pH Regulates Genes for Flagellar Motility, Catabolism, and Oxidative Stress in Escherichia coli K-12

Abstract

Gene expression profiles of Escherichia coli K-12 W3110 were compared as a function of steady-state external pH. Cultures were grown to an optical density at 600 nm of 0.3 in potassium-modified Luria-Bertani medium buffered at pH 5.0, 7.0, and 8.7. For each of the three pH conditions, cDNA from RNA of five independent cultures was hybridized to Affymetrix E. coli arrays. Analysis of variance with an α level of 0.001 resulted in 98% power to detect genes showing a twofold difference in expression. Normalized expression indices were calculated for each gene and intergenic region (IG). Differential expression among the three pH classes was observed for 763 genes and 353 IGs. Hierarchical clustering yielded six well-defined clusters of pH profiles, designated Acid High (highest expression at pH 5.0), Acid Low (lowest expression at pH 5.0), Base High (highest at pH 8.7), Base Low (lowest at pH 8.7), Neutral High (highest at pH 7.0, lower in acid or base), and Neutral Low (lowest at pH 7.0, higher at both pH extremes). Flagellar and chemotaxis genes were repressed at pH 8.7 (Base Low cluster), where the cell's transmembrane proton potential is diminished by the maintenance of an inverted pH gradient. High pH also repressed the proton pumps cytochrome o (cyo) and NADH dehydrogenases I and II. By contrast, the proton-importing ATP synthase F₁F_o and the microaerophilic cytochrome d (cyd), which minimizes proton export, were induced at pH 8.7. These observations are consistent with a model in which high pH represses synthesis of flagella, which expend proton motive force, while stepping up electron transport and ATPase components that keep protons inside the cell. Acid-induced genes, on the other hand, were coinduced by conditions associated with increased metabolic rate, such as oxidative stress. All six pH-dependent clusters included envelope and periplasmic proteins, which directly experience external pH. Overall, this study showed that (i) low pH accelerates acid consumption and proton export, while coinducing oxidative stress and heat shock regulons; (ii) high pH accelerates proton import, while repressing the energy-expensive flagellar and chemotaxis regulons; and (iii) pH differentially regulates a large number of periplasmic and envelope proteins.

Escherichia coli and related enteric bacteria respond to a wide range of pH stresses by regulating gene expression (for reviews see references 21 and 68) and protein profiles (73, 82). Enteric bacteria encounter a wide range of external pHs in their natural habitat, the human digestive tract (17). Colonization of the intestine requires transient survival through the stomach at pH 1 to 2 (fasting) or 2 to 7 (transiently, during feeding) (18), as well as exposure to pancreatic secretions at pH 10 (25) followed by growth and persistence at a range of external pHs of 5 to 8 (20). Growth at a pH substantially higher or lower than the cytoplasmic pH 7.6 induces protective responses with two fundamental aims: to maintain internal pH homeostasis and to prepare the cell to survive future exposure to more extreme pH conditions (below pH 5 or above pH 9) that no longer permit growth (11, 41, 70).

The effects of pH on enteric bacteria contribute to disease. Low pH enhances expression of numerous virulence factors, such as the ToxR-ToxT virulence regulon in Vibrio cholerae (7), the phoP-phoQ regulon of Salmonella enterica (6), and the pH 6 antigen of Yersinia pestis (50). Acid stress contributes to food preservation; many food preservatives are membrane-permeant acids whose uptake is enhanced by acid (60), and acid interacts in complex ways with both temperature and organic food preservatives (65).

While growth in acid challenges pH homeostasis, the pH difference across the inner cell membrane (ΔpH) nevertheless contributes cell energy in the form of proton potential or proton motive force (Δp). The proton potential powers motility, ATP synthesis, and catabolite transport (for a review see reference 29). But low pH also amplifies the uptake of membrane-permeant acids that dissipate the proton potential (59). Thus, we expect low pH to induce a combination of positive and negative responses.

Much of bacterial catabolism affects pH, and in E. coli a growing number of catabolic enzymes and catabolite transporters are known to be regulated by pH (21, 73). Sugar fermentation initially generates short-chain acids that are excreted but accumulate and reenter the cytoplasm, causing acidification. Thus, it is not surprising that sugar transporters such as OmpF and the maltose regulon are down-regulated at low pH (13). Consumption of acids by the tricarboxylic acid (TCA) cycle causes alkalinization, a common result of growth to stationary phase in tryptone-based media (66, 73). Catabolism of amino acids by decarboxylases generates alkaline amines, which help the cell counteract external acidification, for example, the lysine and arginine decarboxylases (4, 27, 45, 47, 71). High pH, however, induces deaminases that generate acids, such as tryptophan deaminase (tnaAB) and serine deaminase (sda) (9, 73, 82).

A complicated case is that of the glutamic acid decarboxylase genes gadA and gadBC (12, 44). The gad system enables cells to survive extreme acid (77), but its expression is induced mainly at high pH, or in Luria-Bertani medium grown to stationary phase, where pH naturally increases (73, 82). An alternative role of gad, particularly under anaerobiosis, may be to channel its product γ-aminobutyric acid into fermentation acids.

Even mild acid (pH 6 to 7) greatly amplifies the uptake of membrane-permeant weak acids such as acetate. Permeant acids pass through the bacterial membrane and dissociate in the cytoplasm, causing accumulation of anions and depression of internal pH (34, 56). Acetate concentrations rise as cell density increases, and acetate induces a large number of genes and proteins (3, 35). Growth inhibition occurs as a result of both lower internal pH and the differential ability of anions to inhibit metabolism (60). The effect of permeant acids is critical in the human colon, where the concentration of short-chain fatty acids totals approximately 100 mM (15).

While numerous responses to pH stress are known, the mechanisms by which E. coli maintains its internal pH at 7.6 remain poorly understood. The electron transport chain pumps protons outside the cell, and the H⁺-ATPase either exports or imports protons, but mutants in these components maintain pH homeostasis. There is evidence that potassium exchange contributes to pH homeostasis in external acid (5, 10, 52, 80), but the precise mechanisms remain unclear. At high pH, the electrical potential (Δψ) is diminished in order to compensate for the inverted ΔpH. The sodium-proton antiporterNhaA contributes to internal pH maintenance under sodium stress (24, 75). High pH also induces major stress systems such as heat shock response (1, 28, 74), the SOS regulon (63), and the CpxP envelope stress response (16).

At more extreme pH values, well below the growth range (as low as pH 1.5 for clinical isolates) E. coli can retain viability for many hours, a phenomenon termed acid survival or acid resistance. Acid resistance is enhanced by many genes induced during growth at the acid end of the pH range (pH 5) or growth to stationary phase. Acid-induced acid resistance factors include periplasmic chaperones such as the hdeA product (23), envelope proteins such as OsmY, and redox modulators such as Tpx (73, 78). A complex acid resistance regulon including the gad system is regulated by transcription factors GadX-GadW and EvgA-YdeO, as well as by RpoS, H-NS, and cyclic AMP (11, 12, 44, 79). E. coli also exhibits base resistance, the ability to survive at or above pH 10 (58, 70). Base resistance requires rpoS and components of the gad system (30).

Finally, pH may affect flagellar motility, although the present picture is unclear. According to one report, growth in acid represses flagellar genes and eliminates motility (72), whereas another group finds motility enhanced by acetate and propionate, which cause acid stress (53).

To investigate acid and base response, we used microarrays to compare E. coli gene expression at low, neutral, or high external pH. Past microarray studies of pH response have been limited by their absence of pH conditions above pH 7 (44, 78); their use of glucose minimal medium (78), in which many catabolic genes are repressed; and their focus on only a single acid resistance regulon (44). Our experimental design included both acid and base conditions, as well as pH 7.0. For each growth condition, five independent cultures were hybridized separately, a number of replicates that ensured detection of virtually all expression ratios of at least twofold. The coregulation of numerous genes within operons confirmed the biological relevance of our expression ratios. Our study revealed unexpected patterns of pH response and clarified the overlap of pH stress with other stress responses.

MATERIALS AND METHODS

Growth conditions.

E. coli K-12 strain W3110 (R. VanBogelen and F. Neidhardt) was grown overnight in unbuffered potassium-modified Luria broth (LBK) (10 g of tryptone/liter, 5 g of yeast extract/liter, 7.45 g of KCl/liter). For pH-controlled growth, media were buffered with 100 mM homopiperazine-N,N′-bis-2-(ethanesulfonic acid) (HOMOPIPES) (pK_a, 4.55 and 8.12). The pH of the media were adjusted to 5.0, 7.0, or 8.7 with KOH solution to avoid extra sodium ions, which stress cells at high pH (24). To maximize aeration and maintain logarithmic growth, the overnight culture was diluted 1,000-fold into 12 ml of buffered medium in a 125-ml baffled flask and rotated at 240 rpm. Cultures were grown at 37°C to an optical density at 600 nm of 0.3. For all cultures, the pH was tested after growth to ensure that the values were maintained at ±0.2 pH unit of the pH of the original uninoculated medium.

To observe motility, we used E. coli K-12 strain RP437 and S. enterica serovar Typhimurium SJW1103 from a laboratory in which strains are maintained for motility (M. Macnab). Culture was spotted on tryptone-KCl soft-agar plates (0.35% Bacto Agar) and incubated at 37°C until cells swam out. Culture was picked from the leading edge of the swimming cells and inoculated into LBK for overnight growth. For quantitative assay of motility, 5 μl of culture was spotted in triplicate on plates containing tryptone-KCl with 100 mM sulfonate buffer of appropriate pK_a (73). After growth for 8 h, the diameter of motile cell growth was measured.

RNA isolation.

Bacterial RNA was isolated using the Qiagen RNeasy kit with on-column DNA digestion (Qiagen), with additional DNA removal with Ambion DNase. To perform this additional DNase digestion, RNA was precipitated and redissolved in 85 μl of nuclease-free water. We then added 10 μl of 10× DNase I buffer and 5 μl of (1-U/μl) DNase I (Ambion). The DNase reaction mixture was incubated at 37°C for 30 min and then chilled on ice. A second RNeasy column purification was performed.

cDNA preparation and array hybridization.

For microarrays, standard methods were used for cDNA synthesis, fragmentation, and end-terminus biotin labeling, based on Affymetrix protocols. Labeled cDNA was hybridized to E. coli Affymetrix Antisense Genome Arrays. Hybridized arrays were stained with streptavidin-phycoerythrin with the use of the Affymetrix Fluidic Station. After staining, arrays were scanned with a GC2500 scanner.

Statistical analysis of gene expression.

The experiment was designed so as to minimize both false-positive and false-negative results for expressed genes. Five full replicates (with respect to E. coli growth, RNA isolation, sample preparation, and array hybridization) were performed for each pH condition.

The median within-group variance in expression for all genes in the data set was 0.031 (or standard deviation, 0.175). To test for significant differences in expression between the pH classes, one-way analysis of variance (ANOVA) was performed at a significance level of 0.001; thus, of every thousand genes tested, only one false positive would be expected. For a gene with average within-group variability, our sample size provided statistical power of 98% to detect a twofold difference in gene expression among pH groups. That is, only 2% of genes that show a twofold difference in expression between any two pH groups would be missed (false negatives).

Model-based expression analysis with dChip software (40) was performed on the probe-level data from Affymetrix's DAT files. The model relates target RNA levels to the probe signals by a linear function that weights the significance of all oligonucleotide probes for each gene. The analysis includes normalization, which rescales data from different arrays so that comparisons can be made among arrays. Each array was normalized to a baseline array from a pH 7 culture, by using local regression on an invariant set of probes (62). Model-based expression indices were calculated for each gene on each array by using only the perfect match probes (61), and outlier detection was performed (39). Only probe sets that received an Affymetrix call of “present” on greater than 50% of the arrays were used in subsequent analyses. “Present” or “absent” calls use information from paired perfect-match and single-base-mismatch probes. Four thousand six hundred fifty probe sets passed this criterion.

For genes whose probe sets passed the 50% screen, one-way ANOVA was performed on the log₂-transformed model-based expression indices, on a gene-by-gene basis. For each gene that displayed significant differences in expression among the classes, pairwise comparisons of pH classes were determined using Tukey's multiple comparisons procedure to control the familywise error rate for the t test.

Additional analyses were performed to explore categories of differential gene expression. Global relationships among arrays were visualized by performing a principal component analysis (81) on the expression data and plotting arrays in two-dimensional space corresponding to the first two principal components. The gene expression profiles of the arrays were visualized in two-dimensional Euclidian space, by using BRB ArrayTools software. In addition, categories of differential expression profiles across the pH classes were generated by a hierarchical cluster analysis of differentially expressed genes, based on the average linkage method (19) with BRB ArrayTools.

RESULTS

Growth range of pH.

To study the full range of pH response, we selected the widest pH range (pH 5.0 to 8.7) in which cultures maintained reasonable doubling times and approximately constant pH throughout growth. Culture media were adjusted to pH 5.0, 7.0, and 8.7. The doubling time for E. coli cultured at pH 5.0 and 8.7 was approximately 25 min and at pH 7.0 was 18 min. All cultures were grown to an optical density of 0.3 in order to facilitate at least five complete replications. The final pH of growth cultures was found to be within ±0.2 of the initial pH. The internal pH of the cytoplasm is approximately 7.6 (69); thus, growth at external pH 7.0 might induce some acid response.

Probe hybridization.

To determine differential gene expression, the log₂ transforms of normalized model-based expression values of genes were compared. Of the 7,231 genes and intergenic regions (IGs) on the array, 4,650 loci were detected on more than half (eight or more) of the 15 arrays. These loci, constituting about 70% of the total array, were taken for further analysis.

Principal component analysis.

Global relationships among arrays were visualized by performing a principal component analysis (81) on the expression data (Fig. 1). Before dimensional reduction, each array existed in 4,650-dimensional space (one dimension for each of the 4,650 intensity values). The array comparisons were plotted in two-dimensional space, corresponding to the first and second principal components of variation. The first principal component for each array is the weighted linear combination of intensity values that shows maximum variation, whereas the second principal component is a weighted linear combination orthogonal to the first component that has maximum variance.

FIG. 1.

Principal component analysis. The gene expression profiles of the arrays were visualized in two-dimensional Euclidian space, by using BRB ArrayTools software as described under Materials and Methods. The first and second principal components are shown. pH 5.0, squares; pH 7.0, circles; pH 8.7, triangles.

The principal component analysis indicated that the microarrays from each of the three pH conditions appeared in distinct groups (Fig. 1). Within-class variability was small relative to variability among pH levels. The pH 8.7 arrays showed the greatest degree of separation, clustering into two groups based on the date on which the arrays were hybridized, but this difference was small compared to the differences between pH classes.

ANOVA for significance of expression profiles.

We compared gene expression among the three pH groups on a gene-by-gene basis using one-way ANOVA at a significance level of 0.001. The significance level indicates the probability of a false positive, and we therefore expect 0.001 × 4,650 = 4.65 false-positive genes (i.e., genes that are not truly differentially expressed but that appear in our differentially expressed list) in our full analysis. Of the 4,650 loci with eight or more “present” calls on arrays, 761 genes and 353 IGs showed a significant F value for differential expression among the three pH classes. Thus, about 17% of E. coli genes showed significant modulation of expression as a function of pH.

Cluster analysis.

As a first attempt at categorizing differentially expressed genes, we performed a hierarchical cluster analysis (19) of differentially expressed genes (Fig. 2). We used average linkage and one minus the centered Pearson correlation as the distance metric. At a correlation value of approximately 0.6, the dendrogram generated six clusters of gene expression profiles.

FIG. 2.

Cluster analysis of differentially expressed genes. The dendrogram was generated based on the average linkage method (19) with BRB ArrayTools. At a correlation of 0.6, six clusters of related gene expression were designated Acid High (AH), Acid Low (AL), Base High (BH), Base Low (BL), Neutral High (NH), and Neutral Low (NL).

Within each of the six clusters, the average profiles were determined for all the gene expression indices (log₂ intensity values) across the three pH conditions (Fig. 3). The clusters were defined by their mean expression profiles across the three pH conditions. The Acid High cluster showed highest expression at pH 5.0, declining at pH 7.0 and 8.7. It included 160 genes and 49 IGs. Acid Low (113 genes, 57 IGs) showed approximately the reverse profile, with its lowest expression at pH 5.0, rising at pH 7.0 and 8.7. Base High (93 genes, 70 IGs) showed low expression at pH 5.0 and 7.0 and higher expression at pH 8.7, whereas Base Low (123 genes, 40 IGs) showed the reverse, higher expression at pH 5.0 and 7.0 than at pH 8.7. The Neutral High cluster (93 genes, 14 IGs) showed highest expression at pH 7.0 and lower expression at both pH extremes. The Neutral Low cluster (181 genes 123 IGs) showed the lowest expression at pH 7.0 and higher expression at both pH extremes, although the mean expression was substantially greater at pH 5.0 than at pH 8.7; a number of acid-induced genes fell in this category.

FIG. 3.

Cluster mean expression profiles. The mean expression profiles over pH are plotted for the six clusters defined in Fig. 2.

Table 1 lists the genes that fell into each cluster; details of description and Blattner open reading frame (ORF) numbers are available online in Table S1 in the supplemental material. In many cases, all or most of the ORFs in a given operon were induced in the same cluster; see, for example, the atp operon (Base High cluster) and the flg and fli operons (Base Low cluster).

TABLE 1.

Clusters of pH-dependent genes

^aAcid induced (68, 73, 82).

^bData from five arrays from pH 5.0; no significant expression at pH 7.0 and 8.7.

^cAcetate induced (3, 35).

^dExtreme acid resistance (44, 78, 79).

^eBase induced (68, 73, 82).

^fBase induced (16).

Known acid-induced genes and acid resistance genes such as sucBC and hdeA (73) generally fell under Acid High, Base Low, or Neutral Low, a cluster whose mean expression indices were actually twofold higher in acid than in base (Fig. 3). These results are generally consistent with the cluster pH profiles and with the structure of the cluster dendrogram, in which the Acid High profile correlates most closely with the Neutral Low profile. Most known base-induced genes, such as alx (ygjT) (8, 73) and tnaA (9), fell under Base High or Acid Low.

For IGs, the cluster assignment and expression ratios are presented online in Table S2 in the supplemental material. Expression of an IG may result from a small regulatory RNA that lies between protein-encoding genes (2, 43), or it may indicate the tail end of mRNA containing pH-regulated genes. For example, the IGs upstream of tnaC (tnaA leader peptide) and downstream of tnaB both were repressed in acid, as are tnaA and tnaB.

Individual gene expression ratios.

For genes whose overall expression profile yielded a significant F value (one-way ANOVA), we used the Tukey procedure to determine ratios of average model-based expression indices from cultures at pH 5.0 versus pH 7.0, at pH 8.7 versus pH 7.0, and at pH 8.7 versus pH 5.0. The full list of individual log₂ expression ratios for all analyzed genes is presented in Table S1 in the supplemental material and for IGs is presented in Table S2 in the supplemental material; for genes of particular interest grouped in functional categories, the data are presented in Tables 3 through 7. Expression ratios that are significant at α = 0.001 are shown in boldface.

TABLE 3.

Flagellar and chemotaxis genes

Gene	Function	Log2 pH ratioa			Classb
		5/7	8.7/7	8.7/5
cheA	Chemotaxis sensor kinase	−1.125	−3.199	−2.0474	BL
cheB	Protein methylesterase	−1.050	−1.578	−0.528	BL
cheR	Chemotaxis MCPc methyltransferase	−0.564	−1.013	−0.448	BL
cheW	Chemotaxis signal transducer	−1.336	−2.785	−1.449	BL
cheY	Response regulator for chemotactic signal	−1.089	−1.310	−0.221	BL
cheZ	CheY-P phosphatase	−1.505	−3.199	−1.694	BL
flgA	Flagellar synthesis	−0.261	−1.056	−0.795	BL
flgB	Basal body rod subunit	−0.192	−1.120	−0.928	BL
flgC	Basal body rod subunit	−0.257	−1.241	−0.984	BL
flgD	Basal body rod modification	0.133	−1.241	−1.107	BL
flgE	Hook subunit	0.272	−0.856	−1.128	BL
flgF	Basal body rod subunit	0.109	−1.239	−1.348	BL
flgG	Basal body rod major subunit	0.220	−1.116	−1.335	BL
flgI	Basal body P-ring	−0.011	−1.218	−1.207	BL
flgJ	Flagellum-specific muramidase	0.030	−0.857	−0.886	BL
flgK	Flagellar synthesis	−0.211	−1.875	−1.664	BL
flgL	Flagellar synthesis	0.111	−1.165	−1.276	BL
flgM	Anti-sigma 28 (FliA); regulates FlhD	−0.292	−1.424	−1.132	BL
flgN	Flagellar synthesis	−0.295	−1.567	−1.272	BL
flhA	Flagellar export pore protein	0.333	−0.528	−0.861	AH
flhE	Function unknown	0.492	−0.454	−0.946	AH
fliA	Sigma 28; regulates class III flagellar genes	−0.150	−1.127	−0.976	BL
fliC	Flagellin subunit, H-antigen	−1.165	−4.561	−3.396	BL
fliD	Hook-associated protein	−0.311	−1.953	−1.641	BL
fliE	Flagellar synthesis; basal body component	−0.478	−1.861	−1.203	BL
fliF	Flagellar basal body M-ring	−0.159	−1.216	−1.057	BL
fliG	Motor switching and energizing	0.182	−1.313	−1.495	BL
fliH	Negative regulator of FliI; flagellar assembly and export	0.389	−1.245	−1.634	BL
fliI	Membrane ATPase, flagellar, axial subunit export	0.270	−1.238	1.508	BL
fliJ	Flagellar biosynthesis	−0.068	−1.404	−1.336	BL
fliK	Hook filament junction	0.278	−1.186	−1.464	BL
fliL	Rotational direction of flagella	−0.102	−1.083	−0.981	BL
fliM	Flagellar synthesis, motor switching and energizing	0.029	−1.053	−1.082	BL
fliN	Flagellar switch	0.273	−0.875	−1.148	BL
fliO	Flagellar synthesis	0.191	−0.921	−1.112	BL
fliP	Flagellar synthesis	0.419	−0.669	−1.087	AH
fliQ	Flagellar synthesis	0.463	−0.980	−1.444	AH
fliR	Flagellar synthesis	0.924	−0.899	−1.822	AH
fliS	Cytosolic chaperone inhibits premature FliC assembly	−0.174	−1.947	−1.772	BL
fliT	Flagellar synthesis	−0.323	−1.021	−0.698	BL
fliY	Cystine-binding protein, periplasmic; may regulate FliA (sigma 28)	0.233	−0.252	−0.484	AH
fliZ	Not required for motility; may regulate FliA (sigma 28)	0.144	−1.165	−1.309	BL
motA	Flagellar rotation	−0.682	−1.847	−1.166	BL
motB	Flagellar rotation	−1.241	−3.238	−1.997	BL
tap	Dipeptide chemoreceptor	−0.622	−1.089	−0.466	BL
tar	Aspartate, maltose chemoreceptor	−0.519	−2.213	−1.694	BL
tsr	Serine chemoreceptor	−1.061	−1.826	−0.765	BL
ycgR	Suppresses hns motility defect	−0.531	−0.783	−0.252	BL
yhjH	Suppresses hns motility defect	−0.972	−2.522	−1.550	BL

^aBoldface for ratios indicates significance (α = 0.001).

^bBL, Base Low; AH, Acid High.

^cMCP, methyl-accepting chemotaxis protein.

The genes most strongly regulated by pH are summarized in Table 2. These genes each showed an expression ratio of at least fourfold (log₂ = 2) between two of the pH classes. Note that the two genes most strongly induced in acid are ORFs with no known function, yhcN and yagU. Other acid-induced genes include those for catabolic enzymes in pathways that consume acids, such as sdhCD (succinate dehydrogenase). Genes repressed at high pH include several members of the flagellar regulon, including the main flagellar subunit fliC (for a review see reference 42).

TABLE 2.

Strongest pH-dependent expression ratios (fourfold or higher)

pH dependence and pH ratio	Gene	Log2 ratio		pH dependence and pH ratio	Gene	Log2 ratio
Acid induced
5.0/7.0	yagU	3.220
	yhcN	3.064
	sdhC	2.728
	lysP	2.662
	sdhD	2.349
	cfa	2.075
	nemA	2.060
5.0/8.7	yhcN	4.199
	yagU	3.962
	fliC	3.396
	fimA	2.579
	cfa	2.555
	gltB	2.271
	ydiY	2.193
	ycdN	2.147
	yncD	2.117
	mqo	2.075
	dhaH	2.074
	cheA	2.074
	motB	1.997
7.0/8.7	fliC	4.561
	malM	3.780
	malK	3.748
	lamB	3.735
	malP	3.373
	motB	3.238
	cheA	3.199
	cheZ	3.199
	flxA	3.007
	malE	2.933
	malQ	2.883
	cheW	2.785
	ibpB	2.618
	yhjH	2.522
	htpG	2.439
	deoC	2.267
	pstS	2.262
	dnaK	2.249
	tar	2.213
	yjdA	2.208
	dnaJ	2.155
	yrfG	2.055
	yheL	2.031
	deoA	2.003
Base induced
8.7/7.0	yifO	2.769
	ymgD	2.221
8.7/5.0	tnaC	5.517
	cpxP	4.234
	tnaA	4.028
	nmpC	3.961
	treB	3.895
	yjiY	3.665
	treC	3.233
	b3913	3.176
	yifO	3.095
	borD	3.088
	tnaB	2.993
	ycfS	2.820
	yghJ	2.762
	ymgD	2.433
	yccA	2.378
	yfiA	2.364
	yebE	2.343
	yjiX	2.292
	nrdD	2.182
	dniR	2.081
	alx	2.009
	mutL	2.000
7.0/5.0	tnaC	5.026
	lamB	4.881
	malK	4.790
	malM	4.643
	yjiY	4.359
	nmpC	4.012
	malP	4.000
	tnaA	3.805
	malE	3.425
	borD	3.378
	malQ	3.322
	cpxP	3.232
	treB	3.156
	yghJ	3.081
	yjiX	2.959
	fecB	2.842
	ompF	2.834
	treC	2.613
	fecA	2.608
	pstS	2.479
	b3913	2.177
	fecE	2.098

The genes most strongly induced at high pH included tnaC, encoding the tryptophanase leader peptide (26), as well as tnaA (tryptophanase) and the Trp transporter gene tnaB, with its leader peptide gene tnaC. Previously in proteomic gels, we found tryptophanase to be the most highly expressed protein observed at high pH (9). The alkali-inducible protease gene cpxP (16) was also strongly induced. Members of the maltose transport regulon (malEKM) were strongly repressed by acid, consistent with previous reports (31, 73). But proteins strongly induced by base also included those from genes of unknown function, such as yifO and ymgD.

Flagellar and chemotaxis regulons.

Motility in E. coli is governed by the flagellar chemotaxis regulon including 50 components in 19 operons, governed by the major regulators FlhC and FlhD (42, 76). The expression of the regulatory operon flhCD is controlled by numerous environmental response systems, such as adenylate cyclase (37), RcsCDB (22), and ClpXP (76).

Nearly all the genes of the flagellar regulons (47 genes) were repressed at high pH (Table 2). Forty-one genes fell in the Base Low cluster, which means that the bulk of significant expression difference occurred between pH 7.0 and 8.7. (The other six genes were Acid High.) These genes were among the most strongly base-repressed genes in the arrays (Table 2); for instance, fliC, encoding the flagellin monomer, had the lowest pH 8.7/pH 7.0 ratio observed, down-regulated about 20-fold (Table 3). Some of the che and mot genes showed a relatively small degree of repression in acid compared to that at pH 7.0 but overall were repressed at high pH.

The major regulator operon flhCD, however, showed no effect of pH. Thus, either the flhCD probes failed to show up in our arrays or pH may affect expression posttranscriptionally.

Motility assays.

The effect of pH on motility was tested by spotting motile cultures of E. coli K-12 RP437 and S. enterica serovar Typhimurium SJW1103 on motility agar buffered at a range of pH values (Fig. 4). Both species showed a steady decline of motility as pH increased. The decline was particularly steep between pH 7.5 and 8.7.

FIG. 4.

Swimming distance as a function of pH. E. coli K-12 RP437 and S. enterica serovar Typhimurium SJW1103 were spotted on soft-agar plates as described under Materials and Methods. Error bars represent standard errors of the means (n = 3); in most cases their size was smaller than the symbol.

Catabolism and proton transport.

Several enzymes for catabolism of sugars and amino acids show a pH dependence that minimizes acid production at low external pH or maximizes acids at high pH (68, 73). Our microarrays revealed many new components, showing the broad scope of pH regulation of catabolism (Table 4).

TABLE 4.

Catabolism and respiration

Group	Gene	Function	Log2 pH ratioa			Classb
			5/7	8.7/7	8.7/5
Sugar catabolism and TCA cycle	aceE	Pyruvate dehydrogenase	1.928	0.527	−1.401	NL
	aceF	Pyruvate dehydrogenase dihydrolipoamide acetyltransferase	1.802	0.402	1.401	NL
	aceK	Isocitrate dehydrogenase kinase/phosphatase	0.667	0.038	−0.629	AH
	acnA	Aconitase A, stationary phase induced	0.769	−0.143	−0.912	AH
	acnB	Aconitase B; 2-methylaconitate hydratase	1.036	−0.041	−1.077	NL
	acrR	Regulator for acrA and acrB	−0.873	0.164	1.037	AL
	dcuR (yjdG)	Fumurate respiration regulator (anaerobic)	−0.220	0.055	0.275	AL
	dcuS (yjdH)	Fumurate respiration regulator (anaerobic)	−0.324	0.065	0.390	AL
	dhaK	Dihydroxyacetone kinase, subunit I	1.041	−0.379	−1.420	AH
	dhaL	Dihydroxyacetone kinase, subunit II	0.742	−0.385	−1.127	AH
	dld	d-Lactate dehydrogenase	0.919	0.566	−0.353	NL
	eno	Enolase; RNA degradosome	0.299	0.052	−0.247	AH
	fucI	l-Fucose isomerase	−0.693	−0.173	0.520	AL
	fucK	l-Fuculose kinase	−1.259	−0.519	0.740	NH
	fucR	Positive regulator, fuc operon	−0.046	0.495	0.541	BH
	galF	Putative regulator of galU				NL
	galK	Galactokinase	−0.616	−0.562	0.053	BL
	galM	Galactose mutarotase; aldose-1-epimerase	−0.619	−0.647	−0.028
	gapA	Glyceraldehyde 3-P dehydrogenase A	0.093	−0.752	−0.844	BL
	gatA	Galactitol-specific enzyme IIA of PTSd	0.929	−0.626	−1.555	AH
	gatB	Galactitol-specific enzyme IIB of PTS	0.632	−0.895	−1.528	AH
	gatC	Galactitol-specific enzyme IIC of PTS	0.973	−0.949	−1.922	AH
	gatD	Galactitol-1-phosphate dehydrogenase	1.037	−0.893	−1.930	AH
	gatY	d-Tagatose-1,6-bisphosphate aldolase, class II	0.846	−0.483	−1.329	AH
	gatZ	Enhances GatY activity	0.774	−0.605	−1.380	AH
	glpA	Glycerol-3-phosphate dehydrogenase large subunit (anaerobic)	−0.200	0.558	0.758	AL
	glpB	Glycerol-3-phosphate membrane anchor (anaerobic)	−0.124	0.437	0.562	BH
	glpC	Glycerol-3-phosphate dehydrogenase (anaerobic) small subunit	−0.169	0.469	0.638	BH
	glpX	Fructose 1,6-bisphosphatase	−0.162	0.375	0.537	BH
	gltA	Citrate synthase	0.288	−0.559	−0.846	AH
	gnd	Gluconate-6-phosphate dehydrogenase	0.895	0.351	−0.544	NL
	gntT	High-affinity gluconate transport	−0.613	−0.362	0.251	NH
	gpsA	Glycerol-3-phosphate dehydrogenase	−0.492	−0.534	−0.041	NH
	gpmA	Phosphoglycerate mutase I	0.303	−0.118	−0.421	AH
	icdA	Isocitrate dehydrogenase	1.211	0.061	−1.272	AH
	lldD	l-Lactate dehydrogenase	0.613	−0.554	−1.167	AH
	lldP	l-Lactate permease; glycolate uptake	1.527	0.219	−1.307	NL
	lpdA	Lipoamide dehydrogenase; E3 component of pyruvate and 2-oxoglutarate dehydrogenase complexes	1.507	0.281	−1.226	NL
	malE	Maltose-binding protein, periplasmic	−3.425	−2.933	0.491	NH
	malF	Maltose transport, inner membrane	−1.016	−0.935	0.082	NH
	malG	Maltose transport, inner membrane subunit	−1.577	−1.502	0.075	NH
	malK	Maltose transport, ATP-binding subunit	−4.790	−3.748	1.041	NH
	malM	Periplasmic protein, mal regulon	−4.643	−3.780	0.863	NH
	malP	Maltodextrin phosphorylase	−4.000	−3.373	0.627	NH
	malQ	Amylomaltase	−3.322	−2.883	0.439	NH
	malT	mal positive regulator	−0.103	0.617	0.720	BH
	pdhR	Pyruvate dehydrogenase operon repressor	0.755	−0.109	−0.863	AH
	pfkB	6-Phosphofructokinase-2	0.363	0.095	−0.268	NL
	pflA	Pyruvate formate lyase I activase	0.964	0.667	−0.297	NL
	pflB	Pyruvate formate lyase I (anaerobic)	0.812	0.578	−0.234	NL
	pgi	Glucose phosphate isomerase	0.654	0.177	−0.477	NL
	pta	Phosphotransacetylase	1.167	0.639	−0.528	NL
	ptsG	Glucose PTS enzyme IIBC	0.841	0.544	−0.287	NL
	ptsH	PTS system histidine phosphocarrier protein Hpr	0.369	0.564	0.195	NL
	ptsI	PTS system enzyme I	0.241	0.339	0.098	NL
	ptsO	NPr, N-regulated HPr-like protein	0.639	0.168	−0.470	NL
	rpiA	Ribose-5-phosphate isomerase A	0.180	0.364	0.184	BH
	srlA	Sorbitol-specific enzyme II of PTS	−1.156	−1.720	−0.564	BL
	srlB	Sorbitol-specific enzyme III of PTS	−0.908	−1.140	−0.232	BL
	srlD	Sorbitol-6-phosphate dehydrogenase	−0.660	−1.334	−0.674	BL
	srlE	srl operon protein	−1.013	−1.613	−0.599	BL
	srlR	srl regulator	−0.151	0.804	−0.654	BL
	sucA	2-Oxoglutarate dehydrogenase, E1 component	0.850	−0.851	−1.701	AH
	sucB	Dihydrolipoamide succinyltransferase component of 2-oxoglutarate dehydrogenase complex (E2)	0.566	−1.073	−1.638	AH
	sucC	Succinyl-CoAc synthase beta subunit	0.494	−1.019	−1.513	AH
	sucD	Succinyl-CoA synthase alpha subunit	0.526	−1.225	−1.751	AH
	tpiA	Triosephosphate isomerase	0.772	0.245	−0.526	NL
	treB	Trehalose-specific PTS enzyme II	−3.156	0.739	3.895	AL
	treC	Trehalose-6-phosphate hydrolase	−2.613	0.620	3.233	AL
Proton transport and electron transport chain	atpA	ATP synthase subunit alpha, F1	0.111	0.460	0.349	BH
	atpB	ATP synthase subunit a, F0	0.125	0.484	0.609	BH
	atpC	ATP synthase subunit epsilon, F1	0.266	1.005	0.739	BH
	atpD	ATP synthase subunit beta, F1	0.542	1.172	0.630	BH
	atpE	ATP synthase subunit c, F0	0.182	0.503	0.321	BH
	atpF	ATP synthase subunit b, F0	−0.055	0.321	0.377	BH
	atpG	ATP synthase subunit gamma, F1	0.372	0.632	0.260	BH
	atpH	ATP synthase subunit delta, F1	−0.172	0.186	0.358	AL
	atpI	ATP synthase subunit, F1F0-type proton-ATPase	0.017	0.499	0.482	BH
	cydA	Cytochrome d (bd-I) terminal oxidase subunit I (microaerobic)	−0.268	0.872	1.140	BH
	cydB	Cytochrome d (bd-I) terminal oxidase subunit II (microaerobic)	−0.097	0.777	0.874	BH
	cydC	Cysteine exporter to periplasm required for Cyd assembly	0.135	0.490	0.355	BH
	cydD	Cysteine exporter to periplasm required for cytochrome assembly	0.232	0.574	0.342	BH
	cyoA	Cytochrome o oxidase subunit II	1.160	0.217	−0.943	NL
	cyoB	Cytochrome o oxidase subunit I	0.812	−0.204	−1.016	NL
	cyoC	Cytochrome o oxidase subunit III	1.026	0.019	−1.007	NL
	cyoD	Cytochrome o oxidase subunit IV	0.829	−0.192	−1.021	NL
	cyoE	Cytochrome o oxidase subunit protoheme IX farnesyltransferase	1.094	0.225	−0.869	NL
	fdoG	Formate dehydrogenase-O, major selenopeptide subunit	0.353	−0.138	−0.491	AH
	fdoH	Formate dehydrogenase-O Fe-S subunit	0.003	−0.414	−0.416	BL
	frdA	Fumarate reductase flavoprotein subunit	−0.305	0.046	0.351	AL
	frdC	Fumarate reductase membrane anchor polypeptide	−0.378	−0.008	0.370	AL
	fumA	Fumarase A	0.595	−0.647	−1.242	AH
	nfsA (mdaA)	Nitroreductase A	0.561	−0.419	−0.980	AH
	mdaB	Probable nitroreductase or quinone reductase	0.444	−0.223	−0.667	AH
	napC	Cytochrome electron source for NapAB, membrane bound	−0.797	−0.382	0.416	NH
	ndh	Respiratory NADH dehydrogenase II; NADH:ubiquinone oxidoreductase II	0.789	0.161	−0.628	NL
	nuoC	NADH:ubiquinone oxidoreductase subunit C	0.440	−0.084	−0.524	AH
	nuoG	NADH:ubiquinone oxidoreductase subunit G; NADH dehydrogenase I	0.525	0.103	−0.442	NL
	nuoH	NADH:ubiquinone oxidoreductase subunit H; NADH dehydrogenase I	0.564	0.185	−0.379	NL
	nuoI	NADH:ubiquinone oxidoreductase subunit I; NADH dehydrogenase I	0.792	0.310	−0.481	NL
	nuoJ	NADH:ubiquinone oxidoreductase subunit J; NADH dehydrogenase I	0.468	0.234	−0.234	NL
	nuoK	NADH:ubiquinone oxidoreductase subunit K; NADH dehydrogenase I	0.766	0.460	−0.306	NL
	nuoL	NADH:ubiquinone oxidoreductase subunit L; NADH dehydrogenase I	0.311	0.078	−0.233	NL
	nuoN	NADH:ubiquinone oxidoreductase subunit N; NADH dehydrogenase I	0.126	−0.210	−0.336	AH
	sdhA	Succinate dehydrogenase flavoprotein subunit	1.438	−0.018	−1.458	AH
	sdhB	Succinate dehydrogenase iron-sulfur protein	1.833	0.329	−1.505	AH
	sdhC	Succinate dehydrogenase membrane anchor subunit, cytochrome b556	2.728	0.890	−1.838	NL
	sdhD	Succinate dehydrogenase hydrophobic subunit	2.349	0.595	−1.754	NL
Amino acid catabolism and transport	artI	Arginine periplasmic binding protein	0.286	0.655	0.369	BH
	artM	Arginine periplasmic binding protein	−0.115	0.293	0.407	BH
	cadA	Lysine decarboxylase, degradative	1.024	0.137	−0.887	AH
	cysK	O-Acetylserine sulfhydrylase A (cysteine synthase)	1.204	1.351	0.147	NL
	dadA	d-Amino acid dehydrogenase	1.273	−0.229	−1.501	AH
	dadX	d-Amino acid dehydrogenase	0.683	−0.393	−1.076	AH
	dppC	Dipeptide permease system	−0.346	−0.032	0.315	AL
	gdhA	Glutamate dehydrogenase	−0.080	−0.496	−0.416	BL
	hisC	Histidinol-phosphate aminotransferase	0.253	0.984	0.731	BH
	hisF	Imidazole glycerol phosphate synthase (cyclase)	0.150	0.703	0.553	BH
	hisH	Amidotransferase of imidazole glycerol phosphate synthase	0.034	0.393	0.359	BH
	hisI	PR-ATP pyrophosphatase and PR-AMP cyclohydrolase	0.234	0.641	0.407	BH
	hisJ	Histidine-binding protein	−0.149	0.400	0.549	BH
	lysC	Aspartokinase III	1.464	−0.226	−1.690	AH
	lysP	Lysine permease	2.662	0.963	−1.698	NL
	lysU	Lysine-tRNA ligase	0.544	−0.077	−0.621	AH
	potD	Putrescine-ornithine transporter	0.053	0.509	0.456	BH
	sdaA	l-Serine deaminase, degradative	−0.409	0.638	1.048	BH
	sdaB	l-Serine deaminase	−1.111	−0.434	0.678	AL
	sdaC	H+/serine symporter; regulator of serine deaminase	−1.205	−0.410	0.794	AL
	tnaA	Tryptophan deaminase, degradative; also deaminases serine and cysteine	−3.805	0.223	4.028	AL
	tnaB	Tryptophan transporter	−1.840	1.153	2.993	AL
	tnaC	tnaA leader peptide	−5.026	0.490	5.517	AL
	tdcB	Threonine dehydratase, degradative	−0.849	−0.296	0.553	NH
	ydfG	l-allo-Threonine, l-serine, d-serine dehydrogenase	0.407	0.287	−0.120	NL

^aValues in boldface are significant (α = 0.001).

^bNL, Neutral Low; AH, Acid High; AL, Acid Low; BH, Base High; BL, Base Low; NH, Neutral High.

^cCoA, coenzyme A.

^dPTS, phosphotransferase.

Many operons encoding processes of glycolysis and the TCA cycle, such as aceEF (pyruvate dehydrogenase), dhaKL (dihydroxyacetone kinase), pta (phosphotransacetylase), and pts (glucose phosphotransferase), showed elevated expression in acid. Others, however, were elevated at high pH. Operons elevated at high pH tended to be those induced by anaerobiosis, such as glpABC (anaerobic glycerol-3-phosphate dehydrogenase), pflBA (anaerobic pyruvate formate lyase), and dcu (anaerobic fumarate respiration). The mal system, however, is strongly repressed by acid (13, 31) and showed up as such in our arrays.

Membrane-bound systems for proton and electron transport were regulated by acid or base along lines largely consistent with their relative degree of export or import of H⁺. An example is the atp operon encoding F₁F_o ATP synthase (32), which imports H⁺ during oxidative respiration. Most of the atp genes were strongly upregulated at high pH, whereas ndh and nuo (the NADH dehydrogenases I and II), which export H⁺, were down-regulated. The sdh gene (succinate dehydrogenase), which contributes electrons for proton export, is also down-regulated at high pH. On the other hand, cytochrome d oxidase (cyd) is expressed in preference to cytochrome o oxidase (cyo) at high pH, presumably because it exports half as many H⁺ per electron (14).

Enzymes for degradation of amino acids showed pH regulation as expected, with high pH favoring deaminase operons such as tna (tryptophan deaminase), sda (serine deaminase), and tdcB (threonine dehydratase). Acid induced only one of the decarboxylase operons, cad (lysine decarboxylase). Several decarboxylases are known to be induced by acid, but their induction is repressed by oxygen (4, 30), which may explain their absence in our highly aerobic cultures.

Oxidative stress and salicylate stress.

Several acid stress genes are known to overlap with oxidative stress, for example, the alkyl hydroperoxide reductase ahpC (9, 84), and certain permeant acids such as salicylate are considered oxidative stress agents (54). We surveyed our pH-regulated genes for overlap with response to H₂O₂, paraquat, and salicylate, as reported in references 54 and 84 (Table 5).

TABLE 5.

pH-regulated oxidative stress response^a

Gene	Function	Log2 pH ratiob			PQ, Sal, or H2O2	Classc
		5/7	8.7/7	8.7/5
acnA	Aconitase A	0.769	−0.143	−0.912	Sal	AH
adhE	Acetaldehyde-coenzyme A dehydrogenase	0.043	−0.380	−0.422	Sal	BL
ahpC	Alkyl hydroperoxide reductase small subunit	1.003	0.436	−0.568	PQ, H2O2	NL
ahpF	Alkyl hydroperoxide reductase large subunit	0.777	−0.222	−0.999	H2O2	AH
aldA	Aldehyde dehydrogenase, NAD linked	0.773	0.764	−0.009	PQ	NL
alx (ygjT)	Membrane protein, alkali induced	−1.317	0.692	2.009	PQ−	AL
artI	Periplasmic arginine binding protein	0.286	0.655	0.369	PQ	BH
aspA	Aspartate ammonia-lyase (aspartase)	−0.592	0.770	1.362	PQ−	BH
carA	Carbamoylphosphate synthase small subunit	0.029	1.059	1.030	PQ−	BH
cfa	Cyclopropane fatty acid synthase	2.075	−0.480	−2.555	Sal	AH
cyaA	Adenylate cyclase	0.691	−0.156	−0.846	Sal	AH
cyoD	Cytochrome o oxidase subunit IV	0.829	−0.192	−1.021	PQ	NL
cysK	Cysteine synthase	1.204	1.351	0.147	PQ, Sal, H2O2	NL
dadX	Alanine racemase	0.683	−0.393	−1.076	PQ	AH
deoA	Thymidine phosphorylase	−1.031	−2.003	−0.972	Sal	BL
deoB	Deoxyribouratase, phosphopentomutase	−0.731	−1.502	−0.771	PQ, Sal	BL
dhaH	Dihydroxyacetone phosphoryl donor	1.547	−0.527	−2.074	Sal	AH
dhaK	Dihydroxyacetone kinase	1.041	−0.379	−1.420	Sal	AH
dnaK	HSP-70-type molecular chaperone	−0.894	−2.249	−1.356	Sal	BL
dps	Stress response DNA-binding protein	1.130	0.105	−1.025	PQ, Sal, H2O2	NL
fliS	Flagellar synthesis; flagellar regulon member	−0.174	−1.947	−1.772	PQ−	BL
fpr	Ferredoxin NADP+ reductase; anaerobic	0.565	0.275	−0.289	PQ, H2O2	NL
gapA	GAPDHd A	0.093	−0.752	−0.844	Sal	BL
gatA	Galactitol-specific enzyme IIA of PTSe	0.929	−0.626	−1.555	PQ, Sal	AH
gatB	Galactitol-specific enzyme IIB of PTS	0.632	−0.895	−1.528	PQ, Sal	AH
gatC	Galactitol-specific enzyme IIC of PTS	0.973	−0.949	−1.922	Sal	AH
gatD	Galactitol-1-phosphate dehydrogenase	1.037	−0.893	−1.930	PQ, Sal	AH
gatZ	Tagatose 6-phosphate aldolase 2	0.774	−0.605	−1.380	Sal	AH
gltA	Citrate synthase	0.288	−0.559	−0.846	PQ, Sal	AH
gltB	Glutamate synthase, large subunit	0.846	−1.425	−2.271	Sal	AH
grxA	Glutaredoxin 1	1.666	−0.106	−1.772	H2O2	AH
gshB	Glutathione synthetase	0.768	−0.081	−0.848	Sal	AH
hdeA	Periplasmic acid chaperone	0.841	−0.326	−1.167	Sal	AH
hdeB	Periplasmic acid chaperone	0.782	−0.622	−1.404	Sal	AH
hisF	Cyclase component of IGP synthase	0.150	0.703	0.553	PQ−	BH
ibpB	Chaperone, HSP20 family	−1.691	−2.618	−0.928	H2O2	BL
katG	Catalase hydrogen peroxidase 1	−0.578	−0.313	0.265	H2O2	AL
lamB	Maltose high-affinity uptake	−4.881	−3.735	1.146	PQ	NH
lldP	l-Lactate permease	1.527	0.219	−1.307	Sal	NL
lysU	Lysyl tRNA synthetase, inducible	0.544	−0.077	−0.621	Sal	AH
malE	Maltose-binding protein, periplasmic	−3.425	−2.933	0.491	PQ	NH
malK	Maltose transport complex, ATP-binding subunit	−4.790	−3.748	1.041	PQ	NH
manX	PTS family, mannose-specific enzyme IIA component	0.000	0.517	0.517	Sal	AL
map	Methionine aminopeptidase	0.510	0.056	−0.454	PQ	AH
marA	Multiple antibiotic resistance	1.321	−0.516	−1.836	PQ, Sal	AH
marB	Regulator for mar	0.894	−0.172	−1.066	Sal	AH
marR	Repressor of mar	0.352	−0.437	−0.789	Sal	AH
mdaB	Drug activity modulator	0.444	−0.223	−0.667	Sal	AH
murF	d-Alanyl:d-alanine adding to cell wall	0.216	−0.170	−0.386	PQ	AH
nfnB	Nitrofurantoin resistance; nitroreductase	0.818	−0.204	−1.022	PQ, Sal	AH
nuoI	NADH dehydrogenase I subunit	0.792	0.310	−0.481	PQ	NL
nuoK	NADH dehydrogenase I subunit	0.766	0.460	−0.306	PQ	NL
ompF	Outer membrane porin	−2.834	−0.892	1.942	PQ−	AL
ompT	Outer membrane protease VII	−0.815	−1.204	−0.389	Sal−	BL
pdhR	Repressor of pdh	0.755	−0.109	−0.863	PQ	AH
pepN	Aminopeptidase N	0.671	0.118	−0.553	Sal	NL
pflB	Pyruvate formate lyase I (anaerobic)	0.812	0.578	−0.234	Sal	NL
pgi	Glucose phosphate isomerase	0.654	0.177	−0.477	PQ	NL
ptsG	PTS family IIC, glucose specific	0.841	0.554	−0.287	PQ	NL
putA	Proline dehydrogenase	−1.062	−0.353	0.708	Sal	AL
pyrB	Aspartate transcarbamylase	0.133	0.608	0.475	PQ−, Sal−	BH
sdhB	Succinate dehydrogenase	1.833	0.329	−1.505	PQ	AH
tnaA	Tryptophanase	−3.805	0.223	4.028	H2O2	AL
treB	Tre-specific PTS enzyme II	−3.156	0.739	3.895	Sal−	AL
yahA	Putative repressor	−0.667	0.422	1.089	PQ−	AL
yaiA	Function unknown	1.154	0.416	−0.738	H2O2	NL
ybjC	Function unknown	0.463	−0.447	−0.910	PQ, Sal	AH
ycfR	Function unknown	0.537	−0.819	−1.356	H2O2	AH
yfiA	Stabilizes ribosome against dissociation	−1.782	0.581	2.364	PQ−, Sal, H2O2	AL
yggJ	Function unknown	0.576	−0.178	−0.755	Sal	AH
yqjD	Function unknown	−0.347	0.378	0.725	Sal	AL
yncE	Function unknown	−0.476	0.331	0.808	PQ, Sal	AL

^aOxidative response is based on data in references 54 and 84. Induction was by H₂O₂, paraquat (PQ), or sodium salicylate (Sal). Repression is indicated by minus sign (Sal−, PQ−).

^bValues in boldface indicate significance (α = 0.001).

^cAH, Acid High; AL, Acid Low; BH, Base High; BL, Base Low; NH, Neutral High; NL, Neutral Low.

^dGAPDH, glyceraldehyde-3-phosphate dehydrogenase.

^ePTS, phosphotransferase.

Of the 73 pH-dependent genes known to be induced by H₂O₂, paraquat, or salicylate, virtually all were induced by acid or repressed by base. This finding confirms our hypothesis of a strong connection between acid stress and oxidative stress. It may be that low pH amplifies the toxicity of oxygen radicals. Genes repressed by paraquat or salicylate were repressed in acid or induced at high pH, such as the base-inducible membrane protein gene alx, the histidine cyclase gene hisF, and outer membrane protein gene ompF. An exception to these generalizations was the maltose regulon (lamB, malE, and malK), which was repressed by acid but induced by paraquat.

Envelope and periplasmic stress.

A large part of E. coli function takes place in the outer membrane and envelope (48) and the periplasm (49), compartments essentially exposed to “extracellular” pH. Thus, it is not surprising that several envelope and periplasmic components show pH-dependent expression (16, 23, 73, 82). Our microarrays revealed an even greater number of such responses (Table 6). Both acid and base induction were observed. Acid-induced periplasmic proteins included the well-known acid chaperone from hdeAB (23), as well as the newly observed TolA-binding protein (ybgF) and the lipoprotein from pal. High pH induced the ferric transporters from fecAB and fhuD, possibly due to low iron solubility at high pH. At high pH, various transport proteins and redox modulators such as that from dsbA are known to be induced. In addition, several additional base-induced periplasmic and envelope proteins appeared, including the vitamin B₁₂ transporter from btuB, the outer membrane protein from nmpC, and the peptidylprolyl-cis-trans-isomerase from ppiA.

TABLE 6.

Envelope and periplasmic genes

Gene	Function	Log2 pH ratioa			Classb
		5/7	8.7/7	8.7/5
artI	Periplasmic binding protein of Arg transport system	0.286	0.655	0.369	BH
artM	Arginine periplasmic binding protein	−0.115	0.293	0.407	BH
btuB	B-12 transporter, outer membrane receptor	−0.667	0.037	0.704	AL
cirA	Colicin I receptor production	−0.885	0.260	1.145	AL
clpX	ATPase subunit of ClpXP protease	−0.361	−0.625	−0.264	BL
cpxA	Periplasmic stress sensor (CpxAR)	−0.461	0.230	0.691	AL
cpxP	CpxAR-regulated periplasmic stress protein	−3.232	1.002	4.234	AL
cpxR	Periplasmic stress response regulator (CpxAR)	−0.662	0.590	1.251	AL
dsbA	Thiol:disulfide interchange, periplasmic	0.086	1.117	1.031	BH
dsbC	Disulfide bond isomerase, periplasmic chaperone	−0.288	−0.534	−0.246	BL
fadL	Fatty acid transport, outer membrane	−0.861	1.069	1.931	BH
fecA	Outer membrane ferric citrate receptor	−2.608	−1.128	1.480	NH
fecB	Periplasmic ferric citrate-binding protein	−2.842	−1.538	1.304	NH
fepA	Ferrienterobactin outer membrane receptor	−0.966	−0.289	0.676	NH
fhuD	Ferric hydroxamate binding protein; hydroxamate-dependent iron uptake	−0.767	−0.180	0.587	AL
fliY	Cystine-binding protein, periplasmic	0.233	−0.252	−0.484	AH
hdeA	Acid periplasmic chaperone	0.841	−0.326	−1.167	AH
hdeB	Acid periplasmic protein	0.782	−0.622	−1.404	AH
hisJ	High-affinity histidine-binding protein	−0.149	0.400	0.549	BH
hlpA	Periplasmic chaperone for OMPsc	0.099	−0.661	−0.759	BL
lamB	Maltoporin, maltose high-affinity uptake; phage lambda receptor	−4.881	−3.735	1.146	NH
lon	DNA-binding, ATP-dependent protease	−0.600	−1.666	−1.065	BL
malE	Maltose-binding protein, periplasmic	−3.425	−2.933	0.491	NH
malM	Maltose operon periplasmic protein	−4.643	−3.780	0.863	NH
mltB	Membrane-bound murein hydrolase	−0.739	−0.419	0.320	NH
nmpC	Outer membrane	−4.012	−0.051	3.961	AL
ompF	Outer membrane porin protein 1a	−2.834	−0.892	1.942	AL
ompT	Outer membrane protease VII	−0.815	−1.204	−0.389	BL
ompX	OMP, induces RNAP-sigma E	1.523	0.439	−1.083	NL
oppA	Periplasmic oligopeptide binding protein	1.358	0.406	−0.953	NL
pal	Lipoprotein associated with peptidoglycan	0.532	0.379	−0.153	NL
potD	Spermidine-binding membrane protein; regulates pot	0.053	0.509	0.456	BH
ppiA	Rotamase; peptidylprolyl-cis-trans-isomerase A	−0.526	0.231	0.757	AL
pstS	High-affinity, periplasmic phosphate binding protein	−2.479	−2.262	0.217	NH
rbsB	d-Ribose binding protein, periplasmic	−0.749	−1.150	−0.401	BL
rseB	Periplasmic, binds RseA; enhances RpoE-RseA cytoplasmic complex formation	0.631	−0.105	−0.736	AH
secD	SecDF-YajC inner membrane secretion complex	−0.165	0.287	0.452	AL
surA	Periplasmic outer membrane porin chaperone, stationary phase	0.310	0.209	−0.101	NL
tatA	Twin arginine translocation	0.043	0.679	−0.636	BH
tatB	Twin arginine translocation	0.075	0.446	0.372	BH
tolB	Group A colicin uptake and tolerance	0.498	0.421	−0.077	NL
tpX	Thiol peroxidase, antioxidant	0.637	0.178	−0.459	NL
tsx	Phage T6, colicin K resistance; nucleoside channel	−0.707	−0.904	−0.196	BL
ybgF	TolA-binding periplasmic protein	0.312	−0.007	−0.318	AH
yceI	Function unknown; periplasmic protein	−0.036	1.038	1.074	BH
yhcN	Periplasmic protein	3.064	−1.136	−4.199	AH

^aValues in boldface indicate significance (α = 0.001).

^bAH, Acid High; AL, Acid Low; BH, Base High; BL, Base Low; NH, Neutral High; NL, Neutral Low.

^cOMPs, outer membrane proteins.

^dRNAP, RNA polymerase.

Universal stress and heat shock.

Various heat shock and universal stress proteins are inducible by the permeant acid benzoate, such as the products of clpB, htpG, dnaK, groS, and uspA (38). Some of these showed pH response in our microarrays (Table 7). The DNA damage response gene uspD was acid induced, as was dps, encoding the DNA-binding protein involved in stationary phase and acid resistance. Acid induced rseAB, the antisigma regulators of the rpoE envelope heat stress system (1). High pH induced the rpoH heat shock sigma 32 gene (28) as well as heat shock proteasome genes hslUV and regulators hslOR.

TABLE 7.

Universal stress and heat shock response genes

Gene	Function	Log2 pH ratioa			Classb
		5/7	8.7/7	8.7/5
ahpC	Alkyl hydroperoxide reductase	1.003	0.436	−0.568	NL
ahpF	NAD(P)H:peroxiredoxin oxidoreductase	0.777	−0.222	−0.999	AH
cfa	Cyclopropane fatty acid synthase; acid resistance in stationary phase	2.075	−0.480	−2.555	AH
clpB	ClpB protease, ATP-dependent chaperone	−0.219	−1.963	−1.744	BL
cysK	Cysteine synthase, o-acetylserine sulfhydrylase A	1.204	1.351	0.147	NL
cysZ	Unknown function	0.440	0.212	−0.228	NL
dinI	Inhibits RecA coprotease	0.132	0.568	0.436	BH
dinJ	Induced by DNA damage	0.683	1.192	0.508	BH
dnaJ	DnaK cochaperone	−0.908	−2.155	−1.247	BL
dnaK	HSP-70-type molecular chaperone	−0.894	−2.249	−1.356	BL
dps	Stress response DNA-binding protein	1.130	0.105	−1.025	NL
grpE	Nucleotide exchange factor for DnaKJ	−0.491	−1.141	−0.650	BL
grxA	Glutaredoxin 1	1.666	−0.106	−1.772	AH
hdeA	Acid periplasmic chaperone	0.841	−0.326	−1.167	AH
hdeB	Acid periplasmic chaperone	0.782	−0.622	−1.404	AH
hslJ	Heat-inducible novobiocin resistance	0.368	−0.676	−1.044	AH
hslU	Heat-inducible ATP-dependent protease	−0.745	−1.688	−0.946	BL
hslV	Heat-inducible ATP-dependent protease	−1.125	−1.913	−0.788	BL
ibpB	Heat-inducible chaperone, HSP20 family	−1.691	−2.618	−0.928	BL
hslO	Hsp33, cytoplasmic heat shock chaperone activated by disulfide bond formation	−1.453	−1.737	−0.284	BL
hslR (yrfH)	Hsp15, heat shock, binds RNA and DNA	−1.773	−1.940	−0.168	NH
katG	Catalase-hydrogen peroxidase I	−0.578	−0.313	0.265	AL
rpoE	RNAPc sigma E, envelope heat stress	0.310	−0.427	−0.738	AH
rpoH	RNAP sigma 32, heat shock regulons	−0.378	0.084	0.462	AL
rseA	Anti-RpoE sigma factor, spans inner membrane	0.419	−0.159	−0.578	AH
rseB	Binds periplasmic domain of anti-RpoE sigma RseA	0.631	−0.105	−0.736	AH
sodB	Superoxide dismutase, Fe; acid inducible	0.752	0.280	−0.472	NL
ycdB	Function unknown, peroxidase homolog	0.654	−0.329	−0.983	AH
ycdO	Acid inducible, function unknown	0.842	−0.625	−1.467	AH
uspD (yiiT)	UV resistance	1.333	−0.497	−1.830	AH

^aValues in boldface indicate significance (α = 0.001).

^bAH, Acid High; AL, Acid Low; BH, Base High; BL, Base Low; NH, Neutral High; NL, Neutral Low.

^cRNAP, RNA polymerase.

DISCUSSION

Overall, our work revealed a large number of genes not previously known to be regulated by pH. Furthermore, many of these genes had no previously known function or response, such as yhcN and yagU (induced by acid) and yifO and ymcG (induced by base).

An important question is to assess the biological relevance of the expression ratios reported (36, 51). Most of the ratios we reported as significant (boldface in Tables 3 through 7) are greater than twofold (log₂ = 1). In many cases, all or most members of an operon fell in the same cluster and show similar expression profiles; the flagellar regulon was particularly consistent (Table 3). The gene probes are synthesized on the array independently of their operon map; thus, parallel expression profiles within operons do not reflect array position. Note that even genes with significant expression ratios of less than 2 (log₂ = 1) tend to group with their operons. In previous studies, comparison with quantitative reverse transcriptase real-time PCR shows that microarray ratios, while quantitatively consistent, generally underestimate the actual differences in mRNA levels between the biological systems compared (83).

Flagellar biosynthesis and motility.

The effects of pH in flagellar biosynthesis and motility remain poorly understood. It has long been known that low external pH (thus, large ΔpH) contributes to the proton motive force that drives flagellar rotation (33). The cytoplasmic pH, however, must remain high; permeant acids such as acetate and benzoate, which depress internal pH and decrease proton motive force, are chemotactic repellents (67) and impair rotation of the flagellar motor (46). Low pH elicits negative chemotaxis (55, 67), whereas a pH increase up to 8.3 elicits a positive response (55).

In recent reports acid stress is associated with low motility (72), yet acetate has been reported to induce the flagellar regulon and enhance motility (53). We believe that the previous reports are limited in several ways. Reference 72 does not compare pH conditions directly but notes repression of flagellar genes in an hns mutant in which acid resistance is increased. The motility assay is not clearly described, and the acid dependence of flhDC-cat expression was observed on plasmids, not in the genome. Reference 53 reports induction of chromosomal flhDC-lacZ fusions by acetate. Those authors' assays of motility, however, show relatively small differences between pH conditions.

Our microarrays showed strong evidence for suppression of motility and chemotaxis at high pH. This evidence was supported by the decrease in motility at high pH, observed for both E. coli and S. enterica serovar Typhimurium, which swims twice as fast as E. coli. We also found weaker evidence for repression of che and mot genes at pH 5, but the flagellar synthesis genes were strongly induced at low pH. Overall, our data point to alkaline suppression of flagellar motility. Work in progress shows that, at high pH, the number of flagella per cell is decreased to one to three per cell (about 20% of normal) (S. Aizawa and J. Slonczewski, unpublished data).

No pH dependence was observed for the flagellar regulators flhD and flhC. On the other hand, in a microarray study of anaerobic cultures, flhD and flhC are induced by acid (E. Hayes and J. L. Slonczewski, unpublished data). Acid induction of these regulators would be consistent with the report of their induction by acetate (53). We did see acid induction of two known activators of flhDC: adenylate cyclase cyaA (37) (Acid High) and dnaK-dnaJ-grpE (64) (Base Low). We saw no acid induction of other flagellar activators such as crp (37), nor did we see alkaline induction of the negative flagellar regulator rcsCDB (22).

An alternative model is that pH regulation of the flagellar regulon is mediated by proteolysis, as in the case of ClpXP proteolysis of FlhD and FlhC (76). We find that ClpX is down-regulated at high pH (Base Low cluster), but a different protease could be involved.

Catabolism.

The picture of catabolism is more complicated, but in general our expression ratios confirm our present hypotheses of pH regulation while extending our knowledge to many more components. Systems that consume acids are enhanced at low pH. On the other hand, initial import and breakdown of some sugars, such as maltose, are favored at high pH, where they may quickly generate a large burst of fermentation acids.

With respect to proton export, E. coli appears to prefer components such as ATP synthase that import protons at high pH (counteracting the alkaline stress on cytoplasmic pH) and prefers to minimize proton export associated with the terminal oxidase cyd in preference to cyo. This observation is consistent with the previous report that cyd expression is higher at pH 7.5 than at pH 5.0 in an fnr mutant (14), although in those experiments cyo expression also increased with pH. It is likely that our broader range of pH classes (up to pH 8.7) provided a clearer picture of pH regulation of cyo and cyd.

Under amino acid catabolism, relatively few new components of pH response were observed. This makes sense, because most amino acid decarboxylases are repressed by oxygen (4, 68), as are deaminases such as sdaA (82). In preliminary experiments, we have repeated our microarray study on cultures grown anaerobically. Under anaerobiosis, several amino acid decarboxylases and deaminases show pH-dependent expression (Hayes and Slonczewski, unpublished).

Stress responses.

Several stress responses are known to interact with pH stress and pH resistance, including oxidative stress, heat shock, and envelope stress (for reviews see references 21 and 68). The overlap with salicylate stress could be explained in part by salicylate's effect as a permeant acid, stressing internal pH (60). The mar drug resistance operon is known to be coinduced by aromatic permeant acids and low pH (69) under regulation by MarR as well as by the superoxide regulator SoxRA (57).

Beyond salicylate, however, a large number of oxidative stress genes inducible by H₂O₂ or by paraquat showed significant pH-dependent expression, nearly all induced by acid or repressed by base. This finding confirms our hypothesis of a strong connection between acid stress and oxidative stress. Since so much of aerobic respiration is stepped up at pH 5, including cytochrome o oxidase, it is likely that acid conditions accelerate the production of oxygen radicals, thus inducing a partial oxidative stress response.

Various envelope and periplasmic stress responses are induced by acid, contributing to acid resistance; the best characterized in terms of mechanism is the acid-induced periplasmic chaperone HdeA (23). Extracellular acid induces a dimer-to-monomer transition in HdeA, which then suppresses aggregation by acid-denatured proteins. Our study reveals additional potential contributors to acid resistance and base resistance, including genes of unknown function such as yhcN, induced by acid, and yceI, induced by base.

Our study presents the most comprehensive picture to date of acid and base response by E. coli grown aerobically in complex medium. Overall, low pH accelerates acid consumption and proton export, while coinducing oxidative stress, possibly through increased production of oxygen radicals. High pH accelerates proton import while repressing the energy-expensive systems of flagellar biosynthesis and chemotaxis. Finally, pH differentially regulates a large number of periplasmic and envelope stress systems, as well as transporters, chaperones, and redox regulators.