Garrigós‑Martínez et al. Microb Cell Fact https://doi.org/10.1186/s12934-021-01577-4 (2021) 20:90 Microbial Cell Factories RESEARCH Open Access Scalable production and application of Pichia pastoris whole cell catalysts expressing human cytochrome P450 2C9 Javier Garrigós‑Martínez1, Astrid Weninger2, José Luis Montesinos‑Seguí1, Christian Schmid2, Francisco Valero1, Claudia Rinnofner3,4, Anton Glieder2,3* and Xavier Garcia‑Ortega1 Abstract Background: Currently, the numerous and versatile applications in pharmaceutical and chemical industry make the recombinant production of cytochrome P450 enzymes (CYPs) of great biotechnological interest. Accelerating the drug development process by simple, quick and scalable access of human drug metabolites is key for efficient and targeted drug development in response to new and sometimes unexpected medical challenges and needs. However, due its biochemical complexity, scalable human CYP (hCYP) production and their application in preparative biotrans‑ formations was still in its infancy. Results: A scalable bioprocess for fine-tuned co-expression of hCYP2C9 and its essential complementary human cytochrome P450 reductase (hCPR) in the yeast Pichia pastoris (Komagataella phaffii) is presented. High-throughput screening (HTS) of a transformant library employing a set of diverse bidirectional expression systems with different regulation patterns and a fluorimetric assay was used in order to fine-tune hCYP2C9 and hCPR co-expression, and to identify best expressing clonal variants. The bioprocess development for scalable and reliable whole cell biocatalyst production in bioreactors was carried out based on rational optimization criteria. Among the different alternatives studied, a glycerol carbon-limiting strategy at high µ showed highest production rates, while methanol co-addition together with a decrease of µ provided the best results in terms of product to biomass yield and whole cell activity. By implementing the mentioned strategies, up to threefold increases in terms of production rates and/or yield could be achieved in comparison with initial tests. Finally, the performance of the whole cell catalysts was demonstrated successfully in biotransformation using ibuprofen as substrate, demonstrating the expected high selectivity of the human enzyme catalyst for 3′hydroxyibuprofen. Conclusions: For the first time a scalable bioprocess for the production of hCYP2C9 whole cell catalysts was suc‑ cessfully designed and implemented in bioreactor cultures, and as well, further tested in a preparative-scale biotrans‑ formation of interest. The catalyst engineering procedure demonstrated the efficiency of the employment of a set of differently regulated bidirectional promoters to identify transformants with most effective membrane-bound hCYP/ hCPR co-expression ratios and implies to become a model case for the generation of other P. pastoris based catalysts relying on co-expressed enzymes such as other P450 catalysts or enzymes relying on co-expressed enzymes for co- factor regeneration. *Correspondence: a.glieder@tugraz.at 2 Institute of Molecular Biotechnology, Graz University of Technology, Petersgasse 14, 8010 Graz, Austria Full list of author information is available at the end of the article © The Author(s) 2021. This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://​creat​iveco​ mmons.​org/​licen​ses/​by/4.​0/. The Creative Commons Public Domain Dedication waiver (http://​creat​iveco​mmons.​org/​publi​cdoma​in/​ zero/1.​0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Garrigós‑Martínez et al. Microb Cell Fact (2021) 20:90 Page 2 of 13 Keywords: Bidirectional promoters, Human CYP2C9, Ibuprofen, Pichia pastoris, Bioprocess optimisation, Whole cell biocatalyst Background associated hP450s seems preferable compared to solu- In order to avoid adverse drug reactions (ADRs) not only ble expression of truncated P450s. Furthermore, to be the drug as such, but also the impact of its metabolites biologically active, CYPs require an electron transport on the organism are considered of great importance [1]. system which provides the electrons to the CYPs for oxy- From the super-family of P450 enzymes from different gen activation and substrate oxidation [15]. hCYPs rely origins, Human cytochrome P450s (hCYPs) are a group on the presence of a cytochrome P450 reductase (CPR), of heme-containing membrane-associated monooxy- which is needed for electron transfer from the co-factor genases involved in the oxidation of monoterpenes, NAD(P)H. The co-expression of hCPR often showed to saturated fatty acids and alkanes, vitamins, steroids and be more effective than taking advantage of the host’s own eicosanoids as well as in the clearance of drugs and xeno- CPR(s) [7]. biotics in humans [2, 3]. Among the human variants, During the last years, to enable industrial biotransfor- next to CYP3A4 and CYP2D6, CYP2C9 is one of the mations that require high conversion rates, easy han- most important drug oxidizing enzyme [4]. In particu- dling and low costs, the recombinant production of CYP lar, CYP2C9 oxidizes numerous drugs such as celecoxib, enzymes has been intensively investigated [6, 10, 12, diclofenac, etodolac, ibuprofen, indomethacin, lornoxi- 16–19]. Some relevant publications have reported several cam, mefenamic acid, suprofen, tenoxicam and, addition- cases for the recombinant expression of human P450s ally, also metabolizes endogenous compounds such as in different cell factories [3, 20]. Specifically, success- arachidonic acid, linoleic acid, and non-drug xenobiotics ful expression of hCYP2C9 has been reported for insect [4]. cells [21], Escherichia coli [22] and Schizosaccharomyces Drugs and drug-metabolite standards are required pombe [23]. Thus, this enzyme was used for the prepara- for toxicological, biological and drug-to-drug interac- tion of milligram amounts of the 4′-hydroxy metabolite tion studies. Their synthesis is often challenging, since of diclofenac using whole cells or isolated membranes selective hydroxylation of non-activated carbon atoms [22]. The use of yeast expression systems as whole cell is difficult to achieve by chemical synthesis. These stud- biocatalysts was described by Neunzig et al., who demon- ies are elucidated by the employment of tissue samples strated the applicability of an ibuprofen conversion pro- (e.g. human liver homogenate) or microsomal prepara- cess using fission yeast for the co-expression of hCYP2C9 tions thereof [5–7]. Both are not scalable for preparative and CPR [24, 25], and efficient and selective 4′-hydroxy synthesis on multi mg scale. Microbial P450 enzymes diclofenac was synthesized on g-scale by the Bureik lab provide an interesting alternative but frequently do not [23]. In a comparative study Geier et al. highlighted the produce the desired specific products, or undesired mix- methylotrophic yeast Pichia pastoris, recently re-classi- tures which are different from the product spectrum fied as Komagataella phaffii, as a preferable cell factory made by human P450s are obtained [6, 8–10]. Adaptation for whole cell bioconversions using human CYP2D6 [7]. by protein engineering is possible but in cases of unex- Due to those experiences and the broad availability and pected arising needs for new drugs such as arising epi- establishment of the P. pastoris expression system in demic outbreaks, but such approach is usually too slow. thousands of labs, we were interested to study the relia- Therefore, based on their versatile and changing applica- bility, scalability and performance of P. pastoris whole cell tions in drug metabolite synthesis, there is an increasing catalysts expressing hCYP2C9. interest and high demand for hCYPs in both pharmaceu- Pichia pastoris is currently considered as an efficient tical and chemical industries, especially when applicable alternative for recombinant protein production com- on a preparative scale level [3, 11]. In this sense, biotrans- bining the simplicity of bacterial expression systems formations performed with purified enzymes are hardly with some essential features of higher eukaryotic hosts feasible on a large scale, since hCYPs usually present low such as mammalian cells. Most of membrane pro- efficiency and stability [11, 12]. To some extend this is teins (MPs) of interest such as CYPs often come from also related to the fact the hCYP’s are membrane asso- eukaryotic organisms, thus bacterial expression sys- ciated, so they can be considered as membrane proteins. tems often fail to express such genes to get biologically However, selective access of some substrates to the active active and corrected folded enzymes. This fact has site is known to rely on the interface between enzyme been described to be due to the lack capacity to per- and the ER membrane [13, 14], expression of membrane form post-translational modifications (e.g. disulphide Garrigós‑Martínez et al. Microb Cell Fact (2021) 20:90 Page 3 of 13 bridge formation or glycosylation), the inefficient Results and discussion secretory pathways needed to direct the protein to the Strain generation membrane, as well as to the differences in lipid bilayer Expression systems based on engineered bidirectional composition [26, 27] and availability of different intra- promoters offer a simple and quick solution to enable cellular compartments. On the other hand, mamma- multi-gene co-expression in which the expression of lian expression systems are able to produce complex each gene can be optimally tuned towards to achieve the proteins, and especially MPs [27]. However, their cell desired objective [37]. To fully exploit the potential of growth is significantly slower and more expensive this bidirectional expression system, a test set of 7 alter- compared to microbial cells. Thus, on the basis of its native promoters with diverse regulation patterns was high efficiency, wide availability and cost-effectiveness, used in different co-orientations to drive the expression P. pastoris has become one of the most frequently used of the full-length genes: CYP2C9 (AL359672) including host systems for the production recombinant proteins, its hydrophobic N-terminal sequence and redox partner including membrane-anchored and soluble proteins— hCPR (accession AAH34277.1). In contrast to self-suffi- either intra or extracellular [27]. Usually, most pub- cient soluble bacterial P450s, here the CPR is covalently lished studies focused on the recombinant expression linked to the heme domain of the P450 catalyst, thus dif- of MPs address issues related to protein degradation, ferent co-expression levels and regulatory profiles were ER folding problems and post-translational modifica- evaluated in order to achieve the best performance of the tions in order to increase final product titers, since it whole cell biocatalyst requiring two separate functionally may have a relevant impact on its production [28, 29]. expressed peptide chains, anchored to the membrane in However, to the date, there has not been reported any optimal orientation and proximity to its redox partner. study that assess the impact of operational strategies This study included the strong methanol inducible pro- in the recombinant production of MPs targeted to be moters ­PDAS1 and ­PDAS2 (and their shortened variants anchored in the cell membrane. So, as a highly relevant ­PDAS1-552 and P ­ DAS2-699), as well as the promoters ­PPDC and factor to design efficient scalable processes, this work ­PPDF, two derepressed promoters in which the recombi- aimed to determine the interrelation between the spe- nant expression is activated upon glucose/glycerol deple- cific production rate (qp) and the specific growth rate tion, and additionally it is further inducible in presence (µ) of the culture for active hP450 catalyst production of methanol. In earlier studies using GFP as a model pro- and biotransformations in bioreactors. This relation- tein, ­PDAS1 presented a strong methanol-inducible per- ship is called production kinetics, and can be consid- formance, reaching at least half of the expression level ered the key to design optimal bioprocess strategies for of the commonly used AOX1 promoter, while ­PDAS2 even recombinant protein production (RPP) [30–36]. outperformed ­PAOX1 expression [38]. On the other hand, The aim of this work was to demonstrate a model the expression of the P ­ PDC promoter (native catalase gene case for the production of highly active and selective promoter) is repressed in the presence of glucose or glyc- CYP2C9/CPR whole cell biocatalysts using the sim- erol and derepressed upon the depletion of these carbon ple, widely available and efficient expression system sources. Its expression can further be induced with meth- P. pastoris. It takes into consideration the essential anol and remarkably also with oleic acid [39]. Besides the scalability required for catalyst manufacturing by the ­PPDC, an orthologous promoter, i.e. P ­ PDF, which exhibits design and implementation of a reliable and optimized a similar regulation profile, was employed to regulate cultivation process for whole cell catalyst production. gene transcription. These promoters, which allow differ- Therefore, in a complementary approach, the cata- ent levels of induction, enable to some extend to uncou- lyst expression bioprocess was optimized by evaluat- ple recombinant protein expression and cell growth for P. ing different co-expression ratios of hCYP and hCPR pastoris and therefore become interesting tools for bio- using a bidirectional and functionally balanced expres- process optimization. sion system and also further combined with a ration- Earlier studies reported the efficiency of CYP2D6 in P. ally designed bioprocess at bench-top bioreactor scale pastoris whole cells, which was detectable by an spectros- taking into consideration the relevant impact of pH, copy method based on carbon monoxide (CO) difference specific growth rate and the use of methanol as co- [7]. However, highest CYP levels determined by CO- substrate on whole cell biocatalyst production. The spectroscopy could not be correlated to highest conver- efficiency of such produced whole cell catalysts was sions levels when using whole cells. Working with whole demonstrated by preparative scale biotransformation cells, their conversion efficiencies are influenced by elec- (0.5 L) of ibuprofen to produce the main hCYP2C9 tron and substrate supply [7], and it also affected by the ibuprofen metabolite 3-hydroxyibuprofen. complex interactions of multiple parameters (e.g. CYP/ CPR ratio, membrane permeability, NADH production/ Garrigós‑Martínez et al. Microb Cell Fact (2021) 20:90 Page 4 of 13 regeneration or expression level of ROS degrading enzymes between among others) [40]. Thus, this work is focused on the achievement of active whole cells with optimized CYP/CPR ratios rather than a maximization of P450 as determined by carbon monoxide spectroscopy following a previously described bidirectional expression strategy [37]. For the identification of the catalyst with the highest activity, two alternative whole cell bioconversions assays suitable for High Throughput Screening (HTS) were used [41]. One based on diclofenac conversion, and the other one based on the fluorescence of the product 7-hydroxy- 4-(trifluoromethyl)-2H-chromen-2-one (HFC). Both analysis methods were used for comparing the best clone’s performance of each construct, as shown in Fig. 1. The qualitative behaviour of both screening methods was rather similar, construct 2C9-PDC/PDF-CPR showed the best conversion (53%) in the diclofenac-based assay (2 mM). The same construct with inverted promoter ori- entation 2C9-PDF/PDC-CPR, also presented good sub- strate conversion reaching 39%, while all other constructs showed only 20% conversion of diclofenac, or even lower. Similar results were obtained using the alternative fluo- rometric MCF-based assay, demonstrating the applicabil- ity of this simple assay for the identification of cells with highest P450 activity in simple 96-deep well plate assays with low cell densities cultures. As for the diclofenac screening, the expression system built on P ­ PDF/PPDC com- bination provided the best results, reaching a maximum value of 28 RFU per ­OD600. However, due to the low sample numbers the activity of samples obtained from Fig. 1 a Diclofenac and 7-methoxy-4-(trifluoromethyl)-2H-chro men-2-one (MCF) oxidation reactions catalysed by P. pastoris CYP2C9/ bench-top fermenters, was directly determined by HPLC CPR whole cell biocatalyst. b Screening of producer clones based for the drug target substrate diclofenac. on bidirectional expression system using different promoter variants The applied screenings identified transformants with a towards to identify the most active CYP2C9/CPR whole cells ­PPDF promoter for CYP expression and P ­ PDC for the CPR (clones 2C9-PDC/PDF-CPR) as most promising for the further bioprocess development in bench-top fermenters. obtained from bench-top fermenters displayed similar diclofenac oxidation activities than cells from shake flask, Production of P. pastoris CYP2C9/CPR whole cell biocatalyst a thorough study of hCYP2C9/hCPR whole cell produc- in bioreactor cultures tion was carried out including chemostat cultivation and For the very first time, CYP2C9/CPR expressing P. pas- bioprocess optimization in fed-batch cultures. toris cells were cultivated successfully in bioreactor. Among the physical–chemical parameters that could Previous unpublished studies in our lab prior to this cul- affect the bioprocess performance, pH was identified to tivation optimization demonstrated unexpected difficul- have significant impact. The control and adjustment of ties in the reproduction of biocatalyst activities observed pH can hardly be controlled or adapted in 96 well plate in shake flasks after cultivation in methanol induced 5 L cultures, with high buffer concentrations as the only way bioreactor cultures. Thus, it was considered interesting to keep at least a desired pH constant. Previous experi- to compare hCYP2C9/hCPR expressing cells cultivated ments in micro scale deep well plate cultures showed in either shake flask and bioreactor regarding their ability best results with a starting pH of 8, which most likely to convert diclofenac. As presented in Additional file 1: drop to lower values during cultivation. Thus, the effect Fig. S1, the conversion rates obtained per mg of biomass of different pH values (pH 5–7) on the production of dry cell weight (DCW) for catalysts generated in this CYP2C9/CPR whole cells in chemostat cultivations at study were rather similar. After confirmation that cells Garrigós‑Martínez et al. Microb Cell Fact (2021) 20:90 Page 5 of 13 intermediate D = 0.10 ­h−1 was assessed systematically. 3 g ­L−1 (FB5). The last strategy was selected since it was Working in steady state conditions for a constant specific previously reported to be the best condition for Rhizopus growth rate (µ), provides a robust and reliable cultiva- oryzae lipase production under the control of P ­ AOX1 pro- tion system that allows studying the effect of a specific moter [42]. operational parameter while keeping the others constant. In designed co-substrate cultivations using the ­MutS Activity tests showed that the best condition was pH = 5, host strain phenotype, it was necessary to maintain a obtaining a specific enzyme activity up to threefold low µ—0.05 ­h−1, otherwise methanol utilisation (MUT) higher than in the other conditions tested (Fig. 2). Con- genes might be repressed [43, 44]. Consequently, metha- sequently, further fed-batch (FB) cultivations were always nol would not be consumed and its impact as promoter carried out at pH = 5. inducer might not be properly assessed. Concerning biomass generation, the identified most Biomass growth characterization in fed‑batch cultures active transformant (named P. pastoris 2C9-PDC/PDF- Since the producer strain identified by screening for best CPR) was able to grow as expected in all three methanol- performing transformants was based on the bidirectional free FB cultures (Fig. 3). Although some growth effect expression construct 2C9-PDC/PDF-CPR, the regula- might have been expected for such membrane protein tion pattern of the used promoters allowed to implement overexpression strain, the recombinant expression did either methanol-induced and/or methanol-free bio- not have any significant limiting effect on the cell growth. process strategies or consecutive induction schemes by Most of the cultures were grown up to reach 100 g ­L−1 derepression followed by induction. Therefore, as the of DCW. Nevertheless, the fermentation performed at current trend of P. pastoris bioprocesses is to avoid the 0.15 ­h−1 (FB1) had to be stopped at a biomass concen- use of toxic and flammable methanol due to its opera- tration around 80 g ­L−1. At this point, the temperature tional associated drawbacks, a primary set of methanol- control could no longer keep constant the tempera- free FB cultures with glycerol as sole carbon source was ture at 25 °C due to the high amount of heat generated performed. In these cultures, the effect of the specific at high specific growth rate. Anyway, numerous samples growth rate (µ) on biocatalyst production was evalu- were taken throughout the process, which allowed the ated by performing glycerol-limiting cultures at different determination of the production-related macrokinetic nominal µ (FBs 1–3). Furthermore, the potential boosting parameters needed for its comparison with the rest of FB effect of methanol on biocatalyst production was addi- cultures. tionally assessed by growing the culture at µ = 0.05 ­h−1 Due to the low biomass-to-methanol yield, YX/MetOH, on glycerol-limiting conditions and adding methanol and the slow specific methanol uptake rate, qMetOH, of the as co-substrate either by pulses as the simplest strategy P. pastoris phenotype (­MutS strains), the use of metha- (FB4), or keeping a constant methanol concentration of nol as a co-substrate did not modify the growth curves Fig. 2 Effect of culture pH for the production of P. pastoris CYP2C9/ Fig. 3 a Biomass time-profile of the fed-batch cultures CPR whole cells biocatalyst. Chemostat cultivations were carried out, performed. Experiment labels: FB1: μ = 0.05 ­h−1; FB2: μ = 0.10 ­h−1; setting the dilution rate to 0.10 ­h−1. Error bars represent the activity FB3: μ = 0.15 ­h−1; FB4: μ = 0.05 ­h−1 + MetOH pulses; FB5: assay standard deviation determined by triplicates μ = 0.05 ­h−1 + constant MetOH = 3 g ­L−1 Garrigós‑Martínez et al. Microb Cell Fact (2021) 20:90 Page 6 of 13 substantially when comparing with methanol-free pro- cesses. Moreover, CYP2C9/CPR production did not alter significantly other relevant cell physiology rates, as indi- cated by the specific glycerol uptake rate (qGly) and bio- mass-to-substrate yield (YX/Gly) (Table 1). The obtained values were similar to those usually reported as standard for P. pastoris strains producing different recombinant proteins [45–47]. Production of CYP2C9/CPR in P. pastoris whole cells in fed‑batch cultures Regarding whole cell biocatalyst activity, the production- time profiles were analysed for the five alternative FB strategies. The obtained results (Fig. 4), demonstrated Fig. 4 Biocatalyst activity time-profiles obtained in fed-batch significant differences in terms of biocatalyst activity cultures 1–5. Experiment labels: FB1: µ = 0.05 ­h−1; FB2: µ = 0.10 ­h−1; between the methanol-free fed-batches cultures (FB 1–3) FB3: µ = 0.15 ­h−1; FB4: µ = 0.05 ­h−1 + MetOH pulses; FB5: µ = 0.05 ­h−1 + constant MetOH = 3 g ­L−1 and those induced with methanol (FB 4–5). In methanol- free processes, a saturation trend was observed in late stage of the profiles at low and mid µ—0.05–1; 0.1 ­h−1, whereas at the highest µ tested—0.15 ­ h−1—always a expressing a heterologous protein regulated by differ- growth-coupled production profile was observed. On ent promoters. In several previous studies was reported the contrary, the methanol induced processes present that cell growth and protein production were coupled an exponential trend up to 3.28 AU ­L−1, thus reaching [33, 45, 48, 49]. As depicted in Fig. 5, the qp in this study biocatalyst activity relevantly higher than methanol-free was clearly correlated with µ within the experimental fermentations. Specifically, when further inducing the range tested. Results pointed out a slight increase of qp system with methanol a twofold increase in biocatalyst when increasing the specific growth rate from 0.05 ­h−1 activity could be observed. (FB1) to 0.10 ­h−1 (FB2). However, a bigger increase was As presented in Fig. 4, a significant but low amount of found when growing cultures at 0.15 ­h−1, reaching up to active biocatalyst was obtained consistently already at the a 2.5-fold qp increase respect to the worst condition. In end of the batch phase in all five cultures. This fact is an contrast, the µ effect on YP/X values might be considered effect of the derepressed expression in non-limiting glyc- negligible as yield values are rather similar for all three erol conditions of the promoters used, ­PPDF/PPDC. How- fed-batches, around 11.5 AU ­gX−1 (FB 1–3). ever, later, at glycerol-limiting conditions the expression With respect to the effect of methanol utilization on was boosted as expected. The similar starting point at the biocatalyst activity production, bioreactor cultivations end of the batches carried out demonstrated also the high with the same µ (FB1, FB4 and FB5) but either adding or reproducibility between the different bioreactor runs. not adding methanol were compared. In all the cultures, It has been widely described that the specific pro- the use of this co-substrate further induced the protein duction rate (qp) is importantly affected by µ when expression, significantly increasing the production rates. Table 1 Main process parameters obtained in the different fed-batch cultures performed for the production of active hCYP2C9/hCPR P. pastoris whole cell biocatalysts Run YX/S qGly qMetOH Biocatalyst qp YP/X YP/MetOH gX ­gS−1 gGly ­gX−1 ­h−1 gMetOH ­gX−1 ­h−1 activity AU ­gX−1 ­h−1 AU ­gX−1 AU AU ­L−1 ­gMetOH−1 FB1 0.47 0.102 – 1.78 0.67 13.4 – FB2 0.51 0.159 – 1.52 0.78 10.4 – FB3 0.46 0.272 – 1.17 1.71 12.9 – FB4 0.55 0.084 0.029 3.28 1.15 21.9 39.7 FB5 0.53 0.087 0.023 2.68 1.10 21.1 47.8 µ: specific groth rate, ­YX/S: Biomass to substrate yield; qGly: specific glycerol uptake rate; qMetOH: specific methanol uptake rate; qP: specific producrtion rate; YP/X: Product to biomassa yield; YP/MetOH: Product to methanol yield; FB1: 0.05 ­h−1; FB2: 0.10 h ­ −1; FB3: 0.15 h ­ −1; FB4: 0.05 h ­ −1 + MetOH pulses; FB5: 0.05 ­h−1 + MetOH constant 3 g ­L−1 Garrigós‑Martínez et al. Microb Cell Fact (2021) 20:90 Page 7 of 13 (YP/X), the process should be prolonged in time, growing the cells not only at a low µ, but also adding methanol as co-substrate, preferably through pulses. In this way, up to 3.28 AU ­L−1 could be achieved, this representing a three- fold increase respect to the worst condition. On the con- trary, if qp is selected as the parameter to be optimized, the best operational strategy would be a carbon-limit- ing fed-batch operation using glycerol as a sole carbon source at high µ. Specifically, setting the µ at the highest value tested—0.15 ­h−1 (FB3); a threefold increase could be achieved respect to worst conditions. However, from an application point of view, in order Fig. 5 Product-related parameters obtained in fed-batch cultures: to select the optimal bioprocess strategy to produce an specific product generation rate (qp) and product-to-biomass active biocatalyst like the presented in this work, the yield (YP/X). Experiment labels: FB1: µ = 0.05 ­h−1; FB2: µ = 0.10 ­h−1; whole workflow including downstream steps should also FB3: µ = 0.15 ­h−1; FB4: µ = 0.05 ­h−1 + MetOH pulses; FB5: µ = 0.05 ­h−1 + constant MetOH = 3 g ­L−1 be taken into consideration during the optimization pro- cess. Since this bioprocess approach uses whole cells, downstream processing can be considered rather simple and not costly. Therefore, the need to reach high bio- More specifically, it led to qp improvements of up to 1.72- catalyst activities, which is essential to be able to afford fold, and up to 1.63-fold of YP/X. On the one hand, this high cost separations, should not be considered as prior- can be due to the effect that recombinant gene expres- ity. Thus, taking into consideration the different criteria sion is highly induced in presence of methanol with both discussed previously, the methanol-free carbon-limiting promoters used, P ­ PDF and P­ PDC. However, the observed strategy based on glycerol at highest µ should be con- effects might also be attributed to other differences in sidered the best alternative that allows maximization of carbon source dependent systems-level organization, the product generation rates (qp). Despite lower biocata- such as for example co-factor regeneration, membrane lyst activity levels reached, avoiding the use of methanol permeability, or the expression levels of other proteins would allow to avoid the important methanol-associ- [40]. As a summary, a comparison of the main process ated operational drawbacks. These relevant drawbacks, parameters is presented in Table 1. When comparing including high oxygen demand, heat production, and both methanol-based culture strategies implemented, additional costs related to methanol storage, transporta- strikingly, controlling the methanol concentration in tion and handling need to be considered for further scal- the culture broth (FB5) only provided a slight increase ing steps of the bioprocess [38]. of about 20% in terms of active biocatalyst produced per gram of methanol fed (YP/MetOH) compared to the methanol-pulses based strategy (FB4). Moreover, other Preparative scale demonstration product-related parameters—qp, YP/X—are rather similar Biomass produced by controlled cultivation in biore- when controlling the methanol at a constant concentra- actors was applied to evaluate its activity for prepara- tion. Consequently, since the results obtained are similar, tive biotransformation of ibuprofen to its respective the methanol pulses-based process can be considered as CYP2C9 metabolites. During Phase I metabolism in the preferable due to its technical simplicity comparing with human liver the drug ibuprofen is mainly converted into a cultivation strategy based on a closed-loop methanol 1-hydroxyibuprofen, 2-hydroxyibuprofen, and 3-hydrox- control. yibuprofen [24]. CYP2C8 and CYP2C9 are known as the In order to maximize the bioprocess efficiency, dif- main enzymes responsible for its stereoselective hydrox- ferent optimization criteria based on diverse per- ylation [50]. However, as reported by Neunzig et al., formance indexes can be considered; among them, CYP2C9 is the only known CYP able to produce relevant biocatalyst activity, productivity and yield are the most amounts of 3-OH-ibuprofen [24]. To evaluate the feasi- often selected. In P. pastoris fermentation processes, bility of a preparative scale production using P. pastoris, the culture is often limited by the maximum amount of CYP2C9 expressing cells made by reliable cultivation in biomass that can be reached, about 100 g ­L−1 in DCW. bioreactors were applied in a whole cell biotransforma- Then, the optimization method should aim to maximise tion approach using a reaction volume of 500 mL and the production before reaching this limiting cell concen- an initial substrate concentration of 2 mM. Within 16 h tration. To maximise the product activity (P) and yield a rather linear increase in conversion to 25% of total product was observed (Fig. 6), which further increased Garrigós‑Martínez et al. Microb Cell Fact (2021) 20:90 Page 8 of 13 confirmed the potential of recombinant CYP2C9 cata- lysed biotransformations using P. pastoris whole cells as biocatalyst. In terms of space–time yield and specific pro- duction rate, values obtained with the P. pastoris whole cell biocatalyst were significantly higher—up to 55% and 21%, respectively—than the reported in the literature, in which the same proteins were expressed in a fission yeast system [24]. Moreover, the ratios between the two prod- ucts (M2/M1) were in average around 5, and therefore, even higher than previous published values [24], con- firming the high selectivity of this biotransformation for the 3-OH-metabolite. The main reaction parameters are presented in Table 2. Despite the promising results obtained in the proof- of-concept ibuprofen biotransformation, the biocatalytic reaction could be further improved by addressing the impact of some important parameters such as dissolved oxygen availability, efficiency of NADPH regeneration system and/or heme group generation. Conclusions Recombinant production of hCYPs has been intensively investigated and it is expected to continue attracting high attention by the pharmaceutical and chemical industry. Fig. 6 a Ibuprofen oxidation reactions catalysed by human CYPs. b Preparative scale ibuprofen oxidation by P. pastoris CYP2C9/CPR Especially, a fast track from catalyst design to biocatalyst whole cells biocatalyst. Substrate concentration: 2 mM, reaction manufacturing and scalable biotransformation is essential volume: 500 mL to react on urgent and sometimes unexpectedly arising needs in drug development. In this work, a whole scala- ble bioprocess development for fine-tuned co-expression of CYP2C9 and CPR in the broadly available and efficient up to 40% within 86 h of reaction. As expected from expression host P. pastoris has been successfully attained. small scale results (data not shown), metabolite 2 (35%), To achieve this objective, an innovative bidirectional which was identified as 3-OH-ibuprofen by NMR spec- expression system using a small representative bidirec- troscopy (Additional file 2), was mainly produced, while tional promoter library with different regulation patterns metabolite 1 was only found to an extent of 5% (2-OH- was successfully implemented to achieve an optimized ibuprofen). No relevant amounts of metabolite 3 could be fine-tuned co-expression clone for the hCYP29 and its detected (1-OH-ibuprofen), which was observed as a typ- natural redox partner hCPR. Furthermore, for the first ical product in previous experiments employing a fungal time, functionally active P. pastoris CYP2C9 whole cell bifunctional P450 homologue of the bacterial CYP102 biocatalysts were obtained from reliable and optimized [6]. bioreactor cultures, which finally were successfully tested A total product concentration of 0.8 mM was obtained for a biotransformation of interest based on the oxidation starting with 2 mM of substrate ibuprofen with 0.70 mM of ibuprofen. The biocatalyst used in the preparative test of 3-OH-ibu and 0.1 mM of 2-OH-ibu, respectively could convert more than 40% of the added substrate in as products and no 1-OH-ibuprofen. These results a simple batch biotransformation and, more importantly, Table 2 Main parameters obtained for the preparative scale conversion of ibuprofen Reaction time (h) Average substrate conversion (%) Product formed Space–time yield Specific prod. rate Yield (µmol (mM) (µmol product ­L−1 (µmol product g product g ­d−1) whole ­cells−1 ­d−1) whole ­cells−1) M1 M2 M1 M2 M1 M2 M1 M2 86 41.7 0.14 0.69 39 194 1.22 6.06 4.39 21.7 Garrigós‑Martínez et al. Microb Cell Fact (2021) 20:90 Page 9 of 13 it demonstrated high selectivity for the product hydroxy- recombination. Due to the Pichia clonal variability, High lated at position three. Throughput Screening (HTS) were used for the further Moreover, a deep study of using CYP2C9/CPR express- identification of the catalyst with the highest activity. ing P. pastoris whole cells focused on the production kinetics has been carried at bench top bioreactor scale. Whole cell cultivation It has been demonstrated that the µ has a positive effect High‑throughput small scale cultures on qp, demonstrating that the production is coupled with Microscale 96 deep well cultivation was performed based the cell growth. Therefore, selecting a high µ in the bio- on Weis et al. [41]. Single colonies were transferred to process is advantageous to maximize qp on methanol- one well containing 250 µL of BMD1% (buffered mini- free strategies. However, if the objective is to maximise mal dextrose 1%) and cultivated for 60 h at 320 rpm and the YP/X and, therefore, the biocatalyst activity, a bio- 28 °C. Subsequently, 250 µL BMM2% (buffered minimal process using mixed substrates based on the co-feeding methanol 1%) were added to induce gene expression. of glycerol and methanol should be recommended. In 8, 24 and 32 h after the first induction, 50 µL BMM10 these strategies, µ is decreased to the lowest value tested (buffered minimal methanol 5%) were added. After 48 h (0.05 ­h−1) to promote the methanol consumption. of methanol induction, cells were harvested by centrifu- Altogether, it is expected that this work will become gation at 500 rpm. Cultivation in shake flask (Thomson a model, not only for the targeted fast development of 2 L ultra-yield flasks) was performed accordingly starting new human CYP catalysts, but also as applicable reli- with 200 mL. able bioprocesses development for respective whole cell biocatalyst production and the respective catalysed bio- transformations employing recombinant human CYPs Chemostat cultures expressed by P. pastoris. In a similar fashion, other bio- Chemostat cultures were performed based on the meth- catalysts relying on optimized co-expression rations can odology previously described by Garcia-Ortega et al. be developed. This progress will support fast and efficient [48], but using glycerol as a substrate instead of glu- drug development pipelines, which become especially cose. In this case, the different cultures conditions com- challenging and important in case of pandemic diseases, pared were performed at a constant specific growth where new drug candidates need to be advanced to the rate (µ) of 0.10 ­h−1, as intermediate value considering next phase within very short timelines. µmax ≈ 0.20 ­h−1. All the operating parameters were kept constant as reported in the reference along the cultiva- tions except the pH, which was controlled at different Materials and methods values of 5, 6 and 7. Strain generation Escherichia coli/P. pastoris shuttle vectors were based on plasmid pPpT4 [51] and adapted as entry vectors for Fed‑batch cultures quick and simple cloning of bidirectional promoters by Main details regarding medium composition, opera- bisy (bisy GmbH, Hofstaetten, Austria). Synthetic DNA tional parameters and process strategies are described sequences were for these human open reading frames in previous works [45]. All the fed-batch cultures were were obtained from IDT (Integrated DNA Technologies based on a carbon-limited operational strategy using a Inc., Coralville, Iowa, USA). By using Gibson assem- pre-programmed exponential feeding profile of glycerol bly [52], a test set of 7 bidirectional promoters obtained feeding solutions, which allows keeping a constant spe- from bisy GmbH (­PDAS1/DAS2, ­PDAS2/DAS1, ­PPDC/PDF, ­PPDF/ cific growth rate (µ) throughout the process, reaching a PDC, ­PDAS1-D6/DAS2-D8, ­PDAS2-D8/DAS1-D6, and ­PDAS2-699/DAS1- pseudo-stationary state. Upon co-substrate processes, 552) and described previously by Vogl et al. [30] to drive two strategies were compared, one based on periodic expression of full length hCYP2C9 (AL359672) includ- methanol pulse additions and another in which the meth- ing its hydrophobic N-terminal sequence and hCPR anol concentration was kept constant at 3 g ­L−1 by means (accession AAH34277.1) were generated using the men- of a closed control loop. For this last case, the methanol tioned cloning plasmids. These cloning plasmids were concentration was determined by a Methanol Sensor Sys- transformed into the P. pastoris strain BSYBG11 ­MutS tem (Raven Biotech Inc., Vancouver, Canada), and con- (Δaox1, ­MutS), which presents a slow methanol utilisa- trolled by performing a predictive-PI control strategy. tion phenotype, which was obtained from Bisy GmbH. Furthermore, off-line methanol determinations by HPLC Generation of competent P. pastoris cells and transfor- were used to validate the control system. Further details mation were performed according to Lin-Cereghino et al. about the methanol control system are described else- [53]. In this strain generation the target expression cas- where [54, 55]. sette is integrated into the Pichia genome by homologous Garrigós‑Martínez et al. Microb Cell Fact (2021) 20:90 Page 10 of 13 Analytical methods diclofenac to reach a 2 mM substrate concentration and Biomass determination reactions were carried at 28 °C and 200 rpm. After 20 h Biomass concentration was measured by DCW using a of incubation, 50 µL were taken and mixed with 450 µL protocol described elsewhere [56]. RSD was estimated to of MetOH/acetonitrile (1:1), to stop the reaction and to be under 3%. Biomass concentration was also measured extract both the substrate (diclofenac) and the product by optical density at 600 nm using a DR3900 spectropho- (4′OH-diclofenac). Both substrate and product concen- tometer (Hach Lange, Colorado, USA). RSD was esti- trations were determined by HPLC–UV. Before its analy- mated to be below 3%. sis, samples were centrifuged at 12000×g for 2 min and supernatants were filtered (0.45 µm). Carbon source and by‑products quantification The product-related results of the fermentations are The amounts of both carbon sources (glycerol, methanol) presented in terms of activity units (AU). One AU is and potential fermentation by products were analysed by defined as the amount of enzyme which catalyses the HPLC. The chromatography procedure and the column production of 1 µmol of 4-hydroxy-diclofenac per hour specifications were described elsewhere [57]. RSD was under the assay conditions. Thus, this parameter pro- estimated to be lower than 1%. vides an enzyme titer value in terms of (AU ­L−1). Whole cell bioconversions High throughput screening (HTS) HPLC–UV/MS Whole cell screenings using diclofenac as substrate were HPLC/UV analysis were performed on a HPLC Dionex carried out in 96-well microtiter plates. After the har- Ultimate 3000 (Dionex Corporation, Sunnyvale, CA, vest, cells were washed twice with 200 µL of assay buffer USA) series using a CORTECSTM C18 + 2.7 μm (100 mM KPi, pH 7.4) and then the biomass concentra- 4.6 × 150 mm column (Waters), at a temperature of 40 °C tion was determined by optical density determination and a detector wavelength of 275 nm. The mobile phase ­(OD600). Cells were frozen at − 20 °C for several days to was composed of Trifluoroacetic acid 0.1% (v/v) in water obtain a better membrane permeability. Subsequently, (Eluent A) and Trifluoroacetic acid 0.095% (v/v) in Ace- 4 µL of a 100 mM diclofenac stock solution in methanol tonitrile 80% (Eluent B). Gradient-elution was performed (MetOH) were added to each well and conversions were at 0.7 mL ­min−1 as follows: 0.00–0.50 min: 95% A, 5% B; performed for 20 h at 28 °C and 320 rpm. Reactions were 0.50–13 min: 55% A, 45% B; 13.00–13.50 min: 20% A, stopped by addition of 200 µL of MetOH/acetonitrile 80% B; 13.5–15.00 min: 0% A, 100% B; 15.00–15.50 min: (1:1) and vigorously mixed. After centrifugation (10 min, 0% A, 100% B; 15.50–20.00 min: 95% A, 5% B. Retention 3200×g) 200 µL of supernatant were transferred into a times were 12.04 for diclofenac and 8.22 4′-hydroxy- fresh 96 well microtiter plate (polypropylene, V-shaped), diclofenac. For calculations, peak areas of the UV chro- sealed and analysed by HPLC–MS. matograms recorded at 239 nm were used. Fluorometric whole cell screenings based on the HPLC–MS was performed on a HPLC 1200 series 7-methoxy-4-(trifluoromethyl)-2H-chromen-2-one (Agilent technologies, Santa Clara, CA, USA) equipped tion (ESI) unit as described previously [58]. A Kinetex® (MFC) assay were performed using 95 μL of the cell sus- with an MSD SL detector with an electron spray ioniza- pension in assay solution using black microtiter plates. The reaction was started by the addition of 5 μL of 1 mM 2.6 μm C18 100 Å, LC Column 50 × 4.6 mm column MFC to reach a final concentration 50 μM. Fluorescent (Phenomenex) was used. The mobile phase was com- metabolite formation was quantified measuring the fluo- posed of water (acidified with 0.1% formic acid) and rescence of the product 7-hydroxy-4-(trifluoromethyl)- acetonitrile (ACN). Gradient-elution was performed at 2H-chromen-2-one (HFC). The fluorescence signal of the 1.3 mL/min as follows: 0.00–1.00 min: 10% ACN; 1.00– product formed (ex 410 nm/em 538 nm) was monitored 3.50 min: 10–100% ACN; 3.50–4.00 min: 10% ACN. To every minute for a period of 1 h using a SynergyMX plate allow a flow rate of 1.3 mL ­min−1, a splitter (2:1 ratio) reader (BioTek Instruments Inc, USA). During the reac- was implemented in front of the ESI unit. 4′-hydroxy- tion the microtiter plates were kept at 30 °C. diclofenac (m/z 312) eluted after 2.2 min and diclofenac (m/z 296) after 2.4 min. For analysis of ibuprofen and its metabolites a stepwise gradient was used at 25 °C as Bioconversion reactor described previously [6]: 0.00–1.00 min: 20% ACN; 1.00– For bioreactor samples, the enzymatic assays were per- 3.00 min: 20–100% ACN; 3.00–5.00 min: 100% ACN; formed in 50 mL falcon tubes. In this case, the washed 5.00–6.00 min: 20% ACN re-equilibration. Metabolites cells were diluted to O­ D600 of 5 as a standard ­OD600 were detected at 239 nm (VWD) and MS spectra were for all reactions. Then, 2 mL of cells were mixed with used to support data evaluation. Garrigós‑Martínez et al. Microb Cell Fact (2021) 20:90 Page 11 of 13 Preparative scale conversion of ibuprofen Standortagentur Tirol, Government of Lower Austria and Business Agency Vienna through the Austrian FFG-COMET—Funding Program. Preparative scale conversions were performed as described previously [6, 58]. A total reaction volume of Authors’ contributions 500 mL including the whole cell catalysts correspond- AW and CS constructed the producer strains and performed the HTS. JGM performed all the cultures in bench-scale bioreactors. CR performed the pre‑ ing to O ­ D600 = 150 was used for oxidation of 206 mg parative scale demonstration. JLM, FV, AG and XGO, conceived and supervised ibuprofen dissolved in DMSO. Conversion was per- the study. JGM, CR, JLM, FV, AG and XGO drafted the manuscript. All authors formed in a baffled 2 L shake flask over a period of 86 h. read and approved the final manuscript. Samples of 200 µL were withdrawn at different times Funding and analysed as described above. This work received funding from the EU project ROBOX (Grant Agreement No. 635734) under EU’s H2020 Programme Research and Innovation actions H2020-LEIT BIO-2014-1. Nuclear magnetic resonance spectroscopy (NMR) Availability of data and materials For recording NMR spectra, a Bruker Avance III The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request. 300 MHz FT NMR spectrometer with autosampler was used (300.36 MHz-H-NMR and 75.5 MHz-C-NMR). Declarations The residual protonated solvent signal serve as inter- nal standard for interpretation of the chemical shifts Ethics approval and consent to participate δ (H-,C-NMR). To facilitate the interpretation, the Not applicable. C-spectra were proton decoupled to gain better identi- Consent for publication fication of the peaks. Not applicable. The chemical shift δ is indicated in ppm (parts per Competing interests million) and the coupling constant J in Hz (Hertz). For The authors of TU Graz and bisy declare that they have commercial interests the signal multiplicities the following abbreviations in the use of bidirectional promoters for protein coexpression. Bisy GmbH were most commonly used: s (singlet), bs (broad sin- owns and commercializes tools for recombinant bidirectional coexpression of genes. glet), d (doublet), t (triplet), q (quadruplet), m (multi- plet). Quaternary carbons are labelled as Cq. Author details 1 Department of Chemical, Biological and Environmental Engineering, School of Engineering, Universitat Autònoma de Barcelona, 08193 Bellaterra (Cerdan‑ Supplementary Information yola del Vallès), Spain. 2 Institute of Molecular Biotechnology, Graz University The online version contains supplementary material available at https://​doi.​ of Technology, Petersgasse 14, 8010 Graz, Austria. 3 Bisy GmbH, Wuenschen‑ org/​10.​1186/​s12934-​021-​01577-4. dorf 292, 8200 Hofstaetten/Raab, Austria. 4 Austrian Centre of Industrial Biotechnology (ACIB), Petersgasse 14, 8010 Graz, Austria. Additional file 1: Figure S1. Whole cell hydroxylation of diclofenac by P. Received: 9 September 2020 Accepted: 7 April 2021 pastoris CYP2C9/CPR whole cells cultivated respectively in bioreactor and in shake flask (UYF = ultra-yield flask). The wild type strain BSYBG11 was used as a negative control. Additional file 2. Nuclear Magnetic Resonance Spectroscopy (NMR) analyses. Figure S2A. Product 1, 1H NMR (300 MHz, DMSO): δ 12.16 (bs, References 1H); 7.15 (s, 4H); 3.96 (bs, 1H); 3.62 (q, J = 7.0 Hz, 1H); 2.61 (s, 2H); 1.31 1. Slaughter RL, Edwards DJ. Recent advances: the cytochrome P450 (d, J = 14.4 Hz, 3H); 1.05 (s, 6H) ppm. Figure S2B. Product 1, 13C NMR enzymes. Ann Pharmacother. 1995;29:619–24. (75.5 MHz, DMSO): δ 175.5; 138.5; 137.4; 130.4; 126.6; 69.3; 49.0; 44.3; 29.2; 2. Bernhardt R. Cytochromes P450 as versatile biocatalysts. J Biotechnol. 18.5 ppm. Figure S2C. Product 2, 1H NMR (300 MHz, DMSO): δ 12.25 (s, 2006;25:128–45. 1H); 7.24–7.07 (m, 4H); 4.51 (s, 1H); 3.62 (dd, J = 13.8; 6.8 Hz; 1H); 3.23 (dd, 3. Schroer K, Kittelmann M, Lütz S. Recombinant human cytochrome J = 15.1; 8.3 Hz, 2H); 2.68 (dd, J = 13.2; 5.7 Hz, 1H); 2.25 (dd, J = 13.1; 8.3 Hz, P450 monooxygenases for drug metabolite synthesis. Biotechnol 1H); 1.76 (dd, J = 12.9; 6.5 Hz, 1H); 1.34 (d, J = 7.0 Hz, 3H); 0.78 (d, J = 6.6 Hz, Bioeng. 2010;106:699–706. 3H) ppm. Figure S2D. Product 2, 13C NMR (75.5 MHz, DMSO): δ 175.5; 4. Daly A, Rettie A, Fowler D, Miners J. Pharmacogenomics of CYP2C9: 139.4, 138.5; 129.1; 127.2; 65.6; 44.3; 38.7; 37.6; 18.6; 16.5 ppm. functional and clinical considerations. J Pers Med. 2017;8:1. 5. Guengerich FP. Human cytochrome P450 enzymes. In: Cytochrome P450. Cham: Springer; 2015. p. 523–785. Acknowledgements 6. Rinnofner C, Kerschbaumer B, Weber H, Glieder A, Winkler M. We would like to acknowledge Elisabeth Höppel for her skillful support in Cytochrome P450 mediated hydroxylation of ibuprofen using Pichia experimental work. The Spanish group is member of 2017-SGR-1462 and pastoris as biocatalyst. Biocatal Agric Biotechnol. 2019;17:525–8. the Reference Network in Biotechnology (XRB, Generalitat de Catalunya). 7. Geier M, Braun A, Emmerstorfer A, Pichler H, Glieder A. Production of J. Garrigós-Martínez acknowledges a PIF scholarship from the Universitat human cytochrome P450 2D6 drug metabolites with recombinant Autònoma de Barcelona. microbes—a comparative study. Biotechnol J. 2012;7:1346–58. 8. Kaluzna I, Brummund J, Schuermann M. Production of diclofenac Disclaimer metabolites by applying cytochrome P450 technology. Chim Oggi/ This publication reflects the author’s view and the Agency is not responsible Chem Today. 2017;35(6):55–8. for any use that may be made of the information it contains. This work has 9. Litzenburger M, Kern F, Khatri Y, Bernhardt R. Conversions of tricyclic also been supported by bisy GmbH and the Austrian BMWD, BMVIT, SFG, antidepressants and antipsychotics with selected P450s from Sorangium cellulosum so ce56. Drug Metab Dispos. 2015;43:392–9. Garrigós‑Martínez et al. Microb Cell Fact (2021) 20:90 Page 12 of 13 10. Worsch A, Eggimann FK, Girhard M, Bühler CJ, Tieves F, Czaja R, et al. A 32. Garrigós-Martínez J, Nieto-Taype MA, Gasset-Franch A, Montesinos-Seguí novel cytochrome P450 mono-oxygenase from Streptomyces platensis JL, Garcia-Ortega X, Valero F. Specific growth rate governs AOX1 gene resembles activities of human drug metabolizing P450s. Biotechnol expression, affecting the production kinetics of Pichia pastoris (Komaga- Bioeng. 2018;115:2156–66. taella phaffii) PAOX1-driven recombinant producer strains with different 11. Urlacher VB, Girhard M. Cytochrome P450 monooxygenases in biotech‑ target gene dosage. Microb Cell Fact. 2019;18:187. nology and synthetic biology. Trends Biotechnol. 2019;37:882–97. 33. Rebnegger C, Graf AB, Valli M, Steiger MG, Gasser B, Maurer M, et al. In 12. Urlacher VB, Girhard M. Cytochrome P450 monooxygenases: an Pichia pastoris, growth rate regulates protein synthesis and secretion, update on perspectives for synthetic application. Trends Biotechnol. mating and stress response. Biotechnol J. 2014;9:511–25. 2012;30:26–36. 34. Nieto-Taype MA, Garrigós-Martínez J, Sánchez-Farrando M, Valero F, Gar‑ 13. Cojocaru V, Balali-Mood K, Sansom MSP, Wade RC. Structure and dynam‑ cia-Ortega X, Montesinos-Seguí JL. Rationale-based selection of optimal ics of the membrane-bound cytochrome P450 2C9. PLoS Comput Biol. operating strategies and gene dosage impact on recombinant protein 2011;7:e1002152. production in Komagataella phaffii (Pichia pastoris). Microb Biotechnol. 14. Williams PA, Cosme J, Sridhar V, Johnson EF, McRee DE. Mammalian 2019;13(2):315–27. microsomal cytochrome P450 monooxygenase. Mol Cell. 2000;5:121–31. 35. García-Ortega X, Cámara E, Ferrer P, Albiol J, Montesinos-Seguí JL, Valero 15. Nebert DW, Wikvall K, Miller WL. Human cytochromes P450 in health and F. Rational development of bioprocess engineering strategies for recom‑ disease. Philos Trans R Soc Lond B Biol Sci. 2013;368:20120431. binant protein production in Pichia pastoris (Komagataella phaffii) using 16. Hausjell J, Halbwirth H, Spadiut O. Recombinant production of the methanol-free GAP promoter. Where do we stand? N Biotechnol. eukaryotic cytochrome P450s in microbial cell factories. Biosci Rep. 2019;53:24–34. 2018;38:BSR20171290. 36. Nieto-Taype MA, Garcia-Ortega X, Albiol J, Montesinos-Seguí JL, Valero F. 17. Olaofe OA, Fenner CJ, Gudiminchi RK, Smit MS, Harrison ST. The influence Continuous cultivation as a tool toward the rational bioprocess develop‑ of microbial physiology on biocatalyst activity and efficiency in the termi‑ ment with Pichia Pastoris cell factory. Front Bioeng Biotechnol. 2020;8:632. nal hydroxylation of n-octane using Escherichia coli expressing the alkane 37. Vogl T, Kickenweiz T, Pitzer J, Sturmberger L, Weninger A, Biggs BW, Köhler hydroxylase, CYP153A6. Microb Cell Fact. 2013;12:8. EM, Baumschlager A, Fischer JE, Hyden P, Wagner M, Baumann M, Borth 18. Quehl P, Hollender J, Schüürmann J, Brossette T, Maas R, Jose J. Co- N, Geier M, Ajikumar PK, Glieder A. Engineered bidirectional promoters expression of active human cytochrome P450 1A2 and cytochrome enable rapid multi-gene co-expression optimization. Nat Commun. P450 reductase on the cell surface of Escherichia coli. Microb Cell Fact. 2018;9:3589. 2016;15:26. 38. Vogl T, Sturmberger L, Kickenweiz T, Wasmayer R, Schmid C, Hatzl A-M, 19. Forman V, Bjerg-Jensen N, Dyekjær JD, Møller BL, Pateraki I. Engineering Gerstmann MA, Pitzer J, Wagner M, Thallinger GG, Geier M, Glieder A. A of CYP76AH15 can improve activity and specificity towards forskolin toolbox of diverse promoters related to methanol utilization: functionally biosynthesis in yeast. Microb Cell Fact. 2018;17:181. verified parts for heterologous pathway expression in Pichia pastoris. ACS 20. Hanlon SP, Friedberg T, Wolf CR, Ghisalba O, Kittelmann M. Recombinant Synth Biol. 2016;5:172–86. yeast and bacteria that express human P450s: bioreactors for drug 39. Fischer JE, Hatzl A-M, Weninger A, Schmid C, Glieder A. Methanol inde‑ discovery, development, and biotechnology. In: Schmid RD, Urlacher VB, pendent expression by Pichia pastoris employing de-repression technolo‑ editors. Modern biooxidation, vol. 10. Weinheim: Wiley; 2007. p. 233–52. gies. J Vis Exp. 2019;143:e58589. 21. Rushmore TH, Reider PJ, Slaughter D, Assang C, Shou M. Bioreactor 40. Rußmayer H, Buchetics M, Gruber C, Valli M, Grillitsch K, Modarres G, Guer‑ systems in drug metabolism: synthesis of cytochrome P450-generated rasio R, Klavins K, Neubauer S, Drexler H, Steiger M, Troyer C, Al Chalabi A, metabolites. Metab Eng. 2000;2:115–25. Krebiehl G, Sonntag D, Zellnig G, Daum G, Graf AB, Altmann F, Koe‑ 22. Vail RB, Homann MJ, Hanna I, Zaks A. Preparative synthesis of drug llensperger G, Hann S, Sauer M, Mattanovich D, Gasser B. Systems-level metabolites using human cytochrome P450s 3A4, 2C9 and 1A2 with organization of yeast methylotrophic lifestyle. BMC Biol. 2015;13:80. NADPH-P450 reductase expressed in Escherichia coli. J Ind Microbiol 41. Weis R, Luiten R, Skranc W, Schwab H, Wubbolts M, Glieder A. Reliable Biotechnol. 2005;32:67–74. high-throughput screening with Pichia pastoris by limiting yeast cell 23. Drăgan C-A, Peters FT, Bour P, Schwaninger AE, Schaan SM, Neunzig I, death phenomena. FEMS Yeast Res. 2004;5:179–89. et al. Convenient gram-scale metabolite synthesis by engineered fission 42. Barrigón JM, Montesinos JL, Valero F. Searching the best operational yeast strains expressing functional human P450 systems. Appl Biochem strategies for Rhizopus oryzae lipase production in Pichia pastoris Mut+ Biotechnol. 2011;163:965–80. phenotype: methanol limited or methanol non-limited fed-batch cul‑ 24. Neunzig I, Göhring A, Drăgan C-A, Zapp J, Peters FT, Maurer HH, et al. tures? Biochem Eng J. 2013;75:47–54. Production and NMR analysis of the human ibuprofen metabolite 43. Arnau C, Casas C, Valero F. The effect of glycerol mixed substrate on the 3-hydroxyibuprofen. J Biotechnol. 2012;157:417–20. heterologous production of a Rhizopus oryzae lipase in Pichia pastoris 25. Neunzig I, Widjaja M, Peters FT, Maurer HH, Hehn A, Bourgaud F, Bureik system. Biochem Eng J. 2011;57:30–7. M. Coexpression of CPR from various origins enhances biotransforma‑ 44. Capone S, Horvat J, Herwig C, Spadiut O. Development of a mixed feed tion activity of human CYPs in S. pombe. Appl Biochem Biotechnol. strategy for a recombinant Pichia pastoris strain producing with a de- 2013;170:1751–66. repression promoter. Microb Cell Fact. 2015;14:101. 26. Ramón A, Marín M. Advances in the production of membrane proteins in 45. Garcia-Ortega X, Ferrer P, Montesinos JL, Valero F. Fed-batch operational Pichia pastoris. Biotechnol J. 2011;6:700–6. strategies for recombinant Fab production with Pichia pastoris using the 27. Byrne B. Pichia pastoris as an expression host for membrane protein constitutive GAP promoter. Biochem Eng J. 2013;79:172–81. structural biology. Curr Opin Struct Biol. 2015;32:9–17. 46. Looser V, Lüthy D, Straumann M, Hecht K, Melzoch K, Kovar K. Effects of 28. Suades A, Alcaraz A, Cruz E, Álvarez-Marimon E, Whitelegge JP, Manyosa glycerol supply and specific growth rate on methanol-free production of J, Cladera J, Perálvarez-Marín A. Structural biology workflow for the CALB by P. pastoris: functional characterisation of a novel promoter. Appl expression and characterization of functional human sodium glucose Microbiol Biotechnol. 2017;101:3163–76. transporter type 1 in Pichia pastoris. Sci Rep. 2019;9:1203. 47. Rajamanickam V, Metzger K, Schmid C, Spadiut O. A novel bi-directional 29. Claes K, Vandewalle K, Laukens B, Laeremans T, Vosters O, Langer I, et al. promoter system allows tunable recombinant protein production in Modular integrated secretory system engineering in Pichia pastoris Pichia pastoris. Microb Cell Fact. 2017;16:152. to enhance G-protein coupled receptor expression. ACS Synth Biol. 48. Garcia-Ortega X, Adelantado N, Ferrer P, Montesinos JL, Valero F. A step 2016;5:1070–5. forward to improve recombinant protein production in Pichia pastoris: 30. Çalık P, Ata Ö, Güneş H, Massahi A, Boy E, Keskin A, et al. Recombinant from specific growth rate effect on protein secretion to carbon-starving protein production in Pichia pastoris under glyceraldehyde-3-phosphate conditions as advanced strategy. Process Biochem. 2016;51:681–91. dehydrogenase promoter: from carbon source metabolism to bioreactor 49. Jungo C, Rérat C, Marison IW, von Stockar U. Quantitative characteriza‑ operation parameters. Biochem Eng J. 2015;95:20–36. tion of the regulation of the synthesis of alcohol oxidase and of the 31. Looser V, Bruhlmann B, Bumbak F, Stenger C, Costa M, Camattari A, Foti‑ expression of recombinant avidin in a Pichia pastoris Mut+ strain. Enzyme adis D, Kovar K. Cultivation strategies to enhance productivity of Pichia Microb Technol. 2006;39:936–44. pastoris: a review. Biotechnol Adv. 2014;33:1177–93. Garrigós‑Martínez et al. Microb Cell Fact (2021) 20:90 Page 13 of 13 50. Hamman MA, Thompson GA, Hall SD. Regioselective and stereoselec‑ under ­PAOX1 control: a modeling aided approach. J Chem Technol Bio‑ tive metabolism of ibuprofen by human cytochrome P450 2C. Biochem technol. 2018;93:3208–18. Pharmacol. 1997;1:33–41. 56. Cos O, Serrano A, Montesinos JL, Ferrer P, Cregg JM, Valero F. Combined 51. Näätsaari L, Mistlberger B, Ruth C, Hajek T, Hartner FS, Glieder A. Deletion effect of the methanol utilization (Mut) phenotype and gene dosage on of the Pichia pastoris KU70 homologue facilitates platform strain genera‑ recombinant protein production in Pichia pastoris fed-batch cultures. J tion for gene expression and synthetic biology. PLoS ONE. 2012;7:e39720. Biotechnol. 2005;116:321–35. 52. Gibson DG, Young L, Chuang R-Y, Venter JC, Hutchison CA, Smith HO. 57. Garcia-Ortega X, Valero F, Montesinos-Seguí JL. Physiological state Enzymatic assembly of DNA molecules up to several hundred kilobases. as transferable operating criterion to improve recombinant protein Nat Methods. 2009;6:343–5. production in Pichia pastoris through oxygen limitation. J Chem Technol 53. Lin-Cereghino J, Wong WW, Xiong S, Giang W, Luong LT, Vu J, et al. Biotechnol. 2017;92:2573–82. Condensed protocol for competent cell preparation and transformation 58. Migglautsch AK, Willim M, Schweda B, Glieder A, Breinbauer R, Winkler of the methylotrophic yeast Pichia pastoris. Biotechniques. 2005;38:44, 46, M. Aliphatic hydroxylation and epoxidation of capsaicin by cytochrome 48. P450 CYP505X. Tetrahedron. 2018;74:6199–204. 54. Ponte X, Montesinos-Seguí JL, Valero F. Bioprocess efficiency in Rhizopus oryzae lipase production by Pichia pastoris under the control of P ­ AOX1 is oxygen tension dependent. Process Biochem. 2016;51:1954–63. Publisher’s Note 55. Ponte X, Barrigón JM, Maurer M, Mattanovich D, Valero F, Montesinos- Springer Nature remains neutral with regard to jurisdictional claims in pub‑ Seguí JL. Towards optimal substrate feeding for heterologous protein lished maps and institutional affiliations. production in Pichia pastoris (Komagataella spp.) fed-batch processes Ready to submit your research ? Choose BMC and benefit from: • fast, convenient online submission • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations • maximum visibility for your research: over 100M website views per year At BMC, research is always in progress. Learn more biomedcentral.com/submissions