INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS ANALYSIS DISCUSSION REFERENCES

Escherichia coli grows over more than a hundred-fold range of external osmolality (Osm), extending from as low as 0.015 Osm (Baldwin et al., 1995) up to ~1.9 Osm (McLaggan et al., 1990; Cayley et al., 1991) in minimal medium and up to ~3.0 Osm in rich medium (Record et al., 1998a). To grow over this range of external water activity (1.0 > a_H2O  0.95, where a_H2O = e^Osm/55.5) requires a high degree of thermodynamic sophistication. In general, growing cells may adapt to changes in osmolality of the growth medium i) by making compensating changes in the intracellular osmolality by changing the amounts of water and/or solutes in the cytoplasm and periplasm, so that the osmolality difference Osm and turgor pressure  = RTOsm across the cell wall are maintained, and/or ii) by allowing Osm to change so that turgor pressure changes with external osmolality (Record et al., 1998a). E. coli makes large and systematic changes in the amounts of cell and cytoplasmic water and in the amounts of periplasmic and cytoplasmic solutes in response to changes in osmolality of growth (reviewed by Record et al., 1998a; Csonka and Epstein, 1996). How do these changes affect the activity of cytoplasmic water? Of particular relevance for the present study are observations that the amount of periplasmic membrane-derived oligosaccharide (MDO, heterogeneous anionic glucose oligomers that are too large to pass through pores in the outer membrane; Kennedy, 1996) increases as the osmolality of growth decreases for all wildtype E. coli K-12 strains examined (Kennedy, 1982; Kennedy and Rumley, 1988; Sen et al., 1988; Lacroix et al., 1989). Do these changes in amount of periplasmic MDO demonstrate that turgor pressure changes with osmolality of growth? Knowledge of turgor pressure as a function of osmolality of growth is the only way to determine the physiological range of cytoplasmic water activity, which, in turn, is needed for analyses of the thermodynamics of cytoplasmic biopolymer processes as a function of growth osmolality.

The activity of a small subset of genes and gene products varies with the osmolality of the growth medium, although the nature of the signal(s) controlling these osmoregulated changes is not well understood (Wood, 1999). Changes in turgor pressure have been considered as a possible osmoregulatory signal (Wood, 1999; Csonka and Epstein, 1996). Few estimates of turgor pressure of E. coli are available under any conditions, however, and the questions of whether turgor pressure exists only across the cell wall/outer membrane or across the inner (cytoplasmic) membrane and whether turgor pressure changes with osmolality of growth have been controversial and unresolved. Studies of osmotic and Donnan properties of nongrowing suspensions of the closely-related bacterium Salmonella typhimurium led Stock et al. (1977) to conclude that the cytoplasm and periplasm are isoosmotic, and that turgor pressure is maintained across the cell wall. However, studies of the osmoregulation of the expression of the kdpABC operon in E. coli have been interpreted as indicating that changes in turgor pressure are sensed by the cytoplasmic membrane-bound kdpD sensor kinase (Laimins et al., 1981). Phase and electron microscopy studies of growing cells have led to the proposal that turgor pressure is maintained across the cytoplasmic membrane (Koch, 1995, 1998).

To address these issues, we have indirectly quantified the variation of turgor pressure with external osmolality by measuring the effects of osmolality of growth and of plasmolysis with NaCl on the volumes (i.e., amounts) of cell, periplasmic, and cytoplasmic water, and by analyzing the dependence on growth osmolality of the concentration of periplasmic MDO, calculated from published amounts of MDO and from our measurements of periplasmic volume. We interpret the observed changes in amounts or volumes of water and in concentration of periplasmic MDO in terms of simple physical models. We conclude that the periplasm and cytoplasm are isoosmotic and that E. coli systematically varies turgor pressure with external osmolality.

Background on passive and active responses of E. coli to changes in external osmolality

The cytoplasm of E. coli exhibits both passive and active responses to changes in external osmolality with NaCl or other cytoplasmic membrane-impermeable solutes (Record et al., 1998a). As an example of a passive response, consider a so-called plasmolysis titration (Cayley et al., 1991, 1992) in which a fresh, nongrowing cell suspension harvested from exponential growth in minimal medium at low osmolality (e.g., 0.1 Osm) is titrated with NaCl. Na⁺ and Cl equilibrate across the outer cell membrane, subject to the Donnan distribution for ionic species, which results from the presence of outer-membrane-impermeable anions in the periplasm (Stock et al., 1977; Sen et al., 1988; and see below). Because the cytoplasmic membrane is impermeable to NaCl and incapable of supporting an osmotic pressure difference, the cytoplasm loses water to increase its osmolality to that of the periplasm; this passive response reduces the amount of cytoplasmic water at 1.0 Osm to ~30% of its original value without changing the amounts of any cytoplasmic solutes (Cayley et al., 1991; Record et al., 1998a). (Plasmolysis beyond 1.0 Osm results eventually in removal of all unbound [osmotically active] water from the cytoplasm.) To recover from the plasmolyzed state at 1.0 Osm and resume growth requires an active osmoregulated response, initiated by increased uptake of extracellular K⁺, with the end result (in minimal growth medium without added osmoprotectants) that cytoplasmic amounts of K⁺, glutamate (and other organic anions), trehalose, and water increase, and the amount of cytoplasmic putrescine (2+) decreases (cf., Csonka and Epstein, 1996; Record et al., 1998a for reviews). We propose that the fundamental reason for the increases in amounts of cytoplasmic K⁺, glutamate, and trehalose is to allow the cell to increase the amount of cytoplasmic water (by almost twofold over that characteristic of the nongrowing plasmolyzed state at 1.0 Osm) and thereby achieve the highest growth rate possible in a minimal medium without osmoprotectants at this osmolality (Cayley et al., 1991, 1992; Record et al., 1998a).

In the present study, we use suspensions of E. coli lacking external K⁺ to prevent cells from adapting to osmotic stress, and analyze measurements of the passive changes in cell and compartment volumes in plasmolysis titrations of cells with NaCl to quantify their osmotic properties. This novel analysis provides the basis for our use of cell and cytoplasmic volume measurements to determine the osmotic properties and intracellular water activity of E. coli as a function of osmolality of growth.

The balloon model for turgor pressure of E. coli: why measurements of cell volume provide information about turgor

The bacterial cell wall has been considered balloon-like (Doyle and Marquis, 1994) because it can stretch under pressure. Outwardly directed turgor pressure (the osmotic pressure difference between the cell interior and the external medium) is the analog of the difference in air pressure, which inflates a balloon. Turgor pressure stretches the peptidoglycan of the E. coli cell wall (Woldringh, 1994 and references therein) elastically (Koch and Woeste, 1992) relative to the unstressed state existing in the absence of an osmotic pressure difference. E. coli has two compartments (periplasm and cytoplasm) with different permeabilities to solutes, so osmotic responses of both compartments must be considered to interpret effects of osmotic stress on cell volume. In this work, we provide additional evidence that the periplasm and cytoplasm are isoosmotic, in agreement with the conclusion of Stock et al. (1977) and Sen et al. (1988), which simplifies the physical situation. Thus, the balloon analogy is instructive, and serves as the qualitative basis for our model that changes in cell volume reflect changes in turgor pressure.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS ANALYSIS DISCUSSION REFERENCES

Bacterial strain, growth media, buffers, and chemicals

All experiments were performed with E. coli K-12 strain MG1655. Cells were grown aerobically at 37°C in a very low osmolality MOPS (3-(N-morpholino)-propanesulfonate)-buffered glucose minimal medium (VLOM; 0.03 Osm; Capp et al., 1996), in the MOPS-buffered glucose minimal medium (MBM; 0.1 Osm; Cayley et al., 1989), or in MBM with NaCl used to adjust osmolality of the growth media. Wash buffer is growth medium in which all K⁺ (present as KH₂PO₄) and glucose were replaced by an isoosmotic amount of NaCl. Plasmolysis buffer is wash buffer with additional NaCl to increase the osmolality.

³H₂O (1 mC_i/g) and [³H] polyethylene glycol (1.51 mC_i/g) were obtained from DuPont (Boston, MA). [¹⁴C] sucrose (621 mC_i/mmol), [¹⁴C] inulin (9.4 mC_i/mmol), [¹⁴C] urea (54.0 mC_i/mmol), and [³H] sucrose (12.0 C_i/mmol) were obtained from Amersham (Arlington Heights, IL). All radiochemicals except ³H₂O and [¹⁴C] urea were purified of radiolabeled contaminants that interfere with volume measurements by preincubation with cell slurries as previously described (Cayley et al., 1991, 1992). NH₄OH (5.08 N in water), 1-bromododecane, and silicone oil were obtained from Aldrich (Milwaukee, WI).

Measurement of amounts of cellular and cytoplasmic water in nongrowing cell suspensions

Volumes of cell water (_cell^wa) and cytoplasmic water (_cyto^wa) in units of µL per mg cell dry weight (µL/mg DW) of fresh nongrowing cell suspensions were obtained from comparisons of the volume of cell pellets accessible to ³H₂O to that of ¹⁴C inulin (a polymer that is outer-membrane impermeable) or ¹⁴C sucrose (which freely diffuses into the periplasm but is not transported into the cytoplasm) using the method of Stock et al. (1977) as described previously (Cayley et al., 1991, 1992). (These volumes are equivalent to amounts of water in mg water/mg DW assuming a density of intracellular water of 1.0 g/mL). Briefly, cells were harvested from exponential growth (at ~3 × 10⁸ cells/mL, <0.2 mg DW/mL) by centrifugation at 7,000 × g for 8 min. Cell pellets were suspended in isoosmotic wash buffer, recentrifuged, resuspended to a final cell density of ~5 mg DW/mL with wash buffer and swirled periodically for the ~30 min typically needed to complete a series of measurements. Three µC_i/mL of ³H₂O and 0.2 µC_i/mL of either ¹⁴C sucrose or ¹⁴C inulin were then added per mL of suspension, after which the samples were immediately centrifuged at 12,000 × g for 30 s (a sufficient time to pellet cells completely). The cpm in samples of the supernatant and cell pellets were then assayed by dual isotope scintillation counting and used to determine _cell^wa and _cyto^wa (in µL/mg DW) as described previously (Cayley et al., 1991). Measurements of _cyto^wa immediately after harvest and 40 min after harvest were the same, demonstrating that the amount of cytoplasmic water did not vary during the time needed to perform a series of measurements. In addition, _cyto^wa was independent of time of incubation of ¹⁴C sucrose in suspensions before volume assay for at least 10 min, the longest time tested. Control experiments showed no significant differences in volumes determined with ³H PEG instead of ¹⁴C inulin or with ¹⁴C taurine instead of ¹⁴C sucrose or ³H sucrose, indicating the absence of any specific interactions of these probes with cell components. Bubbling suspensions with O₂ or incubation with 5 mM fructose, 11 mM glucose or 1.3 mM KH₂PO₄ for five min prior to assay did not affect _cyto^wa, showing that any deprivation of oxygen or nutrients that occurred during preparation of suspensions did not lower _cyto^wa. All steps after cell growth were performed at room temperature.

In NaCl plasmolysis titrations to determine the passive responses of _cell^wa and _cyto^wa to increases in external osmolality, suspensions of cells grown at 0.03 Osm (in VLOM) or at 0.83 Osm (in MBM+0.4 NaCl) were assayed as described above except that, immediately before addition of radiochemicals, samples were diluted fivefold with plasmolysis buffer to achieve the desired range of final NaCl concentrations and a final cell density of ~5 mg DW/mL.

Most previously published values of _cyto^wa (Cayley et al., 1991, 1992) referred to in this paper were determined using ¹⁴C taurine in place of ¹⁴C sucrose. McLaggan and Epstein (1991) found that taurine can be accumulated with a K_M of ~30 mM at high osmolality by strains of growing E. coli defective in the osmotically regulated accumulation of cytoplasmic trehalose. However, in addition to controls that previously demonstrated that neither purified ¹⁴C taurine (typically used at a concentration of 0.28 mM) nor ¹⁴C sucrose is accumulated by nongrowing suspensions of our wild-type strain under the conditions of our volume assays (Cayley et al., 1991, 1992), we observe that 1) _cyto^wa of suspensions measured with purified ¹⁴C sucrose or ¹⁴C taurine for cells grown in MBM+0.5 M NaCl are the same within error, 2) dilution of ¹⁴C taurine to 1 mM with unlabeled taurine does not alter measured values of _cyto^wa of cells grown in MBM+0.2 M NaCl, and 3) variation of the concentration of ¹⁴C taurine from 0.1 mM to 1 mM does not significantly affect measured values of _cyto^wa of cells plasmolyzed with 1 M NaCl. Use of taurine therefore introduced no systematic errors in determinations of _cyto^wa.

We also tested whether the centrifugal harvest method used to prepare suspensions caused physiological changes relative to growing cells. Upon completion of a volume assay with suspensions of cells grown at 0.1 Osm, samples examined in a phase microscope exhibited motility, indicating that these suspensions retained and were capable of utilizing endogenous energy reserves. Cell viability (the number of colony forming units determined by plating on LB agar) and _cyto^wa also remained constant over the course of an assay, as determined by comparison of these values before and after a series of volume measurements. Moreover, cell suspensions had not entered stationary phase by completion of a typical volume assay, because suspended cells resumed a normal growth rate immediately, with no observable lag, upon dilution into fresh growth medium, and the thermotolerance of centrifugally-harvested suspensions grown at 0.1 Osm (determined by the reduction in viability with time after heating at 48°C; see Hengge-Aronis et al., 1991) was identical to that of growing cells and not enhanced as it was in cells grown into stationary phase at 0.1 Osm (data not shown).

Measurements of amounts of cellular and cytoplasmic water and of protein in growing cells

Volumes of water in growing cells were determined by measuring the distribution of radiolabeled probes in pellets of cultures harvested by brief centrifugation through oil. Briefly, 2 µC_i/mL of ¹⁴C inulin, ¹⁴C sucrose, or ³H₂O was added to midlog phase cultures growing in MBM (0.1 Osm, an osmolality at which the density of cells is ~1.08-1.09 g/mL; Baldwin et al., 1995) immediately before centrifuging 1.4 mL aliquots in microfuge tubes containing 200 µL of 1-bromododecane (BDD;  = 1.038 g/mL) at 12,000 × g for at least 90 s. (Centrifugation for less than ~90 s incompletely pelleted cells, whereas results were independent of centrifugation times longer than 90 s). After removing 100 µL of supernatant from each tube, the remaining supernatant and BDD was carefully removed by aspiration, the pellets suspended in 100 µL of water, and the cpm in the supernatant and pellet samples determined by scintillation counting. (The cpm in the suspended pellets ranged from ~1,000 cpm for samples containing ¹⁴C inulin up to ~2,000 cpm for samples containing ³H₂O, and no cpm was detected in the discarded BDD). The amount of protein in each sample was determined by comparing the A₅₅₀ of the culture at the time of assay to a standard curve of mg protein/mL of culture versus A₅₅₀ determined separately on multiple cultures grown in MBM, using the assay of Lowry et al. (1951) with BSA as standard as previously described (Cayley et al., 1991). Values were normalized to the dry weight of samples using measurements of the protein/dry weight ratio determined separately. Volumes of cytoplasmic and cellular water (in µL/mg dry weight) were then calculated as for suspensions by subtracting the sucrose- or inulin-accessible volumes, respectively, from the water-accessible volume. In experiments to control for the effects of centrifugation of suspensions through BDD, the procedure used for assaying volumes of suspensions was used except that, after radiochemical addition, samples were centrifuged in tubes containing 200 µL of BDD for 120 s.

Preliminary experiments showed that the error in the measurements of volumes of growing cells assayed as above was much lower in cells centrifuged through BDD than in cells centrifuged in the absence of BDD. Preliminary experiments also showed that the firmness of pellets and reproducibility of results obtained with cells spun through BDD ( = 1.038 g/mL) was greater than for silicon oil ( = 1.050 g/mL). Evaporation of ³H₂O was an insignificant source of error because the membrane-permeable solute ¹⁴C urea (which readily equilibrates into cytoplasmic water; Mitchell and Moyle, 1956) gave results that were identical to those obtained with ³H₂O as long as sufficient time (~90 s) was allowed for urea to diffuse into the cytoplasm of cells before centrifugation of samples through BDD.

Calculations of periplasmic MDO concentration and of the contribution of MDO to turgor pressure

To calculate concentrations of periplasmic MDO at different osmolalities of growth, molar amounts of MDO in E. coli K-12 from the data of Kennedy (1982) and Lacroix et al. (1989) were divided by the corresponding amounts of free water in the periplasm obtained from our volume measurements. Kennedy (1982) and Lacroix et al. (1989) measured the relative amounts of MDO (in cpm/mg DW, using ³H-glycerol to label MDO) as a function of growth osmolality varied with NaCl. Values of the radioactivity of the extracted MDO taken from Fig. 1 of Kennedy (1982) were corrected for non-MDO material by subtracting the reported osmotically-invariant amount of non-MDO radioactivity in MDO extracts. Lacroix et al. (1989) reported relative amounts of MDO determined after purification from non-MDO-labeled material. The relative amounts of MDO from Kennedy (1982) were normalized to nmol MDO/mg DW using the value of 136 nmol MDO glucose/mg DW determined by Kennedy and coworkers in cells grown at 0.07 Osm and their assumption of 9 glucose/MDO (Rumley et al., 1992). The relative amounts of MDO from LaCroix et al. (1989) were converted to nmol MDO/mg DW using their value of 125 ± 20 nmol MDO glucose/mg DW determined in cells grown at 0.07 Osm and by assuming 9 glucose/MDO. (The principal MDO species have 8-9 glucose monomers; Kennedy, 1996).

To calculate periplasmic concentrations of MDO (C_MDO, expressed as moles of MDO per L of free periplasmic water), the amount of periplasmic water _peri^wa (in µL/mg cell DW) as a function of osmolality were obtained by interpolation of the linear fit (cf., Fig. 3 B below) of measurements of _peri^wa determined in this study and in Cayley et al. (1991). Values of _peri^wa were converted to volumes of free periplasmic water (_peri,f^wa) by subtracting an osmolality-independent estimate of the volume of bound water of hydration of periplasmic biopolymers (_peri,b^wa = 0.09 µL/mg cell DW) obtained by assuming that periplasmic biopolymers are hydrated to the same extent as cytoplasmic biopolymers (0.5 g H₂O/g biopolymer; Cayley et al., 1991, 1992) and that the protein/DW ratio of the periplasm and cytoplasm are the same. Periplasmic protein was taken as 20% of cell protein (Ames et al., 1984; Cronan et al., 1987).

The contribution of periplasmic MDO (concentration C_MDO and apparent valence Z_MDO^app) to turgor pressure across the cell wall as a function of salt concentration was calculated from a Donnan equilibrium analysis, in which the osmotic pressure difference across the cell wall is ascribed to anionic MDO in the periplasm and to the unequal distribution of salt ions between the periplasm and the external solution required for periplasmic electroneutrality. For this calculation, the ionic composition of the minimal growth medium with added NaCl was approximated as that of a univalent salt with an anion concentration C_X equal to the sum of the anion concentrations in the growth medium (primarily Cl and the MOPS anion). Then the MDO contribution to turgor pressure (_MDO) is given by

(1)

where R = 0.0821 liter atm mol¹ deg¹ and T is Kelvin temperature. In Eq. 1, nonideality effects are neglected. Subject to this approximation, Eq. 1 is valid over the entire range of concentration ratios C_MDO/C_X, and gives the correct limiting results _MDO = RTC_MDO for C_X C_MDO and _MDO = RT(|Z_MDO^app| + 1)C_MDO for C_MDO C_X.

Other methods

The osmolalities of the growth media, wash buffer, and low-osmolality plasmolysis solutions (0.1 Osm) were measured using a Wescor model 5520 vapor pressure osmometer. Osmolalities of other plasmolysis buffers were calculated assuming additivity of osmotic contributions from the wash buffer and added NaCl. The latter contribution was calculated using osmotic coefficients of NaCl from Robinson and Stokes (1959). This procedure was verified to be accurate to within 1% for representative samples measured by osmometry. All fittings were performed using the program NONLIN (Johnson and Frasier, 1985; Straume et al., 1991), a nonlinear functional-form, least-squares fitting program. All errors reported for fittings were obtained from NONLIN using a 67% confidence probability.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS ANALYSIS DISCUSSION REFERENCES

Variation of the volume of cell water of E. coli in NaCl titrations of cells grown at very low (0.03) and moderately high (0.83) osmolality

Figure 1 plots the volume of cell water (_cell^wa) per unit DW of E. coli grown at 0.03 Osm (in VLOM) and during the subsequent course of plasmolysis titrations with NaCl. At the growth osmolality of 0.03 Osm, _cell^wa = 2.96 ± 0.10 µL/mg DW. Numerically, values of _cell^wa correspond to amounts of cell water in mg water/mg DW, assuming a density of 1 mg/µL. Figure 1 shows that addition of NaCl reduces _cell^wa of these cells monotonically to a high osmolality minimum value (designated _cell,min^wa) of 1.9 ± 0.05 µL/mg DW, attained within error at plasmolyzing osmolalities in excess of ~1 Osm. Figure 1 also plots the behavior of _cell^wa in plasmolysis titrations of cells grown at 0.83 Osm (in MBM+0.4 M NaCl), at which osmolality the amount of periplasmic MDO is very low (Kennedy, 1982; Lacroix et al., 1989). The initial value of _cell^wa is much smaller for cells grown at 0.83 Osm than at 0.03 Osm (cf., Fig. 1 and Table 1), in agreement with previous observations (summarized in Table 1) that _cell^wa decreases with increasing osmolality of growth (Richey et al., 1987; Larsen et al., 1987; Cayley et al., 1991). Notably (cf., Fig. 1), after plasmolysis to any osmolality greater than 0.83 Osm, the amount of water remaining in cells grown at 0.83 Osm is always significantly less than the amount of water in cells grown at 0.03 Osm. Figure 1 shows that, for cells grown at 0.83 Osm and titrated with NaCl, increasing external osmolality reduces the volume of cell water to a high osmolality minimum value _cell,min^wa of 1.56 ± 0.05 µL/mg DW. This plateau value is significantly less than the value of _cell,min^wa obtained from plasmolysis titrations of cells grown at 0.03 Osm (1.90 ± 0.05 µL/mg DW).

View larger version (17K): [in this window] [in a new window]   FIGURE 1   Reduction in water-accessible cellular volume cellwa of suspensions of cells grown at 0.03 Osm (in VLOM; ) and at 0.83 Osm (in MBM+0.4 M NaCl; ) with increasing plasmolyzing osmolality adjusted with NaCl. Each point shows the average (±1 SD) of 2-7 independent determinations, each performed in triplicate on separate cultures. The curves are empirical hyperbolic best fits to the data, and were used to estimate cell,minwa.

To verify that the variation of _cell,min^wa with osmolality of growth is only the result of changes in the volume of water per cell, and not in the amount of dry weight per cell, we examined whether the dry weight per cell varies with osmolality of growth. Table 1 lists the growth rate and mass of protein per viable cell for cultures grown from 0.1 to 1.0 Osm. Whereas the growth rate exhibits a maximum near 0.28 Osm, the amount of protein per viable cell shows no systematic variation with osmolality and has an average value of 0.41 ± 0.03 pg per cell, which is within the range of published values for E. coli B/r (Bremer and Dennis, 1996). Because the protein/DW ratio of cells grown over this range of conditions is 0.68 ± 0.07, independent of osmolality (Cayley et al., 1991), the dry weight per cell is therefore also constant over the range of osmolalities and growth rates examined.

Table 1 also lists volumes of water per viable cell, calculated from determinations of the amount of protein per viable cell, the protein/DW ratio, and the volume of cell water per mg DW. The calculated amount of H₂O per cell decreases from ~1.8 fL for cells grown at 0.03 Osm to ~1.1 fL for cells grown at 1.0 Osm (Table 1). For cells grown at 0.1 Osm (in MBM), the volume of water per cell is ~1.5 fL. Because the water-inaccessible volume of these cells is 0.63 ± 0.09 µL/mg DW (or ~0.4 fL per cell), independent of osmolality (Cayley et al., 1991), therefore, the total volume of the average cell grown at 0.1 Osm is ~1.9 fL per cell. The corresponding dimensions of a cell with this volume, assuming a cylindrical shape with hemispherical ends and an overall 2:1 length:width ratio, is approximately 2.2 µm × 1.1 µm, consistent with accepted dimensions of E. coli (Neidhardt et al., 1990).

Changes in water-accessible cytoplasmic and periplasmic volume during NaCl plasmolysis titrations of E. coli grown at very low osmolality (0.03 Osm)

Figure 2 plots the volumes of cytoplasmic and periplasmic water of E. coli grown at 0.03 Osm and subsequently plasmolyzed with NaCl. The initial volumes of water in these compartments are _cyto^wa = 2.53 ± 0.08 µL/mg DW and _peri^wa  _cell^wa  _cyto^wa = 0.43 ± 0.13 µL/mg DW. As the concentration of NaCl is increased, _cyto^wa decreases monotonically to an apparent plateau volume of ~0.4 µL/mg DW, approached at plasmolyzing osmolalities in excess of 3 Osm. (For reference, the curve for the reduction in _cell^wa with plasmolyzing osmolality from Fig. 1 is also shown in Fig. 2.) Plasmolysis of cells grown at 0.03 Osm affects _cyto^wa much more than _cell^wa, reducing _cyto^wa to less than 20% of its initial value, whereas _cell^wa is reduced to 60% of its original value. Consequently _peri^wa (the difference between _cell^wa and _cyto^wa) increases during these plasmolysis titrations by more than 300%, to a plateau value of approximately 1.45 ± 0.07 µL/mg DW. This increase in _peri^wa indicates that outer-membrane impermeable periplasmic solutes (including MDO) are diluted to about 30% of their initial concentration upon plasmolysis with high concentrations of NaCl. The observed variations in cell and compartment volumes in NaCl plasmolysis titrations are larger but otherwise quite analogous to those reported by Stock et al. (1977) for sucrose plasmolysis titrations of cells grown at 0.14 Osm.

View larger version (23K): [in this window] [in a new window]   FIGURE 2   Reduction in water-accessible cytoplasmic volume () and increase of water-accessible periplasmic volume () of cells grown at 0.03 Osm (in VLOM) with increasing plasmolyzing osmolality adjusted with NaCl. Each point shows the average (±1 SD) of 6-12 measurements performed on 2-4 cultures. The curve through the cytowa data is an empirical best-fit hyperbolic function. The corresponding fit to the cellwa data for cells grown at 0.03 Osm from Fig. 1 is shown for comparison. Periplasmic data points were obtained by subtraction of cytoplasmic volumes from cell volumes. The periplasmic volume curve was determined by subtraction of the fitted curves for cell and cytoplasmic volumes.

Volume measurements on fresh cell suspensions are applicable to growing cells

K⁺-deprived suspensions of cells were used in the plasmolysis titrations to study the passive response of cells to osmotic stress, because increased uptake of K⁺ from the medium is the initial response required for adaptation of E. coli to hypertonic shock (Csonka and Epstein, 1996). The suspensions also lacked glucose to prevent new synthesis of organic osmolytes, further ensuring that plasmolyzed cells are unable to respond to increases in external osmolality by active osmoregulated mechanisms, and, instead, exhibit only the passive response of loss of cell water. The absence of an active response in plasmolyzed cell suspensions was verified by our observation (data not shown) that the amount of cytoplasmic water in cells after plasmolysis with 1 M NaCl remained unchanged for at least 15 min, the longest time tested.

To test directly whether the amounts of cell and cytoplasmic water in cell suspensions and in growing cells are the same, we used a radiochemical method to measure volumes of growing cells harvested by brief centrifugation through BDD oil as described in Methods. Table 2 shows that the water-accessible cell and cytoplasmic volumes of mid-log phase cells growing at 0.1 Osm are the same, within error, as in fresh suspensions, indicating that fresh suspensions have the same osmotic properties as growing cells (and see Discussion). Table 2 also shows that centrifugation of cells through BDD per se does not perturb the amount of intracellular water, because cell and compartment volumes of suspensions assayed by centrifugation through BDD are the same as suspensions assayed without BDD. A bathing atmosphere of medium must therefore accompany cells as they sediment through BDD, a conclusion verified by our observation that the extracellular (i.e., inulin-accessible) volume in pellets of samples spun through BDD contained ~1.8 µL extracellular water/mg total cell protein. Although significantly less than the ~2.5 µL extracellular water/mg protein of samples of pellets spun in the absence of BDD, the 1.8 µL/mg protein greatly exceeds that estimated for a monolayer coverage (~0.02 µL/mg protein) of a smooth surface with the dimensions of the E. coli cell.

The equivalence of the amounts of cell and cytoplasmic water in fresh suspensions and growing cells validates our use of suspensions in this study. Measurements on suspensions are preferable to measurements of growing cells because they are of higher accuracy (see Table 2), in large part because the cell density of log phase cultures is ~25-fold lower than that of suspensions. Moreover, to measure only the passive response of cells to osmotic stress, plasmolysis titrations must be performed on nongrowing samples lacking extracellular K⁺, because the initial active response to the increase in external osmolality of increased uptake of K⁺ from the medium (Csonka and Epstein, 1996) begins immediately after external osmolality increases (Epstein and Schultz, 1965).

Variation of periplasmic MDO concentration with osmolality of growth: implications for turgor pressure

Figure 3 A summarizes amounts of MDO calculated by us as described in Methods from the data of Kennedy (1982) and Lacroix et al. (1989) for E. coli K-12 strains grown at various osmolalities. Amounts of MDO decrease monotonically from ~15 nmol MDO/mg DW at 0.07 Osm to very low levels at high osmolality. Figure 3 B shows our determinations of the volume of free periplasmic water _peri,f^wa, obtained from the difference between measured volumes of cell and cytoplasmic water as a function of osmolality of growth as described in Methods. At least above 0.3 Osm, _peri,f^wa increases slightly as osmolality of growth increases, in agreement with previous measurements of _peri^wa (Richey et al., 1987; Larsen et al., 1987). The experimental uncertainty is too large to allow us to conclude whether a different behavior occurs at lower osmolality, and, consequently, we have fit all these data to a line for purposes of interpolation. Interpolated values of _peri,f^wa (Fig. 3 B) were used with determinations of n_MDO (Fig. 3 A) to estimate concentrations of MDO (c_MDO; see Methods) in the free water of the periplasm as a function of the osmolality of growth. Figure 3 C shows that c_MDO decreases from ~0.05 mol/L at a growth osmolality of 0.07 Osm to ~0.003 mol/L at a growth osmolality of 0.8 Osm. Uncertainties in absolute MDO concentrations are approximately ±35%; this uncertainty, although large, has no effect on the semiquantitative conclusions obtained from analyses of these results below. (The trend in relative amounts of MDO is known to higher accuracy; Lacroix et al., 1989). In particular, the contribution of MDO (which Kennedy [1996] concluded are free and not bound to other periplasmic components) to turgor pressure across the cell wall may be estimated from these c_MDO and from the concentration of anions of the medium, as described in Methods and analyzed below.

View larger version (18K): [in this window] [in a new window]   FIGURE 3   Reduction of the amount and concentration of periplasmic MDO with increasing osmolality of growth. Panel A plots the amounts of MDO (nMDO) in nmol/mg cell DW calculated as described in Methods from the analytical data of Kennedy (1982; ) and Lacroix et al. (1989; ) versus growth osmolality. The curve is the empirical best hyperbolic fit to all the data. Panel B plots the volumes of free periplasmic water peri,fwa (determined as described in Methods) in µL/mg cell DW versus osmolality of growth. The line is the best linear fit to the data and is shown for purposes of interpolation. Panel C shows the decrease in concentration of periplasmic MDO (cMDO) (in mol/L free periplasmic water) calculated from values of nMDO (A) and interpolated values of peri,fwa (B) with increasing osmolality of growth.

	ANALYSIS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS ANALYSIS DISCUSSION REFERENCES

Analysis of _cyto^wa in plasmolysis titrations of cells grown at low osmolalities (0.03 Osm to 0.28 Osm) demonstrates that turgor pressure decreases with increasing osmolality of growth

For cells grown at 0.1 and 0.28 Osm, Cayley et al. (1991) quantitatively analyzed the behavior of the volume occupied by cytoplasmic water _cyto^wa in plasmolysis titrations to obtain both the volume occupied by bound water of hydration _cyto,b^wa and the amount of osmotically-significant solutes _cyto( n_j)_cyto (see below) in the cytoplasm. Here, we extend this analysis to determine turgor pressure of these cells and of cells grown at 0.03 Osm. Turgor pressure is fundamentally related to the difference between the osmolality of the cytoplasm (Osm_cyto) and the external medium (Osm_ex) by

(2)

The osmolality of the cytoplasm is related to the molar amount of cytoplasmic solutes ( n_j)_cyto and to the volume occupied by free (unbound) cytoplasmic water (_cyto,f^wa) by

(3)

where _cyto is the osmotic coefficient of the cytoplasm (Cayley et al., 1991). In Eq. 3,

(4)

where _cyto,b^wa is defined as the volume occupied by bound cytoplasmic water (expected to be primarily water of macromolecular hydration; Cayley et al., 1991; Record et al., 1998a). Combining Eqs. 2-4 gives

(5)

Plots of _cyto^wa versus 1/Osm_ex for data obtained in NaCl plasmolysis titrations of cells grown at 0.1 Osm and 0.28 Osm (Cayley et al., 1991) and at 1.02 Osm (Cayley et al., 1992) are linear for all plasmolyzing NaCl concentrations investigated, although the initial (unplasmolyzed) value of _cyto^wa at 0.1 Osm deviates from the best fit line. Interpreted using Eq. 5, this linear behavior indicates most simply that /RT is negligible relative to Osm_ex in the range of plasmolyzing NaCl concentrations investigated, and that both _cyto( n_j)_cyto and _cyto,b^wa are independent of plasmolyzing NaCl concentration (Cayley et al., 1991). A similar plot of the _cyto^wa data from Fig. 2 versus Osm_ex¹ for cells grown at 0.03 Osm is highly nonlinear (Fig. 4), indicating most simply that residual turgor pressure is significant relative to RTOsm_ex at low NaCl concentrations. To simplify determination of _cyto( n_j)_cyto and _cyto,b^wa, we therefore linearly fit _cyto^wa versus 1/Osm_ex for cells plasmolyzed to an osmolarity exceeding 1 Osm to Eq. 5 assuming turgor pressure is zero. (Above 1 Osm, _cell^wa ceases to vary significantly with Osm_ex, from which we deduce that residual turgor pressure is insignificant relative to RTOsm_ex). These data are well fit by a line (r² = 0.99) with best-fit slope _cyto( n_j)_cyto = 0.31 ± 0.04 µmol/mg DW and intercept _cyto,b^wa = 0.45 ± 0.03 µL/mg DW, labeled A in Fig. 4. (Fitting these data assuming cells grown at 0.03 Osm exhibit a small residual turgor pressure (~0.4 atm) at high plasmolyzing osmolalities (see below) yields insignificantly different values of _cyto,b^wa and _cyto( n_j)_cyto.) The value of _cyto,b^wa for cells grown at 0.03 Osm is within error of previous determinations of _cyto,b^wa of cells grown at higher osmolality (see Table 3). The value of _cyto( n_j)_cyto is approximately equal to that for cells grown at 0.1 Osm (Cayley et al., 1991), consistent with our previous conclusion (Capp et al., 1996; see also Record et al., 1998a) that the primary difference in the pool of cytoplasmic osmolytes in these cells is that cells grown at 0.03 Osm have ~0.04 ± 0.02 µmol/mg DW more putrescine(2+) and, thus, ~0.08 ± 0.04 µmol/mg DW less K⁺ than cells grown at 0.1 Osm.

View larger version (25K): [in this window] [in a new window]   FIGURE 4   Effects of periplasmic MDO on cytowa in a NaCl plasmolysis titration of cells grown at 0.03 Osm (VLOM). Contributions to turgor pressure from the Donnan osmotic pressure of periplasmic MDO as a function of Osmex were calculated with Eq. 1 using a lower bound estimate of nMDO = 15 nmol/mg DW (estimated by extrapolation of Fig. 3 A as described in the text), interpolated values of peri,fwa from Fig. 3 B, and several choices of MDO valence ZMDOapp. These contributions to were used in Eq. 6 to predict the behavior of cytowa versus 1/Osmex, where Osmex was varied with NaCl in a plasmolysis titration (cf. Methods). Curve A shows the linear fit to the data of cells plasmolyzed to ~1 Osm and above (see text). Curve B shows the fit assuming ZMDOapp = 0 to calculate ; curve C shows the fit assuming ZMDOapp = 3 to calculate ; curve D is the best fit of Eq. 6 to the data and corresponds to the situation in which |ZMDOapp| is initially 2 in unplasmolyzed cells and increases to 3 as the concentration of NaCl increases (see text). Inset: Reduction in turgor pressure of cells grown at 0.03 Osm (VLOM) with increasing osmolality of plasmolysis. Turgor pressure was predicted from the concentration of periplasmic MDO by Eq. 1 with a value of ZMDOapp that varied with [NaCl] as in curve D.

The linearity of the high-osmolality region of the plasmolysis plot in Fig. 4 indicates that _cyto,b^wa and _cyto( n_j)_cyto for cells grown at 0.03 Osm are independent of plasmolyzing osmolality. Because we use nutrient-deprived suspensions to ensure that the sum of the amount of cytoplasmic solutes ( n_j)_cyto is constant throughout these plasmolysis titrations (see above), _cyto is also constant at high plasmolyzing osmolalities. We assume that _cyto does not vary at lower plasmolyzing osmolality, which is consistent with the K⁺-nucleic acid polyion model used to interpret _cyto (Cayley et al., 1991). Because _cyto,b^wa is independent of growth osmolality (see Table 3), we assume it is also independent of low plasmolyzing osmolality. With these plausible assumptions, the turgor pressure of these cells before or in the initial stages of plasmolysis may be estimated from experimental values of _cyto^wa using

(6)

From the value of _cyto^wa of unplasmolyzed cells (2.53 ± 0.08 µL/mg DW) and the values of _cyto( n_j)_cyto and _cyto,b^wa obtained above and tabulated in Table 3, the calculated turgor pressure of cells grown at 0.03 Osm is  = 3.1 ± 0.4 atm. This value significantly exceeds the turgor pressures calculated from Eq. 6 from the published values of _cyto( n_j)_cyto and _cyto,b^wa for cells grown at 0.1 Osm (MBM;  = 1.5 ± 0.3 atm) and at 0.28 Osm (MBM+0.1 M NaCl;  = 0.7 ± 1.1 atm, i.e.,   1.8 atm; see Table 3 and Cayley et al., 1991).

A limitation of our plasmolysis titration method seen in these results is that the propagated error becomes comparable to the turgor pressure itself for growth osmolalities above 0.1 Osm, because the absolute error in _cyto( n_j)_cyto/_cyto,f^wa increases and the turgor pressure decreases with increasing osmolality. Hence, other approaches (developed below) are needed to establish whether turgor pressure varies with osmolality of growth above 0.1 Osm.

Contribution of MDO to periplasmic osmolality and turgor pressure as functions of the osmolality of growth and of plasmolysis

We evaluated periplasmic MDO concentration as functions of osmolality of plasmolysis and growth from the amounts of MDO and volumes of unbound periplasmic water in Fig. 3. These values were then used in Eq. 1 to quantify the effect of periplasmic MDO concentration C_MDO, MDO apparent valence Z_MDO^app, and external salt (anion) concentration C_X on . These calculations, described below, make a strong case for our conclusions that 1) the periplasm and cytoplasm are isoosmotic under all conditions, 2) the turgor pressure exerted across the cell wall at any condition of plasmolysis is primarily determined by C_MDO, Z_MDO^app, and C_X, and 3) periplasmic MDO concentration and turgor pressure decrease together with increasing osmolality of growth or of plasmolysis with NaCl.

Donnan analysis of effects of residual turgor from periplasmic MDO on the initial stages of a plasmolysis titration of cells grown at 0.03 Osm

1

Role of MDO as a determinant of the extent of cell wall stretch and cell volume in plasmolysis titrations

Contribution of MDO to turgor pressure as a function of growth osmolality

View larger version (13K): [in this window] [in a new window]   FIGURE 5   Reduction in turgor pressure () with increasing osmolality of growth. () Turgor pressure calculated with Eq. 6 from the cytoplasmic water-accessible volume cytowa determined for cells grown at 0.03, 0.10, and 0.28 Osm, using the values of cyto( nj)cyto and b,cytowa in Table 3. Uncertainties in these values are ±(0.3-1) atm (Table 3). () Turgor pressures calculated with Eq. 1 from the values of cMDO from Fig. 3 C and from the variation of ZMDOapp with [NaCl] inferred from the best-fit curve D of Fig. 4; uncertainties in these values are approximately ±35%. (+) Turgor pressures determined by measuring collapse pressures of gas vacuoles in Microcystis sp. grown at 15 mOsm and 45 mOsm (Reed and Walsby, 1985). (×) Turgor pressure determined by measuring collapse pressures of gas vacuoles in Ancylobacter aquaticus grown at 0.1 Osm (Koch and Pinette, 1987), which has an error of ~15%. The osmolalities of the media used by Reed and Walsby (1985) and Koch and Pinette (1987) were estimated from their composition.

Relationship between cell volume and turgor pressure

As an independent method of quantifying turgor pressure to compare with our estimates based on analysis of periplasmic MDO concentration, we applied an empirical relationship between changes in cell volume and changes in turgor pressure previously used to characterize turgor-volume relationships in plant (Zimmermann, 1978; Cosgrove, 1988) and microbial (Walsby, 1980; Reed and Walsby, 1985) cells.

Changes in turgor pressure () and the corresponding changes in total cell volume (_cell^total) have been related by the empirical equation (Broyer, 1952; Phillip, 1958; Zimmermann, 1978)

(7)

where  is the volumetric elastic modulus (Cosgrove, 1988), _cell^total = _cell^total  _cell,0^total, and _cell,0^total is the total cell volume in the zero-turgor reference state. Total cell volumes are defined as the sum of the water-accessible cell volume (_cell^wa) and the water-inaccessible cell volume (_cell^wi)

(8)

For the conditions of the present study, _cell^wi = 0.63 ± 0.09 µL/mg DW, independent of external osmolality (Cayley et al., 1991). The empirical elastic modulus  is apparently independent of both _cell^total and of turgor pressure over a range of volumes and pressures for some plant cells (Cosgrove, 1988) and for the gram n