A New Signal Sequence for Recombinant Protein Secretion in Pichia pastoris

Abstract

Pichia pastoris is one of the most widely used expression systems for the secretory expression of recombinant proteins. The secretory expression in P. pastoris usually makes use of the prepro $MAT{\alpha}$ sequence from Saccharomyces cerevisiae, which has a dibasic amino acid cleavage site at the end of the signal sequence. This is efficiently processed by Kex2 protease, resulting in the secretion of high levels of proteins to the medium. However, the proteins that are having the internal accessible dibasic amino acids such as KR and RR in the coding region cannot be expressed using this signal sequence, as the protein will be fragmented. We have identified a new signal sequence of 18 amino acids from a P. pastoris protein that can secrete proteins to the medium efficiently. The PMT1-gene-inactivated P. pastoris strain secretes a ~30 kDa protein into the extracellular medium. We have identified this protein by determining its N-terminal amino acid sequence. The protein secreted has four DDDK concatameric internal repeats. This protein was not secreted in the wild-type P. pastoris under normal culture conditions. We show that the 18-amino-acid signal peptide at the N-terminal of this protein is useful for secretion of heterologous proteins in Pichia.

Introduction

Signal peptides control the entry of virtually all proteins to the secretory pathway, both in eukaryotes and prokaryotes [12, 16, 26]. They comprise the N-terminal part of the amino acid (aa) chain and are cleaved off while the protein is translocated through the membrane. In some cases, the presence of N-terminal signal sequences leads to strong stimulation of the expression of recombinant proteins [20]. With very few exceptions, all the proteins secreted from eukaryotic cells require a N-terminal signal peptide. There are a couple of online tools available for predicting signal peptides and corresponding cleavage sites in protein constructs based on their aa sequence [3]. Stephan Klatt and Zoltan Konthur have successfully demonstrated that minor modifications in the aa sequence of the natural SP from SAP1 based on in silico predictions, following the (-3, -1) rule, resulted in different expressionsecretion rates of the protein of interest [21].

Pichia pastoris (P. pastoris) has very few known secretion leaders for secretion of heterologous proteins. The most widely used signal sequence is the S. cerevisiae MATα signal sequence. That apart, the P. pastoris acid phosphatase PHO1 signal sequence is being used in a few cases [7, 17] and the S. cerevisiae SUC2 gene signal sequence has been used occasionally [15]. The classical mammalian bovine β-casein signal peptide was evaluated for the secretory expression of recombinant xylanase in Pichia. It was found that more recombinant xylanase was trapped intracellularly with the β-casein signal sequence when compared with the MATα signal sequence [6].

Insulin and insulin analog expressed in P. pastoris have O-glycosylated species [4, 10]. The inactivation of various protein mannosyltransferase (PMT) genes present in the Pichia genome helps in reduction of O-mannosylated species, which is considered an impurity. We have reported previously that the PMT1 gene knockout in P. pastoris insulin strain reduces the O-glycosylation of insulin precursor by ~60% [4]. Analysis of the extracellular protein profile of all these PMT-gene-inactivated strains revealed that a ~30 kDa protein was being secreted into the medium from the PMT1-gene-inactivated strain (Fig. 1). No other PMT-gene-inactivated strains showed this protein band. All other PMT-gene-inactivated strains were similar to the parent strain. The protein band was subjected to Nterminal sequencing by Edman degradation method. The amino acid sequence deduced was “DQTFGVLLIRSG”. This sequence was matching with the mature portion of a hypothetical protein (XP_002492554.1). The 18 amino acids at the N-terminus were not part of the secreted protein. We were keen to understand why this protein is being secreted only upon PMT1 gene inactivation. We attempted to investigate whether these 18 amino acids have any signal sequence property and can be used for heterologous protein secretion in P. pastoris. We have first cloned this 30 kDa protein under the control of the strong AOX1 promoter inframe with this new signal sequence and checked for secretion in the wild-type P. pastoris strain. Later, we developed various expression constructs and transformed them into P. pastoris host to evaluate and characterize the signal sequence. The resulting strains were induced with methanol in shake flask experiments. Our results show that the 18-amino-acid signal peptide indeed has potential to secrete the heterologous proteins efficiently into the medium. Moreover, this signal sequence is cleaved off independently as efficiently as with the use of Kex2p cleavage sites. Hence, this signal sequence is more useful for expressing recombinant proteins with internal dibasic amino acids such as KR and RR (Kex2p cleavage sites). We also studied the expression of porcine carboxypeptidase B (CPB) [14] and Erythrina trypsin inhibitor (ETI) [23] from plant origin to show that this signal sequence is robust and can be used for secretion of many recombinant proteins that can be expressed in P. pastoris.

Fig. 1.Secretion of a 30 kDa DDDK protein in the PMT1 gene inactivated host. The P. pastoris strains were grown and induced with methanol using the standard protocol described in the Invitrogen manual under identical conditions. Equivalent volumes (15 μl) of secreted samples were analyzed on Tricine SDS-PAGE. Lane 1: P. pastoris GS115; lane 2: Insulin precursor (IP)-secreting P. pastoris clone; lane 3: PMT1 gene inactivated IP-producing clone; and lane 4: low molecular weight protein marker (Thermo Scientific, Part No. 26632). It is evident from the gel that a 30 kDa protein is being secreted to the medium in the PMT1-gene-inactivated host strain.

 

Materials and Methods

Strains, Plasmids, and Media

The P. pastoris strain BICC 9450 (MTCC 5835) [5] and its derivative BICC 9682 (his−) hosts were used in the current studies. The modified pMBL 208 expression vector, which provides the promoter and terminator sequences of the P. pastoris AOX1 gene, were employed for expression. The S. cerevisiae MATα signal sequence was replaced with the “MFNLKTILISTLASIAVA” signal sequence for secretion of proteins into the medium. E. coli strain DH5α was used for routine cloning and propagation of plasmids. Yeast extract-peptone-dextrose (YPD) medium containing 10 g/l yeast extract, 20 g/l peptone, and 20 g/l dextrose was used for routine growth and subculturing of Pichia strains. YNBD medium used for selection contained 13.4 g/l yeast nitrogen base (YNB) without amino acids and 20 g/l dextrose (D). Luria broth/ agar was used for culturing E. coli. Media components used were either from Himedia (Mumbai, India) or Difco (USA).

Amino Acid Sequencing

The 30 kDa protein secreted into the medium by the PMT1-gene-inactivated P. pastoris strain was separated on 10% SDSPAGE using a standard procedure [18]. The protein bands were immobilized on to Sequi-blot-PVDF membrane, 0.2 μM (Bio-Rad Laboratories, USA) by transfer using 25 mM Tris, 192 mM glycine, and 20% methanol [24]. After transfer, the membrane was rinsed thoroughly with distilled water, followed by 50% methanol solution and once again with distilled water to remove any glycine traces that would be carried over from the transfer buffer. The protein bands were visualized on the PVDF membrane by staining with 0.1% Coomassie blue in 1% acetic acid and 40% methanol. The membrane was destained using 50% methanol. The 30 kDa band was excised and sequenced at ProSeq (3 King George Drive, Boxford, MA 01921, USA).

Construction of the Vectors for Secretory Expression in P. pastoris

The gene corresponding to the DDDK protein ORF was amplified along with the signal sequence using primers DDKFP and DDKRP (Table 1) from P. pastoris genomic DNA. The 870 bp PCR product was cloned into pTZ57R vector and the sequence verified. The gene was excised using BamHI and EcoRI restriction enzymes and subcloned into pMBLNSS208 vector in BamHI and EcoRI sites to obtain DDK/pMBLNSS208 vector, which replaces the MATα signal sequence. The DDDK gene was now under the regulatory control of the AOX1 promoter (Fig. 2B). The DDDK protein ORF without the signal sequence was amplified and cloned similarly to have a control for the above (Fig. 2A), but the forward primer used was DDK(-SS)FP in place of DDKFP. These two constructs were used to study whether the strong promoter can lead to secretion of this protein in the wild-type Pichia. The constructs in Fig. 2C are similar to those in Fig. 2B but with the Kex2p site (KR) at the end of the signal sequence. This construct was developed to evaluate the level of secretion with and without Kex2p cleavage site.

Table 1.Primers used for genetic manipulation of the strains developed in this study.

Fig. 2.Constructs developed to evaluate the signal sequence properties of the 18 amino acids at the N-terminus of DDDK protein under the regulatory control of the AOX1 promoter. The signal sequence part is designated with a single letter amino acid code. (A) The control construct with DDDK protein mature sequence without any signal sequence under the AOX1 promoter. (B) Complete open reading frame of the DDDK protein with native signal sequence. (C) Same as (B) but with Kex2p cleavage site ‘KR’ introduced after the native signal sequence, which is generally being used in the P. pastorissystem for efficient secretion of proteins to the medium. (D) The heterologous protein porcine CPB coding sequence cloned inframe with the signal sequence. (E) The heterologous protein ETI coding sequence cloned inframe with the signal sequence.

Synthesis of Signal Sequence and Development of CPB and ETI Clones

The signal sequence was synthesized using the primers NSSFP1, NSSFP2, NSSFP3, NSSRP1, NSSRP2, and NSSRP3 (Table 1). The primers were phosphorylated by treating with T4 polynucleotide kinase, annealed, and ligated. The full-length signal sequence was amplified from the ligated template using NSSFP1 and NSSRP1. The PCR product was cloned into pTZ57R vector and the sequence verified. The signal sequence was cloned into pMBL208 vector to replace the MATα signal sequence.

The CPB and ETI coding sequences were amplified using the forward primers NSSVACPB and NSSVAETI and pPIC9K reverse primer from the plasmids CPB/pPIC9K and ETI/pPIC9K, which we had expressed previously in our laboratory. The forward primer has the ClaI restriction site. The PCR product was digested with ClaI and EcoRI restriction enzymes and cloned into the DDK/pMBLNSS208 vector in the identical sites to yield the constructs in Figs. 2D and 2E.

Transformation of P. pastoris

The plasmid constructs were digested with SacI and transformed into the P. pastoris (his−) strain. Transformation was carried out by electroporation of freshly prepared competent cells in 0.2 cm cuvettes. The pulses were delivered by a Gene Pulser (BioRad) at 1,500 V, 25 μF, and 200 Ω. The electroporated cells were allowed to recover for 1 h in 1 M Sorbitol at 30℃ and then spread on to YNBD agar plates. The resulting transformants were selected on YPD plates, containing 0.5 mg/ml of G418 concentration, to identify multicopy clones. Several clones that were viable at 0.5 mg/ml concentration were tested in small-scale induction experiments. The proteins secreted to the extracellular medium were analyzed on 10% SDS-PAGE.

Induction of Protein Expression in P. pastoris

The clones were grown on 10 ml of YNB medium (1.34% yeast nitrogen base w/o amino acids and 2% dextrose) overnight. It was subcultured into 50 ml of BMGY (1% yeast extract, 2% peptone, 1.34% YNB, 100 mM potassium phosphate at pH 6.0, and 1% glycerol) to 0.5 OD600. This culture was incubated at 30℃ for 48 h with shaking at 220 rpm. The cells were harvested by centrifugation at 5,000 rpm for 5 min at room temperature and resuspended in induction medium (1% yeast extract, 2% peptone, 1.34% YNB, and 100 mM potassium phosphate at pH 6.0). Methanol (20%) was added to the culture medium to give a final concentration of 1.0% methanol as an inducer. Samples were collected every day and analyzed by SDS-PAGE.

Western Blotting

Western blotting was carried out as per the method described by Towbin et al. [25] with a few modifications. The protein samples were separated on 12% SDS-PAGE and then transferred to a Nitrocellulose sheet, 0.2 μM (Bio-Rad Laboratories, USA), as described above. After transfer, the membrane was soaked in 1% BSA in Tris-NaCl buffer (20 mM Tris, pH 7.2, 150 mM NaCl) and incubated overnight to block the nonspecific binding. The membrane was washed with wash buffer (20 mM Tris, pH 7.2, 150mM NaCl, 0.05% Tween-20) thoroughly, and then probed with an anti-DDK monoclonal antibody (OriGene Technologies, Inc, Rockville, MD, USA) at 1:100 dilution with Tris-NaCl buffer containing 0.1% BSA. The DDK protein was identified with a polyclonal sheep anti-mouse IgG-HRP conjugate (product code: NA931; GE Healthcare, UK) using TMB substrate.

 

Results

Amino Acid Analysis and Identification of Signal Sequence

The 30 kDa protein secreted in the PMT1-gene-inactivated P. pastoris strain was immobilized onto Sequi-blot-PVDF membrane and submitted for N-terminal aa sequencing at ProSeq Inc., USA. The first 12 aa were found to be “DQTFGVLLIRSG”. This sequence was searched against the NCBI database using the BLASTp program to find out the probable homologous protein. It resulted in the identification of a 289-aa-containing hypothetical protein in the P. pastoris GS115 genome database [19]. Interestingly these 12 aa are matching from the aa 19D to 27G in the complete ORF of the hypothetical protein (XP_002492554.1). There are 18 aa at the N-terminal of the protein acting as a signal sequence (Fig. 3), which seem to be cleaved off during the secretion of the mature protein into the medium. The theoretical mass of this protein was found to be 31.66 kDa with a pI of 4.44 when calculated using ExPASY protein parameter tools (http://web.expasy.org/ protparam/). It is a highly acidic protein with four ‘DDDK’ aa repeats at a stretch in the primary aa sequence.

Fig. 3.Amino acid sequence of the 30 kDa DDDK protein identified by searching against the P. pastoris database. The N-terminal sequencing of the 30 kDa band resulted in the amino acids D19 to G30, which are underlined. The protein has four repeats of DDDK in a stretch from D133 to K148, which are italicized and underlined. Hence, we named this as the DDDK protein. The amino acids M1 to A18, which are in bold, correspond to the signal sequence studied for heterologous protein expression in this study.

Secretion of DDDK Protein Under AOX1 Promoter in Wild-Type P. pastoris

In order to evaluate whether this protein will be secreted if fused with the strong promoter AOX1, the 870 bp complete DDDK ORF (which includes the coding sequence of 18 aa signal peptide) was cloned under the regulatory control of the AOX1 promoter (Fig. 2B). This construct was transformed into the P. pastoris wild-type strain. The clones obtained were screened for the secretory production of the DDDK protein. We found good secretory expression of the DDDK protein in the wild-type P. pastoris strain (Fig. 4). This indicates that the promoter of the native protein is indeed very weak and may not be able to produce a sufficient amount of protein to secrete into the medium, although it has a potential signal sequence at the Nterminus. The results shown in Fig. 4B indicate that the proteins secreted with and without the Kex2p cleavage site are similar.

Fig. 4.SDS-PAGE analysis of the secreted 30 kDa DDDK protein. (A) The 30 kDa DDDK protein secreted into the medium when expressed under the regulatory control of AOX1 promoter in wild-type P. pastoris. Lane 1: P. pastoris host strain; lane 2: P. pastoris host carrying DDDK gene CDS without the 18-amino-acid signal peptide construct (Fig. 2A); lanes 3-5: Clones #1, 2, and 3 of P. pastoris host carrying construct of Fig. 2B that has the DDDK gene CDS along with 18-amino-acid signal sequence. (B) Comparison of 30 kDa DDDK protein secretion when Kex2p cleavage site was inserted after the 18-amino-acid signal sequence. Lane 1: P. pastoris host carrying construct that has DDDK gene CDS without the 18-amino-acid signal peptide (Fig. 2A); lane 2: protein molecular weight marker; lane 3: P. pastoris host carrying construct that has DDDK gene CDS along with the 18-amino-acid signal sequence (Fig. 2B); and lane 4: P. pastoris host carrying construct that has DDDK gene CDS along with the 18-amino-acid signal sequence plus KR (Fig. 2C), the Kex2p cleavage site. There was no difference in the level of protein secretion in lanes 3 and 4. This indicates that the signal sequence can secrete the protein efficiently without the need for Kex2p.

Identification of Production of DDDK Protein Using Antibody in Wild-Type P. pastoris Host

To understand whether the DDDK protein is being secreted only in strains with PMT1-gene-inactivated host and also to identify the presence of any intracellular proteins, a western blot experiment using a monoclonal antibody that binds to proteins that have the DDDK motif was carried out. The protein band that corresponds to DDDK protein was not detected in the supernatants of Pichia wild-type strain. (lane 1, Fig. 5), indicating that the protein is being secreted only when the PMT1 gene is inactivated. It was also noticed that there was no band observed in the intracellular fraction of Pichia (lanes 5 and 6, Fig. 5), indicating that all the protein expressed had been successfully secreted to the medium with use of the DDDK protein signal sequence.

Fig. 5.Western blot analysis of the 30 kDa DDDK protein to identify whether any protein accumulates intracellularly as a result of non-processing of signal sequence. (A) Day 4 methanol-induced culture supernatant samples were loaded (15 μl) in each lane, except lanes 5 and 6 where 0.5 OD600 cell lysates were loaded. Lane 1: P. pastoris wild-type strain; lanes 2 and 3: P. pastoris host carrying construct shown in Figs. 2B and 2C, respectively; lane 4: Protein molecular weight marker; and lanes 5 and 6: Intracellular proteins of P. pastoris host carrying construct shown in Figs. 2B and 2C, respectively. (B) The proteins from the replicate of the gel shown in Fig. 5A was transferred to a nitrocellulose sheet and probed using anti-DDDK antibody as described in the Methods. The detection of 30 kDa bands in lanes 2 and 3 indicates that the antibody is binding specifically to DDDK protein. The absence of the corresponding bands in lanes 5 and 6 indicates that all the protein made has been secreted efficiently to the medium.

Secretion of CPB and ETI Using DDDK Protein Signal Sequence

In an attempt to determine whether the signal sequence functions well for secreting heterologous proteins, we developed constructs having CPB and ETI coding sequences fused inframe with this signal sequence (Figs. 2D and 2E). The constructs were transformed into the P. pastoris wildtype strain. The clones obtained were screened for the secretion of these proteins. We found the good secretory expression of CPB and ETI proteins, which were comparable to the quantity of the same proteins produced using the MATα signal sequence (Fig. 6).

Fig. 6.Porcine carboxypeptidase B expressed using the DDDK signal sequence has been secreted into the medium as efficiently as that secreted with the MATα signal sequence. Lane 1: Protein molecular weight marker; lane 2: P. pastoris wild-type strain; lanes 3, 4, and 5: Clones #1, 2, and 3 of P. pastoris host carrying construct that has Pro-CPB CDS with the 18-amino-acid signal peptide (Fig. 2D); and lane 6: P. pastoris host carrying Pro-CPB CDS with the MATα signal sequence. (B) The level of Erythrina trypsin inhibitor (ETI) expressed using the DDDK signal sequence secreted into the medium is comparable to the one secreted with the MATα signal sequence. Lane 1: Protein molecular weight marker; lane 2: P. pastoris host carrying ETI CDS with the MATα signal sequence; lanes 3 and 4: Clones #1 and 2 of P. pastoris host carrying construct that has ETI CDS with the 18-amino-acid signal peptide (Fig. 2E); and lane 5: P. pastoris host strain. All lanes carry the equivalent volume of secreted samples (15 μl).

 

Discussion

The DDDK protein is secreted into the culture medium only in the PMT1-gene-inactivated P. pastoris strain and not in any other PMT-gene (PMT4, 5, and 6)-inactivated strains. It was not clear why it was secreted in this strain alone. PMT1 is an integral membrane protein [22]. Although the doubling time of the PMT1-gene-inactivated insulin precursor producing clone was slightly lower in YPD broth, the cell mass accumulation was not affected and similar results were observed for all the other PMT-gene-inactivated strains [4]. It was also noticed from the tricine SDS-PAGE [8] gel picture that the cells were not lysed, owing to the PMT1 gene inactivation (Fig. 1). If there was cell lysis, one could find other cellular proteins, as shown in Fig. 5, lanes 5 and 6. We sequenced the N-terminal 12 amino acids of the 30 kDa protein band and found the PIR gene family of proteins [11] having four concatameric repeats of amino acids “DDDK”. This protein is found to be unique to P. pastoris when analyzed using the protein databank. The PMT1 gene was previously inactivated in S. cerevisiae and shown by Strahl-Bolsinger et al. [22] to be a key enzyme in the protein O-glycosylation. When we analyzed the 12- amino-acid sequence from the N-terminus using protein BLAST searches, homology to cell wall protein CPW1 from the Pichia genome was identified based on the S. cerevisiae database. When the full-length DDDK protein was compared with the Saccharomyces cell wall protein (CPW1/YKL096W) sequence using NCBI BLAST2p protein search, however, we found only 33% identity and 58% similarity (data not shown). The DDDK repeats were not present in the S. cerevisiae cell wall protein. However, when the stretch of four DDDK repeat sequence was used for search using the NCBI BLAST2p program, a few sequences (GenBank Accession No. ELA36046.1, EKG21897.1, CCH44912.1, and XP_001349335.1) were identified. All these deposits are part of sequencing projects and were reported as hypothetical proteins. The N-terminal 18 aa (Fig. 3) was not part of the secreted protein, emphasizing once again that the DDDK protein has not come out of the cell by cell lysis but because of the presence of the signal sequence. We were curious to understand why this signal sequence is not secreting the DDDK protein in the wild-type strains and also to characterize the same sequence for heterologous protein expression in P. pastoris.

Currently, the P. pastoris expression system to secrete heterologous proteins is mainly based on the 85-aminoacid S. cerevisiae MATα signal sequence [1, 2]. A Kex2p cleavage site “KR” at the end of the MATα signal sequence cleaved off during the export of recombinant proteins to the medium [9]. One of the drawbacks of this powerful expression technology is that, if there are any accessible internal dibasic amino acids present in the recombinant protein sequence, one cannot use the current system. The cleavage and productivity will also depend on the protein N-terminus [13]. The signal sequence identified will be useful to express recombinant proteins having internal Kex2p cleavage sites using the Kex2p-disrupted P. pastoris strain. The Kex2 protease-inactivated Pichia strain is viable [27].

In the current study, we have demonstrated that the signal sequence of a DDDK protein can be used for the expression of recombinant proteins. Initially, we cloned the DDDK protein along with its signal sequence under the AOX1 promoter (Fig. 2B) and successfully secreted the protein in wild-type P. pastoris strain (Fig. 4). The results of this experiment indicate three things. First, the 18-aminoacid signal sequence is able to secrete the proteins into the medium; secondly the strong promoter is able secrete the protein in wild-type P. pastoris using the native DDDK signal sequence; and thirdly, the levels of protein secreted with and without the Kex2p sites are similar (Fig. 4B lanes 3 and 4). It was further demonstrated with the use of anti-DDK antibody that the protein is being secreted to the medium only if the PMT1 gene is inactivated, as there was no detectable DDDK protein band observed in the culture supernatant of wild-type Pichia, as shown in lane 1, Fig 5. The induced cells were lysed using glass beads and separated by SDS-PAGE along with the corresponding secretory protein. The western blot of the same gel demonstrated that the DDDK protein signal sequence was very efficiently processed, and all protein expressed was secreted to the medium as there was no intracellular protein accumulation, as shown in Fig. 5 lanes 5 and 6. There was another prominent protein band that showed up between 116 and 66 kDa. We have not yet characterized this protein band. This could be a protein complex, being formed by interaction between the mature DDDK protein and PMT1 protein. The mRNA transcripts levels of DDDK protein in wild-type Pichia and PMT1-inactivated Pichiastrains are similar (data not shown). However, the mature DDDK protein is not being secreted from the wild-type strain, possibly due to interaction between DDDK and PMT1 proteins. This hypothesis regarding the interaction between these two proteins as a controlling step preventing the secretion of DDDK protein could be an interesting role that could be the object of a later study.

We also successfully expressed the heterologous recombinant proteins such as porcine CPB and ETI (Fig. 6). This is an indication that the signal sequence can be used for heterologous protein expression. We have determined the intact mass of the ETI after gel-eluting the protein band from clone #1 (Fig. 6B lane 3) and found that the intact mass is matching with the theoretical mass (data not shown). This indicates the signal sequence is cleaved off at the appropriate position. The reported new signal sequence will be a useful addition to efficient signal sequences available for heterologous protein expression in Pichia.