The recombinant protein market has grown steadily over the past years due to an increasing demand caused by the biotechnological industry (De Brabander et al., 2023; Rettenbacher et al., 2022; Bachhav et al., 2023). One of the key representatives of the recombinant protein market are enzymes (Barone et al., 2023). Recombinant enzymes initially became popular due to their widespread use in different domains of life and the significant role they play in multiple metabolic reactions (Spohner et al., 2015; Deveryshetty and Antony, 2021). An example thereof is the enzyme class of oxidoreductases (EC 1), which are present in all kingdoms of life (Kareem, 2020). Unspecific peroxygenases (UPO, EC 1.11.2.1) belong to the enzyme class of oxidoreductases that can catalyze non-activated C-H- and C=C- bonds as well as C-C bond cleavage while requiring only H₂O₂ for catalytic reactions (Ullrich et al., 2004; Kinner et al., 2021). However, due to the highly complex structures present in UPOs (disulfide bridges and heme moiety), soluble expression in prokaryotic hosts is challenging (Manta et al., 2019). As a result, eukaryotic hosts are commonly employed to produce UPOs recombinantly (Besleaga et al., 2024). The recombinant expression of UPOs was tested in various eukaryotic and prokaryotic hosts, whereas Komagataella phaffii (K. phaffii) was found to be the most favorable expression system (Hofrichter et al., 2022). Recently, for different short UPOs (AbrUPO, MweUPO-1), fed-batch bioreactor cultivations of recombinant K. phaffii strains were reported, which resulted in protein yields of around 0.7 g/L (Sánchez-Moreno et al., 2024; Schmitz et al., 2023).

Yeasts possess many advantages, such as fast growth, cheap media requirements, easy genetic manipulation tools, high cell density fermentations, the ability to perform post-translational modifications, and high secretory efficiency (Barone et al., 2023; Bernauer et al., 2021; Vijayakumar and Venkataraman, 2023; Karbalaei et al., 2020). Komagataella phaffii was implemented successfully for the recombinant production of various oxidoreductases, like peroxidases (EC 1.11.1.7 and 1.11.1.13) (Zhang et al., 2020; Pekarsky et al., 2018), peroxygenases (EC 1.11.2.1) (Gomez de Santos et al., 2023; Ebner et al., 2023; Tonin et al., 2021), catalases (EC 1.11.1.6) (Gómez et al., 2019), heme oxygenases (EC 1.14.99.3) (Mei et al., 2024) and cytochrome P450s (EC 1.14.14.1) (Hausjell et al., 2020; Garrigós-Martínez et al., 2021b).

Recombinant protein production in K. phaffii is often driven by the promoter of the major alcohol oxidase gene, P _AOX1 , which is inducible by methanol (Bernat-Camps et al., 2023; Bustos et al., 2022). The P _AOX1 promoter shows tight regulation and strong induction by methanol as a sole carbon source (Bustos et al., 2022; Wurm and Spadiut, 2019), allowing a distinct separation between biomass and product formation. However, methanol imposes several disadvantages in industrial-scale production, like high flammability, cell toxicity, high oxygen consumption, and high heat of combustion (Dalvie et al., 2022; Pan et al., 2022; Wollborn et al., 2022). Due to the disadvantages of methanol, alternative induction systems are often required, especially for continuous cultivation (Nieto-Taype et al., 2020). For instance, for many carbon source-dependent promoters of methylotrophic yeasts, methanol-free promoter activation can be achieved by derepression (applying low feeding rates of repressing carbon source) (Vogl et al., 2020; Weinhandl et al., 2014). However, derepression of P _AOX1 leads to low promoter activation, with recombinant protein expression levels of only 2%–4%, when compared to methanol-induced P _AOX1 (Vogl et al., 2018; Yang and Zhang, 2018; Vogl et al., 2020). Alternative promotors, however, are well designed to be induced under carbon derepression, with one of them being the orthologous formate dehydrogenase promoter from Hansenula polymorpha (P _DF , a commercial variant of the promoter) (Vogl et al., 2020). Induction of the P _DF by derepression (in K. phaffii) reached expression levels of approximately 75% compared to the methanol-induced P _AOX1 . Moreover, induction by derepression allows tight regulation of the promoter by a simple variation in feeding the repressing carbon source (Vogl et al., 2020).

Despite novel promotor systems available, many oxidoreductases suffer from low production yields due to their complexity (Ebner et al., 2023). A strategy to facilitate the expression levels is the co-expression of either chaperones or foldases (Irvine et al., 2014; Raschmanová et al., 2021). Here, for convenience reasons, a bi-directionalized promoter system can be implemented to express both proteins simultaneously from a single expression construct (Rajamanickam et al., 2017; Navone et al., 2021). Multiple chaperones are available for co-expression in K. phaffii, some of them, such as PDI (protein disulfide isomerase) and HAC1 (the unfolded protein response transcription factor), showed a significant improvement in protein expression and increase in product titers (Raschmanová et al., 2021). Moreover, a recent study also showed the feasibility of PDI co-expression for the production of the long archetype UPO (PaDa-I) in K. phaffii, as the product titer of recombinant UPO in the supernatant was doubled (Zhao et al., 2024).

When it comes to bioreactor-scale cultivation mode, the fed-batch is the current state-of-the-art for microbials to produce recombinant proteins (García-Ortega et al., 2019; Duman-Özdamar and Binay, 2021). However, continuous cultivation proved to be a powerful tool to achieve high space-time yields (Rahimi et al., 2019; de Macedo Robert et al., 2019; Nieto-Taype et al., 2020). Unfortunately, in K. phaffii chemostat cultivations (secreting recombinant human serum albumin) with specific growth rates (µ) set below 0.075 h^-1, pseudohyphae growth was observed (Rebnegger et al., 2014). This was presumably caused by the upregulation of floculin (FLO11) gene (De et al., 2020). This is a major issue in recombinant protein production, as pseudohyphal growth is known to hinder protein secretion, leading to decreased productivity (Puxbaum et al., 2016). Interestingly, Garrigós-Martínez et al. recently employed a derepressible bi-directionalized promoter system to co-produce hCYP2C9 and the complementary reductase (hCPR) in chemostat cultivation with a D of 0.05 h^-1 and did not report any pseudohyphae growth (Garrigós-Martínez et al., 2021b). This is surprising, considering the induction mechanism of derepressible promoters. Low dilution rates are required in chemostat cultivation, which is reported to trigger pseudohyphae growth. To our knowledge, no studies have further investigated pseudohyphal growth in chemostat cultivations using a derepressible promoter system for recombinant protein production with K. phaffii.

As literature is contradictory, this study aimed to shed more light on pseudohyphal growth in derepressed chemostat cultivations whilst expressing a recombinant target enzyme. For this purpose, we used a bi-directionalized promoter system for derepressible production of an unspecific peroxygenase (UPO, EC 1.11.2.1) and constitutive co-expression of a chaperon (PDI). This study aimed to mechanistically describe the impact of dilution rates on pseudohyphae growth, with the goal of determining stable production conditions in chemostat cultivations.

In a pre-study, unspecific peroxygenase (AnoUPO) expression employing a derepressible bi-directionalized promoter combined with PDI co-expression was already described as a viable production strategy in fed-batch cultivations (Besleaga et al., 2024). In order to potentially improve space-time yields achieved within fed-batch production, continuous cultivations were evaluated in this study. However, considering the induction mechanism of the P _DF , low growth rates are required for induction. Unfortunately, this triggers pseudohyphae growth in chemostat cultivations with K. phaffii, hindering extracellular recombinant protein production (Puxbaum et al., 2016; Aggarwal and Mishra, 2022). Pseudohyphae growth was well described in chemostats using a monodirectional promoter system for recombinant protein production when applying dilution rates below 0.075 h^-1 (Rebnegger et al., 2016; Rebnegger et al., 2014). Contrary to these results, another study exercising recombinant enzyme production of a different target protein using a derepressible promoter system in chemostat cultivations with K. phaffii did not report any pseudohyphae growth at D = 0.05 h^-1 (Garrigós-Martínez et al., 2021a). Additionally, for a derepressed bi-directionalized promoter system in chemostat cultivations, another study reported no pseudohyphae formation at a growth rate of 0.10 h^-1 (Garrigós-Martínez et al., 2021b). Thus, limited knowledge with contradictory statements is available on pseudohyphae formation in K. phaffii chemostats with derepressed recombinant protein production. Therefore, this study aimed to investigate pseudohyphae growth in chemostat cultivations with a derepressible bi-directionalized promoter system and identify the potential and limitations.

Three different constructs were initially generated, expressing the chaperone protein disulfide isomerase (PDI) with different constitutive promoters (P _GAP , P _UPP , and P _HHT1 ). The strain consisting of the AnoUPO-PDI _UPP construct showed the highest productivity compared to other cultivations (Figure 1) and was thus selected for further experiments. We hypothesize that the high production levels of the AnoUPO-PDI _UPP strains were achieved due to the strength of the P _UPP , which provided more co-expressed chaperones to assist in AnoUPO folding.

For the initial chemostat cultivation, the dilution rate (D) was set to 0.05 h^-1 to verify the findings reported in the literature (Garrigós-Martínez et al., 2021a). Still, the highest productivities in fed-batch cultivations for a similar construct were achieved at a µ set between 0.02 h^-1 and 0.05 h^-1 (Besleaga et al., 2024). Hence, additionally, we investigated a chemostat at a dilution rate of 0.02 h^-1 to assess the transferability of fed-batch conditions to chemostat cultivation. As shown in Figure 2, the biomass-specific enzyme activity in the chemostats was nine-fold lower compared to the fed-batch cultivation. A decrease in productivity of 24% was observed in a different study using induction via derepression of P _DF for Candida antarctica lipase B (CalB) production when comparing chemostat to fed-batch cultivation (Garrigós-Martínez et al., 2021a). A reason for decreased productivity could be pseudohyphae growth, which was observed in the chemostats operated at D of 0.02 h^-1 and 0.05 h^-1 after approximately eight generations. The occurrence of pseudohyphae growth indicates a similar µ dependent behavior as previously published for untransformed K. phaffii strains and strains expressing a recombinant protein under the control of the constitutive P _GAP (Rebnegger et al., 2014; Rebnegger et al., 2016).

To avoid pseudohyphae growth and thus facilitate the recombinant UPO expression, the D was increased to 0.08 h^-1, above the pseudohyphae threshold of 0.075 h^-1 reported in literature (Rebnegger et al., 2014). Applying a D of 0.08 h^-1, enzyme production increased approximately two-fold compared to initial chemostats (Figure 4A). Surprisingly, pseudohyphae growth still occurred (Figure 4B), despite the set D. Even though PDI-co expression was shown to improve the activity of the recombinant UPO (Besleaga et al., 2024), the authors hypothesize that constitutive PDI co-expression might increase the maintenance metabolism. This could lead to an adaptation mechanism of the pseudohyphae phenotype even at higher dilution rates than originally described in literature (Rebnegger et al., 2014). Once pseudohyphae growth was observed within the microscope, the D in the chemostat was increased to 0.12 h^-1 for five residence times, trying to eliminate pseudohyphae growth. However, the pseudohyphae subpopulation only increased over process time, which is in accordance with the literature (Mösch, 2002). Mösch et al. attributed this to the advantages of pseudohyphal cells, such as a higher cell surface area, which allows for better carbon source assimilation, triggering pseudohyphae growth in budding yeast cells (Mösch, 2002). On top of pseudohyphae formation potentially hindering the secretion of AnoUPO, the derepressible promoter (P_DF) was potentially repressed already at D = 0.12 h^-1. Combining both effects, it is no surprise that AnoUPO productivity decreased at D = 0.12 h^-1. Similar behavior was observed previously, indicating that pseudohyphae growth is a heritable phenotype (De et al., 2020), as upshifts in D did not eliminate pseudohyphae growth and it interfered with the secretion of the target protein (Puxbaum et al., 2016). Visual determination of pseudohyphae was taken as an indicator for process termination for ongoing experiments. Since pseudohyphae growth also occurred at D of 0.08 h^-1, a decelerostat was performed to screen different dilution rates on the effect of pseudohyphae growth and enzyme expression. An accelerostat was not exercised as we obtained initial pseudohyphae formation to be irreversible.

In the decelerostat experiment, a gradual decrease of D was exercised (Table 1). No enzyme expression was observed at steady-states operated at D of 0.14 h^-1 and 0.12 h^-1, which can be attributed to promoter repression at screened growth rates. This would also be in accordance with the previously performed chemostat cultivation. When the D was decreased to 0.11 h^-1, the promoter was derepressed, and the specific enzyme activity after a timeframe of five residence times was found at 0.43 ± 0.02 U/g biomass (Table 1). With a further decrease of D to 0.09 h^-1, the specific enzyme activity increased by 32.5% (0.57 ± 0.01 U/g biomass, Table 1). However, the decelerostat was stopped after the steady state was achieved at D of 0.09 h^-1, since pseudohyphae growth was detected, which was expected to decrease productivity over time (Puxbaum et al., 2016). Even though D of 0.09 h^-1 showed the highest productivity in the decelerostat, the aim was to avoid pseudohyphae growth for follow-up experiments. Furthermore, initial results revealed that pseudohyphae formation triggers time-instable product formation. Therefore, a D of 0.11 h^-1 was selected to establish a continuous process for derepressed AnoUPO production.

Performing the experiment at a D of 0.11 h^-1, the AnoUPO levels in the chemostat reached similar values compared to the fed-batch cultivation during the first 24 h of the induction (Figure 6A). The productivity during the cultivation fluctuated, but samples taken in the first 100 h of induction showed no significant decrease in cell-specific AnoUPO production. After 119 h of induction, the specific enzymatic activity decreased promptly, and pseudohyphae cells could be observed via microscopy analysis (Figure 7B). To confirm pseudohyphae growth for this experiment, we evaluated the expression level of the gene reported to be responsible for pseudohyphae growth (FLO11) (De et al., 2020) using reverse transcribed qPCR (Table 2). The samples taken at the beginning of induction up until 96 h of induction showed results within the boundaries of the RT-qPCR, while the sample taken at 119 h of induction showed a five-fold upregulation of FLO11, indicating that pseudohyphal growth occurred within the given time frame (Table 2).

Evaluating pseudohyphae growth in the performed chemostat, we noticed that pseudohyphae appeared after 110 ± 10 h of induction in the cultivation with a D set to 0.11 h^-1. For chemostats operated at D of 0.02 h^-1 and 0.05 h^-1, pseudohyphae growth was detected after 115 h of induction. The chemostat operated at D of 0.08 h^-1 showed pseudohyphae growth after 112 h of induction. Still, the timeframe for pseudohyphae growth equaled for all of the cultivations at approximately 110 ± 10 h of induction. Hence, results indicate that pseudohyphae growth in chemostats might be time-dependent once µ is beyond a certain threshold. Further research with different strains, constructs, and target proteins is still required to confirm these findings. Still, the results of this study revealed pseudohyphae growth in derepressed chemostats to be dependent on µ and the induction time.

Despite the pseudohyphae formation at later time points, results showed that chemostat cultivations have a small operating window where they can be operated with derepression for at least 110 ± 10 h. The stable production phase, as well as the recombinant protein titer, could be potentially increased if a floculin-deficient strain is used to reduce pseudohyphae growth, as reported in a recent study (Rebnegger et al., 2024). Even though fed-batch cultivation outperformed chemostat cultivation for the production of the selected UPO, productivity in different cultivation modes might vary between different UPO production. The results of this study indicated a screening method to target stable operational levels for recombinant proteins. As shown for AnoUPO production in continuous cultivation, production could compete with fed-batch cultivations for 110 h of induction time despite pseudohyphae formation occurring in chemostats. Additionally, this study revealed that pseudohyphae phenotype formation is both µ- and time-dependent, aiding more knowledge on the topic of derepressed recombinant production in continuous cultivations with K. phaffii.