Zeocin

Studies to analyse the relationship between IFNα2b gene dosage and its expression, using a Pichia pastoris-based expression system

Abstract

Human interferon α2b (hIFNα2b) is the most important member of the interferon family. Escherichia coli, yeasts, mammalian cell cultures and baculovirus-infected insect cells have been used for expressing recombinant human interferon. Recently a Pichia pastoris-based expression system has emerged as an attractive system for producing functional human recombinant IFNα2b. In this regard, gene dosage is considered an important factor in obtaining the optimum expression of recombinant protein, which may vary from one protein to another. In the present study we have shown the effect of IFNα2b gene dosage on extracellular expression of IFNα2b recombinant protein from P. pastoris. Constructs containing from one to five repeats of IFNα2b-expressing cassettes were created via an in vitro multimerization approach.

P. pastoris host strain X-33 was transformed using these expression cassettes. Groups of P. pastoris clones transformed with different copies of the IFNα2b expression cassette were screened for intrachromosomal integration. The IFNα2b expression level of stable transformants was checked. The copy number of integrated IFNα2b was determined by performing qPCR of genomic DNA of recombinant P. patoris clones. It was observed that an increase in copy number generally had a positive effect on the expression level of IFNα2b protein. Regarding the performance of multicopy strains, those obtained from transformation of multicopy vectors showed relatively high expression, compared to those generated using transformation vector having only one copy of IFNα2b. It was also observed that an increase in drug resistance of a clone did not guarantee its high expression, as integration of a marker gene did not always correlate with integration of the gene of interest.

Keywords: Pichia pastoris; in vitro multimerization; expression cassettes; transformation; Zeocin; gene dosage effect

Introduction

Human IFNα2b is the most widely used member of the interferon family and is of immense clinical importance. It affects several biological functions, including wide-ranging antiviral, antibacterial activities (Rönnblom et al., 1983), inhibition of tumour cell proliferation (Fidler et al., 1987) and improvement of immune functions. The remedial functions of IFNα2b have led to its approval as a potential agent for therapy of hairy cell leukaemia (Talpaz et al., 1986), chronic myelogenous leukaemia (Allan et al., 1995), gastrointestinal stromal tumours, chronic hepatitis B and C (Ariyasu et al., 2005), malignant melanoma, condylomata accuminata, follicular lymphoma (Koster et al., 2005) and AIDS-related Kaposi’s sarcoma (Shepherd et al., 1998).

The cost of HCV therapy is expected to reach US $5 billion by 2013 as mentioned by IMS health 2005 data, which 64% comprises of interferon therapy (News reported on http://www.pharmaessentia. com/english/news_200901.html). The rising inci- dence of certain cancers and viral hepatitis has increased the demand for human recombinant IFNα2b, which indicates the need to optimize production of hrIFNα2b. Among all the expression systems [Escherichia coli (Barbero et al., 1986; Valente et al., 2004; Srivastava et al., 2005), yeast (Shi et al., 2007; Ayed et al., 2008; Gasmi et al., 2011a, 2011b), mammalian cells (Loignon et al., 2008) and several plant species (Ohya et al., 2001; Luchakivskaya et al., 2011)] used for rhIFNα2b production, P. pastoris appears to be most economical, with the capacity to yield up to hundreds of milligrams of purified IFNα2b/litre of culture (Ayed et al., 2008). The advantages of higher eukaryotic post-translational modifications (Cereghino and Cregg, 2000), a low level of native secreted proteins, leading to simple purification (Cregg et al., 1993), absence of hyperglycosylation and terminal α1,3-bound mannose residues (potent antigens) in contrast to S. cerevisiae (Schmidt, 2004; Schuster et al., 2000), accumulation of considerable biomass on low-cost simple defined medium, almost > 130 g/l dry cell weight on meth- anol (Wegner, 1990) and occurrence of a powerful and efficiently regulated promoter derived from the alcohol oxidase 1 gene (PAOX1) of the methanol utilization pathway (Cregg et al., 2000) made this system highly acceptable in contrast to E. coli and mammalian cell culture.

Although the successful production of IFNα2b in P. pastoris has been reported previously, it is interesting to exploit factors influencing gene expression to improve yield, especially in view of its large pharmaceutical interest (Shi et al., 2007; Ayed et al., 2008; Ghosalkar et al., 2008). Among various other factors important for the optimization of protein expression, gene copy number (gene dosage) has emerged as a ‘rate-limiting’ step (Sreekrishna et al., 1989). For IFNα2b overexpression, the effect of copy number has not been systemati- cally studied. P. pastoris has an impressive record of high yield by increasing gene dosage in the expression of tumour necrosis factor (TNFα) (Sreekrishna et al., 1989), tetanus toxin fragment C protein, mEGF (Clare et al., 1991), aprotinin (Vedvick et al., 1991), HIV-1 envelope protein (Scorer et al., 1993), Bordetella pertussis peractin (P69) (Romanos et al., 1991), insulin-like growth factor-1 (Brierley et al., 1994), hepatitis B surface antigen (HbsAg) (Vassileva et al., 2001), β-galacto- sidase (β-gal) (Sunga and Cregg, 2004) and miniproinsulin (MPI) (Mansur et al., 2005). However, by increasing gene dosage, reduced expression has been noted in human trypsinogen (Hohenblum et al., 2004) and Na-ASP1 (Inan et al., 2006). This may be due to deteriorated cell physio- logical conditions of multicopy transformants, emerging from higher oxidative stress related to protein-folding stress in endoplasmic reticulum (Hohenblum et al., 2004; Cos et al., 2005; Zhu et al., 2011). All these studies indicate that the in- fluence of copy number on recombinant protein expression levels could not be predicted, as highlighted by Thill et al. (1990).

Hence, we investigated the effect of gene dosage on the expression level of IFNα2b in P. pastoris. Extracellular production of IFNα2b in P. pastoris and its gene dosage were found to have close correspondence with each other, as recombinant P. pastoris clones having a high expression level of IFNα2b were also found to have a high copy number of integrated IFNα2b gene. It was also observed that increase in the drug resistance of a clone does not guarantee its high expression, as resistance to a higher level of selection marker does not always correlate with integration of the gene of interest.

Materials and methods

Strains and plasmids

The cloning and construction of recombinant expression vectors was carried out in E. coli TOP10F′. P. pastoris yeast strain X-33 (Invitrogen) was used as a host for protein expression studies. Vector pPICZαC (Invitrogen, San Diego, CA, USA) was used for multicopy construction. pPICZαIFNα2b, a pPICZα-derived plasmid with the IFNα2b gene (610 bp) inserted at a unique BamHI site downstream of the AOX1 promoter, was taken from the BPL CEMB laboratory. This IFNα2b gene was chemically synthesized and codon-optimized for expression in Pichia pastoris by utilizing the services of Gene script (NJ, USA).

Enzymes and reagents

DNA-modifying enzymes, restriction endonucle- ases, 1 kb ladder, dNTPs and Gel Extraction Kit were from Fermentas. Materials for sodium dodecylsulphate–polyacrylamide gel electrophoresis (SDS–PAGE) were from Bio-Rad. Components for microbial media were from Bacto™. TaqDNA polymerase and HindIII-digested λDNA standards were obtained from the enzyme production labora- tory at the Center of Excellence in Molecular Biology. All other reagents used were obtained from Sigma Biochemicals (USA) or Merck Bio- chemicals (Germany). All chemicals were of ana- lytical grade.

Construction of expression plasmid with multicopy expression cassettes

IFNα2b expression cassette, encompassing the AOX1 5′-promoter, the MF-IFNα2b fusion and the AOX1 3′-terminator, was isolated from pPICZαIFNα2b as a 2099 bp BglII–BamHI fragment (Figure 1) by digesting it with BamHI and BglII. BglII digestion in the IFNα2b gene (as it carries the BglII restriction site) was avoided, first by the complete digestion of IFNα2bC-1 with BamHI, followed by partial digestion with BglII. The digested mixture was run on with 1% agarose gel along with ladder. The 2.1 kb band was eluted using a Fermentas Gel Elution Kit. This purified insert was then inserted into the unique BamHI site of the original plasmid to create a tandem repeat of two copies by ligation. Subsequent repetition of this process was done to generate vectors carrying up to five copies of the IFNα2b gene (Figure 2). These recombinant plasmids were given the names pPICZαC-2, pPICZαC-3, pPICZαC-4 and pPICZαC-5. For a negative control, the strain X-33/pPICZα, having pPICZα (vector without IFNα2b gene), was generated.

Transformation of P. pastoris

Easycomp™ Transformation and Electroporation were used for transformation of P. pastoris. Verified multicopy constructs and parent vectors were isolated from E. coli TOP10F′, using the alkaline lysis method. SacI-linearized pPICZα and pPICZαIFNα2bC-1 were transformed by electroporation and Easy Comp™ Transformation Kit. Circular pPICZαIFNα2bC-2, pPICZαIFNα2bC-3, pPICZαIFNα2bC-4 and pPICZαIFNα2bC-5 were transformed by electroporation to obtain high-copy number clones. Electroporation was done according to Invitrogen protocols, using modified steps for more colonies. X-33 transformants were selected on YPDS plates with Zeocin at 100 μg/ml. Independent colonies were selected from each transformed plasmid.

Screening for multicopy transformants

All transformed colonies were picked and restreaked on YPD plates with Zeocin at 100 μg/ml. Colonies showing growth were grown in a selection-free environment (in YPD broth without Zeocin) for 96 h in a shaking incubator at 28°C. Restreaking was done on YPD plates with 100 μg/ml Zeocin. Selection of the clones with intragenomic insertion of the vectors was ensured by this step. These transformants were screened for insert at the AOX1 locus by performing genomic PCR assay on yeast DNA. For this purpose, AOX1 sequencing primers, 5′-GACTGGTTCCAATTGACAAGC-3′ and 5′-GCAAATGGCATTCTGACATCC-3′, were used as forward and reverse primer, respectively (supplied by Invitrogen).

Determining methanol phenotype and extent of Zeocin resistance

As the methanol phenotype has a substantial im- pact on growth capacity, the methanol phenotype of the selected clones was determined by replica plating on MD and MM media plates without Zeocin. For this the clones were first grown on MD for 2 days at 30 °C by making grids. Then clones were restreaked, first on MM, followed by MD in replica. KM71 Muts was used as the negative control and GS115 Mut+ as the positive control. The results were noted by comparing the growth patterns of colonies with controls after 2 days. The methanol phenotype of transformants was also verified by genomic PCR assay with AOX1 primers (see section 2.5).

For the extent of Zeocin resistance, selected clones were grown on YPD plates with 100, 250, 500, 750 or 1000 μg/ml Zeocin. For this each clone was grown in 3 ml YPD without Zeocin overnight and then streaked on these plates using sterile toothpicks. The results were noted after 2 days of incubation at 30 °C.

Preparation of genomic DNA and design of primers for qPCR analysis

Isolation of the chromosomal DNA used for simple PCR and for real time qPCR was carried out using the method described by Tomita et al. (2002). For consistent qPCR, assay template DNA extracted from P. pastoris transformants was normalized to 150 ng/μl with deionized and distilled water. The concentration of isolated DNA was estimated using a NanoDrop 1000 spectrophotometer. After isolating the genomic DNA we ran it on 0.8% agarose gel containing 0.5 μg/μl ethidium bromide. The purity of DNA was verified by OD260:OD280 absorption ratio.
Primer 3 software was used for primer design. The sequences of the primers to detect the amplicon of the GAP and IFNα2b genes within the genomic DNA of the transformants are shown in Table 1.

Quantification of IFNα2b gene copy number

For quantification of IFNα2b gene copy number integrated in the P. pastoris genome, real-time qPCR analysis was performed, using genomic DNA as the template. Recommendations from the SYBR Premix Manual were followed for the establishment of reaction conditions. Real-time PCR reactions were performed in 15 μl mixtures. The mixture for one reaction contained 7.5 μl Maxima Syber Green/ROX qPCR Master mix, 1 pM each forward and reverse primer, 150 ng sample genomic DNA and nuclease-free water. The IFNα2b target gene was detected with Ifnα2bF and Ifnα2bR primers. At the same time, co- amplification of the GAP gene was carried out with gapF and gapR primers in other reactions as the internal control. The reaction consisted of a predenaturation step at 50 °C for 2 min and then at 95 °C for 2 min, followed by 40 cycles of denaturation at 95 °C for 35 s and annealing at 52 °C for 35 s and elongation at 72 °C for 45 s. For the negative control, double-distilled water was used as the template. All real-time qPCR reactions were run on Applied Biosystems 7500 Real-Time PCR System, and crossing point (Cp) values of each sample were measured three times. Analysis of the melting curve after 40 cycles and agarose gel electrophoresis was done to verify the specificity of the amplicon.

Construction of the standard curve for IFNα2b gene copy number determination

To establish the standard curves for IFNα2b and GAP, 10-fold serial dilution series of the pPICZα–IFNα2b and pPICZα–GAP plasmids in the range 10–1–10–7 copies/μl were utilized. The Cp values in every dilution were measured three times using real-time qPCR with the IFNα2b and GAP primer sets, using the conditions described earlier. The average Cp value of each dilution was plotted against the logarithm of the corre- sponding template copy number. Each standard curve was generated by linear regression of the plotted points. Optimization of the qPCR assay was evaluated by equation of linear regression and by coefficient of determination (R2 ). The R2 value of a standard curve describes how well the experimental data fit the regression line. The standard equation (1) was applied to calculate the amplification efficiency, E, of qPCR (Rebrikov and Trofimov, 2006): For each DNA sample, the copy concentration of the target gene (GAP or IFNα2b) was determined by relating the Cp value to a standard curve. Finally, the ratio of total IFNα2b copy number to total GAP copy number of each sample was designated as the IFNα2b copy number integrated in the genome.

Shake-flask cultures

Transformants containing the integrated vectors were tested in a shake flask to evaluate IFNα2b expression levels with respect to cassette copy number. The inoculum was prepared by growing a single colony of each of the selected transformants in 5 ml BMGY, taken in 50 ml Falcon tubes, overnight at 28–30°C in an incubator shaker set at 250 rpm. The overnight culture was transferred to the 25 ml BMGY taken in a 250 ml conical flask without baffles and grown at 28–30°C in an incubator shaker at 250 rpm for about 20 h. The culture was harvested by centrifugation on 6000 rpm at room temperature. The supernatant was discarded without disturbing the pellet. To induce expression, the pellet was resuspended in 30 ml BMMY in a 250 ml conical flask without baffles by covering it with sterile aluminium foil. The incubator was set at 25–28°C and 250 rpm. During the induction phase, after every 24 h 100% methanol was added to a final concentration of 0.5% v/v for Mut+ and 0.3% v/v for Muts. To determine the optimal time post-induction to harvest, samples were analysed by SDS–PAGE for expression at 24, 48, 72, 96 and 120 h by taking a 1 ml sample in a 1.5 ml microcentrifuge tube. To analyse the extracellular expression of IFNα2b, the sample in the microcentrifuge tube was spun for 10 min at 14 000 rpm and the supernatant was stored for analysis by silver- and Coomassie- stained SDS–PAGE.

SDS–PAGE, densitometric analysis and determination of protein concentration

SDS–PAGE was performed in a Hoffer’s gel apparatus according to the method of Laemmli (1970). A 15% resolving and 5% stacking gel with 1.5 mm thickness was used. The samples were dissolved in 10× sample buffer and incubated at 100 °C for 5 min. About 30 μl of each sample was loaded in the gel at a constant voltage of 100 V. After running, the gel was stained with Coomassie. Densitometric analysis of the SDS–PAGE gel was performed using ImageJ software (developed by the National Institutes of Health, USA). Greyscale image was used to calculate the protein concentration in the relative lane of the gel image. Total protein concentrations in the culture supernatant at 72,96 and 120 h after induction were determined by the method of Bradford (1976).

Immunoblot assay

The immunoblot assay was performed using culture supernatant at 120 h after induction. The nitrocellulose membrane was wetted in 1× PBST and air-dried on blotting paper for 5 min before use. A 3 μl sample was spotted onto nitrocellulose membrane and air-dried for 1 h. The nitrocellulose membrane was immersed in 5% skimmed milk for 1 h on a shaking platform. The membrane was incubated in anti-human IFNα polyclonal antibody diluted at 1:1000 in 1× PBST. After incubation for 60 min at 37 °C with gentle shaking, the nitrocellu- lose membrane was washed three times for 10 min each with 1× PBST, after which the membrane was incubated for 1 h in secondary antibody solution–anti-sheep IgG horseradish peroxidase conjugated monoclonal antibody (Sigma, St Louis, MO, USA) diluted at 1/10 000 in 1× PBST. After washing, IFNα2b was visualized by adding Nitro Blue Tetrazolium (NBT) alkaline phosphatase substrate 5-bromo-4-chloro-3-indolyl phosphate (BCIP) solution. After about 3 min, a colour change was noted. The positive control (external human interferon standard) and the negative control (untransformed X-33 culture supernatant) were also run in each set of assays.

Results

Construction of multicassette expression vectors and Pichia recombinant strains

Multicassette expression vectors carrying two, three, four and five copies of the IFNα2b gene were constructed by using BamHI-digested and calf alkaline phosphatase-treated pPICZαIFNα2bC-1, pPICZαIFNα2bC-2, pPICZαIFNα2bC-3 and pPICZαIFNα2bC-4 as vectors and ligating them to the expression cassette, followed by transforma- tion in TOP10F′ E. coli followed by screening and restriction digestion analysis. Finally, all five BamHI-digested expression vectors were com- pared on the basis of their size by performing agarose gel electrophoresis. These five BamHI- digested plasmids appeared at 4, 6, 8.1, 10.2 and 12.3 kb, successively, from pPICZIFNα2bC-1 to pPICZIFNα2bC-5, which is equal to the size calculated theoretically (Figure 3).

Figure 3. Agarose (1%) gel of BamHI-digested pPICZαIFNα2bC-1 to pPICZαIFNα2bC-5 constructs with λDNA/HindIII Plus marker constructed using an in vitro multimerization approach. Lane 1, λDNA/HindIII Plus marker; lane 2, BamHI-digested pPICZαIFNα2bC-5; lane 3, BamHI-digested pPICZαIFNα2bC-4; lane 4, BamHI-digested pPICZαIFNα2bC-3; lane 5, BamHI-digested pPICZαIFNα2bC- 2; lane 6, BamHI-digested pPICZαIFNα2bC-1.

All constructs were transformed in circular form ex- cept pPICZα and pPICZαC-1, which were linearized with SacI. Wild-type X-33 and X-33 transformed with pPICZα without insert (IFNα2b gene) were used as control. To reduce the number of potential false positives, initially the transformed colonies were picked from 100 μg/ml Zeocin plates, restreaked several times and then finally shifted to Zeocin-free YPD broth for 96 h. This resulted in substantial reduction of recombinant clones (Table 2), only those having intragenomic integration of transformed constructs surviving.

Genomic DNAs of 47 clones obtained from transformation of strain X-33 with different constructs were isolated; 0.8% agarose gels were run to check that there was no shearing of genomic DNA (Figure 4). The levels of Zeocin resistance (Table 3, Figure 5) and methanol utilization phenotypes (Table 4, Figure 6) of all selected recombinant clones were determined. The results of methanol phenotype were the same both with MM medium and by PCR (Figure 7). The majority of transformants gave two bands, one of 1 kb and another of approximately
2.2 kb. These were Mut+ integrants (in these the AOX1 gene is intact); their 2.2 kb bands correspond to the AOX1 gene and the 1 kb band to the 510 bp of IFNα2b gene and 497 bp of our parent plasmid. Some were Muts integrants; these gave only one band (in these the AOX1 gene was replaced as the consequence of a second recombination, due to the accumulation and later expression of IFNα2b was induced by methanol in BMMY medium. After 120 h of induction, each culture supernatant was removed by centrifugation and was directly applied for qualitative and quantitative study of the IFNα2b protein. SDS–PAGE analysis of the samples obtained at various induction times during the shake-flask culture showed the presence of an IFNα2b band at 48 h and continuous increase of the target protein up to 120 h (Figure 8), with a rapid increase during the 48–72 h period. The same finding was obtained using the Bradford assay.

Figure 5. Determination of Zeocin resistance level of finally selected P. pastoris transformants

Figure 6. Difference in growth of Mut+ and Muts clones on MMH and MDH media. (A) Left plate contains MMH, on which C-3(3) growth is less and resembles KM71 (Muts control), so C-3(3) has the Muts phenotype. Right plate contains MDH medium. (B) All the clones are Mut+

Figure 7. PCR of the P. pastoris genomic DNA of transformants, using 5′ AOX1 promoter and 3′AOX1 terminator primers. Lanes 1–12, recombinant P. pastoris clones. A 1 kb band was observed in lane 10 → C-2(18); lane 9 → C-3(3); and lane 1 → C-5(1); these are Muts (AOX1 replaced). Transformants producing two bands are Mut+ integrants (AOX1s remain intact); their 2.2 kb band corresponds to the AOX1 gene and the 1 kb band to 510 bp of the IFNα2b gene and 497 bp of our parent plasmid. Lane 13, pPICZα vector transformed yeast cells (control); the expected band of 500 bp was observed. Lane 14, wild- type X-33 (control); the expected 2 kb band was observed. Lane 15, 250 bp to 10 kb ladder.

Protein concentration was determined by using NIH ImageJ analysis software. As the highest IFNα2b levels were achieved at 120 h post-induction, Coomassie-stained SDS–PAGE gels of 120 h culture supernatants were scanned.

The expression levels of all recombinant clones were calculated (in mg/l) against an external human interferon standard, which was run in different concentrations to establish a standard curve for ImageJ analysis. While relating expression levels to the levels of Zeocin resistance, average expres- sions of 14.91 (± 4.2), 16.97 (± 2.5), 22.9 (± 5.8), 10.82 (± 3.2) and 8.7 (± 1.5) mg/l were achieved for 100, 250, 500, 750 and 1000 μg/ml Zeocin resistance, respectively (Figure 9). This shows that a high level of expression could not be related to high antibiotic resistance. Immunoblot analysis (Figure 10) using anti-human IFNα2b-specific mono- clonal antibody verified that the putative protein bands observed in SDS–PAGE were IFNα2b.

Discussion

IFNα2b is a member of the multifunctional cytokines produced by vertebrates as early inflammatory proteins in response to foreign invaders. It is secreted by virus-infected cells and plays an important role in the first line of defence protecting the host against parasitic and viral infections, along with many kinds of malignancies (Ortaldo et al., 1983; Pestka et al., 1987; Pestka, 2007). The goal of this study was to examine the influence of gene dosage at the expression level of IFNα2b in relation to Zeocin resistance. To achieve this purpose, we constructed a series of P. pastoris X-33 transformants carrying progres- sively increasing copies of the hrIFNα2b gene. To produce multicopy P. pastoris strains, in vivo and in vitro methods are commonly-used strategies. However, the in vivo method involves the selection of clones highly resistant to selection marker (Nordén et al., 2011), accompanied by extensive screening to find high-copy ‘jackpot’ clones (Wu et al., 1999; Mansur et al., 2005). In traditional in vivo methods of His+ selection or post-transformational vector amplification, a large number of false positives are obtained even after extensive screening (Sunga et al., 2008; Zhu et al., 2009; Nordén et al., 2011). Selection of clones by using high drug resistance may result in skipping of good producers having low antibiotic resistance. To avoid immense screening and to ensure the enrichment of multicopy transformants, we have employed a two-step strategy. In the first step, we performed in vitro multimerization of the IFNα2b gene in expression vectors (Figure 2), followed by transformation of P. pastoris X-33 stain with these plasmids and initial screening of transformants at 100 μg/ml Zeocin. In the second step, to obtain stable transformants, the yeast cells were allowed to grow in selection-free medium (without Zeocin) for several generations, so that transiently transformed recombinant strains having unintegrated plasmids lost their plasmids and stable transformants having integrated plasmids could be selected after 96 h, when the cells were again shifted to Zeocin-containing plates (selection pressure). We observed substantial reduction of clones after this step (of 174, only 61 showed growth after 96 h selection growth; Table 2). In addition, the number of clones surviving after 96 h growth in a selection-free environment showed close correspondence with the pattern of growth that was observed before selection-free growth (Table 2). This indicates that clones with excellent growth are the true transformants, while the others appear to be pseudotransformants or false positives. With regard to intragenomic integration of the IFNα2b gene, more integration efficiency of linearized DNA transformed by Easycomp™ transformation was observed in contrast to the circular form transformed by were not integrated into genomic DNA are retained in the transformed cells for a few generations but eventually would be lost, since they cannot replicate in the cell due to the absence of yeast origin of replication in these vectors. This perhaps lowers the integration efficiency of circular transformed vectors.

An expression study of all P. pastoris clones constructed for this study was carried in shake- flask cultures. Although culture conditions for expression were not fully optimized, the best expression clone for hrIFNα2b production could be chosen by making comparisons at this level as preliminary criteria. The use of ImageJ has been reported by Ayed et al. (2008) and Nordén et al. (2011) for comparing the expression levels of hrIFNα2b and aquaporins, respectively. We also analysed expression levels in culture supernatants by performing densitometric analysis, using ImageJ software, of the SDS–PAGE Coomassie-stained gels of 120 h culture supernatants. The SDS–PAGE study of the samples taken at different induction times showed a continuous increase of the target protein up to 120 h (day 5 post-induction; Figure 8). The optimum expression was generally obtained at 20–25 °C after 120 h post-induction.

In contrast to many other proteins expressed in P. pastoris (Zhu et al., 2009; Mansur et al., 2005; Werten et al., 1999; Liu et al., 2005), recombinant human IFNα2b expression was not shown to be closely associated with the level of resistance to Zeocin (Figure 14). Commonly, clones selected at high Zeocin concentrations are expected to express high levels of the recombinant protein of interest, in contrast to ones selected at low Zeocin concentrations, as high levels of Zeocin re- sistance indicate the integration of multiple plasmid copies, and hence the gene of interest (Higgins et al., 1998), whereas in our study the average expression level was not found to increase with increasing Zeocin concentration, but a marked decrease in expression at 500 and 1000 μg/ml Zeocin was observed (Figure 9). This indicates the loss of expression cassettes in Zeocin hyper-resistance clones. During homologous recombination in yeasts, loss of several expression cassettes has also been reported (Nordén et al., 2011), where low- copy clones were recovered at high antibiotic concentrations.

Instead of traditional Southern blotting, recently real-time PCR to quantify integrated expression cassettes in the P. pastoris genome was successfully applied (Zhu et al., 2009; Nordén et al., 2011; Li et al., 2012). A similar approach has been used in the present study. The reproducibility of qPCR is judged by its efficiency value (as reviewed by Rebrikov and Trofimov, 2006). The amplification efficiency, E, of GAP (as standard) was 101.8% and of IFNα2b was 101%, which is in the acceptable range. qPCR results demonstrated that increase in IFNα2b copy number significantly enhanced the yield of IFNα2b protein, suggesting a close correspondence between gene dosage and final expression level (Figure 12). This higher yield may be attributed to the enhanced transcriptional and translational level, making this relationship very simple (Sreekrishna et al., 1997; Vassileva et al.,2001). This also correlates with many reports of heterologous expression in P. pastoris, where a direct association between copy number of inserted sequences and expression level was observed (Clare et al., 1991; Vedvick et al., 1991; Brierley, 1998; Mansur et al., 2005). Besides copy number, the methanol phenotype, the drug resistance of a clone and the number of integrated cassettes also seemed to have influence on the expression of IFNα2b. Recombinant strains C-4 (2) and C-2 (18) gave high production of IFNα2b as compared to other clones having the same copy numbers of IFNα2b gene. This might be because of their lower Zeocin resistance, indicating lesser burden on the transcriptional and translational machinery by shble genes and hence making the machinery available for high production of IFNα2b. As we have observed, transformants which were resistant to high concentrations of Zeocin did not perform well when IFNα2b expression was checked. The methanol phenotype of C-2 (18) was MutS, which might make this clone an efficient producer. It has already been reported that some methanol phenotypes may be more advantageous than others, which may vary from case to case, as in the case of hepatitis B surface antigen production; Muts strains have been found to be better producers (Cregg et al., 1987). In our study, a clone having high copy number, less Zeocin resistance and fewer integrated cassettes appeared more advantageous for IFNα2b expression than clones having high copy numbers and high Zeocin resistance.

In conclusion, these results suggest that gene dosage has crucial importance in optimizing rhIFNα2b expression, as a positively relationship was found between them. Therefore, recombinant strains having maximum integrated IFNα2b expression cassettes should be applied for large- scale industrial production. These results also suggest that drug hyper-resistance does not always lead to higher expression, as integration of a drug-resistance gene does not correlate to integration of the gene of interest. The present study proposes that instead of drug hyper- resistance clone selection criteria, all the trans- formants should be screened individually when the optimum expression is desired. All these findings would be helpful in optimizing rhIFNα2b production in P. pastoris, adding to its efficiency and economy.