ABSTRACT
The ability of Escherichia coli and Bacillus subtilis to regulate their cytoplasmic pH is well studied in cell suspensions but is poorly understood in individual adherent cells and biofilms. We observed the cytoplasmic pH of individual cells using ratiometric pHluorin. A standard curve equating the fluorescence ratio with pH was obtained by perfusion at a range of external pH 5.0 to 9.0, with uncouplers that collapse the transmembrane pH difference. Adherent cells were acid stressed by switching the perfusion medium from pH 7.5 to pH 5.5. The E. coli cytoplasmic pH fell to a value that varied among individual cells (range of pH 6.2 to 6.8), but a majority of cells recovered (to pH 7.0 to 7.5) within 2 min. In an E. coli biofilm, cells shifted from pH 7.5 to pH 5.5 failed to recover cytoplasmic pH. Following a smaller shift (from pH 7.5 to pH 6.0), most biofilm cells recovered fully, although the pH decreased further than that of isolated adherent cells, and recovery took longer (7 min or longer). Some biofilm cells began to recover pH and then failed, a response not seen in isolated cells. B. subtilis cells were acid shifted from pH 7.5 to pH 6.0. In B. subtilis, unlike the case with E. coli, cytoplasmic pH showed no “overshoot” but fell to a level that was maintained. This level of cytoplasmic pH post-acid shift varied among individual B. subtilis cells (range of pH, 7.0 to 7.7). Overall, the cytoplasmic pHs of individual bacteria show important variation in the acid stress response, including novel responses in biofilms.
INTRODUCTION
Neutralophilic bacteria, such as Escherichia coli and Bacillus subtilis, maintain a cytoplasmic pH within a narrow range, even when exposed to a wide range of extracellular pHs in the environment (16, 25, 28, 30, 39). But bacterial pH is nearly always observed in isotropic cell suspensions and rarely in adherent cells (2, 3). Adherent cells offer the opportunity to track a single cell's response to pH changes over time, which cannot be done either in suspension or through fluorescence-activated cell sorting (FACS) analysis. The single-cell response is important because bacterial cells age differently as they divide, yielding a physiologically diverse population that would be expected to vary in abilities such as pH homeostasis (33). The adherence of cells is important because in natural environments and in medical settings, many bacteria grow as biofilms or communities attached to a surface (1, 9). Biofilms of E. coli O157:H7 (27) and of Streptococcus mutans (38) show altered resistance to acid stress. Bacterial biofilms also show changes in properties related to acid, such as antibiotic resistance (13) and iron requirements mediated by the acid/iron regulator Fur (10). Thus, it is of interest to observe the cytoplasmic pH homeostasis of cells within biofilms and pursue their differences from planktonic cells.
In suspension, planktonic E. coli cells maintain a cytoplasmic pH from pH 7.2 to 7.8 while growing over an external pH range of 5.0 to 9.0 (29, 30, 39). The Gram-positive bacterium B. subtilis maintains a comparable degree of pH homeostasis while growing over a pH range of 6.0 to 9.0 (14, 25, 35). Both organisms respond to rapid external pH perturbation with an initial drop in cytoplasmic pH, followed by some degree of recovery. E. coli pH recovers to a value within 0.2 units of the original within 2 min (15, 36). Surprisingly, the molecular mechanisms of pH homeostasis in both model organisms remain poorly understood. In E. coli, the ability to maintain cytoplasmic pH during acid stress depends on the presence of potassium or other osmolytes in the medium (6, 15, 34). At a high external pH, cytoplasmic pH maintenance involves sodium/proton antiport (17, 20).
Most current methods of cytoplasmic pH measurement report average values from cell suspensions (14, 36), thus overlooking individual cell heterogeneity in a population (2). Single-cell analysis can reveal new evidence for previously untestable hypotheses, such as how individual cells vary in their vulnerability to pH stress both in isolation and as biofilms. Previously, ratiometric fluorophores have been used to measure pHs of individual bacteria (21, 22, 26), but few studies have used bacteria expressing a fluorescent protein. A highly effective pH reporter is the pH-sensitive derivative of green fluorescent protein (GFP) known as ratiometric pHluorin (18). Ratiometric pHluorin produces two excitation peaks: one that increases in intensity with rising pH and one that decreases with rising pH. The use of a ratiometric fluorescent protein has many advantages over other methods in that it is rapid and noninvasive and allows for continuous determination of intracellular pH in individual cells (19, 36).
Here we report the cytoplasmic pHs of adherent E. coli and B. subtilis cells over a range of external pHs using fluorescence microscopy with ratiometric pHluorin. We report for the first time the kinetic responses of individual bacteria to external acid shift and the response to acid shift of E. coli biofilms.
MATERIALS AND METHODS
Strains and plasmids.For cytoplasmic pH measurement, pH reporter plasmids were constructed to express the GFP derivative ratiometric pHluorin (18). As pH increases, ratiometric pHluorin shows increased excitation at 410 nm and decreased excitation at 470 nm.
For pH measurement in E. coli, we constructed pGFPR01, in which ratiometric pHluorin is expressed from the arabinose-induced promoter PBAD. For plasmid construction, the sequence encoding ratiometric pHluorin from pGM1 (GenBank accession no. AF058694.2), kindly provided by Gero Miesenböck and the Sloan-Kettering Institute for Cancer Research, was amplified with the primers 5′-GGCCGAATTCATGAGTAAAGGAGAAGAACTTTTCACTGG-3′ and 5′-GGCCAAGCTTTTATTTGTATAGTTCATCCATGCCATG-3′. The primers generate an insert with engineered start and stop codons and EcoRI and HindIII recognition sites. These two restriction enzymes were used to digest the PCR product and vector pBAD322, a low-copy-number plasmid with the arabinose-inducible PBAD promoter supplied by John Cronan (5). The linearized vector and insert were ligated and transformed into Top10 OneShot E. coli (TA cloning kit; Invitrogen), selecting on 50-μg/ml ampicillin LB plates with 0.2% l-arabinose. Colonies expressing pHluorin were detected by fluorescence at 410 nm. pGFPR01 was then transformed into E. coli W3110 (32), generating strain JLS1105.
For pH measurement in B. subtilis, pMMB1437 was constructed with ratiometric pHluorin expressed from the constitutive blasticidin S resistance promoter (Pbsr). The wild-type gfp gene from pBSVG101 was replaced with the gene encoding pHluorin (12). The primers 5′-CCTGTTCCATGGCCAACAC-3′ (inside the beginning of gfp) and 5′-GAGGAATTCTACGAATGCTATTTGTATAGTTCATCCATGCCATG-3′ (including an engineered stop codon and EcoRI recognition site) were used to amplify the 576-bp partial gfp sequence that encodes the ratiometric pHluorin mutations (E132D, S147E, N149L, N164I, K166Q, I167V, R168H, and S202H) from pGM1 (18). The PCR product was doubly digested with the restriction enzymes NdeI and EcoRI, generating a 497-bp insert. pBSVG101 was doubly digested with NdeI and EcoRI, generating a 5,318-bp vector that was then treated with Antarctic phosphatase and purified. The vector and ratiometric GFP insert were ligated and transformed into NEB 5-α E. coli on 100-μg/ml ampicillin LB plates. Plasmid pMMB1437 was purified from E. coli and transformed into B. subtilis AG174 (JH642 trpC2 pheA1) on 10-μg/ml tetracycline LB plates, producing strain MMB1440. The ratiometric GFP sequence was verified to match the GenBank sequence for pGM1. pMMB1437 was also transformed into E. coli K-12 W3110, producing strain JLS1013.
Bacterial culture and sample preparation for microscopy.E. coli strains JLS1013 (W3110/pMMB1437) and JLS1105 (W3110/pGFPR01) and B. subtilis strain MMB1440 (AG174/pMMB1437) were cultured for fluorescence microscopy in 2 ml LBK (10 g/liter tryptone, 5 g/liter yeast extract, 100 mM KCl) (36) with 0.2% l-arabinose and 50 μg/ml ampicillin for cells expressing pGFPR01 and in 2 ml LBK with 10 μg/ml tetracycline for cells expressing pMMB1437. E. coli bacteria were cultured either to stationary phase or to mid-log phase (approximate optical density at 600 nm [OD600] = 0.4) in LBK buffered with 50 mM 3-(N-morpholino)propanesulfonic acid (MOPS), pH 7.5, at 37°C, rotating at 260 rpm. Growth phase did not affect the pHluorin fluorescence of E. coli. For observations of B. subtilis, cells were grown to mid-log phase (OD600 = 0.4 to 0.5). The fluorescence level in B. subtilis was monitored to ensure optimal expression of the ratiometric GFP, pHluorin, without reaching extreme fluorescence intensities.
Cultures were resuspended in 1 ml M63 minimal medium [0.4 g/liter KH2PO4, 0.4 g/liter K2HPO4, 2 g/liter (NH4)2SO4, 7.45 g/liter KCl] supplemented with 2 g/liter casein hydrolysate (referred to as M63A) and buffered to the desired pH with a 50 mM concentration of the appropriate buffer [pH 5.0, homopiperazine-N,N′-bis-2-(ethanesulfonic acid) (HOMOPIPES); pH 5.5 to 6.0, 2-(N-morpholino)ethanesulfonic acid (MES); pH 6.5, piperazine-N,N′-bis(2-ethanesulfonic acid)(PIPES); pH 7.0 to 7.5, MOPS; pH 8.0 to 8.5, N-Tris(hydroxymethyl)methyl-3-aminopropanesulfonic acid (TAPS); and pH 9.0, N-(1,1-dimethyl-2-hydroxyethyl)-3-amino-2-hydroxypropanesulfonic acid (AMPSO)].
Resuspended cultures were spotted on 40-mm round coverslips that were coated with 0.01% alpha-poly(l-lysine) (Sigma-Aldrich). Prior to use, coverslips were immersed in 1 M HCl for 1 h, rinsed in deionized H2O, immersed in 2 ml 0.01% poly(l-lysine) solution for 1 h, rinsed briefly with deionized H2O, and left to air dry, creating a thin and uniform coating, similar to the “rinse” method (4). Immobilized cells were observed in a FCS3 flow cell (Bioptechs) with a chamber volume of 250 μl, perfused with supplemented M63A medium drawn through by gravity at a rate of approximately 1 ml/min.
Fluorescence ratio imaging.Fluorescence of adherent cells expressing pHluorin (18) was observed using excitation wavelengths of 400 to 425 nm (excitation increases with pH) and 460 to 480 nm (decreases with pH). The excitation ranges were defined using filters D410 and D470, respectively (Chroma Technology Corp.), contained by a Lambda 10-3 filter wheel on a xenon arc lamp (LB-LS/OF17; Sutter Instrument). The designated excitation ratio was 410/470. Emission was observed at 510 to 560 nm (filter HQ535). Microscopy was performed on an Olympus BX61WIF-5 microscope with an 100× oil-immersion objective.
Images and fluorescence intensities were recorded for both wavelengths using MetaFluor software (Molecular Devices) to calculate excitation intensity ratios (410/470). E. coli JLS1105 (W3110/pGFPR01) and B. subtilis MMB1440 (AG174/pMMB1437) were observed with the following settings: 250 gain and 1-binning. Exposure times for each wavelength were calibrated for each replicate. When calibrating exposure times, two factors were considered. Cells become overexposed at high exposure times, giving a fluorescence intensity of zero, which will give a nonexistent ratio. Generally, exposure times above 250 ms would overexpose cells. The second factor was maintaining the “fluorescence ratio-to-pH” ratio, where a fluorescence ratio of 1 was equal to approximately pH 7.1. For all replicates, observed E. coli JLS1105 (W3110/pGFPR01) exposure times were between 100 and 200 ms for each wavelength unless otherwise noted. E. coli JLS1013 (W3110/pMMB1437) cell images were acquired with an exposure time of 5 ms. In all studies, images were acquired with 410/470 nm fluorescence ratios and wavelength intensities.
To determine fluorescence ratios for an individual cell, each cell was defined and numbered using the trace and autotrace (length, 50; angle, 5; hole size, 3; without using threshold) functions of MetaFluor software (Molecular Devices). Regions were defined within the edge of the visible cell to ensure that the average pixel ratios were of similar intensities. Cells containing pixels with intensities lower than 500 were excluded from cell counts because ratios at these intensities are unreliable on the standard curve, and those with pixel intensities greater than 4,000 were excluded due to overexposure.
Cytoplasmic pH measurement.Standard curves for pH as a function of fluorescence ratio were obtained separately for E. coli and for B. subtilis. The curve was generated with 410/470 values, using cells in which the difference between the external pH and the internal pH (ΔpH) was collapsed to equalize the pH of the medium with the cytoplasmic pH. For E. coli, the transmembrane ΔpH was collapsed by the addition of 40 mM potassium benzoate and 40 mM methylamine hydrochloride. For B. subtilis, 40 mM methylamine and 10 μM nigericin, a K+/H+ antiporter ionophore (23), were added. A Boltzmann sigmoid best-fit curve was applied to the standard curve ratio data, assuming that the difference in excitation levels follows a titration curve for the protonation-deprotonation of the ratiometric pHluorin (18). The standard curve was highly reproducible and showed no effect of growth phase of the deenergized cells. Nonetheless, the curve was repeated regularly, especially following any change in the light source, filters, optics, or fluorescence measurement parameters; also, the curve was obtained again for each bacterial strain observed. Any use of the fluorescence ratio method should include publication of a standard curve.
The time course of response to acid shift was observed in adherent cells from 16-h overnight cultures or from late-log-phase cultures (OD600 = ∼0.7) diluted 200-fold from overnight cultures. Isolated E. coli cells were subjected to acid stress by switching the flow cell inlet line from supplemented M63A buffered with 50 mM MOPS at pH 7.5 to supplemented M63A buffered with 50 mM MES at pH 5.5 (biofilms and B. subtilis cells were switched to M63A buffered with 50 mM MES at pH 6.0). For pH shift experiments, images and fluorescence intensities were recorded approximately every 5 s for the initial 60 s of acquisition before adjusting to 30-s increments for the remaining 3 to 4 min, unless noted otherwise. Fluorescence ratios (410/470) were recorded, and the cytoplasmic pH was calculated according to the standard curve.
Biofilm growth for microscopy.A poly-(l-lysine)-coated coverslip was spotted with 10 μl of a 14-h overnight culture of JLS1105 (W3110/pGFPR01). The coverslip was placed in a FCS3 flow cell chamber (Bioptechs). LBK buffered at pH 7.0 with 100 mM MOPS with 0.2% l-arabinose and 50 μg/ml ampicillin was perfused through the chamber overnight (14 h) at a rate of approximately 5 ml/h using the P720 peristaltic pump (Instech) and a tube set with an inner diameter of 0.381 mm. Biofilm culture was carried out at 37°C in an incubator. Fluorescence was observed as described above.
RESULTS
Ratiometric pHluorin expressed in E. coli and in B. subtilis.We constructed strains for expression of ratiometric pHluorin as a measure of cytoplasmic pH and observed the cells to ensure normal morphology. In E. coli, cells containing the plasmid pGFPR01 with pHluorin expression under PBAD showed normal size and morphology under phase-contrast microscopy and under fluorescence (Fig. 1A and B). We also explored the use of pMMB1437 as an E. coli/B. subtilis shuttle plasmid that expresses pHluorin from a constitutive promoter, Pbsr. E. coli cells carrying pMMB1437 showed 40-fold-higher fluorescence intensities than those carrying pGFPR01 (Fig. 1). Thus, comparable fluorescence intensities were observed with an exposure time of 5 ms for pMMB1437 (Fig. 1D) and 200 ms for pGFPR01 (Fig. 1B). However, E. coli cells containing pMMB1437 formed dense inclusion bodies that appeared phase bright and nonfluorescent (Fig. 1C and D). The inclusion bodies may be composed of misfolded pHluorin, a common problem associated with overexpression of protein-encoding genes (11). For all cytoplasmic pH measurements in E. coli, pGFPR01 was used.
E. coli cells expressing GFP from PBAD (A and B) or from Pbsr (C and D). Mid-log-phase cultures of JLS1105 (W3110 pGFPR01; A, B) and JLS1013 (W3110 pMMB1437; C and D) were resuspended in M63A medium at pH 6.5 containing 40 mM benzoate. Microscopy was conducted as described under Methods. Phase contrast images (A and C) and 410/470 excitation ratio images (B and D) were recorded for each strain. Cultures carrying pMMB1437 with GFP under the constitutive Pbsr promoter formed nonfluorescent inclusion bodies (arrow).
In B. subtilis, pMMB1437 did not cause inclusion body formation, and the cell fluorescence intensities were similar to that of pHluorin produced by pGFPR01 in E. coli. Thus, for cytoplasmic pH measurement in B. subtilis, pMMB1437 was used.
pH dependence of pHluorin fluorescence in adherent E. coli cells.To correlate fluorescence with cytoplasmic pH, E. coli cells were observed in the presence of a permeant acid (benzoic acid) and a permeant base (methylamine) (Fig. 2). The permeant acid collapses the transmembrane proton gradient at low external pH, whereas the permeant base collapses the inverted pH gradient at high external pH. Thus, in the presence of both agents, the cytoplasmic pH becomes equal to the external pH in both acid and base ranges.
Ratiometric fluorescence microscopy of E. coli JLS1105: standard curve and pH homeostasis. E. coli JLS1105 was cultured to mid-log phase as described in Materials and Methods. (A) Bacteria were resuspended and observed in M63A medium containing 40 mM benzoate and 40 mM methylamine to collapse ΔpH. The false-color scale bar represents 410/470 excitation ratios. Representative images are shown for cells in which ΔpH is collapsed (standard curve). (B) Bacteria were resuspended and observed in M63A (green line), and following addition of benzoate and methylamine to collapse ΔpH (black squares). From these data, the standard curve was derived using the Boltzmann equation (black line). (C) Cytoplasmic pH values were calculated based on the standard curve. Error bars represent SEM (n = 20 random cells); bars are not shown when smaller than the symbol.
Cells were initially suspended in buffered M63A medium and immobilized on poly(l-lysine)-coated coverslips. The adherent cells were placed in a flow cell that was perfused with M63A medium buffered at various values of pH, and the 410/470 fluorescence excitation ratios were obtained (Fig. 2B). The perfusion medium was then switched to a medium containing benzoate and methylamine, and further 410/470 ratios were obtained. Thus, in Fig. 2B the data are presented as pairs of measurements with and without benzoate/methylamine for the same culture. Fluorescence ratios (with or without benzoate/methylamine) were found to be consistent among biological replicates.
Representative images of cells in the presence of benzoate and methylamine, where the ΔpH was collapsed, are shown in Fig. 2A, visualized with a false-color scale bar calibrated for fluorescence ratios. The standard curve of the fluorescence ratio versus the pH of the permeabilized cells indicates that the effective range of pH measurement was between pH 5.6 and 9.0 under our conditions. The fluorescence properties of pH-sensitive GFP variants are stable from pH 5.5 to pH 10.0 (18); however, the pHluorin excitation shows pH dependence only in the range of pH 5.5 to pH 8.5 (18, 19, 36). The standard curve is approximately sigmoidal, with a nearly linear section where most pH data were collected.
We then used the standard curve to measure cytoplasmic pHs of cells maintaining pH homeostasis. Fluorescence ratios of cells maintaining pH homeostasis (Fig. 2B, green line) were converted to pH values (Fig. 2C) using the Boltzmann equation of the standard curve (Fig. 2B, black line). In supplemented M63A minimal medium buffered at pH 5.0 to 8.0, E. coli maintained a cytoplasmic pH within a range of approximately pH 7.6 to 7.8; most cytoplasmic pH values in neutral to acidic external pH were close to the expected value of pH 7.6 (28). Above external pH 8.0, however, E. coli did not maintain a cytoplasmic pH lower than that of the external medium. This finding differs from the longstanding observation of inverted ΔpH in suspended cells (30, 36). An average of 20 cells yielded a standard error of the mean of 0.1 pH units. These results were consistent and repeatable among biological replicates.
Photobleaching alters fluorescence ratios over time.GFP-derived fluorophores generally show some photobleaching over time (24), an effect that could distort our pH measurements. We observed photobleaching of the GFP fluorophore exposed with rapid 5-s intervals between fluorescence measurements for each wavelength, at 50/70-ms exposure times for 410/470 wavelengths, respectively (Fig. 3). For cells set at cytoplasmic pH 5.5, the 410-nm excitation signal decreased more slowly than the 470-nm signal. At pH 7.5, the 410 signal decreased faster than that at the 470 wavelength. The resulting ratio values also showed an upward slope at pH 5.5 but a downward slope for pH 7.5 over the 4-min period.
In vivo GFP photobleaching alters fluorescence ratios with a 5-s acquisition interval in E. coli. JLS1105 was cultured overnight to stationary phase for 16 h and resuspended in M63A medium containing 40 mM benzoate, 40 mM methylamine, and a buffer: 50 mM MES adjusted to pH 5.5 (A) or 50 mM MOPS adjusted to pH 7.5 (B). Fluorescence intensities for 410 nm (purple lines) and 470 nm (blue lines) were recorded, and 410/470 ratios (black lines) were calculated for 25 and 27 individual cells, in pH 5.5 and pH 7.5, respectively. Average wavelength intensities and ratios for 50/70-ms exposure for 410/470 wavelengths were calculated and plotted over time.
For subsequent experiments, the effect of photobleaching was minimized by increasing the time interval between measurements during the recovery phase of the pH response (presented below). The longer time intervals (30 s instead of 5 s) meant that photobleaching had an insignificant effect on pH measurements during observation of pH recovery (from 1 to 7 min).
Individual E. coli cells respond to acid shift.We measured the response of E. coli cytoplasmic pH to a shift in external pH. E. coli cells adhered to poly(l-lysine)-coated coverslips were exposed to a rapid acid shift by switching the flow cell source medium from pH 7.5 to 5.5 (Fig. 4). A majority of cells showed a temporary loss of pH homeostasis, with an average dip of 1.1 pH units. The dip was followed within 1 to 2 min by a nearly full recovery, to within 0.4 units of the original value of cytoplasmic pH. The magnitude of the initial dip in pH varied among individual cells, within a range of about 0.5 pH unit. Most individual cells maintained pH homeostasis, as reported for cell suspensions (36), but a significant number failed to recover after their pH declined to a value equal to the external pH of 5.5. The percentage of cells failing to recover varied among samples on different days, from less than 2% to as high as 23%.
Cytoplasmic pHs of individual E. coli cells during a shift from external pH 7.5 to pH 5.5. Stationary-phase cultures of JLS1105 (W3110/pGFPR01) were resuspended in M63A medium–50 mM MOPS (pH 7.0) and adhered to poly(l-lysine)-coated coverslips. M63A medium–50 mM MOPS (pH 7.5) was perfused, and 410/470 ratios were recorded for at least 10 s before the medium source was switched to M63A–50 mM MES (pH 5.5). Fluorescence images were acquired with a 200-ms exposure time for each wavelength every 5 s for no longer than 1 min following the external pH shift. Images and ratios were acquired every 30 s thereafter. Individual cell cytoplasmic pHs were calculated from fluorescence ratios in accordance with the standard curve. Representative images of cells at three time points are presented. Different colored lines indicate individual cells representative of the population.
Biofilm cells showed pH homeostasis.Cells grown in a biofilm express different genes and show many physiological differences from suspended planktonic cells (37). Their pH resistance is likely to show differences as well. We tested the ability of E. coli grown in a biofilm to maintain pH homeostasis after being subjected to a shift in external pH. Biofilms were grown overnight for 14 h as described in Materials and Methods.
During microscopy, a biofilm sample was subjected to a shift in external pH from pH 7.5 to pH 6.0. Representative fields of cells are shown in Fig. 5, and representative cytoplasmic pH traces are shown in Fig. 6. Most cells within the biofilm showed a sharp drop in cytoplasmic pH after the initial pH shift, with an average drop of 1.0 ± 0.5 units (Fig. 5C and 6A). The biofilm cells recovered their pH more slowly than the individual adherent cells. After 6 min, a majority of cells recovered their cytoplasmic pH at or near the original value (Fig. 5D and 6A). The proportion of cells whose pH recovered varied among biofilm samples.
Cytoplasmic pH within an E. coli biofilm. Biofilms were grown overnight (14 h) from strain JLS1105 in LBK medium, 100 mM MOPS (pH 7.0). During microscopy, the biofilm was perfused with M63A medium buffered with 50 mM MOPS (pH 7.5) drawn through by gravity at a rate of approximately 1 ml/min. The flow cell inlet was then switched to M63A buffered with 50 mM MES (pH 6.0). Images and fluorescence intensities were recorded approximately every 5 s for the initial 1 min of acquisition before adjusting to 30-s increments for 4 to 6 min. Phase-contrast images of biofilms were acquired (A). Ratiometric images were acquired during perfusion with M63A medium–50 mM MOPS (pH 7.5) (B), following external shift to pH 6.0 (C), up to 6 min following the switch (D), or following the introduction of 40 mM benzoate (E) to collapse the ΔpH of the cells. The arrow in panel D marks a cell that failed to recover cytoplasmic pH. Scale bar = 5 μm.
Cytoplasmic pHs of individual E. coli cells in a biofilm during a shift from external pH 7.5 to pH 6.0 show a range of recovery responses. Individual cells within the biofilm (grown overnight for 14 h) displayed a range of responses to a drop in external pH. A majority of cells displayed complete cytoplasmic pH recovery (A), with less than 3% failing to recover cytoplasmic pH by the end of the time course (B). Fluorescence intensity ratios were recorded at an external pH of 7.5 for a minimum of 30 s before switching to an external pH of 6.0 (indicated by an arrow). Fluorescence ratios were recorded at 30-s intervals for about 6 min after acid introduction to allow for recovery of pH homeostasis before introduction of 40 mM benzoate (indicated by an arrow) to collapse the ΔpH of cells within the biofilm. More than 75 cells within the biofilm were defined, and their 410/470 nm fluorescence ratio values were determined over time. The ratios were converted to cytoplasmic pH in accordance with the standard curve. Representative cell pH values are shown.
The observation of pH recovery was confirmed by the addition of benzoate, where all observed cells displayed a collapsed ΔpH (Fig. 5E and 6). Before benzoate treatment, the biofilm cells recovered to a range of pH 6.2 to 7.2, whereas isolated adherent cells (Fig. 4) recovered cytoplasmic pH to values of pH 7.0 to 7.5. Within the biofilm (Fig. 6A), individual cells recovered more slowly from acid shift than did the separated adherent cells (Fig. 4).
In the sample shown for Fig. 5 and 6, more than 97% recovered (Fig. 5D and 6A) and less than 3% of cells failed to recover cytoplasmic pH (Fig. 6B). In other biofilm trials (not shown), up to 35% of cells failed to recover; the number of cells failing to recover varied among biofilms. Cells that failed to recover were located at separate positions in the biofilm (Fig. 5D). Some of them actually began to recover pH (Fig. 6B) and then collapsed at various times; representative examples are shown. This unusual pattern of aborted recovery was never seen outside the biofilm, in adhered planktonic cells.
Cytoplasmic pH in B. subtilis.For comparison with E. coli, pHluorin 410/470 excitation ratios were used to measure cytoplasmic pH in the Gram-positive neutralophile B. subtilis. To obtain a pH conversion curve, poly(l-lysine)-adherent cells containing pMMB1437 were observed with ΔpH collapsed using nigericin and methylamine (Fig. 7). Separate cultures were assayed with or without nigericin-methylamine because of difficulties involved with eliminating the ionophore nigericin from the flow cell between replicates. Representative images of individual B. subtilis cells in the presence of nigericin-methylamine where ΔpH was collapsed are presented in Fig. 7A, visualized with a false-color scale bar calibrated for fluorescence ratios. As for E. coli, the standard curve values (in which ΔpH is collapsed) form a sigmoid curve (Fig. 7B), At external pH 5.5 to 7.5, individual B. subtilis cells maintained cytoplasmic pH from pH 7.5 to 7.9 (Fig. 7C). Under our conditions, B. subtilis maintained a cytoplasmic pH higher than or equal to the medium pH at values greater than 7.5.
B. subtilis ratiometric pHluorin standard curve and pH measurement. B. subtilis strain MMB1440 was cultured for the standard curve to mid-log phase (OD600 = ∼0.4) as described in Materials and Methods. Bacteria were resuspended and observed at an external pH of 5.5 to pH 8.5 and then with the same media containing 10 μM nigericin and 40 mM methylamine to collapse ΔpH. (A) Representative cell images in which the ΔpH is collapsed by nigericin and methylamine. The false-color scale bar represents fluorescence ratios. (B) Bacteria observed at external pH 5.5 to pH 8.5 (green line) and then with 10 μM nigericin and 40 mM methylamine (black squares). The Boltzmann fit curve is shown (black line). (C) Cytoplasmic pHs were measured based on the standard curve. Error bars represent SEM (n = 20 random cells) and are not shown when smaller than the symbols.
B. subtilis maintains cytoplasmic pH during an external pH shift.We investigated the ability of B. subtilis cells to maintain their cytoplasmic pH after external acid shift. The initial cytoplasmic pH values were pH 7.9 to 8.1, which is higher than those seen for E. coli perfused under the same conditions (Fig. 4). Adherent cells were perfused with M63A pH 7.5 medium before the pH was decreased to 6.0 (Fig. 8). After this shift, cytoplasmic pH decreased by 0.5 to 1.0 pH units (Fig. 8). Within 1 to 3 min, most cells had partly recovered their cytoplasmic pH to values of pH 7.2 to 7.8. Some B. subtilis cells showed only a small pH drop, without recovery (pink and yellow traces).
Cytoplasmic pHs of individual B. subtilis cells during a shift from external pH 7.5 to pH 6.0. B. subtilis strain MMB1440 was grown in LBK–100 mM MOPS (pH 7.0) overnight. Mid-log-phase cells were grown from overnight cultures to an OD600 of 0.4 and resuspended in M63A medium–50 mM MOPS (pH 7.0). The flow cell was perfused with M63A medium buffered with 50 mM MOPS (pH 7.5) for a minimum of 30 s before the perfusion medium was switched to M63A buffered with 50 mM MES (pH 6.0). Representative images of cells at three time points are presented. Different colored lines indicate individual cells representative of the population. Different colored lines indicate individual cells representative of the population. Scale bar = 5 μm.
DISCUSSION
We show that pH homeostasis varies among individual bacterial cells, both among planktonic cells adhered to a slide and among cells within a biofilm grown on a slide. This variability implies that the pH response and recovery measurements obtained for cell suspensions miss important aspects of pH homeostasis of individual cells. The variation we saw could be associated with substrate adherence and with variability among suspended cells. Properties of adherent cells are highly important for medical, environmental, and industrial applications.
To measure pH homeostasis, we used ratiometric GFP combined with fluorescence microscopy, both for E. coli (Fig. 4) and for B. subtilis (Fig. 8). Ratiometric pHluorin (18) combines the advantages of GFP fluorimetry (14, 15, 36) and fluorescence microscopy. Fluorimetry reports only the average response of cell populations, with no capacity to monitor individual cells. For single cells, cytoplasmic pH has been measured previously by ratio imaging with a fluorescent dye, such as carboxyfluorescein diacetate succinimidyl ester (CFDA-SE), in studies of laser light stress in E. coli and in Listeria (21), exposure to antimicrobial compounds (3, 7), and lactic acid exposure in Campylobacter jejuni (31). However, the use of CFDA-SE or other dyes requires a loading step in which the dye is incubated with the culture, which could harm cell physiology and regulation of cytoplasmic pH. Our use of endogenously expressed pHluorin eliminates the drawbacks of fluorescent dyes (2, 19).
The steady-state pH measurements for E. coli (Fig. 2C) and B. subtilis (Fig. 7C) were both consistent with observations of steady-state cultures in the range below pH 8 (14, 36). The comparison of the two organisms is of interest because, while they both grow over a similar range of external pH, they are very divergent genetically, they inhabit different environments, and their molecular mechanisms of pH response are different (14, 35). Under our conditions of adherent cells, no inversion of the ΔpH was observed at values of high external pH, although both species invert their ΔpH in cell suspension (28, 36). We are currently exploring whether altered culture conditions enable adherent cells invert their ΔpH.
Kinetic data on cytoplasmic pH following a rapid pH shift have not been observed previously for single cells of E. coli or B. subtilis. One report on Listeria monocytogenes using the pH indicator CFDA-SE (3) describes single-cell heterogeneity in the cytoplasmic pH response to nisin, a pore-forming bacteriocin.
In E. coli, we observed responses to rapid pH shifts (Fig. 4) that are consistent with observations using suspended cell cultures (29, 36). Upon acid shock, the E. coli cells showed a rapid “dip” in cytoplasmic pH followed by a rapid recovery (Fig. 4). A comparable dip and recovery are observed by fluorimetry of suspended cells (36). The degree of the dip and recovery, however, varied among individual cells, and a substantial fraction of cells failed to recover. A possible source of variation is that pH recovery fails in cells that have “aged” farther after cell divisions (33).
In B. subtilis, some cells showed a more robust response to acid shift in that the cytoplasmic pH fell slightly (by only 0.2 units) without requiring recovery (Fig. 8). This result is also consistent with suspended cells observed by fluorimetry, in which B. subtilis shows a pattern of kinetic pH response differing from that of E. coli (14). Again, however, the individual cells showed significant variability in the temporal response. We can now use pHluorin microscopy to test further what factors are associated with the heterogeneity and level of recovery.
For the first time, our pHluorin technique enabled us to visualize pH variation among cells in an E. coli biofilm (Fig. 6). Biofilms grew well upon our poly(l-lysine)-coated coverslips. Some studies have reported that poly(l-lysine) affects cell viability (4, 8), but under our conditions we found no evidence of growth inhibition. Following acid shift (Fig. 6), the biofilm cells took longer to recover cytoplasmic pH (up to 7 min) than did adherent planktonic cells (Fig. 4). The failure of pH recovery in some biofilm cells is of interest given the heterogenous nature of cells in biofilms, which show diverse capacities for metabolism and reproduction.
Our observations are important because previous reports of pH effects in biofilms are mixed and inconclusive. Biofilms of Streptococcus mutans (38) show greater or less acid resistance than planktonic cells, depending on the experimental conditions. And in E. coli O157:H7 (27), complex effects on biofilm acid resistance have important consequences for food safety. We will undertake further exploration of E. coli biofilm heterogeneity using confocal microscopy to allow three-dimensional mapping of pH responses.
ACKNOWLEDGMENTS
We thank Gero Miesenböck and the Sloan-Kettering Institute for Cancer Research for providing pGM1 and John Cronan for pBAD322.
This work was supported by grant MCB-1050080 from the National Science Foundation.
FOOTNOTES
- Received 8 February 2012.
- Accepted 7 March 2012.
- Accepted manuscript posted online 16 March 2012.
- Copyright © 2012, American Society for Microbiology. All Rights Reserved.
