Bioprocess development to produce a hyperthermostable S‐methyl‐5′‐thioadenosine phosphorylase in Escherichia coli

Nucleoside phosphorylases are important biocatalysts for the chemo‐enzymatic synthesis of nucleosides and their analogs which are, among others, used for the treatment of viral infections or cancer. S‐methyl‐5′‐thioadenosine phosphorylases (MTAP) are a group of nucleoside phosphorylases and the thermostable MTAP of Aeropyrum pernix (ApMTAP) was described to accept a wide range of modified nucleosides as substrates. Therefore, it is an interesting biocatalyst for the synthesis of nucleoside analogs for industrial and therapeutic applications. To date, thermostable nucleoside phosphorylases were produced in shake flask cultivations using complex media. The drawback of this approach is low volumetric protein yields which hamper the wide‐spread application of the thermostable nucleoside phosphorylases in large scale. High cell density (HCD) cultivations allow the production of recombinant proteins with high volumetric yields, as final optical densities >100 can be achieved. Therefore, in this study, we developed a suitable protocol for HCD cultivations of ApMTAP. Initially, optimum expression conditions were determined in 24‐well plates using a fed‐batch medium. Subsequently, HCD cultivations were performed using E. coli BL21‐Gold cells, by employing a glucose‐limited fed‐batch strategy. Comparing different growth rates in stirred‐tank bioreactors, cultivations revealed that growth at maximum growth rates until induction resulted in the highest yields of ApMTAP. On a 500‐mL scale, final cell dry weights of 87.1–90.1 g L−1 were observed together with an overproduction of ApMTAP in a 1.9%–3.8% ratio of total protein. Compared to initially applied shake flask cultivations with terrific broth (TB) medium the volumetric yield increased by a factor of 136. After the purification of ApMTAP via heat treatment and affinity chromatography, a purity of more than 90% was determined. Activity testing revealed specific activities in the range of 0.21 ± 0.11 (low growth rate) to 3.99 ± 1.02 U mg−1 (growth at maximum growth rate). Hence, growth at maximum growth rate led to both an increased expression of the target protein and an increased specific enzyme activity. This study paves the way towards the application of thermostable nucleoside phosphorylases in industrial applications due to an improved heterologous expression in Escherichia coli.

In industry, NPs are attractive biocatalysts for the synthesis of nucleoside analogs which are widely used for the treatment of viral infections and cancer (Bonate et al., 2006;De Clercq & Li, 2016;Flexner, 2007).Chemo-enzymatic production of nucleoside analogs has many advantages compared to chemical synthesis, which is still the "golden standard" in industrial production.Chemical methods for the synthesis of nucleoside analogs are mainly multi-step synthesis routes involving several protection and deprotection steps.Intermediates have to be isolated in almost every step due to the poor regio-or stereoselectivity of the reactions (Anderson et al., 2008;Cen & Sauve, 2010;Mikhailopulo, 2007;Tennilä et al., 2000;Yehia et al., 2018).The advantages of chemo-enzymatic methods, on the other hand, include the limited use of organic solvents, mild reaction conditions and high stereoand regioselectivity (Almendros et al., 2012;Westarp et al., 2022).Furthermore, chemo-enzymatic methods meet the requirements of "green chemistry" to a great extent (Anastas & Eghbali, 2010;Kaspar et al., 2021a;Mikhailopulo & Miroshnikov, 2010).
One of the limitations of the widespread application of chemoenzymatic nucleoside production is the low solubility of nucleobases requiring the use of solvents and/or increasing the reaction temperature to achieve high product concentrations.While mesophilic NPs do not withstand these reaction conditions, their thermostable counterparts show an impressive tolerance towards harsh reaction conditions (Boone et al., 2015;Bruins et al., 2001;Kaspar et al., 2021b).A drawback of thermostable enzymes, however, are the challenges regarding their efficient expression in mesophilic hosts.Poor expression results mainly from codon bias between the thermophilic donor and mesophilic host, insufficient disulfide bond formation, a need for optimal folding temperature or specific activation factors.Nevertheless, strategies have been developed to overcome these limitations.These include coexpression of rare tRNAs, codon optimization, (Wang & Zhang, 2009) optimization of 5′-mRNA (Szeker et al., 2011) or the expression at elevated temperature (Koma et al., 2006;Szeker et al., 2011).
Despite this progress, the expression of thermostable NPs has so far been carried out only on a shake flask scale.This has the disadvantage that only very low final optical densities can be achieved, which is accompanied by comparatively low volumetric protein yields in the range of 35-235 mg enzyme/L culture (Table 1).The efficient expression of (thermostable) NPs in high cell density (HCD) cultivations, however, would be an important prerequisite for their economic application in industrial processes.Using the fed-batch technology, oxygen limitation and over-flow metabolism, which are common complications of batch cultivations, can be avoided by controlling the growth rate via substrate limitation (Schaepe et al., 2014).This allows to reach a significantly higher biomass and therefore also higher volumetric yields of recombinantly expressed proteins (Krause et al., 2016).
In this study, the bioprocess development to improve heterologous expression of the hyperthermophilic ApMTAP in Escherichia coli is described.Before the study, the production of the thermostable enzyme was performed in terrific broth (TB) medium.This resulted in volumetric protein yields of 11.2 mg L −1 (Table 1).To improve protein yields HCD cultivation protocols were established.To identify suitable reaction conditions, 24-well plate experiments were performed using a fed-batch medium as it allows the results to be reliably transferred to bioreactor cultivations.Afterwards, bioreactor cultivations were performed and the impact of varying growth rates on ApMTAP yields was studied as it has been shown that the growth rate can have a significant effect on the recombinant protein yield (Norsyahida et al., 2009;Sandén et al., 2003;Yee & Blanch, 1992).
T A B L E 1 Yields of different PNPs during expression in E. coli.

| Expression system
Protein expression was performed with the recombinant E. coli BL21-Gold strain bearing a pKS2-ApMTAP expression vector, a derivative of the vector pCTUT7, (Zhou et al., 2013) carrying the ApMTAP gene.
This vector harbors the gene that encodes the ApMTAP with an N-terminal histidine-tag, an inducible lac promotor and an ampicillin resistance.The vector map is presented in Figure S1.

| Growth media and preculture cultivation conditions
Either Luria broth (LB) medium or mineral salt medium (MSM) was applied in this study.MSM composed of: 14.6 g L −1 K 2 HPO 4 , 3.6 g L −1 NaH 2 PO 4 • 2H 2 O, 2.0 g L −1 Na 2 SO 4 , 2.47gL −1 (NH 4 ) 2 SO 4 , 0.5gL −1 NH 4 Cl, 1.0 g L −1 (NH 4 ) 2 -H-citrate, 0.1 g L −1 thiamine hydrochloride, 2 mM MgSO 4 • 7H 2 O, 0.1 mL L −1 antifoam PPG2000 and 2 mL L −1 trace elements solution.Trace elements solution was prepared as a 500x stock were supplemented with 100 μg mL −1 ampicillin.For the microwell plate experiments, a one-step preculture was performed, whereas a two-step preculture was needed to test different media additives and bioreactor cultivations.For the first preculture, 10 mL of LB medium were inoculated with 100 µL of a glycerol stock containing the E. coli BL21-Gold pKS2-ApMTAP and incubated in an orbital shaker (Lab-Therm LT-X, Adolf Kühner AG; 50 mm amplitude) at 37°C, 250 rpm for 6-8 h, attaining an OD 600 of around 5. To prepare the second preculture, 100 mL of MSM were inoculated in 500-mL Ultra Yield Flasks (UYF, Thomson Instrument Company) and closed with an AirOtop membrane (Thomson Instrument Company).After inoculation with the first preculture to an OD 600 of 0.15, the flask was incubated in an orbital shaker at 250 rpm, 30°C for 12 h.For the testing of different media additives, 0.24 g L −1 tryptone and 0.48 g L −1 yeast extract were also added to the preculture.

| Pre-experiments in microwell plates
Experiments were carried out in 24-well plates with integrated optical oxygen sensors (OxoDish OD24; PreSens GmbH) with a working volume of 1 mL per well.The 24-well plates were covered with oxygen-permeable sandwich covers (Duetz et al., 2000) (Enzyscreen B.V.).For oxygen measurement, the plates were placed on a sensor tray reader (SDR; PreSens GmbH).The plate and the reader were fixed with a clamp system (Enzyscreen B.V.) on an orbital shaker (Lab-Therm LT-X, Adolf Kühner AG; 50 mm amplitude) and incubated at 30°C and 250 rpm.The concentration of dissolved oxygen (DO [%]) was recorded online over the whole cultivation period in 1-min intervals using the software SDR_v4.0.0 (PreSens GmbH).
The EnBase technology (Enpresso GmbH) enables fed-batch mode culture conditions in small culture volumes.The release of glucose, as a sole carbone source, is controlled by the enzymatic degradation of a polysaccharide in the mineral salt medium containing essential nutrients to support growth.EnPump 200 and EnPresso B are both based on the EnBase technology.EnPump 200 is a powder-based feed system, which can be added to your individual growth medium whereas EnPresso B is a ready-to-use medium in the form of tablets that only needs to be dissolved in sterile water.In both systems an enzyme (Reagent A) is added, which releases glucose from the contained polysaccharides.
The main cultures in EnPresso B medium and MSM supplemented with 30 g L −1 EnPump 200 were inoculated to an OD of 0.15 using a first preculture.Each medium was supplemented with 1.5 U L −1 enzyme (Reagent A) for glucose release and 0.1 μL mL −1 antifoam PPG2000 according to the manufacturer's instructions.The inoculated medium was mixed and distributed in 1 mL aliquots into the 24-well plate and incubated overnight.After overnight cultivation, 10 μL sample were collected from each well and manually diluted 1:11 for the OD 583 and glucose concentration analysis using the Cedex Bio HT Analyzer (Roche Diagnostics International AG) with the Glucose Bio HT and OD Bio HT kits (Roche Diagnostics International AG).When an OD 583 between 9 and 11 was reached, the enzyme expression was induced by isopropyl β-D-1-thiogalactopyranoside (IPTG) addition using varying concentrations (0, 20, 50, 100, 200, 500, and 1,000 μM).At the same time, the booster solution containing 1.5 U L −1 enzyme (Reagent A) was added to selected wells, according to the manufacturer's specifications, while a blank of sterile water was added to the others.
In previous studies it was observed that the addition of EnBase Booster improves the cell growth and the amount of soluble protein (Glazyrina et al., 2010;Nancib et al., 1991).The utilization of complex medium additives can further lead to an improved pH balance (Krause et al., 2010).
For the booster preparations, the booster tablet was dissolved in 5 ml ddH 2 O and 25 μL enzyme (Reagent A, final concentration 1.5 U L −1 ) were added.95 μL of this mixture were added to each well.Additionally, 5 μL of the required IPTG solution or 5 μL sterile water were added to the respective well.For the preparations without booster addition, 95 μL sterile water were added to the selected wells.The cultures were further incubated for another 24 h at 30°C and 250 rpm.At the end of the cultivations OD 583 and glucose concentration were determined using the Cedex Bio HT Analyzer.Additionally, OD 583 = 10 samples were taken from each well.

| Preparations for the Fed-batch bioreactor cultivations
The reactors containing an initial volume of 0.5 L MSM were autoclaved and the medium was complemented with sterile solutions of ampicillin, thiamin, MgSO 4 x 7H 2 O and trace elements solution as described previously (Ongey et al., 2019).For the batch phase, 15 or 18.9 g L −1 sterile glucose were added.The feeding solution contained 650 g L −1 glucose.The bioreactors were inoculated with the MSM precultures, in the exponential growth phase, to OD 583 values of approximately 0.15.induction can be beneficial, because when the growth rate decreases because of substrate limitation in a fed-batch culture, the concentration of ribosomes is reduced and leads to a reduced translation (Enfors, 2019).After glucose was consumed a short exponential feeding with a feeding rate of 0.7 times µ max was included followed by a constant feeding and the start of induction by IPTG addition.

| Bioreactor cultivations
Condition B: An exponential fed-batch with a feeding rate of 0.7 times μ max was applied.Reaching a stirrer speed of 1200 rpm and an aeration rate of 2 vvm, the feed was changed to a constant feed and the culture was induced by IPTG addition.
Condition C: An exponential fed-batch with a feeding rate of 0.7 times μ max was applied.After reaching a stirrer speed of 1200 rpm and an aeration rate of 2 vvm, the feed was changed to a constant feed and 1 h later, the culture was induced by IPTG addition.
Condition D: An exponential fed-batch with a feeding rate of 0.35 times μ max was applied.Reaching a stirrer speed of 1200 rpm and an aeration rate of 2 vvm, the feed was changed to a constant feed and the culture was induced by IPTG addition.
Cultivations were initially performed in batch mode.The 8-10.5 h batch phase was followed by either a fed-batch phase or The initial feed rate (F 0 , L h −1 ) was calculated according to Equation 2, where Y X/S (g g −1 ) is the biomass/substrate yield (calculated from the batch phase), S i is the concentration of the feeding solution ), X 0 is the biomass concentration (g L −1 ) at the end of the batch phase (calculated from a correlation between previous OD 583 and cell dry weight (CDW) measurements) and V 0 is the total liquid volume in the bioreactor at the end of the batch phase.
The complete cultivation process took 35-41 h.To avoid undesired magnesium limitations, 2 mL of 1 M MgSO 4 solution were added each time the OD 583 increased by 20 increments.Foam formation was treated with Antifoam PPG 2000 in 50 μL additions (a maximum amount of 450 μL was added).

| Sample preparation and analytical methods during the bioreactor cultivations
During the cultivation, 10 mL samples were collected every 2 h to determine OD 583 , CDW, glucose, acetate, ammonium and magnesium concentrations.The OD 583 was determined either with a photometer or the Cedex Bio HT Analyzer (OD 583 -Cedex).
OD 583 = 10 cell pellet samples were withdrawn before induction then regularly with the following sampling and stored at −20°C until further use for protein analysis.
The supernatant of the samples was analyzed using the Cedex Bio HT Analyzer to determine glucose and acetate concentration every 2 h

| Evaluation of ApMTAP expression during bioreactor cultivations
To evaluate protein expression during the cultivation process, OD 583 = 10 samples were collected during bioreactor cultivations.
Mechanical lysis was performed using an ultrasonic treatment with an ultrasonic probe (UP200S, Hielscher; 2 mm sonotrode diameter) on ice for 5 min with 30% power input and 30 s on/off intervals.80 μL of the sample were centrifuged in a table centrifuge for 30 min at 4°C and 15,000 ×g to obtain soluble and insoluble protein samples.

| Purification of ApMTAP
An equal quantity (~4 g) of wet cell pellets was purified for each bioreactor cultivation condition.The pellets were resuspended in lysis buffer (50 mM sodium phosphate buffer, 300 mM NaCl, 10 mM imidazole, 0.6 mg mL −1 DNAse I, 1 mg mL −1 Lysozyme, 1 mM MgCl 2 , 0.1 mM PMSF, pH 8.0) and the cells were lysed by sonicating on ice using the UP200S sonicator (7 mm sonotrode diameter) for 5 min with 30% duty cycle and 30 s on/off intervals.The lysates were centrifuged for 15 min at 8,000 ×g at 4°C to remove cell debris.
Clarified lysates were purified using Ni-NTA Agarose and Polypropylene Columns (1-mL) (Jena Science).The loaded column was washed with washing buffer (50 mM sodium phosphate, 300 mM NaCl, 20 mM imidazole, pH 8.0) and all samples were eluted in 1.5 mL elution buffer (50 mM sodium phosphate, 300 mM NaCl, 250 mM imidazole, pH 8.0).The protein concentrations were estimated using the NanoDrop ND-1000 Spectrophotometer (Peqlab Biotechnologie) using the extinction factor E1% 8.8 of the ApMTAP.Samples from each purification step were processed for SDS-PAGE analysis.

| Activity assay for ApMTAP
In a final volume of 1 mL, a reaction mixture containing 1 mM adenosine in 50 mM potassium phosphate buffer (pH 7) was incubated at 60°C.The reactions were started by the addition of enzyme.After 15 min, samples were taken for further analysis and transferred to an equal volume of methanol.After centrifugation at 15,000 ×g for 10 min, the supernatant was analyzed by HPLC as described below.One unit of enzyme activity was defined as the amount of enzyme required for the conversion of 1 µmol adenosine per minute under the given reaction conditions.
Gradual elution was performed according to the following gradient: from 97% 20 mM ammonium acetate and 3% acetonitrile to 60% 20 mM ammonium acetate and 40% acetonitrile in 10 min.Substrate (adenosine) and product (adenine) were identified on account of their different retention times and their concentrations were calculated based on available standard curves.Standard deviations were calculated from two independent experiments.

| Optimization of ApMTAP expression in 24-well plates
In our group, cultivation of ApMTAP was performed in shake flasks and using TB medium before this study.The final optical density was around 3.4 and the volumetric protein yield was 11.2 mg L -1 .Since protein yields were too low to allow the use of ApMTAP in largescaled enzymatic syntheses for the production of valuable nucleoside analogs, alternative production methods were explored.HCD cultivations were identified as a suitable alternative.To identify suitable reaction conditions, experiments were first performed in 24-well plates.The EnBase system, a fed-batch medium that promises reliable transfer into bioreactor cultivations, was used.
The impact of varying IPTG concentrations and complex media additives on ApMTAP expression was screened in 24-microwell plates.
Increasing IPTG concentrations between 0 and 1,000 µM were studied.
At the time of induction, half of the cultures were supplemented with booster, a solution of complex media components (Figure 1a).The booster is part of the commercial EnPresso B medium and provides a balanced level of C-and N-sources for increased volumetric protein yield.
The 24-microwell plates were equipped with online monitoring of DO, which allows to estimate cell growth during cultivation.
During the entire cultivation, no oxygen limitation occurred (Figure 1b,c).Complete consumption of glucose (end of the batch phase) was observed after 12 h.Protein expression in the cultures was induced after 22 h at an OD 583 of 9-11.
A comparison of cultures with and without booster addition revealed that booster had a strong impact on oxygen consumption as an additional polysaccharide substrate, complex media additives and glucose-cleaving biocatalyst were added to the cultures.After induction, a strong drop in DO, up to 21%, was observed in the case of booster addition, while without booster, the DO did not drop below 75%.Booster addition did not only impact DO but also final ODs.In all tests without booster addition, final OD 583 values of only 12-14 were achieved, while with booster addition, final OD 583 of 25-30 were reached.In contrast to the addition of booster, varying IPTG concentration did not significantly impact oxygen consumption as comparable DO curves were observed.However, final ODs decreased with increasing IPTG concentrations.
To validate ApMTAP expression OD 583 = 10 samples were analyzed using SDS-PAGE (Figures 1c,d and S4) and a densitometric evaluation of the gels was conducted (Figure 2d,f,h,j).In MSM without booster addition, an increasing ApMTAP expression level up to approximately 2% of total protein was observed with increasing IPTG concentration until 100 µM.Afterwards, ApMTAP levels dropped to approximately 1.2% of total protein.A slightly higher overall ApMTAP expression in MSM was observed in cultivations with booster addition (approximately by a factor of 2).The beneficial effects of complex additives might be explained by an increase in intracellular protein accumulation, improved product stability and reduced product proteolysis (Krause et al., 2010;Nancib et al., 1991).Similar to cultures without booster addition, ApMTAP expression increased up to 100 µM IPTG concentration after which the yields were comparable.This observation is in good accordance with literature data showing that high inducer concentrations that are often used to fully induce the Iac promoter do not necessarily lead to maximal expression of a target protein (Glick, 1995).The inducer concentration must therefore be chosen individually for each case to balance the decreasing yield of recombinant cells after induction and the increasing cellular concentrations of the target protein (Bentley et al., 1991).
In conclusion, ApMTAP expression was shown to be possible in MSM-based fed-batch medium.IPTG concentrations of 100 µM were chosen for further studies as higher concentrations led to either constant or lower protein yields.Compared to the originally used TB cultures expression levels were comparable, however, obtained OD values were significantly increased by a factor of 2.2-2.7 which impacts protein yields.

| Expression of ApMTAP in 1.4-L stirred-tank bioreactors
HCD cultivations of E. coli BL21-Gold carrying the expression vector pKS2-ApMTAP were performed in 1.4-L stirred-tank bioreactors.Four different feeding strategies were tested and their influence on ApMTAP production was investigated in duplicates (Figure 2a).All four feeding strategies started with a batch phase.For the batch phase, a maximum specific growth rate (µ max ) of 0.5 h −1 was calculated.The end of the batch phase (sudden increase in the DO) was reached after approximately 10.5 h for condition A, and after approximately 8 h for conditions B, C and D. After the batch phase, either a fed-batch phase (condition B, C and D) or a second batch phase (condition A) followed to validate the impact of different growth rates on ApMATP expression.Feeding rates of µ max (condition A), 0.7 × µ max (condition B, C) and 0.35 × µ max (condition D) were used (Table S1).When the stirrer speed reached 1,200 rpm and the aeration rate 2 vvm, the feed was switched to constant.Cultures with feeding strategies A, B and D were directly induced with 500 μM IPTG (Figure 2a), while cultures with condition C were induced with 500 μM IPTG 1 h after the feed was changed.The Glucose and acetate concentrations were measured throughout the whole cultivation to analyze whether overfeeding occurred (Figure 2b).
For all conditions, the glucose concentration decreased as expected until the end of the batch phase(s) and remained at zero level during exponential and constant feeding until just before the end of cultivation except for condition D. Here, glucose and acetate accumulation were observed approximately 36 h after inoculation.Although the feed rate was decreased, glucose concentrations further increased.Therefore, feeding was stopped for condition D, which led to a reduction of the glucose concentration to zero.Acetate concentration also decreased but did not reach zero until the end of cultivation.Acetate accumulation was also observed for condition A in the second batch phase and for feeding conditions A to C towards the end of cultivation (10-20 h after induction).Measured acetate concentrations were between 1 and 5 g L −1 .Probably cell lysis, occurring towards the end of cultivation, led to an increase in acetate concentrations.

| Validation of ApMTAP expression by SDS-PAGE
Expression of ApMTAP was analyzed by SDS-PAGE using OD 583 = 10 samples that were taken regularly after induction.For all bioreactor cultivations, expression of ApMTAP was already observed before Densitometric analysis revealed that the ratio of ApMTAP compared to total protein for condition A was approximately 3.8% (Figure S5).
For the other tested conditions, ApMTAP ratio was in the range of 1.9%-2.2%.Hence, the use of the maximum growth rate led to the best ApMTAP expression, which is in good accordance with observations made in 24-well plates, where booster addition at the time of induction enhanced ApMTAP expression.
ApMTAP was purified from defined amounts of cell pellet for one of the replicate bioreactor cultivations by nickel affinity chromatography.In good accordance with the densitometric analysis of OD 583 = 10 samples, feeding condition A yielded the highest ApMTAP levels (~0.4 mg g −1 cell pellet wet weight) (Figure 3e).The protein yield for the other three feeding conditions was about a factor of 10 lower.

| Validation of ApMTAP activity
The reversible phosphorylation of adenosine to adenine and α-Dpentofuranose-1-phosphate was used to determine the specific activity of the four ApMTAP preparations produced under different expression conditions (A-D).The reactions were analyzed by HPLC and specific activities [U mg −1 ] were calculated (Figure 4).
Surprisingly, the specific activity was highest for condition A with approximately 4 U mg −1 .For the other conditions, the specific activity was significantly lower and ranged between 1 U mg −1 (condition B) and 0.21 U mg −1 (condition C).Hence, the growth rate used before induction not only affected the expression of ApMTAP but also the specific activity.Based on the existing literature, it is difficult to explain why a reduced growth rate during expression leads to reduced specific activity of the expressed ApMTAP, however, one explanation might be that chaperons are more readily available at higher growth rates, thus favoring the folding of ApMTAP in its active form.

| DISCUSSION
HCD cultivations of recombinant E. coli allowed the production of various proteins with high yields and productivities (Choi et al., 2006).
Despite the importance of thermostable NPs for the enzymatic synthesis of nucleoside analogs, their expression in HCD bioreactor cultivations was not yet shown.In this study, we successfully upscaled the heterologous expression of the thermostable ApMTAP in E. coli BL21-Gold from multi-well plates to 1.4-L stirred-tank bioreactors using a fully synthetic medium and glucose as the sole carbon source.After optimizing the inducer concentrations in multiwell plates, the process was successfully transferred to 1.4-L stirredtank bioreactors.ApMTAP yield was well comparable between smallscale and bioreactor cultivations.
In good accordance with Sandén and colleagues, we observed the best ApMTAP expression with the highest tested growth rate.
This could be attributed to the maximum number of ribosomes available while operating at the maximum specific growth rate (Enfors, 2019;Gausing, 1977).An additional advantage is that high glucose concentrations in the medium before IPTG addition suppressed the lac promoter and reduced basal expression before induction (Stülke & Hillen, 1999).
In comparison to other NPs, the specific yield for ApMTAP was comparably low under the tested conditions (Table 1).For other PNPs, specific yields between 12 mg g −1 and 21 mg g −1 were described, (Breer et al., 2008;Lee et al., 2001;Silva et al., 2003;Ubiali et al., 2012) which is by a factor of 3-5 higher than that observed for ApMTAP.An explanation might be the leaky expression system applied in this study.Lac promoters are known for leaky expression which might lead to cell stress and reduced growth and heterologous protein expression which cannot be completely prevented even by a high glucose concentration in the medium (Horn et al., 1996;Penumetcha et al., 2010).A more tightly regulated promoter like the arabinose PBAD promoter would be another alternative to minimize leaky expression (Guzman et al., 1995).In contrast to a lower specific yield, the volumetric yield, however, was by a factor of 7-40 higher compared with other described NPs.This is due to the HCD cultivation performed compared to standard shake flask cultivations applied for the other enzymes (Breer et al., 2008;Lee et al., 2001;Silva et al., 2003;Ubiali et al., 2012).
To further increase the expression of ApMTAP, the 5′-coding sequence of the target gene might be optimized.Ohashi and colleagues developed a statistical model for predicting the translation efficiency by evaluating the first four codons after the start codon (Ohashi et al., 2005).The initial translation index (ITI) indicates the translation efficiency based on the first codons of the coding sequence.The higher the ITI, the better the translation efficiency (Ohashi et al., 2005).As low expression can also be caused by translation-inhibiting 5' mRNA structures, (Griswold et al., 2003;Kudla et al., 2009) protein expression might be enhanced by the substitution of base pairs in the 5'-coding sequence of the target gene and the increase of the cultivation temperature above the standard used for recombinant protein expression (Szeker et al., 2011).The combination of both approaches led to an increased expression of correctly folded thermostable DgPNP from Deinococcus geothermalis over 18-fold.Primary and secondary protein structures also have a significant impact on translation, protein solubility and toxicity, therefore, protein optimization can be envisioned.As shown for human TP, the rational design of N-terminus-truncation constructs led to significantly increased protein expression (Karamitros et al., 2021).

| CONCLUSION
The production of a hyperthermostable NP in stirred-tank bioreactors using a fully synthetic medium and glucose as the sole carbon source was successfully shown in this study.Comparable protein yields were obtained in both multi-well plate format and stirred-tank bioreactors.In 1.4-L stirred-tank bioreactors, growth at maximum growth rate before induction led to the best expression potential of ApMTAP and highest specific enzyme activity.To optimize ApMTAP yield, further studies can focus on more tight expression systems or the application of varying expression temperatures.

Fed-batch cultivations
were carried out in 1.4-L Multifors bioreactors (Infors AG).The cultivation temperature was kept constant at 30°C and the pH was maintained at 7.0 ± 0.3 through the controlled addition of 25% (v v −1 ) ammonia solution.The initial stirring speed was set to 200 rpm and the initial air flow rate was set to 0.05 vvm.Both values were increased over time to ensure a sufficient oxygen supply (DO > 20%).A stirrer with two six-blade Rushton impellers at speeds ranging from 200 rpm to 1,500 rpm was used and the aeration rate was increased from 0.05 to 4 vvm over the cultivation course.Shortly after induction, the stirrer speed and aeration rate reached their maximum limits and the increasing oxygen demand was served by increasing the oxygen concentration in the supplied air.Three foam breakers, each consisting of two cable ties fixed at the impeller shaft above the working volume were used to mechanically destroy foam.The impact of four different feeding strategies on the expression of ApMTAP were analyzed in this study (Figure1): Condition A: A second batch with 42 to 45 g L −1 glucose was included to grow the cells at µ max and to further repress protein expression.A high glucose concentration in the medium up to a second batch phase.It was started when the complete consumption of the carbon source led to an abrupt increase of the DO values.Different exponential feeding rates were set according to the four cultivation conditions.When the stirrer speed reached 1,200 rpm and the aeration rate 2 vvm, the feed was switched to constant.Cultures with conditions A, B, and D were directly induced with 500 μM IPTG (Figure2).Cultures with condition C were induced with 500 μM IPTG 1 h after the feed was changed.The culture in the reactor is diluted by the glucose feed.Therefore, IPTG was added to the feed to keep the IPTG concentration in the reactor constant.The feed pump was controlled by the Eve software (Software for Multifors, Infors AG) and an exponential feed was set according to Equation (1):

F
I G U R E 1 Screening for optimum expression conditions for ApMTAP in microwell plates.The impact of varying IPTG concentrations and the addition of booster at the time of induction were validated.(a) Seedtrain for the expression of ApMTAP in MSM medium starting from a glycerol stock.(b and c) Dissolved oxygen (DO) curves using different IPTG concentrations for induction without (b) and with (c) booster addition.DO curves are exemplarily shown for one of the triplicates.(d, e) Final ODs were determined at the end of cultivation by photometric measurement.ApMTAP expression was validated by SDS-PAGE and densitometric analysis.ApMTAP, Aeropyrum pernix S-methyl-5′thioadenosine phosphorylase; MSM, mineral salt medium; SDS-PAGE, sodium dodecyl sulphate-polyacrylamide gel electrophoresis.F I G U R E 2 Expression of ApMTAP in 1.4-L stirred-tank bioreactors using four different feeding strategies.(a) Feeding strategies applied for ApMTAP production.(b) Feed rate, CDW, acetate and glucose were validated during bioreactor cultivations.Mean values and standard deviations of biological duplicates are shown for the optical density, glucose and acetate concentrations.The feeding rate is shown exemplarily for one replicate each.Full data sets are shown in Figure S3.ApMTAP, Aeropyrum pernix S-methyl-5′-thioadenosine phosphorylase; CDW, cell dry weight.and ammonium and magnesium concentration every 4-6 h with the test kits: Glucose Bio HT, Magnesium Bio HT, Phosphate Bio HT, NH3 Bio HT, and Acetate V2 Bio HT (Roche Diagnostics International AG) according to the manufacturer's recommendation.Cultivation was carried out for 24 h after induction and finally, 50 mL of the culture were harvested by centrifugation (4°C, 15,000 ×g, 10 min).
IPTG concentration used for induction was chosen based on the 24-well experiments, considering the higher cell mass in the bioreactors at the time of induction.During cultivation, CDW was regularly measured offline.The increase in CDW among the different reactors was very comparable during the initial batch phase.Afterwards, CDW increased depending on the feeding profile: with lower feeding rates, CDW increased slower and cultures grown under condition D (growth rate of 0.35 × µ max ) needed a significantly longer time until induction than the other cultures.At the time of induction, however, CDWs were highest for condition D as the slower growth of the cells led to lower oxygen demand.After induction, cells grew very comparably (Figures2b and S2).Final optical density (OD 583 ) values were in the range of 170 to 183.The CDW at the end of the cultivation was comparable for all feeding strategies with values between 85 and 99 g L −1 .

F
I G U R E 3 Validation of ApMTAP expression in high cell density bioreactor cultivations by SDS-PAGE.(a)-(d) OD 583 = 10 samples were taken regularly after induction.(a) Condition A, (b) Condition B, (c) Condition C, (d) Condition D. (e) At the end of cultivation, samples with a defined cell pellet wet weight were disrupted and purified.Soluble and insoluble fractions were analyzed by SDS-PAGE and densitometric analysis.Mean values were calculated for the parallel bioreactors.After purification, concentrations of the ApMTAP in the elution fractions were measured and used to determine ApMTAP content per g cell pellet wet weight.Arrows indicate ApMTAP in SDS polyacrylamide gels.ApMTAP, Aeropyrum pernix S-methyl-5′-thioadenosine phosphorylase; M, protein ladder; P, pellet after cell lysis; S, soluble protein fraction; SDS-PAGE, sodium dodecyl sulfate polyacrylamide gel electrophoresis.induction(Figure3a-d) showing the basal expression of the applied expression system.ApMTAP expression did not significantly increase over time except for condition A (maximum specific growth rate).

F
I G U R E 4 Specific activity of purified ApMTAP for the formation of adenine after expression in 1.4-L stirred-tank bioreactors under different conditions.Error bars indicate standard deviation from duplicates.ApMTAP, Aeropyrum pernix S-methyl-5′-thioadenosine phosphorylase.