Open (non-sterile) cultivations of Debaryomyces hansenii for recombinant protein production combining industrial side-streams with high salt content

The halotolerant non-conventional yeast Debaryomyces hansenii can grow in media containing high concentrations of salt (up to 4 M), metabolize alternative carbon sources than glucose, such as lactose or glycerol, and withstand a wide range of temperatures and pH. These inherent capabilities allow this yeast to grow in harsh environments and use alternative feedstock than traditional commercial media. For example, D. hansenii could be a potential cell factory for revalorizing industrial salty by-products, using them as a substrate for producing new valuable bioproducts, boosting a circular economy. In this work, three different salty by-products derived from the dairy and biopharmaceutical industry have been tested as a possible feedstock for D. hansenii ’ s growth. The yeast was not only able to grow efficiently in all of them but also to produce a recombinant protein (Yellow Fluorescent Protein, used as a model) without altering its performance. Moreover, open cultivations at different laboratory scales (1.5 mL and 1 L) were performed under non-sterile conditions and without adding fresh water or any nutritional supplement to the cultivation, making the process cheaper and more sustainable.


Introduction
Traditionally, Saccharomyces cerevisiae has been used as a model organism to study heterologous expression in eukaryotic systems, being one of the preferred platforms for the industry to produce a wide range of bioproducts [1,2].However, it presents some drawbacks that make it unsuitable for all biotechnological purposes: it is Crabtree-positive; it cannot utilize alternative carbon sources like lactose, xylose, arabinose or glycerol; its metabolism is impaired at extreme temperatures or pH; and it is unable to tolerate high salt concentrations [3][4][5].To avoid extensive engineering to increase its tolerance towards most of the aforementioned, non-conventional yeasts are gaining interest as alternative cell factories for bioproduction, as many present inherent capabilities to withstand harsh environmental conditions [6][7][8].
D. hansenii is a halotolerant yeast isolated from salty environments [3,9] that tolerates salt concentrations of up to 4 M [10].Several studies have demonstrated that its metabolism is enhanced when salt is present in the media.For example, one study [11] concluded that Na + is not toxic for D. hansenii but improves its growth.The effect of NaCl on D. hansenii in extreme temperatures or acidic pH was later studied [12], concluding that Na + also confers protection to the cells under these conditions.Follow-up studies proved that sodium also protects the yeast against oxidative stress [13,14].A physiological analysis of D. hansenii in controlled cultivations (1 L bioreactors) growing at different concentrations of NaCl or KCl (0.5, 1, and 2 M) observed the highest growth rate when a concentration of 1 M NaCl was present in the medium, proposing a survival strategy for the yeast to prevail in high-salinity environments [15].From a biotechnological perspective, D. hansenii is also interesting for its ability to metabolize a broad spectrum of carbon sources and its tolerance to extreme temperatures and pH (5-35 • C and 3-10, respectively) [10,14,16].
All these characteristics make D. hansenii an interesting platform for the so-called green biotechnology transition [17].For example, some authors have suggested the possibility of using seawater as culture Abbreviations: WT, Wild Type; YFP, Yellow Fluorescent Protein; YNB, Yeast Nitrogen Base; YPD, Yeast Extract-Peptone-Dextrose; NTC, Neurseothricin; YAN, Yeast Assimilable Nitrogen; DLP, Delactosed Whey Permeate; SW, Salty Waste; SF, Salty Flowthrough; LSU, Light Scattered Units; OUR, Oxygen Uptake Rate; RQ, Respiratory Quotient; DO, Dissolved Oxygen.media (avoiding the use of pure water) [7,18], or growing D. hansenii using alternative substrates instead of commercial media, e.g., hemicellulosic hydrolysates [19].Salty by-products generated by industrial activities could also serve as feedstock for D. hansenii.We recently used D. hansenii to revalue dairy by-products to produce a reporter protein (YFP), using different side-effluents derived from cheese production: (i) Delactosed Whey Permeate (DLP), rich in lactose (120 g/L) [20], and (ii) a salty effluent containing 1.5 M NaCl (SW) [21].D. hansenii efficiently grew and produced YFP using both by-products while outcompeting other microorganisms present in the broth (derived from cheese production) by only pasteurizing the media, avoiding the traditional (and costly) sterilization step.However, both by-products needed to be supplemented with commercial nitrogen (YNB) due to the lack of a nitrogen source.
Our present study evaluates the use of another salty industrial waste derived from perfusion mammalian cultivations at Novo Nordisk A/S.Perfusion processes can reach and keep high cell densities for a large period of time, generating large quantities of clarified harvest with the same chemical composition as the culture in the bioreactor but containing no cells [22].The salt content of the clarified harvest is adjusted, filter sterilized and run through a chromatography column to capture the product, discarding the flowthrough generated.Although the nutrients present in this salty flowthrough (SF) could be potentially reused as a substrate for other processes, the presence of salt makes it unsuitable for most microorganisms but not for D. hansenii.The SF used in this study was obtained during the exponential growth phase of the perfusion process and had a high concentration of assimilable nitrogen but a low concentration of glucose (below 10 mM).To increase productivity, mixing it with dairy SW or DLP (rich in lactose and salt but lacking nitrogen) was analyzed to find the best balanced feedstock, leading to higher D. hansenii biomass levels.
In all the experiments performed in this work, D. hansenii was inoculated directly to the by-products (SF, SW, or DLP) without adding fresh water or any nutritional supplement, and without working in sterile conditions.The salt in the media was enough to prevent or slow the growth of most microorganisms but not D. hansenii, whose metabolism was enhanced.The production of YFP by D. hansenii was used for two purposes: (i) to demonstrate that the yeast can produce a recombinant protein from industrial salty by-products, and (ii) to specifically monitor the growth of D. hansenii among the other microorganisms present in the dairy by-products (DLP and SW), measuring the fluorescence instead of the optical density.

Strains and preculture conditions
Two strains of D. hansenii IBT22 were used for this study: the Wild Type (D. hansenii WT) and a transformant strain expressing the Yellow Fluorescent Protein (D. hansenii + YFP).The WT strain was obtained from the collection at the Technical University of Denmark.Yeast cells were streaked from glycerol stocks to YPD plates containing 1% yeast extract, 2% peptone, 2% agar, and 2% of D-(+)-glucose monohydrated (Sigma-Aldrich, Darmstadt, Germany) and incubated at 28 • C for 48 h.From YPD plates, liquid precultures containing 6.7 g/L of synthetic complete Yeast Nitrogen Base (YNB) medium (Sigma-Aldrich, Darmstadt, Germany) and 2% D-(+)-glucose were prepared using 500 mL baffled flasks (100 mL culture).The pH was adjusted to 6.0 with NaOH.
The precultures were incubated at 28 • C, 150 rpm for 24 h.All media (YPD and YNB) were autoclaved at 121 • C for 20 min.The glucose was autoclaved separately and added afterwards.
The transformant D. hansenii IBT22 strain expressing the YFP used in this study was constructed by our group as described in [21].Briefly, a linearized plasmid cassette containing the YFP gene (encoded by a TEF1 promoter and a synthetic terminator) and the resistance marker NTC (used for selection) was inserted by random integration into the yeast genome.The YFP was expressed intracellularly.The cultivation conditions for the transformant D. hansenii + YFP strain were the same as for the WT, but the medium for both the agar plates (YPD) and the precultures (YNB), was supplemented with 100 µg/mL of Nourseothricin (NTC) after sterilization.
S. cerevisiae BY4741 (ATCC, Rockville, US) was also included in this work.Cells were grown in YPD plates from a glycerol stock (18-20% final glycerol concentration) stored at − 80 • C. Complete YNB medium (6.7 g/L) plus 2% D-(+)-glucose was used for the liquid precultures, following the same conditions as for D. hansenii.

Industrial by-products used as feedstock
A salty effluent derived from a perfusion bioprocess with mammalian cells was provided by Novo Nordisk (Gentofte, Denmark) and used as a feedstock (SF), consisting of the flowthrough obtained after running the bioreactor's clarified harvest after salinity adjustment through a chromatography column.In addition, two different by-products from Arla Foods Ingredients (Viby, Denmark) were used as feedstock: Delactosed whey Permeate (DLP) and a salty waste effluent (SW) that contains 1.5 M NaCl.Both are derived from cheese production.The concentration of glucose, lactose, salt, assimilable nitrogen, and fat of the by-products are summarized in Table 1.
The total Yeast Assimilable Nitrogen (YAN) of the SW and SF was estimated using the Formol Titration Technique.Briefly, 10 mL of each previously diluted by-product was titrated with 0.05 N NaOH until neutralized (pH 8).Then, 2 mL of previously neutralized formaldehyde solution (37%) (Merck, Darmstadt, Germany) was added to the byproducts, which reacted with the YAN sources and released H + ions, lowering the pH.Finally, the samples were titrated again until pH 8 with 0.05 N NaOH using a burette.The final volume of NaOH added was used to calculate the YAN level using the following equation (Eq.1):  C for at least 24 h.A BioLector II (m2p Labs, Aachen, Germany) was used to perform microcultivation experiments (1.5 mL working volume) in biological triplicates using a 48-well FlowerPlate (MTP-48-B, Aachen, m2p Labs).Each well was filled with media and inoculated with the corresponding yeast (D. hansenii WT, D. hansenii + YFP, or S. cerevisiae) to an initial OD 600 0.1.
Three industrial by-products (SF, SW, DLP) were used as culture media for the microfermentations.They were used independently and mixed in different ratios.The different combinations used are summarized in Table 2 and Table 3.All media were supplemented with 20.4 g/ L of potassium phthalate monobasic ≥ 99.5% (Sigma-Aldrich, Darmstadt, Germany), used as a pH buffer.The pH was adjusted to 6.0 with NaOH.The SW and DLP were nonsterilized, and none of the by-products was manipulated under sterile conditions.The osmolarity of each media was measured in technical duplicates using the Osmomat 3000 equipment (Gonotec, Logan Utah, US).Biomass formation (Light Scattered Units, LSU) and fluorescence production (arbitrary units) were monitored throughout 75 h of cultivation using two filters, biomass (gain 1) and mVenus (gain 7), respectively.The temperature was kept at 28 • C, agitation speed set at 1000 rpm, humidity maintained at > 85%, and oxygen supply was 20.85%.The data obtained were analyzed using R version 4.2.1.
D. hansenii + YFP specific growth rate values (µ max ) were calculated based on the fluorescence profiles obtained for all the conditions tested.The µ max calculation is an average considering a simultaneous coassimilation of both sugars, as it is not possible to completely separate growth in glucose from the growth in lactose: even though it appears sequential, there is an overlapping period where co-consumption occurs.

Statistical analysis
Regression analysis performed with Microsoft Excel ® 2016 were used to calculate D. hansenii µ max values from the microcultivations.The coefficient of determination (r2) was used to determine the statistical significance of the fit, where a value above 95% was considered statistically significant.OriginPro (OriginaLab Corporation, version 2022, Northampton, MA, USA) was used to compare the µ max and the endpoint fluorescence mean values.One Way ANOVA Tukey test (p < 0.05 confidence) was performed, which allowed classifying the mean values into groups (asterisks).Those grouped under the same number of asterisks did not present any statistically significant difference.

Cultivations in bioreactors
Biological triplicates were performed in 1 L Biostat Qplus bioreactors (Sartorius Stedim Biotech, Goettingen, Germany) with a working volume of 0.5 L. The programmed settings were: volumetric flow rate (aeration) set at 1 vvm, temperature 28 • C, stirring 600 rpm, and the pH was controlled and maintained at 6.0 by automatic addition of 2 M NaOH/H 2 SO 4 .
The transformant D. hansenii strain expressing the YFP (D. hansenii + YFP) was precultured in YNB + NTC medium and incubated at 28 • C for at least 24 h.Then cells were washed twice with distilled water and inoculated at OD 600 = 1.Two different media were used for the cultivations: (i) SF and (ii) a mixture of SF and SW in a proportion of 1:1 v/v.The pH was adjusted to 6.0 with NaOH.SW was not sterilized before use and all media were manipulated under non-sterile conditions.The bioreactor's off-gas line was connected to a mass spectrometer (Prima PRO Process MS, ThermoFisher scientific, Loughborough, UK), which allowed monitoring the concentration of CO 2 during the cultivation.

Bioreactor sampling
Samples for OD, fluorescence, and HPLC were taken during the cultivations once CO 2 levels reached above 0.1% and until the stationary phase.To measure the OD, a UV-1800 spectrophotometer was used with a wavelength set at 600 nm.YFP fluorescence was measured using a

Table 2
Combinations of SF and SW tested during microcultivations in the BioLector II.Osmolarity, endpoint fluorescence units, µ max values, and glucose, lactose and salt concentration are specified in the table.All values represent the mean of three biological replicates with their corresponding standard deviation (SD).

Media
Glucose  Nevertheless, all the cultivations were inoculated with the same starting OD 600 (0.1) of D. hansenii, so the total fluorescence values provides sufficient information to determine D. hansenii's optimal growth.It is also worth mentioning that the nature of the side-streams used as feedstock in this study, result in a complex and with not a completely liquid consistence of the cultivation media, which slightly affects the accuracy of YFP detection compared to any other "clean" commercial media.
Finally, the concentration of glucose, lactose, and ethanol was monitored by HPLC (Agilent Technologies, Munich, Germany).The temperature of the column (Bio-Rad Aminex HPX-87 H) was kept at 60 • C, the flow rate set at 0.6 mL/h, injection volume 20 µL, and the mobile phase 5 mM H 2 SO 4 .The data obtained were compared with a standard solution containing glucose (20 g/L), lactose (20 g/L), and ethanol (20 g/L).

Oxygen Uptake Rate (OUR) and Respiratory Quotient (RQ) calculation
Dissolved oxygen (DO) sensors (OxyFerm-FDA 160, Hamilton, Reno, US) were used to measure the DO concentration during cultivations.From the data obtained, OUR [mmol/L/h] was calculated using the following equation (Eq.2): Where the aeration was 60 L/h; the working volume 0.5 L; the pressure 1 atm; the ideal gas constant 0.082 [L*atm/K*mol]; the gas temperature 301.15 K; the inlet O 2 20.96% and the outlet O 2 the one measured with the mass spectrometer at each specific time.
Respiratory quotient (RQ) was also calculated throughout the cultivation, which is the volume of carbon dioxide released over the volume of oxygen absorbed during respiration [23].

D. hansenii's performance using salty flowthrough (SF) as feedstock
The performance of D. hansenii on the SF by-product was first assessed in microcultivation experiments performed with the BioLector.S. cerevisiae was used as a control.The results showed that both D. hansenii strains, the WT and the transformant expressing YFP, were able to grow in this feedstock (Fig. 1) without adding any nutritional supplement or fresh water.Moreover, both D. hansenii strains (WT and YFP) performed equally well.The only difference was the fluorescence levels: whereas the WT strain did not produce fluorescence (as it did not express the YFP gene), the transformant strain (D. hansenii + YFP) produced YFP throughout the cultivation without altering its performance compared to the WT (Fig. 1).
As expected, the production of YFP was coupled to the biomass formation, as the promoter used (TEF1) is constitutive and only active during the exponential phase.Nevertheless, its activity decreases when the cells reach the stationary phase.Hence, the fluorescence can be used as an alternative to the OD 600 or LSU to monitor the yeast growth, as both measurements follow the same tendency.Besides, almost no delay was observed between the increase in biomass and the production of YFP (Fig. 1).In contrast, S. cerevisiae could not grow on the SF, as no increment of biomass was detected throughout the cultivation (Fig. 1).

YFP production using the Salty Flowthrough (SF) as feedstock in 1 L bioreactors
The growth of the recombinant D. hansenii (D. hansenii + YFP) in SF was also studied in bioreactors.This approach was important for several reasons: (i) to ensure reproducibility of the process at higher scales; (ii) to provide a controlled environment during the cultivation, such as controlled pH, temperature, or airflow, which can prevent oxygen limitations; and (iii) to monitor other parameters such as carbon and O 2 consumption, and CO 2 production, allowing a deeper insight into the bioprocess.
As shown in Fig. 2, D. hansenii grew and produced YFP throughout the cultivation, similarly as observed in the micro-fermentations (Fig. 1).After around 5 h of lag phase, the yeast started to consume the glucose from SF, growing and producing fluorescence and CO 2 until the sugar was depleted (after 17.5 h).Around 5 h after, the yeast stopped its growth and, consequently, YFP production.As aforementioned, YFP production was linked to the growth of the yeast, also in 1 L scale, where the fluorescence profile followed the same trend as the biomass profile (Fig. 2).
Although D. hansenii can proliferate on the SF, the concentration of glucose in this by-product is very low (below 10 mM) to achieve an optimal productivity if this side-stream is to be used for an industrial production scale.For that reason, adding an additional carbon source to the SF would be necessary to generate higher biomass and higher production of recombinant protein.This additional carbon source could be ideally obtained from another industrial side-stream.As mentioned in the introduction, a previous work performed by our group demonstrated that D. hansenii is able to grow using salt-rich dairy by-products (SW and DLP) as feedstock containing high lactose concentrations [21].Hence, in order to avoid the use of commercial sugar, a better option of supplementing the SF with these dairy by-products was evaluated.Table 1 summarizes the concentration of sugars, salts, and nitrogen in both side-streams.

Assessing D. hansenii's growth and YFP production using different combinations of SF and SW in micro-bioreactors
The SF was first mixed at different ratios with the SW.The different combinations tested are summarized in Table 2.Although the SF comes from a filtration step, thus free of cells, the SW contains some autochthonous microorganisms present in the milk and derived from cheese production.This represents an extra challenge when working with mixed cultures: the methods traditionally used to measure the biomass (e.g., OD 600 or Dry Weight) are not accurate for exclusively monitoring the growth of D. hansenii, as they quantify the increment of the whole biomass of the culture, including the other microorganisms.For that reason, YFP fluorescence, produced only by the transformant D. hansenii strain, has been used as an alternative method to monitor the growth of the yeast while serving as proof of concept of recombinant protein production.
Fig. 3B shows that the fluorescence produced in control conditions (media not inoculated with D. hansenii) was basically zero in all the combinations tested.This means that none of the microorganisms from the SW emitted fluorescence in the same range as the YFP.Hence, the increment in the fluorescence levels observed in all the media combinations in Fig. 3A was only due to the production of YFP by the Fig. 2. D. hansenii + YFP growth, glucose consumption, CO 2 , and YFP production throughout cultivations performed in 1 L bioreactors using the SF as the only feedstock.The YFP production is expressed in fluorescence intensity (arbitrary units).Each profile shows the average of three biological replicates with their corresponding standard deviation represented in vertical lines (for the glucose, biomass, and fluorescence profiles) and shadow (for the CO 2 profile).SW contains lactose (15-22 g/L) and a high salt concentration (1.5 M).Therefore, when the SF is supplemented with SW, the concentration of the total carbon but also the salinity increases compared to when SF is used alone.Table 2 summarizes the osmolarity, glucose, lactose and salt concentration, the D. hansenii µ max values, and the endpoint fluorescence reached at the end of the cultivation for all the media combinations.
The lowest µ max values were observed when the yeast grew in both by-products independently (either SF or SW), with no statistically significant difference detected (Fig. 4A).However, the lag phase observed when the yeast grew in the SW was longer than in the SF (15 h vs. 7 h, respectively), as the concentration of salt in the SW is more than 1 M higher than in the SF, as well as the osmolarity, which is more than double in the SW (Fig. 3A).Additionally, the SW is more complex than the SF in its composition, as it contains fats, proteins and other microorganisms.
Generally, when both by-products were combined, the calculated µ max values were higher (Fig. 4A).When the SF was mixed with 10% or 25% (v/v) of SW (Table 2, media c and d), D. hansenii performed better than when both by-products were used separately, resulting in shorter lag phases and higher µ max values (Figs.3A and 4A).Besides, when the proportion of SW increased up to 85% of the total volume (Table 2, media e, f, and g), D. hansenii µ max values were even higher, reaching the maximum average µ max value when both by-products were mixed at 50% (v/v) (Table 2, media e).However, longer lag phases were also observed when the proportion of SW increased from 25% of the total volume (Table 2, media d), being more prolonged as the ratio of SW was higher (Fig. 3A).
Therefore, D. hansenii's metabolism was enhanced when more salt and carbon were present in the media (shorter lag phases and higher µ max values), reaching the optimal performance when the SF was mixed with SW in a ratio of 1:1 (v/v) (Table 2, media e, Figs.3A and 4A).In this case, the concentration of total sugar (glucose plus lactose) and salt were 12.27 g/L and almost 1 M, respectively.From this point, higher proportions of SW (Table 2, media f and g) resulted in similar µ max values (no statistically significant difference observed with respect to media e, Fig. 4A) but longer lag phases (Fig. 3A), as the osmolarity of the media, as well as the concentration of salt, were higher (more than 1 M, the optimal for D. hansenii's growth).
Similar results were observed when the fluorescence produced at the end of the cultivations was measured, which can be related to the final biomass achieved by D. hansenii.The lowest endpoint fluorescence was detected when the yeast used only the SF as feedstock (Table 2, media a) since there was a low concentration of glucose in the media (Fig. 4B).When the SF was supplemented with SW, which contains lactose, the concentration of the total carbon in the media increased.Hence, as the proportion of SW was increased with respect to the total volume, more fluorescence was detected at the end of the cultivation, as more sugar was metabolized to produce biomass and YFP (Fig. 4B).The maximum endpoint fluorescence was detected when both by-products were mixed 50% (v/v) (Table 2, media e).From this point, when the proportion of SW increased (Table 2, media f and g), even though there was more concentration of sugar in the media, the osmolarity and the concentration of salt were also higher (over 1 M), resulting in a reduced growth pace and less fluorescence detected (Fig. 4B).All in all, mixing the SF and the SW in a ratio of 1:1 (v/v) (Table 2, media e) was the best combination in terms of D. hansenii's specific growth rate (Fig. 4A) and YFP production (Fig. 4B).

Fig. 4. D. hansenii + YFP A) µ max mean values and B) endpoint fluorescence mean values reached at the end of microcultivation experiments performed with
BioLector II using different combinations of SF and SW as feedstock.µ max values were calculated from D. hansenii fluorescence profiles represented in Fig. 3A for each media tested.The endpoint fluorescence represents the last fluorescence intensity value recorded by the BioLector at the end of each cultivation.For both graphics, each dot represents the µ max (A) or the endpoint fluorescence (B) mean value of three biological replicates with their corresponding standard deviation represented in a vertical line (95% confidence interval).For each graphic, the µ max and endpoint fluorescence values are classified in groups of statistically significant difference according to the one-way ANOVA Tukey test (p < 0.05 confidence).Those values grouped under the same number of asterisks did not present any statistically significant difference, whereas those not grouped together are statistically different.

Characterization of D. hansenii's performance and YFP production using a combination of SF and SW (1:1 v/v) as feedstock
The growth and YFP production of D. hansenii using a combination of SF and SW (1:1 v/v) as culture media was further studied in bioreactors.A control condition was also included (non-inoculated tanks).Fig. 5A shows the OUR during the cultivation.As observed, D. hansenii grew faster and consumed more oxygen than the microorganisms from the SW (no yeast inoculated).The yeast started to grow after around 5 h of lag phase, whereas around 20 h were needed for the control to show some microbial growth (Fig. 5A).Therefore, the presence of other microorganisms does not interfere with the performance of D. hansenii growing in the selected combination of side-streams.
The dissolved oxygen (pO 2 ) in the media was also monitored throughout the fermentation.This is the result of a balance between the oxygen rate consumed by the cells and the oxygen transfer rate from the gas to the liquid [24].Fig. 5B shows that, even though D. hansenii's metabolic activity was enhanced by the presence of salt, resulting in higher oxygen demand, there were no oxygen limitations, as the pO 2 values were above 0 during the cultivation.Even during the short period where pO 2 levels got close to 0, the calculated RQ values were always close to 1 but not above, which means that the cells did not suffer oxygen deprivation (Fig. 5C).The control tank was not oxygen limited either, and as expected, its oxygen consumption was significantly lower (Fig. 5B).
Glucose and lactose consumption and biomass, YFP, and CO 2 production were also monitored during the cultivation (Fig. 6).From Fig. 6A, it can be observed that D. hansenii consumed the sugars sequentially: first the glucose and then the lactose (although coconsumption was observed when glucose levels were getting lower) until both sugars were depleted while producing biomass, fluorescence, and CO 2 .On the other hand, in the control reactors (Fig. 6B), glucose was consumed, but almost no lactose was utilized.Moreover, the biomass formation achieved at the end of the cultivation was significantly low compared to the one reached by D. hansenii, and no fluorescence was detected at the range of the YFP (Fig. 6B).These results indicate that the microorganisms from the SW had difficulties growing in the mix of side-streams, not representing a competition for D. hansenii, whose metabolism was faster.
It is worth mentioning that in both conditions, a decrease in OD 600 was observed during the first hours of fermentation due to the presence of fat in the SW.At the beginning of the cultivation, the fat was homogenized in the medium, making it thick.However, after some hours of agitation, most of the fat was "trapped" in the bioreactor's impellers, Fig. 5. A) Oxygen Uptake Rate (OUR), B) Dissolved Oxygen (pO 2 ), and C) Respiratory Quotient (RQ) profiles obtained throughout cultivations performed with 1 L bioreactors using SF and SW as feedstock (1:1 v/v).Both D. hansenii + YFP strain and the control are represented in the three graphics.The control corresponds to the autochthonous microorganisms already present in the SW (cultivations not inoculated with D. hansenii).Each profile shows the average of three biological triplicates with their standard deviation represented in shadow.leaving a clearer media (Suppl.Fig. 1).This effect was reflected in the OD 600 measurements, which were higher at the beginning of the fermentation due to the interferences caused by the presence of fat.Then, once the fat was "removed" from the media, the OD 600 values decreased, increasing again due to biomass formation.

Assessing the growth of D. hansenii and YFP production in different combinations of SF and DLP
In order to study other possible industrial by-products rich in sugars that could complement the lack of carbon source of the SF, the addition of Delactosed Permeate (DLP) was analyzed.DLP contains other microorganisms from cheese production but was not sterilized before using it as feedstock.YFP fluorescence was used again as an alternative to the OD 600 measurement to precisely monitor the growth of this yeast (Fig. 7B).Four different media combinations of both by-products (SF and DLP) were studied (Table 3), and the emission of fluorescence was monitored in inoculated (Fig. 7A) and in the controls without yeast (Fig. 7B).From the fluorescence profiles, µ max values were calculated (Table 3).
As depicted in Fig. 8A, the lower µ max was detected when D. hansenii grew in only DLP (Table 3, media a).However, the µ max values increased when both by-products were mixed (Table 3, media b, c, d, e).Once again, these results indicate that a combination of both by-products (SF and DLP in this case) boosted the growth of the yeast compared to when the by-products are used separately.
The higher µ max value was detected when the SF was mixed with the DLP at the lowest proportion rate, 25% of the total volume (Table 3, media b and Fig. 8A), which was also the combination with the lowest osmolarity and highest concentration of salt (0.37 M).From the rest of the combinations tested, we observed that the µ max decreased as the percentage of added DLP increased (Table 3, media c, d, e).The same result was observed when D. hansenii's lag phase was analyzed (Fig. 7A).As the percentage of added DLP increased, longer lag phases were observed, reaching 20 h in the worst case (Table 3, media e) compared to the best one (Table 3, media b).The opposite was observed for the fluorescence detected at the end of the cultivations (Table 3 and Fig. 8B).A higher percentage of added DLP with respect to the total volume resulted in higher fluorescence produced (Table 3, media c, d, and e), as more sugar was available.
Finally, the lowest fluorescence was detected when D. hansenii grew on the DLP alone (Table 3, media a).Even though this media is the one with a higher lactose concentration, it is also the most complex since it was not diluted with SF.Therefore, higher osmolarity, less nitrogen, and more lactose and microorganisms are present, which made D. hansenii to have more difficulties to grow.

Discussion
The results obtained in this study demonstrate that D. hansenii can grow and produce a recombinant protein using an alternative feedstock: a salty flowthrough (SF) generated by the biopharmaceutical industry.This serves as further evidence of D. hansenii's capacity not only to withstand high salt concentrations but to thrive in a complex and high osmotic environment, which may contain high concentrations of sugars, secondary metabolites, debris, salts, growth inhibitors, proteins, etc.In contrast, S. cerevisiae could not grow in the SF waste, even though it can metabolize glucose and survive at low salt concentrations [3,25].
The SF used in this study was obtained during the exponential growth phase of a perfusion process with mammalian cells.However, the glucose concentration was too low to achieve an optimal productivity (below 10 mM).In order to increase the carbon concentration, supplementation with another industrial by-product rich in sugars was evaluated in this work.This approach allows the simultaneous revalorization of two different waste streams while avoiding using fresh water, making the process more cost-effective.
Our previous work demonstrated that D. hansenii can grow in two different by-products from the dairy industry (DLP and SW) [21].Both contain lactose, and the SW has a higher salt concentration (1.5 M).Hence, supplementing the SF with either SW or DLP in different ratios was evaluated.Higher growth rates were observed when the SF was combined with either SW or DLP, compared to when the yeast grew in the by-products independently.This suggests that the three side-streams complement each other well, as combining them results in a final media composition that is more favorable for D. hansenii's growth.In our previous work [21], the media had to be supplemented with commercial YNB due to the lack of a nitrogen source in both (DLP and SW).The need of an additional nitrogen source to supplement DLP as feedstock for microbial bioproduction has already been reported in previous studies [26,27].In another study [28], the authors used the bacteria Corynebacterium glutamicum to produce ethanol from DLP, but they additionaly had to supplement the media with NH 4 + , Mn 2 + , Fe 2 + , and trace minerals to achieve optimal biomass levels.However, in our case, when the DLP and SW are mixed with the SF that contains nitrogen, neither supplementation with YNB nor any other nutrient is necessary, as the SF provides the missing elements.At the same time, when the SF is mixed with DLP or SW, the concentration of total carbon source (glucose plus lactose) in the media increases, enhancing yeast growth.Besides, adding SW also increases the salt concentration of the final broth, reaching values closer to 1 M, which is the optimal for the growth of D. hansenii [15].In summary, when the SF was mixed with SW, higher µ max values were observed compared to when it was combined with DLP, regardless of the ratio.This might be due to the high osmotic pressure and complexity of the DLP, which contains more lactose, proteins, and microorganisms than the SW, hindering the growth of the yeast.
The highest µ max and fluorescence were obtained when the SF was mixed with SW in a ratio of 1:1 (v/v).In this case, the concentration of total sugar was 12.27 g/L, and the concentration of salt was almost 1 M. Higher proportions of SW resulted in longer lag phases, which confirms what was previously observed in the literature [15]: when the concentration of salt is above 1 M, D. hansenii's metabolism is slower but not inhibited.On the other hand, when the SF was mixed with DLP, higher concentrations of carbon were present in the media compared to when mixed with SW, resulting in higher biomass and YFP production at the end of the cultivation.However, longer lag phases (more than 20 h in the worst case) and lower µ max values were detected, which were more significant as the proportion of DLP increased.Several hypotheses could explain these results.For example, the C/N ratio may not be optimal for D. hansenii's growth when the proportion of DLP is too high, as it can be an excess of C and a lack of N in the media [29].Moreover, even though D. hansenii is an osmotolerant yeast, the high osmotic pressure of the DLP can also slow down its growth [30,31].Alternatively, given the high complexity of DLP, some metabolites might be present in the by-product that require prior adaptation of the yeast's metabolism to grow optimally.All these reasons could explain the long lag phases observed.Further research is, therefore, necessary to determine the optimal ratio of DLP and SF to enhance D. hansenii's growth.
Considering a balance between the µ max values, lag phases, and fluorescence production, mixing the SF with SW in a ratio of 1:1 (v/v) was selected as the best option for D. hansenii's growth and YFP production.For that reason, the growth of the yeast in this media was upscaled to 1 L bioreactors, where it is possible to have controlled conditions (airflow, temperature, pH, stirring) and more parameters can be analyzed.At this scale, a high oxygen demand was observed, which confirms the stimulated metabolic activity of the yeast when salt is present in the media [12,15,32].
Although some authors have previously reported that D. hansenii produces small amounts of ethanol in aerobic cultivations [33,34], no ethanol production was detected in the present experiments which supports the Crabtree negative behavior of D. hansenii [15,35].
Several studies have suggested that using media with a high salt concentration not only can reduce the use of freshwater but also make the process more affordable [7,18,36].In these conditions, the traditional media sterilization step is not required since the risk of contamination in this high-salinity environment is very low.Avoiding media sterilization can make the process more sustainable since it is one of the most energy-intensive and costly step in a production plant [37].

Conclusions
The biological revalorization of industrial salty by-products reduces the cost and the environmental impact of disposing of such effluents, but also boosts a circular economy by creating new and valuable products from industrial side-streams.This study confirms the potential of D. hansenii as the superior microbial cell factory to revalue wastestreams rich in salts.It has demonstrated that the yeast is not only capable of growing properly in this environment without additional requirements (e.g., the addition of fresh water), but it is also able to produce a model recombinant protein (YFP) from this feedstock.Moreover, it can proliferate faster than other microorganisms present in these by-products, thus avoiding the need to work under sterile conditions and making the process even more sustainable.
Considering future developments, a stronger effort must be dedicated for advancing towards production of secreted recombinant proteins in this yeast, and also to assess the impact on high osmolarity/ salinity on the secretion process.While this work demonstrates that recombinant production of intracellular proteins using saline feedstocks is feasible, it is still far from being able to successfully produce secreted proteins due to the lack of fundamental information (e.g. the existence of secretion signals that can efficiently work in Debaryomyces).Current research efforts are very much focused on this aspect, and it is strongly believed that sooner than later this could become another interesting story per se.

Declaration of Competing Interest
The authors declare that there are no competing interests.

Fig. 1 .
Fig. 1.D. hansenii (WT and D.hansenii + YFP) and S. cerevisiae growth and YFP production throughout microcultivations performed in BioLector II using the SF as feedstock.For each strain, the biomass is represented in a continuous line and expressed in Light Scattered Units (LSU), whereas the YFP production is represented in a discontinuous line and expressed in fluorescence intensity (arbitrary units).Each profile shows the average of three biological replicates with their corresponding standard deviation represented in shadow.

Fig. 3 .
Fig. 3. YFP production profiles represented in fluorescence intensity (arbitrary units) obtained from A) the transformant D. hansenii expressing the YFP and B) the autochthonous microorganisms from the SW (control) throughout microcultivations performed in BioLector II.Different combinations of SF and SW were used as feedstock for the cultivations.Each profile (for D. hansenii and control) shows the average of three independent replicates with their corresponding standard deviation represented in shadow.

Fig. 6 .
Fig. 6.Biomass, YFP, and CO 2 production, and glucose and lactose consumption profiles by A) D. hansenii + YFP and B) control throughout cultivations performed in 1 L bioreactors using SF and SW as feedstock (1:1 v/v).The control represents the autochthonous microorganisms already present in the SW (cultivations not inoculated with D. hansenii).The YFP production is represented in fluorescence intensity (arbitrary units).Each profile shows the average of three biological replicates with their corresponding standard deviation represented in vertical lines (for the glucose, lactose, biomass, and fluorescence profiles) and shadow (for the CO 2 profile).

Fig. 7 .
Fig. 7. YFP production profiles represented in fluorescence intensity (arbitrary units) obtained from A) the transformant D. hansenii expressing the YFP and B) the control, corresponding to the autochthonous microorganisms already present in the DLP (cultivations not inoculated with D. hansenii) throughout microcultivations performed in BioLector II.Different combinations of SF and DLP were used as feedstock for the cultivations.Each profile (for D. hansenii and control) shows the average of three biological replicates with their corresponding standard deviation represented in shadow.

Fig. 8 .
Fig. 8. D. hansenii + YFP A) µ max mean values and B) endpoint fluorescence mean values reached at the end of microcultivation experiments performed with BioLector II using different combinations of SF and DLP as feedstock.µ max values were calculated from D. hansenii fluorescence profiles represented in Fig. 7A for each media tested.The endpoint fluorescence represents the last fluorescence intensity value recorded by the BioLector at the end of the cultivation.For both graphics, each dot represents the µ max (A) or the endpoint fluorescence (B) mean value of three biological replicates with their corresponding standard deviation represented in a vertical line (95% confidence interval).For each graphic, the µ max and endpoint fluorescence values are classified in groups of statistically significant difference according to the one-way ANOVA Tukey test (p < 0.05 confidence).Those values grouped under the same number of asterisks did not present any statistically significant difference, whereas those not grouped together are statistically different.

Table 1
Salt, sugar, fat, and nitrogen content of the different by-products used in this study.

Table 3
Combinations of SF and DLP tested during microcultivations performed in BioLector II.Osmolarity, endpoint fluorescence units, µ max values, and glucose, lactose and salt concentration are specified in the table.All values represent the mean of three biological replicates with their corresponding standard deviation (SD).
CLARIOstar Plus microplate reader (BMG Labtech, Ortenberg, Germany), with excitation and emission wavelengths of 510 and 545 nm, respectively.Black Nunc 96-well microplates (Nunc 167008, Sigma-Aldrich, Darmstadt, Germany) were used for the measurement, which were shaken at 100 rpm for 15 s before each read.The measured fluorescence values correspond to total fluorescence, not normalized by the amount of cell per sample because other microorganisms (who do not produce YFP) from the dairy side-streams are also present in the media.
* non-detected metabolite by HPLC analysis† arbitrary units