Adaptive laboratory evolution increased biofilm formation by Sporomusa ovata through a mutation in galU

Abstract

Acetogenic bacteria are attractive biocatalysts for the conversion of CO₂ with H₂ into acetate, as in gas fermentation. Gas fermentation reactors may benefit from biofilm formation, but attachment by acetogens is often limited. This study indeed found that the acetogen Sporomusa ovata 2663 was mainly planktonic and aimed to increase its biofilm formation through adaptive laboratory evolution. The adaptation strategy consisted of growing S. ovata on plastic (HDPE) carriers in bottles with a H₂:CO₂ headspace and transferring two carriers to a bottle with fresh carriers over eight serial transfers. This procedure resulted in the evolved S. ovata 2663-BF, which had a consistent increased propensity to attach. In heterotrophic growth conditions, four times more cells attached to the bottom of well plates in comparison to the wild type, while the adapted S. ovata produced 1.8 times more extracellular polysaccharides. Moreover, twice as many cells adhered to carriers when grown on H₂:CO₂. This improved attachment, however, did not lead to higher acetate production rates in simple trickle bed reactors, as the experimental setup rather stimulated planktonic growth. Only medium replacement sometimes favored the rate of the adapted strain. Interestingly, the evolved S. ovata had a point mutation in the gene galU, which likely increased the activity of the encoding UDP-glucose pyrophosphorylase, i.e. an enzyme involved in the synthesis of extracellular polysaccharides. Overall, this study demonstrates that cell attachment by S. ovata was increased through adaptive laboratory evolution, offering the prospect of investigating the importance of biofilm formation in biofilm-based gas fermentation reactors.

1. Introduction

Acetogenic bacteria are attractive biocatalysts for the development of CO₂ utilization and energy conversion technologies. Acetogens use CO₂ as a terminal electron acceptor with H₂ as an electron donor for energy conservation via the Wood-Ljungdahl pathway, leading to the formation of mainly acetate [1,2]. The H₂ can be produced from renewable electricity by electrolysis and brought together with CO₂ in an acetogenic bioreactor, as in gas fermentation applications [3]. Alternatively, an electrode (cathode) can deliver H₂ directly into a bioreactor, as in microbial electrosynthesis (MES) [4]. Several studies have described the acetogenic bacterium Sporomusa ovata as a highly interesting biocatalyst for both type of applications, due to its high acetate production rates [[5], [6], [7]].

Several reactor systems for gas fermentation are biofilm-based. For instance, in trickle bed reactors for gas fermentation, H₂ and CO₂ are converted by microorganisms attached to carrier materials present in the gas phase, while a liquid phase trickles over the carriers to provide nutrients [8,9]. Also, membrane biofilm reactors are used for gas fermentation and are based on biofilm formation of microorganisms on a membrane, through which gaseous components diffuse [8,10]. Similarly, in microbial electrosynthesis, microorganisms often attach to the cathode [5,11], which allows the microorganisms to consume H₂ as soon as it is generated by the electrode. Biofilm formation in these systems has the additional advantage that the microorganisms and their catalytic activity remain contained in the bioreactor, while their products can be harvested from the liquid phase by continuous operation. On the other hand, excessive biofilm formation can negatively impact the activity of cells in a biofilm, as the supply of substrates and the removal of products in thick biofilms gets limited by diffusion [8]. Investigations of the role of biofilm formation in those bioreactors thus forms an essential step towards the further development of CO₂ utilization biotechnologies.

Several studies have already reported biofilm formation by acetogenic bacteria, including S. ovata, on various substrates (e.g. plastic, glass, graphite electrodes) (Table S1). Attachment of pure cultures of acetogens was demonstrated in trickle bed and membrane biofilm reactors, as well as on cathodes in MES bioreactors (Table S1). Interestingly, Sporomusa was present in methanogenic biofilms grown on H₂ and CO₂ [12], while S. ovata strain 2663 had more pronounced biofilm formation on cathodes than various other Sporomusa strains [13] (Table S1). Despite these findings, only few studies have specifically placed biofilm formation by acetogens in focus [14], while the molecular mechanisms underlying their attachment remain completely uncharacterized. Furthermore, most existing studies only reported thin biofilms and sparse attachment, leading to incomplete surface coverage (Table S1). Enhancing biofilm formation by acetogens could therefore offer valuable improvements for biofilm-based production biotechnologies. However, it is unclear whether acetogens can spare sufficient energy to synthesize a biofilm matrix consisting of extracellular polymeric substances (EPS) (i.e. extracellular polysaccharides (exoPS), proteins or nucleic acids) [15], as acetogenesis from H₂ and CO₂ yields very little energy [2].

One commonly used strategy to improve a not well-understood phenotypic trait of a microorganism is Adaptive Laboratory Evolution (ALE). By exposing microorganisms to a selection pressure over a prolonged period of time, beneficial (random) mutations are enriched and better adapted microorganisms are obtained [[16], [17], [18]]. ALE has already been applied to various acetogenic bacteria to improve or establish desired characteristics [[19], [20], [21]]. For instance, ALE was used to increase the tolerance to O₂ [22] and methanol of S. ovata [23]. ALE has also been successful in increasing biofilm formation of aerobic microorganisms. For example, ALE enhanced the cathodic autotrophic biofilm growth capabilities of the aerobe Kyrpidia spormanii [24]. In addition, ALE increased biofilm production of the heterotrophic aerobe Burkholderia cenocepacia [25]. The latter study selected for biofilm formation by repeated serial transfers of plastic beads (with adhered cells) to fresh medium with new clean beads. So far, ALE has not been tested to increase biofilm formation of anaerobic bacteria, such as acetogens, while overall the role of acetogenic biofilms in biofilm-based bioreactors remains not well understood.

This study aimed to investigate the native biofilm formation properties of biotechnologically relevant acetogen S. ovata strain 2663, as well as to increase its attachment and biofilm formation capabilities through ALE. Hereto, S. ovata was grown in bottles filled with carriers incompletely covered with liquid medium in an atmosphere containing H₂ and CO₂ (resembling trickle bed reactors). The ALE strategy consisted of transferring two carriers to a bottle with fresh carriers over multiple serial transfers. Attachment, planktonic growth, as well as exoPS production and growth yields, of the resulting biofilm-increased S. ovata 2663-BF were compared with those of the wild type (WT) S. ovata in heterotrophic or autotrophic growth conditions. Furthermore, the performance of S. ovata 2663-BF was compared with that of the wild type in simple trickle bed reactors as an attempt to investigate the importance of biofilm formation in such reactor systems. Finally, insights into the molecular basis of the altered phenotype were obtained through genome comparison.

2. Materials and methods

Strain and culture conditions. Sporomusa ovata DSM-2663 wild type (WT) was purchased from DSMZ, while Sporomusa ovata 2663 BF was obtained through ALE, as described in this study. S. ovata strain 2663 was selected for this study because of its more outspoken biofilm formation in comparison to other Sporomusa strains [13] (Table S1). Both cultures were regularly cultured at 30 °C under a N₂:CO₂ (80:20) atmosphere in an anoxic medium containing per L of medium 0.5 g NH₄Cl, 0.5 g MgSO₄ x 7H₂O, 0.25 g CaCl₂ x 2H₂O, 2.25 g NaCl, 2 mL FeSO₄ x 7H₂O solution (0.1% in 0.1 N H₂SO₄), 0.35 g K₂HPO₄, 0.23 g KH₂PO₄, 4 g NaHCO₃, 0.3 g L-cysteine-HCl x H₂O, 1 mL trace element solution (34), 1 mL selenite-tungstate solution (34) and 10 mL Wolin's vitamin solution (DSMZ 141). Heterotrophic growth was supported by the addition of 6 g L⁻¹ betaine x H₂O and 2 g L⁻¹ yeast extract. No betaine was added to autotrophic growth medium, which had only 0.05 g L⁻¹ yeast extract and the addition of 50 mM 3-(N-morpholino)propane sulfonic acid (MOPS, pH 7). Frozen stocks (10% DMSO) were stored at −80 °C.

Inoculation and replicates. For all experiments, the S. ovata cultures were revived from frozen stocks in heterotrophic growth medium and transferred once in the same medium. Cells in mid-exponential phase were used as inoculum to obtain a starting OD of 0.05. Only for trickle bed reactors under continuous gas supply, the inoculum was first washed twice with autotrophic growth medium to remove acetate and organic substrates. In addition, this inoculum was adjusted to obtain an initial cell density of about 5⋅10⁶ cells mL^-1 in each reactor, as was measured using Impedance Flow Cytometry (IFC) (detailed below). All experiments were performed at least in triplicate, while abiotic controls were included in all experiments.

Trickle bed reactors. Simple trickle bed reactors consisted of 100 mL gas-tight bottles (with a butyl rubber stopper) filled with HDPE carriers (25 carriers with batch gas supply, 36 carriers with continuous gas supply) (high-density polyethylene plastic, HEL-X, HXF12KLL, diameter and length 12 mm, surface area 10 cm², Christian Stöhr GmbH & Co., Germany). These carriers were used, as they successfully stimulated biofilm formation in another study [12]. The carriers were incompletely covered with autotrophic growth medium (15 mL with batch gas supply, 10 mL with continuous gas supply). The bottles were incubated at 30 °C and were regularly turned upside down and along their axis (approximately once per day, static incubation the rest of the time) to replenish nutrients and create a thin layer of adsorbed water on the carriers, simulating a trickle bed reactor.

The trickle bed reactors were operated in two different ways: either batch or continuous gas supply. Reactors with batch gas supply had a headspace consisting of N₂:CO₂ (80:20, 1 bar), which was over-pressurized with H₂:CO₂ (80:20, 0.3 bar) gas at the beginning of the experiment, as well as after the overpressure had reached zero due to microbial consumption. Gas consumption was monitored by measuring the headspace overpressure using a manometer (Bourdon).

In contrast, in trickle bed reactors with continuous gas supply, a H₂:CO₂ (67:33) gas mixture was continuously supplied to maintain a constant overpressure of 0.5 bar, with a maximum flow rate of 50 mL min⁻¹ as regulated by a mass flow controller (Brooks SLA5800). The liquid phase of these reactors was regularly sampled using a syringe to measure the acetate concentration, optical density (OD), and planktonic cell numbers.

Media replacement. In the trickle bed reactors with continuous gas supply, medium was completely replaced to stimulate attachment and biofilm formation. To replace the medium, the bottles were gently positioned upside down to accumulate the liquid above the rubber stopper. A syringe (flushed with N₂:CO₂) was used to withdraw 10 mL liquid and subsequently reinject 10 mL fresh medium.

Gas consumption rate measurements. After specific medium replacements, gas consumption rates in trickle bed reactors were measured. The continuous gas supply in these bottles was first turned off, after which the decrease in overpressure was monitored with an online manometer (Go Direct, Vernier USA). Gas consumption rates were calculated as the average decrease in pressure over time during the second hour, as the first hour was considered as equilibration time.

Adaptive laboratory evolution. ALE was started by transferring two carriers from trickle bed reactors (with batch gas supply), initially inoculated with the original S. ovata 2663, to new bottles with fresh medium and fresh carriers. The carriers to be transferred were randomly taken from the middle of the bottle, where no liquid medium had been accumulated. This transfer was performed using sterile tweezers inside an anaerobic chamber (Jacomex, France) with a N₂:CO₂ atmosphere (80-82% N₂ and 18-20% CO₂). After the transfer, a headspace pressure of N₂:CO₂ (80:20, 1.1 bar, to prevent underpressure) and H₂:CO₂ (80:20, 0.3 bar) (total pressure of 1.4 bar) was established in the new bottle. The new bottles were incubated at 30 °C and were turned upside down several times per week, while the gas consumption was monitored by measuring the decrease in headspace pressure. When the pressure approached 1 bar, H₂:CO₂ (80:20) gas was reinjected to reestablish a total pressure of 1.3 bar. After a second or third injection of H₂:CO₂ gas, the procedure of transferring carriers to a new bottle was repeated. In total, eight serial transfers were performed over a period of six months. At the later transfers, a liquid sample from the bottle (after shaking) was also taken to propagate the obtained culture in heterotrophic growth medium and prepare frozen stocks. The resulting adapted culture with increased biofilm formation is called Sporomusa ovata 2663-BF or BF culture in short, while the ancestral strain (Sporomusa ovata 2663) is called the wild type (WT) culture.

Biofilm quantification. To assess biofilm formation characteristics, S. ovata (both WT and adapted BF culture) was grown in well plates (6-well plates, untreated polystyrene) with 5 mL heterotrophic growth medium per well. These well plates were placed in a plastic bag to restrict evaporation and incubated inside an anaerobic chamber (Jacomex, France) with a N₂:CO₂ atmosphere (80-82% N₂ and 18-20% CO₂) at 30 °C. Two to four days after inoculation, the well plates were taken out of the anaerobic chamber to assess biofilm formation using a crystal violet assay [14]. First, the planktonic part of each well was removed and its OD₆₀₀ was measured. Next, the biofilms remaining in the wells were stained with 5 mL 0.2% crystal violet solution for 30 min. After staining, the crystal violet solution was removed, and the stained biofilms were washed twice with 5 mL PBS buffer (pH 7). Finally, the stain was extracted from the biofilm with 5 mL methanol for 30 min, after which the absorbance of the coloured methanol solution was measured at 570 nm as a quantitative measure for biofilm formation.

Cell number quantification. Planktonic cell numbers, as well as cell numbers attached to well plates or carriers, were quantified using impedance flow cytometry (IFC). To count the cells attached to the well plates, the planktonic phase was removed, and the remaining biofilm was carefully washed twice with PBS buffer, before the biofilm was suspended in 5 mL fresh medium. Biofilms were manually suspended by shaking, as they detached and disintegrated easily. To guarantee complete detachment, the cell suspension was vigorously pipetted in the well and also agitated by vortex for 30-60 s in 15 mL centrifuge tubes. Measured cell densities were used to calculate the number of attached cells per cm² of the bottom of the wells (9.6 cm²).

To quantify the cell numbers attached to the carriers of the trickle bed reactors, the carriers were first carefully washed two times inside of the reactor with autotrophic growth medium, after which all liquid was removed and the reactor was opened inside the anaerobic chamber. Seven carriers were moved carefully and with sterile tweezers to a Falcon tube to which 5 mL autotrophic growth medium was added. This tube was manually shaken and also vortexed at high speed for at least 60 s to guarantee the suspension of all the cells attached to the carriers. The measured cell densities were used to calculate the average number of cells per carrier. This procedure and measurement was performed in triplicate for each reactor.

In case cell densities could not be measured directly using IFC, they were estimated from the OD using the previously determined correlation factor for S. ovata 2663 of 2.17 ⋅ 10⁸ cells mL⁻¹ per OD₆₀₀ unit [26].

ExoPS visualization and quantification. ExoPS attached to the cell surface (also called capsules) were visualized using the Anthony's Capsule Stain [27]. First, the BF and WT S. ovata were grown in heterotrophic growth medium with the extra addition of sterile low-fat milk (10% v/v) to create a stainable proteinaceous background. For BF, 200 μL of cells, that had adhered to the tube's wall, was carefully collected using a needle and gently mixed with 500 μL sterile low-fat milk. For the WT, 200 μL of the planktonic cells were mixed with the milk. A 20 μL aliquot of the cell suspension was spread onto a clean glass slide and dried to the air without heat fixation. Slides were stained with 1% crystal violet and rinsed with 10% copper sulphate. The slides were examined using brightfield microscopy (Leica ICC50 W) with a 100X oil-immersion objective lens. Identical imaging settings were used for all samples. Image cropping and the addition of scale bars were performed using ImageJ [28].

Furthermore, exoPS was extracted and quantified. The BF and WT cultures were grown in heterotrophic growth medium for two days, after which they were homogenized using vortexing and the OD was measured. Next, the cell suspensions were heated at 90 °C for 15 min to detach any exoPS attached to the cells [29], after which the cells were cooled (1h) and removed by centrifugation (5000 rpm, 20 min, 4 °C) and the sugar concentration in the supernatant was measured using the Dubois method [30] on a spectrophotometer (490 nm, Denovix DS11). A standard curve was obtained using glucose solutions.

Growth yield measurement. Growth yields of the BF and WT S. ovata were measured both in heterotrophic (betaine) and autotrophic (H₂:CO₂) growth conditions. The cultures were first grown in quintuplicate bottles with 40 mL medium and 10 carriers. For autotrophic growth, the headspace was flushed with H₂:CO₂ (80:20) gas and pressurized to 1.5 bar total pressure (batch gas supply). These bottles were regularly repressurized with the same gas to avoid underpressures. After four days, the bottles were vigorously shaken to detach adhered cells. A volume of 1 mL was sampled from each bottle to measure OD₆₀₀ and cell densities using IFC. The entire liquid phase of each bottle was then filtered through a pre-dried, pre-weighed 0.25 μm polyethersulfone filter (Frisenette, DK) using a vacuum filtration system. The filter with retained biomass was dried at 55 °C for 24 h, cooled down in a desiccator, and weighed on an analytical balance (±0.0001 g precision) (VWR, LD17 4XN). Dry weight was calculated as the difference between the filter's weight before and after filtration. In addition, the filtrate was analyzed for volatile fatty acids (VFAs). Growth yields were calculated by dividing either the dry weight or the cell numbers by the amount of acetate produced.

Analytical methods. Impedance flow cytometry was performed using a Bactobox benchtop device (SBT Instruments A/S, Denmark) [31], placed under normal atmospheric conditions. A small volume of the homogenized cell suspensions (100-550 μL) was diluted in 10-11 mL dilute PBS buffer with a conductivity between 1900 and 2100 μS/cm and vortexed for 30-60 s. All reported cell numbers reflect the total particles presumed as the number of total cells in the sample.

Optical densities (OD₆₀₀) were measured at 600 nm using a spectrophotometer (Spectronic 200, Thermo Scientific).

VFAs were analyzed by GC- Flame ionization detection (FID) (system 7890, column: HP-INNOWAX 19091N-133, 30 m, 0.25 μm, 0.25 ID, Agilient Technologies, USA) with helium as the carrier gas and H₂ and air as FID detector gases. The detector temperature was 250 °C. Pivalic acid (175 mg L⁻¹ in 0.3 M oxalic acid) was used as internal standard. The detection limit for acetate was 0.3 mM. Ethanol was measure using high-performance liquid chromatography (HPLC).

Statistical Analysis. Student's t-tests were performed to differentiate statistically significant experimental results using GraphPad Prism. The p-value was set at 0.05 for all tests.

Mutation screening. The WT and BF culture were revived from frozen stocks and genomic DNA was extracted using the DNeasy Blood & Tissue Kit (Qiagen) following the manufacturer's protocol for Gram-positive bacteria. Genomic DNA was prepared for sequencing using the Nextera XT kit (Illumina) and sequenced on an Illumina MiSeq sequencer with the MiSeq reagent kit v3. Poor quality sequences (mean < 20 with a sliding window of four bases) were removed using Trimmomatic 0.39 [32]. Reads were mapped to the reference sequence of Sporomusa ovata DSM-2662 (RefSeq accession GCF_000423685.1 -no genome sequence is available for DSM-2663, but these strains have an average nucleotide identity of 99.95%) and mutations were identified using Snippy 4.4. [33]. Mutations found in Snippy were confirmed by assembling sequencing reads with SPAdes 3.15.5 [34] and annotation using Prokka 1.14.5 [35], checking predicted protein-coding sequences for the mutations detected by Snippy. Mutations identified in the adapted S. ovata 2663-BF relative to the reference sequence of S. ovata DSM-2662, that were not already present in the WT S. ovata DSM-2663, were manually inspected to check for positions relative to genes and adjacency to homopolymer regions to determine whether they were true mutations from the ALE experiment. Transcription terminator regions were predicted using BacTermFinder version 1.0 [36] with a mean probability threshold of 0.3 (normal stringency). Raw sequencing data have been deposited in the NCBI sequence read archive under BioProject accession number PRJNA1081306.

Computational structure analysis of UGPase. The structures of the UGPase enzyme encoded by WT and the identified L145H mutant were predicted by the AlphaFold 3 server as a homotetramer and subsequently filled with the product UDP-α-d-glucose and magnesium ion by the AlphaFill server [37,38]. The final 3D structures were compared and visualized by Chimera 1.17.3 [39], while enzyme tunnel analysis was performed by Caver Web (v2.0) [40].

3. Results

3.1. Trickle bed experiments with wild type S. ovata

Initial experiments evaluated the performance of WT S. ovata 2663 in simple trickle bed reactors and assessed its intrinsic capacity to attach. Simple trickle bed reactors were bottles filled with plastic carriers and a small amount of medium in a H₂:CO₂ containing headspace. In a first experiment, gas consumption by S. ovata was monitored in bottles with batch gas supply and with or without carriers (manually turned around daily, otherwise static incubation). In the abiotic controls, a gradual drop in pressure was observed, illustrating the loss of gas due to repeated pressure measurements (Fig. 1A). Both biotic treatments showed a faster drop in the pressure than the abiotic controls, showing the gas consumption activity of S. ovata. Interestingly, in the bottles with carriers, the pressure drop was at least ten times faster than without carriers (Fig. 1A). This relates to the ten times larger surface area of the gas-liquid interface (Supplementary Calculation), demonstrating that the carriers made the bottles function as trickle bed reactors.

Fig. 1.

Performance of S. ovata DSM-2663 (WT) under autotrophic growth conditions with and without the presence of carriers. All bottles were statically incubated and manually turned around at least daily. (A) Gas (H₂:CO₂) consumption (measured as a decrease in overpressure) by S. ovata DSM-2663 in bottles with or without carriers and in abiotic control bottles. At the start, these bottles contained 1 bar of N₂:CO₂ (80:20) gas (represented by the dashed line in the bottom) and 0.3 bar overpressure of H₂:CO₂ (80:20) gas. For the bottles with carriers, H₂:CO₂ (80:20) gas was reinjected on day 5, 8 and 10, when the overpressure approached or reached zero. (B) (C) and (D) Increase of respectively the acetate concentration, optical density (OD₆₀₀) and planktonic cell numbers by S. ovata DSM-2663 in bottles with or without carriers. In these bottles, H₂:CO₂ (67:33) gas was continuously supplied to maintain a constant overpressure of 1.5 bar. All points represent the average of at least triplicated bottles, while the error bars show the standard deviation.

In a second experiment, acetate concentrations and planktonic growth were monitored in reactors with continuous H₂:CO₂ gas supply and with or without carriers. This condition led to similar acetate concentrations with and without carriers after four days of incubation (Fig. 1B), showing that carriers did not have the same advantage, as observed under batch gas supply (Fig. 1A). Interestingly, however, in the bottles with carriers, the optical density (OD₆₀₀) of the liquid phase was about 40% lower than without carriers after four days of incubation (Fig. 1C), while planktonic cell densities were about 30% lower (significant difference, p < 0.05) (Fig. 1D). A significant number of cells had thus attached to the carriers, as similar overall cell numbers are expected from the comparable acetate concentrations (Fig. 1B). This suggests that WT S. ovata 2663 has the capacity to attach to surfaces and does not stay purely planktonic in autotrophic growth conditions.

ALE towards increased biofilm formation. Starting from one of the bottles with carriers of the first trickle bed experiment (Fig. 1A), ALE was carried out by repeatedly transferring two carriers to new trickle bed reactors with fresh medium and fresh carriers (Fig. 2A). In this way, only cells associated to the carriers were transferred, creating a selection pressure for cell attachment. Over the course of the different serial transfers, new transfers gradually started to consume gas faster. For the first transfers, the time between the transfer and the consumption of 0.3 bar of gas overpressure was about 16 to 23 days, while for the last transfers, this time was about 6 to 8 days. As a consequence, the intervals between the serial transfers became shorter during the course of the ALE experiment (Fig. S1). This suggests that increasingly more biomass adhered to the carriers, and that S. ovata attachment was enhanced over the transfer series. At the later transfers, the culture was also propagated in heterotrophic growth medium to allow the preparation of frozen stocks. When grown under heterotrophic growth conditions, the evolved S. ovata 2663 culture obtained after eight serial transfers, visibly differed from the WT S. ovata (Fig. 2B). The WT S. ovata homogenous grew throughout the liquid phase and was visibly planktonic, in contrast to the evolved S. ovata, which formed a visible biofilm on the sides of glass tubes. As this was the desired phenotype, ALE was ended at this timepoint. The adapted culture with increased biofilm features obtained at the end of the ALE procedure is called Sporomusa ovata 2663-BF (or BF culture in short).

Fig. 2.

(A) Conceptual scheme illustrating the adaptive laboratory evolution strategy used to increase biofilm formation by S. ovata. The WT S. ovata DSM-2663 was grown in simple trickle bed reactors under an atmosphere of N₂, CO₂ and H₂ (batch gas supply, manual turned around several times per week). When the gas at overpressure was consumed, H₂:CO₂ gas was reinjected. After two to three reinjections, two carriers were transferred to a new reactor with fresh medium and fresh carriers. The change in the grey colour of the carriers in the scheme illustrates the transfer between different bottles. Eight transfers were performed to obtain the final adapted S. ovata 2663-BF with increased biofilm formation properties. (B) Picture showing the visual difference between the WT S. ovata and the adapted BF culture, when grown in glass Hungate tubes under heterotrophic growth conditions. The WT was mainly planktonic, while the BF culture was rather associated to the side of the tube.

3.2. Comparison of biofilm formation in heterotrophic growth conditions

Heterotrophic growth conditions (betaine addition) were chosen to compare the biofilm formation characteristics between the WT and the BF S. ovata, since this allowed growth in well plates in an anaerobic chamber. After three days of growth in well plates, the OD₆₀₀ of the planktonic phase, as a measurement of planktonic growth, was about four times lower for the BF culture compared to WT (Fig. 3A, left). Conversely, the biofilm formation of the BF culture was about ten times higher than the wild type, as quantified by a crystal violet assay (Fig. 3A–right & 3B). In addition, the increased biofilm formation by the BF culture corresponded with an at least four times higher cell numbers attached to the well plates than with the WT (Fig. 3C, right), while in the planktonic phase of the BF culture had seven times less cells than for the WT (Fig. 3C, left). Similar experiments using the crystal violet assay were further repeated to acquire a time resolution, in which the same trends were reproduced and biofilm formation was found to be stable over at least a few days (quantified from day 2-4) (Fig. 3D). In all, these results demonstrate that the ALE procedure increased the capacity of S. ovata to attach to surfaces and form biofilms, at least under heterotrophic growth conditions.

Fig. 3.

Comparison of planktonic growth and biofilm formation by the WT S. ovata DSM-2663 versus the adapted S. ovata 2663-BF in heterotrophic growth conditions and when grown in well plates. (A) Planktonic growth was assessed by the optical density (OD₆₀₀) (left) and biofilm formation was quantified by the crystal violet assay (A₅₇₀) (right) after 3 days of incubation. (B) Representative pictures of biofilms stained with crystal violet before extraction with methanol from different repeated runs, all after 3 days of incubation, with WT (left) and BF (right). (C) Cell density of the planktonic phase (left) and the number of cells per cm² in the biofilm attached to the bottom of the well (right) as measured using IFC after 3 days of incubation. (D) Planktonic growth measured by the optical density (OD₆₀₀) (left) and biofilm formation quantified by the crystal violet assay (A₅₇₀) (right) at three different time points. The height of all bars or points represents the average of measurements of up to nine wells, while the error bars show the standard deviation.

3.3. Visualization and quantification of exoPS

ExoPS was visualized and quantified for BF and WT S. ovata cultures grown in heterotrophic conditions. The microscopy analysis showed that the WT had a thin and smooth exoPS layer around its cells (Fig. 4A-left), while the BF has a thicker and more uneven exoPS layer (Fig. 4A–right). Similarly, the exoPS quantification showed that the BF produced significantly more exoPS per cell than the WT (p < 0.05) (Fig. 4B), demonstrating that the ALE procedure resulted in increased exoPS production by S. ovata.

Fig. 4.

Comparison of extracellular polysaccharide (exoPS) formation by the WT S. ovata DSM-2663 versus the adapted S. ovata 2663-BF in heterotrophic growth conditions. (A) Microscopic visualization of the exoPS layer around S. ovata WT and BF cells using a capsule staining. The cells and the background were stained purple using crystal violet, while the polysaccharides were not stained and are visible as the white halo around the cells. (B) Quantification of exoPS after extraction from cells grown in heterotrophic medium for two days. Data points represent the average of three replicates, while the error bars show the standard deviation. The star indicates a statistically significant difference (p < 0.05).

3.4. Comparison of growth yields

The growth yield of the BF and WT S. ovata was determined both under heterotrophic and autotrophic growth conditions (Fig. S2). Growth yields, incorporating both the increase of attached and planktonic cells, were not significantly different between the two cultures, both when expressed as the increase of cell numbers or as cell dry weight per mole of acetate produced (p > 0.05), indicating that the increased propensity of the BF culture to form exoPS did not impair its capacity to grow.

3.5. Comparison of performance in trickle bed reactors

The performance of the BF and WT S. ovata cultures was further compared in simple trickle bed reactors using continuous gas supply. Even though both cultures were inoculated with equal cell numbers, the WT started to accumulate acetate faster, while a lag phase of two days was observed in the acetate production by the BF culture (Fig. 5A). After four days, the acetate concentration was twice as high for WT (3.1 ± 0.3 g L⁻¹) than for BF (1.4 ± 0.7 g L⁻¹), which was also reflected by a lower pH and higher OD₆₀₀ (Fig. 5B and S3). Ethanol concentrations remained under the detection limit throughout the entire experiment.

Fig. 5.

Comparison of the WT S. ovata DSM-2663 and the adapted S. ovata 2663-BF in autotrophic growth conditions when grown in simple trickle bed reactors. In all bottles, H₂:CO₂ (67:33) gas was continuously supplied to maintain a constant overpressure of 1.5 bar. Bottles were manually turned around once per day to create a trickling effect, otherwise static incubation. Medium was replaced to stimulate biofilm growth, as indicated by the vertical lines. (A) and (B): Change of respectively acetate concentrations and optical densities (OD₆₀₀) over time, OD₆₀₀ values reflect planktonic cells. (C) The number of attached cells per carrier after 4 and 15 days, as measured using IFC after detachment of the cells of the carriers. (D) The total number of attached and planktonic cells at day 4 and 15, as calculated from the number of attached cells per carrier (Fig. 5C) with 36 carriers and the OD₆₀₀ at day 4 and 13 (before medium replacement) (Fig. 5B) with 10 mL liquid phase. Data points represent the average of at least three replicates, while the error bars show the standard deviation, with the exception of the number of attached cells on the carriers at day 4, for which only one replicate bottle was measured.

At day four, one of the four replicated bottles of each culture was harvested to quantify the cells attached to the carriers. The BF strain had 2.3 times more cells attached (2.0⋅10⁷ cells per carrier) than the WT (9⋅10⁶ cells per carrier) (Fig. 5C). It was further calculated that 67% of the BF cells were attached, while this was only 26% for WT cells at day 4 (Fig. 5D). This confirms that also in autotrophic growth conditions, the BF strain has an increased tendency to attach and form biofilms in comparison to the WT strain.

Medium was completely replaced at day 4 in the remaining bottles, which was repeated every few days over a period of 10 days. After the medium replacements, the OD₆₀₀ was not completely reset to zero (Fig. 5B), indicating that not all planktonic cells were removed, or that some attached cells got suspended by the addition of fresh medium. During the 10-day period, the acetate production rate of both strains strongly increased (Fig. 5A), suggesting increasing cell numbers in the reactors. Moreover, the BF and WT culture reached comparable acetate rates (up to 2.27 ± 0.19 g L⁻¹ day⁻¹ for WT and 2.42 ± 0.17 g L⁻¹ day⁻¹ for BF after the fourth medium replacement at day 9), but in general the acetate titers remained slightly lower for BF culture than for WT (e.g. 3.23 ± 0.08 g L⁻¹ for BF versus 4.47 ± 0.19 g L⁻¹ for WT at day 9) (Fig. 5A). Accordingly, the pH dropped less in between the medium replacements for BF than for WT (e.g. pH 5.71 for BF vs 5.32 for WT at day 9) (Fig. S3). For both strains, the OD₆₀₀ increased between medium replacements (Fig. 5B), indicating that at least a part of the growth occurred planktonically, or otherwise resulted from the suspension of biofilm grown cells. The OD₆₀₀ increase was less pronounced for the BF culture than for WT (OD₆₀₀ of 0.405 for BF vs OD₆₀₀ of 0.729 for WT at day 9), which might be due to a higher number of attached BF cells (Fig. 5C). Overall, the acetate and OD₆₀₀ measurements indicate that the BF culture performed comparable to the WT in a system in which planktonic growth apparently dominated.

In order to assess the activity of the attached cells (and any remaining planktonic cells), the continuous gas supply was halted after some of the medium replacements to enable measurement of the gas consumption rates (i.e. decrease of pressure per time unit). At all instances, the pressure dropped much faster in the biological treatments than in the abiotic controls (Fig. 6A and S4), confirming the activity of both cultures. At two instances (day 4 and 9), the BF culture had a gas consumption rate twice as high as the WT (Fig. 6A). For day 4, the higher BF gas consumption rate might relate to the fact that the WT had to recover from unfavorable conditions (pH < 5.5 at day 4 for WT) (Fig. S3). At day 9, in contrast, both strains had a comparable performance before the medium replacement (Fig. 5A and B). The higher BF gas consumption rate at day 9 is thus likely due to its higher attached cell numbers. Gas consumption rates were also measured at other time points (day 6 and 11), but were then not significantly different between the two cultures (Fig. 6A).

Fig. 6.

(A) Gas consumption rates of the WT S. ovata DSM-2663 and the adapted S. ovata 2663-BF at four different time points, when grown in autotrophic growth conditions in simple trickle bed reactors. To measure these gas consumption rates, the continuous gas supply was halted after the medium replacements. All gas consumption rates were calculated from the pressure decrease during the second hour (first hour as equilibration time). The height of the bars shows the average of triplicate bottles, while the error bars reflect the standard deviation. (B) Pressure decrease measurement for a period of 15 h during day 9, comparing gas consumption by the WT and BF culture, as well as in abiotic conditions. The darkest line represents the average of three replicate bottles, while the shaded area shows the standard deviation.

Several pressure measurements were continued overnight (up to 16 h). Interestingly, around day 9, this resulted in an almost constant gas consumption rate (linear pressure drop) for the BF culture, while the gas consumption rate of the WT increased over time (exponentially decreasing pressure) (Fig. 6B). This difference in the shape of the curves was also recorded after other medium replacements (Fig. S4), suggesting that gas consumption by the WT depended on (exponential) growth, while less growth occurred for the BF culture, possibly because more cells remained in the reactors after the medium replacement, thanks to its higher attachment.

At the end of the experiment (after 15 days), all remaining bottles were harvested to quantify the cells attached to the carriers. The adapted S. ovata had about 2.5 times more attached cells (2.09⋅10⁷ cells per carriers) than the WT (8.2⋅10⁶ cells per carrier) (Fig. 5C). Moreover, about 42% of the BF cells were attached, while this was only 14% for WT cells at the end of the experiment (Fig. 5D). The higher attachment of BF S. ovata thus persisted throughout the experiment. Nevertheless, the relative number of attached cells decreased over the course of the experiment (Fig. 5D), likely because the system favored planktonic growth.

Genomic comparison. To obtain insights into the biofilm formation mechanisms of the evolved S. ovata 2663-BF, we performed a genome analysis comparing the genome of the WT and BF culture. Mapping the sequencing reads to the reference genome indicated several mutations in the BF versus the WT S. ovata 2663 (Table S2). Overall, 11 mutations were found in protein-coding genes, while there were eight mutations in intergenic regions, and two mutations in ribosomal RNA genes. Of the eight mutations in intergenic regions, five were adjacent to long homopolymer stretches and thus likely the result of sequencing errors (Table S2). The three remaining intergenic mutations (on contig NZ_AUIL01000025) were not upstream of any coding sequences for proteins or non-coding RNA, but were in a putative terminator region on both strands, as predicted by BacTermFinder. This tool, however, predicted an unrealistic high number of terminator regions (74,910 for a genome with 5024 protein-coding genes), and thus identifies many false terminator regions. It is thus doubtful whether the three intergenic mutations were in actual terminator regions. Of the 11 apparent mutations in protein-coding genes, nine were found in genes with multiple copies (cdhC and phage tail protein) and likely result from inaccurate mapping of reads between the different copies (Table S2). The two remaining mutations in protein-coding sequences included a frameshift mutation that was adjacent to a long homopolymer stretch and thus likely the result of sequencing errors in hypothetical protein G571_RS0122020, as well as interestingly a missense variant in galU (Table S2). This single-nucleotide mutation at position 434 in the galU gene changed the encoded amino acid from a leucine to a histidine (L145H). The gene galU encodes the enzyme UTP-glucose-1-phosphate uridylyltransferase (UGPase) (Fig. 7), which catalyses the formation of UDP-glucose from glucose-1-phosphate and UTP and has previously been related to the synthesis of exoPS [41,42]. This mutation was found in 89 of the 90 sequencing reads mapped to this gene (Table S2), showing that the mutation had become almost completely established in the BF culture.

Fig. 7.

Schematic representation of (A) a pathway for exoPS production in which UTP-glucose-1-phosphateuridylyltransferase (UGPase) is expected to be involved and (B) the location of the mutation in the galU gene causing an altered amino acid sequence of the UGPase enzyme.

3.6. Structural comparison of the UGPase of WT and BF S.ovata

We investigated whether the point mutation in galU could have affected the enzyme activity of the encoded UGPase, through the generation of structural models of WT and the mutant enzyme (L145H). In UGPase enzymes, glucose-1-phosphate makes a nucleophilic attack on the α-phosphate of UTP to form UDP-glucose, while magnesium ion (Mg²⁺) at the active-site plays a key role in the catalysis by charge neutralization and by positioning the substrates [[43], [44], [45], [46], [47], [48]]. According to our comparative structure analysis using AlphaFold and AlphaFill, the mutated residue (L145H) is far from the active-site pocket, with over 16 Å distance to Mg²⁺ (Fig. 8A). However, the imidazole side chain of H145 in the mutant is within interaction distance to the backbone carbonyl oxygen of a tyrosine 210 (Y210) (4.2 Å distance, Fig. 8A). Y210 is highly conserved among UGPase orthologs and structural studies assigned Y210 as one of the residues forming a hydrophobic cap at the base of the glucose ring in the active-site [43]. Moreover, in our structural model of the S. ovata UGPase, the hydroxy group of Y210 is within 4.4 – 6.0 Å distances from the oxygen atoms of the glucose moiety of UDP-glucose, indicating a role for Y210 in binding and/or catalysis. Thus, interaction between H145 and Y210 can possibly lead to higher catalytic efficiency in the mutant by affecting substrate binding or through better positioning of the substrate(s) for catalysis. In the WT enzyme, L145 is not able to make this interaction with Y210 due to its inert side chain.

Fig. 8.

(A) Overlaid active-site of WT and mutant S.ovata UGPase highlighting the mutation (L145H) site and possible interactions between His145 side chain and Tyr210. The brown structure is WT UGPase and the blue structure is the mutant. The distances between Tyr210 side chain and glucose moiety of UDP-glucose are shown. The green sphere is the magnesium ion. (B) Overlaid structures of WT and mutant S.ovata UGPase. Brown structure is WT enzyme and blue structure is the mutant. The surface representation is only shown for mutant enzyme for clarity. Highlighted are the Arg227 residue that is positioned differently between the two models and the likely interaction with Lys194 in the WT enzyme.

In addition, structural comparison of WT vs. L145H mutant UGPase revealed a significant difference in the position of Arginine 227 (R227) (Fig. 8B), which resides on a mobile gatekeeping loop at the entry of an access tunnel to the active site pocket, according to earlier studies [43] and our analysis by Caver Web software. In the WT enzyme, R227 is 7.8 Å closer to the active-site and partially closes the access tunnel, with its side chain possibly interacting with the backbone of two active-site residues, Lysine 194 (K194) and Glutamate 193 (E193) [43] (Fig. 8B). In the L145H mutant, these interactions are absent. Although AlphaFold 3 models of the enzymes had a higher degree of uncertainty in the loop region including R227 (pLDDT scores per residue in Fig. S5), such differences in the mutant possibly affect the size, access and organization of active-site pocket and/or conformational changes during catalysis or binding events. It has been well documented that various conformational changes and active-site dynamics play a role in the bi-substrate mechanism of UGPases [46].

Additionally, it is possible that the L145H mutation favorably affected the stability of the enzyme, increasing overall catalytic performance. In short, our structure analyses show that the point mutation in galU most probably increased the activity of UGPase enzyme, thereby leading to increased exoPS formation and cell attachment.

4. Discussion

4.1. ALE increased attachment and exoPS formation by S. ovata 2663

Inherent biofilm formation of acetogens, such as S. ovata, is often considered thin or with sparse attachment of cells (Table S1). Throughout this study, we found evidence that WT S. ovata 2663 was capable of attaching to solid surfaces to a certain extent. Some attachment to the bottom of well plates was observed in heterotrophic growth conditions (Fig. 3), while some WT cells also attached to the carriers of the trickle bed reactors (Fig. 1, Fig. 5). In one MES study, S. ovata 2663 was found to have the most pronounced cathodic biofilm formation of various Sporomusa strains (Table S1) [13]. Nevertheless, in our study, the WT S. ovata 2663 mainly grew planktonic both in auto- and heterotrophic growth conditions (Fig. 1, Fig. 2, Fig. 5D).

To increase attachment and biofilm formation by S. ovata 2663, we performed an evolution experiment. Our ALE strategy consisted of transferring carriers with attached cells to fresh trickle bed reactors over a period of six months. This strategy was highly successful, as cell attachment by the adapted S. ovata was consistently higher than for the WT S. ovata in both heterotrophic and autotrophic conditions (Fig. 3, Fig. 5). Moreover, the BF culture had a reduced planktonic growth (Fig. 3) and significantly produced more extracellular polysaccharides (Fig. 4). Even in the trickle bed experiment, where planktonic growth dominated, the BF culture maintained a higher relative fraction of attached cells in comparison to the WT (Fig. 5D), demonstrating the stability of this evolved phenotype. In general, ALE is a widely used tool for altering phenotypic traits with a poorly understood genetic basis. ALE has previously been used to optimize other traits of S. ovata [22,23], while it has also been used to increase biofilm formation properties of aerobic microbes [24,25]. Our study demonstrates that ALE is thus also a valuable method to increase attachment of anaerobic and chemolithoautotrophic microbes, such as S. ovata.

Interestingly, the adapted and the WT S. ovata had comparable growth yields, both under heterotrophic and autotrophic growth conditions (Fig. 6), suggesting they have comparable amounts of ATP available for growth. Biofilm formation likely comes at the expense of some ATP for the production and excretion of exoPS, while acetogens already live on the thermodynamic limit of life [2]. The lack of a significant difference in growth yields is possibly due to the mixed growth states of both the BF and WT cells, as none grew exclusively as a biofilm or planktonic (Fig. 5D). In addition, the ATP required for biofilm formation can likely be saved from not expressing flagella or other traits of planktonic cells. Biofilm and planktonic cells indeed have completely different physiological states, leading to strong differences in gene expression [14,49,50]. S. ovata thus likely reallocates the small amount of energy available from acetogenesis depending on its planktonic or attached growth. Future transcriptomic and proteomic analyses could help to further map this reallocation of energy.

4.2. Increased attachment did not lead to increased performance in simple trickle bed reactors

Biofilm-based reactors systems, including trickle-bed and membrane biofilm reactors, have shown promise for H₂ and CO₂ conversion by gas fermentation [8,9], but so far, the importance of biofilms in these systems in not well understood. This work therefore aimed to investigate the importance of biofilm formation in biofilm-based reactor systems, by comparing the adapted and WT S. ovata in simple trickle bed reactors. We did not find a strong difference between the performance of the two cultures (Fig. 5A), while the BF S. ovata even accumulated slightly less acetate than the WT S. ovata (Fig. 5A). Nevertheless, an advantage for the adapted S. ovata was expected, as repeated medium replacements were performed to remove planktonic cells and favour attachment and biofilm formation. After two of such medium replacements, higher gas consumption rates were recorded for the BF culture (Fig. 6A), suggesting an advantage for attached cells in the trickle bed reactors, at least for a certain period of time after the medium replacement. This advantage for attached cells clearly did not last throughout the experiment, as over the course of the experiment the proportion of planktonic cells strongly increased for both the BF and WT culture (Fig. 5D).

Most likely, the experimental setup used in this study was insufficient to stimulate biofilm growth. Our trickle bed reactors were simply bottles filled with carriers, which were turned around only once a day to spread the medium. We did observe that carriers made the bottles function as trickle bed reactors, as the gas consumption of WT S. ovata was strongly increased in the presence of carriers (Fig. 1A). However, this observation was based on batch gas supply, while acetate titers with carriers were not higher than without carriers using continuous gas supply (Fig. 1B). Under batch gas supply, the gradually decreasing gas pressures likely led to fairly low acetate production rates, as H₂ consumption rates linearly decrease with decreasing H₂ partial pressures [26]. Those low rates likely entailed that daily trickling was sufficient to avoid to accumulation of too high acetate concentrations in the biofilms. However, to compare the performance of the BF an WT S. ovata, we opted for continuous gas supply and thus a fixed H₂ partial pressure, to avoid the kinetic effects of changing H₂ partial pressure. Unfortunately, this continuous gas supply likely caused the daily trickling to be insufficient, as the high acetate production rates likely resulted in high local acetate concentrations in biofilms, causing product and pH inhibitions. Continuous gas supply thus likely stimulated planktonic growth, rather than biofilm formation. A proper investigation of the role of biofilms in biofilm-based reactors thus requires the use of experimental setups more resembling actual reactor operation conditions, for instance using continuously trickling and the continuous addition of new medium and removal of planktonic cells, as in more realistic, but also more complex, trickle bed reactors [9].

4.3. Mutation in galU gene likely contributed to the altered biofilm formation

As the evolutionary engineering led to phenotypic changes of S. ovata (Fig. 2, Fig. 3, Fig. 4, Fig. 5, Fig. 6), genome comparisons were performed to investigate whether the ALE had resulted in genotypic changes. The most interesting mutation was a point mutation in the galU gene (change in the encoded amino acid from a leucine to a histidine (L145H)), encoding for the enzyme UTP-glucose-1-phosphate uridylyltransferase or also called UDP-glucose pyrophosphorylase (UGPase) (Table S2). Moreover, structural comparison of the WT and the mutated UGPase enzyme showed that this mutation possibly altered the substrate binding and the conformations of amino acid residues within the active-site pocket, leading to enhanced activity (Fig. 8). UGPases catalyse the synthesis of UDP-glucose, which has a role in the synthesis of extracellular polysaccharides [41,51,52]. ExoPS is often an important part of EPS, forming the binding matrix of microbial biofilms [15]. Consequently, the identified mutation likely explains the increased synthesis of exoPS by the adapted strain (Fig. 4). Previous studies have already linked the galU gene to changes in biofilm formation capabilities in other bacteria, such as Salmonella [53], Vibrio [41], and Haemophilus [52]. These studies deactivated the galU gene through genetically engineering and found impaired exoPS production and biofilm formation when the galU gene was deleted. Further investigations are required to verify that the specific point mutation found in the present study is indeed responsible for the increased exoPS synthesis by the adapted BF culture. So far, genetic engineering techniques to alter S. ovata are still in their infancy [54], but being better developed for some other model acetogens [55,56]. Once genetic engineering methods get well established for S. ovata, thorough investigations of the molecular mechanisms of its exoPS synthesis and attachment will become feasible. Moreover, comparing the performance of WT S. ovata with a strain deficient in biofilm formation (e.g. GalU deletion mutant), will form an interesting strategy to further investigate the role of biofilm formation in biofilm-based gas fermentation reactors.

5. Conclusion

In this study, adaptive laboratory evolution (ALE) was applied with a selection pressure for cell attachment to enhance biofilm formation in the acetogenic bacterium Sporomusa ovata strain 2663. The resulting evolved culture, S. ovata 2663-BF, showed consistently increased cell attachment compared to the wild-type strain under both heterotrophic and autotrophic growth conditions. This biofilm-improved S. ovata did not have a better performance than the WT in simple trickle bed reactors, likely because the operational conditions (infrequently trickling) favored planktonic growth over cell attachment. Notably, the adapted S. ovata produced more extracellular polysaccharides, i.e. a common biofilm matrix component, which is likely due to a point mutation in the galU gene, which encodes the enzyme UDP-glucose pyrophosphorylase. Moreover, the improved cell attachment of the adapted S. ovata did not negatively affect its growth yields, indicating that energy for extracellular polysaccharide production may have been reallocated from traits associated with planktonic growth. Overall, this evolved S. ovata offers interesting opportunities to further study the importance of biofilm formation for gas fermentation systems. The benefit of biofilm growth is most likely depending on both the reactor design and operating conditions. Given S. ovata 's high performance in microbial electrosynthesis, the biofilm-improved strain is also of particular interest to unravel the role of cathodic biofilms in bioelectrochemical applications.

Author contributions

LVG, LMD and JP designed the experiments. LVG performed the adaptation procedure and the biofilm characterization. LVG and LMD performed the trickle bed experiment with continuous gas supply. LMD quantified the growth yields. YM and ME performed the microscopy work. LVG and LMD analyzed the experimental data and constructed the figures. IPGM performed the mutation analysis. NKJ and BEE performed the structural analysis of galU. LVG and JP wrote the manuscript. All other authors edited the manuscript. JP and KK acquired funding and JP, KK and MWVK performed supervision.

Funding and acknowledgements

This work was supported by the Novo Nordisk Foundation, as Biotechnology-based synthesis & production project grant (NNF19OC0057633). LMD was supported by a Green Transition Project I grant (MES‐MODEL, 1127‐00048B) from the Independent Research Fund Denmark (DFF). JP and YM were also supported by a grant from the Villum Foundation (VIL53083). IPGM was supported by the Danish National Research Foundation (DNRF136). Part the computing for this project was performed on the GenomeDK cluster. We thank Britta Poulsen, Miguel Oliviera, Simon Fruergaard and Nanna Jensen for their laboratory assistance. We also thank Heidi Skov Johansen for the exoPS quantification and her technical support. Thomas Dyekjær, Cecilie Nielsen, Gitte Hastrup and Linn Estel Avalon Sommer are also acknowledged for their technical support. A thanks also to Christian Stöhr GmbH & Co. for donating the carriers applied in this study. Finally, we thank Julia Doris Bechtner and Thomas William Seviour for their advice on the exoPS visualization and extraction methods.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests:Jo Philips reports financial support was provided by Novo Nordisk Foundation. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article:

Data availability

Raw sequencing data have been deposited in the NCBI sequence read archive under BioProject accession number PRJNA1081306.