Biotechnological synthesis of water‐soluble food‐grade polyphosphate with Saccharomyces cerevisiae

Inorganic polyphosphate (polyP) is the polymer of phosphate. Water‐soluble polyPs with average chain lengths of 2–40 P‐subunits are widely used as food additives and are currently synthesized chemically. An environmentally friendly highly scalable process to biosynthesize water‐soluble food‐grade polyP in powder form (termed bio‐polyP) is presented in this study. After incubation in a phosphate‐free medium, generally regarded as safe wild‐type baker's yeast (Saccharomyces cerevisiae) took up phosphate and intracellularly polymerized it into 26.5% polyP (as KPO3, in cell dry weight). The cells were lyzed by freeze‐thawing and gentle heat treatment (10 min, 70°C). Protein and nucleic acid were removed from the soluble cell components by precipitation with 50 mM HCl. Two chain length fractions (42 and 11P‐subunits average polyP chain length, purity on a par with chemically produced polyP) were obtained by fractional polyP precipitation (Fraction 1 was precipitated with 100 mM NaCl and 0.15 vol ethanol, and Fraction 2 with 1 final vol ethanol), drying, and milling. The physicochemical properties of bio‐polyP were analyzed with an enzyme assay, 31P nuclear magnetic resonance spectroscopy, and polyacrylamide gel electrophoresis, among others. An envisaged application of the process is phosphate recycling from waste streams into high‐value bio‐polyP.

due to its beneficial physicochemical properties, whereas the chain length determines which properties are more pronounced. For example, polyP 2 and 3 reactivate the water holding capacity of meat after the rigor mortis and intermediate-chain polyP complexes higher valent cations in soft cheese production.
The biotechnological synthesis of polyP appears promising because the polyP chain length in microorganisms can reach up to a thousand P-subunits (Rao et al., 2009). Furthermore, microorganisms can produce polyP from impure P i, whereas chemical polyP synthesis relies on pure P i . We recently reported a process to obtain Saccharomyces cerevisiae (baker's yeast) containing up to 28% polyP (as KPO 3 ) in cell dry weight . S. cerevisiae was chosen as the polyP production host because it is generally regarded as safe (GRAS) and other food-related fungi, such as Pichia pastoris, Kluyveromyces lactis, and Hansenula saturnus, produced only little polyP (<7%) in our hands (data not shown).
In S. cerevisiae, the transport of P i across the cell membrane is mediated by the low-affinity transporters Pho87 and Pho90 (K m ∼1 mM) and two high-affinity P i transporters Pho84 and Pho89 (K m ∼10 μM, Figure 1; Canadell, Gonzalez, Casado, & Arino, 2015).
During polyP synthesis, S. cerevisiae primarily metabolizes glucose via alcoholic fermentation to ethanol, which produces both the required energy and activated P i in the form of adenosine triphosphate (ATP; . PolyP is synthesized by the vacuolar transporter chaperone (VTC). The enzyme complex is located in the vacuolar membrane, couples the synthesis of polyP to its translocation across the vacuolar membrane, consumes ATP, consists of the five subunits VTC 1-5, and is presumably dependent on the proton gradient that is created by the V-ATPase (Desfougeres, Gerasimaite, Jessen, & Mayer, 2016). Langen and Liss (1958) showed that, after P i starvation, long-chain polyP is produced de novo, and later hydrolyzed to shorter chain polyP. The high polyP content during P i feeding is due to an increased polyP synthesis and not due to reduced polyP degradation (Liss & Langen, 1962). The polyP degradation in the vacuole is facilitated by S. cerevisiae endopolyphosphatase 1 and 2 . P i that is generated in the vacuole is transported to the cytosol by Pho91 (Eskes, Deprez, Wilms, & Winderickx, 2018).
To the authors' best knowledge, there are no reports on the biosynthesis of a food-grade water-soluble polyP with a highly scalable biotechnological process. The overall goal of the here presented study was the biosynthesis of polyP from P i with the focus on the purification (so-called preparative polyP extraction) of the synthesized polyP from polyP-rich S. cerevisiae. The desired characteristics of such a biotechnologically synthesized polyP included: appearance as a dry white water-soluble powder, food-grade quality, a linear molecular structure, a purity comparable with chemically produced polyP, and one polyP with mostly sodium and one with mainly potassium as the counterion. The term "bio-polyP" is proposed for the product of the here described synthesis. All process steps were designed to be highly scalable for intended large scale production. The physicochemical properties of the bio-polyP were analyzed and compared with chemically synthesized polyP.

| Synthesis of bio-polyP
NaCl and NaOH were used in Steps 10-12 to produce sodium bio-polyP. KCl and KOH were used to synthesize potassium bio-polyP.
Nondenatured ethanol was used. All steps were performed at room temperature except Steps 1 and 6. 1. PolyP-rich S. cerevisiae was produced according to   Verduyn, Postma, Scheffers, and Van Dijken (1992), and pH 5 with HCl/NaOH) with mild agitation. After cell harvesting and one washing with sterile water, starved S. cerevisiae was stored overnight at 4°C, and incubated at 7.5 g cell dry weight × L −1 and 30°C anaerobically for 2.5 hr in feeding medium (250 mM glucose, 60 mM KH 2 PO 4 , 20 mM MgCl 2 , and pH 6 with HCl/KOH) with mild agitation. PolyP-rich S. cerevisiae was washed twice with sterile water and dried for 5 min on P i -free filter paper to obtain a wet cell mass that contained ca. 25% dry matter. 2. The wet cell mass (weight w 2 ) was transferred to an aluminum flask and stored at −20°C. 3. The cell mass was thawed. 4. To the cell mass, 5 ml autoclaved Milli-Q water was added per g wet cell mass (referring to w 2 ) and mixed. 5. The flask was incubated for 10 min at 70°C in a vigorously rocking water bath. An aluminum or stainless steel flask was used for optimal heat transfer. 6. The flask was placed on ice for 5 min to cool the content down to room temperature. The content was moved to a centrifugation bucket. 7.
The insoluble matter was removed by centrifugation (10,000 g, 5 min). The pellet was discarded, the volume of the supernatant measured (v 7 ), and the supernatant moved to a new centrifugation bucket. 8. To precipitate the protein and nucleic acid, 0.02 vol of v 7 HCl (2.5 M) was added, and the content mixed. Because polyP hydrolyzes at low pH values, Steps 9-11 were carried out quickly. It was important not to change the order of Steps 8-10 because the HCl precipitation was reversible by alkali addition.
9. The insoluble protein and nucleic acid were removed by centrifugation (10,000 g, 15 min). The prolonged centrifugation time was necessary due to the fine nature of the suspended solids. The supernatant was moved to a new centrifugation bucket, while the pellet was discarded. 10. To neutralize the HCl from Step 8, 0.02 vol of v 7 NaOH or KOH (both 2.5 M) was added after the centrifugation, and the solution was mixed. 11. The pH was set to 7 with the help of a pH electrode with NaOH or KOH.
The used alkali volume was noted. The overall volume (v 11 ) was calculated by adding the volumes of Steps 8, 10, and 11 to v 7 . 12.

| PolyP analytics
To determine the cellular polyP content and the average polyP chain length in S. cerevisiae, the polyP was extracted from the cells with an analytical polyP extraction (Christ & Blank, 2018). Briefly, S. cerevisiae was suspended in a pH-buffered ethylenediaminetetraacetic acid (EDTA) solution and lyzed with phenol. The lysate was washed with chloroform and then used for further analysis. The total polyP (only linear polyP and no cyclic polyP), P i , and the average polyP chain length were determined enzymatically (Christ, Willbold, & Blank, 2019). Briefly, P i was assayed colorimetrically after the addition of a P i detection agent, which contained antimony, molybdate, ascorbate, and sulfuric acid. For total polyP analysis, polyP n was enzymatically hydrolyzed to n P i by S. cerevisiae exopolyphosphatase 1 and S. cerevisiae inorganic pyrophosphatase 1. The released P i was measured colorimetrically. The average polyP chain length was measured as the quotient of the total polyP concentration and the polyP chain concentration. The polyP chain concentration was quantified in an enzyme cascade with the enzymes S. cerevisiae exopolyphosphatase 1 (polyP n → polyP 2 ), ATP sulfurylase (polyP 2 + AMP-sulfate → ATP + sulfate), hexokinase (ATP + glucose → ADP + glucose 6-phosphate), and glucose 6-phosphate dehydrogenase (glucose 6-phosphate + NADP + → 6-phosphogluconolactone + NADPH, NADPH measured fluorometrically). For the study of the precipitation behavior of chemically produced polyP (Figure 2), the total polyP was measured gravimetrically after drying the dissolved polyP at 120°C. To determine the water solubility and pH of the polyP, a 1% (w/v) polyP suspension was vigorously stirred for 5 hr. If some of the polyP did not dissolve, the suspension was centrifuged (5 min, 10,000 g), the pellet dried in a desiccator for 7 days, and the insoluble matter weighed. The

| RESULTS
The envisaged workflow for the synthesis of bio-polyP included six process steps. In process Step 1, S. cerevisiae was starved in P i -free medium (P i starvation). The starved cell mass was subsequently moved to P i -containing medium, where S. cerevisiae took up P i and intracellularly polymerized it into polyP (P i feeding, Step 2). The combination of Steps 1 and 2 is called polyP hyperaccumulation and was already developed in a previous study . In process Step 3, the polyP was liberated from the yeast cell and brought into aqueous solution. In process Step 4, the dissolved polyP was recovered and purified by precipitation. Afterward, the polyP was dried and milled (process Steps 5 and 6). Process Steps 3-6 were developed in this study.
3.1 | Optimal conditions for the precipitation, drying, and milling of polyP The first aim was to understand which process conditions were required for the precipitation, drying, and milling of polyP. The process conditions that were developed here with chemically produced polyP were later used for the synthesis of bio-polyP. To verify that polyP can be precipitated with an organic solvent, a sodium polyP (Budit 4) was precipitated with 2 vol of ethanol, propanol, or acetone with or without NaCl (Figure 2a). PolyP was collected as a sticky viscous gel after precipitation. The recovery without NaCl was unsatisfactory with all organic solvents (≤44%). With the combination of either NaCl and ethanol, or NaCl and acetone, almost the entire polyP was recovered (95% and 97%, respectively). The recovery with isopropanol and NaCl was somewhat lower (92%). Because bio-polyP will be used as a food additive, ethanol was chosen as it has the lowest toxicity of the three tested compounds. The different polyP cation compositions were achieved by displacing the counterions with either Na + or K + .
Budit 4 was precipitated with a combination of different concentrations of NaCl or KCl, and 2 vol ethanol (Figure 2b). At a concentration of 50-750 mM of either salt, Budit 4 was fully recovered (≥95%). Because the recovered polyP was measured gravimetrically, it was concluded that NaCl itself did not precipitate. The precipita- (92% and 91% recovery with NaCl and KCl, respectively). Drying of the polyP gel was done in a desiccator that was filled with dried silica (without vacuum, Figure 2d). The initial water content of the sodium and potassium polyP gels amounted to 49.0% and 42.6%, respectively. After 7 days, water contents of 0.9% and 1.6% were measured in the sodium and potassium polyP gels, respectively. The water content of unprocessed Budit 4 was 0.2 ± 0.0% (mean ± standard error of the mean (SEM), five replicate measurements). The obtained water content was considered adequate for storage and milling.
Budit 4 was recovered as a coarse white crust after drying. To obtain a homogenous fine-grained powder, polyP was milled for 2 min in a bead beater. The fine-grained sodium and potassium polyP powders dissolved readily in water. As for all polyPs, prolonged vigorous stirring was necessary during dissolution to avoid the formation of clumps. The pH of a 1% (w/v) sodium polyP solution amounted to 7.4 ± 0.0, and the pH of the potassium polyP solution was measured at 7.2 ± 0.0 (mean between two independent batches ± SEM). With a pH of 7.0 before the precipitation, this indicated that the pH remained almost unchanged throughout precipitation, drying, and milling.

| Biotechnological synthesis of polyP
The polyP content in polyP-rich S. cerevisiae amounted to 26.5 ± 0.8% polyP (as KPO 3 ) in cell dry weight with an average polyP chain length of 24 ± 1 P-subunits (mean ± SEM from three analytical extractions).
The starting protocol for the preparative extraction included a heat treatment to release the polyP from cells, pH neutralization, and precipitation with NaCl-ethanol. The parameters of the heat treatment (1 hr, 70°C) were inspired by Kuroda et al. (2002) who employed those parameters to release polyP from sewage sludge. Two dependent variables (extraction efficiency and average polyP chain length) were analyzed during the optimization experiments. The amount of extracted polyP and the chain length was constant for 3.5-8 ml water per g wet cell mass (Figure 3a). About 5 ml water per g of wet cell mass was chosen. Interestingly, the bio-polyP did not precipitate as a gel but as a solid due to the presence of higher valent cations. An incubation time of 10 min led to the highest extraction efficiency and was, thus, chosen (Figure 3b). The chain length decreased significantly by 1 P-subunit per 10 min due to the heat catalyzed hydrolysis of the polymer (multiple correlation coefficient r = .983, p < .001). No shorter incubation time was tested because quicker heating and cooling would increase the process cost in an upscaling. Because the highest extraction efficiency was found at an incubation temperature of 70°C, this temperature was chosen ( Figure 3c). The pH after the heat treatment was acidic. Dilute NaOH was tested as an extractant instead of pure water to immediately neutralize the pH (Figure 3d). About 1 mM NaOH showed the same performance as pure water. Five and 10 mM NaOH decreased the extraction efficiency and the polyP chain length. Fifty and 100 mM NaOH increased the extraction efficiency but decreased the chain length profoundly, and led to precipitate formation during pH neutralization. Pure water was chosen. Dilute NaCl is commonly used to liberate RNA, which behaves chemically somewhat similar to polyP, from yeast cells (Kuninaka, Fujimoto, Uchida, & Yoshino, 1980). All tested NaCl concentrations 1-200 mM) reduced both the extraction efficiency and the chain length (Figure 3e). To remove protein and nucleic acid, an HCl precipitation was inserted before the ethanol precipitation. PolyP keeps its negative charge, even at very low pH, due to the low pK a value (pK a = 0-3) of all but two hydroxyl groups per polymer. In contrast, protein and nucleic acid protonate and precipitate at low pH. The ratio of polyP to protein and nucleic acid was increased from 4.1 to 6.2 (w/w), while the extraction efficiency and chain length decreased only by 2.2% points and 0.6 P-subunits, respectively, if 50 mM HCl was used (Figure 3f). Intermediate-chain polyP (41 P-subunits) was recovered in a fractional precipitation with 0.15 vol ethanol (26.9% extraction efficiency, Figure 3g,h). The remaining short-chain polyP (18 P-subunits, 52.5% extraction efficiency) was recovered by adding 1 final vol ethanol. Overall, 80% of the polyP was recovered, which agreed with the extraction efficiency that was obtained with only one precipitation step with 1 vol ethanol.

| Physicochemical characterization of the bio-polyPs
The synthesis of bio-polyP was scaled up by Factor 200 (from 5 mg to 1 g). In the last paragraph, the bio-polyP was liberated from the cells and precipitated with ethanol. The analytics were done with the pellet that was obtained after the ethanol precipitation. In this paragraph, the bio-polyP was dried and milled. A potassium bio-polyP was produced as well as a sodium bio-polyP. The molecular structures of the four newly synthesized bio-polyPs are displayed in Figure 4. The results of the physicochemical characterization of the bio-polyPs in comparison to the three longest, in bulk available commercial polyPs are shown in Table 1. All polyPs appeared as a fine-grained white powder, except for polyP P100, which was delivered as large Liberation of bio-polyP from polyP-rich Saccharomyces cerevisiae and fractional precipitation of bio-polyP. The extraction efficiency was calculated by dividing the amount of recovered polyP by the amount of polyP that was extractable with the reference method (analytical polyP extraction from Christ & Blank (2018). (a-h) Show individual experiments that build upon each other. 100 mg wet cell mass (25% dry matter) was suspended in different volumes of water in a 2 ml reaction tube. The suspension was incubated for 1 hr at 70°C and 750 rpm with one 3.2 mm stainless steel bead per reaction tube to facilitate agitation. After the insoluble matter was removed by centrifugation, the supernatant was transferred to a new reaction tube. The pH was neutralized and the polyP precipitated with 100 mM NaCl and 1 vol ethanol. (a) The volume of water, in which the cell mass was suspended, was varied and set to 5 ml per g wet cell mass. P100 contained a small amount of P i (0.7%). The intermediate-chain bio-polyPs did not contain P i , as well. The sodium and potassium short-chain polyPs contained 1.8% and 0.6% P i , respectively. The nucleic acid content of the bio-polyPs was estimated spectrophotometrically, a nonspecific method, to be-as desired-low (0.1-1.7%). NaCl or KCl was used to aid the ethanol precipitation. Both did not precipitate, because no chloride was detected in the bio-polyPs. Arsenic, cadmium, calcium, chromium, copper, iron, lead, nickel, and vanadium were neither

| Mass fluxes in the biosynthesis of polyP
The substrates and products of the P i starvation and P i feeding are displayed in Reaction 1 and Reaction 2, respectively. The production of the cell mass, which was required for Reaction 1, and the vitamins and trace elements in the P i starvation medium were not included in the reactions. The mass balance of the preparative sodium polyP extraction is depicted in Reaction 3. The increase in the water volume stemmed from cell water. For the preparative extraction of  Currently, there is only the chemical route to produce food-grade polyP on an industrial scale. Chemical polyP synthesis (a condensation reaction) is done by heating pure P i at 400-800°C for several hours.
The main challenge of chemical polyP synthesis lies in its dependence on pure substrate (P i ). P i is mined from P i rock, purified, and imported as phosphoric acid into countries that do not possess P i rock reserves.
The substrate for chemical polyP synthesis (P i ) is obtained by pH neutralization of phosphoric acid with NaOH. Problems associated with P i rock mining include an uneven global P i rock distribution, the limited nature of P i rock, environmental destruction and pollution during P i rock mining, contamination of P i rock with toxic and radioactive elements, and transportation cost (Reta et al., 2018). Strategies for the more efficient use of P i and the recycling of P i from unused P i waste streams must be developed to sustain human life on earth .
We developed a green biotechnological process to synthesize pure food-grade polyP with S. cerevisiae. The biotechnological polyP synthesis F I G U R E 5 PAGE analysis of bio-polyP and chemically produced polyP. The DNA low range ladder (NEB) was used in lane 1. The DNA fragments measured 766,500,350,250,200,150,100,75,50, and 25 base pairs. The depicted polyP chain lengths were calculated according to Smith et al. (2018). The lanes 2 and 3, 4 and 5, 6 and 7, and 8 and 9 show individual batches. Abbreviations: Inter., intermediate; K, potassium; Na, sodium. consumes less energy than the chemical synthesis because it is done at ≤ 30°C. In comparison with chemical polyP synthesis, S. cerevisiae can directly utilize low P i concentrations (ca. 10-60 mM P i ) from impure sources. Chemical polyP synthesis cannot be done from such waste streams without extraction and extensive purification of the P i . In this study, we used pure P i to feed S. cerevisiae. Different kinds of waste streams should be tested as P i source in future studies. The primary requirements for our process include that the P i is dissolved, and other dissolved substances do not inhibit S. cerevisiae excessively. Food-grade P i waste streams, such as agricultural plant waste (Carraresi, Berg, & Bröring, 2018;Herrmann, Ruff, & Schwaneberg, 2020;Herrmann, Ruff, Infanzon, & Schwaneberg, 2019) and some spent fermentation broths, would allow the production of food-grade bio-polyP. There are many applications of polyP not related to food (e.g., paint, fertilizer, cleaning agents, and flame-retardants). Nonfood-grade P i waste streams, such as industrial wash water and sewage sludge ash, can be used for biotechnological technical-grade polyP production.
The released polyP is precipitated with Ca 2+ and used as fertilizer.
The Heatphos polyP cannot be used in food because the product is neither food-grade (P i source: sewage sludge) nor water-soluble (calcium polyP). The water-solubility of polyP is of importance for food applications where polyP can only display its desired physicochemical properties when dissolved. The here described process bypasses the disadvantages of the Heatphos process by extracting the polyP from S. cerevisiae and polyP precipitation with NaCl (or KCl) and ethanol.