The synergetic effect from the combination of different adsorption resins in batch and semi‐continuous cultivations of S. Cerevisiae cell factories to produce acetylated Taxanes precursors of the anticancer drug Taxol

Abstract In situ product recovery is an efficient way to intensify bioprocesses as it can perform adsorption of the desired natural products in the cultivation. However, it is common to use only one adsorbent (liquid or solid) to perform the product recovery. For this study, the use of an in situ product recovery method with three combined commercial resins (HP‐20, XAD7HP, and HP‐2MG) with different chemical properties was performed. A new yeast strain of Saccharomyces cerevisiae was engineered using CRISPR Cas9 (strain EJ2) to deliver heterologous expression of oxygenated acetylated taxanes that are precursors of the anticancer drug Taxol ® (paclitaxel). Microscale cultivations using a definitive screening design (DSD) were set to get the best resin combinations and concentrations to retrieve high taxane titers. Once the best resin treatment was selected by the DSD, semi‐continuous cultivation in high throughput microscale was performed to increase the total taxanes yield up to 783 ± 33 mg/L. The best T5α‐yl Acetate yield obtained was up to 95 ± 4 mg/L, the highest titer of this compound ever reported by a heterologous expression. It was also observed that by using a combination of the resins in the cultivation, 8 additional uncharacterized taxanes were found in the gas chromatograms compared to the dodecane overlay method. Lastly, the cell‐waste reactive oxygen species concentrations from the yeast were 1.5‐fold lower in the resin's treatment compared to the control with no adsorbent aid. The possible future implications of this method could be critical for bioprocess intensification, allowing the transition to a semi‐continuous flow bioprocess. Further, this new methodology broadens the use of different organisms for natural product synthesis/discovery benefiting from clear bioprocess intensification advantages.


| INTRODUCTION
Microbial cell factories are becoming an important route for the sustainable production of different natural products and chemicals of different commercial (therapeutics, food additives, cosmetics, or biofuels) or academic relevance Walls, Otoupal, et al., 2022;Wong et al., , 2018. To date, a majority of studies involving an integrated process of product synthesis and recovery have not been performed at a lab scale, which has contributed to the slow translation of microbial cell factories to fulfill industry demands Nowrouzi et al., 2023;Walls, Martinez & del Rio Chanona, & Rios-Solis, 2022;. Different bioprocess strategies have been used to enhance the recovery of desired products, for instance, using two or more phase partitions, foam phase recovery or in situ product recovery (ISPR) (Zainuddin et al., 2021). ISPR is a common integration bioprocess strategy that incorporates a step that sequesters microbe products released to the media, making the production and extraction a single-step method (De Brabander et al., 2021). Many studies have developed an ISPR using liquid solvents, such as dodecane (Ajikumar et al., 2010), polypropylene glycol 4000 (Nicolaï et al., 2021), monoterpenes (Brennan et al., 2012) or ionic liquids (Zhong et al., 2012). The use of a liquid extractant is very practical if the solvent has an affinity to the molecule of interest (Najmi et al., 2018). However, this method has some limitations, such as the liquid extractant must be nontoxic to microbial cell factories, and the product recovery from these solvents can be problematic as they can be hard to evaporate such as dodecane, while not being environmentally friendly (Santoyo-Garcia et al., 2022;Walls, Martinez & del Rio Chanona, & Rios-Solis, 2022).
In addition, solid extractants for ISPR have also been widely used, where common materials used to achieve the recovery of products involve: activated charcoal, silica gel and adsorbent porous resins among others (Abdullah Thaidi et al., 2023;Lee et al., 2003;Othman et al., 2018).
An important advantage of using a solid extractant is the easiness to be separated from the media by centrifugation or filtration steps as well as the easy recovery of the desired product from the solid phase (Aguilar et al., 2019). In other studies, these solid adsorbers have been used in external column configurations to perform the ISPR (Galindo-Rodriguez et al., 2023). Various studies have assessed the efficacy of single resin beads via in situ cultivation. Phillips et al. (2013) showcased the use of several resins (XAD-7, XAD-8, XAD-1180, HP-20, and HP-21 among others) in situ with different microorganisms (Streptomyces spectabilis, Penicillium chrysogenum, Salinispora tropica, Sorangium cellulosum, Actinoplanes teicomyceticus among others) to get an ample range of natural products such as anticancer drugs (Epothilones, Leinamycin, Dynemicin A among others), antibiotics (Streptovaricin, Pristinamycin, Brotrallin, etc.), antioxidants (anthocyanin, stilbenes, among others) and many more (Luo et al., 2014;Phillips et al., 2013;Shin & Kim, 2016). One of the most studied resins for ISPR cultivation is HP-20, which was shown to increase the yields of anticancer drugs such as thailandepsin from less than 10 mg/L in Burkholderia thailandensis to 237 mg/L (Liu et al., 2012).
Another advantage of ISPR, and more specifically the use of solid extractants, is the potential capacity of cell-waste removal in addition to product recovery, which could be crucial to optimize the bioprocesses (Ojovan et al., 2011). The most common cell-waste generators are ethanol, ketones, ammonia, and lactate among others (De Brabander et al., 2021), as they have been linked to poor cell growth or product inhibition if their concentration surpasses any established tolerance level (Wang et al., 2011). Another undesired effect of cell waste is the unintended activation of stress genes that affect cell homeostasis, oxidative stress and reactive oxygen species (ROS) (Rodacka, 2016). As previously mentioned, ISPR could alleviate the waste load produced during the cultivation by selectively adsorbing it into the resin beads surface or the solvent extractant.
While there are clear advantages to this strategy, the value of using a combination of solid extractants has not been sufficiently investigated. By contrast, the combination of solvents has been more commonly used in the extraction step by coalescing solvents to maximize the recovery of the targeted molecule (González-Peñas et al., 2014;Park et al., 2007). In previous studies, the blend of solid extractants had been used primarily for the removal of hard-chemical industrial waste from water or for capturing CO 2 , therefore, this approach has been more studied in the environmental field rather than for industrial bioprocess intensification (Brennan et al., 2012;Marin & Stanculescu, 2022;Mortezaeikia et al., 2015).
Finally, another key advantage of using solid extractive methods is the natural transition towards continuous or semi-continuous fermentation systems with ISPR. Maintaining the solid extractants during semi-continuous cultivation, the evaporation of the extractant could be minimized, which has been reported as a common drawback in the dodecane overlay method (Nowrouzi et al., 2020Walls, Malcı, Nowrouzi, Li, d'Espaux, et al., 2020. This study, therefore, set out to assess different resins with polar, non-polar and aliphatic adsorption ranges for ISPR to enhance the cell-waste and product removal. As few studies are focusing on the optimization of combining solid extractants with different polarities in cultivations using microorganisms, this work aims to follow a design of experiment (DoE) approach to test the effect of three commercial adsorbent resins simultaneously on the total taxanes recovery, further allowing the adsorption for characterized or uncharacterized secondary metabolites. DoE is a well-known statistical tool applied in scientific experiments for different purposes, either for the identification of the most important factor or variable that affects a determined response, to a process optimization (Fisher, 1925). As previously mentioned, because the combination of different macro-porous resins in an ISPR cultivation has been poorly studied, a DoE capable of screening the best resins configuration will be used. As there are a vast number of DoE with different applications, it is important to frame the main aim wanted from the DoE, for this study is screening the best resin configuration (concentrations of the resins acting simultaneously in the ISPR cultivation) to recover the highest taxane titers. Recently, a DoE called definitive screening design (DSD) was developed by Jones and Nachtsheim, this DoE is capable of estimating main effects without confounding with second-order effects in just 2N + 1 runs for N factors (Jones & Nachtsheim, 2011). For its practicality, potent screening scope and possible optimization with a considerably little SANTOYO-GARCIA ET AL. | 2161 number of runs, DSD was selected in this study to find the best resin configuration.
Further, this study seeks to maximize total taxanes titers in a high throughput system by using semi-continuous cultivation on a microplate scale (Wilk et al., 2019). Three resin beads were used simultaneously for engineered S. cerevisiae fermentations with ISPR, these include Diaion © HP-2MG (polar), Diaion © HP-20 (nonpolar), and Amberlite © XAD7HP (aliphatic) in a range of concentrations and combinations to find the effect on cell growth and total taxane recovery. HP-2MG is an adsorbent resin that is made mainly of methacrylic ester copolymer giving it a higher polarity (Unsal et al., 2012). HP-20 resin beads are conformed by polystyrene and divinylbenzene with no oxygen in their structure making them nonpolar (Aguilar et al., 2019). Finally, XAD7HP resin is primarily made of acrylic ester giving it a medium polarity between the previous resins (Marin & Stanculescu, 2022). A summary of the physical-chemical properties of the resins as well as the Natural products recovered by these resins could be seen in Table 1. A schematic description of the reaction pathway and methodology followed in this study to perform the ISPR can be seen in Figure 1.
T A B L E 1 Physicochemical properties and previous NP recovered of the three resins used in the ISPR cultivations. F I G U R E 1 Different adsorbent porous resin beads were used simultaneously for an engineered S. cerevisiae fermentation to produce and recover taxanes. By combining the adsorption capacities of the three resin beads (Diaion © HP-20, Amberlite © XAD7HP, and Diaion © HP-2MG), it was intended to increment biomass accumulation and the total taxanes production by helping the cell remove cultivation wastes while improving product extraction.
This research was focused on constructing an engineered S. cerevisiae strain EJ2 by modifying the previously built strain LRS6  that was able to produce the oxygenated Taxadiene-5α-ol taxane. An extra copy of the gene of Taxadiene-5α-ol O-acetyltransferase was integrated into the genome using CRISPR-Cas9 to produce Taxaxadiene-5α-yl acetate ( Figure 1). Finally, after optimizing the bioprocess using DoE in microplate batch cultivations, the best concentration and combination of resin beads was selected for further semi-continuous microplate cultivations to optimize the methodology according to total taxanes and oxygenated acetylated taxanes recovered.

| Yeast strain and media
The S. cerevisiae strains used in this study were a CEN.PK2-1C-originated yeast strain (EUROSCARF), and a CEN.PK2-1C yeast strain previously engineered by  labeled as LRS6 with the following genotype: MATa, leu2-3, All cultivation reagents were sourced from Fisher Scientific at the highest available purity unless otherwise stated. The media used for the yeast cultivations was yeast extract-peptone (YP, yeast extract 1% (wt/vol), peptone 2% (wt/vol)) supplemented with 2% (wt/vol) galactose (YPG) or 2% (wt/vol) glucose (YPD) unless otherwise stated.

| Biomass and resin beads partition for taxane quantification
The partition of resins and biomass was made after the batch cultivations in ISPR treatments. Partition was performed using a 200 μL micropipette where the yellow tip was placed as close to the microplate bottom as possible. Once the tip was in direct contact with the plastic wall of the microplate, media and biomass removal was conducted. The space generated between the tip and microplate and the top of the micropipette was small enough to restrict the passing of any resin beads with a diameter bigger than the 400 μm of the resins used in this study. Each separated phase (biomass and resins) was then put into the extraction of taxanes protocol detailed in Section 2.4.

| Extraction of taxanes
Extraction of taxanes was performed after the cultivation process using only the solid phase (biomass + beads), which was recovered and dried under nitrogen gas at room temperature before adding acetone (99.8%, Thermo Fisher Scientific) at a 1:1 ratio of the cultivation volume and incubating the resulting mixture at 30°C and 350 rpm for 4 h. Following the extraction incubation, the mixture was centrifuged at 4500 rpm for 10 min and the organic phase was recovered for taxadiene analysis via Gas Chromatogram-Mass Spectrometry (GC-MS). Taxanes from control runs were extracted from the solid phase (biomass only) at the end of the cultivation using acetone at a 1:1 ratio. The resulting mixture was incubated at 30°C and 350 rpm for 4 h and sent for GC-MS analysis. in a 9-day total cultivation process. This procedure is further detailed in Figure S1 and Table S1 from the Supporting Information Material.
Sacrifice sampling was used in triplicates to determine the total taxanes adsorption kinetics. At the end of the cultivation, extraction was performed as described in Section 2.3 for taxanes analysis in the GC-MS.

| Analytical methods
Taxanes identification and quantification were achieved via GC-MS.
A 1 μL sample of each organic solvent extract was injected into a TRACE™ 1300 Gas Chromatograph (Thermo Fisher Scientific) coupled to an ISQ LT single quadrupole mass spectrometer (Thermo Fisher Scientific). Chromatographic separation was achieved using a Trace Gold TG-SQC gas chromatography column (Thermo Fisher Scientific) using a previously described method . To identify and quantify total taxane production by the S. cerevisiae strain EJ2, pure standards of  Table 2.
Statistical analyses were performed using GraphPad Prism 8.0 software. One-way analysis of variance (ANOVA) was used to determine whether combined resin bead treatments yielded a significant impact on biomass accumulation and cell-waste removal compared with controls that lacked any extraction aid. A Dunnett's multiple comparison test was used to compare each treatment to the control. One-way ANOVA was also used to determine whether the quantity of resins was significant in the total taxane titers in the semicontinuous runs. A Tukey's honest significant difference test was subsequently employed to compare the total resins concentrations of 3%, 6%, and 12% (wt/vol) of treatment 9 (

| RESULTS AND DISCUSSION
3.1 | DoEs to optimize the resin beads ratio and concentration based in the cell growth and total taxane production The in situ extraction material used in this study has been used separately in previous studies for product recovery enhancement (Halka & Wichmann, 2018;Lee et al., 2003;Qi et al., 2014). However, the use of these materials must be well studied as they can negatively affect cell growth and reduce the efficiency target molecule recovery However, beyond total taxane recovery, it can be seen that the total oxygenated ratio [oxygenated taxanes (mg/L)/nonoxygenated taxanes (mg/L)] in every single treatment was lower (around 0.9) compared with the results given in treatment 9 which was 1.2 ( Figure S10 from the Supporting Information Material and Figure 2b). This result is in agreement with previous studies focusing on the P450 for the synthesis of Taxadiene-5α-ol, highlighting that careful tuning is required to optimize the oxygenated ratio of taxanes Nowrouzi et al., 2022).
The change of the concentration in total resin beads also had an important effect on the in situ adsorption of total taxanes. For instance, treatments 1, 2, 7, and 13 had the same resin proportion; however, treatment 1 had a total resin concentration of 1.5% (wt/vol), which was the lowest total taxane titer from the 4 treatments with 57 ± 12 mg/L. Comparing this treatment with treatment 2, which had the highest resins concentration of 4.5% (wt/vol), resulted in 148 ± 17 mg/L of total taxanes (increment of 2.6-fold). These results support a previous study that found using HP-20 resins had a detrimental effect on taxadiene recovery for resin concentrations below 3% (wt/vol) (Santoyo-Garcia et al., 2022). However, not only was the total concentration of resins an important factor for total taxane titers but the proportion of the individual resins was also seen to affect the total taxane yields as well. Treatments 1, 5, 8, 11, and 12, which all possessed a high concentration of HP-20 [1%-1.5% (wt/vol)] resin and low HP-2MG concentration (0.5% wt/vol), showed the lowest taxane recovery. Treatment 11 resulted in the lowest total taxane titer at 39 ± 7 mg/L. On the contrary, the highest titer of total taxanes were found in treatments 2, 3, 4, and 9 where the main (a) (b) F I G U R E 2 S. Cerevisiae strain EJ2 growth and total taxane titers according to DSD microplate cultivations. (a) Cultivations were performed using 2 mL working volume in microplates, YPG and different sets of resin beads combinations which are detailed in Table 2. All cultivations were performed in microscale at 30°C and 350 rpm shaking using HP-20, XAD7HP, and HP-2MG as solid extractants. Blue bars represent OD 600 at the end of cultivation and green bars represent the total taxanes retrieved. (b) Breakdown of the total taxanes retrieved in all treatments after 72 h. of cultivation. Blue bars represent the oxygenated taxanes and green bars the nonoxygenated taxanes. Bar values represent mean ± standard deviation (n = 3). Statistical significance for biomass accumulation was assessed by Dunnett's test versus their corresponding controls. *p < 0.0332; **p < 0.0021; **p < 0.001 and n.s. = not significant p > 0.1234. DSD, definitive screening design.
Treatment 9 had the highest total taxanes titer with 190 ± 20 mg/L, which represented a 4.8-fold increment compared with treatment 11 and 7.9-fold increase compared to the control where no extraction aid was used. Therefore, using 1.5% (wt/vol) of total resins with equal resin proportion (treatment 1), resulted in the lowest taxane titer value.
The differences in the total taxane recovery could be also related to the different physicochemical properties that treatment 9 had, as this treatment had a configuration of 0.5%, 1%, and 1.5% wt/vol of HP-20, XAD7HP, and HP-2MG, respectively. This means that the physical- The resulting statistical model revealed that HP-2MG resin concentration (wt/vol) (p < 0.02) had the most significant effect. The statistical model was subsequently used to predict the optimal settings for each of the significant factors, as summarized in Figure 3.
According to the DoE analysis, the main effect to obtain higher taxane titers was the polar resin HP-2MG. Therefore, the concentration of the HP-2MG resin should be kept at the upper level of the treatments tested, which was 1.5% (wt/vol) (Figure 3). Interestingly, the model also showed a curved behavior for the aliphatic resin XAD7HP, making the optimum concentration for that resin to be 1.2% (wt/vol). Last, the prediction profiler for the selected model showed that the nonpolar resin concentration of HP-20 should be left at the minimum tested concertation of 0.5% (wt/vol). Therefore, the best resin combination of adsorbent resins in the cultivation to obtain the higher taxanes titers according to the selected model was 0.5%, 1.2%, and 1.5% (wt/vol) of the resins HP-20, XAD7HP, and HP-2MG, respectively. This result agrees with the combination that was set up in treatment 9, with a small difference in the concentration of the XAD7HP resin which had 1% (wt/vol) in treatment 9% and 1.2% (wt/vol) in the optimized model.
To further investigate taxane adsorption using the ISPR, the same protocol for cultivation and extraction was performed, with the exception that the resins were separated from the media and biomass for the final quantification step. Only treatments that gave the best taxane titers were run (treatments 2, 7, and 9). Results are detailed in Figure 4, where taxanes adsorbed by the resins represented 80% on average of the total taxanes produced after the cultivation period, leaving the rest in the media and biomass.
Total taxanes were distributed almost equivalently in the three selected treatments, which were retrieved into the resins up to 144 ± 7, 165 ± 18, and 184 ± 21 mg/L, whilst having a total taxane concentration in the biomass up to 35 ± 2, 27 ± 1.5, and 44 ± 5 mg/L from treatments 2, 7, and 9, respectively ( Figure 4). These results represent a partition into the resins for 80%, 85%, and 80% of the total taxanes for treatments 2, 7, and 9, respectively. These results support previous studies where sesquiterpene (+)-zizaene was F I G U R E 3 Prediction factors profiler for the optimum resin concentration. The top-half section of the Figure indicates the optimum concentrations of each resin for a maximized desirability (0.83). The first three graphs of the top-half show the optimum resins concentrations (x axis) in red numbers (0.5%, 1.172%, and 1.5% wt/vol for HP-20, XAD7HP, and HP-2MG, respectively) to obtain the highest taxane concentration, which was predicted to be 179 mg/L (dashed red line in y axis). The shadowed area between the blue lines of the same graphs indicates the confidence interval for each factor. Graphs in the bottom half indicate the maximum desirability score to maximize taxane recovery. Besides this, the method for obtaining taxadiene partition was using a cell strainer, which could affect partition yields.

| Natural products discovery
An interesting advantage of this methodology is the higher capacity to recover many different natural products or secondary metabolites, since, as mentioned, it has the benefit of retrieving metabolites that had been secreted along with intracellular metabolites, leveraging the extended absorbing capacity of the combined beads. When compared to other retrieval methods (no extraction aid and the most commonly used the dodecane overlay method), the in situ cultivation using different resins showcased not only a higher total taxanes titer as previously discussed, but it also retrieved more taxane compounds in the GC-MS chromatograms ( Figure 5).
In total, 18 taxanes were found in treatment 9 with the resin's combination of 0.5%, 1%, and 1.5% (wt/vol) of HP-20, XAD7HP, and HP-2MG, respectively (Figure 5a). Using the same cultivation conditions, a dodecane cultivation and a cultivation without any extraction aid were run, from which only 11 and 10 total taxanes were found respectively in their GC-MS chromatograms ( Figure

| Cell-waste removal of ammonia and ROS
The removal of different cell-waste products using the beads was evaluated in the S. cerevisiae strain EJ2. The main results observed were a clear reduction of ROS and a slight decrease in ammonia present in the media ( Figure 6).
The final ammonia concentrations in treatments 9 and 7 were 176 ± 8 and 169 ± 5 mg/L, respectively, compared to the control value of 205 ± 7 mg/L at the end of cultivation (Figure 6b). The accumulation of ammonia in the control with no adsorption aid was 1.15-fold higher than treatment 9 and 1.2-fold higher than treatment 7, indicating that the interaction with this chemical during cultivation is reduced. This suggests that cell viability and optimal culture conditions are improved in resin treatments (Figure 6b). In previous studies, it has been demonstrated that ammonia both reduces yeast growth at high concentrations and can inhibit important enzymes, such as Acyl-CoA Synthases (ACS) (Zheng et al., 2012). This could also explain why total taxane production was enhanced by the resin treatments, as acetyl CoA is the initial step for terpenoid synthesis in strain EJ2. Increasing ammonia from 337 to 673 mg/L produced an increase of pH in Cryptococcus c. yeast from 7.5 to 8.5, having an inhibitory growth effect towards the production of a biofuel (Zheng F I G U R E 4 Taxane concentration in the partition between resin and biomass in treatments 2, 7, and 9. All treatments had the same cultivation parameters of 30°C, 350 rpm shaking speed, 2 mL of the working volume and media composition, with the only difference being the composition of the three resins in each treatment (Table 2). Both biomass and resins had the same extraction procedure using 99% acetone. Bars represent the mean, and the error bars the standard deviation (n = 2). et al., 2012). For this reason, trying to maintain ammonia levels near the initial cultivation concentration could help optimize cell growth and desired NP production.
In contrast, ROS levels were statistically higher in control cultivations ( Figure 6a) compared to treatments with beads after 24 h of cultivation. After 72 h, the difference between the fluorescence emitted by the resin combination treatments was on average 30 ± 6 compared with 44 ± 5 in the control, meaning ROS concentration was 1.5-fold lower in treatments 9 and 7. Interestingly, no statistical difference was seen between the two resin combination treatments (Figure 6a). ROS production by yeast has an important effect on its growth and cell viability. Tert-butyl hydroperoxide (t-BuOOH), hydrogen peroxide (H 2 O 2 ) and menadione are common ROS produced by S. cerevisiae, and it has been shown that ROS concentrations greater than 0.4 mM or 0.6 mM can inhibit the growth of these yeast (Gazdag et al., 2014). Importantly, engineered S. cerevisiae could also induce the production of additional ROS by the action of cytochrome P450s enzymes (CYPs), which are necessary for the synthesis of oxygenated taxanes, provoking a pro-oxidative imbalance (Zhou et al., 2015).

| Semi-continuous cultivations at microscale
Using the same resin combination of treatment 9, a set of runs was conducted using a microplate to evaluate the possibility of semicontinuous production of taxanes. Cultivation parameters were not changed in comparison to batch cultivations; the main difference was the removal of spent media and biomass before reaching the stationary phase for new fresh YPG media in each cycle after 48 h. In addition, 3%, 6%, 9%, and 12% (wt/vol) total resin bead concentrations were used, keeping the same bead ratio of treatment 9 [0.5%, 1%, and 1.5% (wt/vol) of HP-20, XAD7HP, and HP-2MG, respectively]. The kinetics of biomass and total taxane production were evaluated throughout three cycles of media replacement and the total taxanes were recovered and compared with the batch cultivations (Figure 7). Before the removal of the first media, biomass was around OD 600 = 35 and 37 in the experiments, with total resin amounts of 3% and 6% (wt/vol), respectively, and around OD 600 = 30 for 12% (wt/vol). At the end of the semi-continuous cultivations, the final biomasses were OD 600 = 50 ± 4, 53 ± 3, and 42 ± 3, for 3%, 6%, and 12% total resins (wt/vol), respectively (Figure 6a). This biomass behavior at the end of the semi-continuous cultivation was similar to that obtained in the batch cultivations, which indicates that the renewal cycles did not interfere with biomass generation. Another important result seen in Figure 7a is how the engineered strain EJ2 endured the high resin concentration of the 12% (wt/vol) treatment.
It has been previously reported that a very similar strain (LRS5), which could only produce taxadiene due to its heterologous expression, did not grow on a 12% (wt/vol) cultivation using only HP-20 resin beads in microplate cultivation at the same conditions (Santoyo-Garcia et al., 2022). This result could demonstrate that the combination of resin beads in the cultivation of a delicate strain could be used to diminish the mechanical stress that the in situ cultivation might provoke.
Total taxane accumulation was the key result in this approach, as the resins were maintained during the whole semi-continuous cultivation, with the resins managing to continuously adsorb more quantities of taxanes in each cycle. This method highlights the suitability of using microscale tools to characterize fed-batch and semi-continuous fermentation as previously described (Teworte et al., 2022). Another important result was the paclitaxel early precursor titers, T5α-ol and T5α-yl acetate, which have a final concentration of 251 ± 25, 135 ± 6, and 95 ± 4 mg/L, respectively ( Figure 7b). Higher titers of oxygenated taxanes could be crucial when new metabolic pathways are studied to complete the final paclitaxel pathway in heterologous organisms. These final titers of T5α-ol and T5α-yl acetate are 5.7-and 25.7-fold higher than titers of the same compounds in previous studies . All the collected data in Figure 7b,c represent the total taxanes recovered after the extraction of intracellular and secreted taxanes that came from the resins and the biomass.
As the adsorption of the taxanes curve has not flattened, therefore, it can be concluded that the process could be continued to more cycles as the saturation point of the resins has not been reached.
F I G U R E 7 Biomass kinetics and total taxane production for the semi-continuous microplate cultivations. (a) Biomass kinetics during the cultivation time for three cycles of media replacement. The cultivations had the same parameters as batch cultivations with 30°C, 350 rpm and media YPG concentrations. (b) Kinetic production of the paclitaxel precursor's taxadiene, T5α-ol and T5α-yl Acetate, for treatment 9 during the semi-continuous cultivation using 6% (wt/vol) resins. (c) Total taxanes recovered from treatment 9 maintaining the same three resin proportions for 3%, 6%, and 12% (wt/vol). Green bars represent the total taxane concentrations (mg/L) for batch cultivations and blue bars the total taxane concentrations (mg/L) for semi-continuous cultivations. Bar and kinetic point values represent the mean (n = 3), and error bars represent the SD.
Salvia miltiorrhiza hairy root cultures (Yan et al., 2005). Besides this, as seen in Figure 7c, the maximum total taxanes retrieved from the cultivations at 6% wt/vol reached 783 ± 33 mg/L after 9 continuous days of cultivation. This product recovery behavior could be compared with previous studies, where the adsorption of natural 2-phenylethanol in an ISPR semi-continuous process had concentrations at the end of each cycle of 10.9, 21.1, and 32.3 g/L, showing a constant increase per cycle in the product yield (Wang et al., 2011).
The productivity of both cultivation approaches in 3% wt/vol were 3.2 (mg/L)/h for semi-continuous and 2.6 (mg/L)/h for batch cultivation, which represented a 1.2-fold increment. However, as the resins were not removed from the cultivation, no pretreatments of washing and autoclaving were necessary compared to using batchseries cultivations. This could improve the previously presented productivity making the ISPR semi-continuous approach a better candidate for industry application.

| CONCLUSIONS
In this study, it was shown that by combining three macro-porous resins with different chemical properties in the cultivation of engineered S. cerevisiae, it was possible to recover more taxane compounds at higher titers. A combination of the three adsorbent resins with 0.5%, 1%, and 1.5% (wt/vol) of HP-20, XAD7HP, and HP-2MG, respectively, resulted in a 7.9-fold increase in total taxane recovery compared to the control with no extraction aid. Moreover, by using the semi-continuous approach with 3% (wt/vol) of total resins, the total taxanes increased to a maximum concentration of 639 mg/L, which represented a 3.8-fold increase compared to batch cultivation using 3% (wt/vol). Within this total concentration of taxanes, it was found that the concentration of the oxygenated taxanes T5α-ol and T5α-yl acetate were 135 ± 6 and 95 ± 4 mg/L, respectively, the highest titer values reported to date. Treatments with resin combinations resulted in a lower concentration of cellwaste ROS compared to control with no extraction aid, which could lead to improved cell homeostasis during the cultivation and hence increase taxane production. This work, therefore, highlights a practical and novel ISPR method that leverages different combinations of bead materials for the sustainable and enhanced discovery and recovery of high-value products using microbial cell factories.
Finally, we believe that all the gathered data in this study could lead to interesting unstructured modeling approaches to future scale-up bioprocess.