Engineering Saccharomyces cerevisiae for the production and secretion of Affibody molecules

Background Affibody molecules are synthetic peptides with a variety of therapeutic and diagnostic applications. To date, Affibody molecules have mainly been produced by the bacterial production host Escherichia coli. There is an interest in exploring alternative production hosts to identify potential improvements in terms of yield, ease of production and purification advantages. In this study, we evaluated the feasibility of Saccharomyces cerevisiae as a production chassis for this group of proteins. Results We examined the production of three different Affibody molecules in S. cerevisiae and found that these Affibody molecules were partially degraded. An albumin-binding domain, which may be attached to the Affibody molecules to increase their half-life, was identified to be a substrate for several S. cerevisiae proteases. We tested the removal of three vacuolar proteases, proteinase A, proteinase B and carboxypeptidase Y. Removal of one of these, proteinase A, resulted in intact secretion of one of the targeted Affibody molecules. Removal of either or both of the two additional proteases, carboxypeptidase Y and proteinase B, resulted in intact secretion of the two remaining Affibody molecules. The produced Affibody molecules were verified to bind their target, human HER3, as potently as the corresponding molecules produced in E. coli in an in vitro surface-plasmon resonance binding assay. Finally, we performed a fed-batch fermentation with one of the engineered protease-deficient S. cerevisiae strains and achieved a protein titer of 530 mg Affibody molecule/L. Conclusion This study shows that engineered S. cerevisiae has a great potential as a production host for recombinant Affibody molecules, reaching a high titer, and for proteins where endotoxin removal could be challenging, the use of S. cerevisiae obviates the need for endotoxin removal from protein produced in E. coli. Supplementary Information The online version contains supplementary material available at 10.1186/s12934-022-01761-0.

is HER3. Elevated HER3 expression is associated with malignant cancer in ovarian, prostate, gastric, bladder, lung, melanoma and colorectal tissue [6]. Current applications for Affibody molecules are mostly within diagnostics and therapeutics where the main advantage of Affibody molecules compared to alternatives, like antibodies, is their small size which allows them to penetrate tissue more easily [7,8].
Affibody molecules can be designed in a range of configurations, either containing solely Z-domains or Z-domains together with other peptide sequences. Affibody molecules composed of only the Z domain exhibit a relatively short half-life in the human body. Due to their small size, 6.7 kDa for a single Z domain, Affibody molecules are taken up by the kidneys and degraded [9]. To prolong the half-life of the Affibody molecules an albumin binding domain (ABD) can be added. The ABD originates from the GA3 module of streptococcal protein G and binds to serum albumin. Serum albumin, with a size of 67 kDa, is above the glomerular filtration barrier of the kidneys and adding ABD to Affibody molecules thus increases both their half-life and efficacy [5,10]. Different configurations of the Z domain and the ABD domain, like the presence of multiple Z domains or an alternative localization of the ABD within the molecule, have shown to be effective in a range applications [7,10,11].
Currently, Affibody molecules are produced in the bacterial production host Escherichia coli. Bacteria are popular hosts for recombinant production due to rapid growth and high yields of the recombinant proteins [12]. However, recombinant protein production in E. coli leads to contamination with lipopolysaccharides from the bacterial cell wall, most of which exhibit endotoxin properties and can be difficult to remove during product purification [13,14]. It is therefore of interest to explore other established and competitive production hosts like yeasts that do not have toxic components in their cell wall.
Yeasts are eukaryotic unicellular organisms, which secrete recombinant proteins and have been implemented as industrial production hosts for several pharmaceutical proteins [15,16]. The most widely used yeast species for the industrial production of recombinant pharmaceutical proteins are Saccharomyces cerevisiae and Komagataella phaffi [16]. K. phaffi has been used as production host of a fusion construct of human serum albumin and a HER2 Affibody molecule; however, S. cerevisiae has not been tested as an Affibody molecule production host yet [17]. S. cerevisiae on the other hand is particularly known for its industrial production of insulin, human serum albumin, and hepatitis vaccines [18].
In this study, we tested an engineered S. cerevisiae strain for the production and secretion of three Affibody molecules that bind to the cancer marker protein HER3 [6]. Initially, we found that S. cerevisiae degraded the produced Affibody molecules. We were able to identify several proteases responsible for this degradation, and upon removal of these, all three Affibody molecules were secreted in an intact state. We verified high-affinity HER3 binding by one of the secreted Affibody molecules and performed a fed-batch cultivation where a high final titer was reached thus demonstrating that S. cerevisiae is a competitive host for Affibody molecule production.

The albumin-binding domain is degraded by S. cerevisiae
We first tested whether Affibody molecules could be produced and secreted by S. cerevisiae. We included three different Affibody molecules in this study. These three molecules have different configurations of either one or two Z HER3_1 domains, the Z domain variant that binds to HER3, and one albumin binding domain (ABD) (Fig. 1). S. cerevisiae strain B184k, previously evolved for high levels of protein secretion, was used as the host [19,20]. Genes encoding the three different Affibody molecules shown in Fig. 1 were cloned into the backbone of a CPOT plasmid, thus generating the three plasmids pNatZACPOT (Z HER3_1 -ABD), pNatZ-ZACPOT (Z HER3_1 -Z HER3_1 -ABD), and pNatZAZCPOT (Z HER3_1 -ABD-Z HER3_1 ) [21]. CPOT is a recombinant protein expression plasmid for use in S. cerevisiae tpi1Δ strains. The CPOT plasmid contains the POT1 gene encoding triose phosphate isomerase from Schizosaccharomyces pombe, which partially complements the removal of the TPI1 gene and restores the ability to grow on glucose as a carbon source. The partial complementation will lead to a high abundancy of the plasmid in the cell, which is combined with the expression regulation of the recombinant gene by the native TPI1 promoter and terminator [21]. The combination of the CPOT expression system and B184 has shown effective for high-level production of several  [19,20]. The plasmids pNat-ZACPOT, pNatZZACPOT and pNatZAZCPOT were used to transform B184k and the positive transformants were grown for 48 h in liquid SD2xSCAA media [22]. The supernatant was analyzed by a reducing SDS-PAGE followed by western blot using antibodies against both the ABD and the Z domain ( Fig. 2A-C). Although all three Affibody molecules were secreted the bands did not appear with the expected sizes on the blot. We expected a 12 kDa band for Z HER3_1 -ABD and 18.9 kDa for Z HER3_1 -ABD-Z HER3_1 and Z HER3_1 -Z HER3_1 -ABD. The supernatant derived from the Z HER3_1 -ABD-Z HER3_1 expressing strain showed three separate smaller bands on the anti-Z-domain blot with sizes compared to the Z HER3_1 -ABD standard around 12 kDa instead of one intact band seen in the Affibody standard around 18.9 kDa (Fig. 2B). The supernatant of the Z HER3_1 -ABD expressing strain showed two bands instead of one on the anti-Z-domain blot, which appeared around the expected size of 12 kDa (Fig. 2B). The anti-Z-domain blot for Z HER3_1 -Z HER3_1 -ABD showed an apparently intact band around the correct size of 18.9 kDa (Fig. 2B). Finally, the ABD domain showed none or little signal on the anti-ABD blot for any of the Affibody molecules (Fig. 2C). The absence of bands on the anti-ABD blots indicated an absence of intact ABD in the secreted Affibody molecules.
We hypothesized that the secreted molecules were degraded by S. cerevisiae proteases leading to fragmentation ( Fig. 2B and C). We also checked if the presence of BSA might have an influence. Since the albumin-binding domain binds to serum albumin [5], and one of the components of the SD2xSCAA media that was used for cultivation was bovine serum albumin (BSA), we wanted to exclude any BSA-related interference. We tested our hypotheses by incubating one Affibody molecule standard for the Z HER3_1 -ABD in four different solutions. The solutions were fresh sterile medium, spent supernatant, demineralized water with 0.1% BSA and demineralized water. The spent supernatant is the supernatant of cultures in which yeast cells had been cultivated and were removed by centrifugation. If the absence of the intact ABD was due to proteolytic activity, we expected to see degradation by incubating the Affibody molecule standard in the spent supernatant. The spent supernatant for this experiment was derived from a 24-h culture of B184k producing α-amylase. After the incubation, the proteins were separated by reducing SDS-PAGE and analysed by western blotting against the ABD (Fig. 2D

Affibody molecules are degraded by aspartyl protease(s)
Proteolytic activity can drastically decrease recombinant protein yield; however, this can often at least partially be prohibited by the identification and removal of the responsible proteases [23]. First, we wanted to check if the degradation was a characteristic of the engineered yeast strain B184k, since a potential solution for the degradation would be to switch to another strain. We checked the supernatant of the parental strain of B184k, AAC [19]. The influence of expression of a recombinant protein on the degradation was also tested by the expression of α-amylase by AAC. Both B184k and AAC containing either pNatAmyCPOT or an empty CPOT plasmid were cultivated. The Affibody standard of Z HER3_1 -ABD-Z HER3_1 was incubated in spent medium from those cultivations, and the mixture with the proteins was separated by reducing SDS-PAGE (Additional file 1: Fig. S1). We included supernatant after 24, 48, 72, and 96 h of cultivation and tested the presence of a protease inhibitor cocktail to confirm that the degradation was due to proteases. Degradation of the Affibody molecules was also observed in medium spent by the strain AAC (Additional file 1: Fig. S1). In the supernatant containing protease inhibitor cocktail, we did not observe degradation, confirming that the fragmentation of the Affibody molecules was a result of proteolytic activity. We observed a similar degradation pattern as for B184k. Degradation, however, seemed slightly elevated in supernatant derived from strains with recombinant α-amylase production, which could potentially indicate an induction of protease expression or activity upon recombinant protein production. Since the degradation was observed also in the parental strain of B184k we concluded that the degradation is not a characteristic of B184k. Therefore, we continued with B184k as the production host. The next step was to identify the proteases responsible for the degradation. Protease inhibitors mostly block the activity of a specific class of proteases. Therefore, a mixture of different protease inhibitors is combined in protease inhibitor cocktails to ensure inhibition of all sort of proteases. Since in the experiment with AAC (Additional file 1: Fig. S1) a protease inhibitor cocktail showed effective against the degradation we tested the isolated inhibitors from that specific cocktail. The spent medium of B184k expressing α-amylase was incubated overnight with Affibody molecule standards and the individual components of the protease inhibitor cocktail. The mixture was subsequently analyzed by a reducing SDS-PAGE. The results of the SDS-PAGE show that the Affibody molecules were degraded in the presence of AEBSF, aprotinin, bestatin, E-64 or leupeptin (Fig. 3). Only incubation with pepstatin A prohibited degradation and resulted in Affibody molecules with the correct size (Fig. 3). Both the Affibody molecules Z HER3_1 -Z HER3_1 -ABD and Z HER3_1 -ABD-Z HER3_1 clearly showed the absence of proteolytic processing on the SDS-PAGE in the presence of pepstatin A. For Z HER3_1 -ABD, the difference was less explicit which could be a result of the small size difference between the intact and degraded molecule. Pepstatin A inhibits aspartyl proteases, indicating that the protease(s) responsible for Affibody molecule processing are of this type.

The removal of several proteases results in the secretion of intact Affibody molecules
One of the major proteases in S. cerevisiae and a proven target for improving recombinant protein production yield is a vacuolar aspartyl protease, proteinase A [23][24][25]. We removed the PEP4 gene encoding proteinase A in B184 using CRISPR/Cas9-based gene deletion and expressed all three Affibody molecules in B184 pep4Δ. Z HER3_1 -ABD-Z HER3_1 produced by this strain was intact and showed a signal on western blots using antibodies against either the Z-domain or ABD (Fig. 4), indicating that proteinase A was indeed responsible for its cleavage in B184. In contrast, for Z HER3_1 -ABD and Z HER3_1 -Z HER3_1 -ABD, the western blot against the Z-domain showed a small shift of the bands upon removal of PEP4 but there was no band visible on the blot against the ABD (Fig. 4A). We suspected that at least one other protease was involved in cleaving the ABD domain of Z HER3_1 -ABD and Z HER3_1 -Z HER3_1 -ABD in B184 pep4Δ. We therefore removed two additional proteases in B184 pep4Δ in the hope of producing intact Z HER3_1 -ABD and Z HER3_1 -Z HER3_1 -ABD, namely carboxypeptidase Y and proteinase B. Carboxypeptidase Y is a serine-based exopeptidase active in the vacuole and proteinase B, a serine-based endopeptidase also located to the vacuole. Proteinase B, together with proteinase A, plays an essential role in the activation of several vacuolar proteases including itself and carboxypeptidase Y. The PRC1 gene encoding carboxypeptidase Y and the PRB1 gene encoding proteinase B were removed in B184 and B184 pep4Δ (pep4Δprc1Δ and pep4Δprb1Δ) and a triple deletion strain (pep4Δprc1Δprb1Δ) was constructed. The single, the two double and the triple deletion strains were grown for 48 h in an aerated 24-well plate and the supernatant in which triple deletion strains expressing the Affibody molecules were grown was analyzed for the presence of intact Affibody molecules. In the B184 pep4Δprc1Δ strain we observed intact Z HER3_1 -ABD and Z HER3_1 -Z HER3_1 -ABD ( Fig. 5A and B). Interestingly, the removal of solely PEP4 during cultivation for 48 h instead of 96 h, resulted in very weak bands for the ABD for both Z HER3_1 -ABD and Z HER3_1 -Z HER3_1-ABD on the blot against the ABD. This indicates that the degradation in B184 pep4Δ could primarily occur at later cultivation stages. The single removal of either PRC1 or PRB1 did not result in a band on the western blot using antibody against the ABD. However, the combination of pep4Δ (proteinase A) with either prb1Δ (proteinase B) or prc1Δ (carboxypeptidase Y) resulted in the production of intact Z HER3_1 -ABD as well as intact Z HER3_1 -ABD-Z HER3_1 . The deletion of all three proteases (pep4Δprc1Δprb1Δ) resulted in titers comparable to the two double deletion strains ( Fig. 5A and B).
Proteases fulfill a crucial role in the degradation of macromolecules in the vacuole and have even been reported to benefit aging [26,27]. Therefore, the removal of one to three major proteases can have an impact on vacuolar homeostasis and cellular growth. We monitored the growth profiles of the protease-deficient strains in aerated 96-well plates. The protease deficiency led to a slight reduction of final OD 600 in the later stages of the growth, from 50 to 96 h, but overall growth of the deletion strains, even the triple deletion strain, was similar to B184 with the proteases intact. This was observed for both Z HER3_1 -ABD and Z HER3_1 -Z HER3_1 -ABD ( Fig. 5C and D).
Additionally, we were interested in identifying the cleavage site of the endopeptidase proteinase A in the ABD. It is known that proteinase A has a broad and variable activity for different ligands and seems to favor cut sites between adjacent hydrophobic residues [28,29]. We incubated the Z HER3_1 -ABD-Z HER3_1 standard in spent supernatant for 24 h and analysed the mixture with mass spectrometry. The total ion chromatogram (TIC) showed four peaks, two large peaks and two smaller ones (Fig. 6A). The peptides within the two larger peaks were analysed. In the first peak, several fragments of Z HER3_1 -ABD-Z HER3_1 were identified where either the N-or C-terminus was intact and the other terminus was located within the ABD indicating a cut site of an endopeptidase (Fig. 6B). The peptides in the fourth peak were fragments which had the N-terminus in the ABD and the C-terminus in the second Z HER3_1 sequence (Fig. 6B). Based on these fragments it seems that proteinase A cleaves at several sites within the ABD and interestingly the C-terminal Z HER3_1 domain was also not intact. We next aimed to test whether the engineered S. cerevisiae strain is a competitive host for the Affibody molecule production with regards to product quality. Since the three Affibody molecules consist of the same two subunits, Z HER3_1 and ABD, we decided to include only one version in the subsequent studies as a proof of concept. The Z HER3_1 -ABD-Z HER3_1 molecule produced and secreted by B184 pep4Δ was tested in a surface plasmon resonance binding assay with the substrate of Z HER3_1 , HER3. The Z HER3_1 -ABD-Z HER3_1 Affibody molecule has two binding sites for HER3; therefore, the bivalent binding kinetics were determined and presented as average value of duplicates in Table 1. The equilibrium dissociation constant of the first site has a value in the nano molar range, which indicates high affinity. The values of the Z HER3_1 -ABD-Z HER3_1 molecules secreted by E. coli and S. cerevisiae, respectively, show comparable kinetics which shows that the molecules produced by S. cerevisiae are fully functional ( Table 1, Additional file 1: Fig. S2).

B184 pep4Δ produces a high titer of Z HER3_1 -ABD-Z HER3_1 in a fed-batch bioreactor
For S. cerevisiae to be a suitable host for Affibody molecule production it must be able to produce competitive titers of Affibody molecules. So next, we wanted to assess the productivity of S. cerevisiae in bioreactors. We decided to use B184 pep4Δ producing Z HER3_1 -ABD-Z HER3_1 as a proof of concept. To exclude any major impact of the deletion of pep4Δ on cellular growth of B184 while producing Z HER3_1 -ABD-Z HER3_1 we tested B184 and B184 pep4Δ in a micro-cultivation experiment in an aerated 96-well plate (Additional file 1: Fig. S3). Deletion of pep4Δ conferred only a minor impact on growth (Additional file 1: Fig. S3). After this confirmation the productivity for Z HER3_1 -ABD-Z HER3_1 of B184 pep4Δ was tested in a bioreactor experiment. The cultivation set-up was a batch fermentation followed by fed-batch fermentation. During the batch phase, a specific growth rate of 0.31 h −1 was measured (Additional file 1: Fig. S4) which is the same growth rate as for B184 carrying an intact PEP4 gene and producing α-amylase which confirmed that B184 pep4Δ did not suffer from a growth impairment [20]. The feeding of a low-glucose feed into the bioreactors was started after 36 h and was switched to a high-glucose feed after 130 h for the remaining duration of the fermentation until 180 h. At the end of the fed-batch cultivation, the biomass concentration exceeded 100 g/L (Fig. 7A). We also observed an increasing trend of Affibody molecule titer with the duration of the experiment especially after the switch to the high-glucose feed (Fig. 7A). During the cultivation, minor production of byproducts was observed. Ethanol and glycerol were present in the medium after the batch phase but were rapidly consumed after the feed was started (Fig. 7B). At the end of the fed-batch, residual glycerol increased again and reached a final concentration of 10 g/L (Fig. 7B). We determined the exact quantity of Z HER3_1 -ABD-Z HER3_1 at three timepoints using a BLI based method. The results showed a final titer of 530 mg/L (Fig. 7C).

Discussion
In this study we examined S. cerevisiae as a host for Affibody molecule production. Currently, the molecules are mainly produced in E. coli. In the current study, we present engineered protease-deficient S. cerevisiae as a suitable alternative.
Our results show that all three Affibody molecules were produced and secreted by S. cerevisiae after the removal of only proteinase A or with the additional removal of proteinase B or carboxypeptidase Y. Removal of only proteinase A was sufficient for the production of intact Z HER3_1 -ABD-Z HER3_1 and a small amount of both Z HER3_1 -ABD and Z HER3_1 -Z HER3_1 -ABD in the case of a shorter cultivation of 48 h instead of 96 h. We suspect that a longer duration of the experiment would lead to more degradation. The additional removal of carboxypeptidase Y and/or proteinase B is required for the intact production of Z HER3_1 -ABD and Z HER3_1 -Z HER3_1 -ABD, respectively. The ABD is in Z HER3_1 -ABD-Z HER3_1 flanked by the two Z HER3_1 domains, which potentially protects the ABD from exopeptidase activity from proteases like carboxypeptidase Y. This would explain why we observed intact Z HER3_1 -ABD and Z HER3_1 -Z HER3_1 -ABD only upon removal of carboxypeptidase Y in proteinase A deficient B184 pep4Δ. Additionally, we confirmed that the removal of carboxypeptidase Y does not further improve the production of Z HER3_1 -ABD-Z HER3_1 in proteinase A-deficient B184 pep4Δ (Additional file 1: Fig. S5). Interestingly, we also observed production of intact Z HER3_1 -ABD and Z HER3_1 -Z HER3_1 -ABD upon removal of proteinase B, which is an endopeptidase, in B184 pep4Δ and the additional deletion of proteinase B in B184 pep4Δprc1Δ (lacking proteinase A and carboxypeptidase Y) or carboxypeptidase Y in B184 pep4Δprb1Δ (lacking proteinase A and proteinase B) did not influence the titer of intact Z HER3_1 -ABD and Z HER3_1 -Z HER3_1 -ABD. This observation can be explained by the fact that proteinase A, proteinase B and carboxypeptidase Y are all three vacuolar proteases that before reaching their final destination, the vacuole, pass through the secretory pathway as inactive precursors with a prepro-signalpeptide [30]. The pre-signal peptide is removed after arrival in the ER and the prosignal peptide during trafficking to or upon arrival in the CDW (g/L) Affibody molecules (mg/L)

Glucose added (g)
Pyruvate (g/L)  [30]. Proteases can have multiple pro-signal peptides, which prohibits the proteases to reach their active form. Proteinase A and B are responsible for the removal of pro-signal peptide(s) on other vacuolar proteases and are therefore a central part of the activation cascade of vacuolar proteases.
For maturation of our deletion targets proteinase A, proteinase B and carboxypeptidase Y, activity of both proteinase A and B is required to reach their activate form. The maturation of proteinase A starts with an auto activation step to the pseudo form upon entering the vacuole, which is followed by removal of a second prosignal peptide by proteinase B [31,32]. The intermediate form of the protease with one pro-signal peptide and one already removed is called the pseudo-form. The exact activation trigger for the proteases remains unknown but has been linked to the acidic pH in the vacuole [33,34]. Proteinase B has two pro-peptides, one at the N-terminus and another at the C-terminus. The removal of the N-terminal pro-peptide is autocatalytic and occurs in the Golgi apparatus, and the C-terminal pro-peptide is removed by proteinase A [35]. Carboxypeptidase Y has an N-terminal pro-peptide which is partly removed by proteinase A and the other part by proteinase B upon arrival in the vacuole.
As explained here, proteinase B catalyzes the activation of carboxypeptidase Y and based on our results we expect that proteinase B fulfills a role in the activation of proteinase A and carboxypeptidase Y rather than to recognize Affibody molecules as a substrate. We expect, based on our results, that in a pep4Δprb1Δ mutant carboxypeptidase Y remains in an inactive pro-form, which limits the capacity of carboxypeptidase Y to degrade the ABD [36]. We, however, observe digestion of the ABD by strain B184 pep4Δ. It has been shown that in a PEP4 deficient S. cerevisiae strain pro-proteases of both carboxypeptidase Y and proteinase B accumulate, alongside the reduction of their activity [37]. The combination of deleting proteinase A and proteinase B for improved recombinant protein production has been reported for several microbial hosts, including S. cerevisiae [25,38,39]. Removal of solely PEP4 has however also proven effective for production of several proteins [25,41,42].
The proteases travel through the secretory pathway to reach their destination, the vacuole. The initiation of the activation should start in the proximity of the vacuole or upon transport into this organelle [40]. Proteinase A and B are trafficked directly from the Golgi apparatus to the vacuole, but carboxypeptidase Y travels through a multivesicular body intermediate [26]. Even though the proteases are targeted for the vacuole, residual secretion has been reported and linked to the glucose concentration in the media. In the experiments performed in this study 2% glucose concentration in the media was used, a concentration that previously was shown to result in extracellular activity of proteinase A [34]. This supports the observations of proteinase A activity in the spent supernatant.
Another interesting aspect is the propensity of the proteases for cleavage of ABD. Since potentially even pro-Prc1 and pseudo-Prc1 seem to digest ABD, it could well be that even partly active yeast proteases retain a high propensity for cleaving the ABD as a defense mechanism against bacteria. Proteolytic activity targeting the ABD was previously found in bacterial hosts [43].
The mass spectrometry performed for the analysis of Z HER3_1 -ABD_Z HER3_1 showed that the molecular weight of the molecule produced by S. cerevisiae consisted of an extra 700 Da compared to the molecule produced by E. coli. This is visible on several SDS-PAGEs and western blots (Figs. 4, 5, Additional file 1: Figs. S1 and S5). In the construct expressed by S. cerevisiae, a Kex2 recognition site is located behind the α-leader sequence and is processed in the ER. Between the Kex2 site and the start of the protein, a linker peptide with the amino acid sequence Glu-Glu-Gly-Glu-Gly-Ser-Met was inserted, which exhibits a molecular weight of 737 Da and thus explains the additional mass of the produced Z HER3_1 -ABD-Z HER3_1 . Since the binding kinetics indicate no interference of this additional peptide, we did not include the removal of the linker peptide in the scope of this study, but for future applications its removal may be desirable.
As a final experiment to show the potential of S. cerevisiae as interesting production host for Affibody molecules, B184 pep4Δ producing Z HER3_1 -ABD-Z HER3_1 was grown in fed-batch bioreactors. During the fed-batch cultures, buildup of residual glycerol was observed. The quantity of glycerol accumulating showed high variation between the reactors. Glycerol is a by-product of fermentation to reoxidize NADH to counteract redox imbalances. However, we do not expect this to be the cause of the buildup [44]. During later stages of the fedbatch, the feed was controlled by the dissolved oxygen in the reactor leading to pulse feed to ensure respiratory growth. Besides, in the case of alcoholic fermentation, there should be buildup of ethanol as well [45]. Therefore, the oxygen shortage does not seem to be the source of the potential redox imbalance. Besides redox imbalance, another explanation could be the expression platform used. TPI1 deficiency, which is the deficiency complemented by the CPOT plasmid, is known to result in the accumulation of high levels of glycerol [46]. During the later stages of the experiment, this partial complementation could become limiting leading to the accumulation of excess glycerol. Nevertheless, high biomass concentration and Affibody molecule titer were achieved during the fermentation. The titer that was reached was 0.530 g/L. For E. coli expression titers for different recombinant proteins are reported in the range from 0.25 g/L to 8.5 g/L [47][48][49]. The titer of Z HER3_1 -ABD-Z HER3_1 is within this range and can be the starting point for improvement by further engineering S. cerevisiae or by optimizing the cultivation method in a sequential study.
The results of this study indicate that S. cerevisiae shows potential as a host for Affibody molecule production. Here, we tested only three different Affibody molecules, and for the future it would be interesting to expand the selection of Affibody molecules and potentially test more complex Affibody molecule based proteins like AffiMabs [7]. Secondly, our experiments show that the removal of proteases can be a powerful approach to increase productivity of heterologous proteins in S. cerevisiae.

Strains and plasmids
Two previously constructed S. cerevisiae strains were used in this study, CEN.PK 530.1C [MATα URA3 HIS3 LAU2 TRP1 SUC2 MAL2-8 c tpi1 ] (AAC) and B184k. In a previous study, AAC was evolved for improved protein production by UV-mutagenesis which led to the construction of strain B184k [19]. B184k was shown to be an effective host for the production of several recombinant proteins in combination with the CPOT plasmid [19]. The CPOT plasmid contains the Schizosaccharomyces pombe gene POT1, which partially complements the TPI1 deficiency. The recombinant protein expression cassette contains the native TPI1 promoter and terminator and an α-leader sequence. The empty CPOT is a CPOT plasmid without the recombinant protein gene. The previously constructed pNatA-myCPOT was cut by restriction digest with enzymes XhoI and KpnI to remove only the α-amylase gene. Genes for Z HER3_1 -ABD, Z HER3_1 -Z HER3_1 -ABD and Z HER3_1 -ABD-Z HER3_1 were synthesized by GenScript. The Affibody molecule genes were codon optimized for expression in S. cerevisiae but without repetitive DNA sequences to reduce the risk of homologous recombination within the ORFs. The three Affibody molecule genes were amplified with homologous overhangs of the α-leader and the CPOT backbone using primers presented in Additional file 1: Table S1. The backbone and Affibody molecule genes were assembled by Gibson assembly and the final constructs verified by sequencing using primer #7 that binds in the α-leader sequence. All primers for the construction of the plasmids can be found in Additional file 1: Table S1. After the plasmid construction, the newly assembled CPOT plasmids, pNatZAC-POT, pNatZZACPOT and pNatZAZCPOT were used to transform AAC and B184k. B184k still contained a kanamycin resistance marker in the TPI1 gene. In AAC, the marker had been removed previously. The marker has two flanking loxP sites. pSH66 from the Euroscarf deletion marker set was used to remove the kanamycin resistance marker [50]. pSH66 contains an expression cassette for Cre recombinase under control of the GAL1 promoter and a nourseothricin resistance gene. After the strains had been transformed with the pSH66 plasmid with selection for nourseotricin, the positive transformants were streaked out on solid media with YPGal with nourseothricin to activate Cre recombinase expression. Removal of the kanamycin marker was confirmed by absence of growth on solid media with G418. PEP4, PRC1 and PRB1 were deleted by using plasmids pECAS9-gRNA-kanMX-tPEP4, pECAS9-gRNA-kanMX-tPRC1, pECAS9-gRNA-kanMX-tPRB1, which contain both a cas9 gene and a gRNA expression cassette [51]. The plasmids pECAS9-gRNA-kanMX-tPEP4, pECAS9-gRNA-kanMX-tPRC1 and pECAS9-gRNA-kanMX-tPRB1 were constructed using pECAS9-gRNA-kanMX-tHFD1 as template [51]. First, the backbone of pECAS9-gRNA-kanMX was obtained by linearizing pECAS9-gRNA-kanMX-tHFD1 by digestion with MunI and EcoRI. The 'left' fragment was constructed with primer #14 in combination with either #9 (PEP4), #11 (PRC1) or #13 (PRB1) and the 'right' fragment was constructed with primer #15 in combination with either #8 (PEP4), #10 (PRC1) or #12 (PRB1). The correct assembly of the plasmids was confirmed by sequencing using primer #16. The genes were deleted in B184 pNatZACPOT, B184 pNatZZAC-POT and B184 pNatZAZCPOT. The genomic deletions were verified using forward primers #17 and #18 with reverse primer #19 for PEP4, forward primers #20, #21 and reverse primer #22 for PRC1, and forward primers #23, #24 and reverse primer #25 for PRB1. The CRISPR plasmids were removed by subsequent cultivation in liquid YPD confirmed by absence of growth on solid media with G418. All the primers are presented in Additional file 1: Table S1 and repair fragments for CRISPR in Additional file 1: Table S2. The oligonucleotides were aligned by heating the two corresponding oligonucleotides to 98 degrees for 5 min in equimolar amounts and let to mixture cooldown to room temperature. E. coli DH5α was used for plasmid amplification. The transformation protocol used for E.coli was according to a known protocol [52]. The strains and plasmids used and constructed in this study are listed in Tables 2 and 3.

Media and culture conditions
The media used for S. cerevisiae strain construction were YPD, YPGal, YPE and YPEG. The experiments were always performed at 30 °C and for liquid cultures at 220 rpm. YPD medium contained 10 g/L yeast extract, 20 g/L peptone, and 20 g/L glucose and was used for regular cultures. YPGal media contained 10 g/L yeast extract, 20 g/L peptone, and 20 g/L galactose and was used for induction of the Cre recombinase gene on pSH66. For the selection of the kanMX marker on the CRISPR plasmid, 200 mg/L G418 was added to the YPD medium. For the selection of pSH66-containing cells, 100 mg/L nourseothricin sulfate was added to YPD and YPGal medium. The YPE medium contained 10 g/L yeast extract, 20 g/L peptone, 20 g/L absolute ethanol and was solely used as a solid medium. For liquid cultivations, 30 g/L glycerol was added, and the medium referred to as YPEG. Both YPE and YPEG were only used for S. cerevisiae strains without CPOT plasmids since those are unable to ferment glucose as the sole carbon source [54]. To solidify the media 20 g/L agar (Merck Millipore) was added. The protein expression and physiological experiments were performed in SD2xSCAA media at 30 °C and 220 rpm. SD2xSCAA medium contained 20 g/L glucose, 6.9 g/L yeast nitrogen base without amino acids,

Molecular biology techniques
Saccharomyces cerevisiae strains were transformed using the LiAc/SS carrier method [55]. One µg of DNA was used for the transformation of plasmids and an additional 1 to 2 μg repair fragment if required. To verify deletions or test for the presence of the CPOT plasmids, colony PCR using SapphireAmp fast PCR mix (TaKaRa Bio) was performed. For DNA construction, Phusion High Fidelity DNA polymerase (Thermo Scientific) or Herculase II Fusion DNA polymerase (Aligent) was used. Restriction digestion was done using FastDigest (Thermo Scientific) products. All techniques were used according to the manufacturers' protocols unless stated otherwise.

Growth profiler experiments
Three independent transformation colonies per strain were grown for 24 h in 1 mL SD2xSCAA media in a 14 mL-cultivation tube. Those precultures were used to inoculate the main cultures of the growth experiment in technical triplicates with a starting OD of 0.01. The S. cerevisiae strains were cultivated for 96 h in 250 μL SD2x-SCAA media at 30 °C and 1200 rpm in 96-well plates (Enzyscreen CR1496d). The growth curves were measured using a Growth Profiler 960 (Enzyscreen).

Protease activity experiments
For the incubation experiments, supernatants of cultivations were used. The initial cultivation was a 24 h-cultivation, if not mentioned otherwise, of a single colony in 2 mL of SD2xSCAA media. After the incubation, the culture was harvested by centrifugation at 6000 rpm for 3 min. The supernatant was kept on ice or frozen at − 20 °C. The incubation was performed in 1.5-mL Eppendorf tubes with 200 μL volume of supernatant. As positive controls, standards of purified Z HER3_1 -ABD (1.97 mg/mL), Z HER3_1 -Z HER3_1 -ABD (0.77 mg/mL) and Z HER3_1 -ABD-Z HER3_1 (1.34 mg/mL) provided by Affibody AB were used with a concentration of 0.01 g/L or stated otherwise. The experiments with the protease inhibitors were done by adding the protease inhibitor before adding the Affibody standard at the concentration suggested by the supplier. We used the Halt Protease Inhibitor Cocktail EDTA-free (100x) (Thermo Fisher) and 4-(2-aminoethyl) benzenesulfonyl fluoride hydrochloride (AEBSF) (final concentration 1 mM), aprotinin (final concentration 6.5 μg/mL), bestatin hydrochloride (final concentration 50 µM), leupeptin (final concentration 20 µM), E-64 (final concentration 15 µM), and pepstatin A (final concentration 10 µM). All these chemicals were purchased from Merck.

SDS-PAGE and western blotting
The samples and controls were loaded and separated with reducing SDS-PAGE. Stain free 4-20% gels were used (Bio-rad). The proteins were transferred to 0.45-micron PVDF membranes (Bio-rad) using the Trans-Blot Turbo transfer system (Bio-rad). The blot was blocked using the Western blocker solution (Sigma Aldrich) and incubated in with either anti-Z-domain (2.87 mg/ mL) (1:1000) or anti-ABD (1 mg/mL) (3:1000) antibodies, both obtained from Affibody AB followed by incubation with either anti-mouse antibody (1:5000) for anti-Z-domain or anti-rabbit (1:5000) antibody for anti-ABD antibodies, respectively. Both secondary antibodies were HRP-conjugated and visualized by using West Pico Plus HRP substrate (Thermo Fischer) and measured with a ChemidoC XRS image analyzer (Bio-Rad). As positive controls, standards of purified Z HER3_1 -ABD (1.97 mg/mL), Z HER3_1 -Z HER3_1 -ABD (0.77 mg/mL) and Z HER3_1 -ABD-Z HER3_1 (1.34 mg/mL) provided by Affibody AB were used with a concentration of 0.01 g/L or stated otherwise.

Mass spectrometry analysis
Intact mass analysis was performed on an Agilent (1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC) and N-hydroxysuccinimide (NHS)) on a CM5 sensor chip (Cytiva, Marlborough, MA) at ~ 1000 RU. The coated chip was preconditioned by three regeneration rounds to stabilize surfaces prior to injection of analyte. Binding of Affibody molecules to HER3 was analyzed by single cycle kinetics via injection of analyte at five different concentrations of purified Affibody molecule (1.25, 2.5, 5, 10 and 20 nM) over immobilized HER3/Fc. The experiment was performed in duplicates. Biacore Insight Evaluation Software was used to process, analyze and fit data.

Fed-batch bioreactor cultivations
For the Bioreactor experiments, the Na 2 HPO 4 and NaH 2 PO 4 •2H 2 O were replaced by KH 2 PO 4 with a concentration of 2 g/L in the SD2xSCAA media. The batch fermentations were performed in 1-L bioreactors (Das-Gip) with a start OD 600 of 0.01 in 500 mL of SD2xSCAA media. The conditions in the reactors were controlled and maintained at a pH of 6.0 by 4 M KOH, agitation at 600 rpm, temperature at 30 °C degrees and airflow of 30 L/h. The batch experiments were performed in biological quadruplicates. After all the glucose was consumed, culture broth was removed until a start volume of 250 mL. The conditions of the fed-batch were set to a pH of 6.0, maintained by 4 M KOH and 3 M HCl, agitation was set to 600 rpm, the temperature was set to 30 °C degrees, a constant airflow of 18 L/h and an exponentially increasing feed rate of 0.05 h −1 . When the dissolved oxygen in the bioreactor would decrease below 30% the airflow and agitation would increase. The airflow would increase to 48 L/h and agitation would increase to 1000 rpm. When both the agitation and airflow were at their maximum levels the feeding was changed to a pulse feed controlled the by dissolved oxygen level. The feed would stop if the dissolved oxygen was below 25% which resulted in a pulsed feeding rate around 8 mL/h. Two types of media were used for the fed-batch experiments. The first medium was the low-glucose feed with 200 g/L, 69 g/L yeast nitrogen base without amino acids, 50 g/L casamino acids (Formedium), 1 g/L BSA, and 20 g/L KH 2 PO 4. After adding an average of 320 mL of the low glucose feed the low-glucose feed was replaced by a highglucose feed with the same composition as the low-glucose feed except for a glucose concentration of 400 g/L. The fermentation was ended after addition of 230 mL of the high-glucose feed. The fed-batch experiments were conducted in biological triplicates.

Affibody molecule quantification by Bio-Layer Interferometry (BLI)
The broth samples were centrifugated for 20 min at 4000 rpm to remove the biomass and the supernatant was stored at − 80 °C. After thawing some precipitation appeared in the samples. To redissolve the precipitation the samples were 1:1 diluted in 1 NaCl in two duplicates per fermenter per sample point and transported on dry ice. Upon arrival the sample were thawed and either not diluted or diluted 1:3.75, 1:7.5 in sample diluent (1 × PBS, 0.1% BSA, 0.02% Tween 20) based on the expected concentration. 225 µl of the pre-dilution was mixed with either 75 µL sample diluent (unspiked) or 75 µl of 100 µg/mL Z HER3_1 -ABD-Z HER3_1 (produced by E. coli) (spiked) was added resulting in a 1:5, 1:10 or 1:1.33 final dilution. 200 µL of the spiked and unspiked sample was moved to a black flat-bottomed 96-well plate (Greiner, #655209) together with 200 µL of the samples for the standard curve, 200 µL sample diluent (reference well) and 200 µL cultivation medium (negative control). High precision streptavidin (SAX) biosensors (ForteBio, #18-0037) which were previously loaded offline (Technical Note #10, ForteBio) with 50 µg/mL biotinylated HSA (Recombumin Prime, Albumedix; Thermo Fisher #A39259) (MCR 1:1) were placed for 10 min in sample diluent before the measurements. The data was acquired using an Octet HTX system and Octet Data Acquisition software (ForteBio, ver. 12). The measurement settings were as follows: 8 channel read head, 5.0 Hz acquisition rate, quantitation step: 120 s, 400 rpm, plate temperature 30 °C. The quantity was determined based on the six-point standard curve with Z HER3_1 -ABD-Z HER3_1