Abstract
Cost-effective manufacturing of biopharmaceuticals in non-mammalian hosts still requires tremendous efforts in strain development. In order to expedite identification of novel leads for strain engineering, we used a transposon-mutagenized yeast genomic DNA library to create a collection of Saccharomyces cerevisiae deletion strains expressing a full-length IgG antibody. Using a high-throughput screening, transformants with either significantly higher or lower IgG expression were selected. The integration site of the transposon in three of the selected strains was located by DNA sequencing. The inserted DNA lay within the VPS30 and TAR1 open reading frame, and upstream of the HEM13 open reading frame. The complete coding sequence of these genes was deleted in the wild-type strain background to confirm the IgG expression phenotypes. Production of recombinant antibody was increased 2-fold in the Δvps30 strain, but only mildly affected secretion levels in the Δtar1 strain. Remarkably, expression of endogenous yeast acid phosphatase was increased 1.7- and 2.4-fold in Δvps30 and Δtar1 strains. The study confirmed the power of genome-wide high-throughput screens for strain development and highlights the importance of using the target molecule during the screening process.
INTRODUCTION
The yeast Saccharomyces cerevisiae has been widely used as the expression host for the production of heterologous proteins mostly for industrial applications but also for production of biopharmaceuticals. Although the yeast has proven its value for large-scale industrial production processes, for many commercially interesting proteins such as biopharmaceuticals the production titers are too low to compete with existing mammalian cell-based production platforms. Most notably, this gap can be seen in the production of recombinant full-length antibodies.
Reduced production rates can be the consequence of shortcomings in various processes including protein folding, N-glycosylation, trafficking steps such as ER to Golgi transport, Golgi to plasma membrane transport, vacuolar sorting, or secretory vesicle fusion to the plasma membrane and missorting, or proteolytic degradation (Idiris et al.2010). Up to date, only few systematic global surveys have been pursued to identify and overcome the rate limiting steps in protein production in fungal hosts (Graf et al.2009).
These shortcomings can be tackled from two sides, by modifying either the target protein through protein engineering or by engineering of the expression host (Hou et al.2012; Kazemi Seresht et al.2013; Wu et al.2013; Xu et al.2014; de Ruijter, Koskela and Frey 2016). Whereas the first approach is applicable for industrial enzymes, protein engineering is no option for the production of therapeutic proteins as the protein sequence defines their structure and specificity. However, the second strategy, to tailor the expression host, can be applied more widely. To optimize the yeast as a heterologous protein production cell factory, one of the current focal points in research is the adaptation of the secretory pathway for high-level protein expression, either through overexpression of genes promoting secretion efficiency or through deletion of genes with an inhibitory role. In both cases, rational or non-targeted approaches can be used.
In the rationally designed approach, either prior knowledge of the bottlenecks and the genes causing them is required or a sound hypothesis of the possible limiting steps has to be available. Recent examples include the overexpression of the yeast peptidyl-prolyl-isomerase CPR5 in the ER to aid folding (de Ruijter, Koskela and Frey 2016); the overexpression of SSO1, SLY1 and SEC1 genes involved in vesicular transport steps (Hou et al.2012; Xu et al.2014); and the deletion of selected proteases to diminish proteolytic degradation or deletion of vacuolar sorting receptors like VPS10 to prevent missorting of proteins to the vacuole for degradation (Wu et al.2013; Xu et al.2014). And even though the targeted approach has been successfully used to improve the secretion of various heterologous proteins, like cellobiohydrolase (Van Zyl, Den Haan and Van Zyl 2014; Xu et al.2014), amylases (Liu et al.2014; Rodríguez-Limas, Tannenbaum and Tyo 2015), and single chain and full length antibodies (Xu et al.2005; Rakestraw et al.2009), its applications are limited.
In contrast, in the non-targeted approach, libraries of yeast strains are screened for clones that show an improved production of the target protein, after which the responsible genetic element is analyzed. Examples include the screening of cDNA overexpression libraries (Wentz and Shusta 2007), deletion strain libraries and random mutagenized strains (Davydenko et al.2004; Liu et al.2014) for improved production of the target protein. The value of these approaches lies in the opportunity to find new, unexpected targets for cellular engineering. This can include uncharacterized genes, genes with a function not directly linked to the secretory pathway, and other genes providing an accessory function. It's noteworthy to keep in mind that only around 80% of all 6604 ORFs of Saccharomyces cerevisiae are characterized, and still close to 20% are without any known function or a dubious (Peña-Castillo and Hughes 2007).
Often, the positive results obtained in the rationale and non-targeted approaches are highly specific for the used target protein, and so far it proved difficult to find genes that stimulate heterologous protein secretion in a general way. Despite this strong connection between successful genetic modification of the strain and the protein to be overexpressed, a large fraction of the available screening research is done using an artificial product as model protein, instead of using the protein of interest. Most convenient for screening setups are readily measurable proteins such as fluorescent proteins like GFP or luciferase (Kanjou et al.2007), or conveniently measurable enzymes like β-galactosidase or acid phosphatase (AP) (Wingfield and Dickinson 1992). Additionally, fusion proteins have been studied in order to ease the measurement of the expressed protein of interest, for example, in the use of yeast surface display systems (Sun et al.2014). Although in some cases there appears to be a good correlation between surface display of fusion proteins and the secretion levels of the native proteins, the structure of the protein that targets the fusion construct to the membrane can be of influence. It can for example influence the folding or activity of the target protein, so that the modifications of a strain with a high surface display are not necessarily beneficial for the native structure of the target protein. Overall, these methods need a second round of confirmation to verify the suitability in a real production strain.
The ease of genetic manipulation of microbial host systems, and especially of S. cerevisiae, permits to straightforwardly screen genome-wide libraries for proteins affecting secretory capacity. In this study, we present the screening of a randomly generated deletion strain collection created using a transposon-inactivated yeast genomic DNA library for the identification of strains with altered full-length human IgG antibody production. Interesting clones were contrasted for IgG and AP secretion, after which the genetic disruption caused by the transposon is determined. Finally, targeted deletion strains are made of two of these clones to confirm the expression phenotypes.
MATERIALS AND METHODS
All used FastDigest — restriction enzymes were obtained from Thermo Fisher Scientific (Waltham, MA, USA). All media components and reagents were obtained from Sigma-Aldrich (St. Louis, MO, USA) unless stated otherwise. Yeast nitrogen base without amino acids was obtained from BD (Vantaa, Finland).
Strains and plasmids
Saccharomyces cerevisiae strain W303α (MATα leu2-3,112 trp1-1 can1-100 ura3-1 ade2-1 his3-11,15) was used for all experiments. The transposon mutagenized yeast genomic DNA library has been kindly provided by Kumar et al. (2004). The human IgG expressing plasmids pAX550 and pEK5 and yeast AP expressing plasmids pAX466 and pAX469 have been described elsewhere (Parsaie Nasab et al.2013; de Ruijter and Frey 2015; Frey and Aebi 2015). All plasmids are derived from either pRS415 or pRS416 plasmid series and contain CEN/ARS sequences and either LEU2 or URA3 selection markers (Mumberg, Müller and Funk 1995). Expression of IgG and AP are under control of the galactose inducible GAL1 promoter.
Directed gene deletion strains were constructed using the method from Hegemann and Heick (2011). The target gene loci were deleted by replacement with a PCR product containing the KanMX4 cassette and flanking target gene coding sequences. Plasmid pRS305K (Taxis and Knop 2006) was used as template for amplification of kanamycin resistance (KanMX4) marker cassettes. After transformation, cells were grown on medium supplemented with 200 μg/mL G418. Deletions were confirmed using colony PCR. A genetic cassette containing a GFP gene under control of the UPR control elements (UPRE-GFP) was introduced into TRP1 locus of W303α and VPS30 deletion strain after linearization of plasmid pDEP107 (Pincus et al.2010). The lithium acetate method was used for all transformation of yeast cells (Gietz and Schiestl 2007).
Library amplification and yeast transformation
The Tn7 mutagenized yeast insertional library was propagated in Escherichia coli TOP10 as described previously (Kumar et al.2004). A portion of the 100 ng of DNA pool was used to transform E. coli using electroporation. Transformants were selected on solid LB medium (5 g/l Bacto yeast extract, 10 g/l Bacto tryptone and 5 g/l NaCl) supplemented with 1.5 g/l agarose, tetracycline (3 μg/ml) and kanamycin (40 μg/ml) on plates 14 cm in diameter. Approximately 10 000 transformants were obtained per pool following overnight growth at 37°C.
Transformants were washed from selective agar plates using 6 ml LB medium preparing a homogenous suspension. An aliquot of this suspension was diluted into 45 ml LB medium supplemented with tetracycline (3 μg/ml) and kanamycin (40 μg/ml) to yield a culture of nearly saturated cell density. The culture was incubated at 37°C with aeration for 2–3 h. Plasmid DNA was isolated using Qiagen Midi plasmid isolation kit (Qiagen, Germany).
A portion of 1 μg of the purified DNA library was digested with NotI and the released transposon mutagenized yeast genomic DNA was isolated after separation of DNA on 1% agarose gel using the NucleoSpin® Gel and PCR Clean up kit (Macherey-Nagel, Switzerland).
Competent yeast W303α cells carrying pAX550 were transformed with 500 ng of linearized DNA using the lithium acetate method. Transformants were plated on square plates with agar-solidified synthetic dropout medium lacking uracil and leucine (SD-Ura-Leu) and supplemented with 2% glucose and incubated at 30°C for 48 h. The amount of cells spread per plates was adjusted in order to obtain approximately 500 colonies per plate.
Primary screening
Colonies were picked from rectangular SD-Ura-Leu plates using a HAMILTON Star line liquid handling station. Ninety-six-well deep well plates (VWR, Finland) were filled, for inoculation with single colonies, with 1 ml of SD-Ura-Leu and supplemented with 2% raffinose, 20 mM sodium phosphate buffer, pH 6.5 and 50 μg/ml BSA. Plates were incubated at 30°C and 250 rpm for 24 h. A total of 50 μl of saturated precultures were transferred into 950 μl fresh media. Two rounds of preculturing were performed to even out the cell amounts and growth phase of the cells. After the preculturing procedure, individual cultures were spotted onto solid SD-Ura-Leu square plate and plates were grown for 72 h at 30°C. These plates served as master clone banks. Expression cultures were inoculated with 50 μl of the precultures, grown for 5.5 h at 30°C, 250 rpm and IgG expression was induced by the addition of galactose to a final concentration of 0.5%. For expression, the plates were incubated at 20°C for 24 h.
Samples for IgG determination were collected after 24 h of growth. Cultures were mixed with 50 μl of 20× PBT solution (1 × PBT (PBS (135 mM NaCl, 2.5 mM KCl, 10 mM Na2HPO4, 1.75 mM KH2PO4) containing 0.5% Tween-20)), culture supernatants were cleared by centrifugation of plates at 5000 g for 10 min, 4°C and 3 aliquots of 300 μl cleared culture supernatant were transferred to fresh 96-well plates and stored at –20°C until analysis.
Secondary screening
Strains A, B and C were selected for the secondary screening. In addition, four random colonies were picked from the master plates and used as controls. Lead and control strains were restreaked several times on non-selective media to remove plasmids. Plasmid loss was verified by absence of growth on SD-Leu plates.
Growth of strains was analyzed using Bioscreen analyzer. Precultures of strains were grown in 3 ml of YPD medium at 30°C, 250 rpm. Precultures were diluted to A600 = 0.2 and allowed to grow to A600 = 1. These exponentially growing cultures were diluted to A600 = 0.1. A total of 200 μl of the diluted cultures were transferred to the multi-well plates. Plates were grown for 24 h at 30°C with constant shaking, and A600 was monitored every 30 min. Eight replicate cultures were grown per strain.
For confirmation of screening results, the empty strains were transformed with plasmids pAX466 and pAX550 as described above and transformants were selected on SD-Leu plates. Precultures were grown in 4 ml media at 30°C, 250 rpm for 16 h. Cells were collected and cell density was adjusted to A600 = 5. Deep-well plates containing 0.9 ml of media were inoculated with 0.1 ml of preculture reaching at starting A600 of 0.5. Deep-well plates were grown for 3 h at 30°C and 250 rpm. Temperature was adjusted to 20°C and 30°C, and expression of proteins was induced by the addition of galactose to final concentrations of 0.5% and 2.0%, respectively. Twenty-four hours after induction, samples for IgG determination were cleared by centrifugation and the supernatant was adjusted to 1 × PBT and stored at –20°C until analysis. Samples for AP activity measurements were immediately processed and analyzed.
The strain harboring the transposon in the regulatory region of HEM13 gene and transformed with plasmid pAX550 was grown in the presence of hemin. A 4 mM hemin stock solution was prepared in 100 mM NaOH. Serial dilutions of media containing 50–0.11 mg/l of hemin were prepared. Three replicate cultures per hemin concentrations were prepared and cells were grown as described above. IgG titers were determined after 24 h of growth.
Identification of integration site using vectorette PCR
Genomic DNA was isolated from clones using Qiagen Genomic DNA isolation kit. A portion of 1–2 μg of yeast genomic DNA was digested with FastDigest AluI in a total volume of 20 μl according to the manufacturers’ instructions. Primers ABP2 (5΄-GACTCTCCCTTCTCGAATCGTAACCGTTCGTACGAGAATCGC-TGTCCTCTCCTTC-3΄) and ABP3 (5΄-AAGGAGAGGACGCTGTCT-GTCGAAGGTAAGGAACGGACGAGAGAAGGGAGAG-3΄) were annealed to each other to form the adaptor anchors by mixing 1 pmole of each primer in 200 μl of annealing buffer containing 10 mM Tris, 10 mM MgCl2 and 50 mM NaCl. The primer mixture was heated for 5 min at 95°C and allowed to slowly cool to 37°C. A portion of 1 μl of the annealed primers were ligated to 20 μl of restricted DNA. The ligation reaction was allowed to proceed for 2 h at 22°C. The DNA region between the inserted transposon and annaeled primers was amplified using primer pair UV (5΄-CGAATCGTAACCGTTCGTACGAGAATCGCT-3΄) and M13-47 (5΄-CGCCAGGGTTTTCCCAGTCACGAC-3΄) using 1 μl from the ligation mixture as template. PCR program consisted of one cycle of 2 min at 92°C, followed by 35 cycles of 20 s at 92°C, 30 s at 67°C and 90 s at 72°C with a final extension of 150 s at 72°C. Each PCR product was gel purified using standard protocols into a final volume of 30 μl water. The purified product (50 ng) was used for one sequencing reaction from SEQCORR primer (5΄-CGACGGGATCCAACGCGTAT-3΄).
Characterization of TAR1 and VPS30 deletion strains
The Δtar1, Δvps30 and the wild-type strain were transformed with plasmids pEK5 and pAX469. For expression in 96 deep-well plates, six to eight different transformants were inoculated in triplicates into 1 ml of SD-Ura, supplemented with 2% raffinose, 20 mM sodium phosphate buffer, pH 6.5 and 50 μg/ml BSA, and grown for 21 h at 30°C, 250 rpm. The culture was diluted 1:4 into fresh media in a new 96 deep-well plate and grown at 30°C, 250 rpm for 5.5 h before protein expression was induced by the addition of 0.5% galactose. For the shake flask expression, six clones of the preculture plates of each strain were used to inoculate 10 ml liquid cultures in 100 ml shake flasks at a starting A600 = 0.2 and were grown at 30°C, 250 rpm for 5.5 h before protein expression was induced by adding 0.5% galactose. For all experiments, protein expression was continued for 24 h after the addition of galactose, followed by the analysis of AP activity or IgG titers as described above.
Analytical procedures
The enzyme-linked immunosorbent assay (ELISA) for determination of IgG titers was run on a HAMILTON Star line liquid handling station as described before (de Ruijter and Frey 2015). Capture (anti-human IgG, Fc specific, I2136) and detection (anti-human IgG, Fc specific, HRP conjugate, A0170) antibodies were obtained from Sigma-Aldrich, the human IgG1 standard (#400120) was obtained from Calbiochem (Espoo, Finland).
AP activity was measured from cleared culture supernatant using an endpoint method as described before (Frey and Aebi 2015). Absorbance was read at 405 nm using a BioTek Synergy 2 spectrophotometer. Data evaluation was done with Gen5 software (BioTek).
Reporter strains were cotransformed with IgG expression plasmids (pAX550 and pEK5) and an empty plasmid with the complementary marker (pRS415 and pRS416) enabling growth in the identical medium. For UPR induction measurements, strains harboring the UPRE-GFP construct and the IgG expression plasmids were grown overnight and diluted to an A600 = 0.1 in fresh media. Cell suspension (100 μl) was transferred to a well of a transparent round-bottom 96 well plate. A600 and GFP fluorescence levels were measured using a Cytation 3 Mircoplate spectrophotometer. DTT was added to a final concentration of 5 mM. Induction of IgG expression was achieved by the addition of 5 μl of 10% galactose solution to each well.
RESULTS AND DISCUSSION
Primary screening
In order to assess the influence of technical effects on the variability of the screening procedure, we conducted experiments using the screening workflow, but, instead of using different transformants, 72 replicates of the wild-type strain transformed with the IgG expression plasmid were inoculated into deep-well plates. The average IgG titer in the replicate culture was 34.8 ± 7.50 μg/l after 24 h of expression and most of the standard deviation was due to two clones deviating more than average from the mean. The calculated z-score over the plate was 3.17 × 10−5. Therefore, we considered that the variations due to sample handling and slightly differing growth conditions were small enough for the screen.
For the initial screen, we picked around 900 uracil auxotrophic colonies, which we obtained after transformation of the wild-type strain harboring the IgG expression plasmid with the transposon mutagenized yeast genomic DNA library. After colony picking, two rounds of preculturing were performed in order to synchronize growth phase and cell densities.
After 24 h of expression, the IgG titers were determined from the cleared culture supernatants using an ELISA. In order to compare the results of the strains grown on different plates, z-scores were calculated from the IgG titers (Fig. 1A). We defined a z-score of 2.0 as the criteria for inclusion of hits into confirmatory tests. Both transformants with an increased and with a decreased IgG expression were included in the confirmation tests. In addition, randomly selected transformants were included. The transformants were grown as above and IgG titers were determined (Fig. 1B). After the confirmation of the primary screening results, which included 80 transformants, 3 transformants were selected, designated as strains A, B and C, and these strains were used for further experiments. Transformants A and B displayed improved production of IgG reaching titers of 101 and 122 μg/l, respectively, and, in contrast, expression of IgG in strain C was very low.
Overview of effects of deletions on antibody production. IgG titers of yeast strains grown for 24 h under inducing conditions at 20°C were determined using an ELISA. In order to compensate for interplate variations, IgG titers were converted into z-score. The data showed a normal distribution. Hits were selected based on the criteria of z-score ≥ ±2.0. The samples marked A, B and C indicate the strains finally selected for further characterization (A). IgG titers obtained of strains selected for further testing, the strains included strains which had a z-score ≥ ±2.0 and randomly picked control strains. Strains A and B reached highest IgG titers after 24 h of cultivation (B). Strain C was selected due to its low but not zero expression level.
Initial characterization of strains
For the initial characterization, the previously selected transformants A, B and C, and four randomly picked strains without a clear IgG expression phenotype in the primary screen, were selected as controls. After ensuring loss of the original expression plasmids, the strains were assessed for growth phenotypes through the recording of growth for 24 h (Fig. 2). As can be seen, only transformant A showed a significant lower growth rate.
Growth of candidate and control yeast strain. Strains selected from primary screen and four random strains were depleted of IgG expression plasmid. Strains were grown for 24 h in Bioscreen analyzer and A600 was recorded every 30 min. These data show the average and standard deviations of eight independent replicate cultures.
Next, all seven strains were transformed with either the IgG expression plasmid or with the identical plasmid but encoding the yeast endogenous AP. The transformants were tested in the similar growth format as in the primary screening, and expression of the proteins was studied with two different concentrations of the inducer galactose at 20°C and 30°C. Samples were taken 24 h after induction, and IgG titers and AP activity were determined (Figs 3 and 4, respectively). The absence of IgG production of transformant C was confirmed under all tested conditions. However, although strongly reduced, AP production in transformant C was still substantial, indicating that the expression of the endogenous protein was less affected in this strain. The original phenotype of transformants A and B was most strongly conserved when the IgG expression was induced with 0.5 % galactose and cells were grown at 20°C (Fig. 3A), i.e. the experimental conditions used in the initial identification. Generally, their positive effects were lower at 30°C. Overall, the secondary screening could confirm the phenotypes of transformants A, B and C.
Secretion of IgG of candidate strains and random clones. Strains were grown in deep-well plates at 20°C (A and B) and 30°C (C and D), and expression of IgG was induced with either 0.5% (A and C) or 2.0% (B and D) galactose. IgG titers were determined after 24 h using an ELISA. Data represent the mean and standard deviations of two (random strain) to four (strains A to C) experiments with each four replicate cultures.
Production of AP of candidate strains and random clones. Strains were grown in deep-well plates at 20°C (A and B) and 30°C (C and D) and expression of AP was induced with either 0.5% (A and C) or 2.0% (B and D) galactose. AP activity in culture supernatants was determined after 24 h using an enzyme assay. Data represent the mean and standard deviations of two (random strain) to four (strains A to C) experiments with each four replicate cultures.
Identification of disrupted genes
Sequencing of the produced fragments from the selected clones through vectorette PCR revealed the integration sites as shown in Fig. 5. The site of integration in transformant A was located in the ORF of the VPS30 gene, in transformant B it was located in the ORF of the TAR1 gene and in strain C it was located just upstream of the HEM13 ORF. A summary of the functions of the disrupted genes is shown in Table 1. Whereas the biological functions of Vps30p and Hem13p are well established, the role of Tar1p is more elusive. The TAR1 gene is located on the antisense strand of the multicopy 25S rRNA gene and the protein product localizes to the mitochondria. The TAR1 gene is conserved among hemiascomycetous species (Coelho et al.2002; Galopier and Hermann-Le Denmat 2011).
Characterization of the sites in the genomic DNA where the transposons were inserted into. Vectorette PCR was used on digested yeast genomic DNA to amplify the region around the inserted transposon. The exact site was determined using BLAST from the sequencing data; the surrounding genetic loci are shown.
Genes with an effect on antibody production.
| Strain | Disrupted genea | Functiona |
|---|---|---|
| Clone A | VPS30/YPL120W | Subunit of the phosphatidylinositol 3 kinase complexes I and II, required for overflow degradation of misfolded proteins when ERAD is saturated |
| Clone B | TAR1/YLR154W-C | Protein potentially involveld in regulation of respiratory metabolism, located in the mitochondria |
| Clone C | HEM13/YDR044W | Coproporphyrinogen III oxidases, oxygen requiring enzyme that catalyzes the sixth step in the heme biosynthetic pathway |
| Strain | Disrupted genea | Functiona |
|---|---|---|
| Clone A | VPS30/YPL120W | Subunit of the phosphatidylinositol 3 kinase complexes I and II, required for overflow degradation of misfolded proteins when ERAD is saturated |
| Clone B | TAR1/YLR154W-C | Protein potentially involveld in regulation of respiratory metabolism, located in the mitochondria |
| Clone C | HEM13/YDR044W | Coproporphyrinogen III oxidases, oxygen requiring enzyme that catalyzes the sixth step in the heme biosynthetic pathway |
Annotation from the Saccharomyces Genome Database.
Genes with an effect on antibody production.
| Strain | Disrupted genea | Functiona |
|---|---|---|
| Clone A | VPS30/YPL120W | Subunit of the phosphatidylinositol 3 kinase complexes I and II, required for overflow degradation of misfolded proteins when ERAD is saturated |
| Clone B | TAR1/YLR154W-C | Protein potentially involveld in regulation of respiratory metabolism, located in the mitochondria |
| Clone C | HEM13/YDR044W | Coproporphyrinogen III oxidases, oxygen requiring enzyme that catalyzes the sixth step in the heme biosynthetic pathway |
| Strain | Disrupted genea | Functiona |
|---|---|---|
| Clone A | VPS30/YPL120W | Subunit of the phosphatidylinositol 3 kinase complexes I and II, required for overflow degradation of misfolded proteins when ERAD is saturated |
| Clone B | TAR1/YLR154W-C | Protein potentially involveld in regulation of respiratory metabolism, located in the mitochondria |
| Clone C | HEM13/YDR044W | Coproporphyrinogen III oxidases, oxygen requiring enzyme that catalyzes the sixth step in the heme biosynthetic pathway |
Annotation from the Saccharomyces Genome Database.
Verification of phenotypes using defined deletions of target genes
For verification of the phenotypes of the identified transformants A, B and C, yeast strains with the complete deletion of the open reading frames of TAR1, VPS30 and HEM13 genes were prepared. However, transformations of four commonly used laboratory strains (W303, SS328, CEN.PK113-7D and BJ3500) with the deletion cassette for the HEM13 gene did not result in viable colonies. It has been shown previously that Δhem13 strains display very slow growth at all temperatures or are not viable at all (Giaever et al.2002; Ben-Aroya et al.2008). As mutations and partial deletion have shown to be viable and show less severe growth defects, this indicates that the integration of the transposon upstream of the coding region of the gene led only to a partial inactivation of the HEM13 gene (Zagorec et al.1988).
Therefore, we examined whether the putatively reduced heme biosynthesis in strain C could be restored by supplementation of exogenous heme to the culture medium, which in turn could rescue IgG expression in this strain. Surprisingly, when complementing the media of the initial strain C with up to 50 mg/l of hemin, the observed IgG expression defect was not restored and secreted IgG titers were still very low (data not shown). However, a 60% increase in final cell density was observed. Normally, heme biosynthesis is under tight control of the transcription factor Rox1p, which prevents cellular heme accumulation via repression of HEM13. Rox1p itself is controlled by Hap1p activity, which in turn is activated by high heme levels (Martínez et al.2016). Interestingly, overexpression of HAP1 gene has recently been shown to have a profound effect on growth and to some extent protein expression (Martínez et al.2015) putatively by induction of oxidative stress response and expression of enzymes involved in respiration. Thus, the observed growth effect might be related to enhanced activation of Hap1p resulting from heme supplementation, but the connection between HEM13 and IgG expression remains obscure.
The Δvps30 and Δtar1 yeast strains were compared with the wild-type strain for their secretion capacity of AP and IgG in 96-deep-well plate and shake flask format (Fig. 6). As can be seen in the figure, the data follow the same trends for both plate and shake flask cultivations. For the IgG production (Fig. 6A), there is a clear increase in secretion in the Δvps30 strain, with an on average 100% increase in final IgG titers. In contrast, the Δtar1 strain only produced slightly more in shake flasks, but not in the plate cultivations. For the AP production (Fig. 6B), both deletion strains showed more enzyme activity in the supernatant than was found in supernatants of the control strain. Remarkably, in the expression of the endogenous AP the Δtar1 strain even produced significantly more than the Δvps30 strain, reaching final yields 2.4- and 1.7-fold higher than the wild-type strain in shake flask and deep-well plate format, respectively. This difference again highlights the importance of contrasting the secretion levels of different types of proteins to get a more general idea of the effects of the mutation.
Normalized secretion titers of IgG (A) and secreted enzyme activity of AP (B) from deep-well plate and shake flask cultivations (left and right graphs, respectively). Protein expression was induced for 24 h. Data represent the mean and standard deviations from five to eight biological replicates. The data are normalized with the corresponding OD600 values.
From these two deletion strains, part of the phenotypes can be explained by the nature of the deleted genes. The gene VPS30 encodes a protein of the vacuolar protein sorting receptor family. The gene has been shown to be involved in the sorting of proteins from the Golgi apparatus to the vacuole, as mutations in or deletions of the gene led to missorting and secretion of the vacuolar protein carboxypeptidase Y (CPY) (Seaman et al.1997). Besides the effect of deleting the VPS30 gene on protein sorting, its deletion was also reported to lead to a low constitutive activation of the cellular unfolded protein response in unstressed conditions. However, its deletion did not further accentuate the UPR levels in tunicamycin stressed cells compared to wild-type controls (Kruse, Brodsky and McCracken 2006). This preactivation of the cellular UPR might help to prime cells for the following unfolded protein stress imposed by the overexpression of the antibody. Therefore, we created a wild-type and a Δvps30 strain where a genetic construct containing four repeats of the UPR transcriptional control elements fused to a GPF gene was integrated as a single copy into the genome of the strains (Pincus et al.2010). The measured basal fluorescence levels in both strains were similar, indicating comparable levels of UPR activation under unstressed conditions of wild-type and Δvps30 strains. In order to verify functionality of the reporter, we tested its response to induction of the UPR by the addition of 5 mM DTT 3 h after initiation of the cultivation. This resulted in GFP expression in both strains; however, the GFP signal in the Δvps30 background reached lower levels than in the wild-type background (Fig. 7A). Next, two different IgG expression vectors were introduced into the two UPR-GFP reporter strains, and the fluorescence signals were recorded for 40 h. The GFP signal increased strongly in the wild-type strain, but only after a lag phase it also started to increase in the Δvps30 background and reached again lower levels (Fig. 7B). However, parallel cultivations of strains reproduced the previously observed expression phenotypes (Fig. 7C). In contrast to what was reported before, we could not detect any preactivation of the UPR in our assay. This could be due to genetic differences in the strain backgroud or to the nature of the used reporter construct, using GFP as opposed to the previously used beta-galactosidase, and using a single integrated copy versus the previously used high-copy number plasmid for expression of the reporter. Overall, these data indicate that the improved IgG secretion observed in the Δvps30 strain is solely due to its role in protein trafficking and not due to its previously reported effect of pre-activating the UPR.
UPR induction in the wild-type and Δvps30 strain harboring a single copy of a UPRE-GFP construct in the genome. Fluorescence of wild-type and Δvps30 strain is similar under non-stressed conditions (-DTT) but lower in the Δvps30 strain relative to wild-type cells when exposed to 5 mM DTT (+DTT) (A). Fluorescence levels determined during IgG expressing conditions in Δvps30 and wild-type cells. IgG expression was induced 3 h after seeding the cultures and cultivation was continued for 37 h (B). Normalized IgG titers determined after 24 h of protein expression (C). The data are normalized with the corresponding OD600 values. Reporter strains were cotransformed with IgG expression plasmids (pAX550 or pEK5) and an empty plasmid with the complementary marker (pRS415 or pRS416) enabling growth in the identical medium.
Additionally, it has been shown that the ERAD also target A1PiZ (the Z variant of the human α-1 proteinase inhibitor), which upon strong overexpression was evading ERAD and leaving the ER, and was escaping vacuolar degradation when the VPS30 gene is inactivated (Kruse, Brodsky and McCracken 2006). In this case, it had been suggested that the vacuolar degradation serves as a rescue system when the ERAD was overloaded. In murine hybridoma and B cells, IgG molecules are a target for ERAD and also in Saccharomyces cerevisiae the IgG secretion levels are influenced by ERAD activity (Lee et al.2012; de Ruijter and Frey 2015). Although more AP was also secreted in the Δvps30 strain, the IgG molecules seemed to be more sensitive to missorting and degradation in cells that do have a normal vacuolar sorting. The growth defect observed in the initial charaterization of transformant A with the inactivated VPS30 gene (Fig. 2) might be explained by the observation that deletion of several autophagy-related genes among which VPS30 that is also annotated as ATG6 negatively affects mitochondria maintenance and thus can affect respiratory growth (Zhang et al.2007).
Contrastingly, the underlying mechanism that improves protein secretion in the case of the TAR1 deletion is unclear. The gene is located on the antisense strand of the multicopy 25S rRNA gene and the protein product localizes to the mitochondria (Coelho et al.2002; Galopier and Hermann-Le Denmat 2011). To date, little is known about the exact function of the protein, other than that it is involved in maintaining the cells oxidative phosphorylation capacity under respiratory conditions (Bonawitz et al.2008). To the best of our knowledge, the function of the TAR1 gene has so far not been connected with protein secretion.
Overall, the expression phenotypes of the Δvps30 and Δtar1 deletion strains were more prominent in the final confirmatory experiments than during the screen. The reason for this could be that in the former experiments the deletion strains have been compared with their true wild-type strain background (W303α). In contrast, the four strains selected from the primary screening for use as control strains in the secondary screening contained themselves as well the transposon insertions. And even though they exhibited no clear IgG expression phenotype, it is still possible that they were slightly more efficient secretors than the wild-type strain background and thereby influenced the apparent expression phenotype of the strains selected for characterization.
For the characterization of the Δvps30 and Δtar1 deletion strains, we used deep-well plates and shake flasks for cultivation. As the cultivation conditions between deep-well plate and shake flasks are different, it is important to consider if the deletion phenotypes are preserved in the different growth formats. In general, the relative increase of both deletion strains compared to the wild type is comparable in both expression formats, indicating that the deletions are also relevant in a slightly larger production scale. Overall, the specific IgG titer and specific AP activity were significantly higher for all the strains in the shake flask than in the deep-well plate cultivations.
In this study, we have established a high-throughput method for screening a randomly generated yeast deletion strain library. Our results confirmed that by using a screening set-up and the protein of interest as the primary read-out of the screen, random deletion libraries can be screened and the mutations can be efficiently identified using DNA sequencing. This indicates that the approach could be successfully extended to a larger scale and used identify novel secretion-related genes. Overall, our data demonstrate that deletion of VPS30 improves the secretion of IgG, while deletion of TAR1 improves the secretion of the IgG and AP.
AUTHOR CONTRIBUTIONS
Jorg de Ruijter created the high-throughput colony-picking method, created the clean deletion strains and verified their expression phenotype, and helped drafting the manuscript. German Jurgens created the yeast deletion library, carried out the primary screening experiments and verified the transposon insertion sites. Alexander Frey conceived the study; carried out the secondary screening experiments, the experiments with UPR reporter and the heme complementation study; and drafted the manuscript.
Conflict of interest.None declared.
