An artificial coculture fermentation system for industrial propanol production

Abstract

Converting plant biomass into biofuels and biochemicals via microbial fermentation has received considerable attention in the quest for finding renewable energies and materials. Most approaches have so far relied on cultivating a single microbial strain, tailored for a specific purpose. However, this contrasts to how nature works, where microbial communities rather than single species perform all tasks. In artificial coculture systems, metabolic synergies are rationally designed by carefully selecting and simultaneously growing different microbes, taking advantage of the broader metabolic space offered by the use of multiple organisms. 1-propanol and 2-propanol, as biofuels and precursors for propylene, are interesting target molecules to valorize plant biomass. Some solventogenic Clostridia can naturally produce 2-propanol in the so-called Isopropanol–Butanol–Ethanol (IBE) fermentation, by coupling 2-propanol synthesis to acetate and butyrate reduction into ethanol and 1-butanol. In this work, we hypothesized propanoate would be converted into 1-propanol by the IBE metabolism, while driving at the same time 2-propanol synthesis. We first verified this hypothesis and chose two propionic acid bacteria (PAB) strains as propanoate producers. While consecutive PAB and IBE fermentations only resulted in low propanol titers, coculturing Propionibacterium freudenreichii and Clostridium beijerinckii at various inoculation ratios yielded much higher solvent concentrations, with as much as 21 g/l of solvents (58% increase compared to C. beijerinckii monoculture) and 12 g/l of propanol (98% increase). Taken together, our results underline how artificial cocultures can be used to foster metabolic synergies, increasing fermentative performances and orienting the carbon flow towards a desired product.

A coculture approach combining solventogenic Clostridia and PAB for microbial propanol production.

Introduction

In the context of the energy transition, research on plant biomass as renewable feedstock for biofuel and biochemical production has intensified over the recent years. Among different approaches, microbial fermentation has in particular received attention for its potential to convert plant carbohydrates directly into a wide panel of value-added products.

Propanol is an interesting target molecule, as it can be used both as a biofuel or as a biochemical (notably as an intermediate for the synthesis of biopropylene; Walther and François 2016). One of the two isomers, 2-propanol, is naturally fermented by a few solventogenic Clostridia strains through the Isopropanol–Butanol–Ethanol (IBE) fermentation (Chen and Hiu 1986). Using these atypical strains, 2-propanol can be produced in moderate quantities (≈ 5 g/l), insufficient for using them directly in an industrial process. Hence, efforts have been undertaken to increase performances, by either characterizing (Máté de Gérando et al. 2018; Hocq et al. 2019) and genetically engineering natural IBE strains (Diallo et al. 2020; Li et al. 2020), or by transferring the 2-propanol pathway to other suitable microorganisms by metabolic engineering approaches (Hanai et al. 2007; Jojima et al. 2008; Collas et al. 2012; Dusséaux et al. 2013; Tamakawa et al. 2013).

Using microbial coculture systems to enhance fermentative processes emerged as an alternative approach (Bader et al. 2010; Arora et al. 2020; Du et al. 2020; Sgobba and Wendisch 2020). Indeed, rationally combining different microorganisms can dramatically expand the metabolic repertoire of a microbial culture. Exploiting metabolic synergies can in turn have significant benefits e.g. converting difficult substrates (e.g. lignocellulosic biomass; Zuroff et al. 2013), increasing fermentative performances (Luo et al. 2017), and/or producing novel interesting metabolites (Rateb et al. 2013).

Coculture systems have already been demonstrated to produce propanol in at least two different systems. In the first, the classical Acetone–Butanol–Ethanol (ABE) fermentation was turned into an IBE fermentation by cocultivating Clostridium acetobutylicum, an ABE- and H₂/CO₂-producing organism, with Clostridium ljungdahlii, an acetogen growing on H₂ and CO₂, and able to reduce acetone to 2-propanol (Charubin and Papoutsakis 2019). In the second, a contamination arose during a continuous fermentation experiment of Alkalibaculum bacchi, an acetogen converting syngas into ethanol (Liu et al. 2014a). At a late stage of the process, ethanol concentration suddenly decreased, and 1-propanol appeared as a novel metabolite. Further experiments revealed the additional microorganism responsible for this phenomenon to be Clostridium propionicum, which converts ethanol into propionic acid, subsequently reduced by A. bacchi into 1-propanol (Liu et al. 2014b).

Hereafter, we describe a novel IBE Clostridium—propionic acid bacteria (PAB) coculture system aiming at complementing 2-propanol production with the biosynthesis of 1-propanol, a process that requires the concerted action of both microorganisms. This simple system allowed the production of up to 12 g/l of propanol, increasing propanol content in the solvent mixture from 45% to 62%. This setup also surprisingly resulted in an increase of the total concentration of solvents (up to 58% more—21 g/l), a concentration never reached before with an IBE strain batch monoculture.

Material and methods

Microorganisms, medium, and growth conditions

Clostridium beijerinckii strains (strain A: DSM 6423, strain B: DSM 15410) and PAB strains (Acidipropionibacterium acidipropionici DSM 4900, Propionibacterium freudenreichii DSM 4902) were grown in an anaerobic chamber (Baker Ruskinn) in modified 2YT medium (2YTm). This medium contained, per liter: 16 g tryptone, 10 g yeast extract, 4 g NaCl, 2 g KH₂PO₄, 0.05 g MnSO₄, 1 g MgSO₄, 7 H₂O, 2.9 g ammonium acetate, 0.007 g FeSO₄, 7 H₂O, and 0.1 g p-aminobenzoic acid. Glucose, sucrose, sodium propanoate, and/or glycerol were added depending on the experiment.

For fermentation assays, 125 ml serum bottles filled with 20–45 ml 2YTm were used. Unless otherwise stated, these were inoculated inside the anaerobic chamber and then sealed with butyl rubber stoppers and punctured with a needle connected to a pressure relief system (VapLock, Cole Parmer) to prevent overpressure. Serum bottles were incubated outside of the anaerobic chamber at 30 or 34°C and 150–180 rpm.

Propanoate production and consumption assays

For propanoate consumption assays, Clostridia strains were cultivated in 2YTm supplemented with 20 g/l glucose for 24 h. From these precultures, 1 ml was used to inoculate 29 ml of 2YTm supplemented with 6.5 g/l sodium propanoate and 60 g/l glucose or glycerol. Serum bottles were next incubated 72 h at 34°C and 150 rpm.

For propanoate production assays, precultures from PAB strains were cultivated 24 h in the anaerobic chamber in 2YTm medium, supplemented with 40 g/l glycerol. A total of 4 ml were used to inoculate 36 ml of the same medium supplemented with 20 g/l CaCO₃. Pressure relief valves were not used in this case, and serum bottles were incubated 96 h at 30°C and 180 rpm.

To test whether propanoate production and consumption could be combined, subsequent cultures were performed by first harvesting the cultures from A. acidipropionici propanoate production assays. Cultures were centrifuged (5000 g, 15 min), and the supernatant pH was equilibrated to pH 7 with 5 M KOH. The medium was again centrifuged (15 000 g, 15 min), and sterile filtered. A total of 11 ml of this medium was mixed with 30 ml 2YTm supplemented with 40 g/l glucose and 20 g/l glycerol. A total of 4 ml of C. beijerinckii DSM 15410 (strain B) preculture grown 24 h in 2YTm (40 g/l glucose and 20 g/l glycerol) was used to inoculate the 41 ml of diluted A. acidipropionici culture, and the serum bottles were incubated 72 h at 34°C and 150 rpm.

Cocultures

For coculture experiments, a cell ratio R was defined as ratio of the initial optical density (OD) at 600 nm of the P. freudenreichii culture over the C. beijerinckii DSM 6423 (strain A) culture. Separate monocultures were prepared the day before the experiment in the anaerobic chamber in 2YTm supplemented with 60 g/l sucrose and 20 g/l glycerol. Precultures were harvested on ice the next day, centrifuged (5000 g, 10 min, 4°C), and resuspended in medium so that an OD of 2 would be attained. OD was remeasured and the Clostridium culture was serially diluted in 2YTm medium. All 20 ml cocultures were started with an equivalent of 0.05 OD value for P. freudenreichii. For Clostridia, serial dilutions were used so that the highest concentration of cells corresponded to an OD value of 0.05, decreasing in 10-time increments. Serum bottles were further incubated 120 h at 30°C, 170 rpm.

Metabolite quantification

Metabolite concentrations were determined by HPLC (Shimadzu), equipped with a refraction index and a PDA detector (RID-10A and SPD-M20A, Shimadzu). An Aminex HPX-87H column (300 × 7.8 mm,  Biorad) was used at a temperature of 60°C and a 0.6 ml/min flowrate. A total of 4 mM H₂SO₄ was used for the mobile phase. Samples were centrifuged 2 min at maximum speed and supernatants were used for quantification. Samples and standards were prepared by mixing with 40 mM H₂SO₄ at a 9:1 ratio and filtering (0.22 µm).

Results and discussion

Metabolic synergy of solventogenic Clostridia and PAB

ABE and IBE Clostridium metabolisms have been well described over the last century (Jones and Woods 1986; Sauer 2016). A total of two metabolic phases, acidogenesis and solventogenesis, alternate over the course of fermentation. Acidogenesis is characterized by the concomitant synthesis of butyric and acetic acid, which results in medium acidification. Overacidification of the medium is prevented by a metabolic switch to solventogenesis. In this metabolic phase, the carbon flux from sugars is diverted from acid synthesis and this process is coupled to acids reconsumption and solvent formation (1-butanol, acetone and, in the case of IBE strains, 2-propanol; Ismaiel et al. 1993).

Importantly, enzymes involved in the acid reconsumption pathway have a promiscuous activity that allows them to convert C2 (two-carbon) and C4 compounds. These enzymes have been purified and characterized in the 80s/90s of the last century (Palosaari and Rogers 1988; Welch et al. 1989; Wiesenborn et al. 1989; Ismaiel et al. 1993) and were also shown to catalyze the conversion of C3 compounds. Hence, we hypothesized that the presence of propionic acid in the culture medium would result in 1-propanol production by solventogenic Clostridia.

Since these bacteria do not naturally produce propionic acid, we sought to find another microorganism that could produce this particular compound. Propionic acid happens to be the main fermentation product of the Wood–Werkman cycle, a metabolic pathway characteristic for PAB, a bacterial group of microorganisms well-known from dairy processes (Hettinga and Reinbold 1972; Bücher et al. 2021). We further hypothesized solventogenic Clostridium and PAB metabolisms could be combined to complement 2-propanol production by an IBE strain with 1-propanol synthesis (Fig. 1A).

Figure 1.

Combining the Wood–Werkman pathway and Clostridial solventogenesis for 1-propanol production. (A) Schematic representation of involved metabolic pathways. (B) 96 h-monocultures of P. freudenreichii and A. acidipropionici in an adapted clostridial-PAB medium (2YT_m) containing glycerol. (C) Propanoate consumption of 72 h-monocultures of C. beijerinckii DSM 6423 (A) and DSM 15410 (B) in 2YT_m medium containing propanoate and glucose or glycerol. (D) Major fermentation products of consecutive monocultures of A. acidipropionici (96 h) and C. beijerinckii strain B (72 h) in 2YT_m medium. First culture was performed with glycerol, then filtered, and diluted in 2YT_m medium containing glucose and glycerol for the second fermentation. All error bars represent the standard error of the mean of three replicates.

Production and consumption of propionic acid

To verify these starting hypotheses, we first designed experiments to show that propionic acid could be produced and consumed in the same culture medium. PAB have been shown to grow well on glycerol, with high yields for propionic acid (Barbirato et al. 1997; Wang and Yang 2013). We, hence selected two well-known strains, A. acidipropionici DSM 4900 and P. freudenreichii DSM 4902, and tested them in modified 2YT (2YT_m) supplemented with 40 g/l glycerol and 20 g/l calcium carbonate for pH buffering (Fig. 1B). Both strains grew well in this medium and converted glycerol into propionic acid with a high yield (≈ 0.8 g/g).

Having established conditions that allow propionic acid production, we next investigated whether propionic acid could be converted to 1-propanol by IBE strains. Clostridium beijerinckii DSM 6423 is the model IBE strain (Máté de Gérando et al. 2018), was shown to grow on glycerol (Forsberg 1987) and was, hence selected as candidate. Clostridium diolis DSM 15410 was recently shown to be an IBE strain and was reclassified as a C. beijerinckii strain (Kobayashi et al. 2020; Sedlar et al. 2021). This strain has mainly been studied for its capacity to ferment glycerol, with which however it does not produce 2-propanol. Because of this recent reclassification, both IBE Clostridia used in this study are from the same species i.e. C. beijerinckii. For clarity the DSM 6423 and DSM 15410 strains are, hereafter, identified as strains A and B, respectively. These two strains were further incubated with 60 g/l glucose or glycerol, and with 5 g/l propanoate (Fig. 1C). In both cases, incubation with glucose and propanoate resulted in the production of 1-propanol, with good yields (≈ 0.7 g/g or 0.85 mol/mol). With glycerol, C. beijerinckii DSM 15410 (B) also produced 1-propanol, though at a much lower yield and concentration (0.27 g/g, 0.1 g/l), while C. beijerinckii DSM 6423 (A) did not grow at all. Without propanoate, the latter grew poorly. This also happened with other media, such as the one reported by Forsberg (1987), though 16S sequencing confirmed the strain received from the DSMZ collection to be a C. beijerinckii strain (data not shown).

Taken together, our results indicate that glycerol is not well-suited as a carbon source for promoting propionic acid conversion into 1-propanol by either C. beijerinckii strains. On the other hand, our data show PAB can produce propionic acid in a Clostridium fermentation medium with glycerol. They also demonstrate that IBE Clostridia can convert propionic acid into 1-propanol in the same medium with glucose as carbon source, which prompted us to proceed to the next step: combining both fermentations.

Successive culturing of A. acidipropionici and C. beijerinckii

Successive culturing of microorganisms can be more easily performed than coculturing, because it does not require precise control over several microorganisms growing concomitantly in a fermentation broth. Such a strategy has proved successful in the past e.g. for 2-butanol production (Russmayer et al. 2019). Therefore, this was our first approach for combined 1- and 2-propanol production. Acidipropionibacterium acidipropionici was chosen as the first strain, as it produced higher amounts of propionic acid from glycerol than P. freudenreichii. This in turn allows a higher dilution of the medium for the second fermentation step, which can be beneficial to remove growth inhibitors accumulated during the first fermentation.

Unlike C. beijerinckii DSM 6423 (A), C. beijerinckii DSM 15410 (B) can grow on glycerol (Fig. 1C) and could hence use any excess glycerol remaining from PAB fermentation. The DSM 15410 strain (B) was therefore chosen for the second cultivation step. Glucose was added in the medium prior to inoculation, since we previously (Fig. 1C) observed glycerol only poorly drove propionic acid conversion into 1-propanol, and since C. beijerinckii DSM 15410 (B) can use simultaneously glycerol and glucose (Xin et al. 2016).

From the initial quantity of propanoate (5 g/l), only a very low quantity was reduced to 1-propanol (0.4 g/l) by C. beijerinckii DSM 15410 (strain B; Fig. 1D). When compared to the control (medium containing glycerol and glucose, not fermented beforehand with PAB), the total amount of solvents produced was decreased heavily, with a 3-fold and a 6-fold decrease of 1-butanol and 2-propanol, respectively. More butyrate was produced instead. These results indicated that the presence of PAB-fermented medium in the broth affected C. beijerinckii DSM 15410 (B) metabolism significantly, stimulating acidogenesis and/or inhibiting solventogenesis. This suggested that the accumulation of one or several compounds produced by A. acidipropionici during the first fermentation were responsible for the observed performance decrease. Hence, a consecutive fermentation approach does not seem appropriate with these strains to produce solvents in general and propanol in particular.

Coculture of P. freudenreichii and C. beijerinckii

We hypothesized that cocultivating PAB and solventogenic Clostridia would be advantageous by allowing simultaneous production and conversion of propionic acid. By growing the microorganisms together and not one after the other, coculturing could also alleviate growth inhibition resulting from the accumulation of substances to toxic levels. Another potential advantage derives from their different sensitivity to oxygen. Solventogenic Clostridia are obligatory anaerobes, while PAB are aerotolerant. Growth of the latter ensures the medium becomes anoxic, rendering it suitable for the growth of the former. Despite the advantages they might confer, such artificial coculture systems, in which member microorganisms are carefully selected to produce a desired compound, can, however, prove difficult to operate, mainly because competitive mechanisms for resources arise, potentially leading to the outgrowth of one member over the others (Zhang and Wang 2016). To minimize this problem, different carbon sources can be chosen so that cocultured microbial strains do not compete for them. A proof-of-concept coculture system featuring orthogonal carbon sources was hence designed to facilitate its control and monitoring. Propionibacterium freudenreichii DSM 4902 grows on glycerol, but not on sucrose (Goodfellow et al. 2012), whereas C. beijerinckii DSM 6423 (strain A, hereafter referred to simply as ‘C. beijerinckii’) does not use glycerol when sucrose is present in the medium. Using 2YTm as fermentation medium, sucrose and glycerol can be supplemented to desired quantities and in principle be used separately by both microorganisms to drive 1-and 2-propanol biosynthesis.

Propionibacterium freudenreichii and C. beijerinckii have significantly different growth rates (T_d ≈ 4 h and 2 h, respectively), which needs to be considered for coculturing both microorganisms, as at similar initial cell densities C. beijerinckii is likely to severely outgrow P. freudenreichii. We, thus defined an initial cell ratio R as

where OD_i corresponds to the initial cell biomass (expressed in OD₆₀₀ units) of each cocultured microbial strain. As C. beijerinckii grows much faster than P. freudenreichii, only R values above 1 were investigated, similarly to what was previously done for C. acetobutylicum and C. ljungdahlii (Charubin and Papoutsakis 2019).

Cocultures were inoculated with an initial OD₆₀₀ of 0.05 for P. freudenreichii, with R values between 1 and 1000 000 determining the initial biomass concentration of C. beijerinckii cells (Figs 2 and 3). Biomass was consistently higher in the cocultures than in the monocultures (Fig. 2A). This was expected, as the OD₆₀₀ value results from the growth of two different microorganisms. However, in all cocultures the maximal cell densities measured were unexpectedly (but consistently) higher than the addition of the maximum OD₆₀₀ values of both monocultures. This constitutes a first hint to a positive growth impact each strain could have on the other.

Figure 2.

Fermentation kinetics of C. beijerinckii and P. freudenreichii cocultivation in 2YTm with sucrose and glycerol. R represents the initial OD₆₀₀ ratio. (A) Biomass. (B) pH. (C) Glycerol consumed. (D) Sucrose consumed. (E) 1-Butanol. (F) 2-Propanol. (G) 1-Propanol. (H) Propanoate. Experiments were performed in duplicates. Error bars show the upper and lower values used to calculate the mean.

Figure 3.

Clostridium beijerinckii and P. freudenreichii cocultivation in 2YTm with sucrose and glycerol, final time point. R represents the initial OD₆₀₀ ratio. (A) Glycerol and sucrose consumed. (B) Acids produced. (C) Solvents produced. (D) 1-Propanol produced. Experiments were performed in duplicates. Error bars show the upper and lower values used to calculate the mean.

This synergy could theoretically be explained by the effect solventogenic Clostridia have on the pH (Fig. 2B). Indeed, by consuming acids they can maintain a pH of around 5–7, which in turn may facilitate the growth of P. freudenreichii (inhibitory pH measured in the literature: 4.5–4.9; Goodfellow et al. 2012). Addition of even the lowest amounts of Clostridial cells (R = 1 000 000) reversed the pH curve after 2 days, allowing it to remain above 5 for the rest of the fermentation. However, glycerol (used by P. freudenreichii) and sucrose (used by C. beijerinckii) consumption kinetics (Fig. 2C and D), indicate pH control is not the only reason why cocultures grew to much higher cell densities than monocultures. Keeping the pH above 5 should improve the growth of P. freudenreichii, and hence increase glycerol consumption, which indeed happens after 3 days forR values above 1000. However, taking R = 1 as an example, early growth (24 h time point) shows relatively similar glycerol consumption compared to the P. freudenreichii monoculture, but a clear increase in sucrose consumption compared to the C. beijerinckii monoculture. This suggests that the increased biomass comes from a higher amount or an increased metabolic activity of C. beijerinckii cells. Consequently, higher amounts of native C. beijerinckii metabolites (1-butanol and 2-propanol) are produced in all cocultures, compared to C. beijerinckii monoculture (Fig. 1E and F). Because solvent tolerance has been assumed to be the bottleneck for production of these metabolites, this result was quite unexpected.

For the main end metabolites (1-butanol, 2-propanol, 1-propanol, and propanoate), varying initial cell ratios has different effects. Indeed, R mostly influences the rate at which 1-butanol and 2-propanol are produced, with higher rates when the initial proportion of C. beijerinckii is higher (Fig. 1E and F). For 1-propanol and propanoate on the other hand (Fig. 1G and H), R directly influences both the production rate and the maximum titers, with higher rates and titers when the initial proportion of P. freudenreichii is higher.

Focusing on the final time point (120 h) can help obtaining a clearer picture of differences among cultures (Fig. 3). Compared to monocultures, cocultures resulted in higher total carbon consumption (Fig. 3A). Glycerol and sucrose consumptions expectedly followed inverted trends with increasing R values. On the other hand, acid production peaked at R = 10 000, and then decreased. Given that glycerol consumption was lower than for higher R values, this must be linked to a higher acid uptake by Clostridial cells. In fact, sucrose uptake kinetics (Fig. 2D) show that for R > 10 000,  C. beijerinckii cells are longer active (i.e. at least for 1 day), which should translate into a longer acid conversion activity. This hypothesis is supported by the increasing proportion of 1-propanol in the solvent mixture with increasing R values (Fig. 3C and D), which can only be the result of acid uptake and reduction by C. beijerinckii cells. Interestingly, increase in 1-propanol concentration happens at the expense of 2-propanol production at high R values, which suggests propanoate conversion into 1-propanol could be partly decoupled to 2-propanol synthesis. This could for instance happen if the Ptb—Buk pathway catalyzed acyl-CoA formation instead of acid synthesis (i.e. in reverse direction), as already shown for butyrate uptake in C. acetobutylicum (Lehmann et al. 2012). Importantly, C. acetobutylicum Ptb and Buk enzymes were shown to have catalytic activity on propanoate (Hartmanis 1987) and propionyl-CoA (Wiesenborn et al. 1989), supporting the possibility that this pathway might be involved in propanoate conversion to 1-propanol by C. beijerinckii.

The cocultures generally result in higher solvent production, compared to the C. beijerinckii monoculture. Indeed, the increasing quantity of 1-propanol in the fermentation broth yields a higher combined proportion of both propanol isomers, with a maximum of 12.0 g/l and 61.9% of solvent mass (R = 1 000 000). Compared to the C. beijerinckii monoculture, this represents a 98% titer increase, as well as a 37% increase of propanol proportion in the solvent mixture. In batch fermentation, such a concentration ranks among the best obtained for microbial propanol production (Walther and François 2016).

Increasing 1-propanol quantities do interestingly not result in higher overall solvent titers in the cocultures. Indeed, 1-butanol and, to a lower extent, 2-propanol titers concomitantly decrease, maintaining solvent titers to around 20 g/l (with a maximum value of 21.0 g/l at R = 100 000). This consistent concentration of solvents suggests this is the physiological limit that can be attained in these conditions. However, such a high concentration could not be reached with C. beijerinckii DSM 6423 alone, in this work (13.3 g/l) or in the literature (Collas et al. 2012; 13.2 g/l, similarly in batch condition), indicating that solvent tolerance is not the main bottleneck for solvent production by this strain (as previously suggested; Gérando et al. 2016). The presence of P. freudenreichii therefore not only results in the synthesis of a novel metabolite, but also in significantly increased fermentative performances for all tested R values, to levels comparable to the best wild-type Clostridial strains. Besides, the Wood–Werkman pathway produces no carbon dioxide, and increasing amounts of PAB, therefore, result in higher solvent and propanol yields (based on the carbon sources glycerol and sucrose combined; up to 0.39 g/g and 0.24 g/g for R = 1000 000, respectively, compared to 0.34 and 0.15 g/g for the C. beijerinckii monoculture). The C. beijerinckii/P. freudenreichii synergy could possibly be due to different factors (such as pH maintenance, secretion of growth factors, consumption of growth inhibitors, or even quorum-sensing-based modulation of gene expression), which will require further investigating.

In addition to gaining deeper insights into the Clostridium/PAB interaction, further work will have to address different challenges before envisioning industrializing a process. Coculture processes with members that have such different growth rates will likely be constrained to batch/fed-batch processes as, without a proper mean of controlling their population, Clostridia are likely to take over rapidly in continuous modes. Even for batch culturing, precisely controlling R might prove difficult in industrial-scale bioreactors. Developing ways to control the proportion of both microorganisms would hence be desirable, by for instance designing and engineering codependencies. Developing a way to rapidly quantify the different populations is another aspect that needs to be addressed, in order to be able to properly monitor the fermentation process. In industrial setups, working with orthogonal carbon sources (and, thus following growth by measuring depletion of carbon sources) will likely not be possible. Flow cytometry is a fast alternative that could be used in that regard, provided Clostridia and PAB can properly be differentiated.

In summary, our work exemplifies in a straightforward fashion how artificial cocultures can be used to bypass monoculture limitations, and to combine metabolisms to produce novel metabolites with high efficiencies. Indeed, we produced as much as 21 g/l of solvents, of which 11.5 g/l was 1- or 2-propanol, the former being a metabolite not produced by either monoculture. This is much higher than what can be produced in similar conditions by an IBE strain alone (Collas et al. 2012; 13.2 g/l solvents, of which 4.5 g/l is 2-propanol). An engineered IBE strain, such as C. acetobutylicum overexpressing the isopropanol pathway (Collas et al. 2012), fares better in terms of total solvent titers (24.4 g/l), but not as well in terms of propanol titers (8.8 g/l). Such an engineered strain would be an ideal candidate to be used in combination with PAB to produce more propanol. Comparatively to other propanol-producing coculture systems, the same observations apply. On the one hand, Charubin and Papoutsakis (2019) produced 21.2 g/l of solvents, of which 7.0 g/l was 2-propanol, with their C. acetobutylicum/C. ljungdahlii coculture. PAB could probably similarly be added to their setup, as C. ljungdahlii mainly feeds off H₂ and CO₂, and thus would neither be in competition with C. acetobutylicum nor the added PAB. Besides, C. ljungdahlii was shown to efficiently reduce propionic acid to 1-propanol (Perez et al. 2013), which would also be advantageous in a coculture with PAB. On the other hand, Liu et al. (2014b) produced 15 g/l (of which 6 g/l was 1-propanol) with A. bacchi and C. propionicum (Liu et al. 2014a), but were not able to reproduce these results in their follow-up work (around 1 g/l 1-propanol for 2.2 g/l total solvents). Regardless, C. propionicum could also be a good candidate for complementing our artificial coculture system, as it feeds off ethanol to produce additional propanol. These examples highlight a few synergies increasing or enabling propanol synthesis, and the microbes involved could be considered as the first building blocks of artificial coculture systems dedicated to propanol fermentation. In this context, our work shows PAB as promising candidates complementing this artificial coculture toolbox and paves the way for further, more complex, microorganism associations.

Conflicts of interest statement

None declared.