Online measurement of the viscosity in shake flasks enables monitoring of γ‐PGA production in depolymerase knockout mutants of Bacillus subtilis with the phosphate‐starvation inducible promoter Ppst

Poly‐γ‐glutamic acid (γ‐PGA) is a biopolymer with a wide range of applications, mainly produced using Bacillus strains. The formation and concomitant secretion of γ‐PGA increases the culture broth viscosity, while enzymatic depolymerisation and degradation of γ‐PGA decreases the culture broth viscosity. In this study, the recently published ViMOS (Viscosity Monitoring Online System) is applied for optical online measurements of broth viscosity in eight parallel shake flasks. It is shown that the ViMOS is suitable to monitor γ‐PGA production and degradation online in shake flasks. This online monitoring enables the detailed analysis of the Ppst promoter and γ‐PGA depolymerase knockout mutants in genetically modified Bacillus subtilis 168. The Ppst promoter becomes active under phosphate starvation. The different single depolymerase knockout mutants are ∆ggt, ∆pgdS, ∆cwlO and a triple knockout mutant. An increase in γ‐PGA yield in gγ‐PGA/gglucose of 190% could be achieved with the triple knockout mutant compared to the Ppst reference strain. The single cwlO knockout also increased γ‐PGA production, while the other single knockouts of ggt and pgdS showed no impact. Partial depolymerisation of γ‐PGA occurred despite the triple knockout. The online measured data are confirmed with offline measurements. The online viscosity system directly reflects γ‐PGA synthesis, γ‐PGA depolymerisation, and changes in the molecular weight. Thus, the ViMOS has great potential to rapidly gain detailed and reliable information about new strains and cultivation conditions. The broadened knowledge will facilitate the further optimization of γ‐PGA production.


| INTRODUCTION
Poly-γ-glutamic acid (γ-PGA) is an anionic, natural biopolymer made from D-and L-glutamic acid units linked between the α-amino and γ-carboxylic acid groups. 1-3 γ-PGA is water-soluble, biodegradable, edible, non-toxic and environmentally friendly. The polymer has different lengths depending on the strain and production conditions used, resulting in a broad spectrum of molecular weights ranging from 10 to more than 2000 kDa. 3,4 These properties make γ-PGA a suitable substance for a variety of applications in multiple fields. For instance, there are reports about γ-PGA use in the food industry, water treatment, agriculture, and medical industry. [1][2][3][4][5][6][7] γ-PGA can be produced through chemical synthesis, peptide synthesis, biotransformation, and microbial fermentation. 8 The properties of γ-PGA strongly depend on its variable molecular size, molecular weight distribution, and enantiomeric composition. 9,10 γ-PGA-producing Bacillus strains can be divided into two groups.
The first group consists of glutamate-dependent strains that produce γ-PGA when glutamate is supplemented as substrate. This group usually produces γ-PGA in high concentrations but has the significant disadvantage of high production costs due to glutamate as substrate. 2,7 The second group comprises glutamate-independent γ-PGA producers. 1,11 Since the γ-PGA concentration is lower in these strains, improving γ-PGA production in this group is essential.
Although Bacillus subtilis 168 possesses the genes that encode the polyglutamate synthase (pgs) required for γ-PGA formation, it does not naturally produce γ-PGA, as it lacks a functional promoter. 1,12 In this study, the phosphate-starvation inducible promoter P pst was introduced to activate the expression of the pgs genes.
Through the characteristics of the P pst promoter, the separation of the growth phase from the product formation phase is possible. In the growth phase, B. subtilis is cultivated as biocatalyst. One essential element for cell growth of B. subtilis is phosphate. Without phosphate, the cell growth and proliferation stops. In this study, the phosphatestarvation inducible promoter P pst is used for γ-PGA synthesis. The P pst promoter becomes active, as soon as phosphate is exhausted. Therefore, γ-PGA is synthesized, when phosphate is depleted and cell proliferation stops. The phosphate-dependent induction of the promoter allows the automatic decoupling of growth and γ-PGA production. 13 This prevents flow of substrate into cell growth, after sufficient biomass as biocatalyst for γ-PGA synthesis has been generated. This two-stage process strategy is well established, for example, for IPTG induced Escherichia coli production processes. 14 Naturally, γ-PGA is formed by Bacillus strains as an extracellular secondary product and serves as a nutrient store for periods of starvation or as a protective molecule in harsh environments. Therefore, γ-PGA producers usually co-synthesize specific depolymerases to degrade γ-PGA again to make the polymer available as carbon and nitrogen source. The monomers and peptides thus made available can then be metabolized. 15,16 This is an undesirable side effect for industrial production of γ-PGA, as it reduces the γ-PGA yield and results in a broad distribution of the molecular weight and thus the polydispersity. Three known specific γ-PGA depolymerases CwlO, Ggt, and PgdS that exhibit PGA depolymerase activity have been reported. [17][18][19] These enzymes have already been investigated in many studies to achieve high level γ-PGA production and low molecular weight dispersity. 17,20 PgdS and Ggt are enzymes specific for γ-PGA, whereas CwlO is a peptidoglycan hydrolase located in the cell wall, which also can cleave γ-PGA. PgdS is an endo-γ-D-L-glutamyl hydrolase, 21 and Ggt is a γ-glutamyl transpeptidase. 22,23 Combined and single knockouts of the depolymerase genes have already been described in previous studies. 20,[24][25][26][27] These studies refer to different γ-PGA-producing Bacillus strains such as Bacillus amyloliquefaciens, 20,25 Bacillus licheniformis, 26 and B. subtilis. 24,27 Depending on the strain, different results are obtained regarding the effect of knockouts on γ-PGA production. For B. amyloliquefaciens, it could be shown that a double knockout of pgdS and cwlO leads to a higher γ-PGA yield than a triple knockout. 20 However, for B. subtilis 168, only a double knockout of pgdS and ggt has been investigated and ascribed to a higher γ-PGA production. 24,27 Since knockouts have different effects depending on the Bacillus strain and no triple knockout has been investigated for B. subtilis 168, a detailed investigation regarding the effect of the knockout of the three depolymerases is of interest. In this study, B. subtilis 168 with knockout of ggt, pgdS, and cwlO are analyzed as single knockouts and as a triple knockout. 13 The strains used in this study are sporulation-negative derivatives of B. subtilis 168, which increases the culture reproducibility. 28 Previous research on γ-PGA production has been based on timeconsuming experiments with sampling and offline determination of γ-PGA concentration. 29,30 In the field of small scale culture research, the online analysis of cultivation parameters, in particular, has achieved resounding success. Online measurements of the oxygen transfer rate (OTR) with the Respiration Activity MOnitoring System (RAMOS) lead to new insides in shaken cultures. 31,32 For faster and more detailed strain and process development, non-invasive online techniques to monitor γ-PGA formation combined with OTR measurements are necessary to complement laborious manual sampling.
Because γ-PGA is a biopolymer, its synthesis during fermentation increases the culture broth viscosity. 30 This parameter is influenced by γ-PGA concentration as well as molecular weight. Measuring the culture broth viscosity offers the possibility to draw qualitative conclusions about the course of product formation. Hence, online measurement of the viscosity during cultivation is a promising approach to investigate γ-PGA synthesis in combination with offline analysis.
For online γ-PGA analysis via culture broth's viscosity, two methods have been reported so far. One uses the power input measurement during the cultivation, which allows the calculation of the viscosity. 33 The other one is the recently published Viscosity Monitoring Online System (ViMOS), which optically measures the cultivation broth viscosity 34 and has already been used to analyze γ-PGA production. 35 The power input measurement method is based on the effect of viscosity changes on the power input in unbaffled shake flasks. 36,37 An increase of fermentation broth viscosity results in an increase of the power input. Thus, detection of the power input by a torque sensor integrated into the drive of a shaker allows for calculating the broth viscosity. 36 A disadvantage of the reported method is its limited throughput, as only one single parameter set can be performed at a time. However, the optical online viscosity measurement system (ViMOS) used in this study allows the parallel, non-invasive viscosity measurement in eight shake flasks on one device. This device provides a contact-free optical measurement and is also suitable for non-Newtonian liquids like microbial culture broths. 34 The measuring principle is based on the defined rotating liquid position in shake flasks relative to the direction of the centrifugal acceleration, which depends on the viscosity. The bulk of the fluid moves along the hydrophilic shake flask wall. With increasing viscosity, the liquid friction and the friction between the liquid and the shake flask wall increase, which results in a shift of the angular position of the fluid relative to the direction of the centrifugal acceleration. That change of the angular position of the liquid's leading edge can be optically detected with a method based on transmitted light with an infrared sensor. Each angular position can be assigned to an apparent shear ratedependent viscosity through previous calibration with viscous model fluids. 34 For calibration, model fluids with various viscosities are measured with a reference cone plate rheometer and the ViMOS in shake flasks. An essential aspect of the calibration is the shear rate. The shear rate in shake flasks depends on shaking parameters like, for example, shaking frequency, filling volume, and shake flask diameter, and on the properties of the fluid, the consistency factor K, and the flow behavior index m. 38 The shear rate can be calculated by the correlation of Giese et al. and was shown to be applicable for the calibration of the online viscosity measurement system. 34 This study focuses on analyzing the newly generated B. subtilis 168 derivatives including strains with the phosphate-starvation inducible promoter P pst and sporulation knockout Δspo and B. subtilis 168 strains with knockout of the single genes ggt, pgdS, and cwlO and triple knockout. First the B. subtilis 168 P pst Δspo is analyzed in detail to understand and control the promoter system and to test the ViMOS with this new γ-PGA production strain. Subsequently, the knockout strains are analyzed with the aim to reduce enzymatic γ-PGA hydrolysis and thus improve γ-PGA production. For the online monitoring of γ-PGA production and degradation during the fermentation process the ViMOS is used. Additionally, the OTR is measured by the RAMOS technique and several offline analytics for the determination of side products, substrate amount in the medium, and γ-PGA concentration and molecular weight are applied. The combination of these online and offline techniques demonstrates the possibilities of online measurements for the analysis and control of microbial synthesis of viscous products throughout the fermentation process.
This study introduces the ViMOS technology as a new analytical tool for process development of γ-PGA production. The feasibility of the ViMOS for viscous fermentation processes is shown by analyzing the potential of the phosphate-starvation promoter P pst and the depolymerase deletion strains.

| Medium composition
Cultivations to prepare cryogenic stocks and the first preculture of the experiments were performed in lysogeny broth (LB) consisting of 10 g/L peptone, 5 g/L yeast extract, and 5 g/L NaCl. There are two commonly used NaCl concentrations for LB medium. One with 10 g/L NaCl and one with 5 g/L NaCl. In this study, the variant with 5 g/L NaCl

| Bacterial strains and strain development
The plasmids used are listed in Table 1, and primers are listed in Table S1.
All cloning steps were carried out in E. coli DH5α. The recombinasepositive E. coli strain JM101 was used to obtain plasmids for the transformation of B. subtilis. For plasmid construction and counter selection, all strains were cultivated at 37 C in LB medium containing 100 μg/ml spectinomycin or 0.5% (w/v) mannose as needed. 39 All developed strains were stored as cryogenic cultures at À80 C.
The developed Bacillus strains presented in this study and the E. coli strains used for plasmid construction and cloning are listed in  Table 2. B. subtilis IIG-Bs2 was used as a platform for genetic modification. For further strain development, sporulation-related genes were knocked out in B. subtilis Δspo with plasmid pBs-8 to allow for reproducible growth. 28 The γ-PGA production with the native B. subtilis pgs operon was enabled by integrating the promoter P pst in B. subtilis Δspo with plasmid pBs-4. To investigate the influence of depolymerases on γ-PGA production and degradation, the three genes ggt, pgdS and cwlO, encoding for depolymerases, were knocked out using plasmids pBs-5, pBs-3, and pBs-12, respectively. Additionally, the performance of a triple depolymerase knockout mutant B. subtilis P pst Δspo ΔpgdS Δggt ΔcwlO was investigated. All genome editing steps were performed using the previously published markerless gene deletion system. 39 For maintenance, strains were plated on LB-agar plates and incubated for 16 h. Next, an LB liquid culture was started with a single colony from an agar plate and grown until the late exponential phase.
Then, a glycerol stock with 30% (w/v) glycerol was prepared and stored at À80 C. The glycerol stocks were used as starting cultures for the performed cultivation experiments.

| Shake flask cultivation
For each experimental culture, a cryogenic culture was thawed and used as inoculum of a culture in 250 ml unbaffled shake flasks. The cultivation parameters were a filling volume (V L ) of 10 mL or 30 ml, a shaking frequency (n) of 250 or 350 rpm, a shaking diameter (d 0 ) of 50 mm, a temperature (T) of 37 C and a starting OD at 600 nm (OD start ) of 0.1 in modified glucose-based V3 MOPS mineral medium.
A two-stage preculture was performed under the same cultivation conditions, as all the following experimental cultures. The first preculture was grown on LB medium, and the second preculture was grown on a glucose minimal medium. For the main culture a master mix of the modified V3 MOPS medium and cells from the second preculture was prepared and equally distributed between all shake flasks. The shake flasks for offline sampling and online analysis were cultivated under identical conditions, to ensure comparable growth conditions.

| Oxygen transfer rate (OTR)
The OTR of the cultivations was online monitored using an in-house manufactured V RAMOS device. 31  where. 34 Using calibration, calculating the corresponding viscosities from the different shifts during the cultivation is possible.

| Offline measured viscosity
In addition to online viscosity measurement, the viscosity was offline measured, to validate the online system for viscosity measurements with γ-PGA cultivations. Viscosity was measured offline with a Physica MCR 301 rheometer (Anton Paar Germany GmbH, Ostfildern-Scharnhausen, Germany) in a range of shear rates between 100 and 5000 s À1 with a cone-plate measuring system from Anton Paar (cone  shear rates were calculated, as published before, for determining the effective viscosity in shake flasks with the offline measured viscosity data. 38

| Cetyltrimethylammonium bromide assay
The γ-PGA concentration was determined with the photometric cetyltrimethylammonium bromide (CTAB) assay. 40 The positively charged CTAB reagent binds to the negatively charged γ-PGA and forms a water-insoluble complex that increases the suspension's turbidity.
Samples of 1 ml from the cultivations were centrifuged for 10 min at

| Gel permeation chromatography
To determine the mass average molecular weight (Mw) and the concentration of γ-PGA, gel permeation chromatography (GPC) was performed. 3 To prepare samples for GPC analysis, culture broth was

| Osmolality
The osmolality, required to calculate the maximum oxygen transfer capacity (OTR max ), was determined with a freezing point osmometer (Osmomat 3000, Gonotec, Berlin, Germany).

| Evaporation
The evaporation of water from the culture broth during the cultivation was quantified with two offline shake flasks, by measuring the weight loss of the flasks over time. The results of HPLC and GPC analysis as well as of the CTAB and phosphate assay were corrected by the calculated evaporation factor.

| Influence of various phosphate concentrations on viscosity and OTR
The B. subtilis strains presented in this study produce γ-PGA under the control of the P pst promoter, hence under phosphate limitation. 13 The P pst promoter activity was investigated by cultivations of   13 The viscosities in group 2 show the same course as in group 1, but reach slightly higher values between 3.5 and 4.5 mPa s after 16 h. Although γ-PGA production starts due to phosphate limitation, the viscosity remains low (Figure 1b), because there is potentially insufficient glucose left for a significant γ-PGA production.  Figure 1, the curves can be assigned to three groups. Initially, the OTR and online viscosity curves for 100% and 50% phosphate (group 1) are similar to those in Figure 1. After an exponential growth phase until 8 h, the OTR max of 19.1 mmol/L/h is reached. This is in very good agreement with the calculations according to Meier et al. 42 The OTR slightly increases for another 6 h. Afterwards, it drops sharply due to glucose depletion. 31 The viscosity, shown in Figure 2b, increases only from 2.5 to 4.5 mPa s for the 100% and 50% batches. For group 2 with 25% and 20% phosphate in Figure 2a, similar high OTR values are achieved as for group 1. Viscosities increase to 6 mPa s for the 20% cultivation and are thus slightly higher than for the 100%, 50%, and 25% experiments. For group 3 with 10% and 5% in Figure 2, similar courses of the OTR and viscosity as in Figure 1 can be seen, as the overall OTR becomes lower and  Figure 2) than for the oxygen-unlimited cultivation with a maximum of 7 mPa s ( Figure 1). This may be due to an increased γ-PGA production under oxygen-limited cultivation conditions. 30 Possibly, it also may depend on the higher molecular weight of the produced γ-PGA induced by the changed cultivation conditions. 1,30,44 It could be shown in this section that γ-PGA is produced in a glucose minimal medium by B. subtilis under the control of the P pst promoter, which can be monitored by online viscosity measurement. As higher viscosities indicate higher γ-PGA formation, the oxygen-limited cultivation conditions were used for further experiments.

| Increasing glucose concentration
To produce more biomass as biocatalyst and to ensure sufficient carbon supply for γ-PGA production, the glucose concentration was increased. Figure 3a shows the OTR for cultivations with 25% phosphate and glucose concentrations of 20, 40, 60, and 80 g/L. Figure 3b displays the corresponding online viscosities. The maximum oxygen transfer capacity (OTR max ) decreases with higher glucose concentrations. This solely is a physicochemical, not a biological effect. With increasing concentrations of glucose the oxygen solubility and the diffusion coefficient and, thus, the mass transfer coefficient is reduced.
The combination of these effects can be correlated to the osmotic pressure, which is easy to measure. 42 After 10 h, due to oxygen limitation under the given fermentation  42 The cultivations were carried out in duplicates in three experimental runs (run 1: 100%, 50%, run 2: 25%, 10%, run 3: 20%, 10%, 5%). For clarity, only every ninth (a) and twentieth (b) measuring point is marked as a symbol. For clarity, average values are shown. The original complete dataset can be found in Figure S1 the viscosity increases until the OTR curve drops sharply. At this point, the glucose is presumably metabolized. 31 For 60 and 80 g/L, the viscosity increases until the OTR curve starts to steadily decrease. Phosphate is already depleted, and glucose is still available. Therefore, the slight and steady decrease after the plateau of the OTR for 60 and 80 g/L could indicate that before glucose exhaustion another substrate necessary for γ-PGA synthesis becomes limiting. It is assumed that the highest viscosities reached with 80 g/L glucose indicate the most elevated γ-PGA production.
However, due to the higher osmotic pressure at a concentration of 80 g/L glucose and the presumed secondary substrate limitation, further experiments are carried out containing 60 g/L glucose.

| Promoter sensitivity to phosphate
To support the previous findings and to analyze in detail the P pst promoter, an experiment with offline sampling at a cultivation with 60 g/L glucose in the modified V3 medium is shown in Figure 4. The P pst promoter naturally regulates the expression of the pst operon in   34 The lag phases of cultures with higher glucose concentrations are longer due to increased osmotic pressure. Therefore, the data have been shifted on the X-axis, so that the exponential growth phase of all experiments is superimposed, making the OTR and viscosity data easier to compare. The curves for 40 g/L glucose were shifted by À0.97 h, those for 60 g/L by À1.47 h and those for 80 g/L by À2.47 h. The dotted vertical lines indicate the assumed time of phosphate depletion and start of viscosity increase due to γ-PGA production. The cultivations were carried out in duplicates. For clarity, only every 15th (a) and 8th (b) measuring point is marked as a symbol. For clarity, average values are shown. The complete dataset can be found in Figure S3 B. subtilis. Its gene products are required for high-affinity uptake of inorganic phosphate (Pi) from the environment at micromolar concentrations. 13 Phosphate in the medium represses the P pst promoter and inhibits the expression of the γ-PGA synthetase genes. De-repression of P pst should hence occur only at negligible remaining phosphate concentrations in the cultivation medium. The depletion of phosphate was assumed to be the reason for the increase in viscosity due to γ-PGA production for Figures 1-3. To verify this hypothesis, phosphate concentrations in the medium were measured. The extracellular γ-PGA accumulation by B. subtilis P pst Δspo was analyzed utilizing the CTAB assay and related to the fermentation broth viscosity (Figure 4). The findings of Figure 4 highlight that promoter P pst is tightly regulated and that its repression system is highly affine to P i , thereby allowing de-repression only at concentrations in a low mg/L range. 13 Furthermore, a correlation between γ-PGA production ( Figure 4d) and increased effective viscosity (Figure 4c) could be demonstrated. Ultimately, the use of phosphate starvation-inducible promoter P pst enables decoupling of microbial growth and γ-PGA production. This decoupling may be exploited to establish a cultivation with two distinct phases: an initial microbial growth under non-limiting conditions to reach a sufficient cell density and a succeeding phase of product formation.  Figure S4.
The γ-PGA producing strains with the P pst promoter (Figure 5a) can be divided by their growth and production behavior into two groups. The first group includes the ΔcwlO and the triple gene deletion strain. The second group consists of the B. subtilis P pst Δspo, the  Figure 5f). These also show higher γ-PGA concentrations for group one strains, than the group two strains. These γ-PGA concentrations must be regarded as high for glutamate-independent γ-PGA production with glucose as the only relevant carbon source, compared to other glutamate-independent γ-PGA producers described in literatures. 20,25,46,47 This increased γ-PGA production with the ΔcwlO and the triple KO mutant fits the deviating OTR curves of these two strains. As shown in other studies, γ-PGA production is likely associated with faster growth rates and substrate uptake rates, demonstrating the driven-by-demand concept. 48 Figure S4. The relation of the microbial growth rate and the OTR is shown in the supplementary data section "Calculation of μ max from OTR data." Data were generated in two experimental runs (run 1: wt, P pst Δspo, P pst Δspo ΔpgdS, P pst Δspo Δggt; run 2: P pst Δspo ΔcwlO, P pst Δspo ΔpgdS Δggt ΔcwlO).
In the case of run 2, time data were shifted by À4 h on the x-axis for lag phase adaption. For clarity, only every 15th (a) and 40th (c) measuring point is marked as a symbol are optimized. Conventionally, the so called medium E is used for the production of γ-PGA. But, since medium E does not have a defined pH buffer and is composed of expensive substrates like glutamic acid, 50 modified V3 glucose minimal medium was used instead. It has to be considered that the modified V3 glucose minimal medium with 60 g/L glucose contains just 54.6% of the carbon provided by medium E. This contributes to lower product titers in this work.
The viscosity of all γ-PGA-producing B. subtilis strains decreases after reaching the maximum (Figure 5c,d). Proceeding viscosity decreases indicate that γ-PGA is degraded, despite the KO of the depolymerases. This hypothesis is confirmed by the molecular weight distribution (Figure 5h), which shows that γ-PGA is depolymerized in all strains, since the molecular weight decreases steadily after 16 hthe start of γ-PGA synthesis. This decrease in the molecular weight may also be due to D/L-endopeptidases other than CwlO and proteases that degrade γ-PGA non-specifically. 51 Mitsui et al. 17 investigated whether the cell wall hydrolases LytE, LytF, and CwlS influence γ-PGA production. Although no effect of KO mutants on γ-PGA concentration was found, it is still possible that γ-PGA is degraded. In particular, the gene products of cwlO and lytE share the same activity.
Therefore, it is likely that LytE can also cleave γ-PGA. In Figure  Furthermore, another γ-PGA specific hydrolase encoded through gene pghC was recently identified, which could be responsible for the remaining γ-PGA degradation activity. 54,55 To rule out if this hydrolase is responsible for the ongoing degradation of γ-PGA, KO derivatives of it remain to be investigated in further studies.
Although the cwlO and the triple KO strain show the strongest decrease of molecular weight from initially 2 to 0.5 MDa during cultivation, these strains reach the highest γ-PGA production (Figure 5f,h).
γ-PGA concentration increased from 3.4 g/L with the B. subtilis P pst Δspo up to 6.2 g/L with the triple KO mutant (Figure 5f). Despite the depolymerization, γ-PGA concentration remains at the same level in all strains or drops only slightly towards the end of the cultivation.
This phenomenon of dropping molecular weight and viscosity, but rising or constant γ-PGA concentrations was also made earlier. 33 The reason for this observation could be that γ-PGA is depolymerized, yet primarily split into smaller chains, but not into monomers. 21,33 This means that the total concentration of γ-PGA remains constant, while the molecular weight decreases.
The viscosity is not only influenced by the molecular weight, but also by the γ-PGA concentration and the pH value. 33,56 Therefore, there may be deviations between the molecular weight changes and viscosity changes during the cultivation. For example, the viscosity may continue to increase as more γ-PGA is produced. However, the average molecular weight may already decrease, due to depolymerization or formation of shorter polymers. It can be seen in Figure 5c that the viscosity decreases most rapidly for the two B. subtilis strains P pst Δspo ΔpgdS, Δggt ΔcwlO and P pst Δspo ΔcwlO after 30 h of cultivation. In addition, the molecular weight of these two strains (Figure 5h) also decreases most rapidly.
One additional effect that might cause the dropping viscosity in Figure 5c,d is a conformational change of the γ-PGA due to a changing pH. It was shown, that γ-PGA in solutions with a pH of ≤7 starts to coil, generating smaller particles. 56 Thus, the polymer molecules get a smaller hydrodynamic radius. This would result in a lower viscosity and an apparently smaller molecular weight, as the GPC separates the particles by its size. This explanation would also fit to the results shown in

| CONCLUSION
This work demonstrates the applicability of a recently developed optical online viscosity measurement system, ViMOS, to determine γ-PGA production, using multiple B. subtilis strains equipped with the P pst promoter. 34 The ViMOS offers the possibility for parallel online viscosity measurements and, therefore, higher throughput in microbial cultivationbased polymer research. P pst was demonstrated to be de-repressed upon phosphate depletion from the cultivation medium. After identifying suitable cultivation parameters using the ViMOS and RAMOS, depolymerase Overall, only small γ-PGA titers are obtained under the influence of the P pst promoter compared to other production strains like B. licheniformis ATCC 9945. 33 However, considering the use of the P pst promoter and glucose as the only C-source, the γ-PGA concentrations achieved can be compared with other glutamate-independent γ-PGA producers. These reach γ-PGA concentrations between 4 and 9 g/L γ-PGA. 20,25 One possibility to further increase the titer would be to perform a phosphate limited fed-batch experiment at low phosphate concentrations. It may be possible to enhance the metabolism of the microorganisms through the maintenance of a low phosphate concentration, allowing the investigation of the influence of the phosphate concentration on the γ-PGA synthesis. The aim of this approach would be to increase product formation. It was shown that by using the P pst promoter, γ-PGA production can be controlled by phosphate concentration. The control over the promoter activity makes it attractive for further studies, as growth and production can be separated. For further and direct evidence of the promoter sensitivity to phosphate, quantitative RT-PCR will be applied in future studies. Furthermore, it should be investigated, whether nitrogen is present in sufficient quantities as the findings of Figure 3 indicate another substrate limitation, resulting in limited γ-PGA production.
Only the triple KO and the ΔcwlO KO strain, deficient in peptidoglycan hydrolase CwlO, showed an increased γ-PGA production. The