Introduction

A value of BCoated is that our technology allows the induction of different properties of bacterial cellulose (BC), and allows the BC to be functionalised by binding different proteins, in line with the desired application. We developed a production platform that can facilitate this. To make it scalable, we focused on optimising BC production efficiency, increasing the possibilities of fine-tuning, and expanding the toolbox for these purpose. During the development of the production platform, these two goals were accomplished through the introduction of Saccharomyces cerevisiae to supplement Komagataeibacter sucrofermentans growth and metabolism, and to expand the programmability of BC. In the final consortium design that we set out to achieve, the presence of S. cerevisiae becomes advantageous by boosting the BC yield through ethanol supplementation, increasing the number of sugars that the bacteria can utilise with the enzyme invertase, and increasing the functionalisation possibilities through the incorporation of different sugar monomers, and the excretion of proteins that can bind to the cellulose^1-5.

The development of the production platform started with initiating the K. sucrofermentans culture and characterising its growth conditions. In particular, we evaluated how different pH and agitation rates can influence BC yield and water holding capacity.

Then, we looked at enabling the two organisms to live together in a stable consortium. To do that, we designed the organisms to depend on each other through a cross-feeding network (Figure 1). In this co-culture, S. cerevisiae consumes acetate, an acidic by-product that inhibits K. sucrofermentans growth and BC production². At the same time, S. cerevisiae produces ethanol, which K. sucrofermentans uses for energy generation¹. We modelled the behaviour of the strains both as monocultures and as a co-culture, to use laboratory data to predict the ideal conditions and test accordingly.

Finally, we sought to increase yield by supplementing the medium with ethanol. For a simplified production process, we aimed to use S. cerevisiae to produce the ethanol within the bioreactor. To maintain the ethanol concentration at the optimal value of 1% (v/v), we designed an ethanol homeostasis circuit. In this circuit, yeast metabolism is inhibited at higher concentrations of ethanol. Then, after ethanol is consumed by K. sucrofermentans, ethanol production restarts. This circuit is designed through iterative DBTL cycles combining wet and dry lab.

Experiments

Effect of different cultivation conditions on BC production

The production platform is a versatile tool for changing the properties of the BC. However, it is important to establish also the physical cultivation conditions, as they can influence BC structure.

The current production of functional BC is still facing several challenges, including challenging in situ functionalisation and sub-ideal bioreactors for high production yield⁶. With these experiments, we aimed to establish the ideal physical cultivation conditions to produce BC in our single-reactor production system. We looked at two key parameters: agitation and pH. General research on BC production is mainly focused on yield and strength. However, we identified water holding capacity as an additional important property for seed coatings.

Agitation

To investigate the effect of mechanical agitation on BC production and material properties, we grew K. sucrofermentans under three agitation rates in Erlenmeyer flasks with 100 mL total volume containing 25 g/L glucose in YPD medium. The medium was prepared according to the YPD media protocol. Low agitation (60 rpm), medium agitation (100 rpm), and high agitation (120 rpm) conditions were maintained using the Innova 2300 orbital shaker. We performed each condition in duplicates and incubated the flasks at 30°C for 7 days under aerobic conditions.

After incubation, we harvested the BC pellicles and subjected them to minimised downstream processing involving the removal of cellular debris and media components by washing with water. The resulting BC sheets were analysed for yield (dry weight determination), water-holding capacity (WHC), and rehydration capacity (RHC) using literature adapted BC protocols.

pH

We investigated the influence of pH on BC production using controlled bioreactor cultivation across four pH set points: 4.0, 5.0, 6.0, and 7.0. The bioreactors had a volume of 500 mL, and automated pH control. We added 3M NaOH as base addition to correct for acidification in the media. Each condition was oxygenated after 15 hours subsequent to inoculation with K. sucrofermentans. We maintained the cultures at 30°C, 30 rpm and 5 L/h of air for 160 hours. Real-time monitoring included continuous pH measurement, base dosage quantification, and glucose consumption through HPLC. The relative base dosage was calculated to serve as an indicator of bacterial metabolic activity and acid production.

After cultivation, we harvested the BC pellicles, processed through minimal downstream procedure as described above, and analysed them for yield, WHC, and RHC properties. SEM analysis was performed for structural characterisation, although material limitations prevented visibility of the internal structure for all pH conditions.

Establishing the cross-feeding and obtaining a basis for the co-culture

To enable K. sucrofermentans and S. cerevisiae to live in a stable consortium, we designed a cross-feeding network. To do that, we needed S. cerevisiae to not consume ethanol. We also planned to make yeast consume acetate from the medium and to ferment it into ethanol. However, within the time frame, we were not able to achieve the consumption of acetate in S. cerevisiae in the wet lab, but we modelled it.

Mitochondrial knockout for the fermentative S. cerevisiae strain.

S. cerevisiae has both fermentative and respiratory metabolism. As the function of the yeast in our consortium is to produce ethanol, we aimed to make all yeast metabolism fermentative, to ensure that growth is coupled to ethanol production. We used the obtained strains for both monoculture and ethanol production experiments.

We obtained respiration-deficient S. cerevisie strains by inducing the Rho- mutation. In this process, the cells were grown in consecutive generations on a medium containing EtBr, after which they lost their mitochondrial DNA⁷. We used S. cerevisiae CEN.PK113-5D, which additionally has a uracil deficiency, so that the uracil biosynthesis gene can be used as a plasmid selection marker.

The Rho- mutation was confirmed with replica plating, a stamping technique in which all colonies are transferred from an initial plate to a cloth and from there to two new plates, in this case with different carbon sources. Glycerol is non-fermentable, so colonies that grow on YPD and not on YPGlycerol have successfully lost their fermentative capabilities.

Growth Kinetics of Rho- S. cerevisiae and K. sucrofermentans on glucose and toxicity of ethanol and acetate

As a benchmark for the co-culture experiments, we performed monoculture growth experiments of both K. sucrofermentans and S. cerevisiae. This allowed us to obtain a baseline for growth characteristics and substrate tolerances of each organism independently. Additionally, it ensured that potential toxicities, metabolic preferences, and kinetic parameters were well understood before introducing the complexity of interactions within the co-culture.

For this, we needed to determine the kinetics of both members of the consortium depending on the relevant substrates. By mapping the results of how each strain responds to these substrates individually, we can make better initial predictions, and, later, interpretations of how the consortium behaves.

These experiments also generated quantitative data across gradients of the cross-feeding metabolites, providing input for parameterising (μ, lag time, carrying capacity) the model. By testing a concentration gradient for gluconate, ethanol, acetate, and the combined effect of gluconate and acetate, we obtained data on the inhibitory ranges of the cross-feeding and acidifying metabolites for both of the species in the co-culture. We assumed the biomass growth to be coupled with BC yield based on literature⁸. The BC yield could not be measured simultaneously with growth as cellulose forms on the air-liquid interface, affecting the OD600 measurements.

We prepared the 96 well plate for each species, according to the Monoculture growth assay protocols . We chose substrate concentrations based on reported optimal and inhibitory ranges for K. sucrofermentans and S. cerevisiae^9,10. These values ensured that physiologically relevant conditions were taken into account for both monocultures. Both K.sucrofermentans and (WT and Rho-) S. cerevisiae were inoculated on 2% YPD, a concentration gradient of (sodium) gluconate, ethanol, (potassium) acetate, and potassium acetate combined with constant (2%) gluconate.

We analysed the results with a script provided by Pieter Candry to fit the data to Richards growth function and to obtain average growth rates and other parameters, and implemented into the model to simulate the behaviour of the species.

Mitochondrial knock-out in a \Deltaglucose utilisation S. cerevisiae strain IMX1812

As mentioned above, we designed the co-culture to promote cross-feeding between S. cerevisiae and K. sucrofermentans. In this way, we would alleviate acidification caused by acetate, a by-product of BC production, and boost the energy metabolism of the bacterium by feeding electrons in the electron transport chain (ETC) to generate ATP. When intracellular ATP levels increase, glucose is redirected away from glycolysis and pentose phosphate pathway (PPP), towards the production of BC¹.

To allow this to happen, we needed to obtain a S. cerevisiae strain that cannot consume glucose, to prevent substrate competition with K. sucrofermentans. We found out that Wijsman et al., 2019¹¹ already generated it in the Department of Biotechnology in Delft University. Consequently, we contacted the corresponding author, Pascale Daran-Lapujade, who provided us with this strain: S. cerevisiae IMX1812 (parental strain: CEN.PK2-1C).

After that, we aimed to transform S. cerevisiae IMX1812 into a fermentative yeast. We made the yeast be deficient of respiration pathways to ensure that the ethanol produced is not consumed by itself. We achieved this purpose by using ethidium bromide (EtBr) to induce the loss of mitochondrial DNA (mtDNA). We performed this experiment according to the Ethidium Bromide Rho-mutation protocol, with the only difference of using maltose instead of glucose, as this strain was prepared to grow primarily on maltose.

We validated the mitochondrial knock out through replica plating on glycerol and maltose medium. We did this by transferring all of the colonies from the third round of the EtBr plates to both mediums (glycerol and maltose), using a stamping technique. Successful colonies could grow on maltose medium but not on glycerol medium, as glycerol cannot be consumed through fermentation.

Co-culturing Rho- S. cerevisiae IMX1812 and K. sucrofermentans

After implementing the growth kinetic data of the monocultures into the model, we were able to use it to predict the inoculum ratios for the co-culture. We prepared monocultures of S. cerevisiae and K. sucrofermentans as a control, to compare growth and evaluate BC yield without interaction.

We prepared the co-culture according to the Co-culturing protocol, and incubated it for four days. We took five time point samples: four on the day of incubation and one at the end of four days. Only the start and end data are presented in the results section as there was no difference in the peaks obtained from the samples taken during the same day.

After incubation, the samples were centrifuged at 6000 rpm for 12 minutes. The supernatant and pellet were frozen separately for High Pressure Liquid Chromatography (HPLC) analysis, to determine the metabolic composition, and for plating, to determine if both species survived.

The HPLC was carried out in an AMINEX hpx 87H column a flow rate of 0.6 ml/min at 60°C with 8 mM sulphuric acid and 100 mM isobutyric acid as an internal standard. For the detection, the refractive index detector was used since the yeast extract and peptone gave unidentifiable sample peaks in the UV detection method. Our compounds of interest were glucose, maltose, acetate, and ethanol. To obtain calibration curves, we prepared standards of 10, 5, 2, 1, 0.5, and 0.25 g/L for each compound of interest and plotted them against the area obtained under the curve (μRIU*min). We used the linear regression equations obtained from these calibration curves to calculate the concentrations in g/L.

At the end of four days, we washed BC according to the downstream processing protocol and freeze-dried it for 24 hours. After this, we used the BC quantification protocol to calculate the yields, titer and productivity of the co-culture against the monoculture and compared them.

S. cerevisiae ethanol homeostasis circuit

We designed an ethanol homeostasis circuit with the aim to regulate ethanol production by S. cerevisiae, which will benefit BC yield. The circuit works by inhibiting ethanol production when a concentration of 1% (v/v) is reached. To accomplish this, we used an ethanol-inducible promoter that regulates the expression of pTEV+. pTEV+ is a targeted protease that causes the degradation of proteins that are tagged with an inactive bidirectional degron. This will be fused to a key enzyme in ethanol production, either PGK1 or GPM1 (Figure 2A). This way, the tagged proteins are degraded when the ethanol-inducible promoter is activated, which limits ethanol production.

The development of the ethanol homeostasis circuit consisted of several steps. First, we developed and characterised a fermentative strain. The characterisation data was used to fine-tune the circuit model. After that, we designed the synthetic circuit using model outputs as guidelines for design choices. In the final production platform, the entire synthetic circuit will be present in the cell, but during the development, the circuit was split up into two parts (Figure 2B&C). The signal transducer and actuator were designed so that their effect upon different levels of induction can be assessed. Separately, ethanol-inducible promoters were developed that respond differently to ethanol concentrations in the medium.

Characterisation of ethanol production by fermentative S. cerevisiae

After inducing the rho- mutation in S. cerevisiae CEN.PK113-5D, we performed a growth assay to characterise ethanol production and to generate data for model calibration. An overnight culture was grown before the experiment. The next morning, we used this culture to inoculate 100 mL Erlenmeyer flasks with different glucose concentrations to an OD600 of 0.1. The glucose concentrations were 2, 5 and 10 % (w/v), all in triplicate. We measured the OD600 at nine time points: eight every hour during that day and a last after 24 hours. Additionally, samples were collected to be analysed with HPLC. We loaded 200 μl of each spun down sample into a 96-well plate. The samples were analysed for glucose and ethanol concentration in the Dionex HPLC with an AMINEX HPX-87H mixed acid column (Thermoscientific). A calibration curve was used to correlate the peak surfaces to ethanol and glucose concentrations.

Signal transducer and actuator

To reach the aim of the circuit, ethanol production needs to be reversibly inhibited when the threshold ethanol concentration is reached. For this purpose, we chose an approach based on targeted degradation of key enzymes in lower glycolysis. This starting metabolite is chosen to allow for ethanol production inhibition during growth on different sugars. Through an S. cerevisiae CEN.PK113-7D genome analysis with the S. cerevisiae Genome Database¹², we identified two target genes that have only a single gene copy: phosphoglycerate kinase (PGK1) and phosphoglycerate mutase (GPM1). They catalyse the formation of 3-phospho-D-glycerate from 3-phospho-D-glyceoyl phosphate and the formation of 2-phospho-D-glycerate from 2-phospho-D-glycerate, respectively. Our model predicted that the circuit performed well independently of the targeted glycolytic protein. Based on this, we decided to continue the experimental design with only one target, namely PGK1.

The actuator of the synthetic circuit is the inducible degradation of PGK1. To achieve the inducibility, a bidirectional inactive degron was fused to this gene in the genome with a non-commercial CRISPR/Cas9 toolkit from the Ellis lab at Imperial College London. We designed the sgRNA to direct Cas9 to cut near the end of PGK1. The sgRNA was cloned into the entry vector with BsmBI using Golden Gate assembly. The Cas9 and sgRNA plasmid vectors were linearised with BsmBI and EcoRV, respectively, before yeast transformation. Simultaneously, we assembled a homology-directed repair (HDR) template from the ordered bidirectional degron sequence from Jungbluth et al., 2010¹³ and we amplified genomic fragments from S. cerevisiae CEN.PK113-5D using Q5 PCR (Q5 high-fidelity polymerase, Thermoscientific). The PCR fragments had extensions with BsaI recognition sites and ssDNA overhangs from the Yeast MoClo toolkit for their orthogonality¹⁴. Additionally, SapI restriction sites were added at the start of the gene fragment and the end of the downstream region. We performed Golden Gate assembly into the MoClo part entry plasmid using BsaI for insert digestion and Esp3I for entry plasmid digestion. Prior to transformation, the HDR template was digested with SapI to remove the plasmid backbone.

We transformed a S. cerevisiae CEN.PK113-5D Rho- culture with the linearised Cas9 vector, sgRNA and homology-directed repair template (Figure 3). The transformed cells were selected for nourseothricin resistance. To confirm the presence of the assembly, we performed colony PCR.

While creating the actuator, we also developed the signal transducer. The signal transducer is the targeted protease pTEV+, which specifically recognises the degron through the pTEV+ binding domain and cuts the pTEV+ cleavage site (Figure 3). After cleavage, the inactive C- and N-terminal degron are exposed, thus activated, targeting the protein for degradation¹³. The signal transducer plasmids, aimed at creating a large diversity in pTEV+ expression, are the pTEV+_Z₃EV plasmids (Figure 4A). They contain the pTEV+ coding sequence from previous literature by Taxis et al., 2009¹⁵, under the regulation of the Z₃EV inducible promoter (Z₃EVpr), obtained from McIsaac et al., 2013¹⁶.McIsaac et al., 2013¹⁶ also constructed the Z₃EV coding sequence. Two different designs contained Z₃EV behind the medium activity promoter pRPL18b and the high activity promoter pTDH3 from the yeast MoClo toolkit¹⁴ (Figure 4B).

When β-estradiol is added to a yeast cell that is transformed with this construct, the β-estradiol complexes with Z₃EV and interacts with the Z₃EV inducible promoter to express pTEV+. Since ethanol production should only be inhibited temporarily, we hypothesised that pTEV+ should also be active for only a short period of time. To test whether destabilising pTEV+ can increase the dynamic flexibility of the system, we made two different designs, in which a weak degron (Ubi-M) or a medium degron (Ubi-Y) is fused to pTEV+ (Figure 4C).

We constructed the Z₃EV plasmids in two consecutive steps. In the first step, cassette plasmids were constructed from MoClo parts and the sequences for the Z₃EV gene, the Z₃EVpr and the pTEV+ gene. These DNA fragments are ordered from TWIST with extensions for BsaI restriction sites that create ssDNA overhangs compatible with MoClo parts 3, 2 and 3b, respectively¹⁶. Five different cassette plasmids were constructed through Golden Gate cloning (GGC) with BsaI (New England Biolabs); two contained the Z₃EV transcriptional unit with the different promoters, the second set contained the pTEV+ transcriptional unit with the different degradation tags, and the fifth plasmid contained the backbone. The backbone had the low copy number origin of replication CEN/ARS, the uracil biosynthesis gene as a selection marker for yeast and a GFP dropout where the transcriptional units can be inserted. All plasmids contained E. coli high copy number origin ColE1 and an antibiotic resistance gene as a selection marker. The E. coli selection marker is AmpR for the transcription units and KanR for the dropout vector. This allowed selection of the final vector after the second cloning step, where the transcriptional units are cloned together into the backbone with Esp3I (New England Biolabs) according to the GGC protocol. All four final vectors were confirmed by sequencing.

Ethanol inducible promoter development

Since the modelling showed the importance of the sensor domain on the functioning of the ethanol homeostasis circuit, we developed a set of ethanol inducible promoters with different dynamic ranges. These promoters are based on the KlADH4 promoter, an ethanol inducible promoter previously used by iGEM12_TU_Munich (BBa_K801020)¹⁷, and designed according to the S. cerevisiae promoter architecture described by Tang et al., 2020¹⁸. They described an inducible promoter as consisting of three parts; a nucleosome disfavouring sequence that acts as an insulator sequence to decrease background transcription, regulatory elements and a core promoter. The pMT951 sequence developed by Tominaga et al., 2024¹⁹, a partial sequence of KpARG4, was used as an insulator sequence. The regulatory elements interact with transcription factors (TFs). They are called UAS if they are activated by TFs and URS if they are involved in repression. An increased number of regulatory element repeats can enhance the activation or repression^18,19. This guided us in our design, where we incorporated one, two, or three repeats of the regulatory element UAS_E to alter the ethanol response of the promoter. The core promoter can enhance promoter activity 2.5 fold¹⁸. To develop promoters with varying dynamic range, we did not just use the KlADH4 core promoter, but we also selected two core promoters from the paper by Wang et al., 2018²⁰: low expression core promoter CYC1 (position -158 to -1 5’ of start codon) and high expression core promoter TYF1 (position -197 to -1 5’ of start codon). Decreasing the distance between the UAS and the TATA box to 40 bp can further increase the promoter activity¹⁹. Therefore, not only the KlADH4 core promoter, but also a shorter fragment, starting at 40 base pairs upstream from the TATA box, was used as a core promoter. The promoters were designed by combining the insulator sequence with a varying number of UAS_E repeats and a core promoter element (Figure 5). The different elements can be combined in twelve ways, resulting in twelve different promoters.

All parts were ordered through Twist Bioscience or PCR amplified with restriction sites that create ssDNA overhangs that make it suitable for GGC with BsaI. The overhangs were copied from the MoClo toolkit, as these overhangs have been tested for orthogonality. ce_CYC1 and ce_TEF1 were amplified from CEN.PK113-5D genomic DNA with Q5 high fidelity polymerase (Thermoscientific). We PCR amplified UAS_E repeats and KlADH4 core promoter elements from the KlADH4 promoter with Q5 high fidelity polymerase (Thermoscientific). UAS_E repeats were amplified as six single UAS_E’s with different overhangs to enable combining a different number of repeats in each GGC reaction. The promoters were assembled directly in the promoter reporter vector through GGC with BsaI (New England Biolabs). In total, we constructed twelve different promoters. After obtaining the correct plasmids, these are used to transform CENPK113-5D wild-type. This was the final step achieved before Wiki freeze, but further experiments to assess the dynamic range of all twelve promoters will be performed.

Ethanol inducible promoter characterisation

To determine the dynamic range of the promoters, we used a linearised reporter plasmid that was previously cloned in the Wageningen University and Research Systems & Synthetic Biology laboratory (Figure 6). This promoter reporter plasmid contains a Yellow Fluorescent Protein gene (YFP) behind an insertion site, where a promoter with MoClo part 2 overhangs can be inserted in a Golden Gate assembly reaction with BsaI (New England Biolabs). In addition, the vector contains a reference fluorescence gene cassette with mCherry under the constitutive TEF2 promoter. This reference fluorescent protein allows correction of YFP fluorescence not only for the number of cells, but also for the plasmid copy number. The backbone contains E. coli AmpR resistance gene and high copy number origin ColE1. For S. cerevisiae the vector contains the low copy number origin CEN/ARS and the URA3 selection marker.

After assembly, the correct plasmid was transformed into S. cerevisiae CEN.PK113-5D wild type. We used this strain for the dynamic range assay, since it can grow on the non-fermentable sugar source glycerol.

In the dynamic range assay, we grew the transformed yeast in different ethanol concentrations. An overnight culture of the strain was first made. Five hours before the experiment starts, the strain was pre-grown in ethanol concentrations of 0%, 1%, 2% or 4%. We performed this pre-incubation step to assess the promoter activity decay time. After five hours, the cultures were spun down, redissolved in sterile demi-water and used to load a 96-well plate to a starting OD600 of 0.2.

Cell growth and fluorescence were measured in ethanol concentrations in a range of 0% to 10%. The plate was loaded in the BioTek Synergy H1 plate reader (Agilent). Every five minutes for a duration of 24 hours, three measurements were taken: the OD600, mCherry fluorescence with an excitation of 587 nm and emission of 620 nm and YFP with excitation at 508 nm and emission at 535 nm. All measurements were corrected for medium blanks and a CEN.PK113-5D strain transformed with an empty vector is used as a negative control. We loaded all the samples in triplicate for a statistical analysis of the results.

At the time of the Wiki freeze, the promoter reporter assays were not yet optimised. The chosen excitation and emission wavelength for YFP had an overlap in the previous assay, resulting in higher values for the measurements of medium blanks than for most samples. Because of this, the results were deemed unreliable. Further experiments with other excitation and emission wavelengths will be conducted after the Wiki freeze.

Results

Effect of different cultivation conditions on BC production

BC, as the matrix for our seed coating, requires modular properties induced by biological and physical stimuli. The production of BC requires a mixture of K. sucrofermentans and a sugar-rich feedstock. Here, YPD (Yeast Extract Peptone Dextrose) medium was used due to its consistency and defined composition. However, a wide variety of waste-streams are suitable for the production of BC.

Production and downstream processing

Through our initial experiments, we confirmed that BC was produced in YPD after 7 days of static incubation (Figure 7A). BC sheets demonstrated structural integrity and flexibility. Downstream processing removed debris and resulted in a consistent and translucent sheet (Figure 7B). Downstream processing efficiency was also examined using SEM, which demonstrated that washing the sheets exposed the internal structure of BC with visible cellulose fibrils. Furthermore, K. sucrofermentans cells were no longer visible after this process (Figure 7B&D).

Effect of agitation on the yield and water capacity of bacterial cellulose

We evaluated the effect of three agitation intensities, corresponding to 60 rpm (low), 100 rpm (medium), and 120 rpm (high) in Erlenmeyer flasks. As a result, the BC samples ranged from sheet-type structures, to irregular clumps, to small grains for low, medium, and high agitation intensities, respectively (Figure 8).

Agitation revealed an optimum yield at medium intensity, showing a maximum yield of 1.48 ± 0.28 g dry weight, representing an 85% increase over low agitation (0.80 ± 0.43 g) and 310% increase over high agitation (0.36 ± 0.15 g) (Figure 9A). In contrast to the yield, the medium agitation material is outperformed, in both water-holding capacity (WHC) and rehydration capacity (RHC), by low and high agitation (Figure 9B).

The relation between agitation intensities indicates that insufficient mixing and excessive shear negatively impact BC production. However, it is a trade-off between high yield and WHC before and after freeze-drying.

Effect of pH on the yield and water capacity of bacterial cellulose

We measured pH as an additional variable to control the properties of BC. For this, set-points were chosen for pH 4.0, 5.0, 6.0 and 7.0. The run was stabilised for 15 hours, after which the production started. Deviations from the pH set-point are due to overshoots of the base-dosing mechanism, which require the acidification by K. sucrofermentans to stabilise the pH again over time (Figure 10A).

The relative dosage of base is spread over the entire culture period (Figure 10B). The amount of base added was the highest in pH 5, with pH 6 and 7 as a follow up. A limited amount of base was added at pH 4.

Yield data indicated that a pH of 5 is optimal, followed by pH 4 (Figure 10C). Higher pH (6 and 7) showed only minimal production of BC (Figure 11). While yield was low at higher pH, first tests indicate that the RHC was higher at these pHs (Figure 10D).

SEM examination was initially planned for porosity determination, but the raw material was not down-stream processed due to the little material availability. This prevented the possibility to investigate the internal structure as shown in (Figure 7D).

Establishing the cross-feeding and obtaining a basis for the co-culture

Mitochondrial knockout for the fermentative S. cerevisiae strain

According to the EtBr induced Rho- mutation protocol, three rounds of growth on EtBr supplemented medium are needed to induce the Rho- mutation. We confirmed the loss of mitochondrial DNA by replica plating on glycerol and glucose-containing medium. During the experiment, the second growth cycle was also replica plated. Very few S. cerevisiae CEN.PK113-5D cells retained their ability to grow on glycerol (Figure 12). This demonstrated that two rounds of growth on EtBr containing medium were sufficient to induce the Rho- mutation in most cells.

We confirmed the Rho- mutation with a growth experiment with pure colonies. An overnight culture of the colonies was made in YPD. We inoculated this culture in YPD with 2% glucose and YPGlycerol with 2% glycerol. We measured the OD600 right after inoculation and after overnight growth. The values are listed in (Table 1). The growth that the Rho- cultures showed in YPGlycerol is negligible and could be due to leftover sugars from the overnight culture. This showed that the Rho- mutation was successfully induced. We further developed the ethanol homeostasis circuit in CEN.PK113-5D Rho-1, later referred to as CEN.PK113-5D Rho-.

Table 1: Growth experiment to confirm the Rho- mutation. OD measurements were taken right after inoculation of the growth medium and after overnight growth.

Growth Kinetics of Rho- S. cerevisiae and K. sucrofermentans on glucose and toxicity of ethanol, acetate and gluconate

The general trend for average growth rate of K. sucrofermentans (Figure 13A) started decreasing above 1.4% (v/v) ethanol. This value is in line with previous literature, as 0.5-1% (v/v) enhances cellulose production by supporting the energy metabolism, while inhibition occurs above 3-4% (v/v)¹⁰. The higher growth rate seen for 4.2% (v/v) might be influenced by contamination, supported by the longer lag time seen for that sample. These concentration ranges were chosen to be relevant for the possible amount within the co-culture medium that might be produced by the fermentative S. cerevisiae. This shows that the ethanol circuit is necessary to ensure that inhibition due to high ethanol concentrations will not occur.

In the concentration gradient for gluconate (Figure 13B), there is a pattern of increase in growth rate up to 2%. A clear decline was observed afterwards. The acetate concentration series (Figure 13C) had a similar behaviour to the gluconate series, except for the 1.5% (w/v) which appears to have a μ of around 0.01. We noted that replicates for this sample did not have a reliable Richards fit. Looking at the combined acidity samples (Figure 13D), the highest growing sample was the 0.5% acetate sample, in which there was 2.5% acid concentration in total. Therefore, we saw the same overall trend where the maximum growth was achieved when there was 2-2.5 % acid in the medium.

We thought that the cells did not grow in the ethanol series, due to the especially low growth rates seen for the ethanol series. However, the carrying capacities of the ethanol samples reached an optical density of 1.0 and were within 0.1 range like other samples. This might suggest that the rate was especially low due to stress, extending the lag time within the exponential phase. This is supported by comparing the growth curves (OD vs time) of the ethanol samples with the rest, where the same OD is reached four hours later.

Overall, as expected, we observed much higher average growth rates (μ) per hour for the S. cerevisiae samples (Figure 14). Again, the ethanol series (Figure 14A) showed lower μ than the rest, with a pattern of slower growth with increasing concentration. We do not observe a trend for the acids.

As mentioned before, we also tested and compared the growth kinetics of the WT S.cerevisiae and the S. cerevisiae lacking mitochondrial function. From (Figure 15), we calculated that the ratio between their average growth rates and doubling times was 1.6.

The growth parameters obtained from these experiments were fed into the model.

Mitochondrial knock-out in a Δglucose utilisation S. cerevisiae strain

Figure 16A shows the three rounds of EtBr platings. We selected the colonies from each round by choosing the smaller ones, since it is an indication of less energy and thus loss of metabolic function. The replica plating (Figure 16B) showed that very few colonies retained their ability to grow on glycerol, which is consumed through respiration in S. cerevisiae. The colonies on maltose grew noticeably less than the colonies on the glycerol plates, supporting the fact that less energy was able to be produced in the Rho- colonies.

Co-culturing Rho- S. cerevisiae IMX1812 and K. sucrofermentans

We compared the BC production within the consortium to the K. sucrofermentans monoculture. We calculated yield, titer and productivity as BC production assessments. However, there was no remarkable increase in cellulose production in the co-culture (Figure 17&18).

Looking at the co-culture sample plated on the YPD+YPM plates, it appears that, after four days, both species survived. This was evaluated as two phenotypically distinct species could be observed, matching the phenotypes previously demonstrated for the monocultures. S. cerevisiae colonies have a whiter hue, convex and round colonies, while the K.sucrofermentans colonies have a yellow undertone, irregular shape, and are flat/smooth (Figure 19).

The HPLC analysis results (Figure 20) showed that glucose was consumed and acetate was produced in both the monoculture and co-culture. Due to the time point samples not being spread across the four days, we cannot determine if K. sucrofermentans is consuming the ethanol produced by S. cerevisiae. However, ethanol was produced in the medium. We think that the small amount of maltose at the start of the monoculture is carryover, since maltose was not added to this medium.

We did expect ethanol to be produced in the monoculture samples. We proposed two hypotheses to account for this. The first one is that there is a metabolic pathway in K. sucrofermentans able to produce ethanol, that we are not aware of. The second one is that the supernatant of the monoculture sample had a small amount of S. cerevisiae cells remaining, that consumed the carry over maltose and produced ethanol. To test this hypothesis, we modelled this small amount of contamination and the levels of acetate and ethanol that would be the outcome of this. The model supported that this would yield more ethanol than acetate, suggesting that this hypothesis might be likely.

Learn more about this in our engineering page

S. cerevisiae ethanol homeostasis circuit

Characterisation of ethanol production by fermentative S. cerevisiae

The first thing we did with the produced S.cerevisiae CEN.PK113-5D rho- strain was to characterise its growth, glucose consumption and ethanol production. The results of the growth experiment are showed in (Figure 21). Growth rates were calculated from the OD600 between one and four hours. They were 0.36 h^-1, 0.41 h^-1 and 0.38 h^-1 for respectively 2 (w/v)%, 5 (w/v)% and 10 (w/v)% glucose. These growth rates were a bit lower than WT yeast growth rates, which can reach 0.44 h^-1, but this is an expected effect of the rho- mutation²¹. The final OD600 was lower when the initial glucose concentration was 2% (w/v), with OD600 values of 1.53, 1.83 and 1.86 after 22 hours of growth for 2, 5 and 10 % (w/v). This was probably due to the fact that the glucose in the medium was depleted sooner than for the 5% (w/v) glucose cultures. When the glucose concentration at inoculation was 10% (w/v), glucose remained present for the entire duration of the experiment.

An ethanol concentration above the desired value of 1% (v/v) was reached at glucose concentrations of 2 and 5% (w/v). This highlights the importance of the ethanol homeostasis circuit to limit ethanol production during prolonged growth on a variety of media. The growth data was used to fine-tune the null model of the digital twin of the ethanol homeostasis circuit. When fine-tuning the model with experimental data, the parameter values were adjusted to better match the experiment. This increased the predictive ability of the model.

Conclusions and Outlook

As a team, we put a step forward into the development of a stable production platform, in which BC can be functionalised while being produced at increased yield.

By changing pH and agitation conditions, we could influence BC characteristics, but at the cost of a decline in yield. Our team is already working on engineering K. sucrofermentans to increase BC yield. However, due to time constraints, we were unable to obtain the desired data. However, after obtaining this engineered strain, we plan to use it for production and to counteract the decline in yield for higher pH and agitation.

We co-cultured K. sucrofermentans and S. cerevisiae and we demonstrated that they can survive together, but the robustness or the contribution to yield was not proved. We plan to test this consortium behaviour in longer periods and with the engineered S. cerevisiae for acetate consumption. Additionally, we obtained data on growth kinetics. This data was used train a model on the behaviour of the individual strains and will be useful to predict ideal lab conditions.

As for the design of an ethanol homeostasis circuit, we sucessfully developed a non-respiratory S.cerevisiae strain and we assembled plasmids that contain all the genetic elements that are needed for the circuit domain. These plasmids were transformed into fermentative S. cerevisiae, in the case of the signal transducer and actuator domain, or respiratory S. cerevisiae, in the case of the sensor domain. However, we still need to assess the effect of the signal transducer and actuator domain. We plan to do this with a growth assay. In addition, we will use HPLC to determine ethanol concentrations.

During the development of the ethanol homeostasis circuit, there was a strong interaction between modelling and experiments. We learned how experimental data can be used to refine models and increase their predictive accuracy. In addition, computational findings can be used to improve the design of wet-lab experiments. By combining the two, an informed selection of parameters can be made to design variants and test their function. This decreases the experimental time needed to develop a functional synthetic circuit.

(1)

Montenegro-Silva, P.; Ellis, T.; Dourado, F.; Gama, M.; Domingues, L. Enhanced Bacterial Cellulose Production in Komagataeibacter sucrofermentans: Impact of Different PQQ-Dependent Dehydrogenase Knockouts and Ethanol Supplementation. Biotechnology for Biofuels and Bioproducts 2024, 17. https://doi.org/10.1186/s13068-024-02482-9.

(2)

Kalashnikova, O.; others. Production of Bacterial Cellulose Using a Symbiotic Culture of Bacteria and Yeast (SCOBY). Carbohydrate Research 2024, 506, 108368. https://doi.org/10.1016/j.carres.2024.108368.

(3)

Badotti, F.; Dário, M. G.; Alves, S. L.; Cordioli, M. L. A.; Miletti, L. C.; Araujo, P. S. de; Stambuk, B. U. Switching the Mode of Sucrose Utilization by Saccharomyces cerevisiae. Microbial Cell Factories 2008, 7. https://doi.org/10.1186/1475-2859-7-4.

(4)

Gilbert, C.; Tang, T.-C.; Ott, W.; Dorr, B. A.; Shaw, W. M.; Sun, G. L.; Lu, T. K.; Ellis, T. Living Materials with Programmable Functionalities Grown from Engineered Microbial Co-Cultures. Nature Materials 2021, 20 (5), 691–700. https://doi.org/10.1038/s41563-020-00857-5.

(5)

Singh, A.; Walker, K. T.; Lee, K. P.; Sinskey, A. J.; Stephanopoulos, G. Engineering Bacterial Cellulose by Synthetic Biology. International Journal of Molecular Sciences 2020, 21 (23), 9185. https://doi.org/10.3390/ijms21239185.

(6)

Shishparenok, A. N.; Furman, V. V.; Dobryakova, N. V.; Zhdanov, D. D. Protein Immobilization on Bacterial Cellulose for Biomedical Application. Polymers 2024, 16 (17), 2468. https://doi.org/10.3390/polym16172468.

(7)

Goldring, E. S.; Grossman, L. I.; Krupnick, D.; Cryer, D. R.; Marmur, J. The Petite Mutation in Yeast: Loss of Mitochondrial DNA During Induction with Ethidium Bromide. Journal of Molecular Biology 1970, 52 (2), 323–335. https://doi.org/10.1016/0022-2836(70)90033-1.

(8)

Rincón, A.; Hoyos, F. E.; Candelo-Becerra, J. E. Dynamic Modeling of Bacterial Cellulose Production Using Combined Substrate- and Biomass-Dependent Kinetics. Computation 2024, 12 (12), 239. https://doi.org/10.3390/computation12120239.

(9)

Karničnik, B.; Accetto, T.; Fanedl, L.; Jugović, I.; Trček, J. Isolation and Characterization of Komagataeibacter piraceti sp. nov. and Novacetimonas labruscae sp. nov.: Two Novel Microaerobic Cellulose-Producing Acetic Acid Bacteria from Vinegars. Microorganisms 2025, 13 (2), 456. https://doi.org/10.3390/microorganisms13020456.

(10)

Mikkelsen, D.; Flanagan, B. M.; Dykes, G. A.; Gidley, M. J. Influence of Different Carbon Sources on Bacterial Cellulose Production by Gluconacetobacter xylinus Strain ATCC 53524. Journal of Applied Microbiology 2009, 107 (2), 576–583. https://doi.org/10.1111/j.1365-2672.2009.04226.x.

(11)

Wijsman, M.; Świat, M. A.; Marques, W. L.; Hettinga, J. K.; van den Broek, M.; de la Torre Cortés, P.; Mans, R.; Pronk, J. T.; Daran, J.-M.; Daran-Lapujade, P. A Toolkit for Rapid CRISPR-SpCas9 Assisted Construction of Hexose-Transport-Deficient Saccharomyces cerevisiae Strains. FEMS Yeast Research 2019, 19 (1), foy107. https://doi.org/10.1093/femsyr/foy107.

(12)

Cherry, J. M.; Hong, E. L.; Amundsen, C.; Balakrishnan, R.; Binkley, G.; Chan, E. T.; Christie, K. R.; Costanzo, M. C.; Dwight, S. S.; Engel, S. R.; Fisk, D. G.; Hirschman, J. E.; Hitz, B. C.; Karra, K.; Krieger, C. J.; Miyasato, S. R.; Nash, R. S.; Park, J.; Skrzypek, M. S.; Simison, M.; Weng, S.; Wong, E. D. Saccharomyces Genome Database: The Genomics Resource of Budding Yeast. Nucleic Acids Research 2012, 40 (D1), D700–D705. https://doi.org/10.1093/nar/gkr1029.

(13)

Jungbluth, M.; Renicke, C.; Taxis, C. Targeted Protein Depletion in Saccharomyces cerevisiae by Activation of a Bidirectional Degron. BMC Systems Biology 2010, 4. https://doi.org/10.1186/1752-0509-4-176.

(14)

Lee, M. E.; DeLoache, W. A.; Cervantes, B.; Dueber, J. E. A Highly Characterized Yeast Toolkit for Modular, Multipart Assembly. ACS Synthetic Biology 2015, 4 (9), 975–986. https://doi.org/10.1021/sb500366v.

(15)

Taxis, C.; Stier, G.; Spadaccini, R.; Knop, M. Efficient Protein Depletion by Genetically Controlled Deprotection of a Dormant n-degron. Molecular Systems Biology 2009, 5 (1), 1–7. https://doi.org/10.1038/msb.2009.25.

(16)

McIsaac, R. S.; Oakes, B. L.; Wang, X.; Dummit, K. A.; Botstein, D.; Noyes, M. B. Synthetic Gene Expression Perturbation Systems with Rapid, Tunable, Single-Gene Specificity in Yeast. Nucleic Acids Research 2013, 41 (4), e57. https://doi.org/10.1093/nar/gks1313.

(17)

Heinze, S. (iGEM12_TU_Munich). BBa_K801020, 2012. https://igem.org/.

(18)

Tang, H.; Zhang, Y.; Keasling, J. D.; Zhang, F.; Lee, S.-H.; Wei, N.; Zhao, H. Promoter Architecture and Promoter Engineering in Saccharomyces cerevisiae. Trends in Biotechnology 2021, 39 (3), 266–281. https://doi.org/10.1016/j.tibtech.2020.10.004.

(19)

Tominaga, M.; Shima, Y.; Nozaki, K.; Ito, Y.; Someda, M.; Shoya, Y.; Hashii, N.; Obata, C.; Matsumoto-Kitano, M.; Suematsu, K.; Matsukawa, T.; Hosoya, K.; Hashiba, N.; Kondo, A.; Ishii, J. Designing Strong Inducible Synthetic Promoters in Yeasts. Nature Communications 2024, 15 (1), 10653. https://doi.org/10.1038/s41467-024-54865-z.

(20)

Wang, J.; Zhai, H.; Rexida, R.; Shen, Y.; Hou, J.; Bao, X. Developing Synthetic Hybrid Promoters to Increase Constitutive or Diauxic Shift-Induced Expression in Saccharomyces cerevisiae. FEMS Yeast Research 2018, 18 (8), foy098. https://doi.org/10.1093/femsyr/foy098.

(21)

Paalme, T.; others. Growth Efficiency of Saccharomyces cerevisiae on Glucose/Mineral Media: The Effects of the Ethanol/Glucose Ratio on Growth Characteristics. Journal of Biotechnology 1997, 54 (3), 200–208. https://doi.org/10.1016/S0141-0229(96)00114-7.