The research group of Prof. Deppenmeier (Rheinische Friedrich-Wilhelms-Universität, Bonn, Germany) kindly provided the strain Phocaeicola vulgatus DSM 1447, obtained from the German Collection of Microorganisms and Cell Cultures (DSMZ, Braunschweig, Germany). Brain heart infusion medium (BHI) for cryogenic stocks was acquired as BD Difco™ (Thermo Fisher, Waltham, USA). BHI powder contained: 7.7 g L− 1 calf brain extract, 9.8 g L− 1 beef heart extract, 10 g L− 1 protease peptone, 2 g L− 1 dextrose, 5 g L− 1 sodium chloride, and 2.5 g L− 1 disodium phosphate, dissolved in deionized water. An active growing BHI culture was used to prepare cryogenic stocks after 24 h of cultivation by mixing 50 vol% culture broth with 50 vol% anaerobic sucrose solution (500 g L− 1) and freezing 1.8 mL aliquots at -80 °C. For all main and precultures, a defined minimal medium with glucose (DMM-G) was used. DMM-G composition was based on Varel and Bryant [41] and Lück and Deppenmeier [10] with 3-(N-morpholino)propanesulfonic acid (MOPS) buffer instead of bicarbonate buffer. If not stated otherwise, DMM-G medium components were obtained from Carl Roth (Karlsruhe, Germany). The medium consisted of 13 individual stock solutions: Base components (pH 7.4), glucose, calcium chloride, magnesium chloride, iron(II) sulfate, SL6-trace elements, Wolin’s vitamin solution, butyrate, vitamin K1, hemin, resazurin (Thermo Fisher, Waltham, USA), L-cysteine hydrochloride, and MOPS buffer (pH 7.4). Stock solutions were stored separately, as premature mixing would have caused precipitation. The base components stock comprised ammonium chloride, dipotassium phosphate, monopotassium phosphate, and sodium chloride. The SL6-trace elements included boric acid, cobalt(II)chloride hexahydrate, copper(II)chloride dihydrate, manganese(II)chloride tetrahydrate (Merck, Darmstadt, Germany), nickel(II)chloride, sodium molybdate dihydrate and zinc sulfate heptahydrate (Merck, Darmstadt, Germany) and were set to pH 7.4 with 5 M sodium hydroxide. The Wolin’s vitamin stock solution contained α-lipoic acid, biotin, folate (Sigma Aldrich, St. Louis, USA), nicotinamide, p-aminobenzoic acid (Sigma Aldrich, St. Louis, USA), pantothenic acid (AppliChem, Darmstadt, Germany), pyridoxine hydrochloride (Sigma Aldrich, St. Louis, USA), riboflavin (Sigma Aldrich, St. Louis, USA), thiamine hydrochloride and vitamin B12. Table S1 lists the final concentrations of all components in the DMM-G medium. Base components, glucose, calcium chloride, magnesium chloride, iron(II) sulfate, and SL6-trace elements stocks were sterilized at 121 °C for 20 min. The remaining heat-sensitive stock solutions were sterile-filtered with 0.22 μm polyethersulfone filters (Merck, Darmstadt, Germany). To prevent premature oxidation, reducing agent L-cysteine was sterile-filtered and stored anaerobically in a serum bottle with a nitrogen atmosphere. Wolin’s vitamin solution, vitamin K1, hemin, and resazurin stock solutions were stored light-protected at 4 °C after sterilization. All other stock solutions were stored at room temperature.

Precultures were grown in serum bottles with a total volume of 250 mL. The serum bottles were filled with 50 mL DMM-G medium and sealed gas-tight with a rubber stopper and clamp. Afterward, the serum bottles were gassed with N2 for 20 min to ensure an anaerobic atmosphere. In the next step, CO2 was added to the serum bottles with a sterile syringe to obtain a CO2 headspace concentration of 10 vol%. Afterward, 0.1 mL L-cysteine solution was added as a reducing agent, and in the final step, the medium was inoculated with 500 μL cryogenic culture, both with a sterile syringe. The serum bottles were inoculated in a temperature-controlled shaker for 24 h at 37 °C with a shaking diameter of 50 mm and a shaking frequency of 100 rpm. The main experiments were performed in a RAMOS device designed by Anderlei and Büchs [27]. The RAMOS is a non-invasive online monitoring device for measuring CO2, O2, and pressure for up to eight shake flasks. Anderlei and Büchs [27], Anderlei et al. [28], and Munch et al. [26] provide a schematic overview of the RAMOS setup and gas measurement phases as well as the calculation of the carbon dioxide transfer rate (CTR), oxygen transfer rate (OTR) and total gas transfer rate (TGTR). Measurement of the increase of produced gases is conducted with pressure sensors (26PCA, Honeywell, Charlotte, USA) and infrared carbon dioxide sensors (MSH-P − CO2, 126 Dynament, Mansfield, UK). The RAMOS device is a proven system and has already been operated with syngas [42] or ethylene [43] in the ingas. As the gas measurement phases needed to be adapted to the specific microorganism, time and gas flows were set for both CO2 and O2 experiments as follows: 20 min measurement phase without gas flow, 2.38 min high gas flow rate at 22.5 mL min− 1, and 40 min low gas flow rate at 10 mL min− 1. Before inserting the shake flasks in the RAMOS device, they were filled with 45 mL sterile DMM-G medium and gassed overnight with the respective cultivation gas at 37 °C in a shaker (ISF1-X, Adolf Kühner AG, Birsfelden, Switzerland) at 100 rpm, with a shaking diameter of 50 mm. The system was tested for gas tightness to ensure anaerobic conditions and to prevent false gas measurements. As a reducing agent, 0.1 mL L-cysteine was inserted with a sterile syringe into each flask before inoculation with 5 mL preculture. Initial samples were drawn after inoculation, and final samples at the end of the cultivation.

The gas mixing system consists of up to four mass flow controllers (MFCs) and one control unit, which can be connected to the RAMOS. Therefore, the signal from the RAMOS controls the gas mixing system, to switch between the aforementioned different gas measurement phases.

Schematic illustration of the experimental setup of the gas mixing system. Change of the (a) CO2 or (b) O2 concentration in the gas supply. In case of (b), the dilution of N2 and CO2 by O2 remains very low. Four mass flow controllers (MFC) were used with following ranges, for (a): MFC 1 & 3: 50–500 mL/min (calibrated with N2), MFC 2: 5–50 mL/min (calibrated with O2), MFC 4: 0.5-5 mL/min (calibrated with N2) and for (b): MFC 1: 5–50 mL/min (calibrated with O2), MFC 2 & 3: 50–500 mL/min (calibrated with N2) and MFC 4: 2–20 mL/min (calibrated with N2). This setup was chosen, as experiments at two different gas compositions can be performed with four shake flasks each

The schematic setup of the gas mixing system with gas supply lines can be found in Fig. 5a for different CO2 concentrations and Fig. 5b for different O2 concentrations. The setup was designed so that four shake flasks within the RAMOS can be operated with one gas concentration and the other four with a second gas concentration. After adjusting the gas supply lines, the gas flows were set prior to the experiments. Desired gas concentrations were configured as a percentage of the total maximum flow of the MFC on the control unit. Afterward, the settings of the MFCs were tested by measuring the total flow from the gas mixing system with a gas flow calibrator, Defender 530 + L (Mesa Laboratories, Inc., Lakewood, USA). Before the experiment, a calibration curve was created for the CO2 and O2 sensors within the RAMOS device. With the help of the calibration curve, the concentrations set by the gas mixing system of CO2 and O2 were checked and, if necessary, adjusted.

Besides CO2, also H2 is produced. As no other gases are formed, the hydrogen transfer rate (HTR) was calculated by subtracting the CTR from the TGTR.

Initial and final samples were collected and directly used for OD600nm measurement at a wavelength of 600 nm with a Genesys 20 spectrophotometer (Thermo Scientific, Germany). Samples were diluted with 9 g L− 1 NaCl. To correlate the optical density and CDW, the equation \(CDW = 0.563 \cdot O}\), derived in [44, in revision] for P. vulgatus, was used. Samples not used for optical density measurement were centrifuged at 18,000 rpm for 5 min. The supernatant was used for HPLC and pH measurement. The pH was measured with a pH electrode (Mettler-Toledo, Columbus, USA). The remaining sample supernatant was stored at -80 °C for further HPLC analysis. Therefore, samples were thawed and filtered with 0.2 μm cellulose acetate filters (Merck, Darmstadt, Germany). The SCFAs, acetate, succinate, lactate, propionate, formate, and remaining glucose were measured by HPLC. The HPLC device (Dionex, Sunnyvale, USA) was equipped with an organic acid resin column of 300 × 8 mm dimensions (CS-Chromatography, Langerwehe, Germany) and set to 60 °C. As an eluent, 5 mM H2SO4 at a flow rate of 0.8 mL min−1 was applied. UV/VIS and a refractive index detector were used during HPLC measurement.

Carbon balances were calculated for all experiments with the following Eq. 1:

$$Carbo}\left[ }} \right] = \frac}\left[ - \right]}}\left[ }} \right]}} \cdot \left[ } \right]$$

(1)

Where X is the specific compound, c is the concentration [g L− 1], MX is the molar mass of the specific compound [g mol− 1], Carbon moleculesin X is the number of carbon atoms in the specific compound [-], and Carbonin X is the molar carbon concentration for the compound [mmol L− 1].

The compounds glucose, acetate, lactate, succinate, propionate, formate, CO2, and biomass of every sample were considered. Initial and final concentrations of glucose, acetate, lactate, succinate, propionate, and formate were measured by HPLC. The microbial biomass of P. vulgatus cells was based on data from Franke and Deppenmeier [21] of P. copri microbial biomass. Molar carbon from CO2 was calculated from the CTR integral based on equations in Munch et al. [26]. First, the volumetric molar carbon [mmol L− 1] for each compound was calculated, and then the values were combined to obtain the total volumetric molar carbon content for every sample. Finally, to achieve relative values for the carbon content of the compounds, the molar carbon value was divided by the total carbon of the sample, as shown in Eq. 2:

$$Carbo}\left[ \% \right] = \frac}\left[ }} \right]}}}\left[ }} \right]}}$$

(2)

Where Sample n is designated to a specific sample number in a specific experiment, Carbonin X, Sample n is the volumetric molar carbon of the specific compound in Sample n [mmol L− 1], and Total CarbonSample n is the sum of all carbon in this Sample n [mmol L− 1].

All graphs were created with OriginPro® version 2021 from OriginLab Corporation (Massachusetts, USA).

Statistical Analyses were performed in order to assess the influence of different oxygen concentrations on different cultivation parameters by Mann-Whitney-U-Test using OriginPro® version 2021 from OriginLab Corporation (Massachusetts, USA). Final OD600nm and final pH values were split in two groups for the statistical analysis, low (0-0.7 vol%) and high oxygen concentration levels (1.3–2.5 vol%). To determine if the distributions between both groups differed, the Kolmogorov-Smirnov-Test was conducted.