vhb/upc, pGAPZA-pdi/vhb/upc and pPIC6A-pdi/vhb, respectively.

Transformation of P. pastoris GS115-P1E12

To construct the engineered strains that co-expressed the molecular chaperone (PDI), Vitreoscilla hemoglobin (VHB) and transcriptional factor (UPC) under the control of P or P, the Sac I linearized pPICZA-pdi, pGAPZA-pdi, pPICZA-vhb, pGAPZA-vhb, pPICZA-upc or pGAPZA-upc (10 μg of each plasmid) were respectively transformed into the strain GS115-P1E12 harboring E12 gene, and the transformants were initially plated on the YPD plates containing 100 μg/mL Zeocin. Then, the multi-copy transformants were screened on the YPD plates containing 1 mg/mL Zeocin. The correct transformants were referred as GS115-P1E12-1, GS115-P1E12-2, GS115-P1E12-3, GS115-P1E12-4, GS115-P1E12-5 and GS115-P1E12-6, respectively. To construct strains that co-expressed the molecular chaperone and the Vitreoscilla hemoglobin simultaneously, 10 μg pPIC6A-pdi digested with Avr II was transformed into GS115-P1E12-3 and GS115-P1E12-4, respectively. The transformants were initially plated on the YPD plates containing 700 μg/mL blasticidin, and the selection was carried out on the YPD plates containing 2 mg/mL blasticidin, which generated the recombinant strains GS115-P1E12-7 and GS115-P1E12-8, respectively. Accordingly, 10 μg pPIC6A-vhb digested with Avr II was transformed into GS115-P1E12-2 to generate the recombinant strains GS115-P1E12-9.

Measurement of β-xylosidase activities of the engineered strains

The engineered yeasts GS115-P1E12-1—9 were further inoculated into the 500-mL shake flasks containing 100 mL BMGY medium. The flasks were incubated at 30 °C and 220 rpm for 48—60 h, and then methanol was added every day to maintain 1% (v/v) for inducing gene expression during the induction period. Meanwhile, the β-xylosidase activity was analyzed based on periodic sampling. The samples were washed twice with dH2O via centrifugation, and the cell pellet was resuspended with dH2O in the same volume of culture broth. 10 μL cell suspension was added to 100 μL of 5 mmol/L PNP-Xyl, and incubated for 20 min at 50 °C for the catalytic activity analysis. The β-xylosidase activity was then evaluated by calculating both U/L (volumetric enzyme activity) and U/g (biomass enzyme activity) as described previously.

Western blotting

The freeze-dried recombinant yeast cells (10 mg) were mixed with 500 μL Yeast Active Protein Extraction buffer (Sangon Biotech Co., Ltd., Shanghai, China) containing phenylmethanesulfonyl fluoride (PMSF) and proteinase inhibitor cocktail, and the mixture was incubated at room temperature for 20 min. Then, the lysed solution was centrifuged for 15 min at 15,000×g, and the supernatant was collected. To quantify the protein concentration, Bradford dye (Bio-rad, CA, USA) was used according to the manufacturer׳s instructions. To remove the carbohydrates from the enzyme, 45 μL of the supernatant was mixed with 5 μL of 10×denatured glycoprotein buffer and boiled at 100 °C for 10 min. 5.5 μL of 10×G5 buffer and 2 μL Endo Hf enzyme (NEB, Ipswitch, MA, USA) were added and incubated at 37 °C for 1 h. Finally, aliquots (10 μg for total protein) of cell lysates were separated on 12% SDS-PAGE, and then transferred onto the PVDF membrane at 200 mA for 100 min. Exposed protein binding sites on the membranes were blocked by incubation in 5% skim milk-Tris-buffered saline (TBS) for 1 h at room temperature. Then, the membranes were further incubated with His-tag mouse monoclonal antibody (1:1000, Applygen Technologies Inc., Beijing, China) at 4 °C overnight. The membranes were then washed with TBST (TBS with 0.05% tween-20) for four times, and incubated with peroxidase-conjugated goat anti-mouse IgG (1:5,000, Applygen Technologies Inc., Beijing, China) for 2 h at room temperature. Afterwards, the membranes were washed with TBST for another four times, and the protein bands were detected using eECL Western blot kit (Cwbiotec, Beijing, China) following the manufacturer׳s instructions. Finally, the protein bands on SDS-PAGE gels were quantified with arbitrary unit using Quantity One software version 4.5.0 (Bio-Rad).

Results

Combining the mutated amino acids from low-active LXYL-P1−1 to the high-active LXYL-P1−2

The 11 aforementioned sites in LXYL-P1−1 (template) were subjected to site-directed mutations in a single or combined manner towards their counterparts in LXYL-P1−2 (Fig. 1). The precise site-directed mutations were shown as follows, Ser3→Pro, Ser69→Ala, Ala72→Thr, Val91→Ser, Gln192→Leu, Ser310→Ala, Gly318→Ala, Glu322→Lys, Glu327→Ala, Thr334→Ala, and Ile368→Thr. Thus, the variants E1 to E17 were generated (Fig. 2).

Figure 2

Specific activities of LXYL-P1−1-derived mutants against PNP-Xyl and PNP-Glc. Data are mean±SD. n=3, *P<0.05 vs LXYL-P1−1, **P<0.01 vs LXYL-P1−1, ***P<0.001 vs LXYL-P1−1; #P<0.05 vs LXYL-P1−2, ##P<0.01 vs LXYL-P1−2, ###P<0.001 vs LXYL-P1−2.

Moreover, the activities of the purified enzymes were measured through hydrolysis of the chromogenic substrates PNP-Xyl and PNP-Glu, respectively. As shown in Fig. 2 and Supporting Information Table S3, apart from E1 (S3P), E5 (Q192L), E6 (E310A) and E7 (E318A), all other single site-directed mutants (variants), including E2 (S69A), E3 (A72T), E4 (V91S), E8 (E322K), E9 (E327A), E10 (T334A) and E11 (I368T), had displayed higher activities of β-xylosidase and β-glucosidase than those of LXYL-P1−1, in which E3 (A72T) showed similar β-d-xylosidase activity but higher β-d-glucosidase activity compared with those of E2 (S69A). These finding suggested the presence of upward bias in their activities after mutation. Moreover, the β-glucosidase activities of E3, E4 and E11, as well as the β-xylosidase activity of E11 even surpassed those of LXYL-P1−2. Subsequently, E3 (A72T) and E4 (V91S) were selected to construct the double mutant E12 (A72T/V91S). Later, the triple mutant E13 (A72T/V91S/I368T) was constructed by combining E12 with E11 (I368T, which came up with LXYL-P1−2 in terms of β-xylosidase and β-glucosidase activities). As shown in Fig. 2 and Supporting Information Table S3, the β-xylosidase and β-glucosidase activities of the double mutant E12 (A72T/V91S) were markedly improved (14.10±1.33 and 35.94±0.63 U/mg×104, respectively), which were 5.7 and 7.2 times as high as those of LXYL-P1−1 (2.46±0.27 and 4.97±0.09 U/mg×104, respectively) and even 2.8- and 3-folds higher than those of LXYL-P1−2 (5.03±0.14 and 12.05±0.49 U/mg×104, respectively). However, no further increase in the enzyme activity was observed in the triple mutant E13 (A72T/V91S/I368T) (2.35±0.22 and 8.19±0.69 U/mg × 104, respectively) compared with those of E12. To investigate whether these mutations had positive synergistic effects, the quadruple mutation (E14), septuple mutation (E15), decuple mutation (E16), as well as the hendecuple mutation (E17) were constructed, respectively, in which E17 contained all of the eleven aforementioned mutation sites. Unfortunately, the activities of these 4 mutants were shown to be lower than those of LXYL-P1−2, except that the β-xylosidase activity of E16 and the β-glucosidase activity of E17 were 61% and 66% higher than those of LXYL-P1−1, respectively (Fig. 2 and Supporting Information Table S3).

Kinetic analysis of the mutated enzymes against XDT

The β-xylosidase catalytic efficiencies (kcat/Km: [s—1· (mmol/L)—1]) against substrate XDT were evaluated in the double mutant E12, as well as the single mutants E3 and E4, respectively. As summarized in Table 2, the Km values of LXYL-P1−1 and LXYL-P1−2 (controls) were similar, but the substrate turnover value (kcat, calculated from Vmax) of LXYL-P1−2 was 2.3 times as high as that of LXYL-P1−1. As a result, the catalytic efficiency of LXYL-P1−2 was 2.6 times higher than that of LXYL-P1−1. The affinity of E3 (A72T) towards XDT was apparently lower than those of the two controls, but its turnover value (5.35 s—1) had surpassed that of LXYL-P1−1 (2.13 s—1) by 2.5 times and was even higher than that of LXYL-P1−2 (4.89 s—1). Therefore, E3 had displayed higher catalytic efficiency relative to that of LXYL-P1−1 (4.54 vs. 3.72 [s—1·(mmol/L)—1]). Similarly, the V91S mutation in E4 resulted in a 1.7-fold increase in the catalytic efficiency (6.26 vs. 3.72 [s—1 (mmol/L)—1]), which could be ascribed to the apparent increase in the kcat and the minor increase in the Km. Typically, E12 had exhibited the highest substrate affinity (depicted as the lowest Km value: 0.30 mmol/L), which had surpassed those of LXYL-P1−1 (Km: 0.57 mmol/L) and LXYL-P1−2 (Km: 0.51 mmol/L), but its kcat value was between those of LXYL-P1−1 and LXYL-P1−2, thus endowing it with the highest catalytic efficiency (11.23 [s—1 (mmol/L)—1]) among the three mutants and the native controls.

Table 2

Kinetic parameters for the mutated enzymes using XDT as the substrate.

Parameter	LXYL-P1−1	LXYL-P1−2	E3	E4	E12
Vmax (μmol/L min)	3.55 (±0.08)	8.16 (±0.15)***	8.92 (±0.85)***	7.93 (±0.59)***	5.54 (±0.29)***
Km (mmol/L)	0.57 (±0.02)	0.51 (±0.01)	1.18 (±0.13)***	0.76 (±0.09)*	0.30 (±0.03)**
kcat (s—1)	2.13 (±0.05)	4.89 (±0.09)***	5.35 (±0.51)***	4.75 (±0.35)***	3.32 (±0.17)***
kcat/Km [s—1· (mmol/L)—1]	3.72 (±0.06)	9.67 (±0.10)***	4.54 (±0.04)**	6.26 (±0.30)***	11.23 (±0.56)***

Note: Data are mean (±SD). n = 3,

P<0.05 vs LXYL-P1−1.

P<0.01 vs LXYL-P1−1.

P<0.001 vs LXYL-P1−1.

Substrate–enzyme docking analysis

GH3 β-glucosidase from Aspergillus fumigatus (Protein Data Bank code 5fji) had displayed the highest sequence identity (44%) and sequence similarity (59%) to those of LXYL-P1−1. As a result, it was selected as the template for the homology modeling of the LXYL-P1−1 structure. Typically, the sequence alignment between LXYL-P1−1 and the template was shown in Fig. 3A. Moreover, the three-dimensional structure of LXYL-P1−1 was predicted using molecular modeling (http://swissmodel.Expasy.org/). According to Fig. 3B and C, LXYL-P1−1 possessed a triosephosphate isomerase (TIM) barrel-like domain at N-terminus, which was in accordance with that of LXYL-P1−2, as predicted previously. In addition, it was found that, the 10 aforementioned amino acids (Ser3 was missing from the modeling), which had different properties to those of LXYL-P1−2 were almost located within the N-terminus of the TIM barrel-like domain. Among them, Ser69, Ala72, Val91, Gln192, Glu322, Glu327, Thr334 and Ile368 were located on the surface of the predicted structure, while Ser310 and Gly318 were situated within the structure. In addition, Val91, Ser310, Gly318, Glu322 and Glu327 were close to the catalytic nucleophile Asp300 12., 35..

Figure 3

The homology model of LXYL-P1−1 predicted based on the template GH3 β-glucosidase from Aspergillus fumigatus (PDB ID: 5fji) using the Swiss Model. (A) Sequence alignment between LXYL-P1−1 and the template GH3 β-glucosidase from Aspergillus fumigatus (PDB ID: 5fji). Identical and similar residues are highlighted in blue. (B) Three-dimensional structure view of LXYL-P1−1. α-Helices are shown in cyan, β-strands are in magentas and loops are in salmon. (C) Surface view of LXYL-P1−1. The catalytic sites (Asp300 and Glu529) and the amino acids different from those of LXYL-P1−2 in properties are indicated in blue.

To further explore the potential molecular mechanism of the improved catalytic property of E12 (A72T/V91S), the molecular docking between the enzyme and the substrate XDT was conducted based on the homology modeling. Fig. 4A and B suggested that, residue Ala72 was located on the LXYL-P1−1 surface, and A72T mutation in mutant E12 might lead to conformational change in the loop near XDT, which might positively influence the entrance of XDT. However, residue Val91 was close to the nucleophile Asp300, which was located near the XDT entrance path to the active pocket. Only two internal hydrogen bonds were found between Val91 and the surrounding amino acid residues in LXYL-P1−1, which were Gln108 and Trp301. By contrast, four hydrogen bonds were formed among Ser91, Gln108, Trp301 and substrate XDT in variant E12, which could enhance the enzyme-substrate interaction (Fig. 4C and D). Thus, among the microenvironment of the LXYL-P1−1 structure, the combination of A72T and V91S might positively promote the enzyme-XDT interaction through enhancing the hydrogen bonds and altering the loop conformation. Eventually this would result in the remarkably increased substrate affinity and higher catalytic efficiency (Table 2).

Figure 4

Partial view of enzyme-XDT docking. (A) Side view of LXYL-P1−1 with XDT, showing Ala72, and Val91. (B) Side view of E12 with XDT, in which Ala72 and Val91 are replaced by Thr72 and Ser91, respectively. The loops near the active pocket are indicated in salmon. (C) Hydrogen bonds formed between Val91 and surrounding amino acid residues. (D) Hydrogen bonds formed between Ser91 and surrounding amino acid residues or substrate. Hydrogen bonds are shown as dotted lines. The carbon atoms of XDT are shown in orange. The nucleophile Asp300 (catalytic site), Ala72/Thr72, and Val91/Ser91 are colored in blue.

Engineering yeast for improved β-glycosidase production

The recombinant yeast GS115-3.5K-P1–2 that produced LXYL-P1−2 had been applied for the high-cell-density fermentation and for the biocatalysis of XDT36., 37.. Specifically, E12 obtained in this study had displayed the highest β-xylosidase and β-glucosidase activities. Therefore, the recombinant yeast GS115-P1E12 harboring E12 gene would be an ideal candidate to replace GS115-P1−2 for industrial application. To further up-regulate E12 expression, the genes of molecular chaperone (pdi), Vitreoscilla hemoglobin (vhb) or transcriptional factor (upc), under the control of P or P, was introduced into GS115-P1E12, respectively, generating the following recombinant strains: GS115-P1E12-1 (co-expressing pdi controlled by P), GS115-P1E12-2 (co-expressing pdi controlled by P), GS115-P1E12-3 (co-expressing vhb controlled by P), GS115-P1E12-4 (co-expressing vhb controlled by P), GS115-P1E12-5 (co-expressing upc controlled by P), and GS115-P1E12-6 (co-expressing upc controlled by P). Subsequently, the resultant strains (Fig. 5A) were further cultivated for 7 days prior to the determination of the volumetric and biomass enzyme (β-xylosidase) activities described previously. The results indicated that, except for UPC, both PDI and VHB could apparently enhance the volumetric and biomass enzyme activities compared with those of control; typically, VHB had exerted the highest effect on day 7 (Figure 5, Figure 7). Additionally, the two promoters (P and P) showed similar effects on driving the expression of both pdi and vhb, in which P had exerted a slightly higher effect on improving the biomass enzyme activities (Fig. 5).

Figure 5

Effects of PDI, VHB and UPC on β-xylosidase (E12) expression. (A) Strains with different plasmids for over-expression of pdi, vhb and upc. (B, D and F) Volumetric enzyme activities. (C, E and G) Biomass enzyme activities. Data are mean±SD. n = 3.

Figure 7

Comparison of different strains on volumetric and biomass enzyme activities after cultivation for 7 days. Data are mean±SD. n = 3.

To examine the synergetic effect, both pdi and vhb genes were co-expressed in GS115-P1E12 host, under different combinations of P and P (Fig. 6A), and three recombinant strains were generated, including GS115-P1E12-7, GS115-P1E12-8 and GS115-P1E12-9. After 7 days of cultivation, all these three strains had surpassed the control (GS115-P1E12) in terms of both volumetric and biomass enzyme activities (Fig. 6B and C). Among them, GS115-P1E12-7 had exhibited the highest volumetric (28.10×106 U/L) and biomass (5.93×105 U/g) enzyme activities, which had amounted to 52.85% and 59.50% higher than those of the control strain GS115-P1E12 (Fig. 7). Western blotting data indicated that E12 expression was up-regulated by approximately 50% in GS115-P1E12-7 compared with that in the control (Fig. 6D and E), which was consistent with the improved volumetric and biomass enzyme activities. Additionally, as summarized in Fig. 7, the biomass enzyme activity of GS115-P1E12-7 was about 7.5 times as high as that of GS115-3.5K-P1−1, while 4 times as high as that of GS115-3.5K-P1−2.

Figure 6

Synergetic effect of PDI and VHB on β-xylosidase (E12) expression. (A) Strains with different plasmids for over-expression of pdi and vhb. (B) Volumetric enzyme activities. (C) Biomass enzyme activities. (D) Western blot analysis of de-glycosylated E12 from GS115-P1E12 and GS115-P1E12-7 using anti-His antibody. The Ponceau-S-stained blot (below) is the loading control. (E) Relative protein amount of each sample quantified with Quantity One Software. The values are indicated as mean±SD (n = 3) of independent assays.

Discussions

The glycosidase activity of LXYL-P1−2 was markedly higher than that of LXYL-P1−1, but the sequence difference between the two enzymes was only less than 3%. Moreover, of the 21 different amino acids, only 11 had exhibited significant differences in terms of their polarity and acidic-basic properties (Fig. 1). Site-directed mutations of these 11 amino acids from the low-active LXYL-P1−1 to the high-active LXYL-P1−2 were carried out, and a total of 17 variants were obtained, in which E16 and E17 had harbored 10 and 11 mutated amino acids, respectively (Fig. 2).

Our results showed that Ala69, Thr72, Ser91, Lys322, Ala327, Ala334 and Thr368 exhibited the upward bias in their activities, with Thr72, Ser91, and Thr368 being the top three amino acids. Besides, different combinations of mutations from 2 to 11 sites were made, in which the double mutant E12 (A72T/V91S) displayed the highest enzyme activities. Specifically, it had even surpassed LXYL-P1−2 by 2.8- and 3-fold with regard to β-xylosidase and β-glucosidase activities, respectively. Consequently, the catalytic efficiency of E12 against XDT was apparently increased compared with that of LXYL-P1−2 due to an increased substrate affinity (Table 2). However, other combinations of mutations would not lead to improved activities compared with those of LXYL-P1−2; alternatively, they would even result in decreased activities compared with those of LXYL-P1−1 (E14, E15) (Fig. 2). Furthermore, E16 and E17, which contained 10 and 11 aforementioned mutations, respectively, shared the highest similarity to LXYL-P1−2 in terms of polarity and acidic-basic properties (Fig. 1). Nonetheless, the enzyme specific activities of these two variants were still lower than those of LXYL-P1−2, even though they had surpassed those of LXYL-P1−1 (Fig. 2). Such finding suggests that other amino acids, including Phe105, Leu155, Asp223, Ile224, Ser278, Pro309 Val337, Glu338 Val337, Glu338, Ala354, and Ile357 among the 21 different amino acids in LXYL-P1−2 may also contribute to the positive effects on enzyme activity. It can be concluded from these results that, firstly, the combination of the 21 different amino acids in LXYL-P1−2 is not the best one, since the 19 amino acids of E12 (with the highest activity) in this study are of LXYL-P1−1 origin, while only 2 amino acids are the same as those in LXYL-P1−2. Secondly, the biocatalytic property is determined by the conformational fitness to the substrate, which is the outcome of interactions among various amino acids in the whole enzyme.

Molecular docking results showed that in the microenvironment of LXYL-P1−1 structure, the combinatorial mutation of A72T and V91S might positively promote the enzyme-XDT interaction. This was achieved through increasing the hydrogen bonds among Ser91 and other surrounding amino acids as well as XDT, apart from altering the loop conformation near XDT (Fig. 4). This might partially explain the reason why the mutant E12 (A72T/V91S) had exhibited apparently higher activity.

To further promote E12 production by the recombinant cells of GS115-P1E12, the chaperone protein PDI, bacterial Vitreoscilla hemoglobin VHB, and transcription factor UPC, were co-expressed with E12 in the yeast host GS115-P1E12. The results indicated that except for UPC, both PDI and VHB could up-regulate E12 expression, with VHB exerting a higher effect (Fig. 5). Moreover, both P and P could also efficiently drive enzyme production. Subsequently, these two helper factors were co-expressed with E12 and driven by different promoter combinations. Notably, the best results were observed when the two genes were controlled by P (Fig. 6). The volumetric and biomass activities of all variants obtained in this study, together with those of the original GS115-3.5K-P1−1 and GS115-3.5K-P1−2, are summarized in Fig. 7. As shown in Fig. 7, the volumetric and biomass enzyme activities of GS115-P1E12-7 reached 28.1×106 U/L and 5.93×104 U/g, respectively, which were 4 times higher than those of GS115-3.5K-P1−2 under the same culture conditions. These findings revealed that over-expressing VHB could enhance oxygen supply and efficient oxygen utilization, so as to up-regulate the expression of heterologous protein. Moreover, over-expressing PDI could also promote the correct protein folding and up-regulate the heterologous protein expression. The recombinant yeast GS115-3.5K-P1−2 had been used for the high-cell-density fermentation and catalysis of hydrolyzing xylosyl group from XDT to form DT for the Taxol semi-synthesis34., 36., 37.. Therefore, replacing GS115-3.5K-P1−2 with GS115-P1E12-7 should greatly reduce the catalytic cost, which might benefit the industrial application.

Conclusion and perspectives

Taken together, site-directed mutations of the 11 different amino acids were carried out from the low-active LXYL-P1−1 to the high-active LXYL-P1−2. The results suggested that, seven amino acids, including Ser69, Thr72, Ser91, Glu322, Glu327, Glu334 and Thr368, were the important amino acids with upward bias in their activities. Particularly, A72T and V91S mutations in E3 and E4 had rendered them the superior enzyme activities among all the single-site mutants of LXYL-P1−1. Moreover, the best variant E12 with double mutation (A72T/V91S) was also obtained, which had notably improved the catalytic efficiency compared with that of the original LXYL-P1−1 and even surpassed that of the high-active LXYL-P1−2. The possible mechanism for the improvement was proposed by homology modeling and molecular docking between E12 and 7-β-xylosyl-10-deacetyltaxol. Besides, the recombinant yeast GS115-P1E12-7 was also constructed based on variant E12, which harbored the molecular chaperone gene pdi (P. pastoris origin) and the bacterial hemoglobin gene vhb (Vitreoscilla origin) driven by the inducible promoter P. Typically, the volumetric and biomass enzyme activities of the recombinant yeast GS115-P1E12-7 were 4 times higher than those of the GS115-3.5K-P1−2 control. Therefore, GS115-P1E12-7 can be used to replace GS115-3.5K-P1−2 for industrial purpose, which will greatly increase the catalytic efficiency and significantly reduce the cost. In the future, saturated mutagenesis on Ser69, Thr72, Ser91 and Thr368, followed by different combinations may be conducted to find more active mutants. Since enzymes with different variants are widely distributed in the nature38., 39., 40., our study may serve as a paradigm for acquiring highly active enzyme variants, which can be used for the drug structural modification or optimization.