Construction of A. woodii [P _bgaL _ldhD_NFP] is described in Mook et al. (2022) while the C. drakei wild-type strain was obtained from the German Collection of Microorganisms (DSMZ 12750). Cultivation medium for both strains was prepared according to Herzog et al. (2022). For C. drakei cultivations, 10 g·L⁻¹ DL-lactate was added as carbon source to the medium. Pre-cultures of A. woodii [P _bgaL _ldhD_NFP] were cultivated heterotrophically using fructose (2 g·L⁻¹) at 30°C in non-agitated anaerobic serum bottles. C. drakei pre-cultures were cultivated under the same conditions with the exception of using a 75:25 ratio (v/v) of lactate and fructose. The pre-cultures were incubated for 30–33 h until reaching an optical density at 600 nm (OD₆₀₀) of 3.3 ± 0.4 for C. drakei and 2.0 ± 0.9 for A. woodii [P _bgaL _ldhD_NFP].

Batch fermentations were carried out in a 2.0 L stirred-tank bioreactor (KSF 2000, Bioengineering AG, Wald, Switzerland) with a working volume of 1.4 L. The medium was sterilized in situ at 121°C for 20 min before each fermentation. Then, the cultivation medium was degassed with N₂ to ensure anaerobic conditions. An oxygen reduction potential (ORP) sensor was installed to ensure an ORP value lower than −280 mV prior to inoculation. After degassing, the reactor was sparged constantly with CO₂ at a gas flow rate of 0.9 L·h⁻¹ with an open tube L-sparger. A. woodii [P _bgaL _ldhD_NFP] cultures were agitated with three Rushton disk turbines (d = 40 mm) at a stirrer speed of 800 rpm (P V⁻¹ = 2.4 W·L⁻¹), while C. drakei culture were agitated at 210 rpm (P V⁻¹ = 0.04 W·L⁻¹). The temperature was controlled at 30°C and pH was measured via a pH sensor and controlled at pH 7.0 ± 0.2 by the addition of a 5 M KOH solution. H₂ for the fermentations with A. woodii [P _bgaL _ldhD_NFP] was supplied by the AiO-electrode via in situ electrolysis. The rod-shaped electrode consists of a platinized titan mesh as working electrode (platin coating thickness d = 2.5 µm, coating density ρ = 50 g·m^-2) on the outside where H₂ is produced and which is in contact with the fermentation medium. The working electrode is separated from the counter electrode by a ceramic separator, creating a counter electrode chamber with an exhaust duct at the top. This enables the produced O₂ to leave the bioreactor without interfering with the cultivation medium. The working electrode surface is 75 cm² while the counter electrode surface is 14 cm² (Utesch and Zeng, 2018). A scheme of the AiO-electrode is shown in the Supplementary Figure S1. The AiO-electrode was operated without reference electrode to maintain a constant current of 600 mA (j = 8 mA·cm⁻², E_cell = 4.2 ± 0.3 V) by a power supply (2231A-30-3; Keithley, Solon, OH, United States). C. drakei cultivations were conducted without the AiO-electrode. The OD₆₀₀ of the pre-culture was used to calculate the necessary inoculation volume to start the fermentations to an OD₆₀₀ of 0.15 for A. woodii [P _bgaL _ldhD_NFP] and 0.2 for C. drakei respectively. For A. woodii [P _bgaL _ldhD_NFP] cultivations, the production of lactate was induced with the addition of 0.3 g·L⁻¹ lactose. Induction was conducted when the batch culture had reached an OD₆₀₀ of 0.48 ± 0.13. The online lactate measurement was carried out with a TRACE C2 Control (TRACE Analytics GmbH, Braunschweig, Germany) using a dialysis probe with a membrane for low lactate concentrations. The automated current adjustment for H₂ production by the AiO-electrode was regulated by a LabVIEW script containing a proportional-integral-derivate controller (PID) algorithm. The process variable lactate concentration (g·L⁻¹) was transmitted with 4-20 mA via an analog digital converter (USB-6001; National Instruments Corp., Austin, TX, United States) to the host computer. The control variable was the electrical current (A) applied to the AiO-electrode by the power supply via RS232 communication. The set point was the desired lactate concentration. The derivative value of the controller was set to 0 while proportional gain was set to 0.01 and the integral time to 2 min with a sampling time of 10 ms. For co-cultivations, A. woodii [P _bgaL _ldhD_NFP] was inoculated first into the bioreactor and grown under the same conditions as in the pure culture experiments (T = 30°C; pH = 7.0; P V⁻¹ = 2.4 W·L⁻¹; F_CO2 = 0.9 L·h⁻¹; I_AiO = 600 mA; V₀ = 1.4 L) until 0.4 g·L⁻¹ of lactate were produced. The automated process control was programmed to switch off the AiO-electrode at this lactate concentration. Subsequently, 90 mL of a C. drakei pre-culture were added to the bioreactor. Cell growth was determined with OD₆₀₀ measurements (photospectrometry), while lactate, acetate, butyrate and formate concentrations were determined via HPLC. The exhaust gas composition and flow were monitored constantly with a mass flow meter (EL-FLOW prestige, Bronkhorst High-Tech B.V., Ruurlo, Netherlands) and a mass spectrometer (Omnistar GDS 300, Pfeiffer Vacuum GmbH, Asslar, Germany). All analytical methods are described in detail in Herzog et al. (2022). The experimental plan of the fermentations conducted in this study is shown in Table 1.

C. drakei was cultivated in a stirred tank reactor as described in Chapter 2.2 with lactate being the sole substrate. The cultivation was reproduced (n = 2) and average values with standard deviation are mentioned in this section. Representative data of only one fermentation are shown in Figure 1 (refer to Supplementary Figure S2 in the additional dataset). C. drakei cells grew with a growth rate of 0.04 h⁻¹ in the first 15 h to an OD₆₀₀ of 0.39 ± 0.12 and subsequently entered stationary phase. This phase lasted 16 h where the average OD₆₀₀ added up to 0.41 ± 0.03. Afterwards, the C. drakei culture entered a second growth phase with µ = 0.04 h⁻¹ which lasted for 20 h reaching an OD₆₀₀ of 0.93 ± 0.08 after 49 h of fermentation time. Subsequently, the culture was stationary for the remaining 17 h of the process at a constant OD₆₀₀ of 0.96 ± 0.04. Lactate concentrations in the medium started decreasing with the start of the fermentation. In the first 21 h, lactate was consumed at a rate of 0.06 g·L⁻¹ h⁻¹, reducing the initial concentration of 10.7 g·L⁻¹ by 11% to 9.5 g·L⁻¹. Then, at the end of the first stationary growth phase, the lactate consumption rate almost doubled to 0.11 g·L⁻¹ h⁻¹ and during the second exponential growth phase it reached a maximum value of 0.24 g·L⁻¹ h⁻¹. After 66 h of fermentation time, lactate concentration decreased to 0.13 ± 0.18 g·L⁻¹. Acetate was produced throughout the fermentation at a constant rate of 0.06 g·L⁻¹ h⁻¹ and reached a final concentration of 4.53 ± 0.53 g·L⁻¹. Butyrate was produced at a similar rate as acetate during the first 15 h. Afterwards, the production rate increased to 0.15 g·L⁻¹ h⁻¹, reaching a peak concentration of 1.57 ± 0.65 g·L⁻¹ after 21 h of fermentation time. In the following 8 h, butyrate concentration decreased to 1.26 ± 0.27 g·L⁻¹ due to reassimilation, before increasing again with a production rate of 0.10 g·L⁻¹ h⁻¹ until 47 h fermentation time. The concentration then increased further to a maximum of 3.14 ± 0.08 g·L⁻¹ after 64 h with a reduced rate of 0.04 g·L⁻¹ h⁻¹. Caproate concentrations were detected in the medium after 23 h of fermentation. The caproate production rate added up to 0.01 g·L⁻¹ h⁻¹ in the first 31 h of fermentation time. Afterwards, the production rate increased to 0.06 g·L⁻¹ h⁻¹ between 39 h and an estimated 58 h of fermentation time. The maximum concentration of 1.56 g·L⁻¹ was measured after 64 h when the culture had already become stationary.

The concept of controlling the lactate production of A. woodii [P _bgaL _ldhD_NFP] by dynamic on-off switching of the AiO-electrode raises the question regarding the impact of a H₂ limited phase on the lactate metabolism of A. woodii [P _bgaL _ldhD_NFP]. Specifically, it is unclear whether the lactate metabolism of A. woodii [P _bgaL _ldhD_NFP] would be halted permanently or if A. woodii [P _bgaL _ldhD_NFP] would resume lactate production when H₂ is present in the medium again. To answer this question, a controlled batch fermentation with the AiO-electrode was conducted where 10 h after induction, the AiO-electrode was manually turned off for an interval of 6 h. It was then turned on again to see whether lactate production is resumed. For comparison, a reference batch fermentation with A. woodii [P _bgaL _ldhD_NFP] where the AiO-electrode was left on for the whole process time was also conducted (refer to Figure 2A). Both cultivations were inoculated from a pre-culture with an OD₆₀₀ of 2.5 ± 0.3 and started growing with the maximum growth rate of 0.07 h⁻¹. The reference fermentation reached the stationary phase after 26 h and stayed at an average OD₆₀₀ of 0.85 ± 0.02 for the rest of the fermentation. The manually controlled cultivation entered stationary phase already after 25 h, at an OD₆₀₀ of 0.62 ± 0.01 and stayed afterwards at an average OD₆₀₀ of 0.64 ± 0.02 until the end of the fermentation. The course of the lactate concentration over time is shown in Figure 2B. The lactate production of the reference fermentation did not start until 6 h after induction. Then, lactate was produced constantly at a maximum rate of 0.05 g·h⁻¹ until 25.5 h after induction, reaching a concentration of 0.60 g·L⁻¹. Subsequently, the lactate production rate decreased to 0.01 g·h⁻¹. The maximum lactate concentration of 0.69 g·L⁻¹ was reached after 66 h of fermentation time. The lactate formation of the controlled fermentation began 2 h after induction with a rate of 0.02 g·h⁻¹ and 8 h later, the production halted at a concentration of 0.10 g·L⁻¹ due to switching off the AiO-electrode. The concentration stayed constant at an average of 0.11 ± 0.01 g·L⁻¹ during the 6 h when no electrolysis was taking place. Afterwards, the lactate concentration increased at a rate of 0.02 g·h⁻¹ for 8 h, reaching a concentration of 0.24 g·L⁻¹. For the last 26 h of the controlled fermentation, the lactate concentration stayed constant at an average of 0.25 ± 0.02 g·L⁻¹. The measured acetate concentration is shown in Figure 2C. The reference fermentation produced acetate constantly at a rate of 0.17 g·h⁻¹ for 46 h. Then, the production rate decreased to 0.04 g·h⁻¹ adding up to a final concentration of 6.8 g·L⁻¹. The acetate concentration of the controlled fermentation increased in the beginning of the fermentation with a production rate of 0.11 g·h⁻¹ until the AiO-electrode was switched off. Afterwards, the acetate concentration stayed constant at an average of 2.26 ± 0.11 g·L⁻¹. In the last 26 h of the process, the acetate concentration increased slowly at a rate of 0.03 g·h⁻¹, adding up to a final value of 3.26 g·L⁻¹. The course of the formate concentration over time is depicted in Figure 2D. In the reference fermentation, almost no formate was accumulated in the first 42 h of the process (average 0.09 ± 0.01 g·L⁻¹). Subsequently, formate concentration rapidly increased with a constant rate of 0.10 g·h⁻¹ adding up to a final concentration of 1.67 g·L⁻¹. The concentration of formate in the controlled cultivation increased at the beginning to an average of 0.51 ± 0.06 g·L⁻¹ until the AiO-electrode was switched off. Then, all formate was consumed during the first 2 h without active electrolysis. Afterwards, formate accumulated at a rate of 0.15 g·h⁻¹ during the 8 h after the AiO-electrode was reactivated. Subsequently, formate production slowed down, adding up to a final value of 1.58 g·L⁻¹.

A co-cultivation of A. woodii [P _bgaL _ldhD_NFP] and C. drakei was conducted to test the automated process control for feasibility. For this, the lactate threshold of the automation was set to 0.35 g·L⁻¹ and C. drakei addition was timed 1.5 h after the process control had turned off the AiO-electrode. The threshold value was chosen to be lower than the average lactate concentration of 0.47 ± 0.11 g·L⁻¹ based on previous data of fermentations with A. woodii [P _bgaL _ldhD_NFP] and the AiO-electrode. The results of the co-cultivation are shown in Figure 3. Cell growth of the A. woodii [P _bgaL _ldhD_NFP] strain increased after about 7 h with a maximum growth rate of 0.1 h⁻¹. At an OD₆₀₀ of 0.5 the cells were induced with 0.3 g·L⁻¹ of lactose for lactate production and afterwards, the cell growth declined to 0.02 h⁻¹. C. drakei was added to the bioreactor at an OD₆₀₀ of 0.65, and afterwards, the combined OD₆₀₀ of both strains kept increasing until reaching a final value of 1.0 after 88 h. After induction the A. woodii [P _bgaL _ldhD_NFP] strain started producing lactate with a maximum formation rate of 0.03 g·h⁻¹. As seen in Figure 3B, the lactate measurement was recalibrated after 38 h due to deviation to a control measurement performed with HPLC. After 42.5 h, the measurement system detected lactate concentrations higher than the threshold and therefore, the AiO-electrode was automatically turned off. Due to residues of lactate in the pre-culture of C. drakei, the concentration increased slightly at the beginning of the co-cultivation phase. Furthermore, the lactate measurement system showed strong fluctuations subsequently which triggered the process control to turn on the AiO-electrode again, after 45.5 h and made a recalibration necessary. Subsequently, lactate concentrations decreased due to its consumption by C. drakei with a rate of 0.02 g·h⁻¹. Caproate concentrations were first detected after 63 h and C. drakei kept producing caproate at a rate of 0.01 g·h⁻¹. Final caproate concentrations added up to 0.08 g·L⁻¹. Acetate was produced by the A. woodii [P _bgaL _ldhD_NFP] strain once it had entered the exponential growth phase reaching 3.8 g·L^−-1 when C. drakei was added. Afterwards, both strains contributed to the increase of acetate throughout the subsequent fermentation. Final acetate concentrations added up to of 7.3 g·L⁻¹. Butyrate was produced only during the co-cultivation phase by C. drakei, reaching 0.6 g·L⁻¹ at the end of the fermentation. As shown in Figure 3D, formate was produced by the A. woodii [P _bgaL _ldhD_NFP] strain and concentrations reached a first peak after 16 h at 0.4 g·L⁻¹. Subsequently, the A. woodii [P _bgaL _ldhD_NFP] strain consumed almost all of the formate before starting to produce again, reaching a second concentration peak at the moment the AiO-electrode was turned off at 0.5 g·L⁻¹. During the period where no H₂ was produced, the A. woodii [P _bgaL _ldhD_NFP] strain consumed parts of the formate and afterwards started producing it a third time, reaching a maximum concentration of 0.7 g·L⁻¹. During the last 20 h of the process, formate was consumed again. The electrical current of the AiO-electrode was maintained constant at 600 mA until the electrode was turned off by the automation. During the co-cultivation of both strains, O₂ started accumulating inside the counter electrode channel, causing an increase in electrical resistance and thus also in electrical voltage. Therefore, the electrical current decreased twice for short intervals, but refilling the counter electrode channel with culture medium resolved the issue. Relevant process performance parameters are listed in Table 2.