bs_bs_banner Rationale-based selection of optimal operating strategies and gene dosage impact on recombinant protein production in Komagataella phaffii (Pichia pastoris) Miguel Angel Nieto-Taype, Javier were determined for both models due to the different Garrigo s-Martınez, Marc Sanchez-Farrando,† production kinetic patterns observed as a trade-off Francisco Valero,* Xavier Garcia-Ortega and for product titre, yields and productivity. The com-  Luis Montesinos-Seguı Jose bined effect of gene dosage and adequate µ enables Department of Chemical, Biological and Environmental rational process development with a view to optimize Engineering, School of Engineering, Universitat K. phaffii RPP bioprocesses.  noma de Barcelona, 08193, Bellaterra, Spain. Auto Introduction Summary Komagataella phaffii, formerly known as Pichia pastoris, Its features as a microbial and eukaryotic organism is a widely used yeast for recombinant protein produc- have turned Komagataella phaffii (Pichia pastoris) tion (RPP), for both biopharmaceuticals and an increas- into an emerging cell factory for recombinant protein ing number of industrial enzymes of interest (Puxbaum production (RPP). As a key step of the bioprocess et al., 2015; Burgard et al., 2017). This host has major development, this work aimed to demonstrate the advantages such as a wide range of genetic modification importance of tailor designing the cultivation strat- tools including genome editing toolkits are available (e.g. egy according to the production kinetics of the cell the CRISPR/Cas9 system); its ability to grow to a high factory. For this purpose, K. phaffii clones constitu- cell density in defined media, to perform eukaryotic post- tively expressing (PGAP) Candida rugosa lipase 1 translational modifications and to release target products (Crl1) with different gene dosage were used as mod- extracellularly. These features in combination make els in continuous and fed-batch cultures. Production K. phaffii a promising cell factory for industrial biotech- parameters were much greater with a multicopy nology (Potvin et al., 2012; Vogl and Glieder, 2013; clone (MCC) than with the single-copy clone (SCC). Weninger et al., 2018). Regarding production kinetics, the specific product The increasing demand for recombinant proteins has generation rate (qP) increased linearly with increas- generated a multibillion-dollar market over the last few ing specific growth rate (µ) in SCC; by contrast, qP decades (Highsmith, 2015; Dewan, 2017). Therefore, exhibited saturation in MCC. A transcriptional analy- important efforts are being dedicated to increase biopro- sis in chemostat cultures suggested the presence of cess efficiency and profitability. Two widely reviewed eventual post-transcriptional bottlenecks in MCC. complementary approaches are currently being devel- After the strain characterization, in order to fulfil oped to reach these goals, namely strain engineering overall development of the bioprocess, the perfor- (Zahrl et al., 2017; Juturu and Wu, 2018; Vogl et al., mance of both clones was also evaluated in fed- 2018a) and bioprocess optimization (Theron et al., 2018; batch mode. Strikingly, different optimal strategies Yang and Zhang, 2018). Using efficient promoters is essential to ensure effi- cient recombinant protein expression in this context. The Received 17 May, 2019; revised 3 October, 2019; accepted 4 methanol-inducible alcohol oxidase promoter (PAOX1) October, 2019. has been widely used in K. phaffii bioprocesses by virtue *For correspondence. E-mail francisco.valero@uab.cat; Tel. +34 93 581 1809; Fax +34 93 581 2013. of allowing a strong and tight regulation for the recombi- † Present address: Purac Bioquımica S.A. (Corbion Montmelo ), nant expression in the presence of methanol (Barrigo n Montmelo , Spain. et al., 2013; Ponte et al., 2016; Vogl et al., 2016). How- Microbial Biotechnology (2020) 13(2), 315–327 doi:10.1111/1751-7915.13498 ever, PAOX1-driven expression bioprocesses are subject Funding information to constraints derived from the use of methanol as indu- This work was funded by MINECO and FEDER under Project cer. Thus, using methanol at the industrial scale requires CTQ2016-74959-R. ª 2019 The Authors. Microbial Biotechnology published by John Wiley & Sons Ltd and Society for Applied Microbiology. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. 316 M. A. Nieto-Taype et al. adopting safety measures that raise production costs increasing l. The results obtained with some fed-batch and is subject to operational problems arising from cultures are also consistent with synergism in these two heavy high heat production and oxygen demand (Priel- variables (Zhao et al., 2008; Garcia-Ortega et al., 2013). hofer et al., 2013), cell metabolic burdens (Hartner and Therefore, because gene dosage is expected to affect Glieder, 2006), culture cell lysis and potential subse- production rates, one should assume that it may consid- quent proteolysis of the target product (Mattanovich erably influence production kinetics in assessing its et al., 2009). effects. Alternative promoters avoiding the use of methanol Candida rugosa lipase (Crl) is one of the most promis- have recently been explored (Liang et al., 2013; Priel- ing lipase enzymes for biocatalytic applications (Ken hofer et al., 2013; Shen et al., 2016; Vogl et al., 2018b; Ugo et al., 2017). At least seven genes of C. rugosa Robert et al., 2019). The glyceraldehyde-3-phosphate lipases (CRL1-CRL7) have been identified; also, all dehydrogenase GAP promoter (PGAP), which is involved except CRL6 and CRL7 have been identified and in a key step of the glycolysis pathway, was the first to sequenced (Ferrer et al., 2001). Crl1, which accounts for emerge as a benchmark for efficient protein expression about 80% of all lipase present in commercial powders, on various carbon sources. Thus, by avoiding all metha- is the most widely studied (Sa nchez et al., 1999) nol-related drawbacks, PGAP-based bioprocesses present Because of the difficulty involved in isolating the pure relevant advantages for large-scale production (Zhang isoenzyme from the native microorganism, it has been et al., 2009; Ahmad et al., 2014; Cß alık et al., 2015). alternatively obtained from K. phaffii cell factory (Valero, Some authors have found copy number integration of 2018). the expression cassette in the genome, also called gene For the present work, the isoenzyme Candida rugosa dosage, to play a central role in specific productivity lipase 1 (Crl1) was selected as model protein to eluci- (Schwarzhans et al., 2016a; Vogl et al., 2018a). Using date the differences in the rational design of optimal bio- large numbers of gene copies results in increased pro- process strategies for two clones of K. phaffii with ductivity in some cases (Norde n et al., 2011; Prielhofer different gene dosage, as example of clone variability in et al., 2013; Zhu et al., 2014) but has the opposite effect terms of protein production. For this purpose, an accu- in others (Zhu et al., 2009; Liu et al., 2014; Ca mara rate characterization of physiological parameters and the et al., 2016).In fact, as claimed, integrating several production kinetics was firstly performed on chemostat expression cassettes in the genome may have adverse cultures. The obtained results allowed the selection of effects owing to the physiological limitations in the tran- the optimal bioprocess strategy to maximize the RPP, in scriptional capacity of gene AOX1, which is governed by order to be applied further in fed-batch cultivations, mara et al., 2017). its transcriptional factors (Ca which is currently considered as the most used opera- Production kinetics, the relationship between specific tional mode for industrial RPP (Garcıa-Ortega et al., production rate (qP) and specific growth rate (l), is con- 2019). As a major outcome, this contribution discusses sidered a key factor to be considered in the bioprocess the influence of gene dosage linked to production kinet- development. It reflects the equilibrium between the vari- ics on the determination of the optimal operating strate- ous steps until the product is secreted, as a balance of gies with a view to maximizing bioprocess production the different processes involved during the protein syn- rates and yields. thesis, folding and secretion. This relationship is crucial to bioprocess development and optimization (Potvin Results and discussion et al., 2012; Looser et al., 2015; C ß alik et al., 2015). Strain construction and gene dosage Thus, Garcia-Ortega et al. (2016) and Maurer et al. (2006) characterized PGAP-based strains producing an In order to get two clones with contrasting production antibody fragment and obtained robust results with them performances, several recombinant clones with different in chemostat systems; so, they found qP to increase gene dosage of CRL1 were obtained by transforming dif- eight times with increasing l. Rebnegger et al. (2014) ferent amounts of plasmid in which the CRL1 expression examined the response of this expression system pro- cassette is placed under PGAP regulation. According to ducing human serum albumin (HSA) at different specific the transformation method used, these cassettes are growth rates at transcriptomic level and observed expected to be integrated by homologous recombination marked upregulation of genes involved in translation. into the native PGAP locus. However, constructing pro- However, genes of the glycolytic pathways such as ducer strains from K. phaffii may result in non-homolo- TDH3, which is the endogenous gene regulated by gous end-joining recombination and/or multiple insertion PGAP, were unregulated or weakly regulated, therefore of the gene expression cassette, which usually expands suggesting that effect in the translational machinery the spectrum of clonal variability (Schwarzhans et al., played a major role in by causing qP to increase with 2016b; Jiao et al., 2018; Vogl et al., 2018a). ª 2019 The Authors. Microbial Biotechnology published by John Wiley & Sons Ltd and Society for Applied Microbiology., Microbial Biotechnology, 13, 315–327 Kinetics in K. phaffii bioprocess design 317 Later, transformed clones were screened in order to and closure found to exceed 95% prior to reconciliation. identify the best producer clone, which was a clone that Fig. 1A and B show the variation of the main physiological integrated five copies of the expression cassette. Thus, variables at different D in both strains. it was selected for further studies in which it was com- As can be seen in Fig. 1A, the specific substrate pared with a clone with a single copy of gene of interest. uptake rate (qS) increased linearly with increasing l. The determination of gene dosage was performed by Interestingly, there were no significant differences in qS ddPCR. This method allows to determine the exact num- between clones, which suggests that this rate was unaf- ber of expression cassettes that were integrated in multi- fected by gene dosage. On the other hand, the overall copy clone (MCC), five copies, and to confirm the biomass substrate yield (YX/S) remained fairly constant presence of only one copy of CRL1 gene in the single- with a value of about 0.5 gX gS–1, which is consistent copy clone (SCC). with most reported values (Garcia-Ortega et al., 2013; ß alik et al., 2015; Adelantado et al., 2017). However, C MCC exhibited a slight decrease in YX/S at the lowest l Strain characterization in chemostat cultures values (0.05 and 0.025 h-1) possibly due to requirement Cell growth. The impact of gene dosage on clone of more energy for maintenance (ms) than the single- production kinetics was assessed with two sets of copy strain (0.029 vs. 0.0023 gS gX–1 h–1). This factor chemostat cultures grown at different dilution rates (D). strongly influenced YX/S, which is consistent with some The specific growth rates spanned the range of 0.025– previous studies where biomass production decreased 0.15 h 1. The carbon and electron balances were verified with decreasing l (Rebnegger et al., 2016). Fig. 1. Main physiological parameters for continuous cultures of multicopy (black symbols) and single-copy clones (white symbols). A. (●, ○) specific glucose uptake rate (qS); (■, □) overall biomass-to-substrate yield YX/S. B. (♦, ◊) specific oxygen uptake rate (qO2 ); (▲, D) specific carbon dioxide production rate (qCO2 ); and (●,○) respiratory quotient (RQ). Error bars represent SE of the mean values. ª 2019 The Authors. Microbial Biotechnology published by John Wiley & Sons Ltd and Society for Applied Microbiology., Microbial Biotechnology, 13, 315–327 318 M. A. Nieto-Taype et al. The lower values observed in maintenance coefficient functional CRL1 copies are integrated on MCC, it exerts a could be expected since K. phaffii has been described relevant demand of resources in comparison with SCC. as a robust system in terms of present lower mainte- On the other hand, if it is compared qO2 and qCO2 nance requirements over other alternative platforms such trend for SCC and MCC, both display similar values as E. coli (Zhu et al., 2019). However, in this work, a rel- across the l until 0.10 h-1. Nevertheless, at higher l’s, a evant difference on mS has been described between relatively slight decrease is detected for MCC in front of SCC and MCC, specifically the difference is about one SCC. It has to be considered that RPP consumes ener- order of magnitude (0.0023 and 0.029 respectively). This getic resources that drain precursors from the central notable change could be related to the RPP. It exerts a carbon metabolism to sustain the productivity, which strong effect on metabolic fluxes that often leads to an probably results in a readjustment of metabolism, being increase in the maintenance requirements (Carnicer more inefficient (Pen ~a et al., 2018). For the MCC, it can et al., 2012; Moser et al., 2017). be hypothesized that at low l, Crl1 synthesis does not Accordingly, when comparing MCC respect to SCC in produce any significant metabolic readjustment; there- relative terms, the cell maintenance requirements con- fore, not significant changes are observed on gas-related sume a higher proportion of the overall energy resources specific rates. However, at higher l this readjustment is obtained from carbon source uptake. shown as a reduction in the oxygen consumption and In Fig. 1B is shown how the specific O2 uptake (qO2 ) carbon dioxide production rates. and CO2 production (qCO2 ) rates increased linearly with l. Slight differences between both strains were observed Target protein production. The specific product at high l. However, the proportion between these two generation rate (qP) and the overall product-to-biomass specific rates are constant, and consequently, the respi- yield (YP/X) were evaluated as main key production ratory quotient (RQ) was always about 1.15. parameters. Both are shown in Fig. 2. Similarly to qS, qO2 and qCO2 showed also to be Most of the studies involving PGAP have shown qP to strongly coupled with l, therefore, fit into a linear equa- increase linearly with increasing µ (Khasa et al., 2007; tion pattern (Herbert, Pirt, Luedeking-Piret), which Rebnegger et al., 2014; Garcia-Ortega et al., 2016). In describes how is distributed a determined bioprocess the present work, similar behaviour was observed for the parameter (specific rate) for cell growth and mainte- SCC. In the previous studies, heterologous protein nance. expression was strongly coupled with cell growth. With Regarding the maintenance coefficient, which is repre- the methanol-inducible promoter, PAOX1, qP may not sented by the intercept, for both specific rates of each increase with increase in µ and substrate inhibition may clone, a rather similar value was obtained. Like qS arise as a result (Cos et al., 2006; Ahmad et al., 2014; (Fig. 1A), MCC maintenance coefficient was slightly Schwarzhans et al., 2017; Ponte et al., 2018). higher, suggesting that this difference is produced due to As expected, qP was greater with MCC than with SCC the gene dosage effect. Thus, considering that five (roughly 3–6 times). Some studies have shown protein Fig. 2. Production parameters for chemostat cultivation at different dilution rates of the multicopy (black symbols) and single-copy clones (white symbols). (■, □) Specific product generation rate (qP); (●, ○) overall product-to-biomass yield coefficient (YP/X). Error bars represent SE of the mean values. ª 2019 The Authors. Microbial Biotechnology published by John Wiley & Sons Ltd and Society for Applied Microbiology., Microbial Biotechnology, 13, 315–327 Kinetics in K. phaffii bioprocess design 319 production to be correlated with gene dosage and hence Protein production kinetics reflects the equilibrium suggest that the number of gene copies influences qP between the various steps involved in the synthesis, up an optimum number above which the synergistic folding and secretion of proteins, which are influenced effect usually is lost (Schwarzhans et al., 2016a; Betan- by a variety of physiological factors. Based on the cur et al., 2017; Dagar and Khasa, 2018; Vogl et al., results for MCC, one may hypothesize that some of the 2018a). processes involved in protein production are subject to a Unlike SCC, the variation of qP for MCC with µ was bottleneck in the high µ range that leads to the satura- not linear. In fact, a saturation effect was observed at tion effect observed in the production kinetic profile. This µ > 0.10 h–1, from which the qP reaches a rather con- result is consistent with previous reports where the stant value at about 390 UAgX 1 h 1. As a result, the expression system seemingly saturated during transcrip- proportional difference in qP between the two clones tion, protein synthesis, post-translational modifications or decreased with increasing µ (from roughly six times in even secretion (Puxbaum et al., 2015; Ca mara et al., the low µ range to only about three times in the high µ 2017). Further research is therefore required to improve range). existing knowledge about the events. In this work, we During the chemostat cultures, the stability of the focused the efforts on the transcriptional analysis of key strains has been demonstrated by confirming the gene genes. copy number of the expression cassette genome integra- tion by ddPCR. Other works also have evaluated strain Transcriptional analysis stability for K. phaffii strains in long-run chemostats describing a high genetic stability of the recombinant As suggested in the previous section, further research strains (Cankorur-Cetinkaya et al., 2018). into subjects such as transcriptional analysis was Looser et al. (2015) described that the strains deemed necessary to throw further light onto the differ- expressing heterologous proteins under PGAP normally ences in qP variation patterns between SC and MC show a linear pattern of production kinetics. Therefore, clones in chemostat cultures. For this purpose, relative the optimum µ value for production purposes must be transcript levels in relevant target genes were deter- close to lmax. However, in this work qP for MCC results mined under all the culture conditions studied in chemo- exhibited saturation at specific growth rates lower than stat cultures, using qPCR as described in the lmax, so identifying the optimum µ value should not be experimental procedures section. straightforward, which was also suggested by other Figure 3 compares the correlation of qP with the rela- authors (Maurer et al., 2006; Buchetics et al., 2011). tive transcript levels of the genes CRL1 and TDH3. Ini- As shown in Fig. 2, the variation of YP/X with µ differed tially, TDH3 transcription increased linearly with between SC and MC clones. While the variation pattern increasing D in both clones regardless the CRL1 gene for YP/X in SCC was quite constant irrespective of D, YP/X dosage. This result can be ascribed to the TDH3 product in MCC exhibited a substantial decrease in the higher corresponding to a critical node in the glucose uptake D range. Thus, the highest values of YP/X were obtained pathway, which is closely associated with biomass for low specific growth rates (0.025 and 0.05 h–1). growth (Nocon et al., 2014; C ß alik et al., 2015). Fig. 3. Specific product generation rate, qP, of the single-copy (□) and multicopy clones (■); and relative transcript level of the clones for CRL1 (black bars) and TDH3 (grey bars) at each dilution rate. Error bars represent SE of the mean values. ª 2019 The Authors. Microbial Biotechnology published by John Wiley & Sons Ltd and Society for Applied Microbiology., Microbial Biotechnology, 13, 315–327 320 M. A. Nieto-Taype et al. Consequently, with high specific biomass growth rates, above-mentioned trends. Interestingly, while no effect high fluxes through the glycolytic pathway can only be was observed for the TDH3 quotient, the positive effect maintained by increasing production in TDH3. These of CRL1 gene dosage in terms of transcription is con- results would be in contrast with those microarrays firmed for all the culture conditions tested. According to results presented by Rebnegger et al. (2014) in which these results, therefore, no limitation in the transcrip- the TDH3 transcription appeared to be unregulated with tional machinery for CRL1 can be stated. Nevertheless, the dilution rate. Remarkably, it is considered that qPCR and despite the CRL1 quotient of transcript levels is thought to provide increased quantitative resolution in remains constant regardless the different cultures condi- transcriptional analyses of genes that present slight dif- tions at different D, there is a significant decrease in the ferences among samples (Morey et al., 2006). Finally, qP quotient when increasing µ. This would support the on constancy of dilution rate (D), TDH3 relative transcript hypothesis that the relevant increases achieved in terms levels were essentially identical for the two strains, which of transcript levels due to the higher gene dosage of the suggest that the levels were influenced by D but not by target gene cannot be therefore converted into functional CRL1 gene dosage. protein of interest. Consistent with the results of the enzymatic activity Strikingly, the gene dosage effect observed in this analysis, CRL1 gene expression was initially greater in PGAP-driven example of protein expression differs mark- MC than in SC clones. In SCC, CRL1 relative transcript edly from that reported by Ca mara et al. (2017) for levels increased linearly with increasing D, which sug- PAOX1-driven Rhizopus oryzae lipase (Rol) recombinant gests coupling of µ, CRL1 relative transcript levels and expression. The increased number of cassettes in Rol qP. These results indicate that PGAP regulates genes attenuated transcription of methanol metabolism and TDH3 and CRL1 similarly and hence that the effects are decreased methanol consumption, cell growth and strongly related with cell growth. This was not the case recombinant protein production as a result. with MCC, however. Thus, although CRL1 gene expres- Because K. phaffii may be limited in terms of protein sion was seemingly not affected by D, qP clearly satu- folding, glycosylation and secretion, reduced product rated at high D levels. This may have resulted from yields and productivity could be expected. These limita- CRL1 expression coupling with biomass growth and the tions were widely discussed by Puxbaum et al. (2015) total amount of product secreted being limited. There- and are related to ER-associated degradation (ERAD), fore, there might be a bottleneck after transcription pre- ER-Golgi trafficking in the secretory pathway and cluding conversion of all CRL1 transcripts into proper unfolded protein response (UPR). There have been folded and secreted Crl1. some attempts at circumventing these constraints and In Fig. 4, in which the transcriptional regulation of enhancing production of various recombinant proteins. TDH3 and CRL1 for different µ is compared by means For instance, some authors have used increased gene of plotting the quotient of the relative transcript levels dosages of target genes involved in product folding and between the MCC and the SCC, can be observed the secretion such as HAC1, PDI1 and/or KAR2 (Bankefa Fig. 4. Product and relative transcript-level ratios between multicopy and single-copy clones at different dilution rates in chemostat cultures. (■) Ratio of specific product generation rates (qP); (●) ratio of relative transcript level for CRL1; and (○) ratio of relative transcript level for TDH3. Error bars represent SE of the mean values. ª 2019 The Authors. Microbial Biotechnology published by John Wiley & Sons Ltd and Society for Applied Microbiology., Microbial Biotechnology, 13, 315–327 Kinetics in K. phaffii bioprocess design 321 –1 et al., 2018; Guan et al., 2016; Liu et al., 2014; Yang biomass concentration about 100 g l to compare the et al., 2016). Other authors (Barrero et.al., 2018) have experiments with a similar biomass concentration and engineered the secretion signal in order to improve always below maximal working volume. Thus, the culture translocation into the endoplasmic reticulum (ER). is stopped before the bioprocess may be limited due to No limitation in TDH3 or CRL1 gene transcription some biological and physical restrictions, such as heat was observed here. With MCC, however, CRL1 relative and mass transfer, and lack of homogeneity, keeping the transcript levels increased linearly with increasing D, pseudo-stationary state in the system. l spanned the but the transcripts could not be converted into func- range of 0.05–0.15 h–1 and included an additional condi- tional secreted Crl1 owing to potential post-transcrip- tion of 0.025 h–1 for MCC. tional constraints. In order to shed further light on the Figure 5 presents the time-course of total lipolytic presumed bottlenecks, the relative expression of other activity in the fed-batch cultures with SC and MC clones. target genes including PGK, KAR2 and HAC1 was Although the figure only shows the results for the fed examined. No limitation in chemostat cultures was phase, it should be noted that lipolytic activity at the end observed at any D value, so we can conclude that the of the batch phase was roughly 3.5 times higher with specific growth rate had no regulatory effect on the MC clone. Since biomass grew at the highest possible µ expression of these genes. Overall, this result reveals during the batch phase, these results seemingly confirm that, although no transcriptional limitation was identified the effect of gene dosage on high µ cultures suggested – and hence CRL1 gene transcription in MCC in the previous section. increased linearly with D – this trend was inconsistent As detailed in Table 1, lipolytic activity peaked at 372 with that in the amount of product formed and hence UA ml–1 at the highest µ value with SCC and at 1009 suggests the presence of post-transcriptional con- UA ml–1 at low µ (0.05 h–1) with MCC. straints. Based on these results for MCC, an additional fermen- tation run at 0.025 h–1 was used to confirm that Crl1 pro- duction would peak at the lowest l level. As Fed-batch cultures hypothesized, lipolytic activity (1414 UA ml–1) was high- In addition to the chemostat cultures, the performance of est under those conditions and four times greater than SC and MC clones was also examined in fed-batch cul- the highest value for SCC. A systematic comparison tures, which is the operational mode typically used for based on the main production parameters obtained at industrial RPP. All fed-batch fermentations were con- the different operating conditions followed with the two ducted according to a carbon-limited pre-programmed producer clones is presented in Table 1. exponential feeding profile in order to maintain a con- In Fig. 6, the mean values of qP and YP/X for the fed- stant µ throughout the cultivation period. The ending cri- batch cultures at different µ values are depicted. As teria selected for these cultures were to reach a final expected for consistency with the chemostat results, qP Fig. 5. Variation of the total lipolytic activity in fed-batch cultures at different specific growth rates (µ) with the multicopy (black symbols) and sin- gle-copy clones (white symbols). Specific growth rates: (●) 0.025 h–1, (▲) 0.05 h–1, (♦) 0.10 h–1, (■) 0.15 h–1, (D) 0.05 h–1, (◊) 0.10 h–1 and (□) 0.15 h–1. ª 2019 The Authors. Microbial Biotechnology published by John Wiley & Sons Ltd and Society for Applied Microbiology., Microbial Biotechnology, 13, 315–327 322 M. A. Nieto-Taype et al. Table 1. Main production parameters obtained with the single-copy and multicopy clones at different specific growth rates (µ) in fed-batch cul- tures. For the comparing rows, cultivation at the lowest µ of the single-copy clone is taken as reference. Single-copy clone Multicopy clone Nominal µ(h 1) 0.050 0.100 0.150 0.025 0.050 0.100 0.150 Experimental µ(h 1) 0.048 0.088 0.133 0.023 0.051 0.087 0.147 Product titre(UA ml 1) 180 195 372 1414 1009 827 600 Product titreincrease (%) +8% +107% +685% +460% +359% +233% qP(UA gX 1 h 1) 77 149 292 374 550 697 777 qP increase(%) +92% +277% +383% +611% +800% +903% QP(103 UA l 1 h 1) 3.3 6.3 22.9 14.6 20.1 31.1 34.9 QP increase(%) +87% +586% +337% +502% +831% +943% YP/S(103 UA gS 1) 0.75 0.71 0.94 7.33 5.54 3.72 2.61 YP/S increase(%) 5% +26% +880% +640% +398% +248% YP/X(103 UA gX 1) 1.61 1.70 2.19 16.1 10.9 8.02 5.28 YP/X increase(%) - +6% +36% +902% +577% +399% +229% µ, specific growth rate; qP, specific product generation rate; QP, volumetric productivity; YP/S, overall product-to-substrate yield; YP/X, overall pro- duct-to-biomass yield. Fig. 6. Production parameters for fed-batch cultures at different specific growth rates of the multicopy (black symbols) and single-copy clones (white symbols). (■,) Specific product generation rate (qP); (●, ○) overall product-to-biomass yield coefficient (YP/X). Error bars show SE from regression analysis. increased linearly with increasing µ in SCC but exhibited be extended to increase product titres until the termina- saturation in MCC. The YP/X variation pattern was similar tion criterion is fulfilled. to that for the chemostat cultures; thus, YP/X decreased In Table 1 are gathered the main production parame- exponentially with increasing µ in MCC, but remained ters obtained in fed-batch cultures for the single-copy fairly constant in SCC. and multicopy clones. Product titre, yields and productivi- As in the chemostat cultures, qP was higher with MCC ties are detailed as key performance indexes along the than with SCC, and their ratio (3–6) was dependent on different specific growth rates applied. Some aspects the µ applied. Although the overall patterns for the two can be assessed for discussion, mainly relating to the operational modes were similar, qP and YP/X were sub- synergistic effect of µ and qP, and how these two speci- stantially higher for the fed-batch cultures. fic rates affect the bioprocess efficiency. Fed-batch cultivation usually ends when biomass pro- Regarding the production kinetics, that is, how affects duction reaches a critical concentration limit imposed by l on qP, it was demonstrated that the kinetic pattern is biological and physical restrictions. As a result, growth- markedly different for the MCC and SCC. The MCC associated biomass in fermentation processes is an exhibited a 30% qP decrease when l decreased from unavoidable by-product and requires controlling the bio- 0.15 to 0.05 h–1, by effect of its rather flat production mass growth in order not to exceed the limiting concen- kinetic curve at the highest l, as shown in Fig. 6 and tration. Thus, the lower l is, the longer a bioprocess can Table 1. In a similar comparison for the SCC, qP ª 2019 The Authors. Microbial Biotechnology published by John Wiley & Sons Ltd and Society for Applied Microbiology., Microbial Biotechnology, 13, 315–327 Kinetics in K. phaffii bioprocess design 323 –1 decreased by 70 % from 0.15 to 0.05 h . It is due to other hand, if the impact of µ is weak (e.g. with MCC), the more marked coupling of qP to the cell growth. low-to-moderate µ values can help to extend the biopro- As expected, productivities (QP) varied on µ similarly cess time until the limit imposed by a critical amount of to that observed for qP in MCC and SCC. Consequently, biomass is reached. In this case, these differences have the rational selection of µ to optimize the bioprocess per- been caused by the different gene dosage. formance would require a trade-off. Summing up, it can be stated as main outcome of this In terms of parameters focused on production and contribution that the production kinetics of the cell factory yield, the SCC at 0.15 h–1 reached product titres 107 % and specific growth rate must be jointly considered to higher compared with the obtained at 0.05 h–1. Accord- tailor operating strategies to production clones for maxi- ingly, YP/S and YP/X increased a 26% and 36% respec- mizing expression in K. phaffii. This work presents how tively. Therefore, high µ should be clearly recommended a comprehensively elucidation of the gene dosage effect for production kinetic patterns similarly to SCC. on production kinetics towards an overall optimization of On the contrary, for the MCC, product titre reached a the RPP bioprocesses can be afforded. Hence, properly 68 % increase when l decreased from 0.15 to 0.05 h–1. understanding these features and their correlation pro- Strikingly, titre was even better at 0.025 h–1, equivalent vides a more rational, robust knowledge of the behaviour to a 140 % increase compared with 0.15 h–1, which lead of cell factories and allows bioprocesses to be engi- to important increases in terms of YP/S and YP/X (180% neered for greater efficiency. and 204% respectively). Experimental procedures Rational-based bioprocess design Strains The results previously exposed support the idea that Two recombinant clones of K. phaffii X-33 from Invitro- combined effect of µ and gene dosage on PGAP-based gen (Carlsbad, CA, USA) expressing CRL1 regulated by K. phaffii clones expressing CRL1 can be used to design the GAP promoter were obtained and used. A chimeric optimal operating strategies. vector assembled with the restriction–ligation method In this way, the results obtained from chemostat cul- was constructed by using commercial pGAPZaA plasmid tures reveal that qP increases linearly with increasing from Invitrogen and a codon-optimized synthetic CRL1 specific growth rate in SCC but exhibits a non-linear pat- coding sequence from GeneScript (Piscataway, NJ, tern suggestive of saturation of the production kinetics in USA). Different amounts of plasmid were transformed by MCC. Interestingly, qP is considerably higher with MCC electroporation in order to obtain clones with different than with SCC (3–6 times depending on the particular µ numbers of integrated expression cassettes. However, value). These results confirm that gene dosage impacts only the clones containing 1 or 5 gene copies were directly on the production kinetic profile, both in the levels used. Both clones can secrete Crl1 to the extracellular of titre achieved as well as the profile pattern. From the medium through the Saccharomyces cerevisiae a-mating transcriptional analysis performed, one can hypothesize factor signal sequence. the presence of a potential bottleneck after transcription. The continuous cultures were used as tool to the Gene copy number design of optimized fed-batch strategies, since this oper- ational mode is the most widely operational mode used Droplet digital PCR (ddPCR) was used to determine the for industrial RPP. Therefore, a rational-based biopro- number of cassette integrations in both clones, using a cess design can be carried out considering the effect of mara et al. slightly modified version of the method of Ca gene dosage and l on the bioprocess efficiency. (2016). Actin gene was used as housekeeping agent, Although trends were similar as those observed in the but the primer sequence was that proposed by Landes chemostat cultures, there were some differences, espe- et.al. (2016). The specific primers used are shown in cially with MCC. Saturation was more marked than in Table S1. the chemostat tests, and a plateau was reached in the higher µ range. Transcriptional analysis When the data for the main production parameters in Table 1 are jointly considered, the different trends in pro- Transcriptional analyses were only done on chemostat duction kinetics and yields can be useful to identify the cultures, where steady-state conditions ensured homo- best operational strategies by characterizing producer geneity in cell population. clones. Usually, when µ has a strong impact on specific production rates and yields (e.g. with SCC), a high µ Total RNA extraction. Samples of 1 ml were withdrawn level should be used to increase qP and YP/X. On the under different chemostat conditions and centrifuged at ª 2019 The Authors. Microbial Biotechnology published by John Wiley & Sons Ltd and Society for Applied Microbiology., Microbial Biotechnology, 13, 315–327 324 M. A. Nieto-Taype et al. 4 C at maximum speed for 2 min. The resulting pellets during the culture reaching a pseudo-stationary state. were resuspended in 1 ml of TRIzolTM (Waltham, MA, The process is described in detail elsewhere (Garcia- USA) and 200-mg glass beads. Total RNA was Ortega et al., 2013). extracted as per the manufacturer’s instructions. Then, RNA integrity and concentration were checked by Analytical methods agarose electrophoresis and Nanodrop analysis (Thermo Determination of biomass as dry cell weight ScientificTM, Waltham, MA, USA) respectively. (DCW). Biomass concentrations were measured in triplicate in terms of DCW as described elsewhere (Cos cDNA synthesis and transcriptional levels. cDNA was et al., 2005). The relative standard deviation (RSD) was synthetized by using the iScriptTM cDNA Synthesis Kit about 3%. from Bio-Rad (Hercules, CA, USA) according to the manufacturer’s instructions. A set of primers was Quantification of the carbon source and by- designed to amplify cDNA for the target genes by qPCR. products. The concentrations of the different carbon The set comprised CRL1 (heterologous product); TDH3, sources used in the batch (glycerol) and fed-batch tests which is the gene natively expressed under PGAP control (glucose), and of potential fermentation by-products (glycolytic); and PGK gene, phosphoglycerate kinase (arabitol or ethanol), were all determined by HPLC. The (glycolytic). Additional genes involved in the unfolded column and procedure used for this purpose are protein response (UPR) and the secretory mechanisms described elsewhere (Garcia-Ortega et al., 2017). RSD such as KAR2 and HAC1 were also studied. was always < 1%. Transcription was assessed by qPCR amplification. For maximum accuracy, mixes were made by the robot Off-gas analyses. BlueInOne Cell gas analysers were EpMotion (Eppendorf, Germany). SYBRTM Select used to monitor the cultures exhaust gas (BlueSens, Master Mix was used as polymerase mix and a Herten, Germany). CO2 and O2 mole fractions were QuantStudio 12K Flex Real-Timer from Thermo measured on-line with provision for off-gas pressure and ScientificTM for amplification cycles and data acquisition. humidity. The data thus obtained were used to estimate The qPCR programme was implemented as pre- the oxygen uptake rate (OUR), carbon dioxide evolution scribed by the manufacturer but using a primer anneal- rate (CER), specific rates (qO2 and qCO2 ) and the ing temperature of 57.4°C. Relative transcript level was respiratory quotient (RQ). RSD was < 5% in all cases. determined by using the MTH1 glucose-responsive tran- scriptional factors, which code for a negative regulator of Lipolytic activity assay. Crl1 activity was determined by the glucose-sensing signal transduction pathway, as using a modified version of an existing enzymatic assay housekeeping agents. Rebnegger et al. (2014) previ- based on p-nitrophenyl butyrate (pNPB) (Chang et al., ously found the specific growth rate to be uninfluential. 2006). The reaction buffer consisted of 1 mM pNPB, 50 mM phosphate buffer at pH 7 and 4 (v/v)% acetone. Cultivation methods A volume of 980 µl of buffer was mixed with 20 µl of sample. The absorbance at 348 nm was measured on- Inoculum cultures for the bioreactor tests were prepared line at 30°C for 2 min on a Specord 200 Plus instrument according to Garcia-Ortega et al. (2013). from Analytic Jena (Jena, Germany). One unit of activity was defined as the amount of enzyme needed to Chemostat cultivation. Chemostat cultures of the two release 1 mmol of p-nitrophenol per minute under assay strains were prepared in duplicate as described conditions. RSD was < 1%. elsewhere (Garcia-Ortega et al., 2016). Different specific growth rates from 0.025 to 0.15 h–1 were evaluated. For Process parameters every dilution rate, the continuous cultures were carried out for at least five residence times. In order to ensure Mass balance and stoichiometric equations. All that steady state was achieved, several samples were equations derived from the mass balances used to taken and analysed since three residence times during calculate yields and rates in the chemostat (Garcia- three consecutive residence times up to confirming the Ortega et al., 2016) and fed-batch cultures (Ponte et al., stability of the studied parameters. 2016) were described elsewhere. Fed-batch cultivation. Both strains were also cultivated Data consistency and reconciliation. Measurement in the fed-batch mode, at different specific growth rates consistency was checked by using the standard test with from 0.025 to 0.15 h–1. This strategy is based on a carbon and electron balances as constraints. Both on- carbon-limited feeding profile, keeping a constant µ line and off-line measurements allowed five key specific ª 2019 The Authors. Microbial Biotechnology published by John Wiley & Sons Ltd and Society for Applied Microbiology., Microbial Biotechnology, 13, 315–327 Kinetics in K. phaffii bioprocess design 325 rates in the black-box process model to be calculated, Burgard, J., Valli, M., Graf, A.B., Gasser, B., and Mat- namely biomass generation (µ), glucose uptake (qS), tanovich, D. (2017) Biomarkers allow detection of nutrient product generation (qP), oxygen uptake (qO2 ) and carbon limitations and respective supplementation for elimination in Pichia pastoris fed-batch cultures. Microb Cell Fact 16: dioxide production (qCO2 ). The methodology used is 117. described in detail in a previous paper (Ponte et al., € Gu ß alık, P., Ata, O., C €nesß, H., Massahi, A., Boy, E., Keskin, 2016). A., et al. (2015) Recombinant protein production in Pichia pastoris under glyceraldehyde-3-phosphate dehydroge- nase promoter: From carbon source metabolism to biore- Acknowledgements actor operation parameters. Biochem Eng J 95: 20–36. This work was funded by MINECO and FEDER under Camara, E., Albiol, J., and Ferrer, P. (2016) Droplet digital PCR-aided screening and characterization of Pichia pas- Project CTQ2016-74959-R. The authors’ group is mem- toris multiple gene copy strains. Biotechnol Bioeng 113: ber of 2017-SGR-1462 and the Reference Network in 1542–1551. Biotechnology (XRB) of Generalitat de Catalunya. M.A. Camara, E., Landes, N., Albiol, J., Gasser, B., Mattanovich, Nieto-Taype acknowledges award by the National Coun- D., and Ferrer, P. (2017) Increased dosage of AOX1 pro- cil of Science, Technology and Technological Innovation moter-regulated expression cassettes leads to transcrip- (CONCYTEC) through its executing unit, the National tion attenuation of the methanol metabolism in Pichia Fund for Scientific, Technological and Technological pastoris. Sci Rep 7: 44302. s- Cankorur-Cetinkaya, A., Narraidoo, N., Kasavi, C., Slater, Innovation Development (FONDECYT) and J. Garrigo N.K.H., Archer, D.B., and Oliver, S.G. (2018) Process Martınez, a PIF scholarship from the Universitat development for the continuous production of heterolo- Autonoma de Barcelona. gous proteins by the industrial yeast, Komagataella phaf- fii. Biotechnol Bioeng 115: 2962–2973. Carnicer, M., Ten Pierick, A., Van Dam, J., Heijnen, J.J., Conflict of interest Albiol, J., Van Gulik, W., and Ferrer, P. (2012) Quantita- tive metabolomics analysis of amino acid metabolism in None declared. recombinant Pichia pastoris under different oxygen avail- ability conditions. Microb Cell Fact 11: 83. References Chang, S.W., Lee, G.C., and Shaw, J.F. (2006) Codon opti- mization of Candida rugosa lip1 gene for improving Adelantado, N., Tarazona, P., Grillitsch, K., Garcıa-Ortega, expression in Pichia pastoris and biochemical characteri- X., Monforte, S., Valero, F., et al. (2017) The effect of zation of the purified recombinant LIP1 lipase. J Agric hypoxia on the lipidome of recombinant Pichia pastoris. Food Chem 54: 815–822. Microb Cell Fact 16: 86. Cos, O., Serrano, A., Montesinos, J.L., Ferrer, P., Cregg, Ahmad, M., Hirz, M., Pichler, H., and Schwab, H. (2014) J.M., and Valero, F. (2005) Combined effect of the metha- Protein expression in Pichia pastoris: recent achieve- nol utilization (Mut) phenotype and gene dosage on ments and perspectives for heterologous protein produc- recombinant protein production in Pichia pastoris fed- tion. Appl Microbiol Biotechnol 98: 5301–5317. batch cultures. J Biotechnol 117: 321–335. Bankefa, O.E., Wang, M., Zhu, T., and Li, Y. (2018) Hac1p Cos, O., Ramo n, R., Montesinos, J.L., and Valero, F. (2006) homologues from higher eukaryotes can improve the Operational strategies, monitoring and control of heterolo- secretion of heterologous proteins in the yeast Pichia pas- gous protein production in the methylotrophic yeast Pichia toris. Biotechnol Lett 40: 1149–1156. pastoris under different promoters: a review. Microb Cell Barrero, J.J., Casler, J.C., Valero, F., Ferrer, P., and Glick, Fact 5: 17. B.S. (2018) An improved secretion signal enhances the Dagar, V.K., and Khasa, Y.P. (2018) Combined effect of secretion of model proteins from Pichia pastoris. Microb gene dosage and process optimization strategies on high- Cell Fact 17: 161. level production of recombinant human interleukin-3 (hIL- Barrigon, J.M., Montesinos, J.L., and Valero, F. (2013) 3) in Pichia pastoris fed-batch culture. Int J Biol Macromol Searching the best operational strategies for Rhizopus 108: 999–1009. oryzae lipase production in Pichia pastoris Mut+ pheno- Dewan, S. S. (2017) Global markets for enzymes in indus- type: Methanol limited or methanol non-limited fed-batch trial applications. BCC Res (BIO030J). cultures? Biochem Eng J 75: 47–54. Ferrer, P., Montesinos, J.L., Valero, F., and Sola , C. (2001) Betancur, M.O., Reis, V.C.B., Nicola, A.M., De Marco, Production of native and recombinant lipases by Candida J.L., de Moraes, L.M.P., and Torres, F.A.G. (2017) Mul- rugosa: a review. Appl Biochem Biotechnol 95: 221–256. ticopy plasmid integration in Komagataella phaffii medi- Garcia-Ortega, X., Ferrer, P., Montesinos, J.L., and Valero, ated by a defective auxotrophic marker. Microb Cell F. (2013) Fed-batch operational strategies for recombi- Fact 16: 99. nant Fab production with Pichia pastoris using the consti- Buchetics, M., Dragosits, M., Maurer, M., Rebnegger, C., tutive GAP promoter. Biochem Eng J 79: 172–181. Porro, D., Sauer, M., et al. (2011) Reverse engineering of Garcia-Ortega, X., Adelantado, N., Ferrer, P., Montesinos, protein secretion by uncoupling of cell cycle phases from J.L., and Valero, F. (2016) A step forward to improve growth. Biotechnol Bioeng 108: 2403–2412. recombinant protein production in Pichia pastoris: from ª 2019 The Authors. Microbial Biotechnology published by John Wiley & Sons Ltd and Society for Applied Microbiology., Microbial Biotechnology, 13, 315–327 326 M. A. Nieto-Taype et al. specific growth rate effect on protein secretion to carbon- Maurer, M., Ku €hleitner, M., Gasser, B., and Mattanovich, D. starving conditions as advanced strategy. Process Bio- (2006) Versatile modeling and optimization of fed batch chem 51: 681–691. processes for the production of secreted heterologous Garcia-Ortega, X., Valero, F., and Montesinos-Seguı, J.L. proteins with Pichia pastoris. Microb Cell Fact 5: 37. (2017) Physiological state as transferable operating crite- Morey, J.S., Ryan, J.C., and Dolah, F.M. (2006) Microarray rion to improve recombinant protein production in Pichia validation: factors influencing correlation between oligonu- pastoris through oxygen limitation. J Chem Technol cleotide microarrays and real-time PCR. Biol Proced Biotechnol 92: 2573–2582. Online 8: 175–193. Garcıa-Ortega, X., Ca mara, E., Ferrer, P., Albiol, J., Moser, J.W., Prielhofer, R., Gerner, S.M., Graf, A.B., Wilson, Montesinos-Seguı, J.L., and Valero, F. (2019) Rational I.B.H., Mattanovich, D., and Dragosits, M. (2017) Implica- development of bioprocess engineering strategies for tions of evolutionary engineering for growth and recombi- recombinant protein production in Pichia pastoris (Koma- nant protein production in methanol-based growth media gataella phaffii) using the methanol-free GAP promoter. in the yeast Pichia pastoris. Microb Cell Fact 16: 49. Where do we stand? N Biotechnol 53: 24–34. Nocon, J., Steiger, M.G., Pfeffer, M., Sohn, S.B., Kim, T.Y., Guan, B., Chen, F., Su, S., Duan, Z., Chen, Y., Li, H., and Maurer, M., et al. (2014) Model based engineering of Jin, J. (2016) Effects of co-overexpression of secretion Pichia pastoris central metabolism enhances recombinant helper factors on the secretion of a HSA fusion protein protein production. Metab Eng 24: 129–138. (IL2-HSA) in Pichia pastoris. Yeast 33: 587–600. Norde n, K., Agemark, M., Danielson, J.A.H., Alexandersson, Hartner, F.S., and Glieder, A. (2006) Regulation of methanol E., Kjellbom, P., and Johanson, U. (2011) Increasing utilisation pathway genes in yeasts. Microb Cell Fact 5: 39. gene dosage greatly enhances recombinant expression of Highsmith, J. (2015) Biological Therapeutic Drugs: Tech- aquaporins in Pichia pastoris. BMC Biotechnol 11: 47. nologies and Global Markets. BCC Res. ( BIO079C). Pen~a, D.A., Gasser, B., Zanghellini, J., Steiger, M.G., and Jiao, L., Zhou, Q., Su, Z., Xu, L., and Yan, Y. (2018) High- Mattanovich, D. (2018) Metabolic engineering of Pichia level extracellular production of Rhizopus oryzae lipase in pastoris. Metab Eng 50: 2–15. Pichia pastoris via a strategy combining optimization of Ponte, X., Montesinos-Seguı, J.L., and Valero, F. (2016) gene-copy number with co-expression of ERAD-related Bioprocess efficiency in Rhizopus oryzae lipase produc- proteins. Protein Expr Purif 147: 1–12. tion by Pichia pastoris under the control of PAOX1 is oxy- Juturu, V., and Wu, J.C. (2018) Heterologous protein gen tension dependent. Process Biochem 51: 1954– expression in Pichia pastoris : latest research progress 1963. and applications. ChemBioChem 19: 7–21. Ponte, X., Barrigo n, J.M., Maurer, M., Mattanovich, D., Ken Ugo, A., Vivian Amara, A., CN, I., and Kenechuwku, Valero, F., and Montesinos-Seguı, J.L. (2018) Towards U., (2017) Microbial lipases: a prospect for biotechnologi- optimal substrate feeding for heterologous protein produc- cal industrial catalysis for green products: a review. Fer- tion in Pichia pastoris (Komagataella spp) fed-batch pro- ment Technol 6: 2. cesses under PAOX1 control: a modeling aided approach. Khasa, Y.P., Khushoo, A., Srivastava, L., and Mukherjee, J Chem Technol Biotechnol 93: 3208–3218. K.J. (2007) Kinetic studies of constitutive human granulo- Potvin, G., Ahmad, A., and Zhang, Z. (2012) Bioprocess cyte-macrophage colony stimulating factor (hGM-CSF) engineering aspects of heterologous protein production in expression in continuous culture of Pichia pastoris. Pichia pastoris: a review. Biochem Eng J 64: 91–105. Biotechnol Lett 29: 1903–1908. Prielhofer, R., Maurer, M., Klein, J., Wenger, J., Kiziak, C., Landes, N., Gasser, B., Vorauer-Uhl, K., Lhota, G., Mat- Gasser, B., and Mattanovich, D. (2013) Induction without tanovich, D., and Maurer, M. (2016) The vitamin-sensitive methanol: novel regulated promoters enable high-level promoter PTHI11 enables pre-defined autonomous induc- expression in Pichia pastoris. Microb Cell Fact 12: 5. tion of recombinant protein production in Pichia pastoris. Puxbaum, V., Mattanovich, D., and Gasser, B. (2015) Quo Biotechnol Bioeng 113: 2633–2643. vadis? The challenges of recombinant protein folding and Liang, S., Zou, C., Lin, Y., Zhang, X., and Ye, Y. (2013) secretion in Pichia pastoris. Appl Microbiol Biotechnol 99: Identification and characterization of PGCW14: a novel, 2925–38. strong constitutive promoter of Pichia pastoris. Biotechnol Rebnegger, C., Graf, A. B., Valli, M., Steiger, M.G., Gasser, Lett 35: 1865–1871. B., Maurer, M., and Mattanovich, D. (2014) In Pichia pas- Liu, H., Qin, Y., Huang, Y., Chen, Y., Cong, P., and He, Z. toris, growth rate regulates protein synthesis and secretion, (2014) Direct evaluation of the effect of gene dosage on mating and stress response. Biotechnol J 9: 511–525. secretion of protein from yeast Pichia pastoris by express- Rebnegger, C., Vos, T., Graf, A.B., Valli, M., Pronk, J.T., ing EGFP. J Microbiol Biotechnol 24: 144–151. Daran-Lapujade, P., and Mattanovich, D. (2016) Pichia Looser, V., Bruhlmann, B., Bumbak, F., Stenger, C., Costa, pastoris exhibits high viability and a low maintenance M., Camattari, A., et al. (2015) Cultivation strategies to energy requirement at near-zero specific growth rates. enhance productivity of Pichia pastoris: a review. Biotech- Appl Environ Microbiol 82: 4570–4583. nol Adv 33: 1177–1193. Robert, J.M., Garcia-Ortega, X., Montesinos-Seguı, J.L., Mattanovich, D., Graf, A., Stadlmann, J., Dragosits, M., Freire, D.M.G., and Valero, F. (2019) Continuous opera- Redl, A., Maurer, M., et al. (2009) Genome, secretome tion, a realistic alternative to fed-batch fermentation for and glucose transport highlight unique features of the the production of recombinant lipase B from Candida protein production host Pichia pastoris. Microb Cell Fact antarctica under the constitutive promoter PGK in Pichia 8: 29. pastoris. Biochem Eng J 147: 39–47. ª 2019 The Authors. Microbial Biotechnology published by John Wiley & Sons Ltd and Society for Applied Microbiology., Microbial Biotechnology, 13, 315–327 Kinetics in K. phaffii bioprocess design 327  nchez, A., Ferrer, P., Serrano, A., Pernas, M. A., Valero, Sa Weninger, A., Fischer, J.E., Raschmanova , H., Kniely, C., F., Rua, M.L., et al. (1999) Characterization of the lipase Vogl, T., and Glieder, A. (2018) Expanding the CRISPR/ and esterase multiple forms in an enzyme preparation Cas9 toolkit for Pichia pastoris with efficient donor integra- from a Candida rugosa pilot-plant scale fed-batch fermen- tion and alternative resistance markers. J Cell Biochem tation. Enzyme Microb Technol 25: 214–223. 119: 3183–3198. Schwarzhans, J.P., Wibberg, D., Winkler, A., Luttermann, Yang, Z., and Zhang, Z. (2018) Engineering strategies for T., Kalinowski, J., and Friehs, K. (2016a) Integration event enhanced production of protein and bio-products in Pichia induced changes in recombinant protein productivity in pastoris: a review. Biotechnol Adv 36: 182–195. Pichia pastoris discovered by whole genome sequencing Yang, J., Lu, Z., Chen, J., Chu, P., Cheng, Q., Liu, J., et al. and derived vector optimization. Microb Cell Fact 15: 84. (2016) Effect of cooperation of chaperones and gene Schwarzhans, J.P., Wibberg, D., Winkler, A., Luttermann, dosage on the expression of porcine PGLYRP-1 in Pichia T., Kalinowski, J., and Friehs, K. (2016b) Non-canonical pastoris. Appl Microbiol Biotechnol 100: 5453–5465. integration events in Pichia pastoris encountered during Zahrl, R.J., Pen~a, D.A., Mattanovich, D., and Gasser, B. standard transformation analysed with genome sequenc- (2017) Systems biotechnology for protein production in ing. Sci Rep 6: 38952. Pichia pastoris. FEMS Yeast Res 17: 7. Schwarzhans, J.P., Luttermann, T., Geier, M., Kalinowski, Zhang, A.L., Luo, J.X., Zhang, T.Y., Pan, Y.W., Tan, Y.H., J., and Friehs, K. (2017) Towards systems metabolic Fu, C.Y., et al. (2009) Recent advances on the GAP pro- engineering in Pichia pastoris. Biotechnol Adv 35: 681– moter derived expression system of Pichia pastoris. Mol 710. Biol Rep 36: 1611–1619. Shen, W., Xue, Y., Liu, Y., Kong, C., Wang, X., Huang, M., Zhao, W., Wang, J., Deng, R., and Wang, X. (2008) Scale- et al. (2016) A novel methanol-free Pichia pastoris system up fermentation of recombinant Candida rugosa lipase for recombinant protein expression. Microb Cell Fact 15: expressed in Pichia pastoris using the GAP promoter. J 178. Ind Microbiol Biotechnol 35: 189–195. Theron, C.W., Berrios, J., Delvigne, F., and Fickers, P. Zhu, T., Guo, M., Tang, Z., Zhang, M., Zhuang, Y., Chu, J., (2018) Integrating metabolic modeling and Komagataella and Zhang, S. (2009) Efficient generation of multi-copy (Pichia) pastoris. Appl Microbiol Biotechnol 102: 63–80. strains for optimizing secretory expression of porcine Valero, F. (2018) Recent advances in Pichia pastoris as insulin precursor in yeast Pichia pastoris. J Appl Microbiol host for heterologous expression system for lipases: a 107: 954–63. review. In: Methods in Molecular Biology. New York, NY: Zhu, T., Sun, H., Li, P., Xue, Y., Li, Y., and Ma, Y. (2014) Humana Press, pp. 205–216. Constitutive expression of alkaline b-mannanase in Vogl, T., and Glieder, A. (2013) Regulation of Pichia pas- recombinant Pichia pastoris. Process Biochem 49: 2025– toris promoters and its consequences for protein produc- 2029. tion. N Biotechnol 30: 385–404. Zhu, T., Sun, H., Wang, M., and Li, Y. (2019) Pichia pas- Vogl, T., Sturmberger, L., Kickenweiz, T., Wasmayer, R., toris as a versatile cell factory for the production of indus- Schmid, C., Hatzl, A.M., et al. (2016) A toolbox of diverse trial enzymes and chemicals: current status and future promoters related to methanol utilization: functionally veri- perspectives. Biotechnol J 14: 1800694. fied parts for heterologous pathway expression in Pichia pastoris. ACS Synth Biol 5: 172–186. Vogl, T., Gebbie, L., Palfreyman, R.W., and Speight, R. (2018a) Effect of plasmid design and type of integration Supporting information event on recombinant protein expression in Pichia pas- toris. Appl Environ Microbiol 84: e02712-17. Additional supporting information may be found online in Vogl, T., Sturmberger, L., Fauland, P.C., Hyden, P., Fischer, the Supporting Information section at the end of the arti- J.E., Schmid, C., et al. (2018b) Methanol independent cle. induction in Pichia pastoris by simple derepressed over- Table S1. Primer pairs used for gene dosage and tran- expression of single transcription factors. Biotechnol Bio- script-level determination by means of ddPCR and qPCR eng 115: 1037–1050. respectively. ª 2019 The Authors. Microbial Biotechnology published by John Wiley & Sons Ltd and Society for Applied Microbiology., Microbial Biotechnology, 13, 315–327