Chapter 98

Effect of Temperature, Pressure, pH, and Osmotic Stress on Growth

JOHN L. INGRAHAM and ALLEN G. MARR

INTRODUCTION

The response of Escherichia coli and Salmonella typhimurium (official designation, Salmonella enterica serovar Typhimurium) to their physical environment is in no respect exceptional among bacteria: these enteric organisms, classified as mesophiles with respect to temperature and as neutrophiles with respect to pH, grow over the mid range of temperatures, pH values, water activities, and pressure in which bacterial growth occurs. In this chapter we will attempt to summarize the information regarding the responses of these organisms to their physical environment, to compare these responses with those of other bacteria, and, where possible, to discuss the physical basis of the responses.

TEMPERATURE

The Arrhenius relationship (1) between the velocity (v) of chemical reactions and absolute temperature (T), v = e ^–AE*/RT (E* is the energy of the reaction, R is the universal gas constant, and A is a constant called the collision or the frequency factor), predicts a straight-line relationship between the logarithm of velocity and the reciprocal of absolute temperature which holds for most simple chemical reactions.

When applied to the specific growth rate, k, of bacteria as a function temperature (for this application the term μ, the temperature characteristic, is substituted for AE*), the form of the plot is similar for almost all bacteria. Over a certain interval of temperature, commonly called the normal range, log k is linear with 1/T. At higher and lower temperatures, the growth rate decreases progressively, approaching a vertical asymptote at both the maximum and minimum temperatures for growth.

Most bacteria, including E. coli and S. typhimurium, can grow over a range of approximately 40°C. The growth rates of most wild-type strains of E. coli and S. typhimurium respond similarly to changes in temperature; the particular response of E. coli B/r is shown in Fig. 1. The normal temperature range extends from 21 to 37°C, over which range μ has a value of 13,000 to 14,000 cal/mol (ca. 54,000 to 59,000 J/mol). The maximum temperature at which balanced growth can be sustained by this strain is approximately 49°C. The minimum temperature for sustained growth of E. coli ML30, and presumably other strains as well, lies between 7.5 and 7.8°C (88).

Fig. 1Growth rate of E. coli B/r as a function of temperature. The specific growth rate (k, hour-1), log scale is plotted against the inverse of absolute temperature (kelvins). Individual datum points are marked with degrees Celsius: , in a rich medium; , in a glucose-minimal medium.

Source: After Herendeen et al. (39).

Measurement of the minimum temperature of growth is complicated by the pattern of growth that follows a shift in temperature to the low temperature range. After the shift, growth ceases for a period of time. Shifting an exponential-phase culture of E. coli ML30 growing in a minimal medium at 37°C to10°C is followed by a 4.5-h period during which no growth occurs (69). This period increases as the shift is made to progressively lower temperatures. Thus, it is difficult to distinguish whether failure of a culture to grow after a shift to a temperature near the minimum means that the temperature is less than the minimum for growth or whether the lag in growth is extended. The minimum temperature for growth can be determined with precision by growing a culture at a low temperature somewhat above the minimum and then shifting it to progressively lower temperatures. Even this sort of experiment must be interpreted with caution, however, because incremental shifts to temperatures below the minimum are followed by prolonged transient periods of increase in mass. A shift of E. coli from 10 to 7.0°C is followed by a period of 1 day of growth during which the growth rate gradually declines.

Effect of Nutrition on the Temperature-Growth Response

In wild-type strains, the extent of the normal temperature range and the value of the temperature characteristic within it are unaffected by nutrition. The values of these parameters are the same for cultures growing in a complex medium or in a mineral salts medium with glucose; in the latter, the rate of growth is correspondingly lower at all temperatures (Fig. 1). Neither does the richness of the medium affect the minimum temperature of growth of E. coli; however, this growth parameter of a number of other bacteria is altered by nutrition (43). In contrast, the growth rate of several strains of E. coli, including K-12 strains, is markedly affected in the high temperature range (40 to 45°C) by the availability of exogenous methionine (80). Possibly all strains of E. coli are so affected. In the absence of exogenous methionine, growth stops at 45°C. Between 40 and 45°C, the growth rate is limited by the absence of methionine. At these temperatures the activity of the first enzyme (homoserine transsuccinylase) of the methionine biosynthetic pathway is rapidly but reversibly inhibited, presumably by the rupture of hydrophobic and hydrogen bonds (81).

Modulation of Cellular Composition by Growth Temperature

No major differences in cellular composition exist among cultures of E. coli grown at temperatures within the normal range. If cultures of E. coli growing at one of these temperatures are abruptly shifted to another, the growth rate, without a detectable transient period, assumes the value characteristic of the temperature to which the shift is made. Also in their classic studies, Schaechter et al. (84) showed that the macromolecular composition of cultures of S. typhimurium grown at 25°C is the same as that of cultures grown at 37°C provided that they are grown in the same medium. In contrast, a shift from the normal to the low temperature range is followed by a complex transient phase of growth: a lag followed by a period of abnormally rapid growth before the characteristic steady-state rate ensues. Similarly, a shift from a temperature in the low range to one in the normal range is followed by an extended period of growth (for about 2.3 doublings) at a low rate prior to growth at the characteristic steady-state rate. The existence of transient growth periods following shifts to or from the low temperature range suggests that cells grown within these two temperature ranges differ substantially in composition.

The pattern of proteins of E. coli changes significantly with growth temperatures outside the normal range (39). The changes that occur at higher temperatures are under the control of the heat shock response discussed in chapter 88. Similarly, a specific group of 13 cold shock proteins, none of which is a heat shock protein, are produced during the period of growth cessation following a shift from 37 to 10°C. Nine of these have been identified as NusA, RecA, the dihydrolipoamide acetyltransferase subunit of pyruvate dehydrogenase, polynucleotide phosphorylase, pyruvate dehydrogenase, initiation factors 2α and 2β (46, 97), the A subunit of DNA gyrase (45), and nucleoid protein, H-NS (54). A 10th polypeptide, designated CS7.4, which is undetectable at 37°C and is the first to be synthesized following a shift down, encodes a hydrophilic protein with 70 amino acids (37) and is homologous to a region of several eukaryotic DNA binding proteins (99). One of these, YB-1, is a transcriptional activator which binds specifically to the CCAAT-containing Y box of HLA class II genes (24). CS7.4 is a transcriptional activator of hns which also contains a CCAAT sequence in its promoter (54), as do nusA and cspA (the gene encoding CS7.4) (75). It seems likely that CS7.4 regulates the expression of cold shock proteins by acting as a transcriptional activator.

Determinants of Growth Temperature Limits

For determinants of the maximum temperature of growth, see chapter 88.

Considerable evidence suggests that the inability to synthesize protein determines the minimum temperature of growth of E. coli and that the sensitive step is initiation of translation. Das and Goldstein (23) showed that after shifting E. coli from 37 to 0°C protein synthesis slowed progressively for 4 h as 70S ribosomes accumulated. Friedman et al. (34) showed that synthesis of f-2 coat protein proceeded to completion but was not initiated at 6°C. Broeze et al. (10) showed that a shift from 37 to 5°C was followed by the accumulation of 70S ribosomes at the expense of polysomes.

PRESSURE

Increased hydrostatic pressure changes the rate and equilibrium of chemical reactions by favoring molecules with smaller molecular volumes. It slows reactions in which the molecular volume of the activated state (Δ V*) is greater than that of the reactants and speeds reactions in which Δ V* is smaller. Similarly, increased pressure shifts the equilibrium depending on whether the molecular volume of the products is greater or less than the volume of the reactants. In many cases molecular volume changes result from changes in solvation (92).

Since most biological reactions are far from equilibrium, the effect of pressure on rate is most evident. In a simple enzyme-catalyzed reaction obeying Michaelis-Menten kinetics, two activation volume changes occur, one associated with the formation of the enzyme-substrate complex and the other associated with the conversion of the enzyme-substrate complex to the activated state. Thus, the effect of pressure on a reaction can differ depending on the enzyme that catalyzes it. For example, Hochachka et al. (41) showed that increased hydrostatic pressure slows the fructose bisphosphatase reaction catalyzed by an enzyme from rainbow trout but speeds the same reaction catalyzed by an enzyme from the abyssal rat-tail fish. However, most biological reactions are slowed at pressures of 30 MPa or more (59). It is reasonable to assume that increased pressure favors particular pathways of protein folding, resulting in enzymes with altered activity.

Jannasch and Taylor (44) suggested that bacteria can be divided into three groups on the basis of their response to increased hydrostatic pressure: (i) those that grow at a pressure of 1 atm (0.1 MPa) and grow more slowly as hydrostatic pressure is increased (the highest pressure under with they can grown is the index of their barotolerance); (ii) those termed barophiles, which grow at 1 atm and grow more rapidly at higher pressures; and (iii) those termed obligate barophiles, which grow only under pressures greater than 1 atm. Representatives of the last two classes have been isolated and studied. One spirillum-like barophile that was isolated from a 5,800-m depth in the ocean tolerates 100 MPa but grows most rapidly over a broad range around 50 MPa and 30-fold more slowly at 1 atm. An obligate barophile isolated at a 10,000-m depth grows well at 100 MPa, grows optimally at 70 MPa, and is unable to grow at hydrostatic pressures less than 35 MPa (44). Other barophiles require pressures greater than 100 MPa for optimum growth (100).

E. coli is a moderately barotolerant organism, growing increasingly slowly as the pressure is increased above 1 atm. The maximum growth pressure for E. coli is about 56 MPa in complex medium and somewhat lower in minimal medium. Even pressures as low as 5 MPa cause a detectable diminution of the growth rate (59). Very high pressures are lethal to E. coli and other bacteria. At 37°C and 200 MPa the exponential death rate constant for E. coli is 0.125 min^–1 (82).

Cellular Target of the Inhibitory Effect of High Pressure

Growth-inhibitory pressures of 69 MPa cause a virtual complete cessation of protein synthesis by E. coli. The pressure-sensitive steps are polysome formation and translocation; other steps of the process, including amino acid transport and transpeptidation, are pressure resistant (86, 87). Interestingly, a streptomycin-resistant strain has increased barotolerance of protein synthesis (73). The proton-translocating ATPase, flagellar formation, and function, as well as cell division and DNA and RNA syntheses, are sensitive to moderate pressures (see reference 98 for a review).

Adaptations to Elevated Hydrostatic Pressure

Welch et al. (98) studied the pattern of protein synthesis following an abrupt increase in pressure to 55 MPa on a culture of E. coli growing anaerobically in a minimal medium. An increase in colony-forming units ceases immediately, but after a lag of approximately 100 min, the optical density of the culture continues to increase but at a diminished rate. It is not clear whether the lag reflects a period of adaptation to elevated pressure. The total number of polypeptides synthesized at the higher pressure is greatly reduced, but 55 of them, termed pressure-induced proteins, are synthesized at an increased differential rate. Eleven pressure-induced proteins are also induced by heat shock and four are induced by cold shock; one, which undergoes the largest induction, has not been observed before. In such experiments, it is not possible to distinguish between direct effects of hydrostatic pressure and indirect effects, for example, the toxic effects of high partial pressures of CO₂.

pH

For additional information on the effects of pH, particularly effects on gene expression, see chapter 96.

In nature a wide range of proton concentrations is encountered, from pH 1 in acidic sulfur springs to pH 11 in soda lakes (53). Bacteria exist in all of these environments, although the number of species (acidophilic bacteria) that can grow in extremely acidic habitats and those (alkalophilic bacteria) that can grow in extremely alkaline ones are more restricted than those (neutrophilic bacteria) that can grow over the mid range, from about pH 5.0 to 9.0. E. coli and S. typhimurium are representative of the large neutrophilic class; they grow at a maximum rate between pH 6.0 and pH 8.0 and more slowly at half a pH unit or so beyond these limits.

The pH is an important parameter of the rate of many types of reactions, not only ionization of acids and bases but also solvolysis and oxidoreduction. As a result, enzymes and other macromolecules function optimally only over a narrow range of pH, usually close to neutrality, a range that is much more restricted than the pH range of the external environment over which growth of the bacterium is possible.

E. coli, as well as many other neutrophiles (71), has evolved remarkably effective mechanisms of homeostasis of intracellular pH. Using ³¹P nuclear magnetic resonance, with P_i and methylphosphonate as probes, Slonczewski et al. (90) measured the internal pH (pH_i) of nongrowing E. coli cells as the pH of the external buffer (pH_o) was slowly decreased from 7.55 to 5.6 and then slowly increased to 8.7. pH_i changes only slightly but progressively as pH_o is raised above or below the crossover pH. These measurements have been confirmed and extended to measurements of Na⁺ gradients at alkaline pH_o (72). Estimations of pH_i by measuring intracellular concentrations of permeant weak acids have given similar results for cells (7, 29, 67, 70) and vesicles (76).

The effect of pH_o on pH_i is magnified in the acid range by permeant acids and in the basic range by permeant bases. Permeant acids (or bases) are organic acids (or bases), the undissociated form of which is lipophilic and, therefore, rapidly diffuses through cell membranes. At equilibrium, the intracellular concentration of the undissociated acid (or base) should equal the extracellular concentration, which is a function of the total concentration, c_o (both dissociated and undissociated), pK, and pH_o. Permeant acids are well-known inhibitors of the growth of microorganisms. Some of them, particularly benzoic, propionic, and sorbic acids, are used as food preservatives. It has long been known that their potency as inhibitors increases as the pH_o decreases (20), in accord with the hypothesis that the intracellular concentration of undissociated acid governs inhibition.

Permeant acids lower pH_i and have been used to manipulate this variable (47, 77). The inhibition of growth of E. coli by permeant acids has been found to correlate closely with pH_i (83). A. G. Marr and N. R. Eaton (unpublished results) found that the rate of growth is strictly determined by the pH_i; the specific growth rate was a sigmoid function of pH_i, with the half maximal rate at a pH_i of 7.2 and complete inhibition below a pH_i of 6.6. Conventional cultivation of E. coli can expose the cells to conditions of pH_o and concentrations of acetic acid, a permeant acid, sufficient to depress the pH_i significantly. At a pH_o of 6.0, 5 mM acetate reduces the specific growth rate by half.

Uncouplers of oxidative phosphorylation such as 2,4-dinitrophenol are permeant protonophores which affect the pH_i by collapsing the ΔpH. In the presence of 200 mM 2,4-dinitrophenol, pH_i is within 0.4 unit of the value of pH_o (32).

Buffering Capacity of the Cell

The buffering capacity of bacteria (B_t) has two components: the surface of the cell (B_o) and its internal contents (B_i). These are defined operationally. B_o is the buffering capacity measured by titrating suspensions of intact cells, whereas B_t is measured by titrating suspensions of cells permeabilized by detergent. B_i is a derived value, the difference between B_t and B_o. This method assumes that the pH_i is not changed during the titration of intact cells.

Careful measurements of this sort (49) comparing E. coli with another neutrophile, Bacillus subtilis, and two obligate alkalophiles, Bacillus firmus and Bacillus alcalophilus, reveal a number of significant facts. B_i is a small fraction of B_t over most of the range of pH_o. Values of B_i vary markedly among bacteria, and those for E. coli are relatively low at most values of pH_o. Although alkalophiles have high values of B_i in the alkaline range, E. coli has a low B_i over the range of pH_o from 6.0 to 8.0 in which it grows at maximal rate; the buffering capacity increases above and below this range (Fig. 2).

Fig. 2Internal buffering capacity (nanoequivalents of H+ per pH unit per milligram of protein) of E. coli as a function of external pH. Squares are data of Krulwich et al. (49). Crosses are calculated from the model.

Source:

Since for E. coli the value of B_i is a small fraction of B_t and, thus, is subject to large errors, we have computed the titration curve for the cytoplasm. We assumed that the concentration of protein is equivalent to 2 M total amino acid with the amino acid composition given by Neidhardt (68), that one-third of the glutamate and aspartate residues are the respective amides, and that the average peptide contains 200 amino acids. To simplify the calculations, the weighted average of pKs was computed for three groups: (i) aspartate, glutamate, and C-terminal carboxyl, 0.135 M with pK = 4.33; (ii) histidine and N-terminal amino, 0.046 M with pK = 7.07; and (iii) arginine, lysine, tyrosine, and cysteine, 0.214 M with pK = 9.75. In most of the calculations, we used 0.05 M glutamate (pKs of 2.2, 4.2, and 9.7) to represent small molecules of the cytoplasm. Thus, the basic calculation for the j th of these n equilibria is given by the following:

in which s and a denote, respectively, the concentration of dissociated and undissociated acid. Now assume that a titrant acid (or base), x, is added, causing a change, ΔpH. At the new equilibrium

and

Sets of equations 2a and 2b were solved numerically by guessing a value of ΔpH and computing a corresponding estimated value of x; the guess was refined until the estimated value approached the actual value with an error in ΔpH of <0.001.

Buffering capacity was computed from the model with the result shown in Fig. 2. Except for higher computed values of B_i at an alkaline pH_o the computations are in reasonable agreement with the experimental results of Krulwich et al. (49). It seems likely that at an alkaline pH_o the cytoplasm is significantly titrated; i.e., pH homeostasis fails.

The model was used to simulate experimental values of pH_i for the wild type and icd mutants of S. typhimurium. The icd mutants accumulate about 50 mM citrate and isocitrate (52) and are more resistant to acid than is the wild type, presumably because of greater buffering capacity. Foster and Hall (33) found that at a pH_o of 3.3 the pH_i of the wild type was 4.4. According to the model, this acidification requires 103 meq of H⁺ per liter of cytoplasm. With 50 mM citrate added, the model predicts a pH_i of 5.41; Foster and Hall found the pH_i of the icd mutant to be 5.5.

Mechanism of Maintaining pH Homeostasis

The value of pH_i is set by a number of factors: the buffering capacity of the cell, the extrusion of protons associated with respiration or hydrolysis of ATP, and the exchange of protons for Na⁺ and K⁺ (5). The value of pH_i is essentially independent of pH_o; thus, as the pH_o is raised, the ΔpH decreases and the membrane potential increases (5, 101). The proportion of transmembrane proton motive force expressed as a ΔpH depends upon the exchange of intracellular protons for extracellular cations (58). Since K⁺ is the principal intracellular cation, it is reasonable to assume that the cytoplasmic concentration of K⁺ contributes in this manner to pH homeostasis but is unlikely to be immediately regulatory since its concentration responds to external osmolarity (27).

The antiport of external protons and internal Na⁺ becomes important as the value of pH_o increases above the crossover. McMorrow et al. (62) found that E. coli failed to grow at pH 8.5 if the concentration of Na⁺ was very low (<15 mM) and that the sensitivity to high pH_o varied inversely with antiporter activity. E. coli has two antiporters specified by nhaA and nhaB (reviewed in reference 85). The expression of nhaA increases with both pH_o and the concentration of Na⁺. The activity of NhaA increases from near zero as the pH_o is increased above the crossover. Thus, NhaA is a candidate regulator of pH_i. However, nhaA nhaB double mutants can grow at an alkaline pH_o but fail to grow if the pH_o is alkaline and the concentration of Na⁺ is high (>100 mM). The means by which Na⁺/H⁺ antiport contributes to pH homeostasis is unclear. It cannot be merely the ionic exchange because the Na⁺ gradient does not decrease monotonically with an increase in pH_o. Pan and Macnab (72), who observed this, suggest that a combination of neutral and electrogenic antiport pumps protons into the cell.

Acid Tolerance

For additional information on acid tolerance, see chapter 96.

If the pH_o is below (or above) the range compatible with normal pH homeostasis or if a permeant acid (or base) is present and the pH_o is below (or above) 7.6, the pH_i will decrease (or increase) significantly. As the pH_i decreases, the rate of growth slows and then stops. For S. typhimurium, a reduction in the pH_i to 5.7 is lethal (32). For E. coli, a pH_i below 5.0 is lethal (91). The bases of inhibition of growth and killing by low pH_i are unknown.

To be successful, an enteric pathogen must survive the low pH (pH 2 to 3) of the stomach for about 2 h (35). Although most strains of E. coli are considerably more resistant to the lethal effect of low pH_o than is S. typhimurium (38), both mount a defense against low pH_o. In principle, this defense could be an increase in the capacity of either pH homeostasis or resistance of the vital target(s) to low pH_i. In fact, both contribute to acid tolerance.

Cells of both E. coli and S. typhimurium from cultures in the stationary phase are more resistant than growing cells to acid, as they are to many other environmental stresses (reviewed in reference 89). Full resistance of stationary-phase cells requires the expression of rpoS (28, 37, 56). rpoS mutants of S. typhimurium acquire some acid resistance in the stationary phase at a low, but not at a neutral or high, pH_o (56). This implies that at least two mechanisms confer acid resistance to cells from cultures in the stationary phase. It is not yet known whether this resistance is due to an increased capacity for pH homeostasis or to a greater resistance of vital targets.

Growing cells are much more sensitive to acid, but acquire resistance if grown at or exposed to a low pH_o. Cells of S. typhimurium exposed to a pH_o of 5.8 for one generation are 100 to 1,000 times more resistant to a subsequent pH_o of 3.3 (32). This treatment was found to augment pH homeostasis, and its effect is dependent upon a functional proton-translocating ATPase (33). Resistance to a pH_o of 3.3 is also acquired by acid shock at a pH_o of 4.5; this resistance does not require augmented pH homeostasis but does require the expression of fur (30, 33). The augmentation of pH homeostasis by exposure to a pH_o of 5.8 apparently permits the subsequent synthesis of proteins at a pH_o of 3.3 that are required for acid resistance and are also induced at a pH_o of 4.5.

Mutants of S. typhimurium that are more resistant than the wild type to direct challenge at a pH_o of 3.3 have been isolated (32). One type, referred to above, is icd and has a higher intracellular buffering capacity than does the wild type. Others are auxotrophic mutants which may have the resistance of cells from the stationary phase. Mutants more sensitive to acid have also been isolated (31). Some of these mutants are defective in induced pH homeostasis; others are defective in the acid shock response; others resemble in phenotype fur mutants; and still others are polA, suggesting that DNA repair is important in acid tolerance.

OSMOTIC STRESS

For additional information on osmotic stress, particularly the effect on gene expression, see chapter 77.

The effect on bacterial growth and survival of osmotic stress correlates well in most cases with a_w, which is defined by Raoult’s law: a_w = P/P_o = n ₁/(n ₁ + n ₂), where P is the vapor pressure of the solution and P_o is that of the solvent, pure water; n ₁ and n ₂ are, respectively, the number of moles of solvent and of ideal solute. Depression of the vapor pressure is but one of the colligative properties of solutions; the others are depression of the freezing point, elevation of the boiling point, and osmotic pressure—the hydrostatic pressure (usually expressed in megapascals) which must be applied to a solution to increase the activity of solvent to equal that of pure solvent. Concentrations of actual solutes may be expressed in terms of the concentration of an ideal solute giving the same a_w and denoted as osmolal (osM). The relationship between a_w and osmolality is given by a_w = x/(x + c) in which c is the concentration of solute in osM and x is the number of moles of water per liter, approximately 55.6.

Species of microorganisms differ significantly with respect to the lower limits of a_w (or upper limits of osmolality) permitting growth; the lower limit tolerated by any microorganism is about 0.62, equivalent to about 30 osM (50). If the solute is not intrinsically toxic, the lower limit is usually the same regardless of whether the solute is ionic or nonionic, although there are exceptions. Saccharomyces rouxii can tolerate a value of a_w of 0.845 if set by glucose but only 0.860 if set by NaCl (7). E. coli falls in the mid range of the tolerance to low a_w exhibited by nonhalophilic bacteria. It is more tolerant than B. subtilis and Pseudomonas aeruginosa but less so than lactobacilli and Staphylococcus aureus.

Osmotic Balance in the Steady State

Unlike animal cells, plant and bacterial cells (with the exception of the mollicutes) are surrounded by a cell wall that resists deformation. Bacteria maintain an intracellular osmotic pressure greater than that of the surrounding medium; osmotic equilibrium is reached by developing internal hydrostatic pressure, termed turgor pressure, such that the a_w of the cytoplasm equals that of the medium.

Maintenance of turgor pressure is essential for growth and division of the cell. Koch has proposed that the stress of turgor pressure on the bacterial cell wall is instrumental in the enlargement of the wall (48). Turgor pressure is communicated from the cytoplasmic membrane to the cell wall either directly or, in bacteria with a periplasm, by means of a gel composed of highly hydrated uncross-linked strands of peptidoglycan that fill space between the cytoplasmic membrane and the wall (40). Periplasmic proteins can diffuse within this gel.

E. coli maintains a turgor pressure of approximately 0.3 MPa over a wide range of osmolality of the external medium. It follows that the concentration of solutes in the cytoplasm must increase linearly with external osmolality in order to maintain turgor pressure. K⁺ is particularly important in this respect (27). The intracellular content of K⁺ increases systematically with the osmolality of the growth medium adjusted with glucose, sucrose, or NaCl. Environmental conditions other than external osmolality—temperature, pH, and external K⁺—have little effect on the intracellular concentration of K⁺ (27). In contrast to intracellular K⁺, the concentration of intracellular Na⁺ does not vary directly with the osmolality of the growth medium; the intracellular concentration of Na⁺ is proportional to the external concentration of Na⁺ and varies with the pH (12).

To maintain electroneutrality, intracellular K⁺ must be balanced by anions. Cells of E. coli grown at low external osmolality (<0.2 osM) retain about two-thirds of the intracellular K⁺ after hypoosmotic shock; the counterions are presumed to be macromolecular anions, particularly RNA, and thus K+ is thought not to be osmotically active(61). Cells grown at high external osmolality accumulate substantial concentrations of glutamate (63, 95). The increase in concentration of K⁺ closely parallels the concentration of glutamate (61). The accumulation of K⁺ (and its counterion, glutamate) is sufficient to maintain proper turgor pressure at external osmolalities up to about 1 osM.

Most strains of E. coli growing at high external osmolality accumulate trehalose (93). Some strains of E. coli K-12 carry an amber mutation which, in the absence of suppressors, prevents trehalose accumulation (79). This and other mutations which prevent the synthesis of trehalose slow or prevent growth in minimal media at high external osmolality (36, 79). The accumulation of trehalose is accompanied by a reduction in the concentration of K⁺ and glutamate (25). The replacement of ionic solutes by a nonionic solute may relieve inhibition of transcription-translation by reducing the ionic strength of the cytoplasm (74). Trehalose not only is a compatible solute permitting growth at high external osmolalities but also specifically reacts with the head groups of phospholipids on the face of membranes, thereby stabilizing the membrane (18, 19).

One cytoplasmic solute, the polyamine putrescine (1,4-diaminobutane), decreases as the external osmolality is increased. Within 5 min after the medium is increased to 0.6 osM, the intracellular concentration is reduced fivefold by excretion of putrescine into the medium (66). Putrescine is replaced by K⁺. Munro et al^. (66) suggested that the physiological advantage of replacing divalent putrescine with monovalent K⁺ is that it allows an increase in cytoplasmic osmolality with a less than proportionate increase in intracellular ionic strength.

Thus, for E. coli (and presumably for S. typhimurium) growing in minimal medium at high external osmolality, the osmolality of the cytoplasm is increased by the accumulation of K⁺, glutamate, and trehalose, and the ionic strength is reduced by the exchange of K+ for putrescine. These adjustments permit growth up to an external osmolality of at least 2 osM (61). However, tolerance of high external osmolality can be increased by the presence of certain organic compounds, called osmotic protectants, in the medium.

Osmotic Protectants

In 1955 Christian (14) made the significant discovery that the addition of proline to the medium increased the tolerance of Salmonella spp. to high external osmolality. He showed that exogenous proline was essential for the growth of Salmonella oranienburg in a defined medium at a_w values of less than 0.97, or about 1.7 osM (15), and that proline stimulated the respiration of this and other bacteria at low a_w values (17). The proline content of cells was inversely related to the a_w value of the medium, reaching a value of 1.5 mmol/g (dry weight), or about 0.6 osM, when the a_w of the medium was 0.95, or about 3 osM (16). Thus, proline is an osmotic protectant; bacteria grow in media of higher osmolality if proline is present in the medium.

Csonka (21) isolated mutant strains of S. typhimurium that were resistant to l-azetidine-2-carboxylate because they overproduced l-proline. These strains grew in media of higher osmolality (0.8 M NaCl) than did their parent. In a minimal medium with 0.65 M NaCl, the specific growth rate of the parent was 0.1 h^–1 while that of the mutant was 0.29 h^–1. The addition of proline to a medium containing 0.65 M NaCl doubled the growth rate of wild-type S. typhimurium; none of the other 19 common amino acids had this effect (21).

Other compounds known to accumulate in plants subjected to osmotic stress were tested for their ability to increase the osmotic tolerance of E. coli. Glycine betaine, choline, betaine aldehyde, trimethyl-γ-aminobutyrate, and proline betaine were effective (57). Subsequently β-alanine betaine, taurine betaine, γ-amino crotonic acid betaine, carnitine, choline-O-sulfate, 3-(N-morpholino)propanesulfonate, and dimethylthetin have been added to the list of osmotic protectants of E. coli and S. typhimurium (reviewed in reference 22).

Glycine betaine is the most effective osmotic protectant of E. coli and S. typhimurium. S. typhimurium at 30°C in a minimal medium containing 0.8 M NaCl has a specific growth rate of 0.10 h^–1. If the medium is supplemented with proline, the rate increases to 0.17 h^–1, but if supplemented with glycine betaine, the rate increases to 0.30 h^–1 (11). Choline and betaine aldehyde are effective as a consequence of oxidation to glycine betaine. Choline is converted to betaine aldehyde by a membrane-associated, O₂-dependent enzyme, and the product, betaine aldehyde, is converted to glycine betaine by a soluble NADP-dependent enzyme. The level of both of these enzymes increases with the osmolality of the medium (57).

At high external osmolality, the concentration of osmotic protectants such as glycine betaine and endogenous compatible solutes such as trehalose is equivalent to a substantial fraction of external osmolality. Their accumulation reduces the cytoplasmic concentration of ionic solutes K⁺ and glutamate (3, 13, 25). The accumulation of glycine betaine reduces the concentration of endogenous trehalose as well as ionic solutes (13, 25). The reduction in the concentration of K⁺ or in ionic strength by compatible solutes may be directly responsible for an increasing growth rate through relief of inhibition of transcription-translation or other enzymatic activities (74, 94). However, Cayley et al. (13) argue that glycine betaine increases the fraction of free water.

Hyperosmotic Shock

The response to an abrupt increase in external osmolality should reveal the mechanisms underlying the osmotic homeostasis observed in the steady state. If the osmolality of a culture of E. coli growing in minimal medium is abruptly and substantially increased to, say, 1 osM, the cells almost instantaneously lose water by plasmolysis, increasing the turbidity of the culture; the synthesis of macromolecules is inhibited (51); and the rate of respiration decreases (42), possibly as a consequence of the reduced rate of biosynthesis.

The kinetics of change of amounts and concentrations of the principal osmolytes were measured by Dinnbier et al. (25). The amount of K⁺ rises rapidly to a maximum by about 20 min. The concentration of K⁺, which reflects both the amount of K⁺ and the water content of the cell, reaches a maximum within a few minutes. The concentration of glutamate rises somewhat more slowly. The pH_i transiently increases, reaching a maximum of pH 8.3 after a few minutes, and then then declines rapidly to the normal value; this excursion of pH_i may reflect faster kinetics for the K⁺/H⁺ exchange than for the accumulation of the counterion, glutamate. Recovery from plasmolysis coincides with these events. The concentration of trehalose increases more slowly, reaching a maximum at about 1 h. The accumulation of trehalose is accompanied by a loss of part of the previously accumulated K⁺ and glutamate. Finally, the addition of external proline or glycine betaine causes a loss of trehalose and further loss of K⁺.

The resumption of biosynthesis lags behind the recovery from plasmolysis and coincides with the accumulation of trehalose and loss of K⁺ and glutamate. This fact suggests that the high concentration of ionic (noncompatible) solutes inhibits biosynthesis.

Regulation of the Osmotic Response

The increase in the intracellular concentration of K⁺ appears to be the key event in the response to hyperosmotic shock. The increased concentration results in part from plasmolysis and in part by accumulation. Accumulation is thought to result from stimulation of the K⁺ transport systems, Trk and Kdp, signaled by a decrease in turgor pressure (78). However, intracellular K⁺ in growing cells is dynamic: K⁺ is accumulated by transport and lost by efflux. The cell membrane contains stretch-activated ion channels (102). Plasmolysis would close such channels and contribute to the accumulation of K⁺.

The accumulation of glutamate depends upon the accumulation of K⁺. Glutamate is not accumulated following hyperosmotic shock in a medium lacking K⁺ (61). Tempest et al. (95) suggested that the accumulation of glutamate might result from an increase in the activity of glutamate dehydrogenase. They found that this enzyme has an unusually sharp dependence on pH, being maximally active at pH 8 and nearly inactive at pH 7. The alternative route for glutamate synthesis via glutamine shows a similar dependence on pH (64). The accumulation of K⁺ is accompanied by a marked but transient increase in pH_i (25, 61), which would increase the rate of synthesis of glutamate by either pathway. However, McLaggan et al. (61) argue that the inhibition of biosynthesis by hyperosmotic shock is sufficient to account for the accumulation of glutamate.

Booth and Higgins (6) have proposed that the high concentration of K⁺ and glutamate is the main secondary signal in adjustment to hyperosmotic shock. High ionic strength inhibits transcription from normal promoters but increases transcription from osmotically activated promoters such as proU (74), possibly by release of the DNA binding protein H-NS (65). The consequence of the secondary signal is replacement of ionic solutes, K⁺ and glutamate, by compatible solutes such as trehalose (by synthesis) or glycine betaine (by transport). This replacement is critical to a resumption of growth at high osmolality.

Hypoosmotic Shock

Low external osmolality per se is compatible with survival and growth of E. coli and S. typhimurium; however, a modest, abrupt decrease in external osmolality of –0.2 to –0.3 osM causes the loss of low-molecular-weight internal solutes but not macromolecules (9, 27, 96) and can be lethal (4). Amino acids and nucleotide pools (9) or previously accumulated solutes such as methylthiogalactoside or α-methylglucoside (96) are lost almost completely. If the extracellular fluid contains salts, almost all K⁺ is lost (26), but if salts are absent, only a fraction is lost (61). K⁺ is thought to be retained by electrostatic attraction of polyanions, principally nucleic acids (61), but can exchange with Na⁺ in the extracellular fluid. Bakker (2) has found that if a culture of E. coli growing in 0.4 M NaCl is abruptly diluted 1:1 with water, 90% of the K⁺ and glutamate and 60% of the trehalose are lost from the cells. Under this condition the loss of solutes is selective: ATP and alanine are retained.

The immediate consequence of hypoosmotic shock is a flux of water into the cell, increasing the turgor. The increased turgor would stretch the wall and the underlying membrane. Britten (8) suggested that such stretching would either cause transient "cracks" or perhaps stretch existing pores in the membrane. This behavior may reflect stretch-activated channels (102). The increased permeability is only transitory: after at most a few minutes the cells can begin to accumulate solutes (96). It seems likely that solute loss results from the opening of stretch-activated channels that may be important to normal osmoregulation. Loss of intracellular solutes by hypoosmotic shock is potentiated by low temperature (26, 96), and temperature shock under isosmotic conditions causes solute loss (55). Low temperature may decrease the fluidity of the cell membrane and thereby contribute to opening stretch-activated channels.