Screening for Glycosylphosphatidylinositol-Modified Cell Wall Proteins in Pichia pastoris and Their Recombinant Expression on the Cell Surface

Abstract

Glycosylphosphatidylinositol (GPI)-anchored glycoproteins have various intrinsic functions in yeasts and different uses in vitro. In the present study, the genome of Pichia pastoris GS115 was screened for potential GPI-modified cell wall proteins. Fifty putative GPI-anchored proteins were selected on the basis of (i) the presence of a C-terminal GPI attachment signal sequence, (ii) the presence of an N-terminal signal sequence for secretion, and (iii) the absence of transmembrane domains in mature protein. The predicted GPI-anchored proteins were fused to an alpha-factor secretion signal as a substitute for their own N-terminal signal peptides and tagged with the chimeric reporters FLAG tag and mature Candida antarctica lipase B (CALB). The expression of fusion proteins on the cell surface of P. pastoris GS115 was determined by whole-cell flow cytometry and immunoblotting analysis of the cell wall extracts obtained by β-1,3-glucanase digestion. CALB displayed on the cell surface of P. pastoris GS115 with the predicted GPI-anchored proteins was examined on the basis of potential hydrolysis of p-nitrophenyl butyrate. Finally, 13 proteins were confirmed to be GPI-modified cell wall proteins in P. pastoris GS115, which can be used to display heterologous proteins on the yeast cell surface.

INTRODUCTION

Glycosylphosphatidylinositol (GPI)-anchored proteins are found in all eukaryotic cells. This GPI anchor is essential for viability and maintenance of normal cell morphology in yeasts (1, 2). GPI-anchored glycoproteins serve as structural components, hydrolytic enzymes, surface receptors, and adhesion proteins during mating and formation of flocs, mats, and biofilms. The primary sequences of GPI-anchored proteins share a general pattern, with an N-terminal signal peptide and C-terminal features that mediate GPI anchor addition at an amino acid residue referred to as the omega (ω) site (3). In addition to these signal sequences, GPI-anchored proteins usually contain a Ser/Thr-rich sequence that provides sites for O glycosylation. Mature GPI-anchored proteins do not contain transmembrane (TM) domains because the whole protein is translocated into the lumen of the endoplasmic reticulum (ER). Moreover, cellular localization of GPI-anchored proteins, at least in the yeast Saccharomyces cerevisiae, seems to be partly determined by basic or hydrophobic residues within the ω region (4–6).

Comprehensive lists of GPI-anchored proteins have been compiled through in silico analysis of several yeast genomes (4, 7–9). Furthermore, biochemical evidence has been obtained through metabolic labeling with anchor components such as [³H]inositol, through the demonstration of a loss of hydrophobicity upon treatment with phosphatidylinositol (PI)-specific phospholipase C, by the loss of surface localization upon site-directed mutation of a putative ω site, or by the demonstration of a covalent, mild alkali-resistant, glucanase-sensitive association with the cell wall, for which the addition of a GPI anchor is a prerequisite. The most recent studies involve mass spectrometry analyses of GPI-modified cell wall proteins (10, 11).

In S. cerevisiae, GPI-anchored proteins are found on the plasma membrane and in an intrinsic part of the cell wall. Many GPI-modified proteins of the cell wall are important to build and maintain the stretch resistance of the cell wall components, as simultaneous deletion of multiple GPI-modified wall proteins (Ccw12p, Ccw13p, Ccw14p, Tip1p, and Cwp1p) can lead to decreased osmotic stability and increased mortality (12). Other GPI-anchored proteins act as enzymes to create and break glycosidic linkages and are required for the development of the cell wall and its reshaping during bud emergence, cell separation, mating, or entry into stationary phase. For instance, Gas1p acts as a β-1,3-glucan-specific transglucosidase, Egt2p is an endoglucanase, and Crh1p and Crh2p are chitin transglycosylases containing the glycosidase motif PF00722. Similarly, the proteins encoded by DFG5 and DCW1 contain an α-1,6-mannanase motif (PF03663) and are considered to cleave the Man–α-1,4-GlcN bond of the GPI anchor in order to covalently link GPI proteins to the cell wall glucans (13, 14). Many GPI-modified wall proteins contain a domain that is exposed at the surface of the cell wall to mediate cell adhesion or biofilm formation, such as the flocculins encoded by FLO1, FLO5, FLO9, and FLO11, the sexual agglutinins encoded by SAG1 and AGA1, and possibly, the proteins of the flocculin-related semipauperin family (encoded by TIR1 to TIR4, TIP1, DAN1, and DAN4) (2). Flo1p from S. cerevisiae, which is linked to the cell wall, has been used as an anchor for protein surface display (15). Many GPI-modified cell wall proteins have been studied in the context of their potential to display heterologous proteins. Hamada et al. screened for GPI-modified cell wall proteins in S. cerevisiae and classified a total of 14 GPI-modified cell wall proteins (9). A positive correlation between the Ser/Thr content of GPI-anchored proteins and their tendency to localize to the cell wall has been found to be highly significant (P = 0.005 or 0.027, depending on the criteria used to define cell wall proteins). Moreover, it has been observed that cell wall proteins have an average isoelectric point (pI) of 4.87 ± 0.22, whereas proteins defined as plasma membrane proteins in the same collection have a significantly higher average pI of 6.67 ± 0.95 (2). The hydropathy of the N-terminal signal sequence of yeast secretory proteins has been observed to determine whether they are inserted into the ER cotranslationally or posttranslationally by the signal recognition particle (SRP)-dependent or SRP-independent pathway, respectively; e.g., Gas1p has a low hydropathy signal peptide and is inserted posttranslationally (16).

The methylotrophic yeast Pichia pastoris is a potent expression system with a strong capability to produce recombinant proteins under the control of the alcohol oxidase 1 (AOX1) promoter, resulting in a 1,000-fold induction upon methanol addition. Gas1p in P. pastoris has primarily been reported to be a glycoprotein anchored to the outer layer of the plasma membrane through a GPI anchor. The disruption of GAS1 in P. pastoris production strains for obtaining human trypsinogen and human serum albumin did not result in an enhancement of product secretion, whereas Rhizopus oryzae lipase secretion could be improved 2-fold (17). In addition, GPI-modified cell wall proteins from S. cerevisiae, such as α-agglutinin, Tip1p, Flo1p, and Sed1p, have been used as anchor proteins for P. pastoris cell surface display of heterologous proteins (18–20). However, very few endogenous GPI-modified cell wall proteins of P. pastoris have been confirmed and used in P. pastoris cell surface display. In this study, we screened for GPI-anchored proteins in the P. pastoris genome and identified 13 GPI-modified cell wall proteins by detection of recombinant protein expression on the cell surface. The identified proteins were fused to the FLAG tag and Candida antarctica lipase B (CALB) for further investigations regarding possible applications.

MATERIALS AND METHODS

Database analysis.

Open reading frames (ORFs) in the genome of P. pastoris were screened for potential GPI-anchored proteins using the big-PI fungal predictor described by Eisenhaber et al. (8). The precursor lists are available at http://mendel.imp.ac.at/gpi/fungi/gpi_fungi.html. The predicted potential GPI-anchored proteins were subsequently analyzed for putative N-terminal signal peptides on the basis of import into the ER (SignalP v.4.0 server; http://www.cbs.dtu.dk/services/SignalP/) (21), glycosylation sites (NetNGlyc v.1.0 server and NetOGlyc v.3.1 server; http://www.cbs.dtu.dk/services/) (R. Gupta, E. Jung, and S. Brunak, unpublished data) (22), internal TM domains (TMHMM v.2.0 server; http://www.cbs.dtu.dk/services/TMHMM/) (23), pIs (ProtParam; http://www.expasy.org/tools/protparam.html) (24), and subcellular localization (PSORT II; http://psort.hgc.jp/form2.html) (25).

Strains, plasmids, and cultivation media.

For cloning purposes, Escherichia coli TOP10 was used. Cells were grown at 37°C in Luria-Bertani medium (0.5% yeast extract, 1% tryptone, 1% NaCl) supplemented with 100 μg/ml ampicillin. P. pastoris GS115 and expression plasmid pPIC9K were purchased from Invitrogen Corporation (Carlsbad, CA). The vector pKNS-CALB, which contained the mature CALB cDNA, has been described previously (20). P. pastoris GS115 cells were grown in yeast peptone dextrose medium (1% yeast extract, 2% tryptone, 2% dextrose) and on minimal dextrose (MD) agar plates (1.34% yeast nitrogen base, 2% dextrose, 2% agar) supplemented with 0.4 mg/liter biotin. The buffered glycerol complex (BMGY) and buffered methanol complex (BMMY) culture media (pH 6.0) contained 1% yeast extract, 2% peptone, 1.34% yeast nitrogen base, 1% glycerol (in the case of BMGY) or 1% methanol (in the case of BMMY), 0.4 mg/liter biotin, and 50 mM potassium phosphate buffer. The BMGY and BMMY culture media were used for growth and induction studies, respectively.

Construction and expression of recombinant GPI-anchored proteins.

Gene fragments encoding the predicted GPI-anchored proteins without internal N-terminal signal peptides were amplified from the genomic DNA or cDNA of P. pastoris GS115 by PCR using primers containing MluI and NotI restriction sites. The pKNS-CALB plasmid contains the AOX1 promoter, the alpha-factor secretion signal, the FLAG tag, the mature CALB cDNA, and AGα1 (the gene encoding α-agglutinin), in addition to MluI and NotI restriction sites. The amplified fragments were subsequently inserted into the MluI and NotI restriction sites of pKNS-CALB, respectively, to construct recombinant expression plasmids in which the AGα1 sequence was replaced with the gene fragments encoding the predicted GPI-anchored proteins. The insertions were confirmed by DNA sequencing using an ABI Prism 310 genetic analyzer (Applied Biosystems, Foster City, CA). The constructed plasmids were linearized and subsequently integrated into the host strain, P. pastoris GS115. The resulting transformants were selected by incubation at 30°C for 72 h on MD plates and tributyrin agar plates that were supplemented with 0.5% tributyrin and to which methanol was added to induce the expression of CALB. The isolated transformants were precultured in BMGY medium at 30°C to an optical density at 600 nm (OD₆₀₀) of 2 to 6. Then, the cultures were centrifuged at 6,000 × g for 5 min and resuspended in BMMY medium containing 1% (vol/vol) methanol to an OD₆₀₀ of 1. To maintain the induction of the expression of fusion proteins for 120 h, 100% methanol was added to the culture every 24 h to a final concentration of 1% (vol/vol).

Measurement of lipase activity in recombinant P. pastoris cells.

A modified method for the lipase activity assay was used (26, 27). The lipase activity on the cell surface was measured spectrophotometrically using p-nitrophenyl butyrate (pNPB) as the substrate. pNPB was emulsified by sonication in ultrapure water containing 0.5% Triton X-100, leading to a final concentration of 12.5 mM. A total volume of 1 ml of the assay mixture contained 940 μl of 50 mM Tris-HCl buffer (pH 8.0), 50 μl of substrate solution, and 10 μl of the cell suspension. The cell density was confirmed by OD₆₀₀ measurement, and an appropriate amount of this cell suspension was used for lipase activity measurement. The assay mixture was incubated at 45°C for 5 min and then centrifuged at 10,000 × g at room temperature for 1 min. The activity of CALB was assayed by measuring the absorbance of the liberated p-nitrophenol (pNP) at 405 nm using a kinetic microplate reader (Molecular Devices, Sunnyvale, CA). A 200-μl aliquot of the resultant supernatant was analyzed in a 96-well plate. Average values were generated from triplicates of each sample. One unit of activity was defined as the amount of enzyme required to release 1 μmol pNP/min from pNPB at 45°C. P. pastoris GS115 transformed with pPIC9K (GS115/pPIC9K) was used as a negative control.

Whole-cell flow cytometry analysis.

Whole-cell flow cytometry analysis was performed as described by Kobori et al. (28). The induced cells were washed twice in ice-cold water and resuspended at 4°C in phosphate-buffered saline (PBS; pH 7.4) supplemented with 10 mg/ml of bovine serum albumin. Subsequently, the cell suspension was incubated with an antibody against FLAG (Agilent) at a dilution of 1:200 to a total volume of 200 μl on a rotator at room temperature for 2 h. Then, the cells were washed with PBS and exposed to the secondary antibody, Alexa Fluor 488 goat anti-mouse IgG (H+L; Invitrogen) at a 1:200 dilution to a total volume of 200 μl for 1 h at room temperature. After three washing steps, the cells were examined using a flow cytometer (Beckman Coulter, Fullerton, CA). A total of 10,000 cells were analyzed for each sample, and the data were examined using EXP032 software (Beckman Coulter). The above-mentioned process was also applied to GS115/pPIC9K, which served as a negative control.

Extraction of the cell wall proteins.

Laminarinase and sodium dodecyl sulfate (SDS) extractions of the cell wall proteins were performed as described previously, with minor modifications (29). Cells induced for the expression of the recombinant proteins were harvested by centrifugation at 6,000 × g for 3 min and washed thrice with ice-cold 10 mM Tris-HCl buffer (pH 7.8) containing 1 mM phenylmethylsulfonyl fluoride (PMSF) and 5 mM EDTA. The cells, buffer, and glass beads (diameter, 0.5 mm) were mixed at a ratio of 1:2:1 (wet wt/vol/wt) in a microcentrifuge tube and agitated vigorously using a vortex mixer at maximum speed for 30 s for eight times with an interval of 1 min of incubation on ice. The supernatant and cell wall fraction were separated by centrifugation for 5 min at 6,000 × g. The supernatant was used to analyze the intracellular protein fraction, while the cell wall fraction was washed thrice with the above-mentioned buffer. Noncovalently bound proteins or proteins bound via disulfide bridges were released from the cell wall fraction by boiling for 10 min in 50 mM Tris-HCl (pH 7.8) containing 2% SDS, 100 mM EDTA, and 40 mM β-mercaptoethanol (30). The SDS-extracted cell wall fraction was pelleted by centrifugation for 10 min at 10,000 × g. The remaining fraction was washed several times in 1 mM PMSF to remove the SDS, treated with 0.2 U laminarinase (β-1,3-glucanase; Sigma) in 100 μl of 50 mM sodium acetate (pH 5.4), and incubated overnight at 37°C. After treatment, the extracts containing the covalently bound cell wall protein fractions were separated by centrifugation for 5 min at 10,000 × g.

Western blot analysis of recombinant cell wall proteins.

SDS- and laminarinase-extracted cell wall proteins were separated by polyacrylamide gel electrophoresis (PAGE) using a 12% (vol/vol) gel and then transferred onto polyvinylidene difluoride membranes, as described by Van Der Vaart et al. (31). The recombinant GPI-modified cell wall proteins from the above-mentioned extracts were detected by binding of the primary rabbit anti-FLAG antibody (MBL, Japan), which reacted with the FLAG tag at the N terminus of the recombinant cell wall proteins, and the secondary goat antirabbit antibody (Invitrogen), which was conjugated with horseradish peroxidase. Protein bands were visualized using enhanced chemiluminescence detection reagents (Applygen Technologies Inc., Beijing, China), followed by chemiluminescent exposure of X-ray film.

RESULTS AND DISCUSSION

Prediction and bioinformatics analysis of GPI-anchored proteins of P. pastoris GS115.

Potential GPI-anchored proteins were selected from the ORFs of the P. pastoris GS115 genome on the basis of three criteria: (i) the presence of a C-terminal GPI attachment signal, (ii) the absence of TM domains in the mature protein, and (iii) the presence of a signal sequence for secretion at the N terminus (including GPI-anchored plasma membrane proteins). The ORFs encoding all proteins in the genome of P. pastoris GS115 (http://www.ncbi.nlm.nih.gov/genome/) were downloaded and analyzed using the big-PI fungal predictor and TMHMM. A total of 61 proteins containing a C-terminal GPI attachment signal and not including internal TM domains were selected. Subsequently, these 61 selected proteins were screened, and those without the N-terminal signal sequence were excluded. Finally, 50 putative GPI-anchored proteins (Table 1), representing 0.94% of all the ORFs, were obtained on the basis of the prediction results, similar to those obtained from S. cerevisiae (9). A new web tool, ProFASTA (32), which allows pipeline filtering of proteins with cell surface characteristics by analyzing the output created with SignalP, TMHMM, and big-PI, was also used to predict potential GPI-anchored proteins of P. pastoris GS115, and the same 50 putative GPI-anchored proteins were selected. Gas1p, a known GPI-anchored plasma membrane protein of P. pastoris (17), was among the proteins selected by this screening procedure.

Table 1.

Putative GPI-anchored proteins in Pichia pastoris GS115^a

^aShading indicates information for the confirmed cell wall proteins.

^bGCW1 encodes a β-1,3-glucanosyltransferase (GAS1) of P. pastoris.

^cPT, the tetrapeptide XPTX; SUN, family of proteins corresponding to the genes SIM1, UTH1, NCA3, and SUN4; EGF, epidermal growth factor.

Yeast GPI-anchored proteins often contain stretches of amino acids rich in Ser/Thr, indicating potential O-glycosylation sites. The 50 predicted GPI-anchored proteins were analyzed using the web tool ProFASTA, and the result revealed that almost all proteins displayed a Ser/Thr-rich domain in the C-terminal region (data not shown). In addition, all proteins were predicted to have glycosylation sites, especially O-linked glycosylation sites (Table 1). Furthermore, the possible subcellular localization of the 50 predicted GPI-anchored proteins was predicted by PSORT II, and the predicted outcome showed that most of the proteins were extracellular, including cell wall proteins (data not shown). It has been observed that cell wall proteins have an average pI of 4.87 ± 0.22, whereas plasma membrane proteins have a significantly higher average pI of 6.67 ± 0.95 (2). The pIs of the 50 predicted GPI-anchored proteins were subsequently analyzed, and it was found that the values were between 3.26 and 8.79 and did not show this trend (Table 1).

Identification of potential GPI-modified cell wall proteins of P. pastoris.

It was previously reported that a majority of GPI-anchored proteins are processed and integrated into the cell wall by covalent attachment to cell wall glucans, while only a few of them reside permanently on the plasma membrane (2). To examine whether the 50 putative GPI-anchored proteins have the ability to direct incorporation of a protein into the cell wall, a chimeric reporter gene encoding the FLAG tag and the mature CALB cDNA was used to construct a recombinant expression plasmid with pPIC9K. The fragments encoding putative GPI-anchored proteins without internal N-terminal signal peptides were inserted at the end of the reporter gene and then used for the transformation of P. pastoris GS115. Yeast transformants expressing a chimeric reporter protein fused to each predicted GPI-anchored protein were isolated on MD plates and tributyrin agar plates.

First, the localization of the fusion proteins and their surface accessibility on the yeast cell surface were analyzed by indirect immunofluorescence labeling using the FLAG tag and flow cytometry. The predicted proteins that could successfully direct the fusion proteins with the FLAG tag to the yeast cell surface were considered GPI-anchored proteins. The transformants based on 50 predicted GPI-anchored proteins were induced for 120 h in BMMY medium. Then, these recombinants were collected and labeled with anti-FLAG antibody as the primary antibody and Alexa Fluor 488 goat anti-mouse IgG (H+L) as the secondary antibody. The flow cytometry analysis revealed that 16 recombinants showed increased fluorescence intensity on the cell surface compared with that of the control strain, GS115/pPIC9K (Fig. 1), whereas the remaining 34 recombinants hardly produced any fluorescence intensity on the cell surface (data not shown). This suggests that these 16 predicted GPI-anchored proteins can direct fusion proteins to the cell surface efficiently in a P. pastoris expression system. Thus, these 16 proteins were preliminarily considered GPI-anchored proteins.

Fig 1.

Flow cytometry analysis of the 16 recombinants. The flow cytometry histograms reflect the mean fluorescent signal after labeling cells displaying fusion proteins with a FLAG tag. The x axis represents the fluorescence intensity, and the y axis represents the cell count. The 16 recombinants (gray background) showed increased fluorescence intensity on the cell surface compared with that for the control strain, GS115/pPIC9K (white background).

It has been reported that both the anchor proteins and heterologous proteins influence the translocation process and efficiency of the final surface display (33). The N-terminal sequence of the GPI-anchored protein is essential for its correct folding. When a GPI-anchored protein fuses to CALB in the N terminus, the GPI-anchored protein and the lipase may not fold correctly, which might result in removal of the misfolded fusion protein from the ER. Therefore, some GPI-anchored proteins could be missed using this selection procedure. To identify whether the fusion expression would have some effect on proteins so that they localized on the cell surface, we randomly chose three confirmed GPI-modified cell wall proteins (Gcw12p, Gcw21p, Gcw42p) and seven predicted GPI-anchored proteins (Gcw2p, Gcw15p, Gcw17p, Gcw36p, Gcw48p, Gcw50p, Gcw56p) which failed to show surface localization in the presence of CALB to construct novel cell surface display systems without CALB. Flow cytometry analysis results indicated that the fusion of CALB with most of the predicted GPI-anchored proteins had little impact on protein expression, except for one cell surface display system which was based on Gcw56p (see Fig. S1 in the supplemental material). We supposed that this impact might be caused by misfolding or dislocalization of the Gcw56p fusion with CALB. In our study, we aimed to screen for endogenous GPI-modified cell wall proteins of P. pastoris by a reporter protein, CALB, and have them display CALB for further investigations regarding possible applications. Therefore, the above-mentioned 16 selected proteins were subjected to further experimentation.

To further confirm the localization of the GPI-anchored proteins and examine whether these fusion proteins were displayed on the yeast cell surface by covalent or noncovalent bonds, the above-mentioned 16 selected proteins were collected by centrifugation and extracted with SDS and laminarinase. SDS treatment in combination with mercaptoethanol allowed the extraction of noncovalently bound cell wall proteins or proteins bound via disulfide bridges, while laminarinase treatment resulted in the release of covalently bound cell wall proteins. To detect whether part of the fusion proteins was retained in the cytoplasm or secreted into the growth medium, samples of intracellular extract and growth medium were also analyzed. Samples of each recombinant were subjected to SDS-PAGE, followed by Western blotting. As shown in Fig. 2, of the 16 fusion proteins, 13 were detected in the cell wall fraction extracted with laminarinase but not in the fraction extracted with SDS. However, CALB-Gcw22p, CALB-Gcw23p, and CALB-Gcw26p were detected in the cell wall fraction extracted using SDS, as well as in the intracellular samples and growth medium. These results at least revealed that the 13 fusion proteins with the GPI-anchored proteins were covalently attached to the cell wall. In addition, the 13 fusion proteins, except CALB-Gcw30p, showed no specific polypeptide fragments in the intracellular samples or growth medium. In the intracellular sample of CALB-Gcw30p, a faint band was detected, indicating that the majority of the fusion proteins were displayed on the cell wall and only a small fraction remained intracellular. Furthermore, the observed molecular sizes of all the fusion proteins were higher than the predicted protein sizes. This could probably be due to glycosylation of the proteins, which has been described previously (29). In conclusion, our results indicated that at least 13 of the 16 fusion proteins with GPI-anchored proteins localized on the cell surface were covalently attached to the cell wall. Therefore, these 13 anchor proteins were confirmed to be GPI-modified cell wall proteins.

Fig 2.

Western blot analysis of the 16 fusion proteins using laminarinase and SDS extracts. Blots of putative GPI-anchored proteins were incubated with a rabbit anti-FLAG IgG antibody and goat antirabbit antibody. The growth medium and intracellular extract supernatant were analyzed in parallel. Medium, sample of growth medium tested; Intra-cell, the supernatant containing intracellular extracts; SDS, SDS extraction of isolated cell walls; Laminarinase, laminarinase extraction of isolated cell walls.

Subsequently, the CALB hydrolytic activity displayed on the cell surface by each recombinant was measured by a spectrometric method using pNPB as the substrate. Samples from the cytoplasm and growth medium were analyzed in parallel. The results showed that the hydrolytic activity of CALB could be detected on the cell surface of the 16 recombinants (Fig. 3). The level of CALB activity varied with the anchored proteins fused. The highest hydrolytic activity of CALB displayed on the cell surface was found in CALB-Gcw61p-expressing cells, which reached 23.4 U/ml in the broth and 2,234 U/g cell dry weight on the cells. With regard to cells displaying CALB-Gcw19p and CALB-Gcw30p fusion proteins, a fraction of the enzyme activity was found to have leaked into the medium, as shown in Fig. 3. Furthermore, a part of the CALB hydrolytic activity of recombinants displaying CALB-Gcw19p, CALB-Gcw51p, and CALB-Gcw61p was found in the cytoplasm. Cells expressing CALB-Gcw26p showed low activity along with flocculation in the medium. These results indicate that CALB fused to the 13 GPI-modified cell wall proteins was efficiently immobilized onto the cell wall and that the CALB molecule was functionally displayed on the cell surface.

Fig 3.

Analysis of hydrolytic activity of CALB displayed on the cell surface by 16 recombinants in the intracellular extract and medium supernatants. Recombinant cells were induced to display CALB in BMMY medium containing 1% (vol/vol) methanol for 120 h. After centrifugation, the cells and growth medium were analyzed for CALB hydrolytic activity. Subsequently, the cells were broken with glass beads and fractionated into an intracellular extract supernatant fraction and a cell wall fraction. On-cell, the CALB hydrolytic activity of the cell suspension; Medium, the CALB hydrolytic activity of the growth medium; Intracellular extracts, the CALB hydrolytic activity of the supernatant fraction when cells were broken; Total, sum of the enzyme activities found in the cell suspension, medium, and intracellular extract fraction.

The expression level of fusion proteins was determined by active-site titration. This method was first reported by Fujii et al. (34) and was subsequently further developed by our team for quantitative analysis of lipase anchored on the cell surface (unpublished data). The amount of fusion proteins determined on the basis of the results for the 13 confirmed GPI-modified cell wall proteins was about 2 mg/g cell dry weight, and specifically, the number of fusion protein molecules was approximately 2 × 10⁵ molecules/cell. This result is consistent with the experimental data reported by Shibasaki et al. (35).

Putative function and application of confirmed GPI-modified cell wall proteins.

GPI-anchored proteins have been established to be a large class of functionally diverse proteins. Amino acid sequence comparisons (http://www.ncbi.nlm.nih.gov/BLAST/) and searches for functional domains (http://www.ebi.ac.uk/InterProScan/) indicated that the 16 selected proteins of P. pastoris may be functional enzymes, surface antigens, adhesion molecules, or ion transporters, and they have been shown to play a structural role in cell wall biogenesis. As shown in Table 1, Gcw5p and Gcw45p contain the flocculin type 3 repeat, which is a cell wall protein that participates directly in adhesive cell-cell interactions during yeast flocculation, a reversible, asexual, and Ca²⁺-dependent process in which cells adhere to form aggregates (flocs) consisting of thousands of cells (36). Gcw12p has a CFEM domain, which could function as a cell surface receptor or signal transducer or as an adhesion molecule in host-pathogen interactions (37). Gcw22p has been predicted to contain a flocculin type 3 repeat and an uncharacterized domain related to Flo11, which is found at the N terminus of the S. cerevisiae Flo11p protein. Flo11p is required for diploid pseudohyphal formation and haploid invasive growth (38). Gcw26p contains a GLEYA adhesion domain related to the lectin-like ligand-binding domain found in the S. cerevisiae Flo proteins (39), while Gcw28p comprises an adhesion domain of bacterial origin which forms part of the adhesive pilus found on the cell surface and is important for receptor binding to host cells during pathogenesis (40). Gcw34p is considered to be an aspartic protease, with a peptidase aspartic active site and peptidase A1 domain. Aspartic proteases are a family of protease enzymes that use an aspartate residue for catalysis of their peptide substrates. In general, they have two highly conserved aspartates in the active site and are optimally active under acidic pH conditions (41). Gcw23p has similarity to proteins that contain receptor L-like domains. In our study, Gcw22p, Gcw23p, and Gcw26p were observed to be expressed and anchored on the cell surface by noncovalent bonds, and Gcw26p caused cell flocculation.

In this study, 13 proteins were confirmed to be GPI-modified cell wall proteins which could be used as anchoring proteins to functionally display CALB on the cell surface as an immobilized biocatalyzer; however, the immobilization efficiency was noted to vary significantly depending on the cell wall proteins used. The length of the anchoring proteins was not a predominant factor influencing immobilization efficiency. Sato et al. developed cell surface display systems using various anchor lengths based on the C-terminal region of Flo1p and showed that anchor lengths over 428 amino acids enhanced the activity of cell surface glucoamylase on polymer substrates (15). On the other hand, Van Der Vaart et al. reported that shortening of the YCR089w anchoring domain from 744 to 313 amino acids resulted in an increase in the immobilization efficiency of α-galactosidase (42). In the present study, in the 13 novel P. pastoris cell surface display systems, the CALB hydrolytic activity of the recombinants displaying CALB-Gcw21p, CALB-Gcw51p, and CALB-Gcw61p was higher than that of the others, reaching 1,828, 1,889, and 2,235 U/g cell dry weight, respectively (Table 1). These GPI-modified cell wall proteins have been applied to display industrial enzymes for the investigation of biotransformations in our laboratory. For example, cells expressing CALB-Gcw21p were collected and combined with cells displaying Rhizomucor miehei lipase as whole-cell biocatalysts to improve biodiesel production in cosolvent media. The results showed that the reaction time and biodiesel yield of the displayed lipase-mediated transesterification reaction were approximately equal to those of commercial lipases (43).

Besides playing a role in biotransformations, GPI-modified cell wall proteins have been reported to be useful in the construction of multiplex protein chips for the analysis of a subset of high- and low-molecular-weight biomarkers and nanoparticles as novel biomaterials or in oral protein delivery (44). The present study provides further insight into these proteins, facilitating advancements in these research fields.