Chemicals and standards
Chemicals were mainly obtained from Carl Roth GmbH & Co. KG (Karlsruhe, Germany). All chemicals were of analytical grade, if not indicated otherwise. Reference standards for HPTLC measurements for surfactin (≥ 98% purity) and glucose (≥ 99.5% purity) were obtained from Sigma–Aldrich Laborchemikalien GmbH (Seelze, Germany). Azocasein was obtained from Megazyme Ltd. (Wicklow, Ireland). The unmodified ComX peptide backbone (ADPITRQWGD, ≥ 99% purity) was obtained from Thermo Fisher Scientific (Rockford, USA).
Microorganism and strain maintenance
The microorganisms used in this study are listed in Additional file 1: Table S1. B. subtilis wild-type strain DSM 10 T was mainly used for cultivation experiments and strain CT2 for ComX pheromone bioactivity assay. For storage at − 80 °C, cells were cultivated to exponential growth phase and glycerol stocks in lysogeny broth (LB) were prepared containing 15–20% (v/v) glycerol.
Construction of mutant strains
Genetic engineering was performed using standard molecular techniques (Harwood and Cutting 1990). Primers (Eurofins Genomics Germany GmbH, Ebersberg, Germany) and plasmids used in this study are listed in Additional file 1: Tables S2 and S3, respectively. For ComX pheromone bioactivity assay, a reporter strain, namely CT2 was engineered. Therefore, a chromosomal integration of PsrfA-lacZ fusion was introduced into the amyE locus of BKK31700 (Koo et al. 2017), carrying a deletion of comX (∆comX::kan), using the plasmid pKAM446 (Hoffmann et al. 2021a). The mutant strains were selected on LB agar plates supplemented with kanamycin (5 µg/mL), spectinomycin (100 µg/mL) or erythromycin (5 µg/mL), respectively. All plates were incubated at 37 °C. Successful integration disrupted the amyE gene, resulting in a loss of amylase activity. This was confirmed with LB agar plates that were supplemented with 1% (w/v) starch and stained with Lugol's iodine. The genotype verification was performed by PCR (peqSTAR 96X VWR GmbH, Darmstadt, Germany) using Q5® Hot Start High-Fidelity DNA Polymerase (New England Biolabs GmbH, Frankfurt am Main, Germany). Sanger sequencing (Eurofins Genomics Germany GmbH, Ebersberg, Germany) revealed a correct chromosomal integration without any point mutations.
Media
LB medium was prepared containing 10 g/L tryptone, 10 g/L NaCl and 5 g/L yeast extract. For LB agar plates, LB medium was supplemented with 12.5 g/L bacteriological agar (Bertani 1951). A mineral salt medium (MSM) for enhanced surfactin production was prepared, which is based on the Cooper medium (Cooper et al. 1981) and further improved by Willenbacher et al. (2015). Varying glucose concentrations of 8, 20, 40 and 60 g/L were employed. The buffer concentration was 0.07 M (0.03 M KH2PO4 and 0.04 M Na2HPO4) for shake flask cultivation and 0.01 M (4.29 × 10−3 M KH2PO4 and 5.71 × 10−3 M Na2HPO4) for bioreactor cultivation (Willenbacher et al. 2014). The pH of the media was adjusted to pH 7.0 for shake flask cultivation prior to sterilization with the autoclave (15 min, 1 bar, 121 °C). In case of tryptophan auxotrophy, sterile filtrated tryptophan (50 µg/mL) was added to the cultivation medium. For cultivation of the reporter strain CT2, the MSM (8 g/L glucose) was supplemented with spectinomycin (100 µg/mL).
Preparation of inoculum cultures
For cultivation experiments, incubation was performed at 120 rpm and 37 °C in an incubator shaker (Newbrunswick™/Innova® 44, Eppendorf AG, Hamburg, Germany), unless otherwise stated. For preculture I, 20 mL LB medium were inoculated with 100 µL of the respective glycerol stock in a 100 mL baffled shake flask and incubated for 15–16 h. Preculture II was performed in the respective MSM used for later cultivation experiments and inoculated with preculture I to a starting OD600 of 0.1. The filling volumes varied, depending on the cultivation experiment. Typically, 50 mL of MSM was used in a 250 mL baffled shake flask for subsequent shake flask cultivations and 200–300 mL in 2000–3000 mL shake flasks for bioreactor cultivations. Preculture II was incubated for 10–15 h to reach exponential phase before inoculating the main culture. Precultures of strain CT2 for bioassay determinations were prepared with minor adjustments, thus allowing easier handling. The incubation times were extended by decreasing the temperature to 30 °C and by inoculating preculture II to an initial OD600 of 0.05. This resulted in incubation times of 24 h for preculture I and 16 h for preculture II.
Cultivation conditions
Shake flask cultivations were carried out in baffled shake flasks. Main cultures were operated with relative filling volumes of 0.1 mL/mL (10%) in respective MSM and inoculated with preculture II to a starting OD600 of 0.1.
Bioreactor cultivations were carried out in 42 L custom-built bioreactors (ZETA GmbH, Graz/Lieboch, Austria) in batch mode with a filling volume of 20 kg. The parameter settings were partially based on a previously published batch process for strain DSM 10 T, as described by Willenbacher et al. (2014; 2015). The dissolved oxygen was set to a minimum of 20% and was consistently regulated, with an initial agitation rate of 300 rpm (Rushton turbine) and an aeration rate of 1.4 L/min, corresponding to 0.07 vvm. Temperature was set to 37 °C and pH was kept constant at 7.0 using 4 M NaOH and 4 M H3PO4. Both a mechanical and a chemical strategy were used to control the intense foaming, as illustrated by Klausmann et al. (2021). A sensor in the headspace of the bioreactor first activated a foam centrifuge set to 2790 rpm. A second sensor in the exhaust line controlled the addition of antifoam agents (Contraspum A4050; Zschimmer & Schwarz GmbH, Lahnstein, Germany and Xiameter® AFE-1520; Dow Silicones Corporation, Midland, USA), which was kept to a minimum (~ 40–60 mL). To prevent blockage of the exhaust filter in the event of potential over-foaming, a 50 L foam trap was connected upstream.
Sampling and sample analysis
Samples for offline measurements were taken regularly starting from the beginning of the cultivation (t0 = 0 h) at intervals between 2 and 8 h. The OD600 was determined immediately (Biochrom WPA CO8000, Biochrom Ltd., Cambridge, UK) and samples were centrifuged for 10 min at 4816 g and 4 °C (Heraeus X3R, Thermo Fisher Scientific GmbH, Braunschweig, Germany). If necessary, centrifugation was repeated until a clear supernatant was obtained, which was then stored at − 20 °C until further analysis. Cell dry weight (CDW) was calculated from OD600 by using a correlation factor of 3.2 ± 0.3 (10.3% RSD), which was determined as described by Geissler et al. (2019a). The cell-free supernatant was analyzed for surfactin, glucose, and ammonium concentrations, as well as peptidase and ComX activity. Thereby, the glucose and surfactin concentrations were determined using a High-Performance Thin-Layer Chromatography (HPTLC) system (CAMAG Chemie-Erzeugnisse und Adsorptionstechnik AG, Muttenz, Switzerland) as described by Geissler et al. (2017, 2019a). For surfactin detection, 1 mL of sampled cell-free supernatant was extracted thrice with 1 mL of chloroform/methanol (2:1; v/v). After evaporation for 40 min at 10 mbar and 40 °C (RVC2-25 Cdplus, Martin Christ Gefriertrocknungsanlagen GmbH, Osterode am Harz, Germany), the sample was resuspended in 1 mL methanol and applied on HPTLC Silica gel 60 plates (Merck KGaA, Darmstadt, Germany). Development was conducted using a mobile phase of chloroform/methanol/water (65:25:4; v/v/v) and surfactin was detected at 195 nm. For glucose detection, no extraction step was required, and the cell-free supernatant could be directly applied on HPTLC Silica gel 60 F254 plates (Merck KGaA, Darmstadt, Germany). Development was conducted using acetonitrile/water (85:15; v/v) as mobile phase. After a derivatization step with diphenylalanine (DPA reagent), glucose was detected at 620 nm. Quantification of ammonium was performed photometrically using the Spectroquant® Ammonium Assay Kit (Cat. No.: 114752, Merck KGaA, Darmstadt, Germany) according to the manufacturer's instructions in 96-well plates with a reduced volume of 5%. Calibration range was defined from 0.05 to 4.0 mg/L for the reduced volume.
ComX pheromone bioactivity assay (ComX bioassay)
Due to the high complexity of ComX purification, the pheromone was measured with a bioassay based on the principle provided by Magnuson et al. (1994). Thereby, reporter strain CT2 (∆comX::kan; PsrfA-lacZ) can be used as an assay reagent and a ComX-dependent expression of lacZ can be determined by measuring the β-galactosidase activity with the Miller assay (Miller 1972). The employed bioassay is further in line with a recently published method, provided by Dogsa et al. (2021). A detailed description of the working principle and further validation experiments are given in the supplementary information. In brief, a main culture of strain CT2 was inoculated with preculture II to a starting OD600 of 0.1 and incubated for 3 h at 120 rpm and 37 °C. 750 µL of the pre-incubated main culture were then mixed with an equal volume of sampled cell-free supernatant and incubated in culture tubes for 3 h (shake flask cultivation) and 5 h (bioreactor cultivation) at 120 rpm and 37 °C. In case of an expected high peptidase activity in the samples, BSA (50 µg/mL) was added to prevent ComX from potential degradation due to extracellular proteases (Magnuson et al. 1994; Spacapan et al. 2018). After measuring the OD600, the samples were used for β-galactosidase assay with means of the Miller assay, according to the protocol described by Hoffmann et al. (2020). The ComX activity is expressedn Miller units (MU) and represents the highest possible induction of PsrfA as a function of the present ComX concentration (Eq. 1).
$${\text{MU}} = 1000 \cdot \frac{{{\text{OD}}_{{420{\text{nm}}}} - \left( {1.75 \cdot {\text{OD}}_{{550{\text{nm}}}} } \right)}}{{{\text{t}} \cdot {\text{v}} \cdot {\text{OD}}_{{600{\text{nm}}}} }}$$
(1)
where t represents the incubation time and v the sample volume. Bnk values were measured using the respective sterile cultivation medium instead of sampled cell-free supernatant. The limit of detection (LOD) and limit of quantification (LOQ) was calculated from 20 samples using Eqs. 2 and 3 (Shrivastava and Gupta 2011).
$${\text{LOD}} = {\overline{\text{m}}}_{{\text{B }}} + 3{\text{S}}_{{\text{B}}} .$$
(2)
$${\text{LOQ}} = {\overline{\text{m}}}_{{\text{B }}} {\text{ + 10S}}_{{\text{B}}}$$
(3)
where \({\stackrel{\mathrm{-}}{\text{m}}}_{\text{B}}\) represents the mean value of the blank and \({\text{S}}_{\text{B}}\) the respective standard deviation. For selection of blank values, a Shapiro–Wilk test was performed, which confirmed a normal distribution, with p ≥ 0.559 for shake flask cultivations and p ≥ 0.200 for bioreactor cultivations. Based on this, values with a z-score ≥ 2.0 were excluded from the calculation. This resulted in an LOD of 23.9 MU and LOQ of 42.7 MU for shake flask cultivations and a LOD of 30.3 MU and a LOQ of 55.4 MU for bioreactor cultivations.
Liquid chromatography-mass spectrometry analysis of ComX pheromone
Cell-free supernatant was adjusted to pH 2 by adding 6 N HCl and samples were incubated for 1 h at 4 °C to precipitate the ComX pheromone. Samples were then centrifuged for 10 min at 21,000 g at room temperature (Centrifuge 5415 D, Eppendorf AG, Hamburg, Germany). The supernatant was discarded, and the pellet was dissolved in 100% methanol. The methanolic solution was centrifuged for 10 min at 21,000 g at room temperature and the supernatant was used for liquid chromatography-mass spectrometry (LC–MS) analysis. Therefore, a 1290 UHPLC system (Agilent Technologies, Waldbronn, Germany) coupled to a Q-Exactive Plus Orbitrap mass spectrometer equipped with a heated electrospray ionization (HESI) source (Thermo Fisher Scientific GmbH, Bremen, Germany) was used. Chromatographic separation of methanolic extracts from B. subtilis strains was performed on a CSH C18 150 mm × 1 mm column (1.7 μm particle size, Waters GmbH, Eschborn, Germany). The column temperature was maintained at 55 °C. In total, 10 µl of each methanolic extract was injected. A synthetic unmodified ComX peptide was used for building a calibration curve. For mobile phase A, 0.1% formic acid in water, and for mobile phase B 0.1% formic acid in acetonitrile was used. A constant flow rate of 0.12 ml/min was used and the gradient elution was performed as follows: 20–35% B from 0 to 3 min, –65% B from 3 to 4 min, 35–100% B from 4 to 13 min, 100% B from 13 to 16 min. The system was returned to initial conditions from 100 to 20% B from 16 to 17 min. The Q-Exactive Plus mass spectrometer was calibrated externally using the manufacturer’s calibration solutions (Pierce™, Thermo Fisher Scientific GmbH, Bremen, Germany). The HESI source was operated in positive ion mode, with a spray voltage of 4.2 kV and an ion transfer capillary temperature of 290 °C. Sweep gas and auxiliary pressure rates were set to 15 and 2, respectively. The S-Lens RF level was 50% and the auxiliary gas heater temperature was 200 °C. A targeted single ion monitoring method including data dependent MS2 scans (tSIM/ddMS2) was used for quantification of the ComX pheromone. The m/z ratios of the + 2 charged precursor ions of the unmodified ComX peptide ([M + 2H]++, m/z 579.78050) and farnesylated ComX peptide (M + 2H]++, m/z 681.87470) were used as predefined target ions. Mass spectra were acquired in SIM mode with an isolation window of 1.6 Da at a resolution of 70,000 FWHM, and an Automatic Gain Control (AGC) target of 3 × 106 and 100 ms maximum ion injection time (MIT). Data dependent MS2 spectra of precursor ions were acquired with a resolution of 17,500 FWHM, an AGC target of 5 × 105, 100 ms MIT and a normalized collision energy of 30. An exemplary extracted ion chromatogram (XIC) and an ESI–MS/MS spectrum can be found in the supplementary material (Additional file 1: Figure S1).
Endopeptidase activity
The activity of extracellular proteases was determined by measuring the endopeptidase activity with the azocasein assay. The method was adapted from Baur et al. (2015), following a measurement principle based on Charney and Tomarelli (1947). To reflect the activity of the peptidases during cultivation as realistically as possible, the parameters were approximated to the cultivation conditions. For this purpose, 5 g/L sulfanilamide azocasein solution was dissolved in MSM buffer of the analyzed cultivation and adjusted to pH 7.0. 100 µL pre-incubated substrate solution (5 min at 40 °C) was added to 100 µL of cell-free supernatant. The mixture was incubated for 1 h at 800 rpm and 37 °C and the reaction was stopped with 20 µL of 2 M TCA. After centrifugation for 10 min at 19,357 g and 4 °C (Centrifuge 5430R, Eppendorf AG, Hamburg, Germany), 150 µL of the solution were transferred to a 96-well microtiter plate. Before the absorption was measured at 450 nm, 50 µL of 1 M NaOH were added. It was verified by means of time-turnover curves (data not shown) that measurements were within linear range and samples were diluted accordingly. For blank measurements, the cell-free supernatant was added after reaction stop using TCA. The volumetric endopeptidase activity [EApeptidase] is defined as the absorbance difference [∆A] between sample and blank at 450 nm per h and mL cell-free supernatant [∆A/(h·mL)].
Degradation experiments
Cell-free supernatant was obtained from bioreactor cultivation with 40 g/L glucose by two centrifugation steps for 30 min at 4816 g and 4 °C (Heraeus X3R, Thermo Fisher Scientific GmbH, Braunschweig, Germany). The time points were selected to represent the exponential phase, the phase around the highest ComX activity, and the stationary phase, around CDWmax, resulting in different peptidase activities at each time point. The supernatant was then filtered using a filter with a retention capacity of 7–12 µm. To prevent further growth, the supernatant was treated with spectinomycin (100 µg/mL), as sterile filtration was not possible, due to immediate filter blockage. Cell-free supernatant was incubated in baffled shake flasks with a relative filling volume of 10%. Incubation was performed in biological duplicates at 120 rpm and 37 °C. Samples were taken after t = 0, 1, 2, 3, 4, 6 and 8 h and analyzed for ComX and peptidase activity and the ComX degradation rate in (MU/h) was calculated from the measured data.
For autodegradation studies, strain DSM 10 T was additionally cultivated in 3000 mL shake flasks using MSM with 40 g/L glucose. The supernatant was collected after 14 h of cultivation, corresponding roughly to the highest peptidase activity in DSM 10 T. The supernatant was prepared as previously described and additionally sterile filtered (2 µm) to prevent growth and incubated for 30 min at 80 °C to inactivate proteases. To mimic bioreactor conditions, the pH was adjusted to 7.0. The incubation was performed as previously described in a biological triplicate. To gather information about the influence of the cell fraction on putative ComX degradation, a ComX deficient strain was cultivated in spent medium. Therefore, B. subtilis BKK31700 (∆comX) and the corresponding wild-type strain B. subtilis DSM 23,778 were both cultivated in MSM, employing 8 g/L glucose. The cells were harvested after 12.5 h by two-fold centrifugation for 15 min at 4816 g and 4 °C (Heraeus X3R, Thermo Fisher Scientific GmbH, Braunschweig, Germany). The cells of strain BKK31700 were washed in between the centrifugation process with sterile saline solution (0.9% w/v) and resuspended in an equal volume of spent filtered medium of strain DSM 23,778. To prolong cultivation time, 4 g/L glucose were added to the culture. The incubation was performed in biological duplicates for 24 h at 120 rpm and 37 °C and samples were taken regularly and analyzed for OD600, ComX and peptidase activity.
Data analysis and process parameters
For evaluation of the cultivations, the biomass yield per substrate YX/S [g/g], the product yield per substrate YP/S [g/g], the product yield per biomass YP/X [g/g] as well as the specific growth rate µ [1/h] and the specific productivity q [g/(g·h)] were determined using Eqs. 4, 5, 6, 7, and 8 (Geissler et al. 2019a). Process parameters were calculated for the respective replicates using absolute values at the time point when at least 90% of the maximum value was exceeded. Hence, YX/S was calculated at X≥90%, YP/S at P≥90% and YP/X and qoverall at P≥90% and X≥90% with t0 used as first time point.
$$Y_{{{\text{X}}/{\text{S}}}} = \frac{{\Delta m_{{\text{X}}} }}{{\Delta m_{{\text{S}}} }}$$
(4)
$$Y_{{{\text{P}}/{\text{S}}}} = \frac{{\Delta m_{{\text{P}}} }}{{\Delta m_{{\text{S}}} }}$$
(5)
$$Y_{{{\text{P}}/{\text{X}}}} = \frac{{\Delta m_{{\text{P}}} }}{{\left( {\frac{{\left( {m_{{{\text{X}}_{1} }} + m_{{{\text{X}}_{2} }} } \right)}}{2}} \right)}}$$
(6)
where X represents the biomass, P the product, here surfactin, and S e substrate, here glucose.
$$\mu = \frac{{\ln \left( {\frac{{m_{{{\text{X}}2}} }}{{m_{{{\text{X}}1}} }}} \right)}}{{\left( {t_{2} - t_{1} } \right)}}$$
(7)
Thereby µmax is defined as the maximum growth rate.
$$q = \frac{{\Delta m_{{\text{P}}} }}{{\left( {\frac{{\left( {m_{{{\text{X}}_{1} }} + m_{{{\text{X}}_{2} }} } \right)}}{2}} \right) \cdot \Delta t}}$$
(8)
The overall specific productivity qoverall determined the entire process, while qmax is defined as the maximum specific productivity.
Modeling platform and software
All cultivation experiments were carried out in biological duplicates or triplicates. For bioreactor cultivations, the offline measurements were additionally carried out in technical duplicates, yielding in at least 4 measurements for each point in time. Mathematic modeling was performed using programming and numeric computing platform MATLAB (The MathWorks Inc., Natick, USA). As described previously (Henkel et al. 2013), all models were implemented as nonlinear differential equations. For simulation of the ordinary differential equations the ODE-solvers “ode23s” and “ode15s” embedded in the MATLAB environment were used. Nonlinear parameter optimizations and fitting were performed using the “fmincon” functions. A least-square error function was assumed for all parameter optimizations. Graphs were generated from obtained results using the scientific graphing and data analysis software Sigma Plot (Systat Software Inc., San Jose, USA), which was also used for statistical analysis. The experimentally recorded carbon and nitrogen data were fitted using a sigmoidal or logistic curve fit with 4 parameters. To correlate specific surfactin productivity qsurfactin to ComX, the experimentally plotted data for biomass, surfactin and ComX activity were fitted using either a sigmoidal curve fit with 4 parameters or an exponential fit with 3 parameters. In this case, the data range was limited up to ComXmax and specific productivity qsurfactin could be calculated from the fitted values. Xcalibur™ software version 4.4.16.14 (Thermo Fisher Scientific, San Jose, USA) was used for data acquisition and data analysis of LC–MS measurements.