The subcloning and construction of recombinant plasmids was carried out in E. coli DH10B (Invitrogen). The P. pastoris GS115 strain (Invitrogen) was used as a host for the expression of heterogenous proteins. The vectors pPIC3.5K and pPICZαA were purchased from Invitrogen. The vector pTSC (Yan et al., 2008) was stocked in this lab. pUC57-MCS7 (the original plasmid used for construction of recombinant expression vectors) was from GeneCreate Biological Engineering Co., Ltd, Wuhan, China. E. coli DH10B was cultured in Luria-Bertani medium [1% (w/v) tryptone, 0.5% (w/v) yeast extract, and 1% (w/v) NaCl, pH 7.0] at 37°C. The P. pastoris yeast strain was cultured in YPD medium [1% (w/v) yeast extract, 2% (w/v) peptone, 2% (w/v) dextrose]. YPDZ plates containing Zeocin (100 μg/mL) were used for the selection of positive P. pastoris transformants. Buffered minimal glycerol-complex medium (BMGY) was prepared with 2% (w/v) peptone, 1% (w/v) yeast extract, 1% (w/v) glycerol, 1.34% (w/v) yeast nitrogen base with ammonium sulfate but without amino acids, and 4 × 10⁻⁵% (w/v) biotin in 100 mM potassium phosphate buffer. The phosphate buffer was adjusted to pH 6.0. Buffered minimal methanol-complex medium (BMMY) was the same as BMGY, except that 0.5% (w/v) methanol was replaced with glycerol.

All primers used in the construction of the recombinant plasmid pMC are shown in Table 1; cleavage sites of restriction endonucleases are underlined. The schematic map of the construction of expression vector pMC is shown in Figure S1 (online resource) and can be briefly described as follows. A fragment with 7 multiple cloning sites was synthesized in vitro and cloned into pUC57-Kan-MCS-free, resulting in a pUC57-MCS7 plasmid. The HIS4 gene was amplified by PCR from pPIC3.5K using the primer pairs HIS4-A-F/R and HIS4-B-F/R and was integrated into the pUC57-MCS7 plasmid. The Xba I on HIS4 was synonymously mutated by overlap extension PCR. The fragment AOX1TT-P_TEF1-P_EM7-Zeocin^R-CYC1TT-lox66 was amplified from pPICZαA using the primer pairs ATEZC-F/R and was integrated into the pUC57-MCS7 plasmid. The cre gene was amplified from pTSC using the primer pairs Cre-F/R and was integrated into the pUC57-MCS7 plasmid. The fragment lox71-P_AOX1 was amplified from pPICZαA using the primer pairs AOX-F/R and was integrated into the pUC57-MCS7 plasmid. Finally, the recombinant plasmid was named pMC.

Overnight cultures of E. coli DH10B (pMC) were harvested for extraction of total RNA. Total RNA was extracted using a Bacteria total RNA Isolation Kit (Sangon Biotech, Shanghai, China) and treated with RNase-free DNase (Takara, Dalian, China) to eliminate any residual DNA. The total RNA was added to the PCR system (used as temple) to check if the total DNA was completely hydrolyzed. This PCR system was used as a negative control. The cDNA was obtained using a PrimeScript RT reagent kit (TaKaRa, Dalian, China), and 2 μl of 1:5-diluted cDNA samples was used as the template for reverse transcription-PCR. The existence of the cre gene in cDNA was identified by RT-PCR. The 16S rRNA gene was used as a positive control. Primers used in RT-PCR are shown in Table 2.

To inhibit the leakage expression of Cre recombinase in E. coli, an upstream ORF or lacO (operator gene) was integrated using the primer pairs AOX-F/R-uORF or AOX-F/R-lacO, respectively. The resulting recombinant plasmids were named pMCU and pMCO. The sites of Xho I, Not I, and Nco I on pMCO were deleted by overlap extension PCR, making them available in the heterogenous expression cassette or linearization site of HIS4. The fragments AOX1TT and P_AOX1-αFactor were subsequently integrated into pMCO. The resulting recombinant plasmids were named pMCO-AOXα.

The sequence the of appA gene according to the literature (Dassa et al., 1990) was codon optimized and synthesized in vitro, then cloned into the pMCO-AOXα vector via EcoR I and Pst I. The resulting recombinant plasmid, named pMCO-AOXα-A1, was digested by Spe I and Xba I to generate a 2.7 kb expression cassette. Meanwhile, pMCO-AOXα-A1 was also linearized by Xba I. The two fragments were connected by T4 ligase to construct the plasmid pMCO-AOXα-A2; the multicopy recombinant vectors pMCO-AOXα-A4 and pMCO-AOXα-A8 were constructed in the same way.

The expression vectors pMCO-AOXα-An (n = 1, 2, 4, or 8) were linearized with Sal I and transformed into P. pastoris GS115 by electroporation. The charging voltage, resistance, and capacitance were 1,500 V, 250 Ω, and 25 μF, respectively. Positive transformants on YPDZ plates were shifted to new YPDZ plates with pre-marked serial numbers and incubated overnight at 30°C. The fresh clones were placed into 3 mL of BMGY liquid medium in test tubes and cultured at 30°C and 200 rpm for ~36 h until a stable saturation was attained; then, the cells were harvested by centrifugation and resuspended in 1 mL of BMMY. After incubation at 30°C and 200 rpm for 24 h, the enzyme assay was immediately performed.

Phytase activity was determined by the ferrous sulfate-molybdenum blue method (Huang et al., 2006). Briefly, 100 μl of 200-fold diluted enzyme solution was incubated with 900 μl of substrate solution (4 mM sodium phytate in 0.25 M sodium acetate buffer, pH 4.5) at 37°C for 10 min. The reaction was stopped by adding 1 mL of 10% (w/v) trichloroacetic acid (TCA). The amount of inorganic phosphate released was analyzed by adding 2 mL of a coloring reagent C [1% (w/v) ammonium molybdate, 3.2% (v/v) sulfuric-acid solution, and 7.2% (w/v) ferrous sulfate solution] and measuring the absorption at 700 nm. One unit (U) of phytase activity was defined as the amount of enzyme that releases 1 μmol phosphate in 1 min at 37°C.

In each group of recombinants with designed copy numbers (copy number = 1, 2, 4, or 8), the clone that exhibited the highest phytase activity was selected, and its genomic DNA was extracted using a Yeast Genomic DNA Isolation Kit (Sangon Biotech, Shanghai, China). GAPDH (PAS_chr2-1_0437), which is present as a single copy in the P. pastoris genome, was chosen as the reference gene. All primers used for real-time quantitative PCR (qPCR) are listed in Table 2. For each qPCR reaction mixture, 2 μl of 1:100 diluted genomic DNA samples was used as the template with 10 μl of SYBR Premix Ex Taq II (Takara, Dalian, China) in a total volume of 20 μl. The amplification reaction was performed according to the manufacturer's instructions. Each reaction was performed in triplicate. Relative quantification of the copy number of the target gene was performed according to the 2^−ΔΔCt method (Livak and Schmittgen, 2001; Zarei et al., 2014), where ΔΔCt = ΔCt of target − ΔCt of calibrator, and ΔCt = Ct of target of calibrator − Ct of reference (GAPDH). A strain with a known copy number (one copy) of the target gene served as the calibrator strain.

The 24-h BMMY cultures were plate streaked onto YPD plates and incubated at 30°C until colonies were visible. Single clones were shifted to new YPD and YPDZ plates and incubated at 30°C overnight. The percentage of cells that lost the Zeocin^R cassette after methanol induction was calculated by comparing colony numbers on YPD and YPDZ plates. Then, the phytase activity assay was performed as previously described to reconfirm the yield of phytase for colonies that were sensitive to Zeocin.

In the group of recombinants harboring 8 copies of the target gene, the clone that exhibited the highest phytase activity was induced in YPM liquid medium in test tube. The obliteration of resistance to Zeocin was performed as previously described. After reexamining the phytase activity of the clones that lost resistance to Zeocin, the linearized expression vectors pMCO-AOXα-A4 and pMCO-AOXα-A8 were transformed into markerless P. pastoris cells by electroporation, resulting in transformants with 12 or 16 copies of the target gene. The enzyme assay of positive transformants was determined in test tube-scale, and the copy number of the target gene was determined by qPCR as previously described.

In each group of recombinants harboring 1, 2, 4, 8, 12, or 16 copies of the target gene, the clone that exhibited the highest phytase activity was cultured and induced on the shake-flask scale for reexamination. Transformants were transferred to 30 mL of BMGY liquid medium in a 250 mL shake flask and were cultured ~48 h at 30°C and 200 rpm until a stable saturation was attained. Then, the cells were harvested by centrifugation, resuspended in 30 mL of BMMY and cultured at 30°C at 200 rpm for 96 h. A total of 150 μl of 100% methanol was added every 24 h to maintain induction, and 150 μl of culture was extracted at the same time for the enzyme assay. After 96 h of induction, culture supernatants were analyzed by SDS-PAGE.

A selection marker rescue vector pMC containing a lox71-Cre-Zeo^R-lox66 cassette was designated for multicopy P. pastoris expression strain generation. The schematic map of the construction of pMC is shown in Figure S1. Briefly, pUC57-MCS7 was used as the original plasmid, and four DNA fragments were subsequently integrated into pUC57-MCS7. To verify the proper assembly of pMC, single-colony PCR was performed using the primer pair lox-F/R, indicated by arrows in Figure 1A. The fragment of Cre-AOX1TT-P_TEF1-P_EM7-Zeo^R-CYC1TT-lox66 was ~2,800 bp, while the fragment plus lox71-P_AOX1 was calculated as ~3,800 bp. Unexpectedly, a single DNA band of ~250 bp was clearly observed by gel electrophoresis analysis (Figure 1B), which indicated that the fragment of Cre-AOX1TT-P_TEF1-P_EM7-Zeo^R-CYC1TT-lox66 confirmed the estimated molecular weight (2.8 kb). The smaller electrophoretic band suggests that the plasmid pMC recombined spontaneously.

To further identify the leakage expression of the cre gene, the total RNA of E. coli DH10B (pMC) was extracted, and reverse transcription-PCR (RT-PCR) was performed. The negative control (Figure 1C, line 3) was not observed, demonstrating that the genome DNA and plasmid DNA were completely digested. The positive control (Figure 1C, line 1) and Cre recombinase (Figure 1C, line 2) were detected, indicating that the AOX1 promoter from P. pastoris could be partially recognized by E. coli and the downstream cre gene was transcribed. We deduced that the leakage expression of Cre recombinase caused the excision of lox71-P_AOX1-Cre-AOX1TT-P_TEF1-P_EM7-Zeo^R-CYC1TT-lox66, generating a lox72 site on the plasmid.

To overcome the leakage expression of Cre in E. coli, two cis elements were integrated between P_AOX1 and the cre gene on the plasmid pMC (shown in Figure 2A). One is an upstream ORF (uORF), which contains eight AGA codons. The other is the lac operator gene (lacO) of the lactose operon. The integrations were performed using primer pairs AOX-F/AOX-R-uORF and AOX-F/AOX-R-lacO, respectively. The resulting recombinant plasmids were named pMCU and pMCO.

To verify the proper assembly of pMCU and pMCO, colony PCR was performed using the primer pair lox-F/R. A 3.8 kb DNA band was observed for both plasmids (Figure 2B, lane 2–3), which matches the predicted size of the whole insertion fragment, indicating the proper integration of P_AOX1-uORF and P_AOX1-lacO. However, when these two recombinant plasmids were extracted, gel electrophoresis analysis showed that the size of the two plasmids was clearly different. The larger band (Figure 2C, lane 1, consistent with the theoretical molecular weight) showed the outstanding stability of pMCO, while the smaller band of pMCU (Figure 2C, lane 2) indicated that recombination between lox71 and lox66 mediated by Cre recombinase still occurred. The stability of pMCO was also verified by RT-PCR using the primer pairs 16s-RT-F/R and Cre-RT-F/R. The gel electrophoresis analysis showed that no cDNA of the cre gene was detected (Figure 2D, lane 2). These results showed that the lacO operator could effectively block the leakage expression of Cre recombinase promoted by the AOX1 promoter in E. coli.

pMCO was applied to construct the full functional expression vector for P. pastoris. First, to make Xho I, Not I, and Nco I available in the heterogenous expression cassette or linearization site of HIS4, these sites on pMCO were deleted by overlap extension PCR. Then, the fragments AOX1TT and P_AOX1-αFactor were subsequently integrated into pMCO. The resulting recombinant plasmid was named pMCO-AOXα (Figure 3).

The expression vector pMCO-AOXα has the potential for commercial application: it can be linearized by Nco I, Sal I, and Nhe I on HIS4, and heterologous genes can be integrated into the expression cassette by utilizing the Xho I, EcoR I, Pst I, and Not I sites. Since Spe I and Xba I were introduced into pMCO-AOXα as isocaudamers, the expression cassette can be inserted in the unique Xba I site repetitively, generating a multicopy recombinant plasmid. Briefly speaking, suitable restriction endonuclease sites facilitate the insertion of target genes, the multimerization of expression cassettes and the integration of linearized recombinant plasmids.

Phytase is generally used as an animal feed supplement to enhance the nutritive value of plant material by liberation of inorganic phosphate from phytic acid. Phytase AppA was from E. coli and can catalyze the hydrolysis of phytic acid (myo-inositol hexakisphosphate). The optimized appA gene was synthesized and successfully cloned into pMCO-AOXα; the resulting vector carrying 1 copy of the appA gene was designated as pMCO-AOXα-A1. The multimerization was performed as described in the “Materials and methods,” producing the vectors pMCO-AOXα-A2, pMCO-AOXα-A4 and pMCO-AOXα-A8, which carried 2, 4, or 8 copies of the appA gene, respectively (Figure 4A). To confirm the correct orientation of the integration of expression cassette, which was inserted in the unique Xba I site, the recombinant vectors pMCO-AOXα-A1, pMCO-AOXα-A2, pMCO-AOXα-A4, and pMCO-AOXα-A8 were digested with Spe I and Xba I (Figure 4B). Gel electrophoresis analysis showed that the molecular weight of the inserted expression cassettes increased gradually, while the molecular weight of the vector frame remained ~10 kb (arrow, Figure 4B). The result verified that pMCO-AOXα-A1, pMCO-AOXα-A2, pMCO-AOXα-A4, and pMCO-AOXα-A8 were constructed correctly.

The plasmids containing 1, 2, 4, or 8 copies of appA were linearized and electroporated into P. pastoris. After selection of recombinants on YPDZ plates, 10 positive clones for GS115-pMCO-AOXα-An (n = 1, 2, 4, or 8) were picked (a total of 40 clones), initially cultured in BMGY, then transferred into BMMY to induce the expression of the target protein with methanol. The supernatants of 24-h BMMY cultures were collected, and the highest-yielding clones were screened by quantification of phytase activity.

In test tube culture, the transformants GS115-pMCO-AOXα-An-Zeocin^R (n = 1, 2, 4, or 8) showed an increasing trend in the recombinant protein expression level, and the GS115-pMCO-AOXα-A8-Zeocin^R obtained the highest phytase yield (Figure 5A). For each group of transformants GS115-pMCO-AOXα-An-Zeocin^R (n = 1, 2, 4, or 8), the production of the strain exhibiting the highest yield of phytase from the 10 clones is listed in Table 3.

Since methanol could also induce the expression of Cre recombinase, the loss of Zeocin resistance gene was checked. The BMMY cultures of the clones denoted as GS115-pMCO-AOXα-A1-Zeocin^R #1, GS115-pMCO-AOXα-A2-Zeocin^R #7, GS115-pMCO-AOXα-A4-Zeocin^R #5, and GS115-pMCO-AOXα-A8-Zeocin^R #3 were streaked onto YPD plates, single colonies were transferred to new YPD and YPDZ plates, and the percentage of cells that lost the Zeocin^R cassette was calculated (Figure 5B, 10 colonies for each Zeocin^R transformant except 20 colonies for GS115-pMCO-AOXα-A8-Zeocin^R #7). All clones were sensitive to Zeocin, indicating the excision of the resistance gene by Cre recombinase, which was verified by single-colony PCR (Figure 5C, and the schematic map of the deletion of Zeocin resistance gene was shown in Figure S2). Then, the phytase yield assay was performed for these 50 colonies (Figure 5D, in test tube-scale as well); the strain exhibiting the highest yield of phytase from each group of 10 or 20 clones is listed in Table 4. The yield of phytase and the copy number analysis are also shown in Table 4.

To obtain transformants with 12 or 16 copies of the appA gene, the vectors pMCO-AOXα-A4 or pMCO-AOXα-A8 were successfully transformed into GS115-pMCO-AOXα-A8-Zeocin^S #17, and the resulting strains were denoted as GS115-pMCO-AOXα-A8A4-Zeocin^R or GS115-pMCO-AOXα-A8A8-Zeocin^R, respectively. The results showed that our marker rescue system worked well in P. pastoris. For each strain, 10 positive clones were cultured and induced; their phytase expression levels are shown in Figure 6A. In addition, the highest-yielding clones for GS115-pMCO-AOXα-A8A4-Zeocin^R and GS115-pMCO-AOXα-A8A8-Zeocin^R were screened (listed in Table 3). The highest phytase yield of the transformants harboring 12 copies of appA gene was 418.23 U/mL, which was 4.45 and 1.17 times higher than the yield of 1-copy transformants (93.97 U/mL) and 8-copy transformants (355.82 U/mL), respectively. However, the phytase yield of 16-copy transformants decreased to 369.88 U/mL, indicating that the abundant over-expression of proteins imposed a metabolic burden on the secretory pathway.

Zeocin resistant genes of the P. pasrois transformants #7(12 copies) and #9(16 copies) were further deleted, the phytase yields of the 10 Zeocin^S colonies were assayed (as shown in Figure 6B). The strains exhibiting the highest phytase yield from each group of 10 clones of GS115-pMCO-AOXα-A8A4-Zeocin^S or GS115-pMCO-AOXα-A8A8-Zeocin^S are listed in Table 4. The yield of phytase and the copy number analysis are also shown in Table 4.

After the generation and identification of a series of variant P. pastoris transformants harboring 1, 2, 4, 8, 12, or 16 copies of the appA gene, the shake-flask cultures were created to reconfirm the phytase yield for each transformant. Data shown in Figure 7 demonstrated that the recombinant phytase yield increased as the copy number of appA gene increased, and GS115-pMCO-AOXα-A8A4-Zeocin^S #5 reached the highest expression level of 179.34 mg/L, based on specific activity of 3,100 U/mg protein (Huang et al., 2006). The phytase yield of GS115-pMCO-AOXα-A8A8-Zeocin^S #7 decreased to 148.38 mg/L, probably because of an ERAD response to restore ER homeostasis.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewer XZ and handling Editor declared their shared affiliation.