The ability to solubilize lignocellulose makes certain ionic liquids (ILs) very effective reagents for pretreating biomass prior to its saccharification for biofuel fermentation. However, residual IL in the aqueous sugar solution can inhibit the growth and function of biofuel-producing microorganisms. In E. coli this toxicity can be partially overcome by the heterologous expression of an IL efflux pump encoded by eilA from Enterobacter lignolyticus. In the present work, we used microarray analysis to identify native E. coli IL-inducible promoters and develop control systems for regulating eilA gene expression. Three candidate promoters, PmarR’, PydfO’, and PydfA’, were selected and compared to the IPTG-inducible PlacUV5 system for controlling expression of eilA. The PydfA’ and PmarR’ based systems are as effective as PlacUV5 in their ability to rescue E. coli from typically toxic levels of IL, thereby eliminating the need to use an IPTG-based system for such tolerance engineering. We present a mechanistic model indicating that inducible control systems reduce target gene expression when IL levels are low. Selected-reaction monitoring mass spectrometry analysis revealed that at high IL concentrations EilA protein levels were significantly elevated under the control of PydfA’ and PmarR’ in comparison to the other promoters. Further, in a pooled culture competition designed to determine fitness, the strain containing pPmarR’-eilA outcompeted strains with other promoter constructs, most significantly at IL concentrations above 150 mM. These results indicate that native promoters such as PmarR’ can provide effective systems for regulating the expression of heterologous genes in host engineering and simplify the development of industrially useful strains.
Citation: Frederix M, Hütter K, Leu J, Batth TS, Turner WJ, Rüegg TL, et al. (2014) Development of a Native Escherichia coli Induction System for Ionic Liquid Tolerance. PLoS ONE 9(7): e101115. https://doi.org/10.1371/journal.pone.0101115
Editor: John R. Battista, Louisiana State University and A & M College, United States of America
Received: December 3, 2013; Accepted: June 3, 2014; Published: July 1, 2014
Copyright: © 2014 Frederix et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was part of the DOE Joint BioEnergy Institute (http://www.jbei.org) supported by the U.S. Department of Energy, Office of Science, Office of Biological and Environmental Research, through contract DE-AC02-05CH11231 between Lawrence Berkeley National Laboratory and the U.S. Department of Energy. Kimmo Hutter was supported by his Master program at the Mannheim University of Applied Sciences and Prof. Matthias Mack. William Turner and Mary Dunlop were supported by the Office of Science (BER) at the U.S. Department of Energy and the NASA Vermont Space Grant Consortium. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: J.D.K. has a financial interest in Amyris, LS9 and Lygos. This does not alter the authors’ adherence to PLOS ONE policies on sharing data and materials.
The polysaccharides present in lignocellulosic biomass provide an attractive raw material for the production of renewable biofuels. Pretreatment of the biomass, e.g. via exposure to dilute acids at high temperatures or ammonia fiber expansion , , , , is necessary prior to saccharification of the material. Recently, hydrophilic ionic liquids (ILs) have emerged as pretreatment solvents because they are highly effective at solubilizing polysaccharides , , , . To minimize the costs associated with washing pretreated biomass, and recycling ILs, a residual level (0.2–5wt/vol%) of IL typically remains in downstream stages of an industrial scale production process . However, these low levels of IL, contaminating the sugar stream used for cultivation, are toxic to biofuel-producing microorganisms , , , . Residual IL may eventually be utilized in an industrial set-up, in combination with host organisms engineered for IL tolerance, to prevent contamination of the cultures.
The natural IL resistance of a rainforest bacterium, Enterobacter lignolyticus, has been investigated by transcriptome analysis, which revealed the differential expression of 688 genes in response to the IL 1-ethyl-3-methylimidazolium chloride ([C2mim]Cl) . Significant increases occur in several genes encoding membrane transporters, including one of the most highly upregulated genes that encodes a member of the multidrug efflux pump of the Major Facilitator Superfamily (MFS). This gene, eilA, was independently identified by functional screening for IL tolerance using a fosmid library of genomic DNA from E. lignolyticus . Heterologous expression of eilA in E. coli, either controlled by its native promoter or by PlacUV5, dramatically increases tolerance to [C2mim]Cl. A low level of IPTG (e.g. 10 µM) induction of PlacUV5 is optimal for expression of the EilA pump, whereas higher levels are inhibitory to microbial growth . A similar effect is also observed with the overexpression of other membrane proteins , .
Well-characterized induction systems like PlacUV5 are useful in laboratory studies of genes and pathways, but they are not amenable for use in industrial processes because of the cost of inducing reagents. Another drawback of PlacUV5 is the lack of flexible induction, since a given concentration of the inducer and time of induction must be selected in advance and remain constant throughout the cultivation period . An engineered microbial host may include numerous pathways and related genes that each require fine control, placing a limit on the number of available externally induced systems.
Because residual concentrations of IL in the saccharified biomass solution are likely to vary from batch to batch, dynamic control of gene expression would provide more uniform regulation and robust cellular growth than that afforded by constitutive expression , . In contrast to constitutive expression systems, which are optimal only under certain constant conditions, a condition-responsive expression system adjusts to the actual reaction environment. Such expression optimization may become particularly important when integrated with the expression of metabolic pathways or other tolerance mechanisms. One way of developing such a dynamic regulatory system is via a transcriptional regulator that induces the expression of the pump in the presence its substrate, as is often found in natural expression systems for efflux pumps , , . To this end, we use transcriptomics and proteomics to determine the usefulness of E. coli IL-responsive promoters to drive expression of the heterologous export pump EilA.
Results and Discussion
Transcriptomic response of E. coli to [C2mim]Cl
The E. coli DH1 toxicity profile to [C2mim]Cl was determined by measuring the optical density of cultures after addition of the IL over a range concentrations (0–400 mM) in mid-log phase. [C2mim]Cl exposure resulted an increase in the lag phase and the doubling time (Figure 1). The impact on the lag phase and the doubling time was more pronounced when [C2mim]Cl was added in the culture at the time of inoculation (Figure S1). The transcriptional response of E. coli to [C2mim]Cl was determined by microarray analysis, 30 and 60 minutes after adding 150 mM [C2mim]Cl at mid-log phase (GSE51731). A sub-lethal concentration was chosen in order to reveal transcriptional changes that are specific to [C2mim]Cl rather than to general differences in the stage of growth. All transcripts with a differential expression greater than three fold are listed in Table S1. Results with a p-value>0.05 in the 30 min sample were not considered. In the samples collected 30 minutes after exposure, 122 genes were significantly upregulated and 66 genes significantly downregulated. After 60 minutes, a subset of these genes were no longer differentially expressed, suggesting that E. coli starts to adapt to [C2mim]Cl. For other genes the transcriptional changes persisted, as 94 genes remained upregulated and 46 genes remained downregulated. The transcriptional changes suggest a general stress response, as several of the induced genes (e.g. ftnA, sodB, marRABC, pspABC, spy, degP) have previously been reported to be involved in oxidative, solvent or salt stresses , , , , , .
Characterization of selected [C2mim]Cl inducible promoters
To validate the microarray analysis, mRNA was prepared using conditions identical to the ones for the microarray analysis. Sixteen upregulated genes were chosen from the microarray dataset and their expression levels were evaluated by qPCR. Using this method, thirteen out of sixteen genes were confirmed to be significantly upregulated (Table 1). We chose three promoters, representing three different levels of induction, for further study: ydfO (high induction), ydfA (medium induction) and marR (low induction).
Of the genes corresponding to the promoters selected, marR is the most extensively studied. MarR is a transcriptional regulator that functions as an autorepressor  and is associated with regulating the expression of multidrug efflux systems, stress response systems, metabolic pathways and virulence factors , . Further, the expression of marR is regulated by several transcriptional regulators involved in stress response, and includes MarA, Rob and SoxS , . In contrast to MarR, very little is known about the primary function and regulation of ydfO and ydfA, both of which are predicted to encode genes of unknown function belonging to the Qin prophage. Interestingly, YdfO overexpression has been found to increase resistance of E. coli to oxidative stress .
To assess whether the upregulation of the selected promoters was specific to the presence of the [C2mim] cation rather than a general response to osmotic stress, the changes in expression of ydfO, ydfA and marR transcripts were measured in the presence of a similar IL, 1-ethyl-3-methylimidazolium acetate [C2mim][CH3COO] and the corresponding sodium salts of these ILs (Figure 2). Lower concentrations of the [C2mim][CH3COO] were selected to provide a comparable toxicity to [C2mim]Cl. The higher toxicity of [C2mim][CH3COO] may be due to the additive toxic effect of the acetate anion. A similar increased toxicity of [C2mim][CH3COO] compared to [C2mim]Cl has also previously been reported in S. cerevisiae . All three genes were induced in 100 mM [C2mim][CH3COO], but not in 150 mM NaCl or 100 mM Na[CH3COO]. Therefore, the ydfO, ydfA and marR promoters appear to respond specifically to the presence of the [C2mim]+ cation.
Cultures were grown until mid-exponential phase, when the compounds were added and RNA samples were collected after 30 min. The fold change on the chart represents the change in expression compared to an untreated control. hcaT was used as an endogenous reference to normalize the data. Error bars represent standard errors. Blue: 150 mM [C2mim]Cl, red: 100 mM [C2mim][CH3COO], green: 150 mM NaCl, purple: 100 mM Na[CH3COO].
Next, [C2mim]Cl was added at the beginning of growth and the expression levels of ydfO, ydfA and marR were determined by qPCR analysis at two time points (mid-exponential and stationary growth), in a range of 0–200 mM [C2mim]Cl (Figure 3). Under these growth conditions, which better represent anticipated IL levels present in fermenters, ydfO was no longer induced by [C2mim]Cl, but rather repressed in stationary phase at higher IL concentrations, indicating that the transcriptional response of the ydfO promoter is indeed growth phase dependent. ydfA and marR showed [C2mim]Cl dependent induction, and during exponential growth a clear dose dependence could be seen for both genes. In stationary phase, the induction pattern and promoter strength were very different. The inducibility of the marR promoter was reduced twofold, while the ydfA promoter became less inducible at higher concentrations of [C2mim]Cl (Figure 3).
[C2mim]Cl was added from the beginning of growth at the indicated concentrations, and cultures were harvested during exponential growth (OD600 = 0.5) and stationary growth. The fold change represents the change in expression compared to an untreated control. hcaT was used as an endogenous reference to normalize the data. Error bars represent standard errors. Blue: 50 mM [C2mim]Cl, red: 100 mM [C2mim]Cl, green: 150 mM [C2mim]Cl, purple: 200 mM [C2mim]Cl.
Evaluation of [C2mim]Cl responsive promoters to drive expression of eilA
It is known that overexpression of membrane proteins can lead to non-optimal cell growth . To evaluate the growth burden of the EilA pump, culture density was measured as a function of increasing IPTG levels. In the growth conditions used in the present study, expression of the pump did not affect the lag phase and slightly reduced the maximum OD, however, it did impact the doubling time during mid and late exponential growth (Figure S2). Based on these results and other studies , 10 µM IPTG was used as the induction level in subsequent assays.
To evaluate which of the [C2mim]Cl inducible promoters was best suited to drive the expression of the EilA pump, plasmid-borne eilA expression systems were constructed using the ydfO, ydfA and marR promoters and compared with the pPlacUV5-eilA construct induced with 10 µM IPTG in E. coli DH10B. A promoterless eilA construct (pP-eilA) served as the negative control. The resulting strains showed very similar lag, log and maximal growth under conditions without ILs (Figure 4).
A–E: growth assays of E. coli DH10B carrying different promoter-eilA constructs. Due to day-to-day variability in the final cell density reached in the microtiter-plate experiments, growth curves were normalized to a start OD of 0 and a maximum OD of 0.5. A) pPydfO’-eilA, B) pPydfA’-eilA, C) pPmarR’-eilA, D) pPlacUV5-eilA, E) pP-eilA. Error bars represent standard errors. Blue: 0 mM [C2mim]Cl, red: 100 mM [C2mim]Cl, green: 200 mM [C2mim]Cl, purple: 400 mM [C2mim]Cl.
Tested over a range of [C2mim]Cl concentrations (0 mM to 400 mM), the pPydfA’-eilA, pPmarR’-eilA, and pPlacUV5-eilA constructs performed similarly in conferring [C2mim]Cl resistance (Figure 4). As may be expected from the lack of inducibility of ydfO, in these test conditions (Figure 3), the pPydfO’-eilA construct did not provide any significant resistance, and growth was similar to the strain carrying the promoterless pP-eilA construct.
Since subtle differences in pump function or expression profile may result in a strain that has greater fitness , these strains were competed to establish which of the promoters was most suitable to drive the expression of eilA. For this, the different strains were pooled in equal proportions and grown in 0–400 mM [C2mim]Cl. For each IL concentration, the pooled cultures were grown for 48 h, during which they were subcultured six times into fresh medium, after which total plasmid DNA was isolated. qPCR analysis on the total plasmid DNA, using primers specific for the different constructs, allowed for quantification of the relative amounts of a specific strain in the pool (Figure 5). pP-eilA and pPlacUV5-rfp containing strains served as negative controls in the pool. The results of the competition assay (Figure 5) showed that the strain containing pPydfO-eilA also disappeared from the pool, at even the lowest levels of [C2mim]Cl. At low concentrations (50 and 100 mM), pPlacUV5-eilA, pPmarR’-eilA and pPydfA’-eilA performed equally well, while at concentrations above 150 mM of [C2mim]Cl, pPmarR’-eilA outperformed pPlacUV5-eilA and pPydfA’-eilA. pPydfA’-eilA outperformed pPlacUV5-eilA at 400 mM [C2mim]Cl (Figure 5).
Cultures were pooled and grown over 48-specific primers. 10 µM IPTG was added to induce the pPlacUV5-eilA construct. qPCR results were normalized using cat (the gene conferring chloramphenicol resistance on the expression plasmid) as endogenous control. Error bars represent standard errors. Dark blue: pPlacUV5-rfp, red: pP-eilA, green: pPydfO’-eilA, purple: pPydfA’-eilA, light blue: pPmarR’-eilA, orange: pPlacUV5-eilA.
Expression levels of EilA
EilA protein levels in the different strains were determined by selected-reaction monitoring (SRM) mass spectrometry  and were normalized to the amount of chloramphenicol acetyl transferase (Cat) protein (expressed from the backbone of all the EilA expression plasmids) in the samples to correct for plasmid copy number variation (Figure 6). To examine the condition-responsive nature of the promoter activity, samples were collected at two time points (exponential and stationary growth) and a range of [C2mim]Cl concentrations (0 mM, 50 mM, 150 mM and 400 mM). No data were obtained for the pP-eilA and pPydfO’-eilA strains at 400 mM [C2mim]Cl since no growth was observed, for these strains, in these conditions. Data for the remaining samples showed that the strains harboring pPydfO’eilA and pP-eilA contain very low EilA levels resulting from basal promoter activity. The basal EilA levels resulting from pPlacUV5-eilA were significantly higher, but lower than those from pPydfA’-eilA and pPmarR’-eilA. The strains harboring pPydfO’eilA and pP-eilA do not show IL-dependent induction in either growth phase. In contrast, the ydfA and marR promoters showed increased EilA levels at higher [C2mim]Cl concentrations during exponential growth. During stationary phase, no induction of EilA levels was detected at any of the [C2mim]Cl concentrations (Figure 6). The qPCR analysis had shown that the ydfA and marR promoters are inducible at this stage of growth (Figure 3) and a possible explanation is that the EilA accumulated in the cell membrane during exponential growth is sufficient to expel the IL, thus eliminating [C2mim]Cl inducibility of the ydfA and marR promoters in stationary phase. EilA expression from both these native promoters provides [C2mim]Cl tolerance in individually tested as well as mixed strain assays.
Strains contain eilA expression constructs are as indicated in the graph. EilA protein levels were quantified during exponential (A) and stationary (B) growth, at the indicated [C2mim]Cl concentrations. EilA protein levels were normalized against Cat protein levels as an endogenous control. No growth was observed for pP-eilA and pPydfO’-eilA in the presence of 400 mM [C2mim]Cl. Error bars represent standard errors. Blue: 0 mM [C2mim]Cl, red: 50 mM [C2mim]Cl, green: 150 mM [C2mim]Cl, purple: 400 mM [C2mim]Cl.
Differences in EilA levels were also seen for the PlacUV5 expression system. During exponential phase 10 µM IPTG resulted in equal levels of EilA over different concentrations of [C2mim]Cl, but this was not the case during stationary growth, where a decrease in EilA levels with increasing [C2mim]Cl levels, was seen. These data suggest that the toxicity of [C2mim]Cl has a direct or indirect impact on PlacUV5 function. While the mechanism of this decrease in promoter function is unclear, it might help explain the poorer performance of PlacUV5 at higher [C2mim]Cl concentrations. Alternatively, the basal expression levels of EilA in the pPlacUV5-eilA strain may generally be lower than the pPydfA’-eilA and pPmarR’eilA strains in the absence of [C2mim]Cl.
A model of EilA expression by different promoter systems
To provide insight into the results obtained with the various pump expression systems, we developed a mathematical model describing the dynamics of cell growth, substrate consumption, intracellular IL concentration, and pump expression. This model was used to compare the performance of the promoters that dynamically regulate pump expression in response to [C2mim]Cl (pPydfO’-eilA, pPydfA’-eilA, and pPmarR’-eilA), a static expression system (pPlacUV5-eilA), and a negative control with no promoter (pP-eilA). The toxicity of ionic liquids has a major impact on cell viability, while pump overexpression shows a mild reduction in cell growth. Therefore, we included both of these terms as negatively impacting the growth of the culture, deriving model parameters from experimental data (Methods). The growth model is coupled with equations describing the diffusion and transport of intracellular ILs and pump expression dynamics.
The mathematical model indicates that PmarR’ and PydfA’ are the most responsive for a range of [C2mim]Cl concentrations, showing good agreement with experimental findings (Figures 4 and 7). In the absence of [C2mim]Cl, PlacUV5 drives expression of the pump when it is not needed, reducing the growth rate. In contrast, condition-responsive induction allows the culture to grow to the maximum population density quite rapidly with less toxicity arising from pump expression. As the IL concentration is increased to an intermediate level of 200 mM [C2mim]Cl, PmarR’, PydfA’, and PlacUV5 exhibit better regulation of pump expression than PydfO’ or no promoter. This result is due to the low levels of expression provided by the fully induced PydfO’ and the lack of promoter in pP-eilA’, both of which are insufficient to eliminate IL toxicity. At high concentrations (400 mM [C2mim]Cl), the PmarR’, PydfA’, and the IPTG-induced PlacUV5 promoters enable growth, though PlacUV5 is at a disadvantage relative to the others due to an increased lag phase and a decreased cell concentration during stationary phase. Therefore, PmarR’ and PydfA’ provide good performance by eliminating the burden of pump expression when IL levels are low and expressing pumps highly when IL levels are high. Subtle differences between these two promoters can be detected with sensitive competition assays, like those shown in Figure 5. These findings have practical implications, as residual IL concentrations are likely to vary from batch to batch. In such varying conditions, static pump expression systems like pPlacUV5-eilA will always be at a disadvantage relative to expression systems that respond to IL levels in a condition dependent manner. A pump expression system that responds to a changing environment will be robust to variations in production conditions.
Promoters such as PlacUV5 are commonly used in metabolic and host engineering proof-of-concept studies , . However, developing economically-viable production systems for products such as biofuels restricts the use of expensive inducers such as IPTG. Further, the control provided by PlacUV5 and other such commonly used systems may not be optimal for regulating mechanisms that are required for tolerance towards compounds whose levels vary from batch to batch or even during the course of a given culture. Examples of these compounds include inhibitors present in the carbon source from pretreated biomass , , , ,  or accumulation of the biofuel product . Dynamic control of gene expression may provide superior regulation of tolerance genes , , and studies have demonstrated that dynamic control of metabolic pathway genes can result in improved strain stability and production . In the case of the EilA pump, an IL inducible system has also been established by using the pump associated E. lignolyticus repressor . However, while tolerance genes can often function in heterologous hosts, it is less common to discover and deploy a corresponding heterologous regulatory system. In this study we used microarray data measuring transcript level responses to select three promoters that may allow condition-responsive control of eilA, PydfO’, PydfA’ and PmarR’. Other promoters that were part of our initial list may have also provided good candidates for the optimal expression of eilA. Alternate resources available for E. coli host engineering can also be utilized for promoter selection. For example the promoter library developed by Zaslaver and coworkers  can be used to test the promoters that respond to ILs. Since the post translational regulation of the tolerance pump is different from that of a reporter protein such as RFP or the native protein being controlled, each selection criteria presents its own caveats and strengths. However, as gene synthesis capabilities become less expensive a larger number of regulatory systems can be tested in order to select the most ideal or alternate candidates.
Despite the fact that the regulatory mechanisms underlying the induction of PydfO’, PydfA’ and PmarR’ by [C2mim]Cl are unknown, we demonstrate the utility of these native host-organism promoters that are responsive to [C2mim]Cl, in developing control systems that provide condition-responsive control of target genes without the use of expensive reagents. Further, the use of orthogonal control systems may permit the integration of such tolerance mechanisms in strains with metabolic pathways with no conflict in control systems used. In this study, plasmid-based expression systems were used to demonstrate the concept of condition-dependent regulation of tolerance genes. For the development of an industrial host, strategies such as chromosomal integration may eliminate the need to use antibiotics, thus providing further avenues of optimization.
Materials and Methods
Bacterial strains, growth conditions and media
The E. coli strains used in this study were DH10B and DH1 (Invitrogen and ATCC33849). Bacteria were grown at 37°C in Luria-Bertani (LB) or M9 minimal medium (per liter: 200 ml 5×M9 salts, 2 ml 1 M MgSO4, 50 ml 20% glucose, 20 ml 5% casamino acids, 100 µl 0.5% thiamine, 100 µl 1 M CaCl2). Unless otherwise mentioned, all chemicals were purchased from Sigma-Aldrich (St. Louis, MO). Antibiotics were added as required to maintain plasmid selection. [C2mim]Cl (98%) was purchased from Sigma-Aldrich (St. Louis, MO) and [C2mim][CH3COO] (90%) was purchased from BASF (Ludwigshafen, Germany). Bacteria were adapted to M9 minimal medium by subculturing three times into fresh medium, after which they were stored as single-use glycerol stocks. For growth assays, strains were inoculated directly from frozen stocks into M9 minimal medium in 24-well plates and grown at 37°C in Tecan F200 or Tecan F200 pro microtiterplate readers (Maennedorf, Switzerland), measuring growth (OD 600) at 20 min intervals. For microarray analysis, strains were grown in 25 mL cultures in 250 mL flasks. For qPCR and proteomics analyses, strains were grown in 5 mL cultures tubes. For competitive growth experiments, 5 mL precultures were grown overnight in M9 minimal medium and pooled before the start of the competition assay.
Construction of plasmids
The plasmids used in this study are listed in Table 2. eilA expression constructs were generated by circular polymerase extension cloning (CPEC)  by replacing the lacI repressor gene and the lacUV5 promoter with the promoter regions of the ydfO, ydfA and marR genes. Primers used for amplification of vector and promoters are listed in Table S2.
RNA isolation, microarray and qPCR analysis
For RNA extraction, cultures were collected in RNAlater (Qiagen, Hilden, Germany) to stabilize the RNA. Total RNA was isolated using the RNeasy Mini kit (Qiagen, Hilden, Germany). RNA quality and quantity were analyzed on 1% agarose gels and quantified using a NanoDrop ND-1000 spectrophotometer (Thermo Fisher, Waltham, MA). Labeling of RNA was performed as described previously , and the resulting cDNA was hybridized on commercially available E. coli K12 microarrays according to the manufacturers instructions (Nimblegen, Madison, WI). Microarray data analysis and normalization was performed using Arraystar (DNASTAR Inc., Madison, WI). To determine differentially expressed transcripts, an arbitrary cut-off of threefold change was used and genes with a p-value higher than 0.05 were omitted from the analysis. Microarray data have been deposited in the GEO database (GSE51731).
For qPCR analyses, purified RNA was additionally treated by Turbo DNase (Ambion Invitrogen, Carlsbad, CA), and reverse transcribed by Superscript III (Invitrogen, Grand Island, New York, USA) according to the manufacturers instructions. 1 µl of tenfold diluted cDNA was used as template in 20-µl reactions using Evagreen qPCR mix (Biorad, Hercules, CA). Primers used for amplification in the qPCR reaction are listed in Table S2. As endogenous controls cat (for competition experiments) or hcaT (other experiments ) were used and the fold change in expression of target genes was calculated by the 2-(ΔΔ-C(T)) method .
Cultures were grown for 18 h, and samples were processed as described in . Relative protein production levels were determined by using targeted proteomics as described . Samples were analyzed on an AB Sciex 5500Q-Trap mass spectrometer (AB Sciex Foster City, CA) operating in MRM-mode coupled to an Agilent 1100 HPLC (Agilent, Santa Clara, CA) operating in micro mode. SRM transitions and collision energies for the peptides were picked via the Skyline software . Briefly, SRM selection excluded peptides containing cysteines, methionine and peptides with missed or repeating tryptic cleavage sites. Two peptides were selected for each protein with 2–3 y-series transitions per peptide. For analysis, 5 ug of each peptide digest was injected with 1 picomole of bovine serum albumin (BSA) as internal standard. Peptides were separated on a Agilent Zorbax 300 SB-C18 column (5 µm bead size, 150×0.5 mm) at 50 µl/min. The separation gradient was as follows: column was equilibrated with 98% A (2% Acetonitrile, 0.1% Formic acid) for 0.5 min and decreased to 65% A, 35% B (98% Acetonitrile, 0.1% Formic acid) over 7.5 min. The gradient was quickly ramped up to 90% B in one min and held at 90% B for 1.5 min after which it was ramped back down to 98% A in 1 min where it was held for 8.5 min to re-equilibrate the system for the next sample. Targeted proteomics data were analyzed using Skyline and quantification was based on the peak area for highest intensity transition for each peptide. All peak area integrations were performed by Skyline and were manually reviewed. Peak areas for each peptide were summed for corresponding protein and normalized to the sum of the Cat peak areas. Error bars represent the standard deviation of three biological replicates.
A system of ordinary differential equations was used to model the rate of change of biomass (N), substrate (S), pump (p), and intracellular IL (ci) concentrations.
Cell growth (Eq. 1) and substrate consumption (Eq. 2) were modeled using a modified form of the Monod equation . The maximum growth rate (µmax), growth yield (γ), and half-saturation constant (Ks) were selected to match the experimentally measured growth dynamics in the absence of [C2mim]Cl (Figure S3). The model includes inhibition by intracellular ILs and pump expression. The half-inhibition constant for IL, Kc, and the Hill coefficient, h, were fit to the experimental data at stationary phase (Figure S4). The value of the half-maximum inhibition constant for pumps, Kp, was set to match the data from Figure S2 for a range of relevant pump expression constants, αp (Figure S5, Eq. 4).(1)(2)
The equation used to model the rate of change of pump protein concentration depends on the promoter being modeled. For all promoter types there is basal expression, αp0, which was set to the exponential phase protein measurement without any IL present (Figure 6A). Also, in all cases, EilA proteins decay at the rate, β, which models both dilution due to cell growth and protein degradation. Thus, equation 3 represents protein production in the case of the promoterless construct, P-eilA, having only basal expression and degradation.(3)
Equation 4 was used to model pump protein expression for the PlacUV5 promoter. Along with the basal expression, this promoter has a constant protein expression level corresponding to the concentration of IPTG in the solution. This is represented by the constant, αp, which was set such that the total expression rate would equal the average of the protein measurements at the stationary phase (Figure 6B).(4)
In the case of the dynamic controllers, PydfO’, PydfA’, and PmarR’, the protein production rate depends on the intracellular IL concentration. Figure 6A indicates that the PydfA’ and PmarR’ promoters respond faster at higher concentrations of [C2mim]Cl due to the higher protein measurements in the exponential phase. Equation 5 represents the rate of change of pump protein for the dynamic controllers. Where the maximum (non-basal) protein production rate, αp, was set to the average of the protein measurements at stationary phase, and the half-maximum, γc, was set to the concentration of IL that yields half of the maximum protein concentration during exponential growth (Figure 6A & B).(5)
In order to model the intracellular IL concentration (ci), an IL mass balance was incorporated into the model to ensure that the total mass of IL in the reactor was constant and equal to the IL mass in the intracellular environment plus the IL mass in the extracellular environment. Passive diffusion of [C2mim]Cl through the cell membrane was modeled using Eq. 6.(6)
The gradient of IL concentration across the cell membrane is the driving force of this diffusion, and the concentration change is accounted for by the ratio between the extracellular and intracellular volumes. Intracellular volume was calculated by converting cell mass to number of cells and scaling by the volume of an E. coli cell. In all simulations IL is also actively exported from the cell at a rate proportional to the intracellular IL concentration and the pump protein concentration. The rate of export per pump protein, αc, was set to match experimental data (Figure 7). All model constants are given in Table 3. All simulations were performed in MATLAB R2012a (MathWorks) using the ode45 solver.
Toxicity of [C2mim]Cl to E. coli DH1 upon addition of [C2mim]Cl. Red: 0 mM [C2mim]Cl, purple: 50 mM [C2mim]Cl, blue: 100 mM [C2mim]Cl, green: 150 mM [C2mim]Cl, yellow: 200 mM [C2mim]Cl, orange: 400 mM [C2mim]Cl. Error bars represent standard errors.
Toxicity of pPlacUV5-eilA construct in E. coli DH10B with increasing IPTG concentrations. Expression of the pump was induced by different concentrations of IPTG. Dark blue: 0 µM IPTG, red: 10 µM IPTG, green: 50 µM IPTG, purple: 100 µM IPTG, light blue: 200 µM IPTG, orange: 300 µM IPTG.
Model fit to experimental data for growth in the absence of inhibitors. Blue dots: Experimental data, error bars represent standard error, light Blue: Simulated growth curve.
Final biomass concentration as a function of IL concentration. Blue asterisk: experimental data, red: Simulated final values.
Simulation of toxicity of pPlacUV5-eilA and no IL present. Blue: αp = 0, red: αp = 0.05, green: αp = 0.1, purple: αp = 0.2, light blue: αp = 0.3.
Microarray results after addition of 150 mM [C2mim]Cl to exponentially growing E. coli cells. Only genes with a change in expression of three-fold or higher and a p-value<0.05 at the 30 min time-point are listed.
Conceived and designed the experiments: MF KH CJP MJD AM. Performed the experiments: MF KH JL TSB WJT. Analyzed the data: MF WJT CFP AM. Contributed reagents/materials/analysis tools: TLR MPT HB BAS PDA JDK. Wrote the paper: MF WJT MJD CJP AM.
- 1. Wyman CE, Dale BE, Elander RT, Holtzapple M, Ladisch MR, et al. (2005) Comparative sugar recovery data from laboratory scale application of leading pretreatment technologies to corn stover. Bioresour Technol 96: 2026–2032.
- 2. Holtzapple MT, Jun JH, Ashok G, Patibandla SL, Dale BE (1991) The ammonia freeze explosion (Afex) process - a practical lignocellulose pretreatment. Applied Biochemistry and Biotechnology 28–9: 59–74.
- 3. Lau MW, Gunawan C, Dale BE (2009) The impacts of pretreatment on the fermentability of pretreated lignocellulosic biomass: a comparative evaluation between ammonia fiber expansion and dilute acid pretreatment. Biotechnology for Biofuels 2.
- 4. Suryawati L, Wilkins MR, Bellmer DD, Huhnke RL, Maness NO, et al. (2009) Effect of hydrothermolysis process conditions on pretreated switchgrass composition and ethanol yield by SSF with Kluyveromyces marxianus IMB4. Process Biochemistry 44: 540–545.
- 5. Li Q, Jiang XL, He YC, Li LZ, Xian M, et al. (2010) Evaluation of the biocompatibile ionic liquid 1-methyl-3-methylimidazolium dimethylphosphite pretreatment of corn cob for improved saccharification. Applied Microbiology and Biotechnology 87: 117–126.
- 6. Mora-Pale M, Meli L, Doherty TV, Linhardt RJ, Dordick JS (2011) Room temperature ionic liquids as emerging solvents for the pretreatment of lignocellulosic biomass. Biotechnology and Bioengineering 108: 1229–1245.
- 7. Singh S, Simmons BA, Vogel KP (2009) Visualization of biomass solubilization and cellulose regeneration during ionic liquid pretreatment of switchgrass. Biotechnology and Bioengineering 104: 68–75.
- 8. Yoon LW, Ngoh GC, Chua ASM, Hashim MA (2011) Comparison of ionic liquid, acid and alkali pretreatments for sugarcane bagasse enzymatic saccharification. Journal of Chemical Technology and Biotechnology 86: 1342–1348.
- 9. Klein-Markuschamer D, Simmons B, Blanch HW (2011) Techno-economic analysis of a lignocellulosic ethanol biorefinery with ionic liquid pre-treatment. Biofuel Bioprod Bior 5: 562–569.
- 10. Pham TP, Cho CW, Yun YS (2010) Environmental fate and toxicity of ionic liquids: a review. Water Research 44: 352–372.
- 11. Ouellet M, Datta S, Dibble DC, Tamrakar PR, Benke PI, et al.. (2011) Impact of ionic liquid pretreated plant biomass on Saccharomyces cerevisiae growth and biofuel production. Green Chem: 2743–2749.
- 12. Quijano G, Couvert A, Amrane A (2010) Ionic liquids: Applications and future trends in bioreactor technology. Bioresource Technology 101: 8923–8930.
- 13. Romero A, Santos A, Tojo J, Rodriguez A (2008) Toxicity and biodegradability of imidazolium ionic liquids. Journal of Hazardous Materials 151: 268–273.
- 14. Khudyakov JI, D'Haeseleer P, Borglin SE, Deangelis KM, Woo H, et al.. (2012) Global transcriptome response to ionic liquid by a tropical rain forest soil bacterium, Enterobacter lignolyticus. Proc Natl Acad Sci U S A.
- 15. Rüegg TL, Kim E, Simmons BA, Keasling JD, Singer SW, et al. (2014) An auto-inducible mechanism for ionic liquid resistance in microbial biofuel production. Nature Communications 5: 3490.
- 16. Wagner S, Baars L, Ytterberg AJ, Klussmeier A, Wagner CS, et al. (2007) Consequences of membrane protein overexpression in Escherichia coli. Molecular & Cellular Proteomics 6: 1527–1550.
- 17. Dunlop MJ, Keasling JD, Mukhopadhyay A (2010) A model for improving microbial biofuel production using a synthetic feedback loop. Syst Synth Biol 4: 95–104.
- 18. Holtz WJ, Keasling JD (2010) Engineering static and dynamic control of synthetic pathways. Cell 140: 19–23.
- 19. Grkovic S, Brown MH, Skurray RA (2001) Transcriptional regulation of multidrug efflux pumps in bacteria. Semin Cell Dev Biol 12: 225–237.
- 20. Harrison ME, Dunlop MJ (2012) Synthetic feedback loop model for increasing microbial biofuel production using a biosensor. Front Microbiol 3: 360.
- 21. Touati D, Jacques M, Tardat B, Bouchard L, Despied S (1995) Lethal oxidative damage and mutagenesis are generated by iron in delta fur mutants of Escherichia coli: protective role of superoxide dismutase. J Bacteriol 177: 2305–2314.
- 22. Hopkin KA, Papazian MA, Steinman HM (1992) Functional differences between manganese and iron superoxide dismutases in Escherichia coli K-12. J Biol Chem 267: 24253–24258.
- 23. Chubiz LM, Rao CV (2010) Aromatic acid metabolites of Escherichia coli K-12 can induce the marRAB operon. J Bacteriol 192: 4786–4789.
- 24. Brissette JL, Weiner L, Ripmaster TL, Model P (1991) Characterization and sequence of the Escherichia coli stress-induced psp operon. J Mol Biol 220: 35–48.
- 25. Quan S, Koldewey P, Tapley T, Kirsch N, Ruane KM, et al. (2011) Genetic selection designed to stabilize proteins uncovers a chaperone called Spy. Nat Struct Mol Biol 18: 262–269.
- 26. Skorko-Glonek J, Zurawa D, Kuczwara E, Wozniak M, Wypych Z, et al. (1999) The Escherichia coli heat shock protease HtrA participates in defense against oxidative stress. Mol Gen Genet 262: 342–350.
- 27. Saridakis V, Shahinas D, Xu X, Christendat D (2008) Structural insight on the mechanism of regulation of the MarR family of proteins: high-resolution crystal structure of a transcriptional repressor from Methanobacterium thermoautotrophicum. J Mol Biol 377: 655–667.
- 28. Wilkinson SP, Grove A (2006) Ligand-responsive transcriptional regulation by members of the MarR family of winged helix proteins. Curr Issues Mol Biol 8: 51–62.
- 29. White DG, Goldman JD, Demple B, Levy SB (1997) Role of the acrAB locus in organic solvent tolerance mediated by expression of marA, soxS, or robA in Escherichia coli. J Bacteriol 179: 6122–6126.
- 30. Barbosa TM, Levy SB (2000) Differential expression of over 60 chromosomal genes in Escherichia coli by constitutive expression of MarA. J Bacteriol 182: 3467–3474.
- 31. Martin RG, Rosner JL (2002) Genomics of the marA/soxS/rob regulon of Escherichia coli: identification of directly activated promoters by application of molecular genetics and informatics to microarray data. Mol Microbiol 44: 1611–1624.
- 32. Wang X, Kim Y, Ma Q, Hong SH, Pokusaeva K, et al. (2010) Cryptic prophages help bacteria cope with adverse environments. Nat Commun 1: 147.
- 33. Dunlop MJ, Dossani ZY, Szmidt HL, Chu HC, Lee TS, et al. (2011) Engineering microbial biofuel tolerance and export using efflux pumps. Molecular Systems Biology 7: 487.
- 34. Redding-Johanson AM, Batth TS, Chan R, Krupa R, Szmidt HL, et al. (2011) Targeted proteomics for metabolic pathway optimization: application to terpene production. Metab Eng 13: 194–203.
- 35. Nichols NN, Sharma LN, Mowery RA, Chambliss CK, van Walsum GP, et al. (2008) Fungal metabolism of fermentation inhibitors present in corn stover dilute acid hydrolysate. Enzyme and Microbial Technology 42: 624–630.
- 36. Palmqvist E, Hahn-Hagerdal B (2000) Fermentation of lignocellulosic hydrolysates. I: inhibition and detoxification. Bioresource Technology 74: 17–24.
- 37. Palmqvist E, Hahn-Hagerdal B (2000) Fermentation of lignocellulosic hydrolysates. II: inhibitors and mechanisms of inhibition. Bioresource Technology 74: 25–33.
- 38. Pienkos PT, Zhang M (2009) Role of pretreatment and conditioning processes on toxicity of lignocellulosic biomass hydrolysates. Cellulose 16: 743–762.
- 39. Dahl RH, Zhang F, Alonso-Gutierrez J, Baidoo E, Batth TS, et al. (2013) Engineering dynamic pathway regulation using stress-response promoters. Nature biotechnology 31: 1039–1046.
- 40. Zaslaver A, Bren A, Ronen M, Itzkovitz S, Kikoin I, et al. (2006) A comprehensive library of fluorescent transcriptional reporters for Escherichia coli. Nat Methods 3: 623–628.
- 41. Quan J, Tian J (2009) Circular polymerase extension cloning of complex gene libraries and pathways. Plos One 4: e6441.
- 42. Ouellet M, Adams PD, Keasling JD, Mukhopadhyay A (2009) A rapid and inexpensive labeling method for microarray gene expression analysis. Bmc Biotechnology 9.
- 43. Zhou K, Zhou L, Lim Q, Zou R, Stephanopoulos G, et al. (2011) Novel reference genes for quantifying transcriptional responses of Escherichia coli to protein overexpression by quantitative PCR. BMC Mol Biol 12: 18.
- 44. Livak KJ, Schmittgen TD (2001) Analysis of relative gene expression data using real-time quantitative PCR and the 2(T)(-Delta Delta C) method. Methods 25: 402–408.
- 45. MacLean B, Tomazela DM, Shulman N, Chambers M, Finney GL, et al. (2010) Skyline: an open source document editor for creating and analyzing targeted proteomics experiments. Bioinformatics 26: 966–968.
- 46. Monod J (1949) The Growth of Bacterial Cultures. Annual Review of Microbiology 3: 371–394.