Introduction

Biological production of biofuel is a promising future technology for sustainable energy conservation in a post-petroleum era. Biobutanol is one of the potential chemicals that can be produced through a microbiological pathway, acetone-butanol-ethanol (ABE) fermentation, by bacterial strains of solventogenic clostridia. Clostridia have a great potential in the field mainly due to their ability to utilize an exceptionally wide range of substrates (Tracy et al. 2012), which predisposes them to be potential producers of bioenergy from various renewable and waste streams (Zhao et al. 2019; Branska et al. 2020). The biggest bottleneck that hinders economically viable production of biobutanol is its low final titre, which is dictated by solvent tolerance (Patakova et al. 2018). Clostridia possess a wide range of mechanisms allowing them to survive in increased concentrations of solvents (Patakova et al. 2018; Cheng et al. 2019) and one of such defence mechanism is enhanced expression of efflux pumps that can pump toxic compounds out of the cell into the extracellular environment so that it improves biotechnological process performance through strengthening the flux of product out of the cell. Although transport of most uncharged low-molecular-weight substances was previously assumed only to be mediated by passive diffusion, there is evidence now that this may not be the case and many substances are transported by more or less promiscuous transmembrane transporters (Fernandes et al. 2003; Kell et al. 2015; Boyarskiy and Tullman-Ercek 2015). Up to now, seven families of prokaryotic efflux pumps have been described, including five well-known groups: small multidrug resistance (SMR), major facilitator superfamily (MFS), multidrug and toxic compound extrusion (MATE), resistance nodulation division (RND) and ATP binding cassette (ABC) transporters, and two newly described groups, proteobacterial antimicrobial compound efflux (PACE) (Hassan et al. 2015; Spengler et al. 2017) and p-amino benzoyl-glutamate transporter (AbtG) (Delmar and Yu 2015). These transporters have been thoroughly studied, in particular in connection with multidrug resistance (MDR) of pathogenic microorganisms; however, similar mechanisms are probably natural to most bacteria for their defence against a wide range of antimicrobial agents including solvents. Indeed, some studies point to the possibility of transport of various drugs across the phospholipid moiety being minor (Kell et al. 2014). The role of efflux pumps in the transport of short-chain alcohols like n-butanol, isobutanol, ethanol, etc. has been intensively researched, but so far with rather contradictory conclusions. Transcriptomic data revealed increased expression of particular efflux pumps in strains that are more tolerant to butanol (Reyes et al. 2011, 2013) and it has been experimentally confirmed that overexpression of selected pumps (Basler et al. 2018) or modification of existing ones leads to increased tolerance to butanol (Bui et al. 2015; Zhang et al. 2018; Basler et al. 2018; He et al. 2019). However, it is likely that not only one particular transporter is involved in the export of butanol from the cell, but multiple different pumps could operate in synergistic mode (Turner and Dunlop 2015). The space within a cell membrane is rather limited when compared to intracellular space, and increased expression of transmembrane pumps can lead to changes in natural membrane proteome (Wagner et al. 2007) and thus affect cell fitness or viability. Some balance should exist between the negative effect of overexpression of transporters and the actual concentration and toxicity of antimicrobial substances (Boyarskiy et al. 2016). Hence, the expression of efflux pumps is tightly regulated and changes throughout the life cycle of bacteria in response to actual needs. Mostly a basal level of expression can be found under standard conditions while overexpression is induced by particular stimuli (Sun et al. 2014).

While the sporulation-inclusive life cycle and biphasic production schedule of solventogenic clostridia barely allow cells to remain in the same physiological state (Branska et al. 2018) for even a short period of time, dynamic changes in efflux activity throughout ABE fermentation are expected. Considering that efflux activity may play a fundamental role in defence against toxic substances, we wanted to know how efflux activity develops during the production of intrinsic stressors such as organic acids and solvents during ABE fermentation. Nowadays, efflux is analysed mostly in clinically important isolates of pathogenic bacteria with enhanced efflux using an indirect minimal inhibition concentration (MIC) assay, but effective protocols for direct monitoring of efflux that depends on changing physiology are scarce. Therefore, we focused on the development of a reliable methodology for the quantification of efflux in the culture of solventogenic clostridia. An optimized methodology based on ethidium bromide (EB) efflux was then applied to monitor changes of efflux during standard ABE fermentation and ABE fermentation with impaired solventogenesis (the so-called “acid crash” phenomenon) for Clostridium beijerinckii strain NRRL B-598. The data obtained were compared with the expression profile of genes annotated as transporters, which were acquired in an independent but similar experiment (Sedlar et al. 2018; Patakova et al. 2019).

Materials and methods

Cultivation

Bacterial strain C. beijerinckii NRRL B-598 from the ARS/NRRL culture collection was cultivated either in an anaerobic chamber Concept 400 (Ruskin) or a bioreactor Multiforce (Infors HT) according to protocols described previously (Kolek et al. 2016a; Patakova et al. 2019). For assay optimization experiments, cultures were inoculated directly from spores; 0.5 ml of a spore suspension was heat-treated (80 °C, 2 min) and transferred into an Erlenmeyer flask containing 50 ml of TYA medium (50 g/l glucose, 6 g/l tryptone (Sigma-Aldrich), 2 g/l yeast extract (Merck), 3 g/l ammonium acetate, 0.5 g/l KH2PO4, 0.3 MgSO4·7H2O and 0.01 FeSO4·7H2O). An exponentially growing overnight culture was then used for tuning the efflux assay. Experiments in bioreactors were conducted under similar conditions for both phenotypes, i.e. standard ABE and impaired solventogenesis, the only differences being in the initial glucose concentration (40 and 50 g/l, respectively) and in an establishment of an anaerobic atmosphere in the bioreactor. For standard ABE fermentation, reactors were bubbled with CO2 for 15 min, which led to a decrease in pH that was adjusted afterwards with 1 M NaOH to a pH value of 6.3 prior to inoculation. For induction of the acid crash-like phenotype, reactors were bubbled with N2 and therefore no pH adjustment was necessary.

Efflux assay

For accumulation experiments, 1 ml of the sample was transferred from an Erlenmeyer flask into an Eppendorf tube and centrifuged for 1 min at 8000 g. After centrifugation, the supernatant was discarded and the pellet was resuspended in either a saline solution or TYA medium containing EB at a concentration of 2 μg/ml. Fluorescence was measured immediately and the mean fluorescence in the FL3 channel (˃ 670 nm) was recorded together with the precise time interval after EB addition. For each point, 50 μl of cell suspension were transferred into 1 ml of deionized water and placed onto the SIP of a BD flow cytometer (Accuri C6). At least 5000 particles were analysed per sample. An accumulation assay with increasing concentrations of EB was carried out similarly, the only change being an increasing concentration of EB in the solutions added and incubation time of 5 min. CCCP (carbonyl cyanide m-chlorophenyl hydrazone) was diluted in DMSO (dimethyl sulfoxide) and added into TYA and saline solution with EB to reach a final concentration of 5 μg/ml. Beside this experiment, a similar concentration of CCCP was added to the growing culture; however, it resulted in sharp declination in culture viability and thus no more data were generated by this design. Control experiments were carried out to show that DMSO had no impact on final fluorescence (data not shown). For the direct efflux assay, the calculated amount of cell suspension corresponding to 1 ml of culture with OD 0.5 ± 0.1 was transferred into an Eppendorf tube and centrifuged and placed into the anaerobic chamber where the supernatant was discarded and the pellet was resuspended in 1 ml of saline solution containing 2 μg/ml EB and incubated for 5 min. After EB loading into cells, the suspension was centrifuged (8000 g, 1 min) and the supernatant was discarded. Then, the pellet was resuspended in 1 ml of saline solution or TYA medium with 20 g/l of glucose, and fluorescence was measured immediately by flow cytometry as described above. From the time course of fluorescence, the efflux activity was quantified as an absolute value of slope calculated using MS Excel function “slope”. The slope was calculated from at least three values of mean fluorescence (arbitrary units) obtained after EB removal at regular intervals within the first 15 min. All mean fluorescence values were calculated automatically using FC software CFlowPlus. The methodology did not allow us to measure the samples in technical replicates; therefore error bars within charts with cytometric data indicate variance between biological replicates and not analytical variations.

Analytics

Glucose and metabolite concentrations were analysed by HPLC with ion exchange-based separation using as a stationary phase IEX H form, 250 × 8 mm (Watrex). The mobile phase was 0.5 mM H2SO4 in demineralized water at a flow rate of 1.0 ml/min. Detection was refractometric for all analytes and quantification was based on calibration curves of pure substances. Cell concentration was measured as absorbance at 600 nm expressed as optical density (OD) using a Varian Cary 50 UV-Vis spectrophotometer. Viability was measured as the number of solely propidium iodide (PI)–positive cells after staining with a combination of PI and carboxyfluorescein diacetate as described elsewhere (Branska et al. 2018) and analysed employing a BD Accuri C6 flow cytometer.

Transcriptomic analysis was performed as a part of another study (Sedlar et al. 2018). Putative transporters were selected based on the publication of Jureckova et al. (2018) and the annotation available in the GenBank database under accession number CP011966.3. Count tables generated during previous studies (Sedlar et al. 2018; Patakova et al. 2019) were normalized using the R/Bioconductor DESeq2 package (Love et al. 2014) using a built-in DESeq2 function. A heatmap representing transcription of selected genes using Z-score was generated with the R package ggplot while similarities in expression profiles were analysed using Ward’s hierarchical agglomerative clustering considering the Manhattan distance (Murtagh and Legendre 2014).

Results

Ethidium bromide accumulation assay

All experiments were performed using ethidium bromide (EB), a fluorescent probe that is one of the most commonly used (Viveiros et al. 2008; Paixão et al. 2009; Patel et al. 2010; Rodrigues et al. 2011; Pal et al. 2020) in efflux assays as it serves as a universal substrate for a wide range of efflux pumps. Actually, two different experimental designs can be used in an accumulation assay. The first is a direct measurement of the increase in fluorescence of EB after its addition to a cell suspension; for this design, lower efflux activity leads to a more rapid increase in fluorescence. The second method is the addition of an increasing EB concentration to cells, in which case cells are able to actively efflux EB up to some particular concentration and EB exceeding this concentration can penetrate the cells, leading to an increase in fluorescence. Both methods were applied to an exponentially growing (motile, rod, non-sporulating) culture of C. beijerinckii NRRL-B598. Fluorescence was measured at the individual cell level using flow cytometry in two environments, a saline solution to mimic an environment without a source of energy and TYA medium, containing both a source of energy and all required nutrients. Results are shown in Fig. 1a from which it is obvious that without an energy source, accumulation takes place much faster than with an energy source, confirming that EB can be used to measure energy-dependent active efflux from C. beijerinckii. The accumulation assay in saline solution also identified two distinct populations (Fig. 1b and c), one of apparently higher fluorescence (probably less active or injured cells) and a second with low fluorescence (viable cells). Focusing on these two subpopulations, we could see a different dynamics of EB uptake. Fluorescence of intact cells with lower fluorescence increased steadily for approximately 20 min and then the rate of increase declined, although the cells continued to take up the EB probe. Less active, injured or dead cells with higher fluorescence took up EB very rapidly, reaching nearly steady fluorescence in approximately 15 min. When the accumulation assay was performed in TYA medium, only one population of lower fluorescence intensity was observed (Fig. 1d).

Fig. 1
figure 1

Increase in mean fluorescence of cells as a function of incubation time with ethidium bromide (EB; 2 μg/ml). (a) Accumulation of EB in time in two different solutions, saline solution (circles) and TYA medium (squares). (b) Dynamic fluorescence accumulation within damaged (triangles) cells and vital cells (diamonds) in saline solution, and FC histograms of cells stained in (c) saline solution showing the division of cells into two subpopulations (left, active vital cells; right, cells with impaired ability to pump out EB) and (d) TYA medium where no apparent division was observed

Full size image

The second method, i.e. cell exposure to increasing EB concentration, was also performed in both solutions with an energy source (TYA medium) and without (saline solution); see Fig. 2. We expected that fluorescence of cells challenged with a low concentration of dye would be comparable to the level of autofluorescence and that by exceeding a certain EB concentration accumulation would occur, resulting in a stepwise increase in fluorescence. However, no such course was observed; instead, fluorescence increased linearly in the range from 0.25 up to 2 μg/mL EB, and at 5 μg/mL, fluorescence was probably influenced by saturation of nucleic acid binding sites with EB. Even though the fluorescence of cells loaded with EB in TYA medium was lower, the course of fluorescence change with increasing EB concentration was similar. To understand this phenomenon better, we added carbonyl cyanide m-chlorophenyl hydrazone (CCCP), a substance that is used in efflux assays for its ability to uncouple a proton gradient across the membrane. CCCP had a negligible effect on the time course of mean fluorescence with increasing EB concentration (Fig. 2). It should be noted that data are presented only for the experiment where CCCP was added to accumulation reaction only; protocols with the addition of CCCP to the growing culture led to a significant decline in viability with almost 100% of inactivated cells after 2 h.

Fig. 2
figure 2

Mean fluorescence of cell culture loaded for 5 min with increasing concentration of EB in the presence of saline solution, TYA medium and with and without addition of carbonyl cyanide m-chlorophenyl hydrazone (CCCP; 5 μg/ml)

Full size image

Direct efflux measurement

Direct measurement of efflux was performed as follows: cells were first pelleted by centrifugation to remove all possible energy sources in the cultivation medium and then loaded with EB in saline solution to enable EB to enter the cells under conditions minimizing the effect of active efflux during dye loading. After loading of the dye, the cells were centrifuged and saline solution with EB was replaced by TYA medium or saline solution without EB. Fluorescence was measured immediately for 60 min. The resulting curves (Fig. 3) showed a clear difference between changes in fluorescence over time in the culture without an energy source (saline solution) versus the culture supplemented with an energy source (TYA medium). In saline solution, fluorescence remained constant or slightly increased, which was probably the result of a slightly increased amount of EB intercalated into nucleic acids (EB increases its fluorescence upon intercalation). In contrast, fluorescence of cells in TYA medium declined quickly due to active efflux of the EB. Thus, the direct efflux assay clearly distinguished between energy-dependent efflux and passive diffusion. Direct efflux measurement in an energy-supplemented environment was therefore chosen for direct monitoring of efflux. Data shown in Fig. 3 indicate that fluorescence declined quickly after EB depletion, and at later stages, the decline slowed down. Therefore, the interval of the first 15 min after EB depletion was chosen for the evaluation of efflux rate and linear regression was fitted to the experimentally obtained time course of mean fluorescence of cells. The absolute value of the slope for at least three points within the interval of 0–15 min after EB removal was used for quantification of efflux in clostridia subjected to two types of cultivation experiment (Figs. 4 and 5).

Fig. 3
figure 3

Fluorescence of cell culture after removal of ethidium bromide (EB) and suspension of cells in TYA medium and saline solution. The linear slope depicted in the chart shows a linear regression of data obtained within the selected time interval for evaluation of efflux

Full size image
Fig. 4
figure 4

Parameters of standard ABE fermentation showing (a) efflux activity and pH, (b) growth and proportion of inactive cells in culture, (c) concentration of solvents, (d) acids and (e) glucose. The vertical line represents the transition between the acidogenic phase (left) and the solventogenic (right) determined by the change in pH

Full size image
Fig. 5
figure 5

Transcriptional profiles of genes coding proteins putatively involved in efflux activity of C. beijerinckii NRRL-B598 visualized on the heatmap using a Z-score related to an average expression of each gene. Genes are clustered using Ward’s hierarchical clustering considering Manhattan distance among transcriptional profiles. T1–T6 are six sampling time points (for more information see Sedlar et al. (2018)), namely, T1 represents an exponentially growing, acid-producing culture, 3.5 h after initiation of ABE fermentation; T2, the switch from acidogenesis to solventogenesis indicated by a pH turning point at 6 h; T3–T6 early to late solventogenesis and sporulation initiation, sampling times 8.5, 13, 18 and 23 h, respectively

Full size image

Efflux changes during standard ABE fermentation

Changes in efflux activity throughout the time course of standard ABE fermentation are shown in Fig. 4, together with the main growth and metabolic parameters and basic analysis of culture fitness using cytometric detection of physiological changes after staining with carboxyfluorescein diacetate and propidium iodide (PI). The efflux rate during a typical ABE process was high throughout the majority of the fermentation. The highest value was reached in late acidogenesis, an apparent drop was recorded in the transient switch stage between acidogenesis and solventogenesis, but it was soon restored and remained high up to the 24th hour when efflux declined strongly to low values, where it remained until the end of the cultivation. We have considered the possible effect of inactive cells in the population; therefore, a viability assay was performed on all samples. While the viability analysis revealed a significant increase in % of PI-positive cells after the shift to solventogenesis, there was no other apparent change up to the 24th hour and metabolic profiles suggested continuing fermentation. The authors would like to note here that, under some conditions, PI positivity does not necessarily mean cell death (Shi et al. 2007) and might reflect some physiological changes including, among others, initiation of sporulation ((Jones et al. 2008; Tracy et al. 2008; González-Peñas et al. 2015; Branska et al. 2018). The first mature spores were observed in the culture (0.8% ± 0.2) shortly after the 24th hour, indicating successful accomplishment of the sporulation cycle.

As there are published data of transcriptomic analysis of the strain tested under similar conditions in the NCBI Sequence Read Archive (SRA) under accession number SRS2757230, we analysed the expression of putative transporters (Jureckova et al. 2018) and grouped them according to their expression profiles (Fig. 5). Ward’s hierarchical clustering considering Manhattan distance among particular genes according to their expression profiles revealed two major groups, each of them further dividing into two well-distinguished sub-groups reflecting different physiological patterns. This gene clustering into two main groups was not limited only to transport-related genes but was probably a common phenomenon observed in independent transcriptomic studies of C. beijerinckii (Wang et al. 2012; Sedlar et al. 2018). The main groups represented a distinction between genes active predominantly in acidogenic plus early solventogenic stages and genes actively transcribed in later stages or during the whole period analysed. The subdivision of the first group reflects differences between acidogenesis (Fig. 5, red depicted cluster) and the switch into the solventogenic stage (Fig. 5, yellow depicted cluster) while the second group is formed either by genes that are more involved in metabolism during the transition into a stationary or sporulating culture or by genes that cannot be unambiguously assigned to any of the above groups.

Seemingly there are various genes switched on and off during particular stages of ABE fermentation, which may explain the phase-dependent fluctuation in efflux activity. Those genes active predominantly in the acidogenic phase have apparently higher RPKM (Reads Per Kilobase Million) values when compared to the group of genes more transcribed in later stages (see Supplemental_material_1). Much deeper and more thorough studies should be performed to reveal interconnection between nonspecific efflux activity assayed with EB and specific gene expression of particular pumps. Nevertheless, some general phenomena can be observed. Within the transmembrane proteins and their subunits depicted in Fig. 5, a clear abundance of ABC transporters can be found within the first main group with a higher prevalence in the acidogenic phase (Fig. 5, red cluster). On the other hand, transmembrane subunits of RND family transporters are more frequent in mid to late solventogenic stage (the second major group), together with MATE and MFS families.

Efflux changes during ABE fermentation with impaired solventogenesis

Since the highest efflux activity was observed for the acidogenic culture in the standard ABE fermentation, we wanted to explore whether this applies generally for acid-producing cultures. The sole acid production is typical for so-called “acid crash”, a phenomenon that sometimes can be observed in a non-pH–controlled cultivation of solventogenic clostridia (Maddox et al. 2000; Schuster et al. 2001; Seo et al. 2017). It is characterized by an accumulation of organic acids while a switch to the solventogenic stage is omitted and cells continue to grow at a decreased rate with acid production leading finally to detrimental acidification of the extracellular environment and cell death. C. beijerinckii NRRL B-598 tends to display a phenotype similar to “acid crash” accidentally but can be induced by particular changes in medium preparation as well, although factual triggers for its induction are poorly understood.

We measured efflux activity during cultivation when increased acid production was reached by slightly changed protocol for preparation of the culture medium (see Materials and methods section). Even though the pH curve (Fig. 6) implied “acid crash” behaviour, subsequent HPLC analysis proved suppressed solventogenesis, and after all, some production of solvents apparently started, even though with significantly lower titres. It seemed more like disrupted acid re-utilization than total inhibition of solvent production. Nevertheless, increased toxicity of the acidic environment (decreased pH) together with increasing acid concentration was reached as originally intended. Efflux rate, surprisingly, decreased strongly at the stage when, in standard ABE fermentation, the switch to solventogenesis would occur (which is indicated by cessation of the sharp pH fall). Very low values of efflux rate were observed up to the 24th hour of cultivation when the culture viability dropped significantly and cells started to die from the accumulation of negative factors in the surrounding environment. A decrease in efflux activity was not the only apparently changed physiological feature observed at the switch. The growth rate was also significantly slowed, possibly reflecting attenuated cell activity related to decreased ability of cells to actively pump EB out of the cells. Moreover, no sporulation events were observed throughout the whole fermentation.

Fig. 6
figure 6

Parameters of cultivation with impaired solventogenesis: (a) efflux activity and pH, (b) growth and proportion of inactive cells in culture, (c) concentration of solvents, (d) acids and (e) glucose. The vertical line represents the transition between the acidogenic phase (left) and expected solventogenic phase (right) determined by the change in pH

Full size image

Discussion

Even though efflux pumps are highly discussed especially in connection with increased antibiotic resistance in bacteria (Sun et al. 2014; Spengler et al. 2017; Du et al. 2018), their field of action can be much more extensive; e.g. in the case of solventogenic clostridia, active efflux might play a role in cell defence mechanisms against toxicity of solvents produced (Patakova et al. 2018; Kusumawardhani et al. 2018). Nowadays, an effort is being made towards the development of high-producing solvent-tolerant strains, e.g. with targeted overexpression of genes encoding heat shock proteins (HSPs) (Tomas et al. 2003; Liao et al. 2017). Increasing the efflux rate of a toxic product might represent another approach for developing hyper-producing mutants. However, knowledge on this issue is still very limited, mainly due to limited tools for studying this phenomenon.

In this work, it was shown that flow cytometry–based methodology could serve for at-line monitoring of active efflux during various cultivation experiments, reflecting the simultaneous activity of a whole range of various transporters. Our original intention was to determine whether the production of solvents accompanied by increasing adverse environmental stress may lead to an increased efflux rate. The results indicated the highest efflux in the acidogenic stage of ABE fermentation, independent of the current butanol concentration. Actually, high efflux pump activity during acidogenesis is not surprising. Increasing acid concentration puts increasing demands on transporting acids and H+ ions (Bowles and Ellefson 1985; Guan and Liu 2020) from the cells in order to maintain a stable intracellular environment. Efflux of H+ ions occurs at the expense of ATP hydrolysis (Senger and Papoutsakis 2008); because the production of acids at this stage of fermentation is rising, an intensive efflux leads to depletion of cellular energy. In accordance with the expected high activity of ABC transporters during acidogenesis, transcription profile of genes related to ABC pumps clustered within the group of genes exhibiting higher expression during acidogenesis and early solventogenesis (Fig. 5). The switch to solventogenesis can be considered a rescue mechanism from the negative impact of decreasing pH caused by the accumulation of butyric and acetic acids. Results similar to ours were obtained for C. acetobutylicum ATCC 824, where a high rate of H+ ion efflux was reported at the beginning of the acidogenic phase, with a decrease towards the end, coinciding with the maximum production of butyric and acetic acids (Senger and Papoutsakis 2008).

Re-utilization of acids was observed after the 7th hour of cultivation (Fig. 4), which could lead to a lower rate of efflux. Simultaneously, low concentrations of solvents were already present in the medium; therefore, activation of other pumps capable of solvent excretion could be possible. Increased expression of genes encoding efflux pumps such as RND transporter subunits belonging to the RND family (the most probable family involved in solvent efflux (Kusumawardhani et al. 2018)) was observed in all four major clusters but with higher prevalence in solventogenic phase, supporting the putative function of this type of pumps in the transportation of solvent across cytoplasmic membrane.

The group of genes encoding putative efflux pumps activated in mid to late solventogenesis (Fig. 5) included various transporters of RND, MATE and MFS efflux pump families and decreased incidence of ABC pumps was observed. At this stage, increased expression of the MerR family of transcriptional regulators was also observed. Regulators belonging to this family are characterized by their response to environmental stimuli, including oxidative stress or antibiotics (Brown et al. 2003), and some members of the family are reported to act as activators of efflux transporters in Bacillus (Ahmed et al. 1994).

Contrary to our expectations, efflux did not increase with rising solvent concentration, but rather the opposite. During late solventogenesis, cells ceased active pumping of EB. At this stage, the first mature spores were released from mother cells and a decrease in viability suggested cells were approaching the end of ABE fermentation. This type of predictive repression of physiological functions and loss of viability prior to attaining an inhibitory concentration of solvent has been reported previously in a culture of C. beijerinckii NRRL-B598 (Branska et al. 2018).

An unambiguous difference between efflux rates during standard ABE and acid crash-like fermentation indicates the possible involvement of efflux pumps in solvent transport. The role of particular efflux pumps in short-chain organic solvent transport has been studied through overexpression of genes for various types of efflux pumps that might be involved. Outcomes of this research mostly admit the active efflux of solvents as one of the possible transport mechanisms; however, the issue is not fully understood yet. Particularly for butanol, the extended screening of non-native RND efflux pumps in E. coli (Dunlop et al. 2011) has disproved the active efflux of butanol. Other reports describe a positive impact of particular pumps expression on the survival of cells in the presence of butanol, or on production, particularly TtgABC (Basler et al. 2018) and SrpABC (Kieboom et al. 1998; Bui et al. 2015), both from Pseudomonas putida and MdtJ (Zhang et al. 2018) and AcrB (Reyes et al. 2011; Fisher et al. 2013; He et al. 2019) from E. coli. Most studies consider the RND-type pumps as the best candidates for solvents extrusion, but another pump families might also be involved. For example, upregulation of ABC-type pumps was observed upon butanol challenge of acidogenic C. acetobutylicum chemostat cells (Schwarz et al. 2012) and improvement of isopentenol production was recorded after expression of certain genes encoding ABC transporter in E. coli (Foo et al. 2014). Many published studies concerning solvent efflux used non-native pumps from aerobic production systems and were tested in Gram-negative bacteria with mainly externally added solvents; therefore, conclusions from them should be taken cautiously if compared to the results obtained for solventogenic clostridia. Unfortunately, there is a limited number of studies available directly for solventogenic clostridia or their native pumps, probably because the genetic transformation of clostridia is an experimental challenge in itself (Kolek et al. 2016b). However, some reports are beginning to emerge, e.g. the very recent study successfully utilized RND-type pump from Pseudomonas putida for the enhancement of tolerance of Clostridium saccharoperbutylacetonicum to lignocellulose-derived inhibitors and butanol (Jiménez-Bonilla et al. 2020). Another example is the overexpression of a putative ABC-type transporter and related regulator in C. acetobutylicum (Yang et al. 2020) which led to faster growth and solvent formation. In general, the role of efflux pumps in solvent tolerance seems to have a solid base with an increasing number of studies supporting this presumption (see recent review (Vasylkivska and Patakova 2020)). Another major issue is the harmful effect of an abundant presence of efflux pumps in the cell membrane. Generally, the ratio of membrane proteins to cytoplasmic ones is low due to limited space in the membrane compared to the cytoplasmic compartment (Kell et al. 2015). An increased expression of membrane-bound subunits of pumps might lead to disruption of membrane homeostasis (Wagner et al. 2007). Therefore expression of efflux pumps must be precisely tuned (Harrison and Dunlop 2012; Turner and Dunlop 2015; Boyarskiy et al. 2016; Basler et al. 2018). Moreover, cooperation between pumps might be needed for increased tolerance (Rojas et al. 2001; Turner and Dunlop 2015). Even though these studies used non-native pumps and artificial regulation systems, we believe that assessment of overall pump activity might be very helpful in searching for interconnections between increased tolerance and pump activity; cooperation and sensitive control might take place in natural systems too.

Screening of pump activities has become a common approach for the detection of antibiotic resistance caused by increased activity of multidrug efflux pumps (Patel et al. 2010). However, in multidrug resistance studies, cultivation-based determination of the MIC of EB serves as an indicator of stably increased efflux activity. With regard to the natural purpose of efflux pumps and the need for tight regulation, it seems obvious that an efflux rate must change according to actual cell requirements. Hence a method that would monitor efflux rate throughout the time course of cultivation might be very useful. Two quantitative assays for bacterial cultures or individual cells that can be used for such monitoring are described in the literature: an indirect accumulation assay (Rodrigues et al. 2011) and direct measurement of efflux of a suitable indicator. In order to identify a suitable method for quantification of efflux in C. beijerinckii, we tested both approaches in various experimental arrangements. An accumulation assay might be influenced by different factors. We think that the most critical is altered levels of EB penetration into cells due to changed permeability of the membrane, caused by an alteration in the composition of the cytoplasmic membrane (Blair and Piddock 2016) at various stages of ABE fermentation. This is a well-known phenomenon in solventogenic clostridia as an adaptive mechanism to increase in the solvent concentration (Kolek et al. 2015; Kusumawardhani et al. 2018). While the efflux of solvents is an energy-demanding process (Matsumoto et al. 2002), we searched for methodologies capable of distinguishing between active and passive fluxes of EB by modification of methodologies described previously (Viveiros et al. 2008; Paixão et al. 2009; Costa et al. 2018). During the experimental examination, we recognized different patterns of accumulation and efflux of EB depending on whether the energy source was present or not. Nevertheless, we chose the direct efflux measurement method, because EB accumulated in the absence of an energy source and afterwards pumped out in a glucose-containing growth medium provided a simpler and clearer quantification of efflux activity. The method based on quantification of the rate of active nonspecific efflux of EB from cells revealed that overall efflux activity changed significantly throughout the cultivation probably as a consequence of induction and suppression of a new set of transporters. Given the original plan to correlate efflux results with butanol production, we would like to mention that performing the analysis with EB we were able to reflect the total efflux rather than the specific efflux of solvents. Even so, we believe that knowledge of efflux changes introduced in this work helps understanding its fluctuation during the ABE fermentation as well as a possible contribution to the tolerance of solventogenic clostridia. At the same time, flow cytometry using different fluorescent probes can be an important tool for studying this area, as shown, for example, in a recent study where extensive screening of both efflux and influx was successfully performed using other fluorescent probes for mutants with a knockout of genes for various transporters (Jindal et al. 2019).