|
|
Appl Environ Microbiol. 2007 July; 73(14): 4639–4647. Published online 2007 May 18. doi: 10.1128/AEM.00527-07. | PMCID: PMC1932837 |
Copyright © 2007, American Society for Microbiology Metabolic Characterization of Escherichia coli Strains Adapted to Growth on Lactate ![[down-pointing small open triangle]](/corehtml/pmc/pmcents/x25BF.gif) Qiang Hua,1 Andrew R. Joyce,2 Bernhard Ø. Palsson,1* and Stephen S. Fong1† Department of Bioengineering, University of California, San Diego, La Jolla, California,1 Bioinformatics Program, University of San Diego, La Jolla, California2 Received March 7, 2007; Accepted May 15, 2007. In comparison with intensive studies of genetic mechanisms related to biological evolutionary systems, much less analysis has been conducted on metabolic network responses to adaptive evolution that are directly associated with evolved metabolic phenotypes. Metabolic mechanisms involved in laboratory evolution of Escherichia coli on gluconeogenic carbon sources, such as lactate, were studied based on intracellular flux states determined from 13C tracer experiments and 13C-constrained flux analysis. At the end point of laboratory evolution, strains exhibited a more than doubling of the average growth rate and a 50% increase in the average biomass yield. Despite different evolutionary trajectories among parallel evolved populations, most improvements were obtained within the first 250 generations of evolution and were generally characterized by a significant increase in pathway capacity. Partitioning between gluconeogenic and pyruvate catabolic flux at the pyruvate node remained almost unchanged, while flux distributions around the key metabolites phosphoenolpyruvate, oxaloacetate, and acetyl-coenzyme A were relatively flexible over the course of evolution on lactate to meet energetic and anabolic demands during rapid growth on this gluconeogenic carbon substrate. There were no clear qualitative correlations between most transcriptional expression and metabolic flux changes, suggesting complex regulatory mechanisms at multiple levels of genetics and molecular biology. Moreover, higher fitness gains for cell growth on both evolutionary and alternative carbon sources were found for strains that adaptively evolved on gluconeogenic carbon sources compared to those that evolved on glucose. These results provide a novel systematic view of the mechanisms underlying microbial adaptation to growth on a gluconeogenic substrate. Biological systems are capable of adapting to environmental changes by invoking a number of different strategies to achieve optimal overall performance under a specific condition. Laboratory evolution methods have been used to understand adaptive changes in a wide variety of microorganisms ( 3, 7, 8, 11, 24). Whereas some evolved phenotypes, such as the growth rate or product secretion rate, are readily detectable, the underlying mechanisms that result in improved cellular properties during the evolutionary process are difficult to identify. A number of technologies are now available to help elucidate the underlying mechanistic changes by providing genome-wide measurements of adaptive changes in gene expression or genome sequence ( 7, 11, 18, 19). While the molecular basis of evolution might be revealed by these technologies, they cannot directly account for the specific metabolic changes responsible for the improved phenotypes. More direct information on metabolic network responses to laboratory evolution is therefore necessary to understand changes in metabolic functions and to link possible genetic mechanisms of evolution to evolved metabolic phenotypes. The characteristics of metabolic networks can be directly assessed by analyzing metabolic flux distributions in cells obtained over the time course of laboratory evolution. Recently, 13C isotopomer-based flux analyses have been successfully applied to investigate metabolic responses and underlying mechanisms in various prokaryotic and eukaryotic systems ( 4, 9, 12, 26, 29, 33). We have previously studied the metabolic flux states for several gene deletion mutants of Escherichia coli that evolved on glucose ( 16, 20). These flux analyses indicated that these mutant strains adapted to the evolutionary process by activating either latent pathways or pathway of which the major enzyme was deleted. Further studies of gene expression and enzyme activity suggested good correlations between increased pathway fluxes and these measurements. Metabolic flux analysis can therefore provide valuable insight into dynamic responses of the evolutionary system by associating transcriptional or translational changes with evolved metabolic phenotypes. A large number of laboratory studies of evolution on glucose have been carried out ( 7, 11, 23, 27, 31), since glucose is the preferred carbon and energy source for most bacteria and eukaryotic cells ( 2, 28). In comparison to adaptive evolution on glucose, microorganisms might be more capable of adapting to various other carbon compounds in the natural environment. Few studies, however, have investigated the evolutionary changes of metabolic networks with carbon sources other than glucose. In recent years, we have used laboratory evolution to study the adaptation of the E. coli K-12 strain and derivative metabolic-gene deletion strains with several alternative carbon sources in addition to glucose. Physiological studies suggest that the effects of most genetic modification, such as growth rates at the end point of evolution, were consistent with that predicted by genome-scale metabolic models using flux balance analysis ( 17). Furthermore, while replicate evolution experiments showed convergence and reproducible growth of E. coli on lactate and glycerol, detailed transcriptional studies revealed different gene expression states among evolutionary end points, with only a few genes showing beneficial expression in common across parallel evolution populations ( 14). In an effort to further elucidate the metabolic mechanisms involved in the adaptive evolution of bacteria on less-studied gluconeogenic carbon sources, we sought to evaluate metabolic flux states of E. coli strains that adaptively evolved in parallel on lactate by using 13C tracer experiments and 13C-constrained flux analysis. Metabolic analyses were conducted at four different time points during laboratory evolution for each of seven populations independently generated using serial passage of exponentially growing cultures in lactate-supplemented minimal medium ( 14). Adaptive characteristics and mechanisms on lactate were therefore investigated by determining cellular phenotypes, metabolic flux distribution patterns, growth on alternative carbon substrates, and previously generated global gene expression data. Strains and culture conditions. The strains used in this study were the wild-type E. coli K-12 strain MG1655 and seven populations (LacA, LacB, LacC, LacD, and LacE [ 14] and Lac2 and Lac3 [ 15]) that adaptively evolved on M9 minimal medium supplemented with l-lactate for up to 60 days in our laboratory. For each evolved population, strains obtained at different days of evolution (day 10 [~100 generations of growth], day 20 [~230 generations], day 40 [~540 generations], and day 60 [~900 generations]) were selected to investigate the effects of adaptive evolution on cellular metabolism. Cultures were grown in 30 ml of M9 minimal medium with 2 g/liter of l-lactate as the carbon source in 250-ml baffled Erlenmeyer flasks using magnetic stir bars for aeration at 30°C. A mixture of 20% (wt/wt) uniformly 13C-labeled l-lactate (U- 13C, >99%; Cambridge Isotope Laboratories, Andover, MA) and 80% (wt/wt) 3- 13C-labeled l-lactate (>99%; Sigma-Aldrich, St. Louis, MO) was used for all 13C labeling experiments. Each culture was inoculated from a preculture with a starting optical density at 600 nm (OD 600) of less than 0.005. Cell samples were taken at mid-exponential-growth phase (OD 600 of ~0.5) for labeling analysis of intracellular fluxes. Physiological assays. The OD600 was measured to monitor batch cell growth. Culture samples were taken periodically, and substrate uptake rates and metabolite secretion rates were measured throughout the course of exponential growth. The depletion of l-lactate and secretion of acetate, ethanol, and other metabolites in the medium were determined using commercial enzymatic assay kits (R-Biopharm AG, Darmstadt, Germany). Growth on alternative carbon sources was evaluated using the VERSAmax microplate reader system (Molecular Devices Co., Sunnyvale, CA). For each experiment conducted with the VERSAmax system, precultures were grown overnight and allowed to reach mid-exponential-growth phase. The appropriate volume of cultures was used to inoculate the 96-well plates containing 200 μl of medium, yielding an initial OD600 of 0.005. Growth on nine different carbon-supplemented M9 minimal media (acetate, alanine, aspartate, glucose, glutamate, glycerol, α-ketoglutarate, pyruvate, and ribose) was tested. The initial carbon content in each medium was approximately 65 to 70 mmol/liter, similar to that used for lactate evolutionary experiments. The plates were incubated at 30°C, and measurements were taken every 15 min with continuous shaking between measurements. Growth rates were averaged across at least three replicate cultures. GC-MS sample preparation and analysis. Cells growing at mid-exponential-growth phase from 3 to 5 ml of culture were harvested by centrifugation. The cell pellet was washed with 1 ml 0.9% (wt/vol) NaCl and then hydrolyzed in 200 μl of 6 M HCl at 105°C for 24 h. The filtrate of hydrolysate was dried in a heating block at 60°C, and proteinogenic amino acids were derivatized at 85°C for 1 h in 75 μl tetrahydrofuran (Sigma) and 75 μl N-methyl- N-[tert-butyldimethylsilyl] trifluoroacetamide (Sigma). After filtration, 3 μl of derivatized sample was injected for the gas chromatography-mass spectrometry (GC-MS) assay. The GC-MS analysis was performed with a Trace GC/Trace MS Plus system (Thermo Electron Corporation, Waltham, MA), and mass spectral data were obtained for fragments of most derivatized amino acids, as previously reported ( 20). Prior to analysis of cellular metabolism, the raw mass isotopomer distribution vector (MDV) (fractional abundance set for amino acid fragment with different mass numbers) was corrected for the natural isotope abundances of carbon atoms introduced from the derivatization reagent and all noncarbon atoms involved in the whole fragment ( 30). Flux ratios and flux distributions in lactate-grown cells. Intracellular flux states of E. coli grown aerobically on lactate were studied based on a metabolic model describing the whole central metabolism. In contrast to glucose-grown E. coli cells, in which the glycolytic pathway is predominantly employed for carbon catabolism, trioses and hexoses required for biosynthesis are generated through gluconeogesis in lactate-grown cells. pps-encoded phosphoenolpyruvate (PEP) synthase (PPS) and fbp-encoded fructose 1,6-bisphosphatase catalyze key steps towards the gluconeogenic synthesis of six-carbon compounds. The model also included metabolic reactions involved in the tricarboxylic acid (TCA) cycle, the pentose phosphate (PP) pathway, carbon anaplerosis, the Entner-Doudoroff (ED) pathway, and pyruvate metabolism as shown elsewhere ( 20). The model system consists of 26 reactions and is thus underdetermined when only material balance constraints are conducted for 21 metabolites in the model network. Analysis of 13C isotopomer distribution of intermediates was therefore employed to provide additional useful constraints for fast flux determination through the quantification of the relative contributions of individual pathways to the formation of target metabolites ( 12, 26, 33). The flux ratio constraints used to facilitate the metabolic analysis of lactate-evolved strains included ratios relating to various metabolic functions, such as gluconeogenesis (PEP from oxaloacetate [OAA] and pyruvate [Pyr] from malate), anaplerosis (OAA from PEP and OAA generated through the glyoxylate bypass), and the degradation of sugar acids (pyruvate generated through the ED pathway). PEP is generally synthesized either from pyruvate via PPS or from OAA via PEP carboxykinase (PCK) in lactate-grown E. coli. Under the lactate-labeling condition in this study, activation of both the TCA cycle and the glyoxylate bypass resulted in increased fraction abundance for OAA 2-3 fragments (the subscript indicates the carbon position) in comparison with the mass distribution of Pyr 2-3 fragment, which was therefore used to determine the contributions of individual reactions to PEP synthesis. The mass distribution of OAA 2-3 was derived from the mass distributions for acetyl-coenzyme A 1-2 (CoA 1-2) (ACA 1-2) and α-ketoglutarate 2-5 (AKG 2-5), and the MDV of phenylalanine 2-9 (Phe 2-9) was employed to estimate the mass distributions of PEP 2-3 and erythrose 4-phosphate 1-4 (E4P 1-4) based on least-squares fitting analysis ( 12). To calculate the fraction of OAA generated via the anaplerotic reaction catalyzed by PEP carboxylase, the equation derived by Fischer and Sauer ( 12) was used. The possible contribution of the glyoxylate bypass to OAA generation was neglected in this expression estimating the ratio of OAA from PEP and the fractional labeling of CO 2 simultaneously. In addition, we also used the mass distribution of OAA 1-2 to quantify both fractions of OAA through anaplerotic PEP carboxylation and OAA derived via the glyoxylate bypass. In this case, the MDV of OAA 1-2 generated from the TCA cycle is the average of the MDVs of OAA 2-3 and ACA 1-2, and the MDV of OAA 1-2 derived through the glyoxylate bypass might be obtained by calculating 0.5 OAA 3-4 + 0.25 (OAA 1-2 + ACA 1-2). Both methods showed similar flux ratios of OAA from PEP for most evolved populations with only a few exceptions that might be attributed to a slight activity of the glyoxylate bypass. The fraction of pyruvate obtained from malate via malic enzyme was determined by comparing the mass distributions of the Pyr 1-2 fragment with those for malate 1-2 (Mal 1-2) and lactate 1-2 (Lac 1-2). The MDV of Pyr 1-2 was calculated from the fraction of PEP from OAA (f PCK), using the equation PEP 1-2 = f PCK × OAA 1-2 + (1 − f PCK) × Pyr 1-2. The lower bound of the ratio was evaluated by assuming that malate was formed exclusively through the TCA cycle, where the MDV of Mal 1-2 was obtained as the average of the MDVs of Lac 2-3 and a two-carbon fragment with the same fractional labeling of Lac 3 for each carbon unit. With the assumption that malate was balanced with oxaloacetate, the MDV of OAA 1-2 was then used to obtain the upper bound of the ratio ( 12). In both cases, the TCA cycle yielded a significantly lowered abundance for the unlabeled Mal 1-2 fragment than for the Pyr 1-2 fragment. By assuming that 6-phosphoglutonate was exclusively generated from the oxidative PP pathway, the fraction of pyruvate formed through the ED pathway was obtained by comparing the MDV of Pyr1-3 with that of PEP1-3. The fractions of pyruvate with the mass of m (m represents the mass number of the unlabeled fragment) and m + 2 should be very low if pyruvate is primarily converted from lactate. PEP might contribute more molecules with the mass m + 2 to pyruvate (with lowered fractions for m + 1 and m + 3 fragments) through consecutive reactions in gluconeogenesis, the oxidative PP pathway, and the ED pathway. Since the nonoxidative PP pathway and the malic enzyme might also result in the cleavage between carbon atoms of pyruvate, the calculated fraction remains an upper bound. Similar to those shown by Fischer et al. ( 13) and Blank et al. ( 1), the above flux ratios were expressed by relevant metabolic reactions, which therefore provided flux calculation with three independent equality constraints and three independent inequality constraints. Quantification of metabolic flux was then performed from a series of randomly generated starting fluxes to get an optimal flux distribution that met the metabolite balance requirements and flux ratio constraints and reproducibly minimized the objective function as described elsewhere ( 13). Transcriptional analysis. The detailed description of transcriptional analysis of gene expression in lactate-grown E. coli was given previously ( 14). Briefly, Affymetrix (Santa Clara, CA) E. coli antisense genome arrays were used for all transcriptional analyses. Each experimental condition was tested at least in triplicate on lactate using biologically independent cultures and processed following the manufacturer's recommended protocols. Total RNA was isolated from exponentially growing cells with RNeasy Mini Kits (QIAGEN, Valencia, CA), using RNAProtect (QIAGEN) to stabilize samples. Total RNA yields were measured using a spectrophotometer ( A260), and quality was checked by visualization on agarose gels and by measuring the sample A260/ A280 ratio. An appropriate amount of isolated total RNA was then used for cDNA synthesis, fragmentation, and terminal labeling, which were all conducted as recommended by Affymetrix. Raw .CEL files were analyzed using robust multiarray average analysis ( 21) for normalization and calculation of probe intensities. Log-transformed expression values were then assessed for statistically significant differential expression relative to values for wild-type cultures using log 2 ratios and t tests. In an effort to characterize the E. coli strains that adaptively evolved on a gluconeogenic carbon source, such as lactate, a variety of detailed experimental analyses were conducted for seven populations that evolved at different generations in M9 medium supplemented with l-lactate. Detailed studies of physiological characteristics, in vivo flux determination, follow-up gene expression analysis, and growth on alternative carbon substrates were performed to elucidate metabolic and molecular mechanisms that accompany the enhanced phenotypic characteristics during laboratory evolution and evolutionary differences among individual populations. Physiological characteristics of lactate-evolved E. coli. Overall, all 28 (7 populations × 4 independent time points) lactate-evolved strains exhibited an improved growth rate, lactate uptake rate, and biomass yield compared with the unevolved wild-type strain. Specific growth rate continued to increase over the course of evolution for all seven evolved populations, resulting in a maximal growth rate of 0.51 ± 0.04 h −1 (more than twofold that of the unevolved strain) at the end point of adaptive evolution (Fig. ). An average 50% increase in biomass yield was also observed for end points of evolution (Fig. ). Whereas both physiological parameters kept increasing over the 60-day (about 900 generations) evolutionary course, approximately 60% phenotypic improvements were achieved within the first 20 days (less than 250 generations) of evolution, indicating rapid fitness gains in the initial stage of evolution ( 10). Generally, the relative change in fitness, calculated from the changes in the growth rate or biomass yield between adjacent days studied, decelerated sharply over the course of laboratory evolution. Only minor improvements were observed during the last 20 days (the last 350 generations) of evolution. However, large differences in fitness gains for both physiological measurements were observed, especially in the early stage of evolution, among seven evolved populations (Fig. and ). The major carbon overflow in E. coli growing aerobically on lactate was acetate, which was secreted at about 0.4 mmol/mmol lactate by the unevolved strains. The acetate secretion rate, however, declined in most evolved populations, and an up to 50% decrease was observed for some strains. Ethanol, another metabolite derived from intracellular acetyl-CoA, was also secreted but at very low concentration levels by wild-type E. coli. Interestingly, ethanol secretion gradually increased, but no more than 0.1 mmol/mmol lactate was observed over the course of evolution (data not shown). Moreover, regardless of the evolution, neither TCA-cycle metabolites nor glycerol was detected in lactate-grown E. coli strains. Metabolic flux states over the course of evolution on lactate. In an effort to understand changes in metabolic mechanisms over the course of laboratory evolution of E. coli on lactate, intracellular flux states were quantified from 13C-labeling experiments for all selected evolved populations and unevolved wild-type strains. Flux estimation indicated that the errors for most pathway fluxes were less than 10% for the 95% confidence interval based on the redundant mass distribution data. Fast adaptation of strains on lactate was characterized by up to 80% increases in lactate uptake fluxes within the first 20 days of evolution, whereas only minor changes in lactate utilization were observed during the late stage of adaptive evolution (lldD boxes; Fig. ). Consequently, metabolic capacities of most reactions were enhanced, as shown by hierarchical clustering of fluxes relative to those of the wild-type strain (Fig. ). The flux of reactions catalyzed by malic enzyme, flux through the glyoxylate bypass, and flux through the ED pathway were excluded from clustering because of extremely low fluxes for the wild-type E. coli and insignificant flux changes in evolved populations. Neither the reaction through the malic enzyme nor the reaction through the ED pathway was active in all cases, which is probably attributable to a high intracellular pyruvate pool size in cells growing on lactate. Whereas fitness gains for growth rates and biomass yields varied largely among evolved populations obtained at day 10 (100 generations) of evolution, relative flux distributions of most day-10 strains fell into the same cluster, suggesting similar overall metabolic responses to selection pressure of a high growth rate at the very early stage of lactate evolution. In contrast, relatively wide spectra of relative flux distributions were observed for strains that evolved for more than 20 days. During the early phase of evolution, two growth patterns were observed for E. coli adaptively growing on lactate. Five of seven parallel evolved populations (LacA, LacB, LacC, LacD, and LacE) exhibited maximal gains for biomass yield between day 10 and day 20 of evolution, which is probably attributable to the decreases in acetate overflow and increases in the TCA cycle flux that therefore provided cells with more precursors and energy for biosynthesis of cellular constituents (Fig. ). Among the five strains mentioned above, LacC and LacD had the most significant repression of acetate flow (approximately a 50% drop from day 10 to day 20), resulting in higher fitness gains for biomass yield (30 to 50% increases) and consequently higher evolutionary gains for the growth rate (40 to 80% increases) (Fig. ). Aside from these strains, maximal gains for biomass yields and growth rates of Lac2 and Lac3 populations were observed even earlier (before day 10 of evolution), with similar patterns of flux rerouting (Fig. ). Interestingly, biomass yield of both evolved populations decreased during the next 10 days without significant changes in flux distribution (Fig. and ). Despite diverse evolutionary paths in the early phase, the evolutionary end points were convergent for all evolved populations, with much lower gains for both the cell yield and the growth rate during the late evolutionary stage, where a general increase was found for the relative overflow fluxes (Fig. ). Transcriptional and metabolic responses for genes related to lactate metabolism. To provide more-comprehensive views of evolutionary growth on lactate, both metabolic pathway and transcriptional responses in lactate metabolism- and gluconeogenesis-related genes were dissected based on the flux states and gene expression profiling of the wild-type strain and seven evolved populations obtained at day 20 and day 60 of evolution ( 14). (i) Genes for lactate utilization. General increases in transcriptional expression of the lldD gene (encoding l-lactate dehydrogenase) were observed for most evolved populations (with the exception of LacC), which was consistent with up to 80% increases in the specific lactate uptake rate for lactate-evolved strains (Fig. ). The l-lactate permease-encoding gene lldP, which appears in the same operon, lldPRD, with the lldD gene, had similar patterns of transcriptional changes, whereas it is not clear why the expression of another operon gene, lldR (encoding operon regulatory protein), did not change significantly. Therefore, lactate uptake flux changes were likely controlled at the transcriptional level over the course of laboratory evolution on lactate. (ii) Genes associated with pyruvate metabolism. Transcriptional analysis suggested that expressions of pyruvate kinase-encoding genes were largely repressed in lactate-grown wild-type E. coli. Transcriptions of these genes were further repressed during evolution on lactate (data not shown). On the contrary, expressions of the gluconeogenic pps gene, which encodes PPS, were significantly up-regulated (up to 1.5-fold increase) over the course of evolution (Fig. ). The pps gene was also the only gene of which the expression was induced in all evolved populations, as concluded previously ( 14). Interestingly, we detected only small increases in absolute flux through PPS for most evolved populations (Fig. ). In addition, the flux of pyruvate dehydrogenation (the initial metabolic reaction of oxidative degradation of pyruvate) was generally increased, albeit moderately, in most evolved populations, whereas expression of genes involved in the pdhR-aceEF-lpdA operon was not necessarily elevated during evolution (Fig. ). The lack of correlation between gene expression and metabolic flux associated with pyruvate metabolism might be related to the energetic demand for cell growth on the nonpreferred carbon substrate of lactate, where pyruvate catabolism through the TCA cycle in evolved strains might be stimulated by low energy charge values ( 6). (iii) Anaplerotic and gluconeogenic genes. The PEP-pyruvate-oxaloacetate node acts as an important switch point for carbon flux distribution by controlling the carbon flux into gluconeogenesis, the TCA cycle, and the anaplerotic pathways ( 25). With the exception of the pps gene previously discussed, transcription of two other groups of gluconeogenic genes, PCK-encoding pckA and malic enzyme-encoding sfcA and maeB, were generally decreased. Metabolic flux through malic enzyme was, irrespective of evolution, extremely low. The flux of the reaction catalyzing the conversion of oxaloacetate to phosphoenolpyruvate was, however, increased markedly in most cases (Fig. ). Similarly, despite general down-regulation of ppc gene expression, anaplerotic PEP carboxylation flux was accelerated up to twofold for seven parallel evolved strains (Fig. ). While the mechanism for the inconsistency between transcriptional and metabolic responses for both genes is unclear, the adjustment of energetic flow through phosphorylation of oxaloacetate by PCK, at least in part, accounted for the flux increases in OAA decarboxylation and PEP carboxylation over the course of lactate evolution. Actually, activities of both enzymes might be largely regulated by several intracellular effectors, as analyzed for glucose-grown conditions ( 32) and summarized by Sauer and Eikmanns ( 25). Moreover, while analysis of the ppc gene and the pckA gene showed similar evolutionary patterns (i.e., decrease in transcription and increase in metabolic flux), the ratio of the PEP carboxylation flux to the OAA decarboxylation flux varied from 1.3 to 3.0 in the evolved populations (2.2 in the wild-type strain). The flexibility of the PEP-oxaloacetate node during evolution might be important for energetic and anabolic purposes under the selective pressure of rapid growth on lactate. (iv) Transporter genes of carbon substrates. As previously stated, the transcription of both the lactate/proton symporter-encoding gene lldP and the lactate dehydrogenase-encoding gene lldD were generally enhanced during evolution on lactate, which was then corresponding well to markedly improved lactate uptake rates of the evolved populations. However, a general decrease in transcription of genes relating to the utilization of many other carbon substrates was observed. Expression levels of the pstI and pstH genes (encoding the sugar-nonspecific proteins enzyme I and Hpr in the phosphotransferase system) were about one-fourth lower than that for the unevolved strain. A similar down-regulation in expression was observed for most sugar-specific phosphotransferase system enzyme II genes (data not shown). For genes that encode ABC transporters, except for the slight up-regulation of some amino acid ABC transporter genes (e.g., glutamate and glutamine), a general down-regulation of transcription was found for genes of most sugar (e.g., ribose, galactose, and xylose), peptide, and other amino acid (e.g., leucine and isoleucine) transporters. The expression of genes encoding ion channels and most secondary transporters did not change significantly over the course of evolution, whereas down-regulation was observed for the glycerol channel-encoding gene glpF and the dctA gene, which encodes a dicarboxylate/cation symporter (data not shown). In the wild-type strains growing on lactate instead of glucose, the expression levels of a large number of genes for utilizing unavailable carbon compounds were also up-regulated, as also was observed for cells growing on other poor-quality carbon sources ( 22). Our results furthermore indicated that during adaptive growth on the nonpreferred carbon source, cells are likely to settle for optimizing their growth on the available substrate by shutting off any potential energy drains, including those involved in the expression of alternative transporters. Growth of lactate-evolved strains on alternative carbon substrates. To assess the growth characteristics of lactate-evolved E. coli strains on nonevolutionary carbon substrates, growth rates were measured for the wild-type and the selected evolved strains on nine alternative carbon substrates, including several amino acids. Overall, on each carbon substrate tested, nearly all evolved populations showed continuous improvements in the growth rate over the course of evolution (Fig. ). Although some growth gains were lower than those in our previously study ( 14) because of different microculture systems used, we still observed an overall growth increase of 65% for end points of evolution. Notably, similar to the growth improvement on lactate, specific growth rates of lactate-evolved strains increased more than onefold on pyruvate. Interestingly, similar to the evolutionary growth on lactate, an average more than 70% growth increase was obtained with strains that evolved for less than 20 days (Fig. ). In particular, despite variable growth adaptation of individual populations with different carbon substrates, the average growth gains were approximately 18% between the wild-type and evolved strains at day 10, 26% between evolved strains at day 10 and day 20, and only 2% to 5% between strains evolved more than 20 days. Moreover, the elevated growth rates, together with probably decreased substrate uptake rates as indicated by the down-regulation of relevant transporter genes, might result in significantly improved biomass yields of all evolved populations on alternative carbon substrates. The metabolic network of E. coli adaptively evolved on lactate was studied based on 13C tracer experiments and 13C-constrained flux analysis. Although large metabolic flux variations were observed for seven parallel evolved populations at four selected days of evolution, there was no new metabolic pathway response found regarding evolutionary effects on lactate. The main feature of evolution was the overall increased fluxes for already-active central pathways, an effect also revealed from the evolution of several knockout mutants ( 16, 20). In addition, flux states varied significantly among seven populations that evolved over the observed time frame, especially among strains generated during the early stage of evolution. Examples are the up to 1.6-fold variation in absolute lactate uptake flux and more than 3-fold flux variation in a reaction catalyzed by gluconeogenic PCK. While the overall metabolic flux states varied markedly among strains that evolved in parallel, the ratio of pyruvate dehydrogenation flux to flux of lactate dehydrogenation remained almost unchanged, especially during the early evolutionary phase (Fig. ). Approximately two-thirds of the carbon source obtained from lactate oxidation was catabolized through the pyruvate dehydrogenase complex for energy requirements. The almost invariable partitioning between gluconeogenic and catabolic fluxes during growth on lactate might be well regulated by the PEP pool size and cellular energy charge ( 5) and suggests the critical role of the pyruvate node in maintaining a balanced supply of biosynthetic precursors and energy for E. coli cells that adaptively evolved on lactate. Early studies on the partitioning of pyruvate in lactate-grown E. coli reported a wide range for the ratio of pyruvate dehydrogenase activity to that of PPS (from as high as 4 in the early stage of culture to about 0.5 in late logarithmic phase) ( 5). Our results indicate that during mid-logarithmic growth, the ratio of pyruvate degradation through the dehydrogenase complex to pyruvate used for gluconeogenesis and anaplerosis was kept at 2.4 ± 0.6 in both wild-type and lactate-evolved E. coli cells. The high ratio was likely attributable to a relatively low energy charge in the cells. Actually, based on the fluxes calculated in this study, we observed a more than 1.7-fold energy requirement for all cell processes, including rapid growth on lactate compared to the exponential growth on glucose, indicating possible energy limitation in cells grown on a gluconeogenic carbon source, such as lactate. In contrast to invariable flux distributions around the pyruvate node, large variations were found for fluxes around the PEP, acetyl-CoA, and oxaloacetate nodes in strains that evolved in parallel. Generally, the ratio of anaplerotic PEP carboxylation flux to PEP synthetic fluxes through gluconeogenic PPS and PCK increased to replenish TCA cycle intermediates that were withdrawn for anabolism. Whereas an average increase of 30% was shown for this ratio, the ratio varied from 0.3 to 0.6 in all evolved populations (Fig. ). Total overflow (acetate and ethanol) fluxes, however, decreased from about 60% to 45% of pyruvate dehydrogenation flux, which therefore allowed more acetyl-CoA oxidation via the TCA cycle for energy generation. The relative oxaloacetate decarboxylation flux to all oxaloacetate synthesized was also increased by more than 30% on average but with large variation among independently evolved populations (Fig. ). The observed changes in the metabolic network suggested that the bacterial cells are able to adapt to a new environment by rapidly examining metabolic functions of central metabolic nodes and determining the extent of variations in flux distribution pattern around these nodes. The relative fluxes around metabolites that are critical in both carbon and energy metabolism might not change significantly (e.g., fluxes around pyruvate in lactate metabolism), while the cells could invoke various metabolic strategies to adjust flux distributions on other flexible metabolic nodes to achieve high fitness gains. Transcriptional and flux studies of lactate-evolved E. coli strains revealed that with a few exceptions, qualitative correlations were not found between the overall gene expression and metabolic flux changes, as is partly shown in Fig. . There were large variations in transcriptional changes even among strains that had evolved for the same length of time. Moreover, gene expressions of independently evolved populations were not convergent over the course of lactate evolution, as indicated by expression differences among strains evolved at the end point of evolution. Random spontaneous mutations in the E. coli genome ( 18, 19) and complex gene expression responses to highly dynamic evolutionary processes probably account for significant transcriptional variations among E. coli strains that evolved in parallel. While flux variations among evolved populations were also found for most pathway reactions, laboratory evolution resulted in overall increases in pathway fluxes that correlated tightly with enhanced lactate utilization and an improved growth rate. While some genetic modifications might provoke modified flux distributions during the course of adaptive evolution, detailed regulation remains unknown. It is clear that multiple layers of information, such as accurate metabolite analysis, are useful in explaining the incoherencies between gene expression and pathway flux and in finding possible changes in regulation phenomena. Such an enhanced global approach would significantly improve the chances of identifying the key causal mechanisms underlying cellular adaptive responses. Laboratory evolution of E. coli in lactate-supplemented medium revealed significant increases in growth on lactate for all populations that evolved in parallel. Similarly, large fitness gains for cell growth were also found for the adaptive evolution of E. coli on other gluconeogenic carbon sources. Examples are a 150% increase while evolving on glycerol ( 14), an above-110% increase on pyruvate, and an approximately 50% increase on α-ketoglutarate ( 15). The growth of glucose-evolved strains was, in contrast, enhanced by only 20% relative to that of wild-type cells grown on glucose. Significant metabolic adjustment, including enhanced capacity for substrate utilization and optimized intracellular flux distributions, might account for the large increase in evolutionary growth on gluconeogenic carbon sources, as shown with lactate-evolved strains. In contrast, flux analysis of glucose-evolved strains did not show much change in relative flux states during evolution on glucose (unpublished data). In addition, for the evolution on gluconeogenic carbon compounds, most changes in flux distribution and transcriptional expression ( 14) were shown in the early stage of evolution, resulting in rapid cell adaptation to less-favorable nutrient conditions. The optimal cell metabolic systems obtained through laboratory evolution on lactate were further evidenced by 50 to 100% increases in growth on a wide variety of nonevolutionary gluconeogenic carbon compounds. Interestingly, the growth of lactate-evolved strains on nongluconeogenic substrates was also largely improved (e.g., 34% ± 11% and 63% ± 15% increases on glucose and ribose, respectively), suggesting overall improvement of metabolic network characteristics during adaptive growth on lactate. On the contrary, only small growth improvements on alternative carbon sources (e.g., about a 17% increase on lactate and a 47% increase on acetate; unpublished data) were shown for glucose-evolved strains, for which the metabolism did not change much relative to that of the unevolved strain. Therefore, more-comprehensive information and the mechanisms underlying the microbial evolution process can probably be learned by the systematic study of adaptive evolution under conditions with less-optimal metabolic or genetic networks and in appropriate evolutionary stages with most significant fitness gains. We thank Marc Abrams for useful discussions regarding the manuscript. This work was supported by NIH grant GM057089. The principal investigator and UCSD have a financial interest in Genomatica, Inc.; although this grant has been identified for conflict-of-interest management based on the overall scope of the project and its potential to benefit Genomatica, Inc., the research findings included in this publication may not necessarily directly relate to the interests of Genomatica, Inc. 1. Blank, L. M., L. Kuepfer, and U. Sauer. 2005. Large-scale 13C-flux analysis reveals mechanistic principles of metabolic network robustness to null mutations in yeast. Genome Biol. 6:R49. [PubMed] 2. Bruckner, R., and F. Titgemeyer. 2002. Carbon catabolite repression in bacteria: choice of the carbon source and autoregulatory limitation of sugar utilization. FEMS Microbiol. Lett. 209:141-148. [PubMed] 3. Buckling, A., M. A. Wills, and N. Colegrave. 2003. Adaptation limits diversification of experimental bacterial populations. Science 302:2107-2109. [PubMed] 4. Christensen, B., A. K. Gombert, and J. Nielsen. 2002. Analysis of flux estimates based on 13C-labelling experiments. Eur. J. Biochem. 269:2795-2800. [PubMed] 5. Chulavatnatol, M., and D. E. Atkinson. 1973. Kinetic competition in vitro between phosphoenolpyruvate synthetase and the pyruvate dehydrogenase complex from Escherichia coli. J. Biol. Chem. 248:2716-2721. [PubMed] 6. Chulavatnatol, M., and D. E. Atkinson. 1973. Phosphoenolpyruvate synthetase from Escherichia coli. Effects of adenylate energy charge and modifier concentrations. J. Biol. Chem. 248:2712-2715. [PubMed] 7. Cooper, T. F., D. E. Rozen, and R. E. Lenski. 2003. Parallel changes in gene expression after 20,000 generations of evolution in Escherichia coli. Proc. Natl. Acad. Sci. USA 100:1072-1077. [PubMed] 8. Cowen, L. E., D. Sanglard, D. Calabrese, C. Sirjusingh, J. B. Anderson, and L. M. Kohn. 2000. Evolution of drug resistance in experimental populations of Candida albicans. J. Bacteriol. 182:1515-1522. [PubMed] 9. de Graaf, A. A., M. Mahle, M. Mollney, W. Wiechert, P. Stahmann, and H. Sahm. 2000. Determination of full 13C isotopomer distributions for metabolic flux analysis using heteronuclear spin echo difference NMR spectroscopy. J. Biotechnol. 77:25-35. [PubMed] 10. Elena, S. F., and R. E. Lenski. 2003. Evolution experiments with microorganisms: the dynamics and genetic bases of adaptation. Nat. Rev. Genet. 4:457-469. [PubMed] 11. Ferea, T. L., D. Botstein, P. O. Brown, and R. F. Rosenzweig. 1999. Systematic changes in gene expression patterns following adaptive evolution in yeast. Proc. Natl. Acad. Sci. USA 96:9721-9726. [PubMed] 12. Fischer, E., and U. Sauer. 2003. Metabolic flux profiling of Escherichia coli mutants in central carbon metabolism using GC-MS. Eur. J. Biochem. 270:880-891. 13. Fischer, E., N. Zamboni, and U. Sauer. 2004. High-throughput metabolic flux analysis based on gas chromatography-mass spectrometry derived 13C constraints. Anal. Biochem. 325:308-316. [PubMed] 14. Fong, S. S., A. R. Joyce, and B. O. Palsson. 2005. Parallel adaptive evolution cultures of Escherichia coli lead to convergent growth phenotypes with different gene expression states. Genome Res. 15:1365-1372. [PubMed] 15. Fong, S. S., J. Y. Marciniak, and B. O. Palsson. 2003. Description and interpretation of adaptive evolution of Escherichia coli K-12 MG1655 by using a genome-scale in silico metabolic model. J. Bacteriol. 185:6400-6408. [PubMed] 16. Fong, S. S., A. Nanchen, B. O. Palsson, and U. Sauer. 2006. Latent pathway activation and increased pathway capacity enable Escherichia coli adaptation to loss of key metabolic enzymes. J. Biol. Chem. 281:8024-8033. [PubMed] 17. Fong, S. S., and B. O. Palsson. 2004. Metabolic gene-deletion strains of Escherichia coli evolve to computationally predicted growth phenotypes. Nat. Genet. 36:1056-1058. [PubMed] 18. Herring, C. D., A. Raghunathan, C. Honisch, T. Patel, M. K. Applebee, A. R. Joyce, T. J. Albert, F. R. Blattner, D. van den Boom, C. R. Cantor, and B. O. Palsson. 2006. Comparative genome sequencing of Escherichia coli allows observation of bacterial evolution on a laboratory timescale. Nat. Genet. 38:1406-1412. [PubMed] 19. Honisch, C., A. Raghunathan, C. R. Cantor, B. O. Palsson, and D. van den Boom. 2004. High-throughput mutation detection underlying adaptive evolution of Escherichia coli-K12. Genome Res. 14:2495-2502. [PubMed] 20. Hua, Q., A. R. Joyce, S. S. Fong, and B. O. Palsson. 2006. Metabolic analysis of adaptive evolution for in silico-designed lactate-producing strains. Biotechnol. Bioeng. 95:992-1002. [PubMed] 21. Irizarry, R. A., B. Hobbs, F. Collin, Y. D. Beazer-Barclay, K. J. Antonellis, U. Scherf, and T. P. Speed. 2003. Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics 4:249-264. [PubMed] 22. Liu, M., T. Durfee, J. E. Cabrera, K. Zhao, D. J. Jin, and F. R. Blattner. 2005. Global transcriptional programs reveal a carbon source foraging strategy by Escherichia coli. J. Biol. Chem. 280:15921-15927. [PubMed] 23. Notley-McRobb, L., and T. Ferenci. 1999. Adaptive mgl-regulatory mutations and genetic diversity evolving in glucose-limited Escherichia coli populations. Environ. Microbiol. 1:33-43. [PubMed] 24. Riehle, M. M., A. F. Bennett, R. E. Lenski, and A. D. Long. 2003. Evolutionary changes in heat-inducible gene expression in lines of Escherichia coli adapted to high temperature. Physiol. Genomics 14:47-58. [PubMed] 25. Sauer, U., and B. J. Eikmanns. 2005. The PEP-pyruvate-oxaloacetate node as the switch point for carbon flux distribution in bacteria. FEMS Microbiol. Rev. 29:765-794. [PubMed] 26. Sauer, U., D. R. Lasko, J. Fiaux, M. Hochuli, R. Glaser, T. Szyperski, K. Wuthrich, and J. E. Bailey. 1999. Metabolic flux ratio analysis of genetic and environmental modulations of Escherichia coli central carbon metabolism. J. Bacteriol. 181:6679-6688. [PubMed] 27. Shaver, A. C., P. G. Dombrowski, J. Y. Sweeney, T. Treis, R. M. Zappala, and P. D. Sniegowski. 2002. Fitness evolution and the rise of mutator alleles in experimental Escherichia coli populations. Genetics 162:557-566. [PubMed] 28. Ullmann, A. 1996. Catabolite repression: a story without end. Res. Microbiol. 147:455-458. [PubMed] 29. van Winden, W. A., J. C. van Dam, C. Ras, R. J. Kleijn, J. L. Vinke, W. M. van Gulik, and J. J. Heijnen. 2005. Metabolic-flux analysis of Saccharomyces cerevisiae CEN.PK113-7D based on mass isotopomer measurements of (13)C-labeled primary metabolites. FEMS Yeast Res. 5:559-568. [PubMed] 30. van Winden, W. A., C. Wittmann, E. Heinzle, and J. J. Heijnen. 2002. Correcting mass isotopomer distributions for naturally occurring isotopes. Biotechnol. Bioeng. 80:477-479. [PubMed] 31. Wick, L. M., H. Weilenmann, and T. Egli. 2002. The apparent clock-like evolution of Escherichia coli in glucose-limited chemostats is reproducible at large but not at small population sizes and can be explained with Monod kinetics. Microbiology 148:2889-2902. [PubMed] 32. Yang, C., Q. Hua, T. Baba, H. Mori, and K. Shimizu. 2003. Analysis of Escherichia coli anaplerotic metabolism and its regulation mechanisms from the metabolic responses to altered dilution rates and phosphoenolpyruvate carboxykinase knockout. Biotechnol. Bioeng. 84:129-144. [PubMed] 33. Zamboni, N., E. Fischer, A. Muffler, M. Wyss, H. P. Hohmann, and U. Sauer. 2005. Transient expression and flux changes during a shift from high to low riboflavin production in continuous cultures of Bacillus subtilis. Biotechnol. Bioeng. 89:219-232. [PubMed]
|
This article has been 
