|Home | About | Journals | Submit | Contact Us | Français|
Studies of steady-state metabolic fluxes in Escherichia coli grown in nutrient-limited chemostat cultures suggest remarkable flux alterations in response to changes of growth-limiting nutrient in the medium (Hua et al., J. Bacteriol. 185:7053-7067, 2003). To elucidate the physiological adaptation of cells to the nutrient condition through the flux change and understand the molecular mechanisms underlying the change in the flux, information on gene expression is of great importance. DNA microarray analysis was performed to investigate the global transcriptional responses of steady-state cells grown in chemostat cultures with limited glucose or ammonia while other environmental conditions and the growth rate were kept constant. In slow-growing cells (specific growth rate of 0.10 h−1), 9.8% of a total of 4,071 genes investigated, especially those involved in amino acid metabolism, central carbon and energy metabolism, transport system and cell envelope, were observed to be differentially expressed between the two nutrient-limited cultures. One important characteristic of E. coli grown under nutrient limitation was its capacity to scavenge carbon or nitrogen from the medium through elevating the expression of the corresponding transport and assimilation genes. The number of differentially expressed genes in faster-growing cells (specific growth rate of 0.55 h−1), however, decreased to below half of that in slow-growing cells, which could be explained by diverse transcriptional responses to the growth rate under different nutrient limitations. Independent of the growth rate, 92 genes were identified as being differentially expressed. Genes tightly related to the culture conditions were highlighted, some of which may be used to characterize nutrient-limited growth.
Microorganisms are capable of varying their metabolic fluxes and sizes of intermediate pools over a wide range in response to changes in environmental conditions. These metabolic responses are usually studied in chemostat cultures, where growth conditions can be well controlled. By designing the composition of the feeding medium carefully, growth in chemostats may be limited by a single or multiple nutrients. Carbon limitation is the most widely used type and has been applied to investigate cell metabolism (9, 17, 18, 30, 37), metabolic differences among knockout mutants (11, 30, 38), metabolic reaction models (7), and so on. Growth in carbon-deficient medium may be regarded as limiting energy or catabolism. On the other hand, although energy generation rates are high in carbon-replete cultures, cell yields are usually low, resulting in anabolism limitation and overflow metabolism (9, 18, 23). Other nutrient-limited chemostat cultures include limitation of the nitrogen source, phosphorus, or sulfur, among which nitrogen limitation is the most frequently studied (9, 18, 30, 37). Growth, metabolism and energetics in chemostat cultures under carbon-limited and nitrogen-limited conditions have been investigated for Saccharomyces cerevisiae, Bacillus subtilis, and other microorganisms (5, 9, 13, 17, 18). For Escherichia coli, the influences of nutrient limitation on metabolic responses has been investigated in recent studies by using information on isotopomer distribution (11, 30). Metabolic changes and kinetic properties during the transition from glucose-excess to glucose-limited growth conditions have also been studied (37).
Recently, we developed the method of flux ratio analysis (30) to identify central reaction networks and obtained reliable metabolic fluxes in wild-type W3110 and some knockout mutants of Escherichia coli grown in either glucose-limited or ammonia-limited chemostats (16, 38). Steady-state flux information suggested remarkable metabolic alterations in these strains in response to changes of the limiting nutrient. It is known that E. coli growing in nutrient-limited chemostat cultures, as in natural conditions, is capable of adapting to these environments quickly and achieving suboptimal growth. To elucidate these physiological adaptations through the flux change and understand the molecular mechanisms underlying the change in the flux, it is therefore very important to investigate the transcriptional responses to different nutrient-limited growth conditions. On the other hand, although a number of genes responsible for cell adaptation to carbon- or nitrogen-limited conditions and many important results relating to its molecular mechanisms have been revealed (2, 12, 13, 19, 20, 28), the transcriptional responses of most genes in the E. coli genome are not yet clear. A DNA microarray is one of the most effective techniques to investigate global gene expression and has been used in many studies (15, 25-27, 34).
In this work we investigated gene expression in E. coli grown in glucose-limited or ammonia-limited chemostat cultures by DNA microarray analyses. Analyses of transcriptional responses to the change of the growth-limiting nutrient were emphasized, which may provide valuable information to elucidate the change in metabolic flux and cell adaptation. Comparisons of gene expression between two nutrient-limited cultures were conducted for both slow- and fast-growing cells to study the influence of growth rate and identify the genes most important to nutrient-limited growth.
E. coli K-12 W3110 [F− λ− IN(rrnD-rrnE)1 rph-1] was used to investigate the difference in global gene expression under two nutrient-limited growth conditions. Cells were grown in glucose- or ammonia-limited chemostat cultures. The composition of the glucose-limited medium was (per liter): 5.0 g of glucose, 1.0 g of NH4Cl, 2.7 g of (NH4)2SO4, 6.8 g of Na2HPO4, 3.0 g of KH2PO4, 0.6 g of NaCl, 0.2 g of MgSO4 · 7H2O, 1.0 μg of thiamine HCl, 2.0 μl of polypropylene glycol 2000 as an antifoaming agent, and 10 ml of trace element solution (30). For the ammonia-limited culture, the concentrations of NH4Cl and (NH4)2SO4 were reduced to 0.20 g and 0.24 g per liter, respectively, while all other components remained constant.
Chemostat cultures were operated at 37°C in a BMJ-02PI bioreactor (ABLE, Tokyo, Japan) with a working volume of 1 liter. The culture medium was continuously fed to the bioreactor at a dilution rate (D) of 0.10 ± 0.005 or 0.55 ± 0.025 h−1, and the working volume was kept constant by withdrawing culture broth through a continuously operating pump. The pH of the culture was maintained at 7.0 by automatic addition of 2.0 M NaOH. An agitation speed of 450 rpm and constant airflow of 1.0 liter/min ensured dissolved oxygen concentrations above 60% saturation. The concentrations of oxygen and carbon dioxide exhausted from the bioreactor were monitored with an exhaust gas analyzer, Off-Gas Jr. DEX-2562 (ABLE). The steady-state condition was ascertained when the optical density of the biomass at 600 nm, the dissolved oxygen concentration, and the exhaust gas concentrations had remained constant for at least three volume changes. In steady-state cultures, almost all cells in the reactor could be considered to be growing at the same specific growth rate, which is equal to the dilution rate. Chemostat cultures were performed at least twice under each dilution rate-nutrient limitation combination.
Cell growth was monitored by measuring the optical density at 600 nm. Cell dry weight was determined by collecting cell pellets from 100 to 200 ml of culture aliquots, washing them with distilled water, and drying them at 85°C until a constant weight was obtained. The concentrations of glucose and ammonium in the medium were determined with enzymatic test kits (Roche Molecular Biochemicals, Mannheim, Germany) on a microplate spectrofluorometer (SPECTRAmax GEMINI XS; Molecular Devices, Sunnyvale, Calif.), following the changes in NAD(P)H fluorescence at the 355- and 460-nm wavelength pair. The concentrations of other metabolites were determined as given elsewhere (16). At least three measurements were performed for each metabolite concentration.
Total RNA was prepared with a modified hot phenol method (1). Cells were first harvested by centrifugation at 11,000 × g for 3 min, resuspended in 0.5 ml of solution A (0.5% sodium dodecyl sulfate, 20 mM sodium acetate, 10 mM EDTA) and mixed by pipetting with 0.5 ml of acidic phenol (pH 5.5) preheated at 60°C. The mixture was incubated at 60°C for 5 min. After centrifugation at 13,000 × g for 3 min at room temperature, the supernatant was recovered and the phenol extraction process was repeated. A phenol-chloroform (1:1, pH 5.5) extraction was then performed, and the RNA was precipitated by the addition of 3 volumes of ethanol. The RNA pellet was dried and dissolved in a DNase solution (100 mM sodium acetate, 50 mM MgSO4) containing 5 U of RNase-free DNase (Takara Shuzo Co., Ltd., Kyoto, Japan), and incubated at room temperature for 1 h. A second phenol-chloroform extraction and RNA precipitation were then performed. Purified total RNA was subjected to 1% agarose gel electrophoresis to check for degradation and whether the 23S and 16S rRNAs were recovered without contamination of genomic DNA.
High-density custom-made DNA microarrays containing 4,071 independent genes cloned from the E. coli K-12 W3110 strain were prepared by the Takara Shuzo Company. In addition to genes of the E. coli genome, 24 spots containing the human β-actin gene were present on the slide as negative controls. The details of the preparation and accuracy checking of the DNA microarray were described elsewhere (26, 27).
The preparation of cDNA fluorescently labeled with indocarbocyanine (Cy3) and indodicarbocyanine (Cy5) and microarray hybridization were performed essentially according to the MGuide (http://cmgm.stanford.edu/pbrown/mguide/index.html) with some modifications. Fluorescently labeled cDNA probes were prepared by random priming methods with 5.3 nmol of random hexamer (Takara Shuzo). Reverse transcriptase reactions were performed twice with 50 U of avian myeloblastosis virus reverse transcriptase XL (Life Sciences, St. Petersburg, Fla.) and 4 nmol of either Cy3-dUTP or Cy5-dUTP (Amersham Pharmacia Biotech, Piscataway, N.J.) with 30 μg of total RNA from E. coli grown under each of the four dilution rate-nutrient limitation conditions. Labeled cDNA probes were purified with Centri-Sep columns (Princeton Separations, Adelphia, N.J.), phenol-chloroform extraction, and ethanol precipitation. After drying, the cDNA probe was dissolved in 9 μl of water. Both Cy3- and Cy5-labeled cDNA probes were then added to a final volume of 21 μl of hybridization buffer consisting of 4× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate), 0.2% sodium dodecyl sulfate, 5× Denhardt's solution, and 100 ng of salmon sperm DNA μl−1 and denatured by heating at 98°C for 2 min.
The denatured cDNA probe was applied to the microarray, and hybridization was carried out at 65°C for 16 h. The slides were then washed consecutively with 2× SSC (60°C, 5 min), 0.2× SSC containing 0.1% sodium dodecyl sulfate (60°C, 5 min), and 0.2× SSC (room temperature, 5 min or more). The fluorescence intensities of the slides were scanned immediately after centrifugation (200 × g, room temperature, 2 min) with a GMS 418 array scanner (Genetic Microsystems, Woburn, Mass.) and recorded to 16-bit image files. The signal intensity of each spot in the microarray was quantified with IMAGENE software (BioDiscovery, Marina del Rey, Calif.).
For each growth-nutrient condition, two steady-state cell samples from two separate cultures were taken for analysis. Differences in gene expression between ammonia-limited growth and glucose-limited growth at the same growth rate were studied. For each comparison, two hybridization samples were prepared from two independent cell sample pairs (with reversal of dyes), and DNA microarrays were duplicated for each cDNA hybridization, generating four transcriptional data per gene. The signal intensity of each gene was first corrected by subtracting the local background value. Then, genes were classified into three groups by comparison with the intensity of the negative control spots. Group I consisted of genes for which both Cy3 and Cy5 signal intensities were greater than the mean + 1 standard deviation (SD) of the negative controls. Group II consisted of genes for which either Cy3 or Cy5 signal intensity was greater than the mean + 1 SD of the negative controls. Group III consisted of genes for which both Cy3 and Cy5 signal intensities were lower than the mean + 1 SD of the negative controls. Expression differences (ratios of the fluorescence intensities of Cy3 and Cy5) of all group I genes were then normalized by defining the mean of ratios of all genes as 1.0. Group II genes with sufficiently high Cy3 or Cy5 signal intensities (usually exceeded the high-intensity cutoff) were considered. Group III genes were ignored due to extremely low expressions under both growth conditions.
To ensure that the observed transcript alterations were really caused by the change of the growth-limiting nutrient, we also assessed the random fluctuation inherent in our microarray system. Cy5- and Cy3-labeled cDNA probes were simultaneously synthesized from the same template RNA, and their intensities were compared after hybridization. Only a few spots were found to have reproducible twofold changes in intensity due to artificial errors and systematic biases. When transcription in ammonia-limited growth was compared with that in glucose-limited growth, significantly more genes with twofold or greater transcriptional changes were observed, suggesting the transcriptional responses to nutrient-limited cultures could be well investigated by this microarray system.
Significant alteration in transcription was then recognized if one of the following criteria was satisfied: (i) for genes whose four measurements were all group I data, expression ratios should be reproducibly >2.0 or <0.5 and in good consistency (P < 0.05, paired Student's t test); (ii) for genes whose four measurements were all group II data, sufficiently high Cy3 or Cy5 intensities should be observed for all measurements; (iii) for genes with both group I and group II data, reproducible twofold or more expression changes should be observed for all group I data and sufficiently high Cy3 or Cy5 intensities should be observed for all group II data.
When E. coli strain W3110 was cultivated at a D value of 0.10 h−1 with either limited glucose or limited ammonia supply, steady-state cultures could be achieved after six to eight volume changes, which was further ascertained by the measurement of residual substrate concentrations at regular intervals. The residual nutrient concentrations were about 23 μmol/liter for glucose and 50 μmol/liter for ammonia when cells were grown at the low growth rate. When the dilution rate of the chemostat culture increased to 0.55 h−1, 20 to 25 volume changes after the initiation of feeding were necessary to ensure steady-state cultures. In this case of high growth rates, there were about 27 μmol of residual glucose and 50 μmol of residual ammonia per liter in the medium of glucose-limited and ammonia-limited cultures, respectively. Cell yields on glucose in cultures deficient in glucose were more than 1.5-fold higher than those in cultures deficient in ammonia under both dilution rate conditions. Several metabolic by-products were excreted from the cells in the case of ammonia deficiency, where the primary by-product was acetate and others included pyruvate, α-ketoglutarate, ethanol, fumarate, and some extracellular polysaccharide. In chemostat cultures with glucose limitation, however, no other metabolic by-products except CO2 were excreted into the medium at detectable levels (Table (Table11).
For each growth rate condition, the number of genes that could be classified in groups I and II after one hybridization exceeded 3,300 (more than 80% of the total genes on the slide). By combining four individual hybridizations, it was found that 400 genes (9.8% of the total genes analyzed) were differentially expressed in slow-growing cells adapting to glucose- or ammonia-deprived environments. The number of genes with significant transcriptional changes was, however, decreased markedly with an increase in the growth rate. The number of differentially expressed genes under two nutrient limitations and the corresponding percentages in their functional categories are summarized in Table Table2.2. For genes with known functions, it was observed that ammonia limitation might upregulate genes involved in amino acid metabolism and transport system to high transcription levels, while a deficiency of glucose in chemostat cultures might greatly induce genes involved in central carbon and energy metabolisms, transport systems, and the cell envelope.
At the low growth rate, expressions of most of the 138 genes encoding enzymes of amino acid metabolism were induced in ammonia-limited cells. Among these upregulated genes, the transcript levels of 32 were significantly elevated in response to nitrogen deficiency. It is known that there are two routes of ammonia assimilation in enterobacteria. One involves the synthesis of glutamine and glutamate by glutamine synthetase (GS, product of the glnA gene) and glutamate synthase (GOGAT, product of the gltB and gltD genes). Another route is direct glutamate synthesis from α-ketoglutarate and ammonia by NADP-linked glutamate dehydrogenase, the product of the gdhA gene. In cells growing under ammonia limitation, unlike the slight increase in expression of the gdhA gene, the transcript level of the glnA gene was significantly elevated (7.8- and 4.4-fold increase at the low and high growth rates, respectively) (Table (Table3),3), indicating that the former route is primarily responsible for the assimilation of ammonia in this condition. The glnA gene is the first gene of the glnA-ntrBC operon, which involves another two genes, ntrB and ntrC, that encode the nitrogen-regulatory proteins NtrB and NtrC, respectively. When the nitrogen source in the medium is abundant, the expression of three genes is regulated by NtrC, resulting in relatively low intracellular concentrations of GS, NtrB, and NtrC. The extremely low residual ammonia concentration in ammonia-limited culture, however, results in a drop in the intracellular concentration of glutamine, which enables PII uridylyltransferase (product of the glnD gene) to uridylylate PII protein to form PII-UMP and release NtrB from its association with PII. NtrC is then phosphorylated by NtrB to form NtrC phosphate, which activates the initiation of transcription at the σ54-dependent promoter glnAp2 and results in high expression of the glnA-ntrBC operon (2, 20). The expression of the glnD gene was also induced markedly (two- to fivefold) in ammonia-limited cells. The resulting product of PII uridylylation (PII-UMP) may then stimulate the deadenylylation (activation) of partly inactive GS (GS-AMP) by adenylyltransferase. Therefore, significantly elevated transcriptional and posttranscriptional levels of glutamine synthetase, together with high expression of the gltB and gltD genes, which encode glutamate synthase, provided high intracellular concentration of GS and GOGAT, capable of harvesting the remaining ammonia in the medium.
Among the genes involved in amino acid metabolism, the asnB gene, encoding glutamine-dependent asparagine synthetase, is one of the genes that showed the most significant increase in expression when culture conditions were changed from glucose limitation to ammonia limitation (Table (Table3).3). Although there is another ammonia-dependent asparagine synthetase encoded by asnA in prokaryotes, this enzyme is not preferred for asparagine synthesis under nitrogen limitation due to insufficient intracellular ammonia (Km for ammonia, about 0.3 mM). Indeed, no change in expression of the asnA gene in response to ammonia limitation was observed in our experiments, which may be due to the repression of asnA expression by the elevated NtrC concentration (28). Most of the other genes with large increases in transcript levels were genes involved in the synthesis of amino acids, including thrA and asd, involved in the common biosynthetic pathway of aspartate family amino acids; lysC, dapB, and dapD, involved in lysine synthesis; hisG, which encodes the first-step enzyme in histidine synthesis; and two genes of the ilvIH operon, encoding acetolactate synthase, which catalyzes first enzymatic step of valine and isoleucine synthesis. Besides these, the expression of the trpEDCAB operon genes (encoding the enzymes of tryptophan synthesis) and the argECBH operon genes (encoding the enzymes of arginine synthesis) were also elevated in slow-growing ammonia-deprived cells.
In contrast to most of the upregulated genes involved in amino acid biosynthesis, no significant expression was found for genes involved in the degradation of amino acids except for the genes of the astCADBE operon (2-fold to 10-fold increase in ammonia-deprived cells) (Table (Table3).3). Genes of the astCADBE operon encode the enzymes of the arginine succinyltransferase pathway, the dominant arginine catabolic pathway in E. coli. Previous studies suggested that some enzymes of the arginine succinyltransferase pathway are nitrogen regulated, and the activities of these enzymes increased greatly when arginine or aspartate was used as the nitrogen source (29, 31). This was also confirmed by our results for gene expression under ammonia limitation.
Of the 515 genes involved in central carbon and energy metabolism, differentially expressed genes were equally distributed between glucose-limited and ammonia-limited cells when the growth rate was low. With an increase in the growth rate, the number of genes with elevated expression in ammonia-deprived cells decreased significantly, whereas an increase in number was observed for genes that were differentially expressed when glucose was deficient. A lack of ammonia in the medium generally induced genes of the Embden-Meyerhof-Parnas and pentose phosphate pathways. Besides these, all of five genes in two glycogen operons (glgBX and glgCAP) were also expressed at significantly higher levels (Table (Table4),4), corresponding to the large amount of glycogen storage (more than 10% of the cell's dry weight) when the cell grows slowly under ammonia limitation (16). Moreover, significant elevation of transcript levels of pyruvate dissimilation genes were also observed in ammonia-deprived cells. These included the pflA gene (encoding pyruvate formate-lyase I activating enzyme), the ackA gene (encoding acetate kinase), and the adhE gene (encoding alcohol dehydrogenase/acetaldehyde dehydrogenase). High transcript levels of these genes were probably responsible for the excretion of acetate, ethanol, and formate as overflow metabolites in cells growing under ammonia limitation (16).
Unlike the general upregulation of genes in the Embden-Meyerhof-Parnas and pentose phosphate pathways, expression of most of the genes of the tricarboxylic acid cycle as well as glyoxylate and dicarboxylate metabolism declined in ammonia-limited cells under both growth rates (Table (Table4).4). Among these downregulated genes, four genes of the sdhCDAB operon that encode succinate dehydrogenase; the sucD gene, encoding succinate-coenzyme A ligase α chain; the fumA and fumC genes, which encode fumarase; and aceA, encoding isocitrate lyase were expressed with significantly lower transcript levels. The expression of a considerable number of oxidative phosphorylation genes was upregulated in glucose-limited cells in response to the limitation of energy. Moreover, with the increase in growth rate, an increase was observed for the oxidative phosphorylation genes that had elevated transcript levels under glucose limitation, among which 8 of 13 genes of the nuo operon encoding NADH dehydrogenase I were differentially expressed in glucose-limited cells. The remarkable elevation in aerobic respiration suggested the further requirement of ATP synthesis by faster-growing cells to satisfy the rapid biosynthesis of cell macromolecules in carbon- and energy-limited conditions.
A considerable number of genes involved in membrane transport were found to have differential expression in cells grown in the two different nutrient limitation conditions (Table (Table2).2). Ammonia deficiency in the medium significantly induced the transcript levels of a number of genes that encode transport proteins for nitrogen-containing metabolites. These included the proP gene (encoding proline/betaine transport protein), the cycA gene (encoding d-serine/d-alanine/glycine permease), the gabP gene (encoding γ-aminobutyrate permease), the nupC gene (encoding a nucleoside permease), and genes of the potFGHI operon (encoding the ATP-binding cassette [ABC] transport system for putrescine) (Table (Table5).5). Besides these genes, the functions of which are well annotated, expression of most of the genes involved in two hypothetical ABC transport systems, YddOPQRS and YliABCD, was also induced significantly by ammonia limitation, indicating the possibility of transport of certain nitrogen-containing metabolites by these systems. Cells grown under ammonia limitation also expressed the transporter gene amtB (encoding ammonium/methylammonium transport B protein [AmtB]) at markedly elevated transcript levels (more than 27-fold and 44-fold changes in slow- and fast-growing cells, respectively).
It was proposed that AmtB and its homologue proteins are responsible for the increase in diffusion rate of ammonia (NH3) across the cytoplasmic membrane rather than actively transporting NH4+ or NH3 (32, 33). Although it was reported that AmtB function is required when the external NH3 concentration drops to 50 nM (about 10 μM external ammonium concentration at pH 7) or less (33), the result of the remarkable increase in expression of the amtB gene implied that its product may function earlier, such as when the residual ammonium concentration was about 50 μM, as in our ammonia-limited chemostats. Another significantly expressed gene was the nac gene, which, like the amtB gene, is a nitrogen-regulatory (Ntr) gene which is transcriptionally regulated by the nitrogen-regulatory proteins NtrB and NtrC (3, 21, 39). The nac gene encodes the nitrogen assimilation control protein NAC, which activates transcription from a number of σ70-dependent promoters to transcribe operons whose products provide the cell with alternative nitrogen sources (21). A >100-fold increase in the transcript levels of this gene in both slow- and fast-growing cells suggests full induction of the nitrogen regulatory system in E. coli to cope with nitrogen starvation and optimize cell growth.
In contrast to the induction of transport systems of possible nitrogen sources under ammonia limitation, expression of a number of genes encoding membrane transporter proteins for various carbon sources was significantly induced in nitrogen-replete but carbon-limited chemostat cultures (Table (Table5).5). About half of these are genes involved in a wide range of high-affinity ABC transport systems that facilitate the uptake of maltose (all genes of the malGFE and malK-lamB-malM operons), galactose (three genes of the mglBAC operon), xylose (xylF), and ribose (rbsB). Other transporter genes induced by glucose limitation included those encoding the galactitol-specific phosphotransferase system (gatB and gatC), mannose-specific phosphotransferase system (manXYZ), and an ion-coupled transporter responsible for the aerobic transport of C4 dicarboxylates (dctA).
Significantly high expression of ABC transporter genes may be under the influence of high internal cyclic AMP (cAMP) levels and the induction of endogenous sugar inducers (13, 35). Among the above-mentioned ABC transporter genes, the expression of the malM gene dropped markedly to below the intensity of the negative controls when ammonia was deficient. The function of this gene is, however, not yet clear. The lamB gene, located in the same operon, also showed a significant change in expression. This gene encodes the maltoporin LamB, which facilitates the diffusion of maltose and glucose across the outer membrane and acts as a general porin for small solutes. The transcript level of the lamB gene was more than 20-fold higher in glucose-deprived cells than in ammonia-deprived cells. Ammonia-limited growth was found to be not greatly impaired by the loss of LamB, possibly due to the small size of NH3 and NH4+ ions (19) and the improved ammonia acquisition ability resulting from the overexpressed amtB gene. Although LamB and a general porin, OmpF, were considered the major players in outer membrane permeability under sugar, iron, or phosphate limitation (13, 19), the results of the insignificant expression changes in the ompF gene between glucose and ammonia limitation indicate that LamB might be the primary porin that contributes to glucose permeability through the outer membrane under glucose limitation.
Other genes that were regulated significantly by nutrients include the codA, guaD, and cfa genes, involved in nucleotide and fatty acid metabolism, and the glnK and cbl genes, involved in regulatory function. All of these genes showed differential expression independent of the growth rate (Table (Table6).6). The codA gene, one gene of the codBA operon, encodes cytosine deaminase, one enzyme involved in pyrimidine catabolism. Since another gene of the codBA operon (codB, encoding cytosine permease) was also upregulated, the remarkable increase in the expression of this operon under ammonia limitation is in good agreement with the recent results that codBA expression is regulated by the nitrogen-regulatory system through the direct activation of nitrogen assimilation control protein NAC (22, 39).
The expression of the guaD gene (encoding guanine deaminase of the pyrine catabolic pathway) was induced most among the genes involved in nucleotide metabolism when ammonia in the medium became deficient (more than 20-fold and 4-fold increases at the low and high growth rates, respectively). Our results support the hypothesis that this pyrine catabolism gene from E. coli is also a nitrogen-regulated gene, though further verification is required. Indeed, the gde gene, which encodes guanine deaminase in Bacillus subtilis, was found to be expressed only during nitrogen-limited conditions, and the control of gde expression required regulatory proteins such as GlnA, GlnR, and TnrA, which are important for regulating the expression of nitrogen metabolism genes (24).
Differential expression was also observed for the cfa gene. This gene encodes cyclopropane fatty acid (CFA) synthase, which catalyzes the postsynthetic modification of the lipid bilayer (8). Although it is known that one of the two promoters of this gene is activated by RpoS and that acid conditions increase cfa transcription (6, 8), there was only a slight expression difference for the rpoS gene and no pH change between the two nutrient-limited chemostat cultures, which may indicate other possibilities that induce the expression of the cfa gene. Further studies are therefore required to show whether the cfa gene is nitrogen regulated or whether a nitrogen assimilation-related activator is involved in the transcription of this gene.
The glnK gene, which encodes signal transduction protein GlnK, was a strict group II gene in this study, which agreed well with the result of Atkinson and Ninfa (3) that no GlnK was found in nitrogen-replete cells and a high concentration of GlnK was found in nitrogen-starved cells. The product of glnK is a paralogue of the signal transduction protein PII (product of glnB) and is known to play a role distinct from that of PII protein in nitrogen regulation (3, 4). It is supposed that most PII has been uridylylated to PII-UMP before the accumulation of GlnK, and the high expression of the glnK gene in ammonia-limited cells may provide additional GlnK-UMP to activate the deadenylylation of GS-AMP. Although the presence of GlnK may also activate the phosphorylase activity of NtrB and adenylylate GS, its effect in regulating adenylyltransferase and controlling the level of NtrC-P appeared less potent than that of PII protein. Further study and quantification of the relative contents of GlnK and GlnK-UMP would help a lot in understanding the role of this protein.
The cbl gene, lying downstream of the nac gene and encoding a CysB-like regulator, Cbl, for sulfur utilization, is probably expressed from the nac promoter under nitrogen limitation (39). Absolutely elevated transcript levels of the cbl gene (85-fold and 13-fold increases in slow- and fast-growing cells, respectively) may imply an important role of the Cbl protein in nitrogen metabolism in addition to its role in sulfate utilization (36).
Although more than 40% of the genes in E. coli have unknown functions, our results suggested that only about 1.1% of them (21 genes) were differentially expressed at both growth rates in response to nutrient-limited growth (Table (Table6).6). In ammonia-limited cells, most genes in the ycdGHIJKLM (JW0991-7) and yeaGH (JW1772-3) operons were expressed at higher transcript levels than under glucose limitation, while the most significant expression changes (more than 16-fold increase) could be observed for the yedL and yhdW genes in ammonia-limited cells and the yjcG gene in glucose-limited cells. Further experiments are required to clarify the functions of these genes in cell growth under nutrient limitation.
Chemostat cultures provide microorganisms with steady-state growth under well-controlled conditions, making them suitable for the study of genome-wide transcriptional responses to environmental changes such as nutrient limitation for S. cerevisiae (5) or E. coli in this study. Regardless of the growth rate, about 1.0% of E. coli genes (42 genes) were significantly upregulated by ammonia limitation, half of which were involved in amino acid metabolism and transport systems for nitrogen-containing compounds. On the other hand, glucose deficiency in the medium induced another 50 genes (mainly involved in central carbon and energy metabolism and the transport or uptake of carbon sources) at high transcript levels.
Comparing transcriptional profiles between steady-state cells grown under different nutrient limitations allows the investigation of genes involved in the nitrogen-regulatory system. It is known that the remarkably decreased intracellular glutamine level under nitrogen limitation triggers the initiation of the σ54-dependent promoter of the glnA-ntrBC operon, resulting in elevated transcription of the three genes of this operon. The nitrogen-regulatory protein of NtrC then activates the transcription of a number of σ54-dependent genes, including the nac gene. The nitrogen assimilation control protein NAC, the product of the nac gene, can induce a number of σ70-dependent genes, which, together with the NtrC-activated genes, encode proteins responsible for scavenging, transport, and utilization of many nitrogen-containing compounds (2, 20, 21).
Although growth rate affected markedly the relative expression of genes between the two nutrient conditions, our results suggest that 42 genes involved in 26 operons may be tightly related to the physiological adaptation to nitrogen limitation independent of the growth rate (Table (Table6).6). Fifteen of the 26 operons were also identified as being nitrogen regulated by Zimmer et al. when the mRNA levels in an ntrB-overexpressing mutant were compared to those in a strain with an ntrC null allele (39). These include four genes involved in nitrogen regulation (ntrB, ntrC, nac, and glnK); nine genes involved in the transport of ammonia (amtB), nucleosides (nupC), γ-aminobutyric acid (gabP), d-serine/d-alanine/glycine (cycA), d-alanyl-d-alanine dipeptide (yddP, yddQ, yddR, and yddS) and amino acid (yhdW, hypothetical); 10 genes of nitrogen-containing compound metabolism or other functions (five genes of the astCADBE operon, oat, glnA, codA, cbl, and ddpX); and another eight hypothetical genes (yedL, yeaGH, and five genes in the ycdGHIJKLM operon). Except for the putrescine transport genes of the potFGHI operon (significantly expressed in slow-growing cells and upregulated in fast-growing cells), differential expression was not observed in our experiments for nine other operons that were determined to be under nitrogen regulation (gltIJKL, glnHPQ, ompF, chaBC, oppABCDF, ydcSTUV, hisJQMP, argT, and dppABCDF) (39). On the other hand, 11 additional genes that are significantly induced by ammonia limitation were identified in our chemostat cultures (Table (Table6),6), suggesting that these genes may play important roles in the adaptation to nitrogen deficiency. Moreover, among 42 significantly upregulated genes, the transcript levels of 6 genes (glnK, nac, amtB, yhdW, yedL, and cbl) were increased by more than 10-fold, which may then be used to indicate ammonia limitation in the medium.
Limitation of carbon source such as glucose obviously induced another group of genes (Table (Table6).6). The first noteworthy result is the ability of E. coli to scavenge carbon from glucose-deficient medium through elevating the expression of a number of transport genes, among which all genes of the malEFG, malK-lamB-malM, and mglBAC operons were expressed at significantly high levels. Since nitrogen limitation has less influence on the expression of these genes in glucose-replete cultures (12), our transcription results indicate the significant induction of these genes by glucose limitation, which is also consistent with the results with lacZ fusions (12) and two-dimensional gel electrophoresis (37). The significance of highly expressed maltose regulon (malEFG and malK-lamB-malM) and galactose transport operon (mglBCA) genes to glucose-limited growth has also been demonstrated by Ferenci's group (10, 12-14, 19). It was found that the LamB porin (product of the lamB gene, which was induced by more than 20-fold) has glucose affinity, through which most glucose may diffuse across the outer membrane rather than through other porins such as OmpF and OmpC when E. coli grows on extremely low glucose. On the other hand, the ABC transporter of galactose (Mgl), which also has high affinity for glucose, showed more than a 10-fold increase in transcription under glucose limitation and is likely to play an important role in transporting periplasmic glucose into the cytoplasm in addition to the glucose-specific phosphotransferase system. Investigations suggested that high cAMP levels and elevated endogenous inducers of maltotriose and galactose are possible regulators that induce the high expression of the mgl/lamB high-affinity system in an environment of extremely low glucose concentration (12, 13). An upregulation in expression of two genes required for endogenous synthesis of galactose, galE and galU (10), was also confirmed by our transcriptional analysis. Similarly, a >10-fold increase in transcription of genes in glucose-deprived cells can be used as indicators for this nutrient-limited condition. These genes include five genes involved in carbon transport systems (mglB, mglA, malE, lamB, and malM) and five other genes involved in central carbon metabolism (pckA and tnaA) and other functions (cstA, yjcG, and acs). For S. cerevisiae, however, glucose-limited growth may be characterized by the transcription of many more genes (5).
Comparison of gene transcription indicates that nearly 10% of the total genes were differentially expressed in slow-growing cells in response to the change of the growth-limiting nutrient. The number of differentially expressed genes, however, decreased markedly when chemostat cultures were carried out at the high growth rate (Table (Table2).2). To investigate the influence of growth rate on gene expression, additional DNA microarray experiments were performed to compare transcriptional profiles between slow- and fast-growing cells grown under nutrient limitation, i.e., glucose-limited growth at D = 0.55 h−1 (C0.5) versus that at D = 0.10 h−1 (C0.1) and ammonia-limited growth at D = 0.55 h−1 (N0.5) versus that at D = 0.10 h−1 (N0.1). Similarly, for each growth rate, two independent steady-state RNA samples were prepared and DNA microarrays were duplicated for each hybridization, yielding four sets of data for each experiment.
Only genes that were differentially expressed at the low growth rate were investigated in this study. Based on the transcriptional responses to growth rate, these genes were roughly divided into two groups. For group A genes, the transcriptional response to the change in growth rate was similar under two nutrient limitations, i.e., kC ≈ kN, where kC and kN indicate the average ratio of transcription in fast-growing cells to that in slow-growing cells grown under glucose or ammonia limitation, respectively. Therefore, for most group A genes, differential expression between the two nutrient-limited conditions could be maintained even if the growth rate varied markedly. Group B includes genes for which kC is very different from kN, suggesting significant differences in transcriptional responses to the growth rate between the two nutrient-limited cultures. Group B genes can be further divided into two subgroups, i.e., genes whose expression ratios (logarithmic ratios) in glucose-limited growth were much higher than those in ammonia-limited growth (kC kN) and vice versa (kC kN).
Among 400 genes that were differentially expressed in slow-growing cells with either ammonia limitation or glucose limitation, the effects of growth rate on the transcription of 35 genes may be independent of the growth-limiting nutrient if the maximum relative difference permitted between kC and kN was 25% (Table (Table7).7). Moreover, the growth rate did not significantly affect the expression of most group A genes, particularly genes such as pckA, yfhT, rbsB, and ybhH (kC ≈ kN ≈ 1). For most of the genes considered, transcriptional responses to growth rate were also affected by the growth-limiting nutrient in the medium. A >5-fold difference in transcriptional response was observed for about one-sixth of the genes investigated when the growth-limiting nutrient was changed. It was also observed that the expression of most genes was much more sensitive to the growth rate in ammonia-limited cells than in glucose-limited cells.
For genes highly induced by ammonia limitation in slow-growing cells, high positive differences in transcriptional responses to the growth rate between the two nutrient-limited growth conditions [log2(kC) − log2(kN)] generally resulted in the decreased induction or even repression by ammonia limitation in fast-growing cells, such as the xasA and trpE genes and genes around the hdeAB (histone-like protein-determined expression) operon (hdeB, hdeA, and yhiE) (Table (Table7).7). On the other hand, for genes highly repressed by ammonia limitation in slow-growing cells, a high negative value for [log2(kC) − log2(kN)] generally resulted in decreased repression by ammonia limitation in fast-growing cells. Most genes in this group were flagellar genes (Table (Table7),7), which agrees well with the results that most flagellar genes that were significantly downregulated under ammonia limitation at the low growth rate were no longer differentially expressed with the increase in the growth rate (data not shown). It can then be concluded that E. coli is capable of regulating gene transcription in a wide variety of ways in response to a change in growth rate and growth-limiting nutrient.
Our results also indicate that an increase in the growth rate may reduce the transcriptional difference between the two nutrient-limited growth conditions. Indeed, although 74 new genes were differentially expressed at the high growth rate, 308 genes were no longer differentially expressed in fast-growing cells (Table (Table2),2), which also suggests that the effects of growth rate on gene expression in E. coli may be much greater than the effects of nutrient-limited growth.
We are grateful to T. Ara and E. Inamoto for technical support and M. Robert for critical reading of the manuscript and helpful discussions.
This work was supported by a grant from the New Energy and Industrial Technology Development Organization of the Ministry of Economy, Trade and Industry of Japan (Development of a Technological Infrastructure for Industrial Bioprocess Project).