REVIEW published: 25 June 2020 doi: 10.3389/fbioe.2020.00632 Continuous Cultivation as a Tool Toward the Rational Bioprocess Development With Pichia Pastoris Cell Factory Miguel Angel Nieto-Taype, Xavier Garcia-Ortega, Joan Albiol, José Luis Montesinos-Seguí and Francisco Valero* Department of Chemical, Biological and Environmental Engineering, School of Engineering, Universitat Autònoma de Barcelona, Bellaterra, Spain The methylotrophic yeast Pichia pastoris (Komagataella phaffii) is currently considered one of the most promising hosts for recombinant protein production (RPP) and metabolites due to the availability of several tools to efficiently regulate the recombinant expression, its ability to perform eukaryotic post-translational modifications and to secrete the product in the extracellular media. The challenge of improving the bioprocess efficiency can be faced from two main approaches: the strain engineering, which Edited by: includes enhancements in the recombinant expression regulation as well as overcoming Manfred Zinn, potential cell capacity bottlenecks; and the bioprocess engineering, focused on the HES-SO Valais-Wallis, Switzerland development of rational-based efficient operational strategies. Understanding the effect Reviewed by: of strain and operational improvements in bioprocess efficiency requires to attain a Dongming Xie, University of Massachusetts Lowell, robust knowledge about the metabolic and physiological changes triggered into the cells. United States For this purpose, a number of studies have revealed chemostat cultures to provide a Dirk Tischler, Ruhr University Bochum, Germany robust tool for accurate, reliable, and reproducible bioprocess characterization. It should *Correspondence: involve the determination of key specific rates, productivities, and yields for different Francisco Valero C and N sources, as well as optimizing media formulation and operating conditions. francisco.valero@uab.cat Furthermore, studies along the different levels of systems biology are usually performed Specialty section: also in chemostat cultures. Transcriptomic, proteomic and metabolic flux analysis, using This article was submitted to different techniques like differential target gene expression, protein description and Bioprocess Engineering, 13 C-based metabolic flux analysis, are widely described as valued examples in the a section of the journal Frontiers in Bioengineering and literature. In this scenario, the main advantage of a continuous operation relies on the Biotechnology quality of the homogeneous samples obtained under steady-state conditions, where Received: 27 September 2019 both the metabolic and physiological status of the cells remain unaltered in an all- Accepted: 22 May 2020 Published: 25 June 2020 encompassing picture of the cell environment. This contribution aims to provide the Citation: state of the art of the different approaches that allow the design of rational strain and Nieto-Taype MA, Garcia-Ortega X, bioprocess engineering improvements in Pichia pastoris toward optimizing bioprocesses Albiol J, Montesinos-Seguí JL and based on the results obtained in chemostat cultures. Interestingly, continuous cultivation Valero F (2020) Continuous Cultivation as a Tool Toward the Rational is also currently emerging as an alternative operational mode in industrial biotechnology Bioprocess Development With for implementing continuous process operations. Pichia Pastoris Cell Factory. Front. Bioeng. Biotechnol. 8:632. Keywords: Pichia pastoris, continuous cultivation, steady-state omics, physiological characterization, rational doi: 10.3389/fbioe.2020.00632 bioprocess development, bioreaction kinetics, heterologous protein production, systems microbiology Frontiers in Bioengineering and Biotechnology | www.frontiersin.org 1 June 2020 | Volume 8 | Article 632 Nieto-Taype et al. Continuous Cultivation Toward Rational Bioprocesses INTRODUCTION interest (Cregg et al., 1993; Paulová et al., 2012; Ponte et al., 2016, 2018; Lee et al., 2017). Although PAOX1 affords outstanding RPP In recent years, recombinant protein production (RPP) levels, using methanol has some relevant drawbacks, especially technology has given rise to a multibillion-dollar market at production scale processes. These are mainly related with offering products for a wide range of industrial activities the increased costs and risk of storing and using methanol, as such as food, detergent, paper, chemical, cosmetic, and well as the increased oxygen demand and high heat production pharmaceutical production. For example, more than 400 RPP- derived from methanol metabolization (Heyland et al., 2010; based pharmaceutical products are currently available on the Prielhofer et al., 2013; Çalik et al., 2015). These shortcomings market (Sanchez-Garcia et al., 2016; Vieira Gomes et al., 2018). have given ground to dedicate important efforts to identify and In parallel, and derived from the RPP technology, metabolic to develop methanol-free alternatives, which can be classified in engineering has opened the door to metabolic flux modification two categories: discovery of alternative efficient promoters for and new heterologous reactions. This enable the manufacture of P. pastoris (PGAP , PPGK , PTHI11 , PPYK , PSDH ) (Periyasamy et al., a wide range of products of interest including vitamins, amino 2013; Landes et al., 2016; Juturu and Wu, 2018; Massahi and acids, ethanol, antimicrobials, enzyme inhibitors, and organic Çalik, 2018; de Macedo Robert et al., 2019) and engineering or acids. The overall economic impact of RPP and metabolite development of synthetic promoters (Ata et al., 2017; Wang et al., productions was estimated at around 143.5 billion US dollars in 2017; Prielhofer et al., 2018; Vogl et al., 2018b). 2014, with upward forecasts until 2020 (Porro et al., 2011; Singh Nevertheless, not only is the promoter selection key in et al., 2017). the strain engineering development to construct efficient cell Various organisms ranging from bacterial hosts to transgenic factories, other relevant factors such as host strain background, animals have been proposed as efficient cell factories for gene engineering and dosage, protein processing helpers or bioprocesses with the mentioned applications. For example, secretion machinery also have a significant impact on the overall for RPP, most of the products including pharmaceuticals and performance of the strain (Ahmad et al., 2014). Furthermore, industrial enzymes are obtained with the main considered it is crucial to combine the mentioned factors with an efficient workhorses, namely: Escherichia coli, Saccharomyces cerevisiae or bioprocess development from a bioprocess engineering approach Chinese hamster ovary (CHO) cells (Sørensen, 2010; Maccani (García-Ortega et al., 2019). For the latter, it should be essential to et al., 2014). However, the use of non-conventional yeasts, carry out a characterization producer clone used as a cell factory which combine eukaryotic ability for protein processing and over different culture conditions to understand its performance major microbial advantages, have lately started to be considered and therefore, to design the rational based bioprocess strategies as promising alternatives. In general, the new choices offer that allows to achieve the optimal performance for industrial many advantages in terms of pathway requirements, desired bioprocesses (Yang and Zhang, 2018). product profile, and gross physiology over S. cerevisiae (Wagner To obtain a proper characterization, cells should be kept and Alper, 2016). Such is the case with Schizosaccharomyces under constant conditions in terms of both physico-chemical pombe, Hansenula polymorpha (syn. Ogataea polymorpha), operating conditions as well as their metabolic state (Hoskisson Kluyveromyces lactis, Yarrowia lipolytica, or Komagataella phaffii and Hobbs, 2005). In this sense, continuous operation, which (Pichia pastoris) (Çelik and Çalik, 2012; Baghban et al., was first reported in 1920 (Cooney, 1979), is considered the most 2019). Actually, at present Pichia pastoris (recently classified simple operational mode for this purpose. Continuous cultures as Komagataella spp.) has arisen as an efficient and versatile are defined as a fermentation systems in which fresh medium is cell factory for obtaining a wide spectrum of biotechnological continuously added to the bioreactor and the components of the products including recombinant proteins and metabolites of culture broth—cells and metabolites included—are continuously diverse origins (Gasser et al., 2013; Peña et al., 2018). removed from the vessel (Fernandes et al., 2015); thereby The key features that make P. pastoris an outstanding host for forcing the cells to proliferate at a fixed rate and in a constant the above-mentioned uses include fast growth at high densities environment reaching the stationary state (Gramelsberger, 2018). in defined media, the availability of advanced tools for genetic The two main types of continuous cultures, chemostat, and modification (e.g., CRISPR/Cas9 system) and the ability to turbidostat, are analogous and differ mainly in their cell growth perform post-translational modifications as well as to secrete control mechanism (Prielhofer et al., 2013; McGeachy et al., the products extracellularly (Vogl and Glieder, 2013; Gasser 2019). On one hand, chemostat is controlled through the and Mattanovich, 2018; Weninger et al., 2018). Furthermore, continuous addition of culture medium in which a single nutrient the increasing knowledge on P. pastoris metabolism gathered is present at growth limiting conditions (Peebo and Neubauer, over the last decades, together with the high RPP potential of 2018). On the other hand, in turbidostat, the goal is to avoid this yeast, have strongly increased the interest in using this nutrient limitation while growth is controlled by using an optical microbial cell factory to obtain a variety of compounds. Thus, sensor which maintains a determined turbidity level by adding strengthening P. pastoris as one of the most suitable chassis for fresh medium through a feedback control loop. In this case, the biotechnological industry. the resulting specific growth rate (µ) obtained is close to the Promoters have proved to be key regulators for RPP processes. maximum specific growth rate (µmax ) of the microorganism used For P. pastoris, the methanol inducible alcohol oxidase promoter (Gresham and Dunham, 2014; Fernandes et al., 2015). (PAOX1 ), a strong and tightly regulated promoter, was the first Additional alternatives such as “changestats” have been used obtaining efficient production rates of several proteins of proposed as a novel tool where an environmental parameter is Frontiers in Bioengineering and Biotechnology | www.frontiersin.org 2 June 2020 | Volume 8 | Article 632 Nieto-Taype et al. Continuous Cultivation Toward Rational Bioprocesses FIGURE 1 | Overview of different continuous cultivation strategies. They are divided in two main groups: the first, corresponding to the “classical” approaches which work at steady state conditions; whereas for the other hand, changestats are variations of classical strategies proposed to obtain similar results but at dynamic conditions allowing faster results. continuously changed in a single experiment. These approaches – Nutristats, where the dilution rate is a function of the set-point provide quick information about the response of cells in front substrate concentration in the bioreactor (Curvers et al., 2002). of determined environmental conditions under what has been – Finally, as special case, adaptastats allow to operate under called steady state growth space analysis (GSA) (chemostat or substrate-limiting conditions but, like turbidostats, work near turbidostat-based approaches) (Adamberg et al., 2015). Several µmax (Tomson et al., 2006). variants have been described in the literature and have been Although changestats have been presented as promising listed and classified below according to the limitation or not of alternatives to classical continuous cultures, it should be noted a nutrient and it is summarized in Figure 1. that in these systems cells grow in a quasi-steady state Nutrient-limited changestats: because the absence of a total stabilization phase prevents – Accelerostats (A-stats) and deceleration-stats (De-stats), them from reaching steady state (Subramanian et al., 2017). which are used to study the increase or decrease the specific This is the reason why chemostats continue to be the main growth rate, respectively (Nahku et al., 2010; Hoekema et al., choice for characterizing microorganisms under substrate- 2014; Wagenen et al., 2015; Gerritzen et al., 2018). limiting conditions (Ziv et al., 2013). In fact, first reported – Dilution rate stats (D-stats), where an environmental in 1950 (Monod, 1950), chemostat cultures have proved the parameter is continuously changed while the dilution rate is best operational choice for precise kinetic and metabolic kept constant (Kasemets et al., 2003). characterization; also, they have provided the most accurate – Retentostats, where biomass is retained by means of internal quantitative understanding at whole-cell level (Hoskisson and or external filter devices. In prolonged retentostats, the energy Hobbs, 2005). Thus, the empirical knowledge obtained from this derived from the consumed carbon source only allows to characterization allows the determination of several parameters sustain the cellular maintenance but can no longer support at different conditions, which therefore can be used to identify biomass growth. This allows experiments to be performed at the optimal culture conditions toward the maximization of very low or near-zero specific growth rates (Herbert, 1961; product related parameters such as production rates and yields. Ercan et al., 2015a,b; Rebnegger et al., 2016). The relation between cultures and production parameters is detailed through different examples in the following sections. On Non-nutrient-limited changestats the other hand, the application of systems biology approaches – Auxo–accelerostats, allows to examine the effects of smooth based on continuous cultures has enabled the development of environmental changes under nutrient-excess conditions new highly sensitive analysis tools capable of detecting slight (Nisamedtinov et al., 2008). changes at different regulation points, which can be deemed Frontiers in Bioengineering and Biotechnology | www.frontiersin.org 3 June 2020 | Volume 8 | Article 632 Nieto-Taype et al. Continuous Cultivation Toward Rational Bioprocesses FIGURE 2 | Valuable information that gives continuous cultivation to systems biology for a deep understanding about cell physiology. significant when integrating experimental results (Rebnegger promoters engineering as an example) or to external stimuli such et al., 2014). as operational strategies or environmental stress. The following sections describe the basis and uses of Transcriptomic, proteomic, metabolomics, and fluxomic data, continuous cultivation with Pichia pastoris. Specifically, it has can be jointly integrated into a systems biology approach (shown been focused on carbon-limited chemostat cultures, which in Figure 2) in order to obtain an all-encompassing picture of according to the literature is the most used strategy for Pichia a biological system used (Graf et al., 2009). A more detailed pastoris in continuous operation. The role of chemostat cultures description is provided below in the next sections. has been reviewed as a powerful tool for the characterization of P. pastoris strains with a view to enabling the rational design of Transcriptional Studies engineered cell factories and optimal operational strategies for Transcriptional analysis is usually the first step in systems maximal bioprocess efficiency. biology studies as it is also the first stage in the information flow from genomic data (Engstrom and Pfleger, 2017). Gene regulation can be examined by absolute or relative quantification CONTINUOUS CULTIVATION PROVIDES of individual transcripts. Integrating this information enables AN EXCELLENT TOOL FOR SYSTEMS further interpretation of data and reveals how the complex BIOLOGY RESEARCH machinery behind a specific phenotype operates as a result of certain experimental conditions. The insight thus gained can be Steady-State Omics, a Key Issue in the contrasted with knowledge derived from other structural levels of Renaissance of Research Into Continuous systems biology. Cultures The contribution of transcriptional regulation to physiological The development of high-throughput molecular biology outcomes is strongly dependent on the host employed. Thus, techniques in the post-genomic era has brought a renaissance although 27% of all protein-coding genes in human cell lines are of continuous cultures. At present, they are used not only in transcriptionally regulated (Vogel et al., 2010), the proportion kinetic characterization and adaptive laboratory evolution (ALE) amounts to 50% in E. coli and up to 70% in S. cerevisiae (Lu studies (Bull, 2010), but also as a powerful tool for gaining et al., 2007). These figures testify to the high significance of deeper insight into cell behavior thanks to the large amount interpretations of transcriptional research in yeasts, where P. of robust information they provide. In this sense, continuous pastoris is taxonomically located. cultivations have been used to understand cells response to Transcriptional data are typically acquired in either of internal traits such as those observed in engineered clones (using two methodologies: Frontiers in Bioengineering and Biotechnology | www.frontiersin.org 4 June 2020 | Volume 8 | Article 632 Nieto-Taype et al. Continuous Cultivation Toward Rational Bioprocesses - One involves using specific primers for selected target genes differentially regulated under different expressing conditions either by real-time quantitative PCR (qPCR) (Landes et al., was suggested. 2016) or droplet-digital PCR (ddPCR) analyses (Cámara et al., On the other hand, endoplasmic reticulum associated protein 2019). Both methods provide numerical results and hence degradation (ERAD) is one other limiting factor for RPP quantitative data. Also, they are commonly used to examine (Dragosits et al., 2010; Adelantado et al., 2017). UPR and the effects on the transcription regulation of certain genes of ERAD have been found to jointly control protein folding to known sequence. a considerable extent and also, generally, heterologous protein - However, other high-throughput technology methods synthesis, in P. pastoris (Delic et al., 2013; Zahrl et al., 2018). such those using microarrays, based qualitative fluorescent The relevance of both mechanisms as a response to RPP was detection, and/or RNA sequencing (RNAseq), which quantify confirmed by a microarray analysis intended to shed light on the the number of transcripts present in a sample by whole transcriptional upregulation of these mechanisms in P. pastoris transcriptome sequencing, have arisen as helpful workhorses. clones expressing a 2F5 Fab fragment relative to the wild type In fact, these methods provide a robust overview of all strain X-33 in chemostat cultures (Gasser et al., 2007). transcripts (Baumann et al., 2010; Ata et al., 2018) and are Furthermore, it has been observed that transcriptional thus especially useful when increased gene coverage is needed. analysis provides useful information for characterizing native and engineered promoters. Thus, Landes et al. (2016) The selection of the method to be used in each case will be characterized the production of recombinant human serum dictated by the specific objectives and information available about albumin (HSA) by using an alternative promoter from the the studied system. thiamine biosynthesis gene THI11, which is thiamine-repressible. Continuous cultivation has proved useful for transcriptional An initial transcriptional assay on batch cultures using a specific analysis. Specifically, chemostats are frequently deemed the most transcriptional analysis with qPCR provided insight into the suitable operational choice for this purpose, as seen with S. promoter performance under non-repressive conditions— cerevisiae strains producing either recombinant proteins (Liu absence of thiamine—and additional batch experiments allowed et al., 2013) or metabolites (Vos et al., 2015). As shown by several to define repression and de-repression effects of the thiamine successful attempts described below, chemostat uses could be concentration. Also, chemostat cultures were used to evaluate extended to P. pastoris, a similar cell factory related in the the performance of the same promoter under carbon-limited taxonomical group of yeasts. conditions at different specific growth rates in order to monitor Expression of heterologous proteins and metabolites in P. production rates, and THI11 and HSA gene transcription. pastoris triggers many changes in cell physiology including Whereas, the specific production rate (qHSA ) improved with stoichiometric drain of energy and cellular resources, or increasing specific growth rate (µ), gene transcription had no restrictions in the biochemical machinery responsible for influence on the latter, indicating a constitutive nature of PTHI11 . bioproduct synthesis (Hesketh et al., 2013; Nocon et al., 2014). In similar work intended to develop alternatives to Identifying different points of regulation through transcriptional the methanol-based expression system, Prielhofer et al. analysis can therefore allow useful information about cell (2013) identified six novel promoters through microarray performance to be obtained. Such information could be transcriptomics analysis under carbon non-limiting and limiting used to address new targets by strain engineering and to conditions (batch and chemostat cultivation, respectively). One develop innovative, more efficient operational approaches by of the promoters, PG1 , corresponding to a high-affinity glucose bioprocess engineering. transporter (GTH1), exhibited strong and tight regulation under Puxbaum et al. (2015) provided a solid discussion about the carbon-limiting conditions. Also, further characterization of this potential limitations of RPP in P. pastoris; they found protein promoter at variable specific growth rates in chemostat cultures synthesis, folding, and secretion to be the main bottlenecks revealed the optimal specific production rate at a dilution for the process. In this sense, Hesketh et al. (2013) directly rate D = 0.07 h−1 . observe stress arising during RPP by expressing under AOX1 As it was observed previously, the useful information promoter two lysozyme variants differing in degree of misfolding derived from the production kinetics of recombinant P. pastoris in chemostat cultures. The effect on unfolded protein response producer clones growing at controlled specific growth rates (UPR) was examined by using the RNAseq SOLiD platform. In in chemostat cultures have also enabled further transcriptional addition, further RT-PCR analysis showed a constitutive splicing characterization and expansion of existing knowledge about of HAC1 gene transcript across the different phases of RPP the impact of this key parameter for improvement of the cell induction independent of the lysozyme misfolding level. The factory. Thus, Rebnegger et al. (2014) performed a transcriptional importance of this finding is based on the fact that Hac1p analysis using microarrays on P. pastoris clone expressing HSA corresponds to a key transcription factor in the triggering of under GAP promoter in chemostat cultures to examine the UPR response (Graf et al., 2008), and data obtained suggests that transcriptional response at different µ values. They found qHSA UPR is dependent on the transcript regulation of HAC1 rather and µ to be linearly related, and upregulation of ribosomal than on the splicing process. As a result, the genes KAR2 and and translation-related genes in response to higher heterologous PDI1, which are further down the line in UPR, also exhibited protein production. On the other hand, upregulation of UPR high relative expression as revealed by RNAseq transcriptome and secretory pathways at high µ levels suggested that, although analysis. Finally, a potential role of untranslated RNA molecules high D rates resulted in no limitation in qHSA , they caused Frontiers in Bioengineering and Biotechnology | www.frontiersin.org 5 June 2020 | Volume 8 | Article 632 Nieto-Taype et al. Continuous Cultivation Toward Rational Bioprocesses heterologous protein production-related stress. Furthermore, the oxygen-limiting effects observed in metabolism regulation, marked upregulation of transcriptional regulators involved in which included upregulation of genes associated to the glycolytic stress response as well as a carbon source and nitrogen responsive pathway, the oxidative branch of the pentose phosphate genes was observed at low µ levels, which suggest an adaptation pathway (PPP), ergosterol and sphingolipid biosynthesis and to the carbon source limitation when the dilution rate was tuned. UPR genes, while the tricarboxylic acid cycle (TCA) revealed Finally, consistent with the widely reported Crabtree-negative transcriptional downregulation. effect on P. pastoris metabolism, mitochondrial transcripts were Osmotic stress has been reported to exert a beneficial positively correlated to µ. It is worth mentioning that Crabtree effect on RPP. Thus, Dragosits et al. (2010) conducted a effect corresponds to the respiro-fermentative phenotype widely transcriptional analysis by using microarrays in combination reported for S. cerevisiae, however a similar phenomenon has with chemostat cultures and found increasing concentrations been observed also in cancer cells (Warburg effect). In order of KCl to result in upregulation of genes involved with UPR, to generate an alternative model for Warburg effect studies, ribosome biogenesis and translation, albeit only in the wild- Ata et al. (2018) convert Crabtree negative P. pastoris into a type clone. Although a Fab-producing clone exhibited higher Crabtree positive. While determination of Crabtree phenotype transcriptional levels than the wild-type strain, such levels were was achieved using short- and long-term studies in chemostat independent of the experimental conditions and no increase in cultures, RNASeq transcriptome and 13 C central metabolism product formation was observed at higher concentrations of KCl. flux analysis were performed in batch cultures. The use of This result suggests that upregulation of these genes as revealed chemostats for further transcriptional and metabolic studies by microarray analysis was previously triggered by a recombinant could be proposed to provide new insight of P. pastoris as better protein effect and that osmolarity induced no further increase. model than S. cerevisiae for Warburg effect studies. Although most transcriptional analyses have been performed One other interesting use of molecular biology in strain in chemostat cultures, some have been conducted in retentostats. engineering is to integrate multiple target expression cassettes Rebnegger et al. (2016) performed retentostat studies to outline in the P. pastoris genome, which usually leads to substantially the physiology and transcriptional response of P. pastoris increased production. Previous reports suggest that gene dosage under this operational condition at near-zero specific growth considerably influences specific productivity (Schwarzhans et al., rates. Among the pool of genes that were differentially 2016; Vogl et al., 2018a). Thus, Cámara et al. (2017) evaluated expressed in microarray hybridization for transcriptome analysis, the effect in chemostat cultures growing at µ = 0.10 h−1 upregulation in transcripts involved in alternative carbon source by using strains with multiple insertions of Rhizopus oryzae consumption (such as ethanol and methanol), global stress- lipase (ROL) under AOX1 promoter (0, 1, 2, 4, 8, and 15 related transcription factors and with nutrient responses reveal copies). As confirmed by ROL transcriptional levels in a ddPCR an adaptation of low substrate availability at near-zero growth analysis, optimal specific extracellular lipase activity was obtained rates. This finding reinforces the reduction observed in the with the 2C clone. By contrast, transcriptomic analyses with maintenance coefficient (ms ), in comparison to chemostat microarrays revealed constraints with increasing ROL gene cultures growing at higher µ. Otherwise, protein synthesis dosage on the transcriptional machinery (specifically, in the in retentostat cultures exhibited insubstantial transcriptional methanol expression regulator 1, Mxr1p) (Cámara et al., 2017). regulation, so further research, for example with RP-expressing This information was used to design new clones carrying several clones, to determine whether the alternative cultures could copies of MXR1 and methanol-induced transcription factor be useful with P. pastoris and to outperform S. cerevisiae (MIT1). As revealed by specific analysis with ddPCR, methanol translational capacity at low µ is proposed. uptake was found to have a positive effect on the transcriptional response; the effect, however, resulted in no increase in product Proteomic Studies activity (Cámara et al., 2019). According to the central dogma of molecular biology, the protein Not only strain engineering, but also new bioprocess level falls at the last stage of the information flow (van Hove strategies, have led to enhanced RPP. Therefore, better et al., 2016; Engstrom and Pfleger, 2017). This, however, is understanding the way cells adapt to different external conditions no reason to underestimate the importance of the information requires accurate physiological characterization (Baumann et al., it provides. Proteomics emerged in the mid-1990s as a new 2008; Dragosits et al., 2010; Garcia-Ortega et al., 2016). Based approach to determine the proportion of functional proteins on the positive effect of oxygen-limiting conditions on RPP containing information translated into the transcriptome pool. (Baumann et al., 2008), the impact of these stress conditions Such information is useful to contrast and supplement data was further examined by using transcriptional analysis with obtained at other levels in systems biology in order to support microarrays in samples obtained from chemostat cultures and strengthen the conclusions of other studies (Szopinska and under normoxic, oxygen-limited and hypoxic conditions to Morsomme, 2010). study the expression of an antibody fragment (Fab) under Proteomic analysis with various microorganisms has provided the constitutive GAP promoter (Baumann et al., 2011). interesting results (Brejning et al., 2005; Rossignol et al., 2009; Supplementary proteomic and metabolic flux analyses, and Arvas et al., 2011; Dragosits et al., 2011). The use of proteomics lipid-profile analyses (lipidomics) support the findings obtained on P. pastoris has been widely used in different cultivation modes previously (Baumann et al., 2010; Adelantado et al., 2017). including chemostat, batch and fed-batch cultures (Baumann The ensuing information enabled thorough characterization of et al., 2010; Dragosits et al., 2010; Pfeffer et al., 2012; Vanz Frontiers in Bioengineering and Biotechnology | www.frontiersin.org 6 June 2020 | Volume 8 | Article 632 Nieto-Taype et al. Continuous Cultivation Toward Rational Bioprocesses et al., 2012, 2014; Lin et al., 2013), though not so widely as understanding the behavior of the cell factory is rather difficult as in transcriptional analysis. This section focuses on applications a result. involving continuous cultivation, specifically in chemostats. The earliest example of proteome characterization was Metabolic Fluxes as Key Physiological accomplished by Dragosits et al. (2009), who compared the effect of temperature changes on recombinant proteome. The greatest Indicators changes were observed at the highest and lowest temperature The physiological status of cells, which is directly reflected in studied (20 and 30◦ C, respectively). In addition to increased their metabolic phenotypes, is a result of the interplay of various specific productivity at 20◦ C, 2D-Gel electrophoresis followed key cell processes and factors such as gene expression, protein by LC-ESI-QTOF tandem MS allowed the identification of 49 production and kinetics, regulation and thermodynamic driving out of 150 spots that exhibited significant differences. Proteomic forces (metabolite concentrations). A metabolic phenotype is data revealed that some cellular processes were affected by low defined by the actual activities of the metabolic reactions in cells temperatures. On the one hand, energy metabolism, oxidative (viz., by reaction rates). Because they are rarely constant, reaction stress response, and protein folding protein levels occurred rates are usually determined under steady-state or pseudo-steady to a lesser extent; whereas, in amino acid metabolism and state conditions and commonly referred to as “metabolic fluxes,” RNA/ribosomal biogenesis, an increase of protein levels might which are thus key indicators of the physiological status of cells. account for the positive effect on RPP at 20◦ C. Determining metabolic fluxes remains difficult despite major Dragosits et al. (2010) conducted a similar study to determine technological advances in quantifying all components involved. the effect of osmolarity on recombinant P. pastoris physiology. In fact, it is not always possible to determine the concentrations Thirty-seven out of the total spots detected, 300 for wild-type of all relevant components; also, in vivo values for kinetic and 150 for recombinant clone, were successfully identified parameters are usually unavailable. A number of techniques have with significant changes. These results strengthen the findings thus been developed to quantify metabolic fluxes collectively of transcriptional analysis as regards energy metabolism and that are known as “metabolic flux analysis” (MFA) or, simply, protein folding in both clones. On the other hand, as it “fluxomics.” Many of them have been applied to P. pastoris and was observed in transcriptional analysis, a positive effect on the most salient are discussed below. Using those techniques UPR response protein levels was observed at high osmolarity for metabolic engineering improves existing knowledge of the conditions, however it was only in control wild-type strain, physiological properties of biological systems, thus enabling whereas it was not the case for recombinant strain where no the rational design of cell factories and efficient metabolic impact was exerted at different KCl concentrations. The author engineering cycles. suggests that although osmolarity not exert a direct effect in producer clone, it could be used for preconditioning cells for RPP, Metabolic Flux Analysis and as outcome it can be considered on improved bioprocess and As stated above, MFA is widely used for comprehensive analysis for generate novel routes for strain engineering. of the physiological effects of environmental changes or genetic Rußmayer et al. (2015) interpreted proteomic data at other engineering modifications. Metabolic fluxes (i.e., metabolic levels of systems biology and demonstrated symbiosis by using reaction rates in the steady state) provide an excellent operational multi-level analysis to expose the cellular processes involved picture of cell operation. A continuous culture provides a simple, in methanol metabolism. They found good positive correlation convenient tool for maintaining a stable (in steady state) cell between transcriptomic and proteomic data, consistent with metabolism over long periods of time. Comparing metabolic the strong transcriptional control over protein abundance flux distributions under different conditions allows a deeper in S. cerevisiae (Lu et al., 2007). Chemostat cultures on understanding of the effects resulting from the genetic or methanol–glycerol mixed feed revealed a positive correlation environmental changes. between protein production and transcripts from methanol Although determining metabolic fluxes in continuous cultures metabolism, and also with peroxisome biogenesis. There was, has the advantage that cell inputs and outputs are stable, it is however, no clear-cut relationship between the increased still a difficult task due to the large number of reactions taking protein levels associated to the translation machinery and place simultaneously in each cell. Because in steady state reaction cytoskeleton organization, which were not reproduced at the rates are constant, it only requires considering the stoichiometry transcriptional level. of the reactions to calculate metabolic fluxes. The mass balances The usefulness of proteomic analysis as a supplement to for each metabolite involved in such reactions allow one to transcriptional and metabolic flux analysis can also be illustrated construct a linear metabolic model. Those metabolic models with the results under hypoxia conditions. Thus, Baumann with a realistic degree of metabolic detail are known as Genome et al. (2010) conducted an integrative multilevel study and Scale Metabolic Models (GEMs). Using GEMs in combination found a strong positive correlation between transcriptional with steady-state data should in theory allow one to calculate levels and protein data. This result provides support for the metabolic fluxes. Unfortunately, the intricate interconnections assumption that transcriptional regulation is stronger than post- among metabolites, and the small number of flux measurements transcriptional regulation in P. pastoris. In any case, proteomic typically available (usually cell inputs and outputs only), prevent data should never be considered irrelevant, particularly in those the accurate determination of most of the metabolic fluxes, which cases were transcript levels do not match protein expression and requires the combined use of mathematical models, optimization Frontiers in Bioengineering and Biotechnology | www.frontiersin.org 7 June 2020 | Volume 8 | Article 632 Nieto-Taype et al. Continuous Cultivation Toward Rational Bioprocesses methods and/or experimental techniques such as 13 C labeling and handling of data. These methodologies have been used in (Ferrer and Albiol, 2014a,b). combination with chemostat cultures to examine the impact Genome-scale metabolic models have become essential tools of heterologous protein production and the effects of different for MFA and rational-based strain engineering. These models environmental conditions on P. pastoris metabolism. are mathematically structured knowledge bases containing Specific growth rate is one of the basic variables influencing descriptions of all biochemical reactions, metabolites, and the distribution of metabolic fluxes, and the macromolecular and metabolic genes for a specific organism—a Biochemical, Genetic elemental composition of biomass. This specific rate is easy to and Genomic (BiGG) knowledge base (King et al., 2016). GEMs regulate in continuous cultures. In P. pastoris, the influence of can be used in a number of ways such as using their elementary the specific growth rate has been studied by applying 13 C labeling modes or extreme pathways to predict phenotypic function, to to chemostat cultures fed with various carbon sources (glycerol, improve understanding of the metabolism underlying structure, methanol) or their mixtures (e.g., glycerol/methanol) (Jordà et al., or using the well-known constraint-based modeling approach 2014a; Tomàs-Gamisans et al., 2018). In general, the results were (Fong, 2014) to predict function, as examples. consistent with what is known as the “growth rate hypothesis” Constructing so detailed BiGG knowledge bases is a which states that increasing the specific growth rate increases continuously evolving process due to the still limited knowledge the RNA and protein fractions of cells, thereby boosting protein available on some cell details. The availability of the P. pastoris synthesis and energy production at the expense of other fractions genome sequence allowed the development of the first metabolic such as carbohydrates. reconstructions. Thus, the DSMZ70382 genome sequence, and Biomass composition has also been found to depend on the sequence from the GS115 strain, allowed the PpaMBEL1254 the particular carbon source. Thus, cells grown on methanol and iPP668 GEM, respectively, to be produced (Chung et al., contain more protein than those grown on glycerol or glucose 2010; Sohn et al., 2010). Although several improved versions (Tomàs-Gamisans et al., 2018). This effect has also been seen of these P. pastoris GEMs have been reported (Ye et al., 2017; with mixed glycerol/methanol or glucose/methanol feeds and Tomàs-Gamisans et al., 2018), this is a continuously evolving can be ascribed to the methanol utilization pathway requiring field and new or upgraded versions of existing models are an expanded metabolic machinery (Jordà et al., 2012, 2014a; expected to appear in the near future. Rußmayer et al., 2015). On the other hand, the amino acid The continuous cultivation methodology has been used to composition of cells is virtually independent of their specific adjust and validate GEMs. Also, GEMs adapted to specific growth rate and only differs among carbon sources. Based on continuous culture conditions have been used with the latest the foregoing, metabolic fluxes are calculated by using different algorithms and optimization techniques to determine metabolic biomass composition equations whichever the approach adopted fluxes under constrains derived from experimental data. This (GEMs included). The environmental conditions also affect approach is highly useful for interpreting such data and also for metabolic flux distribution. For example, altering the chemostat designing effective metabolic engineering strategies. Its greatest dilution rate or using a different carbon source alters the advantage is that it can be used in any type of continuous culture distribution of metabolic fluxes; as a result, specific consumption experiment without the need for a special experimental technique and production rates increase with increasing specific growth such as 13 C labeling. rate or by-products such as arabitol are formed (Jordà et al., The most common alternative to using an optimization 2014a; Tomàs-Gamisans et al., 2018). Interestingly, the amount technique for metabolic flux calculations involves obtaining of methanol directly dissimilated into CO2 is influenced not supplementary experimental information. Adding a carbon- only by the specific growth (dilution) rate, but also by the labeled carbon source to the input medium of a continuous methanol fraction in mixed feeds. Experiments using chemostat culture is one common method for this purpose. A typical cultures have shown environmental factors such as oxygen- experiment involves establishing a continuous culture and, once limited conditions or low temperatures to have a strong impact it has reached steady state, replace the input culture medium on MFA (Mattanovich et al., 2004; Dragosits et al., 2009; with another one containing a 13 C-labeled carbon source. The Baumann et al., 2010; Carnicer et al., 2012). labeled substrate usually consists of a mixture of molecules The other most extensively studied factor in P. pastoris labeled at different carbon positions in proportions that are continuous cultures in addition to the environmental conditions selected according to the expected results. The information is the impact of heterologous protein production on MFA. For to be acquired can be either the distribution of the label this purpose, the effects of different model proteins on metabolic once steady state is reached (Tomàs-Gamisans et al., 2019) flux distribution relative to their non-productive counterparts or changes in the label during the transient labeling period is usually compared under identical cultivation conditions. (Heyland et al., 2011; Jordà et al., 2012, 2014b; Nocon et al., The type of heterologous protein produced and, especially, 2016). In either case, using this information jointly with steady- the amount produced, have proved considerably influential state input/output fluxes and a metabolic model allows one on metabolic flux distributions. Thus, increased heterologous to calculate metabolic fluxes under specific conditions. Using protein production has been shown to shift metabolism toward GEMs in combination with 13 C labeling requires solving a increased energy generation, reduction equivalents (ATP, NADH, very large number of equations. It is thus common to use NADPH) and building block production (enhanced pentose manually or automatically constructed “context-specific” core phosphate pathway, PPP, or TCA cycles) while reducing biomass models (Tomàs-Gamisans et al., 2019) to facilitate calculations yield (Heyland et al., 2011; Jordà et al., 2012, 2014b; Nocon Frontiers in Bioengineering and Biotechnology | www.frontiersin.org 8 June 2020 | Volume 8 | Article 632 Nieto-Taype et al. Continuous Cultivation Toward Rational Bioprocesses et al., 2016). Increased heterologous protein production has also Operating Conditions and Media been found to boost protein misfolding and, ultimately, trigger Optimization an unfolded protein response (UPR), thereby further draining Bioprocess performance is influenced by some physical resources and shifting metabolic fluxes (Graf et al., 2008). conditions of cultures including temperature (T), pH and The effects of environmental factors and heterologous protein dissolved oxygen (DO). The optimum conditions for production production on metabolic fluxes have fostered the development may differ from those for growth due to their effect on several of in silico strategies to optimize metabolic engineering. factors including protein-specific characteristics, kinetics and sensitivity of products to proteolytic degradation. BIOPROCESS CHARACTERIZATION WITH The optimum growth temperature for P. pastoris bioprocesses CHEMOSTAT CULTURES is about 30◦ C, above 32◦ C protein expression is suppressed and cell growth rapidly decays (Çalik et al., 2015). However, some Microbial Physiology Studies/Physiological authors have suggested that operating at lower temperatures of Characterization of the Cell Factory up to 20◦ C may significantly improve production of heterologous Increasing the efficiency of a bioprocess (i.e., increasing yields proteins (Surribas et al., 2007). Decreased operating temperatures and productivities with high-quality proteins) requires a deep have a double but opposite effect. On one hand, the solubility physiological understanding of the cell factory. For the topic of oxygen is increased and consequently the oxygen transfer discussed in this review, such knowledge is expected to rate (OTR) is improved. On the contrary, bioreaction rates such eventually allow the rational development of feasible and as intracellular reactions, proteolysis and cell growth diminish. competitive bioprocesses with shorter times to market. The Reduced proteolysis at low temperatures is the result of decreased overall approach is usually not straightforward due to the protease activity rather than decreased proteases production complexity of the different interactions occurring at many cell (Sirén et al., 2006). Thus, some studies have focused mainly on physiology levels. Thus, since the interactions between the cell this cultivation parameter. Data obtained in chemostat cultures factory performance and the bioprocess conditions define key on the production of an antibody Fab fragment, indicates parameters of interest (KPI) of the production, an empirical that folding stress is generally decreased at lower cultivation determination of those are required. In this sense, to obtain temperatures, therefore enabling a higher efficiency for protein this kind of information accurately, chemostats are considered processing and secretion (Dragosits et al., 2009). The specific the most usual continuous cultivation applied (Hoskisson and productivity was 3-fold higher at the lower temperature of 20◦ C, Hobbs, 2005). As pointed out in the Introduction section, observing also a reduction on the flux through the TCA cycle, variations of the continuous operating mode can be alternatively jointly with lower oxidative stress response and less presence of implemented to determine nutrient and environmental effects on molecular chaperones. metabolism and physiology. Berrios et al. (2017) examined the effect of using co-substrates Chemostat cultures provide important advantages over other with methanol in the production of Rhyzopus oryzae lipase (Rol) alternatives such as batch and fed-batch cultures. The advantages by P. pastoris in continuous cultures growing at the same µ arise from the fact that operating under steady-state conditions but at different temperature (22 or 30◦ C). Overall, the lower provides highly robust and reliable data, thus allowing one temperature led to lower specific productivities irrespective of the to assess the effect of a specific parameter while keeping co-substrate used. Strikingly, lowering temperature produced an all others constant. Therefore, several cultivation conditions increase in biomass concentration. such as medium composition, carbon and nitrogen sources, As a conclusion of the studies focused on the T effect, the dilution rates, and others can be investigated to optimize their efficiency of a low cultivation temperature depends on the target use. Moreover, some additional strain features and bioreaction protein, so this effect should be studied for each case. kinetics can be thoroughly determined using chemostat cultures. The pH of P. pastoris cultures is commonly set at 5.0–6.5 Currently, these parameters are considered of capital importance (Looser et al., 2015). pH values above 8 diminish cell viability, toward bioprocess optimization for the development of improved and can reduce protein activity and stability (Çalik et al., 2015). industrial processes. However, the optimum pH for RPP depends on the properties of Traditionally, shake-flask cultures have been used to the target protein, especially stability related with the isoelectric find optimal conditions of physical parameters (pH, T) and point (pI). Since the selection of the optimal pH should also concretely, batch or fed-batch cultures to determine oxygen consider the effect on the proteases, a pH value of 5.5 is thus requirements (DO, pO2) However, currently, continuous often used to minimize this deleterious effect. In addition, an cultivations should be considered a preferred option to assess the inappropriate pH can cause some components of the medium to effect of any culture conditions. The fact of working in steady precipitate, which should be avoided by choosing a low value in state conditions in whichever cell population is considered the working range (Cos et al., 2006). homogeneous allows to obtain very accurate and reliable results. The dissolved oxygen tension in P. pastoris cultivation is In contrast, dynamic processes such as batch of fed-batch cultures typically maintained at 20–30% (air saturation) by having are expected to be less robust and present more deviations due to the bioreactor operate at a constant agitation rate and using possible drifts in culture conditions. Furthermore, also parameter air or oxygen-enriched air as the inlet gas in order to determination such as kinetic rates and yields is also considered ensure fully aerobic conditions (especially in high-cell density to be more complex in dynamic processes. cultures). Chemostat cultures have been used to examine the Frontiers in Bioengineering and Biotechnology | www.frontiersin.org 9 June 2020 | Volume 8 | Article 632 Nieto-Taype et al. Continuous Cultivation Toward Rational Bioprocesses boosting effect of oxygen-limiting conditions on RPP. For is often the sole carbon source and inducer of protein expression example, Garcia-Ortega et al. (2017) compared different steady- (Barrigón et al., 2015). states under oxygen-limiting conditions and found the optimum Pichia strains exhibits a different phenotype regarding setting to provide a 3-fold increase in a specific production rate methanol assimilation. For MutS phenotype, cell growth is of the recombinant protein. considerably slow relative to the standard strains, which are Using as simple and inexpensive media as possible is essential designated Mut+ strains. Cell growth and productivity in to reduce fermentation and protein purification costs for efficient bioprocesses involving MutS strains can be improved by using RPP. The most common media for P. pastoris cultivation have mixed substrates in combination with a co-feeding strategy been reviewed (Cos et al., 2006; Sreekrishna, 2010). However, (Arnau et al., 2010, 2011). This increases the total concentration some of them present important problems such as unbalanced of the carbon source and the amount of energy supplied composition, formation of precipitates, high ionic strength, etc. reducing process times, as well as reducing heat production and One of the critical points in formulating cultivation media oxygen requirements relative to Mut+ bioprocesses. Glycerol is the nitrogen source. In some cases, nitrogen is only supplied and sorbitol are the most commonly used co-substrate jointly to with the feed medium; in others, it is added when pH control is methanol (Jungo et al., 2007a,c), although in repressed classical required. Nitrogen starvation produces an increase in protease systems, glycerol may lead to lower specific production rates activity which can be avoided by increasing the initial/inlet (Arnau et al., 2011; Potvin et al., 2012). New, unrepressed concentration of the nitrogen source. Protease inhibitors provide, promoters allow glycerol and glucose to be used in combination in general, a more expensive alternative method to the use of with methanol to substantially increase RPP efficiency (García- protease-deficient strains in order to decrease the presence of Ortega et al., 2019). proteolytic activity in culture broth (Shi et al., 2003; Sinha et al., The continuous strategy provides a powerful tool for assessing 2005). the potential impact of using co-substrates. The design of this Cankorur-Cetinkaya et al. (2018) developed a new interesting system is usually based on the selection of a dilution rate (D) far medium formulation that provide an appropriate balance enough from µmax , which should ensure that no accumulation between cell growth and product formation in the synthesis of substrates apart from methanol is produced. Moreover, the of human lysozyme (HuLy) and the anti-idiotypic antibody methanol concentration used should be high enough to allow 3H6 Fab in chemostat cultures of P. pastoris. Tyrosine efficient induction of the target protein, but not so high as to supplementation increased productivity, but tryptophan addition inhibit cell growth or protein production. had virtually no effect; also, phenylalanine addition increased D’Anjou and Daugulis (2001) used a substrate mixture of HuLy expression but decreased 3H6 Fab expression. methanol and glycerol with a Mut+ strain for PAOX1 -driven The effect of biotin deficiency on the growth of P. pastoris heterologous production of a sea raven antifreeze protein. producing a recombinant glycosylated avidin was studied in The specific product formation rate (qP ) and product-based chemostat culture. Replacing biotin with aspartic acid and oleic volumetric productivity (QV ) increased throughout the range acid, although they had a growth promoting effect, wash-out of dilution rates used (0.01–0.10 h−1 ). Similarly, Boze et al. happened and avidin productivity decreased. The addition of (2001) studied the production of recombinant porcine follicle- small amounts of biotin provided stable chemostat cultures on stimulating hormone (rFSH) with mixed substrates of sorbitol methanol and enabled production of biotin-free avidin (Jungo and methanol. Neither protein production nor QV or qP et al., 2007b). increased relative to the use of methanol as sole carbon source The composition of the cultivation medium also has a direct at the same dilution rate (0.01 h−1 ). However, the substrate impact on osmolarity. Dragosits et al. (2010) examined the concentration ratio was not optimized in either study. effect of osmolarity on cell physiology in wild-type and Fab Jungo et al. (2007a) studied the influence of the methanol fragment-producing strains of P. pastoris grown in carbon- fraction in the feed medium on recombinant avidin productivity limited chemostat cultures. Neither total protein nor specific and specific alcohol oxidase activity by using a glycerol–methanol Fab production was affected by osmolarity. However, proteins mixed substrate to grow a P. pastoris Mut+ strain expressing involved in energy metabolism and folding processes were recombinant avidin at a constant dilution rate D = 0.06 h−1 . With affected by an increased osmolarity (especially with the wild-type methanol fractions higher than 0.6 C-mol C-mol−1 , qP was like strain). Also, a high osmolarity decreased biomass yield due to cultures performed with methanol as sole carbon source. the increased energy maintenance requirements. Canales et al. (2015) used the production of Rhizopus oryzae lipase (Rol) as a model system to investigate the application of methanol-glycerol feeding mixtures in the P. pastoris Mut+ Carbon Sources producer clone. Cultures were grown in a simple chemostat The main substrate for a recombinant production bioprocess system and response surface methodology was used to evaluate is usually selected in terms of the promoter regulating the the effects of D and the methanol/glycerol ratio in the feed as expression. With P. pastoris and constitutive promoters such as experimental variables. The optimum conditions provided qP PGAP , glycerol and glucose are widely used as carbon sources values similar to, and QV values higher than those of processes because they result in high specific growth rates that are usually using methanol alone as substrate. Furthermore, Berrios et al. accompanied by adequate specific production rates (Garcia- (2017) compared the effect of glycerol and sorbitol as co- Ortega et al., 2013). In PAOX1 -driven systems, however, methanol substrates for methanol in continuous cultures growing at the Frontiers in Bioengineering and Biotechnology | www.frontiersin.org 10 June 2020 | Volume 8 | Article 632 Nieto-Taype et al. Continuous Cultivation Toward Rational Bioprocesses same rate. Higher QV , but lower qP and volumetric methanol processes with P. pastoris. The table shows the target protein, consumption rates were observed using glycerol as a co-substrate strain, promoter, substrate, operating strategy, and product at either 22 or 30◦ C. formation kinetics. As can be seen, chemostat cultivation was Paulová et al. (2012) used a mixture of glucose and methanol the norm, but alternative strategies based on S-stats (nutriostats) in continuous cultures of a Mut+ P. pastoris strain to obtain (Curvers et al., 2002) or DO-stats were also used (Yamawaki et al., recombinant trypsinogen. No repressive effect of glucose on 2007). methanol was observed under conditions far enough from The most common types of kinetics reported for cell growth µmax . The product was synthesized throughout the D range, (µ), substrate uptake (qS ), and product formation (qP ) are with enhanced productivity at intermediate D values that was Monod for cell growth, Pirt’s as a maintenance energy model and significantly higher than with methanol alone in the feed. Luedeking–Piret as linear µ-dependence for substrate uptake and Niu et al. (2013) used a co-feeding strategy based on product formation, respectively. However, saturation patterns, a methanol–sorbitol mixture with the aim of reducing the non-monotonic behaviors or bell-shaped types have also been high oxygen demand at large-scale operation. Thus, they used observed, especially in substrate-inhibited systems for cell growth transient continuous cultures of a Mut+ /PAOX1 -lacZ strain at a and non-constitutive RPP (Barrigón et al., 2015). fixed dilution rate. While a linear change of methanol fraction The kinetics of oxygen uptake, carbon dioxide production, was conducted, cell growth and metabolism, including O2 and substrate uptake can usually be fitted to a linear model thanks consumption, CO2 and heat production were analyzed. Based to the rather linear relationship between specific rates and the on the results, decreasing the methanol fraction in the feeding specific growth rate. These have been widely discussed in several medium reduced the specific oxygen uptake rate and maintained studies on chemostat cultures of P. pastoris (Maurer et al., 2006; maximal ß-galactosidase qP. Production was optimal with Rebnegger et al., 2014; Looser et al., 2015; Garcia-Ortega et al., methanol fractions over the range of 0.45–0.75 C-mol C-mol−1 . 2016). Although glucose and sorbitol have also been tested as co- Because production kinetics is influenced by a number of substrates jointly with methanol for Mut+ phenotype, glycerol physiological interactions, the overall balance of the different is the most selected co-substrate to improve the productivity and steps from gene transcription to product secretion is usually to reduce the specific oxygen uptake rate and heat generation in considered instead. As discussed by some authors, the correlation continuous cultures in comparison with the sole use of methanol between qP and µ is specific for each case, so it must be as a substrate. However, the methanol:glycerol ratio and the determined experimentally for each individual strain and protein operating dilution rate must be empirically optimized for each under fixed operating conditions (Looser et al., 2015). The target protein. data of Table 1 confirm that production kinetics is not only influenced by the promoter used with each strain but also, often, Bioreaction Kinetics protein-specific. For this reason, the most widely used expression The physiological characterization of the producer strains, systems (viz., that driven by methanol-inducible PAOX1 and the production rates and yields included, is currently considered constitutive glycolytic PGAP ) usually exhibit different production critical for bioprocess development. Environmental factors affect kinetics. Thus, the PAOX1 -driven expression system often exhibits cell factories in terms such as rearrangement of central carbon, non-monotonic kinetics, possibly as a consequence of the amino acid metabolism and other basic functions, all of which metabolic burden resulting from methanol uptake and/or protein have a direct impact on cell growth, folding stress, and vesicular processing, and from secreting limitations due to outstanding transport, and also potential implications in protein secretion expression levels (García-Ortega et al., 2019). In this situation, and other phenomena influencing production rates and yields maximizing bioprocess efficiency entails using intermediate µ (Baumann et al., 2010; Zahrl et al., 2017). values far from µmax . On the other hand, a growth-coupled Continuous cultivation is the most frequently used increasing correlation between µ and qP has been observed in operational mode to obtain accurate physiological data for constitutive expression systems relying on glycolytic promoters reliable characterization. By contrast, fed-batch cultures are such as GAP or PGK (Rebnegger et al., 2014; Garcia-Ortega et al., usually deemed less robust, and more laborious and time- 2016; de Macedo Robert et al., 2019), and also with the thiamine consuming (García-Ortega et al., 2019). Also, stress conditions sensitive THI11 promoter (Landes et al., 2016), qP peaking at associated to high-cell density cultures can eventually occur and higher µ values closer to µmax . render them unsuitable for accurate physiological studies. For most proteins, however, qP is related to µ by a On the other hand, the use of dynamic process conditions growth-coupled production model (D’Anjou and Daugulis, 2001; for fast physiological strain characterization in order to reduce Rebnegger et al., 2014; Garcia-Ortega et al., 2016) and even, experimental time associated to conventional fed-batch and/or possibly, by a Luedeking–Piret or µ-linear model (Curvers continuous cultivation has been reviewed (Spadiut et al., 2013). et al., 2002; Jungo et al., 2006; Khasa et al., 2007; Tang et al., Moreover, the use of transient continuous cultures in order to 2010). Other, partially growth-dependent, exhibit a µ-saturated reduce the time-consuming steady-state experiments was also production pattern (Maurer et al., 2006; Buchetics et al., 2011) implemented in the quantitative study of using mixed feeds or fit a bell-shaped model (Yamawaki et al., 2007). For example, (Jungo et al., 2007a). Yamawaki et al. (2007) found qP for scFv to exhibit either a Table 1 summarizes the continuous cultures used so far to saturation behavior on µ or bell-shaped kinetics depending on characterize the bioreaction kinetics involved in some RPP the operational mode used (continuous or fed-batch cultivation). Frontiers in Bioengineering and Biotechnology | www.frontiersin.org 11 June 2020 | Volume 8 | Article 632 Nieto-Taype et al. Continuous Cultivation Toward Rational Bioprocesses TABLE 1 | Summary of continuous cultures used as a tool toward the rational based bioprocess development with P. pastoris cell factory. Protein expressed Strain Promoter Substrate Continuous Production Optimization References strategy kinetics approach Fab fragment X-33 GAP Glucose Chemostat Growth- Optimized feeding from Maurer et al., 2006 (anti-HIV antibody coupled continuous data 2F5) µ-saturated Fab fragment X-33 GAP Glucose Chemostat Growth- Optimized feeding from Buchetics et al., 2011 (anti-HIV antibody coupled continuous data 3H6) µ-saturated Fab fragment X-33 GAP Glucose Chemostat Growth- Optimal D at Garcia-Ortega et al., (anti-HIV antibody coupled continuous mode 2016 2F5) hGM-CSF (human) GS115 Mut+ GAP Glucose Chemostat Linear Fed-batch designed Khasa et al., 2007 from continuous data phytase X-33 GAP Glucose Chemostat Linear Fed-batch designed Tang et al., 2010 from continuous data srAFP GS115 MutS AOX Glycerol/methanol Chemostat Growth- Fed-batch designed D’Anjou and Daugulis, coupled from continuous data 2001 ◦ scFv GS115 Mut+ AOX Methanol DO-stat Bell shaped Trade-off between Yamawaki et al., 2007 S-stat µ-saturated fed-batch and continuous hCTRB GS115 Mut+ AOX Methanol Nutriostat Linear Optimal D at Curvers et al., 2002 continuous mode rHV2 GS115 Mut+ AOX Methanol Chemostat Bell shaped Fed-batch designed Zhou and Zhang, 2002 from continuous data Porcine trypsinogen X-33 AOX Glucose/methanol Chemostat Bell shaped Optimal D at Paulová et al., 2012 continuous mode Avidin GS115 Mut+ AOX Methanol/glycerol/ Chemostat Linear Fed-batch designed Jungo et al., 2006, sorbitol from continuous data 2007a,c CalB X-33 PGK Glycerol Chemostat Linear Trade-off between de Macedo Robert fed-batch and et al., 2019 continuous Serum albumin X-33 THI11 Glucose Chemostat Growth- Optimized feeding from Landes et al., 2016 (human) coupled continuous data For other proteins, studies performed in chemostat (Paulová growth-dependent pattern. On the other hand, there is not a et al., 2012) or fed-batch cultures (Zhou and Zhang, 2002; common trend in reference to YP/X behavior in front of µ, Zhang et al., 2005) have shown them to fit µ non-monotonically observing different positive and negative linear relationships, increasing or bell-shaped models. bell-shaped and saturation patterns. In Table 2 a comparison of qP and YP/X according to the µ In terms of absolute values, only in the case of Fab 3H6 reached in continuous and fed-batch bioprocesses is presented. and hSA production, the specific production rate (qP ) and the This table includes results obtained with different target proteins, YP/X values were always higher in continuous mode at similar µ promoters, gene dosage, substrates, and feeding strategies. In comparing to those obtained in fed-batch cultivation. continuous mode, YP/X normally decreases when µ increases, but Additionally, since productivities could be one of the in contrast qP increases along with µ, except in the production main performance criteria from an industrial point of view, of hSA where YP/X increases when µ increases, which is volumetric productivity (QV ) has been also included in Table 2. probably associated with the high gene dosage (6 copies) (Landes Nevertheless, the comparison could not be made in a suitable way et al., 2016). In general, this linear behavior between qP and because the biomass concentration and the reactor volume may µ is related with the commonly observed µ-dependence in P. not be equivalent in the two operating modes. Moreover, for a pastoris bioprocesses, which indicates that the product formation proper comparison it had to be also considered a time window is growth-dependent. of several weeks in order to take into account the unproductive When the data obtained in both, continuous and fed-batch time between consecutive fed-batches. For instance, taking these mode are compared, the analysis is often difficult because not factors into consideration, overall CalB production was reported all the examples have been studied for coincident µ’s. However, as 5.8 times greater in continuous culture (de Macedo Robert the qP behavior is quite similar for the two operating modes, et al., 2019). only for the case of rHV2 and Crl1 production regulated by The results of this experimental approach, which is mostly PAOX1, which qP presents a maximum at intermediate µ for performed in chemostat cultures, provide a wealth of knowledge the fed-batch cultivation not following a product formation to tailor rational strategies with a view to optimizing bioprocess Frontiers in Bioengineering and Biotechnology | www.frontiersin.org 12 June 2020 | Volume 8 | Article 632 Nieto-Taype et al. Continuous Cultivation Toward Rational Bioprocesses TABLE 2 | Comparison of specific production rates (qP ), product yields (YP/X ), and volumetric productivities (Qv ) between continuous and fed-batch operating modes with different proteins produced, promoters/gene dosage, substrates, specific growth rates (µ), and feeding strategies used. Continuous Fed-batch References Protein Promoter/gene Substrate µ qP YP/X Qv µ qP YP/X Qv dosage h−1 AU/(gX ·h) AU/gX AU/(L·h) h−1 AU/(gX ·h) AU/gX AU/(L·h) Crl1 PGAP Glucose 0.056 78.2 1,400 1,924 0.048 77 1,610 2,195 Nieto-Taype et al., Sc 0.081 89.1 1,093 2,410 0.087 149 1,700 3,246 2020 0.136 126 930 3,108 0.133 292 2,190 10,442 Crl1 PAOX MetOH 0.020 118 6,017 2,291 0.028 364 13,155 2,291 Garrigós-Martínez sc 0.050 271 5,398 5,564 0.053 588 11,299 5,564 et al., 2019 0.079 310 3,939 6,874 0.084 326 3,911 6,874 CalB PPGK Glucose 0.050 20.4 449 449 0.060 19.4 323 402 de Macedo Robert Sc 0.090 38.4 423 864 0.110 47.5 423 796 et al., 2019 0.160 59.1 377 1,331 0.140 52.3 377 868 Phytase PGAP Glucose 0.050 480 9,600 nd DFP 0.118 567 4,800 20,900 Tang et al., 2010 0.150 820 5,500 nd DFP 0.125 586 4,680 17,400 0.300 1,420 4,700 nd DFP 0.125 628 5,020 18,700 Protein Promoter/gene Substrate µ qP YP/X Qv µ qP YP/X Qv References dosage h−1 µg/(gX ·h) mg/gX mg/(L·h) h−1 µg/(gX ·h) mg/gX mg/(L·h) Human PGAP Glucose 0.050 17.2 0.33 0.34 0.050 7.7 0.15 0.24 Garcia-Ortega et al., FAB 2F5 sc 0.100 32.4 0.32 0.70 0.101 25.7 0.25 0.64 2013, 2016 0.150 45.7 0.30 1.04 0.146 35.2 0.24 0.73 Human PGAP Glucose 0.060 25.1 0.42 0.62 CFR 10.0 0.48 0.39 Maurer et al., 2006 FAB 2F5 sc 0.100 34.0 0.34 0.94 0.150 41.4 0.28 1.14 OFR 30.0 0.48 0.67 Human PGAP Glucose 0.190 45.6 0.24 1.38 FAB 3H6 sc 0.050 48.1 0.96 1.32 CLB2 0.100 60.5 0.60 1.66 OFR 10 0.45 nd Buchetics et al., 0.150 66.1 0.44 1.82 2011 rHV2 PAOX MetOH 0.4 g/L 0.017 nd nd nd 0.01 0.03 3.0 nd MetOH 0.6 g/L 0.020 nd nd nd 0.02 0.20 9.8 nd Zhou and Zhang, MetOH 1.5 g/L 0.040 nd nd nd 0.04 0.08 2.0 nd 2002 CLC 0.018 0.17 9.6 14 hSA PTHI11 Glucose 0.050 0.003 0.06 0.05 6 copies 0.075 0.017 0.22 0.30 Landes et al., 2016 0.100 0.050 0.50 1.00 0.125 0.119 0.95 2.38 0.150 0.180 1.20 3.60 0.150 0.023 0.88 2.60 Avidin PAOX *MS-MS 6 0.03 0.003 0.11 0.03 (% Methanol 25 0.03 0.013 0.42 0.13 C-mol) 43 0.03 0.016 0.54 0.16 0.027 0.049 1.78 3.24 Jungo et al., 2007c 65 0.03 0.016 0.52 0.13 100 0.03 0.016 0.54 0.12 SC: Single Copy; nd: not determined; All the fed-batch cultures were carried out by implementing open-loop feeding strategies except: DFP, Different Feeding Profile; CFR, Constant Feed Rate; OFR, Optimized Feed Rate; CLC, Closed Loop Control; *Mixed-substrates Methanol-Sorbitol. Some values have been estimated from the values indicated in the references or directly from the figures. For rHV2, hSA, and avidin, qP is expressed as mg/(gx ·h). operating conditions for increased production rates and Bioprocess Optimization and Process yields. Although production patterns are expected to be Development similar when using the same expression system to produce Most of the bioprocesses currently used for recombinant protein different target proteins, the outcome usually depends on or metabolite production are performed in the fed-batch mode both. Therefore, optimal bioprocess development requires using on the grounds of the high cell concentrations it affords, and similar experiments not only to maximize production rates, but the increased amounts of product obtained as a result. Therefore, also to identify the most suitable conditions for testing other using the data obtained from chemostats with fed-batch cultures strains with industrial potential. Chemostats therefore provide an requires that the latter operate in a pseudo-stationary state or effective platform for easily characterizing production strains and balanced growth in order to maintain the overall µ value rather elucidating production kinetics by the obtention of accurate and constant during the process. This can be accomplished by using a robust data. fed-batch strategy based on pre-programed exponential feeding Frontiers in Bioengineering and Biotechnology | www.frontiersin.org 13 June 2020 | Volume 8 | Article 632 Nieto-Taype et al. Continuous Cultivation Toward Rational Bioprocesses derived from mass balances. Under these conditions, if µ is specific clones, strains or bioreactors (Looser et al., 2015). This kept constant and the biomass-to-substrate yield rather constant requires combining effective strains and bioprocess engineering throughout the cultivation period, the substrate can be assumed approaches for optimal results, and also avoiding complex media to reach a pseudo-stationary state. to reduce costs and facilitate downstream processes with a view Table 1 summarizes works with continuous cultures as a to increasing economic viability of the bioprocess (Sreekrishna, tool for further rational process development based on the 2010; Potvin et al., 2012). In this sense, due to the relevant physiological characterization of the producer strains. As can advantages of the continuous manufacturing, as once pioneered be seen, three main approaches have been used for optimal in the oil and chemical industry, many industrial bioprocesses RPP. One involves identifying the optimum D conditions are expected to eventually evolve from batch or fed-batch to under continuous operation for key parameters of interest continuous processes (Croughan et al., 2015). (KPIs) including productivities (Curvers et al., 2002; Paulová Currently, continuous biomanufacturing processes based on et al., 2012). Another uses a trade-off between fed-batch and the application of continuous fermentation are being carried continuous operation for yields or productivities depending on out in different industries including ethanol, lactic acid, and the particular design or performance criteria (Yamawaki et al., numerous biopharmaceutical productions. Mears et al. (2017) 2007). The third is a sequential approach to bioprocess design discussed some of these examples performed with continuous (i.e., fed-batch operation is designed and implemented on the microbial systems at industrial scale to evaluate the expected basis of production kinetics or productivity data obtained under wide implementation in biopharmaceutical biomanufacturing continuous operation) (Zhou and Zhang, 2002; Jungo et al., processes. Actually, the US FDA has encouraged the development 2007a,c; Khasa et al., 2007; Tang et al., 2010). of continuous processing for biopharmaceuticals manufacturing Garcia-Ortega et al. (2016) characterized the recombinant (Croughan et al., 2015). In this sense, it is worth mentioning that production of Fab fragment 2F5 driven by the constitutive Novo Nordisk implemented the continuous insulin production promoter PGAP in chemostat cultures. A positive effect of high with the yeast S. cerevisiae since the 1990’s (Peebo and Neubauer, µ levels on productivity was observed, which confirmed that 2018). And currently, several commercial biopharmaceuticals production was growth-coupled. such as monoclonal antibodies or therapeutic proteins are de Macedo Robert et al. (2019) compared Candida antarctica recombinantly produced in continuous perfusion cultures of lipase (CalB) production with the constitutive promoter PGK mammalian cell lines (Mears et al., 2017). in continuous and fed-batch cultures. They found volumetric Industrially, switching from batch to continuous operation and specific productivity to peak at the highest µ levels, which can have several advantages such as reduced processing costs, indicated a direct correlation between growth and production. increased productivity, and product quality, and the ability to Production kinetic profiles for other proteins have also integrate upstream and downstream in a continuous manner been obtained in other continuous cultures that were used (Cankorur-Cetinkaya et al., 2018). However, bottlenecks in as references to develop optimal feeding strategies adjusted to industrial bioprocesses are generally product-specific and must cell factory performance. The result was a marked increase in be identified on a case-by-case basis with provision for economic production in terms of volumetric productivity and product titer constraints (Yang and Zhang, 2018). for some Fab fragments (Maurer et al., 2006; Buchetics et al., Continuous cultures growing under steady-state conditions 2011) and human serum albumin (Landes et al., 2016). are, in theory, more effective in this mode than are batch or D’Anjou and Daugulis (2001) used glycerol–methanol fed-batch cultures. In this mode, which is usually performed mixtures as co-substrates for continuous cultures to obtain at a constant µ, most of the factors involved in the RPP sea raven antifreeze protein (srAFP). They elucidated the as well as a large fraction of the transcriptome, proteome relationship of µ to the yield on methanol, qS on methanol and and fluxome can be considered constant, since it has been qP and then used data collected under continuous operation to reported that they are coordinated by µ (Peebo and Neubauer, predict cell growth and RPP, as well as to develop an exponential 2018). Therefore, continuous cultivations where µ can be feeding strategy for fed-batch cultures with the same two easily fixed, simplifies significantly the interpretation and carbon sources. The outcome was increased product yield and control of the bioprocess. In this way, bioprocesses should be productivity relative to a heuristic approach. easier to characterize, understand, control and maintain under optimal conditions over long periods. Furthermore, the lag IS CONTINUOUS CULTIVATION A REAL and dead times, intrinsically associated to batch and fed-batch operation, result in considerably diminished productivity. Hence, ALTERNATIVE FOR ITS USE IN continuous biomanufacturing processes allow reducing running INDUSTRIAL PICHIA PASTORIS costs, minimizing equipment size, integrating upstream and PROCESSES? downstream steps, increasing product quality, ensuring constant product recovery, and making processes scalable are additional The Transition of Industrial Biotechnology advantages of continuous cultures over batch and fed-batch From Batch/Fed-Batch Mode to cultures (Peebo and Neubauer, 2018). Continuous Manufacturing However, continuous cultivation also has some drawbacks The main current trend in bioprocess optimization is to move arising from the need to ensure long-term stability and sterility away from standard, fixed protocols toward concepts enabling in the cultures, and to shorten average long development times the operator to adapt particular recommendations to any (Croughan et al., 2015). Specifically, the genetic instability of Frontiers in Bioengineering and Biotechnology | www.frontiersin.org 14 June 2020 | Volume 8 | Article 632 Nieto-Taype et al. Continuous Cultivation Toward Rational Bioprocesses recombinant strains may limit the length of RPP bioprocesses; and biochemicals as a real alternative to Saccharomyces cerevisiae. especially for with P. pastoris strains. Clone stability can be Furthermore, Moser et al. (2017) investigated the impact of long- confirmed by testing productivity over a large number of term carbon source adaptation toward improved cell growth generations under steady states (about 20–30 residence times) in in RPP processes with P. pastoris in the context of Adaptive chemostat cultures growing under a steady state. Alternatively, Laboratory Evolution (ALE). Adaptation of the yeast to growth it can also be assured with periodic changes in dilution rate on methanol over 250 generations was examined and observed or the aeration conditions (Cankorur-Cetinkaya et al., 2018). a complex correlation among carbon source, cell growth and Additionally, and concretely for multi-copy strains, the gene RPP. Increased specific growth rates on rich and minimal growth dosage conservation in continuous cultures maintained over long media was studied at the level of both population and single periods is important to be checked. clone. Selected clones displayed strain-dependent variations for the yield of PAOX1 -based recombinant protein expression, one Challenges to Be Addressed in showing up to 2-fold increase in terms product titer. Pichia Pastoris Continuous Cultivations From a bioprocess engineering perspective, process variability An important requirement with continuous bioprocesses is to often results in issues arising from changes in critical process develop effective alternative promoters to replace the classical parameters (CPPs), key parameters of interest (KPIs), or critical PAOX1, which is widely used efficiently in fed-batch processes material attributes (CMAs), all of which in turn affect critical but is not recommended for continuous RPP. In fact, identifying quality attributes (CQAs). CPPs can change throughout a culture strong and constitutively expressed promoters, and testing and require using appropriate control strategies to reduce process them in continuous cultures currently still remains a major variability. Advances in process analytical technology (PAT) goal (Cankorur-Cetinkaya et al., 2018). Obviously, efficiently and quality by design (QbD) approaches have substantially exploiting the potential advantages of continuous cultivation reduced times for process development. According to Hernandez requires integrating upstream and downstream bioprocess flows. (2015), in the context of transition to continuous bioprocesses Thus, the implementation of new approaches is a challenge it is essential to adapt these methodologies from batch to for biotechnology companies in order to reach short process continuous cultures. development times jointly with minimal risks in the context of It has been stated that continuous bioprocesses require real- possible hard quality/regulatory requirements (Rathore et al., time monitoring of process variability in order to identify 2015). At this point a question arises: is continuous cultivation eventual deviations and to ensure consistent performance a real alternative to fed-batch cultures for its use in industrial (Rajamanickam et al., 2018). Variability in KPIs (e.g., cell growth, biotechnology based on the Pichia pastoris cell factory? yields, production rates) is especially important because it has Gasser and Mattanovich (2018) reported some interesting a direct impact on process performance and product quality. guidelines to transform this yeast from an efficient cell factory Thus, continuous bioprocesses require real-time monitoring with to a useful chassis for the production of recombinant proteins standard sensors for direct measurements, or soft sensors that FIGURE 3 | Mean features in the industrial implementation of Pichia pastoris continuous cultivation. Frontiers in Bioengineering and Biotechnology | www.frontiersin.org 15 June 2020 | Volume 8 | Article 632 Nieto-Taype et al. Continuous Cultivation Toward Rational Bioprocesses allow the indirect determination of parameters ensuring a high A chemostat process producing recombinant hepatitis B bioprocess efficiency. Recent advances in the monitoring of small surface antigen (rHBsAg) was maintained working for various cell factories, including soft-sensors, have been reviewed 2 weeks reaching similar levels of product titer than at the by several authors (Valero and López-Santín, 2017; Veloso and end of a fed-batch culture. Specifically, QV and qP were, Ferreira, 2017; Randek and Mandenius, 2018). Nevertheless, respectively, about 1.5 and 1.3 times higher, than in fed- most of the reported monitoring applications for P. pastoris have batch mode. Interestingly, no contamination issues or/and been described for batch and fed-batch operations. Besides the genome instability were detected after those 2 weeks of common sensors used for the monitoring and control of culture continuous operation (Rahimi et al., 2019). In another study, the conditions and gas analysis, just scarce applications have been performance of continuous cultures designed as a combination reported for continuous mode (Jungo et al., 2007a; Fazenda et al., of turbidostat/chemostat modes at constant cell concentration 2013). In this scenario, further implementing PAT and QbD in was evaluated (Wang et al., 2012). As a relevant outcome, it P. pastoris continuous cultures has been deemed essential for was reported that feeding the system with methanol resulted in industrial processes. high recombinant polygalacturonate lyase PGL expression and a substantial performance improvement relative to conventional fed-batch cultures. Moreover, a chemostat producing lipase B How Far Is the Implementation of from Candida antarctica (CalB) was running 6 weeks at µ = Pichia pastoris Continuous Manufacturing 0.14 h−1 . In this case, considering the unproductive time spent in Processes? setting up, draining out, cleaning and reassembling in fed-batch Several RPP industrial processes are currently performed operation, the overall CalB production was 5.8 times greater than with P. pastoris. Including both industrial enzymes and the fed-batch process (de Macedo Robert et al., 2019). biopharmaceuticals, the www.pichia.com website provides a Finally, a summary including the main items required to list of products manufactured with Pichia which are on the address toward the industrial implementation of continuous market of late stage development. However, according to the cultivation with the cell factory P. pastoris is presented in open literature, there are no reported in detail continuous Figure 3. biomanufacturing processes implemented with P. pastoris. Nevertheless, taking into consideration the state of the art CONCLUSIONS discussed in this section, some long terms continuous examples described in the literature could eventually become interesting The yeast P. pastoris is considered an outstanding cell factory candidates for continuous manufacturing with P. pastoris. alternative for its industrial use in production processes of both FIGURE 4 | Use of continuous cultivation as a tool toward the rational bioprocess development with Pichia pastoris cell factory by combining strain and bioprocess engineering. Frontiers in Bioengineering and Biotechnology | www.frontiersin.org 16 June 2020 | Volume 8 | Article 632 Nieto-Taype et al. Continuous Cultivation Toward Rational Bioprocesses metabolites and recombinant proteins. Despite the excellent challenges with the identification of efficient and methanol- features of this host, in general, most of the bioprocesses free alternative promoters, the integration of upstream and based on fermentation technology are considered still far downstream, reaching short process development times with from achieving the maturity of other industries. In this sense, minimal risks on quality/regulatory requirements through PAT the present trends toward their optimization are provided and QbD approaches and, finally, to evaluate the impact by two complementary approaches: Strain development and of long-term continuous operation on genetic stability and bioprocess engineering. In order to achieve relevant progresses cell physiology. in the long road to bioprocess optimization it is essential to acquire robust and reliable knowledge of the system used. This AUTHOR CONTRIBUTIONS knowledge is considered with the objective to be able to design and implement rational developments from any of the both MN-T and XG-O contributed writing the Abstract and the approaches mentioned. In this context, continuous cultivation Introduction sections. MN-T and JA wrote the section: emerges as an excellent tool to accurately characterize the Continuous cultivations provide an excellent tool for systems performance of the cell factory in terms of physiological and biology research. JM-S and FV wrote the section: Bioprocess production parameters. Furthermore, the steady state conditions characterization with chemostat cultures. XG-O, JM-S, and FV achieved during the continuous cultures and the high degree wrote the section: Is continuous cultivation a real alternative for of cell population homogeneity makes this operational mode, its use in industrial Pichia pastoris processes? XG-O, JM-S, and an excellent tool toward the development of a wide range of FV wrote the Conclusions section. All authors read, reviewed, systems biology studies including transcriptomics, proteomics, and approved the submitted version. metabolomics and other cutting edge “-omics” studies. In Figure 4 is summarized how continuous cultivation can be FUNDING used as a tool toward the rational development of bioprocesses with P. pastoris. It illustrates the different approaches that, based This work was funded by MINECO and FEDER under Project on the results obtained in continuous cultivation (chemostat), CTQ2016-74959-R. The authors’ group is member 2017-SGR- allow to achieve rational improvements at both strain and 1462 and the Reference Network in Biotechnology (XRB) of bioprocess engineering level. Generalitat de Catalunya. Besides, the current trends of industrial biotechnology is promoting the transition from batch/fed-batch-based processes ACKNOWLEDGMENTS toward the continuous biomanufacturing due to the numerous advantages that present this operation mode in terms of MN-T acknowledges award by the National Council of Science ensuring a robust and very high product quality as well Technology and Technological Innovation (CONCYTEC) as, among others, to reduce running costs and minimizing through its executing unit and the National Fund for equipment requirements. The potential implementation of Scientific, Technological and Technological Innovation biomanufacturing processes with P. pastoris has the most relevant Development (FONDECYT). REFERENCES promoter for improved heterologous protein production in Pichia pastoris. Biotechnol. Bioeng. 114, 2319–2327. doi: 10.1002/bit.26363 Adamberg, K., Valgepea, K., and Vilu, R. (2015). Advanced continuous Ata, Ö., Rebnegger, C., Tatto, N. E., Valli, M., Mairinger, T., Hann, S., et al. cultivation methods for systems microbiology. Microbiology 161, 1707–1719. (2018). A single Gal4-like transcription factor activates the Crabtree effect doi: 10.1099/mic.0.000146 in Komagataella phaffii. Nat. Commun. 9:4911. doi: 10.1038/s41467-018- Adelantado, N., Tarazona, P., Grillitsch, K., García-Ortega, X., Monforte, S., Valero, 07430-4 F., et al. (2017). The effect of hypoxia on the lipidome of recombinant Pichia Baghban, R., Farajnia, S., Rajabibazl, M., Ghasemi, Y., Mafi, A., Hoseinpoor, R., pastoris. Microb. Cell Fact. 16:86. doi: 10.1186/s12934-017-0699-4 et al. (2019). Yeast expression systems: overview and recent advances. Mol. Ahmad, M., Hirz, M., Pichler, H., and Schwab, H. (2014). Protein Biotechnol. 61, 365–384. doi: 10.1007/s12033-019-00164-8 expression in Pichia pastoris: recent achievements and perspectives for Barrigón, J. M., Valero, F., and Montesinos, J. L. (2015). A macrokinetic heterologous protein production. Appl. Microbiol. Biotechnol. 98, 5301–5317. model-based comparative meta-analysis of recombinant protein production doi: 10.1007/s00253-014-5732-5 by Pichia pastoris under AOX1 promoter. Biotechnol. Bioeng. 112, 1132–1145. Arnau, C., Casas, C., and Valero, F. (2011). The effect of glycerol mixed substrate doi: 10.1002/bit.25518 on the heterologous production of a Rhizopus oryzae lipase in Pichia pastoris Baumann, K., Carnicer, M., Dragosits, M., Graf, A. B., Stadlmann, J., Jouhten, P., system. Biochem. Eng. J. 57, 30–37. doi: 10.1016/j.bej.2011.08.004 et al. (2010). A multi-level study of recombinant Pichia pastoris in different Arnau, C., Ramon, R., Casas, C., and Valero, F. (2010). Optimization of the oxygen conditions. BMC Syst. Biol. 4:141. doi: 10.1186/1752-0509-4-141 heterologous production of a Rhizopus oryzae lipase in Pichia pastoris system Baumann, K., Dato, L., Graf, A. B., Frascotti, G., Dragosits, M., Porro, D., using mixed substrates on controlled fed-batch bioprocess. Enzyme Microb. et al. (2011). The impact of oxygen on the transcriptome of recombinant S. Technol. 46, 494–500. doi: 10.1016/j.enzmictec.2010.01.005 cerevisiae and P. pastoris - a comparative analysis. BMC Genomics 12:218. Arvas, M., Pakula, T., Smit, B., Rautio, J., Koivistoinen, H., Jouhten, P., et al. (2011). doi: 10.1186/1471-2164-12-218 Correlation of gene expression and protein production rate - a system wide Baumann, K., Maurer, M., Dragosits, M., Cos, O., Ferrer, P., and Mattanovich, study. BMC Genomics 12:616. doi: 10.1186/1471-2164-12-616 D. (2008). Hypoxic fed-batch cultivation of Pichia pastoris increases specific Ata, Ö., Prielhofer, R., Gasser, B., Mattanovich, D., and Çalik, P. (2017). and volumetric productivity of recombinant proteins. Biotechnol. Bioeng. 100, Transcriptional engineering of the glyceraldehyde-3-phosphate dehydrogenase 177–183. doi: 10.1002/bit.21763 Frontiers in Bioengineering and Biotechnology | www.frontiersin.org 17 June 2020 | Volume 8 | Article 632 Nieto-Taype et al. Continuous Cultivation Toward Rational Bioprocesses Berrios, J., Flores, M.-O., Díaz-Barrera, A., Altamirano, C., Martínez, I., and D’Anjou, M. C., and Daugulis, A. J. (2001). A rational approach to improving Cabrera, Z. (2017). A comparative study of glycerol and sorbitol as co- productivity in recombinant Pichia pastoris fermentation. Biotechnol. Bioeng. substrates in methanol-induced cultures of Pichia pastoris: temperature 72, 1–11. doi: 10.1002/1097-0290(20010105)72:1<1::AID-BIT1>3.0.CO;2-T effect and scale-up simulation. J. Ind. Microbiol. Biotechnol. 44, 407–411. de Macedo Robert, J., Garcia-Ortega, X., Montesinos-Seguí, J. L., Guimaraes Freire, doi: 10.1007/s10295-016-1895-7 D. M., and Valero, F. (2019). Continuous operation, a realistic alternative to fed- Boze, H., Laborde, C., Chemardin, P., Richard, P., Venturin, C., Combarnous, batch fermentation for the production of recombinant lipase B from Candida Y., et al. (2001). High-level secretory production of recombinant porcine antarctica under the constitutive promoter PGK in Pichia pastoris. Biochem. follicle-stimulating hormone by Pichia pastoris. Process Biochem. 36, 907–913. Eng. J. 147, 39–47. doi: 10.1016/j.bej.2019.03.027 doi: 10.1016/S0032-9592(00)00296-X Delic, M., Valli, M., Graf, A. B., Pfeffer, M., Mattanovich, D., and Gasser, B. (2013). Brejning, J., Arneborg, N., and Jespersen, L. (2005). Identification of genes The secretory pathway: exploring yeast diversity. FEMS Microbiol. Rev. 37, and proteins induced during the lag and early exponential phase of lager 872–914. doi: 10.1111/1574-6976.12020 brewing yeasts. J. Appl. Microbiol. 98, 261–271. doi: 10.1111/j.1365-2672.2004.0 Dragosits, M., Frascotti, G., Bernard-Granger, L., Vázquez, F., Giuliani, M., 2472.x Baumann, K., et al. (2011). Influence of growth temperature on the production Buchetics, M., Dragosits, M., Maurer, M., Rebnegger, C., Porro, D., Sauer, M., et al. of antibody Fab fragments in different microbes: a host comparative analysis. (2011). Reverse engineering of protein secretion by uncoupling of cell cycle Biotechnol. Prog. 27, 38–46. doi: 10.1002/btpr.524 phases from growth. Biotechnol. Bioeng. 108, 2403–2412. doi: 10.1002/bit.23198 Dragosits, M., Stadlmann, J., Albiol, J., Baumann, K., Maurer, M., Gasser, B., Bull, A. T. (2010). The renaissance of continuous culture in the post-genomics age. et al. (2009). The effect of temperature on the proteome of recombinant Pichia J. Ind. Microbiol. Biotechnol. 37, 993–1021. doi: 10.1007/s10295-010-0816-4 pastoris. J. Proteome Res. 8, 1380–1392. doi: 10.1021/pr8007623 Çalik, P., Ata, Ö., Güneş, H., Massahi, A., Boy, E., Keskin, A., et al. Dragosits, M., Stadlmann, J., Graf, A., Gasser, B., Maurer, M., Sauer, M., et al. (2015). Recombinant protein production in Pichia pastoris under (2010). The response to unfolded protein is involved in osmotolerance of Pichia glyceraldehyde-3-phosphate dehydrogenase promoter: from carbon source pastoris. BMC Genomics 11:207. doi: 10.1186/1471-2164-11-207 metabolism to bioreactor operation parameters. Biochem. Eng. J. 95, 20–36. Engstrom, M. D., and Pfleger, B. F. (2017). Transcription control engineering doi: 10.1016/j.bej.2014.12.003 and applications in synthetic biology. Synth. Syst. Biotechnol. 2, 176–191. Cámara, E., Landes, N., Albiol, J., Gasser, B., Mattanovich, D., and Ferrer, P. (2017). doi: 10.1016/j.synbio.2017.09.003 Increased dosage of AOX1 promoter-regulated expression cassettes leads to Ercan, O., Bisschops, M. M. M., Overkamp, W., Jørgensen, T. R., Ram, A. F., Smid, transcription attenuation of the methanol metabolism in Pichia pastoris. Sci. E. J., et al. (2015a). Physiological and transcriptional responses of different Rep. 7:44302. doi: 10.1038/srep44302 industrial microbes at near-zero specific growth rates. Appl. Environ. Microbiol. Cámara, E., Monforte, S., Albiol, J., and Ferrer, P. (2019). Deregulation of 81, 5662–5670. doi: 10.1128/AEM.00944-15 methanol metabolism reverts transcriptional limitations of recombinant Pichia Ercan, O., Wels, M., Smid, E. J., and Kleerebezem, M. (2015b). Molecular and pastoris (Komagataella spp) with multiple expression cassettes under control metabolic adaptations of Lactococcus lactis at near-zero growth rates. Appl. of the AOX1 promoter. Biotechnol. Bioeng. 116, 1710–1720. doi: 10.1002/ Environ. Microbiol. 81, 320–331. doi: 10.1128/AEM.02484-14 bit.26947 Fazenda, M. L., Dias, J. M., Harvey, L. M., Nordon, A., Edrada-Ebel, R., LittleJohn, Canales, C., Altamirano, C., and Berrios, J. (2015). Effect of dilution rate and D., et al. (2013). Towards better understanding of an industrial cell factory: methanol-glycerol mixed feeding on heterologous Rhizopus oryzae lipase investigating the feasibility of real-time metabolic flux analysis in Pichia production with Pichia pastoris Mut + phenotype in continuous culture. pastoris. Microb. Cell Fact. 12:51. doi: 10.1186/1475-2859-12-51 Biotechnol. Prog. 31, 707–714. doi: 10.1002/btpr.2069 Fernandes, B. D., Mota, A., Teixeira, J. A., and Vicente, A. A. (2015). Cankorur-Cetinkaya, A., Narraidoo, N., Kasavi, C., Slater, N. K. H., Archer, D. B., Continuous cultivation of photosynthetic microorganisms: approaches, and Oliver, S. G. (2018). Process development for the continuous production of applications and future trends. Biotechnol. Adv. 33, 1228–1245. heterologous proteins by the industrial yeast, Komagataella phaffii. Biotechnol. doi: 10.1016/j.biotechadv.2015.03.004 Bioeng. 115, 2962–2973. doi: 10.1002/bit.26846 Ferrer, P., and Albiol, J. (2014a). “13C-based metabolic flux analysis in yeast: the Carnicer, M., ten Pierick, A., van Dam, J., Heijnen, J. J., Albiol, J., van Gulik, W., Pichia pastoris case,” in Methods in Molecular Biology, ed V. Mapelli (New York, et al. (2012). Quantitative metabolomics analysis of amino acid metabolism NY: Humana Press Inc.), 209–232. doi: 10.1007/978-1-4939-0563-8_13 in recombinant Pichia pastoris under different oxygen availability conditions. Ferrer, P., and Albiol, J. (2014b). “13C-Based metabolic flux analysis of Microb. Cell Fact. 11:83. doi: 10.1186/1475-2859-11-83 recombinant Pichia pastoris,” in Metabolic Flux Analysis, eds J. O. Krömer, L. Çelik, E., and Çalik, P. (2012). Production of recombinant proteins by yeast cells. K. Nielsen, and L. M. Blank (New York, NY: Humana Press Inc.), 291–313. Biotechnol. Adv. 30, 1108–1118. doi: 10.1016/j.biotechadv.2011.09.011 doi: 10.1007/978-1-4939-1170-7_17 Chung, B. K., Selvarasu, S., Andrea, C., Ryu, J., Lee, H., Ahn, J., et al. Fong, S. S. (2014). Computational approaches to metabolic engineering utilizing (2010). Genome-scale metabolic reconstruction and in silico analysis of systems biology and synthetic biology. Comput. Struct. Biotechnol. J. 11, 28–34. methylotrophic yeast Pichia pastoris for strain improvement. Microb. Cell Fact. doi: 10.1016/j.csbj.2014.08.005 9:50. doi: 10.1186/1475-2859-9-50 Garcia-Ortega, X., Adelantado, N., Ferrer, P., Montesinos, J. L., and Valero, Cooney, C. L. (1979). Continuous culture of microorganisms: an F. (2016). A step forward to improve recombinant protein production in overview and perspective. Ann. N.Y. Acad. Sci. 326, 295–314. Pichia pastoris: from specific growth rate effect on protein secretion to doi: 10.1111/j.1749-6632.1979.tb14170.x carbon-starving conditions as advanced strategy. Process Biochem. 51, 681–691. Cos, O., Ramón, R., Montesinos, J. L., and Valero, F. (2006). Operational doi: 10.1016/j.procbio.2016.02.018 strategies, monitoring and control of heterologous protein production in the García-Ortega, X., Cámara, E., Ferrer, P., Albiol, J., Montesinos-Seguí, J. L., and methylotrophic yeast Pichia pastoris under different promoters: a review. Valero, F. (2019). Rational development of bioprocess engineering strategies for Microb. Cell Fact. 5:17. doi: 10.1186/1475-2859-5-17 recombinant protein production in Pichia pastoris (Komagataella phaffi) using Cregg, J. M., Vedvick, T. S., and Raschke, W. C. (1993). Recent advances in the the methanol free GAP promoter. Where do we stand? N. Biotechnol. 53, 24–34. expression of foreign genes in Pichia pastoris. Nat. Biotechnol. 11, 905–910. doi: 10.1016/j.nbt.2019.06.002 doi: 10.1038/nbt0893-905 Garcia-Ortega, X., Ferrer, P., Montesinos, J. L., and Valero, F. (2013). Fed- Croughan, M. S., Konstantinov, K. B., and Cooney, C. (2015). The future batch operational strategies for recombinant Fab production with Pichia of industrial bioprocessing: batch or continuous? Biotechnol. Bioeng. 112, pastoris using the constitutive GAP promoter. Biochem. Eng. J. 79, 172–181. 648–651. doi: 10.1002/bit.25529 doi: 10.1016/j.bej.2013.07.013 Curvers, S., Linnemann, J., Klauser, T., Wandrey, C., and Takors, R. Garcia-Ortega, X., Valero, F., and Montesinos-Seguí, J. L. (2017). Physiological (2002). Recombinant protein production with Pichia pastoris in continuous state as transferable operating criterion to improve recombinant protein fermentation – kinetic analysis of growth and product formation. Eng. Life Sci. production in Pichia pastoris through oxygen limitation. J. Chem. Technol. 2:229. doi: 10.1002/1618-2863(20020806)2:8<229::AID-ELSC229>3.0.CO;2-9 Biotechnol. 92, 2573–2582. doi: 10.1002/jctb.5272 Frontiers in Bioengineering and Biotechnology | www.frontiersin.org 18 June 2020 | Volume 8 | Article 632 Nieto-Taype et al. Continuous Cultivation Toward Rational Bioprocesses Garrigós-Martínez, J., Nieto-Taype, M. A., Gasset, A., Montesinos-Seguí, flux analysis reveals impact of recombinant protein production on J. L., Garcia-Ortega, X., and Valero, F. (2019). Specific growth rate trehalose and energy metabolism in Pichia pastoris. Metabolites 4, 281–299. governs AOX1 gene expression, affecting the production kinetics of doi: 10.3390/metabo4020281 Pichia pastoris (Komogataella phaffii) PAOX1-driven recombinant producer Jungo, C., Marison, I., and von Stockar, U. (2007a). Mixed feeds of glycerol strains with different target gene dosage. Microb. Cell Fact. 181:87. and methanol can improve the performance of Pichia pastoris cultures: a doi: 10.1186/s12934-019-1240-8 quantitative study based on concentration gradients in transient continuous Gasser, B., and Mattanovich, D. (2018). A yeast for all seasons – is Pichia pastoris cultures. J. Biotechnol. 128, 824–837. doi: 10.1016/j.jbiotec.2006.12.024 a suitable chassis organism for future bioproduction? FEMS Microbiol. Lett. Jungo, C., Rérat, C., Marison, I. W., and von Stockar, U. (2006). Quantitative 365:181. doi: 10.1093/femsle/fny181 characterization of the regulation of the synthesis of alcohol oxidase and of Gasser, B., Maurer, M., Rautio, J., Sauer, M., Bhattacharyya, A., Saloheimo, the expression of recombinant avidin in a Pichia pastoris Mut+ strain. Enzyme M., et al. (2007). Monitoring of transcriptional regulation in Pichia Microb. Technol. 39, 936–944. doi: 10.1016/j.enzmictec.2006.01.027 pastoris under protein production conditions. BMC Genomics 8:179. Jungo, C., Schenk, J., Pasquier, M., Marison, I. W., and von Stockar, U. (2007c). A doi: 10.1186/1471-2164-8-179 quantitative analysis of the benefits of mixed feeds of sorbitol and methanol for Gasser, B., Prielhofer, R., Marx, H., Maurer, M., Nocon, J., Steiger, M., the production of recombinant avidin with Pichia pastoris. J. Biotechnol. 131, et al. (2013). Pichia pastoris: protein production host and model organism 57–66. doi: 10.1016/j.jbiotec.2007.05.019 for biomedical research. Future Microbiol. 8, 191–208. doi: 10.2217/fmb. Jungo, C., Urfer, J., Zocchi, A., Marison, I., and von Stockar, U. (2007b). 12.133 Optimisation of culture conditions with respect to biotin requirement for the Gerritzen, M. J. H., Maas, R. H. W., Van Den Ijssel, J., Van Keulen, L., Martens, production of recombinant avidin in Pichia pastoris. J. Biotechnol. 127, 703–715. D. E., Wijffels, R. H., et al. (2018). High dissolved oxygen tension triggers doi: 10.1016/j.jbiotec.2006.08.001 outer membrane vesicle formation by Neisseria meningitidis. Microb. Cell Fact. Juturu, V., and Wu, J. C. (2018). Heterologous protein expression in Pichia 17:157. doi: 10.1186/s12934-018-1007-7 pastoris: latest research progress and applications. ChemBioChem 19, 7–21. Graf, A., Dragosits, M., Gasser, B., and Mattanovich, D. (2009). Yeast systems doi: 10.1002/cbic.201700460 biotechnology for the production of heterologous proteins. FEMS Yeast Res. Kasemets, K., Drews, M., Nisamedtinov, I., Adamberg, K., and Paalme, T. (2003). 9, 335–348. doi: 10.1111/j.1567-1364.2009.00507.x Modification of A-stat for the characterization of microorganisms. J. Microbiol. Graf, A., Gasser, B., Dragosits, M., Sauer, M., Leparc, G. G., Tüchler, T., et al. (2008). Methods 55, 187–200. doi: 10.1016/S0167-7012(03)00143-X Novel insights into the unfolded protein response using Pichia pastoris specific Khasa, Y. P., Khushoo, A., Srivastava, L., and Mukherjee, K. J. (2007). Kinetic DNA microarrays. BMC Genomics 9:390. doi: 10.1186/1471-2164-9-390 studies of constitutive human granulocyte-macrophage colony stimulating Gramelsberger, G. (2018). Continuous culture techniques as simulators factor (hGM-CSF) expression in continuous culture of Pichia pastoris. for standard cells: Jacques Monod’s, Aron Novick’s and Leo Szilard’s Biotechnol. Lett. 29, 1903–1908. doi: 10.1007/s10529-007-9473-8 quantitative approach to microbiology. Hist. Philos. Life Sci. 40:23. King, Z. A., Lu, J., Dräger, A., Miller, P., Federowicz, S., Lerman, J. A., et al. (2016). doi: 10.1007/s40656-017-0182-x BiGG models: a platform for integrating, standardizing and sharing genome- Gresham, D., and Dunham, M. J. (2014). The enduring utility of scale models. Nucleic Acids Res. 44, D515–D522. doi: 10.1093/nar/gkv1049 continuous culturing in experimental evolution. Genomics 104, 399–405. Landes, N., Gasser, B., Vorauer-Uhl, K., Lhota, G., Mattanovich, D., and Maurer, doi: 10.1016/j.ygeno.2014.09.015 M. (2016). The vitamin-sensitive promoter PTHI11 enables pre-defined Herbert, D. (1961). “A theoretical analysis of continuous culture systems,” in autonomous induction of recombinant protein production in Pichia pastoris. Continuous Culture of Microorganisms, Microbiology Group Symposium Papers, Biotechnol. Bioeng. 113, 2633–2643. doi: 10.1002/bit.26041 eds D. W. Henderson, J. J. H. Hastings, B. A. Southgate, and P. W. Brian Lee, J. Y., Chen, H., Liu, A., Alba, B. M., and Lim, A. C. (2017). Auto-induction (London: Society of Chemical Industry; University College), 51–53. of Pichia pastoris AOX1 promoter for membrane protein expression. Protein Hernandez, R. (2015). Continuous manufacturing: a changing processing Expr. Purif. 137, 7–12. doi: 10.1016/j.pep.2017.06.006 paradigm. BioPharm Int. 28, 20–41. Lin, X. Q., Liang, S. L., Han, S. Y., Zheng, S. P., Ye, Y. R., and Lin, Y. (2013). Hesketh, A. R., Castrillo, J. I., Sawyer, T., Archer, D. B., and Oliver, S. G. Quantitative iTRAQ LC-MS/MS proteomics reveals the cellular response to (2013). Investigating the physiological response of Pichia (Komagataella) heterologous protein overexpression and the regulation of HAC1 in Pichia pastoris GS115 to the heterologous expression of misfolded proteins pastoris. J. Proteomics 91, 58–72. doi: 10.1016/j.jprot.2013.06.031 using chemostat cultures. Appl. Microbiol. Biotechnol. 97, 9747–9762. Liu, Z., Hou, J., Martínez, J. L., Petranovic, D., and Nielsen, J. (2013). Correlation of doi: 10.1007/s00253-013-5186-1 cell growth and heterologous protein production by Saccharomyces cerevisiae. Heyland, J., Fu, J., Blank, L. M., and Schmid, A. (2010). Quantitative physiology Appl. Microbiol. Biotechnol. 97, 8955–8962. doi: 10.1007/s00253-013-4715-2 of Pichia pastoris during glucose-limited high-cell density fed-batch cultivation Looser, V., Bruhlmann, B., Bumbak, F., Stenger, C., Costa, M., Camattari, A., for recombinant protein production. Biotechnol. Bioeng. 107, 357–368. et al. (2015). Cultivation strategies to enhance productivity of Pichia pastoris: a doi: 10.1002/bit.22836 review. Biotechnol. Adv. 33, 1177–1193. doi: 10.1016/j.biotechadv.2015.05.008 Heyland, J., Fu, J., Blank, L. M., and Schmid, A. (2011). Carbon metabolism limits Lu, P., Vogel, C., Wang, R., Yao, X., and Marcotte, E. M. (2007). Absolute protein recombinant protein production in Pichia pastoris. Biotechnol. Bioeng. 108, expression profiling estimates the relative contributions of transcriptional 1942–1953. doi: 10.1002/bit.23114 and translational regulation. Nat. Biotechnol. 25, 117–124. doi: 10.1038/ Hoekema, S., Rinzema, A., Tramper, J., Wijffels, R. H., and Janssen, M. (2014). nbt1270 Deceleration-stats save much time during phototrophic culture optimization. Maccani, A., Landes, N., Stadlmayr, G., Maresch, D., Leitner, C., Maurer, M., Biotechnol. Bioeng. 111, 792–802. doi: 10.1002/bit.25131 et al. (2014). Pichia pastoris secretes recombinant proteins less efficiently than Hoskisson, P. A., and Hobbs, G. (2005). Continuous culture - making a comeback? Chinese hamster ovary cells but allows higher space-time yields for less complex Microbiology 151, 3153–3159. doi: 10.1099/mic.0.27924-0 proteins. Biotechnol. J. 9, 526–537. doi: 10.1002/biot.201300305 Jordà, J., De Jesus, S. S., Peltier, S., Ferrer, P., and Albiol, J. (2014a). Massahi, A., and Çalik, P. (2018). Naturally occurring novel promoters Metabolic flux analysis of recombinant Pichia pastoris growing on different around pyruvate branch-point for recombinant protein production in Pichia glycerol/methanol mixtures by iterative fitting of NMR-derived 13C- pastoris (Komagataella phaffii): pyruvate decarboxylase- and pyruvate kinase- labelling data from proteinogenic amino acids. N. Biotechnol. 31, 120–132. promoters. Biochem. Eng. J. 138, 111–120. doi: 10.1016/j.bej.2018.07.012 doi: 10.1016/j.nbt.2013.06.007 Mattanovich, D., Gasser, B., Hohenblum, H., and Sauer, M. (2004). Stress Jordà, J., Jouhten, P., Cámara, E., Maaheimo, H., Albiol, J., and Ferrer, P. in recombinant protein producing yeasts. J. Biotechnol. 113, 121–135. (2012). Metabolic flux profiling of recombinant protein secreting Pichia doi: 10.1016/j.jbiotec.2004.04.035 pastoris growing on glucose: methanol mixtures. Microb. Cell Fact. 11:57. Maurer, M., Kühleitner, M., Gasser, B., and Mattanovich, D. (2006). Versatile doi: 10.1186/1475-2859-11-57 modeling and optimization of fed batch processes for the production of Jordà, J., Rojas, H. C., Carnicer, M., Wahl, A., Ferrer, P., and Albiol, J. secreted heterologous proteins with Pichia pastoris. Microb. Cell Fact. 5:37. (2014b). Quantitative metabolomics and instationary 13C-metabolic doi: 10.1186/1475-2859-5-37 Frontiers in Bioengineering and Biotechnology | www.frontiersin.org 19 June 2020 | Volume 8 | Article 632 Nieto-Taype et al. Continuous Cultivation Toward Rational Bioprocesses McGeachy, A. M., Meacham, Z. A., and Ingolia, N. T. (2019). An accessible systems better than bacterial production systems? Appl. Microbiol. Biotechnol. continuous-culture turbidostat for pooled analysis of complex libraries. ACS 89, 939–948. doi: 10.1007/s00253-010-3019-z Synth. Biol. 8, 844–856. doi: 10.1021/acssynbio.8b00529 Potvin, G., Ahmad, A., and Zhang, Z. (2012). Bioprocess engineering aspects of Mears, L., Feldman, H., Falco, F. C., Bach, C., Wu, M., Nørregaard, A., et al. heterologous protein production in Pichia pastoris: a review. Biochem. Eng. J. (2017). “Continuous fermentation for biopharmaceuticals?,” in Continuous 64, 91–105. doi: 10.1016/j.bej.2010.07.017 Manufacturing of Pharmaceuticals, eds P. Kleinebudde J. Khinast, and J. Prielhofer, R., Maurer, M., Klein, J., Wenger, J., Kiziak, C., Gasser, B., Rantanen (Chichester: John Wiley & Sons, Ltd), 227–246. et al. (2013). Induction without methanol: novel regulated promoters Monod, J. (1950). La technique de culture continue, theorie et applications. Ann. enable high-level expression in Pichia pastoris. Microb. Cell Fact. 12:5. l’Institut Pasteur 79, 390–410. doi: 10.1186/1475-2859-12-5 Moser, J. W., Prielhofer, R., Gerner, S. M., Graf, A. B., Wilson, I. B. H., Mattanovich, Prielhofer, R., Reichinger, M., Wagner, N., Claes, K., Kiziak, C., Gasser, B., D., et al. (2017). Implications of evolutionary engineering for growth and et al. (2018). Superior protein titers in half the fermentation time: promoter recombinant protein production in methanol-based growth media in the yeast and process engineering for the glucose-regulated GTH1 promoter of Pichia Pichia pastoris. Microb. Cell Fact. 16:49. doi: 10.1186/s12934-017-0661-5 pastoris. Biotechnol. Bioeng. 115, 2479–2488. doi: 10.1002/bit.26800 Nahku, R., Valgepea, K., Lahtvee, P.-J., Erm, S., Abner, K., Adamberg, K., Puxbaum, V., Mattanovich, D., and Gasser, B. (2015). Quo vadis? The challenges et al. (2010). Specific growth rate dependent transcriptome profiling of of recombinant protein folding and secretion in Pichia pastoris. Appl. Microbiol. Escherichia coli K12 MG1655 in accelerostat cultures. J. Biotechnol. 145, 60–65. Biotechnol. 99, 2925–2938. doi: 10.1007/s00253-015-6470-z doi: 10.1016/j.jbiotec.2009.10.007 Rahimi, A., Hosseini, S. N., Karimi, A., Aghdasinia, H., and Arabi Mianroodi, R. Nieto-Taype, M. A., Garrigós-Martínez, J., Sánchez-Farrando, M., Valero, F., (2019). Enhancing the efficiency of recombinant hepatitis B surface antigen Garcia-Ortega, X., and Montesinos-Seguí, J. L. (2020). Rationale-based production in Pichia pastoris by employing continuous fermentation. Biochem. selection of optimal operating strategies and gene dosage impact on Eng. J. 141, 112–119. doi: 10.1016/j.bej.2018.10.019 recombinant protein production in Komogataella phaffii (Pichia pastoris). Rajamanickam, V., Herwig, C., and Spadiut, O. (2018). Data science, modeling, Microb. Biotechnol. 13, 315–327. doi: 10.1111/1751-7915.13498 and advanced PAT tools enable continuous culture. Bioprocess Int. 16, 20–25. Nisamedtinov, I., Lindsey, G. G., Karreman, R., Orumets, K., Koplimaa, M., Randek, J., and Mandenius, C.-F. (2018). On-line soft sensing Kevvai, K., et al. (2008). The response of the yeast Saccharomyces cerevisiae to in upstream bioprocessing. Crit. Rev. Biotechnol. 38, 106–121. sudden vs. gradual changes in environmental stress monitored by expression doi: 10.1080/07388551.2017.1312271 of the stress response protein Hsp12p. FEMS Yeast Res. 8, 829–838. Rathore, A. S., Agarwal, H., Sharma, A. K., Pathak, M., and Muthukumar, S. (2015). doi: 10.1111/j.1567-1364.2008.00391.x Continuous processing for production of biopharmaceuticals. Prep. Biochem. Niu, H., Jost, L., Pirlot, N., Sassi, H., Daukandt, M., Rodriguez, C., et al. Biotechnol. 45, 836–849. doi: 10.1080/10826068.2014.985834 (2013). A quantitative study of methanol/sorbitol co-feeding process of Rebnegger, C., Graf, A. B., Valli, M., Steiger, M. G., Gasser, B., Maurer, M., a Pichia pastoris Mut+/pAOX1-lacZ strain. Microb. Cell Fact. 12:33. et al. (2014). In Pichia pastoris, growth rate regulates protein synthesis doi: 10.1186/1475-2859-12-33 and secretion, mating and stress response. Biotechnol. J. 9, 511–525. Nocon, J., Steiger, M., Mairinger, T., Hohlweg, J., Rußmayer, H., Hann, S., et al. doi: 10.1002/biot.201300334 (2016). Increasing pentose phosphate pathway flux enhances recombinant Rebnegger, C., Vos, T., Graf, A. B., Valli, M., Pronk, J. T., Daran-Lapujade, P., et al. protein production in Pichia pastoris. Appl. Microbiol. Biotechnol. 100, (2016). Pichia pastoris exhibits high viability and a low maintenance energy 5955–5963. doi: 10.1007/s00253-016-7363-5 requirement at near-zero specific growth rates. Appl. Environ. Microbiol. 82, Nocon, J., Steiger, M. G., Pfeffer, M., Sohn, S. B., Kim, T. Y., Maurer, 4570–4583. doi: 10.1128/AEM.00638-16 M., et al. (2014). Model based engineering of Pichia pastoris central Rossignol, T., Kobi, D., Jacquet-Gutfreund, L., and Blondin, B. metabolism enhances recombinant protein production. Metab. Eng. 24, (2009). The proteome of a wine yeast strain during fermentation, 129–138. doi: 10.1016/j.ymben.2014.05.011 correlation with the transcriptome. J. Appl. Microbiol. 107, 47–55. Paulová, L., Hyka, P., Branská, B., Melzoch, K., and Kovar, K. (2012). Use doi: 10.1111/j.1365-2672.2009.04156.x of a mixture of glucose and methanol as substrates for the production of Rußmayer, H., Buchetics, M., Gruber, C., Valli, M., Grillitsch, K., Modarres, G., recombinant trypsinogen in continuous cultures with Pichia pastoris Mut+. J. et al. (2015). Systems-level organization of yeast methylotrophic lifestyle. BMC Biotechnol. 157, 180–188. doi: 10.1016/j.jbiotec.2011.10.010 Biol. 13:80. doi: 10.1186/s12915-015-0186-5 Peebo, K., and Neubauer, P. (2018). Application of continuous culture methods Sanchez-Garcia, L., Martín, L., Mangues, R., Ferrer-Miralles, N., Vázquez, E., and to recombinant protein production in microorganisms. Microorganisms 6:56. Villaverde, A. (2016). Recombinant pharmaceuticals from microbial cells: a doi: 10.3390/microorganisms6030056 2015 update. Microb. Cell Fact. 15:33. doi: 10.1186/s12934-016-0437-3 Peña, D. A., Gasser, B., Zanghellini, J., Steiger, M. G., and Mattanovich, D. Schwarzhans, J.-P., Wibberg, D., Winkler, A., Luttermann, T., Kalinowski, J., (2018). Metabolic engineering of Pichia pastoris. Metab. Eng. 50, 2–15. and Friehs, K. (2016). Integration event induced changes in recombinant doi: 10.1016/j.ymben.2018.04.017 protein productivity in Pichia pastoris discovered by whole genome Periyasamy, S., Govindappa, N., Sreenivas, S., and Sastry, K. (2013). Isolation, sequencing and derived vector optimization. Microb. Cell Fact. 15:84. characterization and evaluation of the Pichia pastoris sorbitol dehydrogenase doi: 10.1186/s12934-016-0486-7 promoter for expression of heterologous proteins. Protein Expr. Purif. 92, Shi, X., Karkut, T., Chamankhah, M., Alting-Mees, M., Hemmingsen, S. M., 128–133. doi: 10.1016/j.pep.2013.09.008 and Hegedus, D. (2003). Optimal conditions for the expression of a single- Pfeffer, M., Maurer, M., Stadlmann, J., Grass, J., Delic, M., Altmann, F., et al. (2012). chain antibody (scFv) gene in Pichia pastoris. Protein Expr. Purif. 28, 321–330. Intracellular interactome of secreted antibody Fab fragment in Pichia pastoris doi: 10.1016/S1046-5928(02)00706-4 reveals its routes of secretion and degradation. Appl. Microbiol. Biotechnol. 93, Singh, R., Kumar, M., Mittal, A., and Mehta, P. K. (2017). Microbial 2503–2512. doi: 10.1007/s00253-012-3933-3 metabolites in nutrition, healthcare and agriculture. 3 Biotech 7:15. Ponte, X., Barrigón, J. M., Maurer, M., Mattanovich, D., Valero, F., and doi: 10.1007/s13205-016-0586-4 Montesinos-Seguí, J. L. (2018). Towards optimal substrate feeding for Sinha, J., Plantz, B. A., Inan, M., and Meagher, M. M. (2005). Causes heterologous protein production in Pichia pastoris (Komagataella spp) fed- of proteolytic degradation of secreted recombinant proteins produced in batch processes under PAOX1 control: a modelling aided approach. J. Chem. methylotrophic yeast Pichia pastoris: case study with recombinant ovine Technol. Biotechnol. 93, 3208–3218. doi: 10.1002/jctb.5677 interferon-? Biotechnol. Bioeng. 89, 102–112. doi: 10.1002/bit.20318 Ponte, X., Montesinos-Seguí, J. L., and Valero, F. (2016). Bioprocess efficiency Sirén, N., Weegar, J., Dahlbacka, J., Kalkkinen, N., Fagervik, K., Leisola, M., et al. in Rhizopus oryzae lipase production by Pichia pastoris under the control (2006). Production of recombinant HIV-1 Nef (negative factor) protein using of PAOX1 is oxygen tension dependent. Process Biochem. 51, 1954–1963. Pichia pastoris and a low-temperature fed-batch strategy. Biotechnol. Appl. doi: 10.1016/j.procbio.2016.08.030 Biochem. 44:151. doi: 10.1042/BA20060001 Porro, D., Gasser, B., Fossati, T., Maurer, M., Branduardi, P., Sauer, M., et al. (2011). Sohn, S. B., Graf, A. B., Kim, T. Y., Gasser, B., Maurer, M., Ferrer, P., et al. (2010). Production of recombinant proteins and metabolites in yeasts: when are these Genome-scale metabolic model of methylotrophic yeast Pichia pastoris and its Frontiers in Bioengineering and Biotechnology | www.frontiersin.org 20 June 2020 | Volume 8 | Article 632 Nieto-Taype et al. Continuous Cultivation Toward Rational Bioprocesses use for in silico analysis of heterologous protein production. Biotechnol. J. 5, Vogl, T., and Glieder, A. (2013). Regulation of Pichia pastoris promoters 705–715. doi: 10.1002/biot.201000078 and its consequences for protein production. N. Biotechnol. 30, 385–404. Sørensen, H. P. (2010). Towards universal systems for recombinant gene doi: 10.1016/j.nbt.2012.11.010 expression. Microb. Cell Fact. 9:27. doi: 10.1186/1475-2859-9-27 Vogl, T., Sturmberger, L., Fauland, P. C., Hyden, P., Fischer, J. E., Schmid, C., Spadiut, O., Rittmann, S., Dietzsch, C., and Herwig, C. (2013). Dynamic et al. (2018b). Methanol independent induction in Pichia pastoris by simple process conditions in bioprocess development. Eng. Life Sci. 13, 88–101. derepressed overexpression of single transcription factors. Biotechnol. Bioeng. doi: 10.1002/elsc.201200026 115, 1037–1050. doi: 10.1002/bit.26529 Sreekrishna, K. (2010). “Pichia, optimization of protein expression,” in Vos, T., de la Torre Cortés, P., van Gulik, W. M., Pronk, J. T., and Daran-Lapujade, Encyclopedia of Industrial Biotechnology, ed M. C. Flickinger (Hoboken, NJ: P. (2015). Growth-rate dependency of de novo resveratrol production in John Wiley & Sons, Inc.), 16. chemostat cultures of an engineered Saccharomyces cerevisiae strain. Microb. Subramanian, G., Erm, S., Abner, K., Seiman, A., Adamberg, K., and Vilu, Cell Fact. 14:133. doi: 10.1186/s12934-015-0321-6 R. (2017). “Study of cells in the steady-state growth space,” in Continuous Wagenen, J., Van Pape, M. L., and Angelidaki, I. (2015). Characterization of Biomanufacturing - Innovative Technologies and Methods, ed G. Subramanian nutrient removal and microalgal biomass production on an industrial waste- (Weinheim: Wiley-VCH Verlag GmbH & Co. KGaA), 233–258. stream by application of the deceleration-stat technique. Water Res. 75, Surribas, A., Stahn, R., Montesinos, J. L., Enfors, S.-O., Valero, F., and 301–311. doi: 10.1016/j.watres.2015.02.022 Jahic, M. (2007). Production of a Rhizopus oryzae lipase from Pichia Wagner, J. M., and Alper, H. S. (2016). Synthetic biology and molecular genetics in pastoris using alternative operational strategies. J. Biotechnol. 130, 291–299. non-conventional yeasts: current tools and future advances. Fungal Genet. Biol. doi: 10.1016/j.jbiotec.2007.04.009 89, 126–136. doi: 10.1016/j.fgb.2015.12.001 Szopinska, A., and Morsomme, P. (2010). Quantitative proteomic approaches and Wang, H., Li, J., Liu, L., Li, X., Jia, D., Du, G., et al. (2012). Increased production their application in the study of yeast stress responses. Omi. A J. Integr. Biol. 14, of alkaline polygalacturonate lyase in the recombinant Pichia pastoris by 639–649. doi: 10.1089/omi.2010.0045 controlling cell concentration during continuous culture. Bioresour. Technol. Tang, S., Potvin, G., Reiche, A., and Zhang, Z. (2010). Modeling of phytase 124, 338–346. doi: 10.1016/j.biortech.2012.08.027 production by cultivation of Pichia pastoris under the control of the GAP Wang, J., Wang, X., Shi, L., Qi, F., Zhang, P., Zhang, Y., et al. (2017). Methanol- promoter. Int. J. Chem. React. Eng. 8:11. doi: 10.2202/1542-6580.2144 independent protein expression by AOX1 promoter with trans-acting elements Tomàs-Gamisans, M., Ferrer, P., and Albiol, J. (2018). Fine-tuning the P. pastoris engineering and glucose-glycerol-shift induction in Pichia pastoris. Sci. Rep. iMT1026 genome-scale metabolic model for improved prediction of growth on 7:41850. doi: 10.1038/srep41850 methanol or glycerol as sole carbon sources. Microb. Biotechnol. 11, 224–237. Weninger, A., Fischer, J. E., Raschmanová, H., Kniely, C., Vogl, T., and Glieder, A. doi: 10.1111/1751-7915.12871 (2018). Expanding the CRISPR/Cas9 toolkit for Pichia pastoris with efficient Tomàs-Gamisans, M., Ødum, A. S. R., Workman, M., Ferrer, P., and Albiol, donor integration and alternative resistance markers. J. Cell. Biochem. 119, J. (2019). Glycerol metabolism of Pichia pastoris (Komagataella spp.) 3183–3198. doi: 10.1002/jcb.26474 characterised by 13C-based metabolic flux analysis. N. Biotechnol. 50, 52–59. Yamawaki, S., Matsumoto, T., Ohnishi, Y., Kumada, Y., Shiomi, N., Katsuda, T., doi: 10.1016/j.nbt.2019.01.005 et al. (2007). Production of single-chain variable fragment antibody (scFv) in Tomson, K., Barber, J., and Vanatalu, K. (2006). Adaptastat—a new method for fed-batch and continuous culture of Pichia pastoris by two different methanol optimising of bacterial growth conditions in continuous culture: interactive feeding methods. J. Biosci. Bioeng. 104, 403–407. doi: 10.1263/jbb.104.403 substrate limitation based on dissolved oxygen measurement. J. Microbiol. Yang, Z., and Zhang, Z. (2018). Engineering strategies for enhanced production Methods 64, 380–390. doi: 10.1016/j.mimet.2005.05.014 of protein and bio-products in Pichia pastoris: a review. Biotechnol. Adv. 36, Valero, F., and López-Santín, J. (2017). “Online analysis for industrial bioprocesses: 182–195. doi: 10.1016/j.biotechadv.2017.11.002 gas analysis” in Current Developments in Biotechnology and Bioengineering, Ye, R., Huang, M., Lu, H., Qian, J., Lin, W., Chu, J., et al. (2017). eds Ch. Larroche, M. A. Sanromán, G. Du, and A. Pandey (Amsterdam: Comprehensive reconstruction and evaluation of Pichia pastoris genome-scale Elsevier), 649–678. doi: 10.1016/B978-0-444-63663-8.00022-7 metabolic model that accounts for 1243 ORFs. Bioresour. Bioprocess. 4:22. van Hove, B., Love, A. M., Ajikumar, P. K., and De Mey, M. (2016). “Programming doi: 10.1186/s40643-017-0152-x biology: expanding the toolset for the engineering of transcription,” in Synthetic Zahrl, R. J., Mattanovich, D., and Gasser, B. (2018). The impact of ERAD on Biology, eds A. Glieder, Ch. P. Kubicek, D. Mattanovich, B. Wiltschi, and M. recombinant protein secretion in Pichia pastoris (Syn Komagataella spp.). Sauer (Cham: Springer International Publishing), 1–64. Microbiology 164, 453–463. doi: 10.1099/mic.0.000630 Vanz, A. L., Lünsdorf, H., Adnan, A., Nimtz, M., Gurramkonda, C., Khanna, N., Zahrl, R. J., Peña, D. A., Mattanovich, D., and Gasser, B. (2017). Systems et al. (2012). Physiological response of Pichia pastoris GS115 to methanol- biotechnology for protein production in Pichia pastoris. FEMS Yeast Res. induced high level production of the hepatitis B surface antigen: catabolic 17:fox068. doi: 10.1093/femsyr/fox068 adaptation, stress responses, and autophagic processes. Microb. Cell Fact. Zhang, W., Sinha, J., Smith, L. A., Inan, M., and Meagher, M. M. (2005). 11:103. doi: 10.1186/1475-2859-11-103 Maximization of production of secreted recombinant proteins in Pichia Vanz, A. L., Nimtz, M., and Rinas, U. (2014). Decrease of UPR- and ERAD- pastoris fed-batch fermentation. Biotechnol. Prog. 21, 386–393. doi: 10.1021/bp related proteins in Pichia pastoris during methanol-induced secretory insulin 049811n precursor production in controlled fed-batch cultures. Microb. Cell Fact. 13, Zhou, X.-S., and Zhang, Y.-X. (2002). Decrease of proteolytic degradation of 1–10. doi: 10.1186/1475-2859-13-23 recombinant hirudin produced by Pichia pastoris by controlling the specific Veloso, A. C., and Ferreira, E. C. (2017). “Online analysis for industrial growth rate. Biotechnol. Lett. 24, 1449–1453. doi: 10.1023/A:1019831406141 bioprocesses: broth analysis” in Current Developments in Biotechnology and Ziv, N., Brandt, N. J., and Gresham, D. (2013). The use of chemostats in microbial Bioengineering, eds Ch. Larroche, M. A. Sanromán, G. Du, and A. Pandey systems biology. J. Vis. Exp. 80:50168. doi: 10.3791/50168 (Amsterdam: Elsevier), 679–704. doi: 10.1016/B978-0-444-63663-8.00023-9 Vieira Gomes, A., Souza Carmo, T., Silva Carvalho, L., Mendonça Bahia, F., and Conflict of Interest: The authors declare that the research was conducted in the Parachin, N. (2018). Comparison of yeasts as hosts for recombinant protein absence of any commercial or financial relationships that could be construed as a production. Microorganisms 6:38. doi: 10.3390/microorganisms6020038 potential conflict of interest. Vogel, C., De Sousa Abreu, R., Ko, D., Le, S.-Y., Shapiro, B. A., Burns, S. C., et al. (2010). Sequence signatures and mRNA concentration can explain two-thirds Copyright © 2020 Nieto-Taype, Garcia-Ortega, Albiol, Montesinos-Seguí and Valero. of protein abundance variation in a human cell line. Mol. Syst. Biol. 6:400. This is an open-access article distributed under the terms of the Creative Commons doi: 10.1038/msb.2010.59 Attribution License (CC BY). The use, distribution or reproduction in other forums Vogl, T., Gebbie, L., Palfreyman, R. W., and Speight, R. (2018a). Effect of is permitted, provided the original author(s) and the copyright owner(s) are credited plasmid design and type of integration event on recombinant protein and that the original publication in this journal is cited, in accordance with accepted expression in Pichia pastoris. Appl. Environ. Microbiol. 84, e02712–e02717. academic practice. No use, distribution or reproduction is permitted which does not doi: 10.1128/AEM.02712-17 comply with these terms. Frontiers in Bioengineering and Biotechnology | www.frontiersin.org 21 June 2020 | Volume 8 | Article 632