Research Grants/Fellowships/SBIR

Final Report: Metabolic Engineering of Solvent Tolerance in Anaerobic Bacteria

EPA Grant Number: R828562
Title: Metabolic Engineering of Solvent Tolerance in Anaerobic Bacteria
Investigators: Papoutsakis, E. T. , Welker, N. E.
Institution: Northwestern University
EPA Project Officer: Richards, April
Project Period: June 1, 2000 through May 31, 2003
Project Amount: $180,000
RFA: Technology for a Sustainable Environment (1999) RFA Text |  Recipients Lists
Research Category: Sustainability , Pollution Prevention/Sustainable Development



Understanding solvent (and other toxic chemical) tolerance of microorganisms is crucial for the production of chemicals, bioremediation, and whole-cell biocatalysis. Also, it is very important basic knowledge. Past efforts to produce tolerant strains have relied on selection under applied pressure and chemical mutagenesis, with some good results, but not consistently so. We examined metabolic engineering (ME) and genomic approaches to determine if they can be used to construct more tolerant strains for bioprocessing. The accepted dogma is that toxicity results from the chaotropic effects of solvents on the cell membrane. Impaired membrane fluidity and function inhibit cell metabolism and result in cell death. We have found that in Clostridium acetobutylicum, several well-defined genetic modifications not related to membrane function impart solvent tolerance (by 40-70 percent) without strain selection. This suggests that we need to reexamine the accepted dogma. The objective of this research project was to identify genes that contribute to solvent tolerance and to use genetic modifications (involving these genes) to generate solvent tolerant strains. In view of the large number of possible genes that may be involved in determining solvent tolerance, we used DNA microarrays for transcriptional analysis.

Summary/Accomplishments (Outputs/Outcomes):

The goal of ME is to alter the metabolism of a cell in a manner that results in desirable cellular traits. In the context of this project, the major objective was to alter the metabolism of the strict anaerobe C. acetobutylicum ATCC 824 to increase the production of solvents (acetone and butanol), which are produced as part of this organism's natural metabolic pathway. A major class of proteins known as heat shock proteins (HSPs) have been implicated in the ability of C. acetobutylicum to tolerate relatively high levels of solvents achieved in wild type fermentations. The main focus of this interdisciplinary research project was on the role of HSPs in the solvent tolerance of C. acetobutylicum. Specifically, the role of the major native HSP family GroESL (groES and groEL) on butanol tolerance was examined. The hypothesis that overexpression of the major HSPs will result in increased butanol titers as a result of increased protein stability was tested. Overexpression of the GroESL HSPs was achieved using replicative plasmids carrying copies of the major heat shock operon genes. Alternatively, gene inactivation of the putative HSP repressor, orfA, could be used to overexpress both GroESL and the other major HSP family, DnaKJ. Although gene inactivation in C. acetobutylicum is difficult, several alternative approaches were developed.

Additionally, a replicative plasmid expressing the aforementioned putative repressor (orfA) was utilized to downregulate the HSPs. The expression levels of the HSPs was confirmed using quantitative reverse transcription (Q-RT) polymerase chain reaction (PCR), DNA-arrays, and Western blot analysis. The in vivo stability of the solvent formation proteins was examined through time-course Western analysis. Product profiles and metabolic flux analysis were used to characterize strains demonstrating increased solvent production. DNA microarrays were used to examine the effect of HSP overexpression on global gene expression and the response of C. acetobutylicum to butanol stress.

Education and Outreach Activities

The project represented a collaborative effort between biochemical engineering and molecular microbiology and includes the education and training of three undergraduates, four graduate students, and one postdoctoral fellow. We have trained students at all levels in modern methods and techniques in biochemical engineering, metabolic engineering, genetics, fermentation technology, molecular biology, and high throughput. We have talked to several biotech companies regarding the importance of our approach and the likely impact of our findings in bioprocessing and bioremediation applications. We also have presented our results at several relevant conferences in both the United States and Europe.

A Replicative Plasmid for Overexpression of orfA, a Putative Repressor of the groESL and dnaKJ Operons

The major HSPs DnaKJ and GroESL are among the most widely studied and best understood sets of proteins. Both groups of proteins are essential to normal cellular function, and they play an important role in cell physiology. HSPs bind proteins in non-native states and assist in their proper folding. Although the mechanism by which HSPs assist in the proper folding of proteins is well understood, the role and regulation of the DnaKJ and GroESL protein families in C. acetobutylicum are not well understood. OrfA is a putative repressor of the groESL and dnaKJ operons, as identified by homology to HrcA in Bacillus subtilis. HrcA has been shown to operate as a negative repressor through interaction with a Controlling Inverted Repeat of Chaperone Expression (CIRCE) element located upstream of both the dnaKJ and groESL operons. An identical sequence has been identified in C. acetobutylicum. The orfA gene was cloned from the dnaKJ operon using PCR primers that amplified the structural sequence, but not the regulatory sequences (i.e., promoter and CIRCE elements). The orfA gene was ligated to an Escherichia coli/C. acetobutylicum shuttle vector (pSOS95del) such that orfA is under the control of the clostridial thiolase (thl) promoter, which has been shown to be constituitively expressed. The resulting plasmid (pORFA1) was transformed into C. acetobutylicum ATCC 824. Overexpression of orfA should result in decreased expression of both the groESL and dnaKJ operon genes.

Northern Blot Analysis of the orfA Overexpression Strain

Northern blot analysis of the orfA overexpression strain 824(pORFA1) was compared to that of the wild type 824 strain to verify increased transcription of orfA from pORFA1. A probe designed to detect the orfA gene resulted in a strong 1.1 kb band in the 824(pORFA1) samples that was not present in the wild type 824 samples. This corresponds to the expected size of the orfA gene from pORFA1. The same orfA probe also resulted in a 3.8 kb band present in the wild type samples, which was not present in the 824(pORFA1) samples. The 3.8 kb band corresponds to the expected 3.8 kb transcript from the complete dnaKJ operon. The fact that this transcript is absent in the orfA overexpressing strain is expected if the OrfA protein is in fact acting as a repressor of the DnaKJ operon. A second probe for the dnaKJ genes was used to confirm this finding. A third probe for the groESL operon genes resulted in a 2.2 kb band corresponding to the expected transcript size. The 2.2 kb band was an order of magnitude more intense on the wild type blot relative to the orfA overexpression strain. This is the first direct evidence that OrfA acts as a repressor of the dnaKJ and groESL operons in C. acetobutylicum. These findings also demonstrate the usefulness of pSOS95del for overexpression of various genes.

Gene Inactivation (orfA) Using Replicative Plasmids

Methods for targeted gene inactivation have proved to be difficult in C. acetobutylicum. A recent improvement in the technology includes use of a replicative plasmid, which allows for a double crossover event to occur. Double crossover integrations do not result in insertion of the entire plasmid; therefore, they are more stable than single crossover integrations, which incorporate the entire plasmid. A replicative plasmid capable of a double crossover was developed in our laboratory.

The vector contains a pIM13 Gram-positive replicon and a chloramphenicol resistance (Cmr) gene. An internal gene fragment can be cloned into the vector. An MLSr marker then is cloned into the center of the gene fragment. The strain carrying the plasmid was grown on plates without antibiotic selection and replica plated daily onto fresh plates. After several days, the culture was plated on antibiotic selection plates, and bacteria that have incorporated the MLSr marker into the chromosome were selected. This method depends on the loss of the plasmid after integration. However, the pIM13 replicon has proved to be quite stable, making it very difficult to isolate inactivation mutants. Cells retaining the plasmid mask any cells that may have undergone an integration event. A plasmid was constructed to inactivate the orfA gene based on this plasmid. However, because of the stability of the replicative plasmid, selection of an inactivation mutant has proved to be difficult. In an attempt to further improve this method, a vector that utilizes a temperature-sensitive replicon from pE194 has been constructed (pRefFColE1). A temperature shift from 32°C to a higher temperature (37°C - 40°C) should result in the loss of the plasmid at a much higher rate than with the pIM13 replicon.

Creation of pAN2 for Methylation of Cmr Plasmids

The gene inactivation method described above utilizes a vector carrying a Cmr gene, as well as an MLSr marker encoding resistance to erythromycin. Methylation of the knockout plasmids with pAN1, which also carries a Cmr gene, can be difficult for two reasons. First, both plasmids carry a Cmr gene, making loss of the pAN1 plasmid possible. Second, use of erythromycin for selection in E. coli is less than ideal because of the relatively high level of natural resistance E. coli has towards erythromycin. Therefore, a sister plasmid, pAN2, has been created in which the Cmr gene has been replaced by an ampicillin resistance (Apr) gene. The Apr gene was obtained by PCR amplification from plasmid pIMP1, a commonly used E. coli/C. acetobutylicum shuttle vector. The Cmr gene was removed from pAN1 by restriction enzyme digestion, and the Apr gene was inserted in its place. This new sister plasmid will allow easy methylation of plasmids containing the Cmr gene. As the use of multiple antibiotics for selection of single recombinant strains becomes more prevalent, this new methylating plasmid will become increasingly useful.

Replicative Plasmids for Overexpression of dnaKJ and groESL

The operons encoding the DnaKJ and GroESL proteins have been previously cloned and sequenced, but the effect of altering their levels in vivo has yet to be studied. As previous studies of the dnaKJ operon only utilized partial fragments of the operon containing no more than one complete gene sequence, it was necessary to clone a fragment of the operon containing all three functional genes (grpE, dnaK, and dnaJ). This was achieved by designing PCR primers such that grpE, dnaK, and dnaJ were amplified without the orfA gene (encoding for a putative repressor of the groESL and dnaKJ operons) or the putative regulatory element (CIRCE) discussed above. Similarly, the groESL operon genes (groES and groEL) were amplified without their natural promoters or the CIRCE element. The CIRCE element was omitted to liberate the genes from control of the chromosomal OrfA produced during normal cell growth. The amplified fragments were ligated to two E. coli/C. acetobutylicum shuttle vectors, pSOS94del and pSOS95del, such that they were under the control of the phosphotransbutyrylase (ptb) and thl promoters, respectively. Initial efforts to identify an E. coli clone in which the groESL and dnaKJ genes were successfully inserted into the vector proved rather difficult. It has been shown that overproduction of HSPs is toxic in a number of hosts (including E. coli and B. subtilis). It was hypothesized that transformation and growth of E. coli at a lower temperature (25°C) may decrease the initial level of stress created by increased HSP production. This hypothesis proved to be correct in that it resulted in the identification of positive clones. It was subsequently shown that positive clones grew up to 1,000 times better at 25°C versus cells grown at 37°C, as determined by colony-forming units/mL counts. Before C. acetobutylicum ATCC 824 can be transformed with these four constructs, it is necessary to methylate the constructs by passing them through the methylating strain E. coli ER2275(pAN1). Plasmid pAN1 contains the capitol phi3T I methyltransferase gene, which methylates Cac824I restriction sites. Similar difficulties were encountered with this transformation. The constructs expressing groESL (pGROE1) and dnaKJ (pDNAK1) under control of the thl promoter have been successfully methylated and inserted into C. acetobutylicum.

Fermentations With 824(pGROE1) and 824(pSOS95del)

A series of duplicate fermentations with the 824(pGROE1), 824(pSOS95del) control strain, and wild type 824 were conducted to examine the effects of groESL overexpression on solvent formation, metabolic fluxes, and transcriptional and protein expression patterns. In addition, the effects of the host-plasmid effect (wild type versus plasmid control) were examined. The fermenters were kept anaerobic with nitrogen, and a low end pH value of 5.0 was maintained with ammonium hydroxide. Glucose was fed once during the fermentation to prevent complete depletion of glucose. Supernatants were collected for product formation analysis by high-performance liquid chromatography, cell pellets were harvested for use in Western blot analysis, and RNA samples were taken for use in microarray and RT-PCR analysis.

Product Formation and Metabolic Fluxes

The presence of the pGROE1 plasmid had dramatic effects on product formation, particularly in the formation of acetone and butanol, when compared to both the wild type and control strains. The 824(pGROE1) strain produced 148 mM and 231 mM of acetone and butanol, respectively, compared to 107 mM and 178 mM in the control strain and 96 mM and 175 mM in wild type. This represents an increase in final acetone and butanol titers of 66 percent and 56 percent, respectively, relative to the 824(pSOS95del) control strain. It also is interesting to note that the onset of solvent production is delayed and appears to occur in two distinct phases for both recombinant strains. This is likely because of a frequently observed host-plasmid interaction. Final ethanol titers were slightly lower in the 824(pSOS95del) and 824(pGROE1) strains (23 mM and 21 mM, respectively) compared to the wild type 824 strain (28 mM). Acetate levels showed no statistical difference between strains, but butyrate levels were slightly lower in the two recombinant strains (73 mM and 70 mM) compared to the wild type (80 mM). The 824(pGROE1) strain grew to higher optical densities than the 24(pSOS95del) control strain, but was slightly lower than the wild type strain. The doubling times for the two recombinant strains were nearly identical (2.01 and 1.99 hours), both exhibiting slower exponential growth than the wild type (1.24 hours).

An examination of specific in vivo fluxes for a number of key metabolic reactions provides an excellent portrait of differences in carbon flows between various strains, while accounting for differences in cell densities. A metabolic flux analysis program, CompFlux, was utilized to calculate the metabolic in vivo fluxes for the wild type and recombinant strains. The specific in vivo fluxes show drastic differences between the wild type 824, 824(pSOS95del), and 824(pGROE1) strains. As previously mentioned, both recombinant strains exhibit two distinct phases, but the wild type exhibits only one. This pattern has been observed for other plasmid carrying strains as well, and appears to be a generalized plasmid effect. Strain 824(pGROE1) exhibited an elevated glucose utilization (rGLY1) and acetyl-CoA utilization (rTHL) relative to the control strain. The acetate and butyrate uptake rates also were higher in strain 824(pGROE1), which results in an increased acetone formation flux (rACETONE). Acetone is formed through the uptake of either acetate or butyrate. For strain 824(pGROE1), acetate uptake (rACUP) appears to play a larger role in increased acetone production than butyrate uptake. The increases in both butyrate uptake (rBYUP) and butyrate formation (rPTBBK) are relatively small when compared to the increases in acetate uptake and acetate formation (rPTAAK). Strain 824(pGROE1) also exhibited butanol (rBUOH) formation fluxes that were significantly higher than in strain 824(pSOS95del). The differences observed in the in vivo fluxes are consistent with the observation of higher final solvent titers in strain 824(pGROE1), and further illustrate the delay in solvent production in both recombinant strains. Active metabolism lasted 2.5 times longer in 824(pGROE1) and 824(pSOS95del) than in wild type. Transcriptional and Western analysis data (below) suggest that this is because of the effect of GroESL overexpression on stabilizing cellular proteins and the biosynthetic machinery, which is consistent with our hypothesis. This was further supported by the finding that growth of 824(pGROE1) was up to 85 percent less inhibited by a butanol challenge relative to the control strain 824(pSOS95del).

DNA-Array Based Transcriptional Analysis of groESL Overexpression Demonstrates Large Changes in Key Cellular Transcriptional Programs

DNA arrays containing a select set of 1019 genes covering more than 25 percent of the whole genome were designed, constructed, and validated for data reliability. Following the identification of differentially expressed genes using our own bioinformatic tools, clustering of expression profiles provided useful functional information. Hierarchical clustering uses pairwise comparison to obtain an expression similarity tree. Partitional clustering groups genes based on the similarity of their expression profiles. Self-organizing maps (SOMs) are one of the most frequently used partitional clustering methods. SOM analysis resulted in the identification of 12 tight gene clusters. Examination of the genes belonging to each cluster leads one to several conclusions. 824(pGROE1) displays an increased expression of motility and chemotaxis genes (mostly in cluster G11). A decrease was measured in the expression of the other major stress response genes (including hrcA, dnaK, dnaJ, hsp90, clpC, and ctsR, hsp18, mostly in cluster G3), but contrary to expectation, such a decrease did not affect the cell's ability to tolerate higher levels of butanol. Decreased expression of the dnaKJ operon upon overexpression of groESL suggests that groESL functions as a modulator of the CIRCE regulon, which is shown here to include the hsp90 gene. Key solvent formation genes also show decreased expression during the first 45 hours of the 824(pGROE1) culture, including ctfB, adc, and bdhA. Several glycolytic genes also have lower expression in 824(pGROE1), including triosephosphate isomerase, enolase, and pfk.

Transcriptional Analysis of Host-Plasmid Interactions Confirms the Hypothesis That Plasmids Elicit a Stress Response Affecting the Expression of Product-Formation Genes

Comparison of the plasmid control strain, 824(pSOS95del), to the wild type strain reveals that the host-plasmid interaction is quite significant at the transcriptional level: many of the stress protein genes (including groES/EL, dnaJ/K, grpE, hsp18, clpA/C, and hsp90) are upregulated during exponential growth (followed by downregulation in the stationary phase), suggesting that the cells respond to the presence of a plasmid as they would to many other stresses. Several sporulation genes (including the sigF operon) also are upregulated. Nineteen genes related to DNA topology, replication, modification, recombination, and competence (DNA gyrases, DNA polymerases III, recombinates, and DNA glycosylases) are downregulated in 824(pSOS95del) compared to the wild type strain. The genes in the glycolytic pathway also are expressed at a lower level in 824(pSOS95del), and this confirms the prior suggestions that DNA topology plays a major role in the regulation of carbon metabolism. Among other downregulated genes in 824(pSOS95del) are those related to motility and chemotaxis, representing all four predicted motility and hemotaxis operons and two glycosyltransferases. These patterns of gene expression agree with previous microscopic observations that plasmid control strains display increased sporulation and decreased motility.

Western Analysis of GroEL, DnaK, AADC, and CoAT Expression Suggests That Elevated GroESL Levels Stabilize Product-Formation Enzymes

Consistent with the transcriptional analysis, GroEL levels were elevated, and DnaK levels were drastically reduced in 824(pGROE1) compared to 824(pSOS95del). In the late stationary phase, DnaK levels also were lower in 824(pSOS95del) compared to the wild type, and this again confirms that DnaK downregulation does not affect the cell's ability to produce higher butanol levels. Acetoacetate decarboxylase (AADC, the product of adc), which catalyzes the last step in acetone formation (and to a lesser extent, the CoAT units CtfA/B, which catalyze the first step of the same reaction) were much higher in the GroESL-overexpressing and plasmid control strains, and this is consistent with the observed fluxes. This demonstrates that elevated levels of stress proteins stabilize key product-formation enzymes in addition to the overall cellular structures.

Butanol Challenges of the Wild Type, Plasmid Control, and spo0a Overexpressing Strain Demonstrate the Importance of Differentiation and Stress Response in Solvent Tolerance

Wild type, plasmid control (both 824[pSOS95del] and 824[pIMP1]), 824(pGROE1), and 824(pMSPOA) (this strain overexpresses the spo0A gene, the regulator of stationary-phase phenomena such as sporulation and solventogenesis, and shows accelerated and enhanced sporulation) were subjected to butanol challenges at several levels. Butanol additions alter growth, product formation, and glucose uptake, depending on the challenge level, with low levels of butanol having a stimulatory effect on glucose utilization and butanol formation. In addition to the tolerant strain 824(pGROE1) discussed above, plasmid control strains also exhibit enhanced tolerance compared to wild type.

Furthermore, the 824(pMSPOA) strain exhibits enhanced tolerance compared to the plasmid-control strain 824(pIMP1), thus demonstrating that higher levels of Spo0A allow the cultures to tolerate butanol.

Transcriptional Analysis of the Butanol Stress Response Shows Large Effects on Key Cellular Programs

In wild type, among the genes that show the earliest upregulation (10 minutes) are the HSPs dnaK, dnaJ, groEL, and hsp18 and the solvent formation genes aad, ctfA/B, and bdhB. Additional stress genes (hsp90 and groES) are upregulated at 60 minutes post butanol. The ntrC gene, responsible for regulation of polyphosphate accumulation under stress, also is upregulated at 60 minutes. Additional upregulated genes include several ATPases, ATP enzymes, phosphatases, pyrophosphatases, and phosphorylases, many of which partake in energy metabolism and stress response. Notable among genes that were downregulated was the sporulation-specific sigma factor F, a lon protease, and two primary metabolism genes. Transcriptional analysis was performed with butanol-stressed cultures of 824(pMSPOA). We identified approximately 100 genes differentially expressed after 10, 30, and 60 minutes of the initial stress.

Upregulated genes include the stress proteins (groEL, dnaJ, hsp18, hsp90), sporulation (spoVAE, spoVS, spoVB), sulfate utilization genes, and the butanol-formation pathway gene ctfB. Genes downregulated include several involved in fatty acid synthesis (fabG, accC, fabZ) and motility and chemotaxis (fli operon, cheC, cheA). It appears that membrane transporters are not prominent in the short-term stress response. Interestingly, several genes related to glycolysis and primary metabolism (glucose-6-P isomerase, 6-P-fructokinase, fructose-bisphosphate aldolase, and triose-P isomerase) were downregulated within 10 minutes of stress, followed by overexpression after 60 minutes. Many of the genes involved in stress response, sporulation, chemotaxis, motility, and metabolism were upregulated in several sets of transcriptional analyses (824 vs. 824[pSOS95del]; 824[pSOS95del] vs. 824[pGROE1]; butanol challenged 824[pMSPOA]), providing a consistent picture of how cells combat various stresses/changes.

dnaKJ Overexpression Results in Severely Impaired Growth

As proposed, we also developed a plasmid based on pSOS95del to overexpress dnaKJ. In both E. coli and C. acetobutylicum, cells grew only at lower temperatures and to very low densities, thus suggesting that, as found in other organisms, large overexpression of DnaKJ is toxic to the cells, possibly because of an expression imbalance relative to other stress proteins.

Stress From Small Exposure to Oxygen Results in Increased and Prolonged Solvent Formation

We found that by exposing shake-flask cultures to a few seconds of atmospheric oxygen (an oxidative stress), up to fivefold higher levels of total solvents were produced. In addition, metabolism was prolonged for about 5-7 days, and biomass levels were higher in oxygen-induced cultures than in strict anaerobic cultures. Optimization of oxygen dosing and fermentation conditions may lead to higher solvent titers and further extension of productive fermentation time.

Journal Articles on this Report : 4 Displayed | Download in RIS Format

Other project views: All 9 publications 7 publications in selected types All 4 journal articles
Type Citation Project Document Sources
Journal Article Alsaker KV, Spitzer TR, Papoutsakis ET. Transcriptional analysis of spo0A overexpression in Clostridium acetobutylicum and its effect on the cell's response to butanol stress. Journal of Bacteriology 2004;186(7):1959-1971 R828562 (Final)
not available
Journal Article Tomas CA, Welker NE, Papoutsakis ET. Overexpression of groESL in Clostridium acetobutylicum results in increased solvent production and tolerance, prolonged metabolism, and large changes in the cell's transcriptional program. Applied and Environmental Microbiology 2003;69(8):4951-4965. R828562 (Final)
not available
Journal Article Tomas CA, Beamish J, Papoutsakis ET. Transcriptional analysis of butanol stress and tolerance in Clostridium acetobutylicum. Journal of Bacteriology 2004;186(7):2006-2018 R828562 (Final)
not available
Journal Article Tummala SB, Junne SG, Paredes CJ, Papoutsakis ET. Transcriptional analysis of product-concentration driven changes in cellular programs of recombinant Clostridium acetobutylicum strains. Biotechnology and Bioengineering 2003;84(7):842-854. R828562 (Final)
not available
Supplemental Keywords:

chemicals, solvents, bacteria, green chemistry, renewable, innovative technology, bioremediation, biology, genetics, engineering, measurement methods, Escherichia coli, E. coli, acetone, anaerobic clostridia, bacterial degradation, bioaccumulation, bioavailability, biodegradation, bioprocessing, bioremediation model, biotechnology, cellular properties, gene expression patterns, metabolic engineering, metabolic pathways, solvent-producing enzymes, solvent tolerance., RFA, Scientific Discipline, Waste, Sustainable Industry/Business, Genetics, Environmental Chemistry, Sustainable Environment, Health Risk Assessment, Technology for Sustainable Environment, Biochemistry, Bioremediation, bioprocessing, gene expression patterns, bacteria, bioremediation model, bioavailability, biodegradation, E. Coli, anaerobic clostridia, solvent tolerance, biotechnology, acetone, metabolic engineering, cellular properties, solvent producing enzymes, metabolic pathways, bioacummulation, toxic organic molecules

Progress and Final Reports:
Original Abstract
2000 Progress Report
2001 Progress Report