Relation of growth and protein synthesis to the adenylate energy charge in an adenine-requiring mutant of Escherichia coli.

When Escherichia coli K-12 strain PC 0294, which is unable to synthesize adenine nucleotides, thereonine, proline, and leucine, was starved for adenine, the concentrations of ATP, ADP, and AMP fell rapidly. The adenylate energy charge was not affected until the adenine nucleotide pool fell to about 30% of its normal value. Similarly, cells in an adenine-limited chemostat grew at intracellular adenine nucleotide pool values as low as 30% of normal, but the energy charge in all cases was approximately 0.90. Incorporation of [14C]leucine into protein in adenine-starved cells continued rapidly as long as the energy charge was in or near its normal range, even when the concentration of ATP was 30% to 10% of its normal value. When glucose was added to cells that had been resuspended in medium lacking both glucose and adenine, the energy charge rose rapidly but the concentration of ATP fell, presumably because of nucleic acid synthesis. The rate of [14C]leucine incorporation into protein rose rapidly while adenine nucleotide concentrations fell, and then declined roughly in parallel with the energy charge. At a given value of energy and of total concentration of adenine nucleotides, the rate of protein synthesis may, of course, vary with the concentrations of precursors and modifiers; thus, the rate cannot be predicted from knowledge only of the energy charge. Our results suggest, however, that the rate of protein synthesis and the capacity for growth are much more sensitive to changes in the value of the energy charge than to changes in the concentration of ATP. Growth occurs when the ATP concentration is reduced to one-third of its normal value, but has not been observed when the energy charge has fallen by as much as 10%.

When Escherichia coli K-12 strain PC 0294, which is unable to synthesize adenine nucleotides, threonine, proline, and leucine, was starved for adenine, the concentrations of ATP, ADP, and AMP fell rapidly.
The adenylate energy charge was not affected until the adenine nucleotide pool fell to about 30% of its normal value. Similarly, cells in an adenine-limited chemostat grew at intracellular adenine nucleotide pool values as low as 30% of normal, but the energy charge in all cases was approximately 0.90. Incorporation of [Y jleucine into protein in adenine-starved cells continued rapidly as long as the energy charge was in or near its normal range, even when the concentration of ATP was 30% to 10% of its normal value. When glucose was added to cells that had been resuspended in medium lacking both glucose and adenine, the energy charge rose rapidly but the concentration of ATP fell, presumably because of nucleic acid synthesis.
The rate of ['YZ]leucine incorporation into protein rose rapidly while adenine nucleotide concentrations fell, and then declined roughly in parallel with the energy charge. At a given value of energy and of total concentration of adenine nucleotides, the rate of protein synthesis may, of course, vary with the concentrations of precursors and modifiers; thus, the rate cannot be predicted from knowledge only of the energy charge. Our results suggest, however, that the rate of protein synthesis and the capacity for growth are much more sensitive to changes in the value of the energy charge than to changes in the concentration of ATP. Growth occurs when the ATP concentration is reduced to one-third of its normal valde, but has not been observed when the energy charge has fallen by as much as 10%.
The adenine nucleotides stoichiometrically couple energyproducing and energy-utilizing metabolic sequences. The adenylate energy charge, defined as [(ATP) + s(ADP)]/ [(ATP) + (ADP) + (AMP)], is a linear measure of the amount of metabolic energy stored in the adenine nucleotide pool. Studies in oitro have shown that the catalytic properties of enzymes from both catabolic and biosynthetic sequences are modified by changes in the energy charge (l-8). Studies in uiuo have shown that the energy charge is stabilized in the range 0.8 to 0.95 in normal cells (9) as predicted from the kinetic studies (10). When cells are subjected to metabolic stress such as limitation in their energy source, nitrogen source, or phosphate supply, or a sudden increase in the concentration of a phosphoryl acceptor, there is usually only a moderate change in the energy charge, indicating that it is a tightly controlled parameter (9,(11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23).
Enzymes that show a response to the value of the energy charge in uitro have been found to be insensitive to variations in the total adenine nucleotide concentration (and thus to the concentration of ATP); within the physiological range of adenylate concentrations, the activity of the enzyme depends on ratios of the nucleotide concentrations (the energy charge) rather than on absolute concentrations (see under "Discussion" for references).
These observations strongly support the prediction, based on the metabolic roles of the adenine nucleotides as coupling agents between metabolic sequences, that the energy charge rather than absolute concentrations should be the effective regulatory parameter in uiuo. This prediction is difficult to test directly because these variables tend to rise and fall together in the intact cell. In this paper, we report that this covariance of the ATP level and the energy charge can be partially uncoupled by adenine limitation of an adenine auxotroph of Escherichia coli. The results suggest that the rate of protein synthesis and the capacity for growth are more closely related to the energy charge value than to the ATP concentration.

MATERIALS AND METHODS
Organism-Escherichi coli K-12 strain PC 0294, purA-, thr-, leu-, proA-, gal-, strA-, mtl-, xyl-, was a gift from Dr. A. J. Clark. The metabolic block in the purine pathway is at the adenylosuccinate synthetase step. Growth Conditions-Cells were grown aerobically in a standard 6930 6931 growth medium containing 5 g of KH,PO,, 17 g of K,HPO,.3H,O, 2 g of (NH,),SO,, 200 mg of MgCl,.GH,O, 5 g of glucose, 200 mg of L-proline, 50 mg of L-threonine, 50 mg of L-leucine, 20 mg of adenine, and 0.2 mg of thiamine hydrochloride/liter. For glucose-limited growth, 300 mg of glucose were used per liter, and for adenine-limited growth, 7.5 mg of adenine were used per liter. The temperature for each experiment is indicated in the figure legends. Cells were checked for adenine requirement at the conclusion of each experiment to guard against reversion.
For the chemostat experiments, a cylindrical glass vessel 15.5 cm long and 11 cm in diameter, with integral baffles and water jacket, was used. The top was sealed with a rubber stopper. Holes were drilled in the stopper to accommodate the level controller probes, tubing for sampling, air input and exhaust, medium inflow and outflow, and antifoam addition. ("Antifoam A" from Dow Corning Corp., a silicone supplied in an aerosol spray can, was removed from the can and centrifuged at 10,000 x g for 15 min. Only the supernatant fraction was used.) The volume of the culture in the chemostat was maintained at 500 ml by peristaltic pumps (one controlling inflow and one controlling outflow) functioning in conjunction with a Cole-Parmer Dyna-Sense level controller (model 7186). Aeration of the culture was achieved by passing moist air through a fritted glass sparger at a rate of 1 liter/min. A rapidly rotating Teflon-coated magnet was used for thorough mixing of the culture.
Perchloric Acid Extraction of Adenine Nucleotides-A l-ml sample of the bacterial culture was removed through narrow tubing by suction and rapidly pipetted into 0.2 ml of 35% HClO,. After 30 min at O", the extract was frozen at -70". Within 24 hours, the extract was thawed, thoroughly mixed, and centrifuged at 12,000 x g for 3 to 4 min at about 4O. A sample of the supernatant fluid (0.8 ml) was neutralized with about 0.26 ml of a solution containing 2.6 M KOH and 0.58 M KHCO,. After at least 15 min at 4', the KClO, was removed by centrifugation. The samples were stored at -70" and assayed within 5 days of extraction.
The ranges of recovered values for ATP, ADP, or AMP added to the perchloric acid at the time of cell addition and carried through the entire procedure were 95 to 109%, 101 to 103%, and 92 to lOO%, respectively.
This extraction procedure has been modified from that previously used in this laboratory (9) by centrifugation of the perchloric acid extract. This removal of acid-precipitable material before neutralization of the extract improves the recovery of externally added adenine nucleotides, presumably by removal of enzymes that are otherwise reactivated on neutralization. With the modified procedure described here, the energy charge in exponentially growing E. coli was found to be about 0.9 rather than about 0.8 as previously reported. While this manuscript was in preparation, a paper appeared (24) recommending this same procedural change and reporting much larger losses of ATP in neutralized whole extracts of Bacillus breuis than we have observed in E. coli extracts.
Luciferuse Assay-ATP was assayed by the luciferase reaction using a Luminescence Biometer (E. I. DuPont de Nemours and Co.). After enzymic conversion of ADP, or of both AMP and ADP, to ATP, the sums (ATP + ADP) and (ATP + ADP + AMP) were determined using the same assay. These assay procedures were essentially as described previously (9). Determination of Intracellular Adenine Nucleotide Levels-The intracellular adenine nucleotide levels were determined by taking the difference between amounts in the complete culture and the amount in the medium alone after cells were removed by filtration. This correction was especially important when the intracellular pool was small. A Swinnex-25 filter holder (Millipore Corp.) containing two Millipore membrane filters-a 1.2.pm filter laid on top of a 0.45~pm filter-was attached to a syringe. A culture aliquot was added directly to the syringe and then filtered. A sample of the filtrate was added to perchloric acid and treated as described above.
For growing cells, the adenine nucleotide level in the medium usually represented 5% or less of the adenine nucleotide level in the complete culture. For adenine-starved cells, the medium contained up to 35% of the adenine nucleotide level in the complete culture. This increased percentage was due to a decrease in the intracellular adenine nucleotides rather than an increase in the extracellular nucleotides. ADP and AMP represented the majority of the extracellular adenine nucleotides.
Computations-For each sample, the ATP, (ATP + ADP), and (ATP + ADP + AMP) values were each assayed five or six times. The using the appropriate propagation of errors formula derived according to Wilson (25) using the defining energy charge equation in the form, energy charge = [(ATP) + (ATP + ADP)1/2(ATP + ADP + AMP). was 0.01 to 0.02 .energy charge unit. 1 Protein Estimation-The total protein was measured by a modified biuret procedure (26). A l-ml sample of culture was added to 1 ml of 10% trichloroacetic acid. The precipitate was collected by centrifugation at 12,000 x g for 10 min, then resuspended in phosphate buffer and assayed. Measurement of Protein Synthesis-A l-ml sample from the bacterial culture was pipetted into a 16.mm test tube containing about 0.2 PCi (in 0.02 ml) of ["C]leucine (final specific activity was about 5 pCi/mol), and incubated for 1 min in a shaking water bath. The incorporation was stopped by the addition of 1 ml of a 20% trichloroacetic acid solution containing 0.5% leucine. Samples were incubated for 20 min at about 90" to hydrolyze any ["Clleucyl-tRNA (27), then cooled to O-4". The precipitated material was collected on glass fiber filters (Whatman GF/C, 2.4 cm diameter) which had been soaked '/L to 2 hours in a solution containing 5% trichloroacetic acid and 0.5% leucine, and then washed with six lo-ml aliquots of a solution containing 5% trichloroacetic acid and 0.05% leucine. A Millipore sampling manifold (model 3025) was used in the collection of the precipitates.
Filters were dried under a heat lamp for at least 25 min. Each filter was placed in a scintillation vial, and 6 ml of a toluene-Omnifluor solution (4 g/liter) were added. The vials were counted with a Packard scintillation spectrometer, model 2002.
Since the unlabeled leucine in the culture medium decreased as cell growth proceeded, the specific activity of leucine in samples to which a fixed amount of ["Clleucine was added increased during an experiment. The results were corrected for this change in specific activity as described in the appendix.' The corrected data differed from the raw data by, at most, 35% for control cultures and by, at most, 15% for nutrient-limited cultures. In no case would the interpretation of the results have been significantly different if the corrections had not been made.

RESULTS
Glucose Staruation-In a preliminary experiment, the adenine auxotroph E. coli PC 0294 was grown in medium containing a limiting amount of glucose. During exponential growth, the ATP level was about 10 nmol/mg of protein, which is similar (assuming that protein represents 50%) of the cell's dry weight (28) to previously published values (29). The sum, ATP + ADP + AMP, was about 11 nmol/mg of protein.
The energy charge was about 0.90. Within 1 hour after glucose was depleted, the energy charge had decreased to about 0.80, and the total adenine nucleotide concentration had decreased to less than 50% of normal.
These results are similar to those reported by Chapman et al. (9) for a wild type E. coli strain, except that the energy charge values are consistently higher by about 0.1 unit, a consequence of the improvement in the assay procedure noted above. Further experiments were designed to study the extent to which the energy charge is stabilized during large changes in the total adenine nucleotide concentration when an energy source is available.
Adenine Starvation-E. coli PC 0294 was grown with a 'The derivation of an equation for use in correction of specific activities is presented in an appendix immediately following this paper (p. 6938). Material published in miniprint form can be easily read with the aid of a large-field reading glass of a type readily available at most onticians.
For the convenience of those who prefer to obtain supplementary material in the form of 6 pages of full size photocopy, these same data are available as JBC as a function of the adenylate concentration (Fig. 3). The cells continue to take up oxygen at a rate at least 70% of the control for 50 min after resuspension without adenine at 25" (Fig. 4). Cells remain viable for at least 60 min during adenine starvation, and recover rapidly. Within 2 min after addition of adenine, the energy charge was normal, and the adenine nucleotide level was at least 80% of normal (Fig. 5).
Adenine-limited Steady State Growth-When the adenine auxotroph was grown in adenine-limited chemostat cultures (Fig. 6)  At intervals, the oxygen probe was placed in each chamber, preventing aeration, and the rate of oxygen uptake monitored for 5 min. The procedure was repeated, comparing cells resuspended in complete medium and glucose-deficient medium. The rate of oxygen uptake in each case was plotted as a percentage of the rate for cells in complete medium, which was 11% of the saturating oxygen concentration per min for a suspension with an absorbance at 540 nm of 1.0. [YZ]leucine incorporation declined to less than 10% of normal during the first 10 min of glucose starvation (Fig. 9). The energy charge was 0. incorporation value, the ATP concentration or energy charge was read off the appropriate curve in Fig. 7. The average of the first five ATP values in Fig. 7, 10.6 nmol/mg of protein, was designated 6934 during growth was 0.93 + 0.02). The ATP concentration during this period was about 80% of the value during growth. In Fig. 10, the rate of ["Clleucine incorporation during the first 20 min of glucose starvation has been plotted as a function of the energy charge or ATP concentration. Comparison of these results with those shown in Fig. 8 indicates that protein synthesis is inhibited more rapidly and severely with regard both to time and to ATP concentration or energy charge. When exponentially growing cells of E. coli PC 0294 were collected and suspended in medium lacking both adenine and glucose, the energy charge was about 0.80 (about 0.1 unit below the control culture), the total adenine nucleotide concentration was about 80% of control, and the rate of [14C]leucine incorporation was about 10% of control (Fig. 11)  Growth stopped about 6 hours later at an A,,, of 0.56. Sampling began just as growth was stopping. Samples were also taken from a control culture containing the standard glucose concentration (0.5%). The cultures were incubated at 25", and the doubling time was 3.0 hours in both cultures. Protein synthesis was plotted as nanomoles of ["Clleutine incorporated per mg of protein during a 1-min pulse. The amount of radioactivity incorporated was converted to the amount of leucine incorporated as described in the appendix. pyridine nucleotides has been reported for dehydrogenases (32,33) and response to the adenylate energy charge has been reported for many enzymes (l-8, 32, 33) studied in vitro.
Enzymes that respond to variation in the energy charge (or in the pyridine nucleotide ratio) have been found to be insensitive to changes in the absolute concentration of the adenine nucleotides (or pyridine nucleotides) over considerable ranges around the presumed physiological levels (4, 7,8,[32][33][34]. These results support the predictions based on the logic of metabolic relationships. A  (Fig. 9) have been replotted. The average ATP concentration for five samples from the control culture containing adequate glucose, 9.5 nmol/mg of protein, was taken as 100%. TIME imlnl FIG. 11. Rate of protein synthesis, adenine nucleotide concentration, and energy charge in Escherichia coli PC 0294 resuspended in medium lacking adenine and glucose. Exponentially growing cells (doubling time 2.7 hours at 27") were harvested by centrifugation (4") and resuspended at 27" in medium containing standard supplement concentrations, except that the leucine concentration was 5 mg/liter and neither adenine nor glucose was present. For the control culture, adenine and glucose were present. Protein synthesis was plotted as nanomoles of ["Clleucine incorporated per mg of protein during a lmin pulse. The amount of radioactivity incorporated was converted to the amount of leucine incorporated as described in the appendix. incorporation value, the ATP concentration or the energy charge was read off the appropriate curve in Fig. 11. The average ATP concentration for five samples from the control suspension (in complete medium), 10.9 nmol/mg of protein, was taken as 100%.
correlation and control of processes in intact cells is difficult to obtain.
If the adenylate pool (the sum of concentrations of ATP, ADP, and AMP) remains constant, an increase or decrease in energy charge will be accompanied by a change in the same direction in the concentration of ATP. Further, the adenylate pool tends to rise and fall with the energy charge (17-23). The basis for this relationship is not well understood, but it probably rests in part on the properties of adenylate deaminase (35) and AMP nucleosidase (36). Whatever the explanation, this covariation increases the magnitude of change in ATP concentration that may accompany a change in the value of the adenylate energy charge. By use of an adenine auxotroph, we were able to obtain a degree of control of the adenine nucleotide pool level and to decouple partially the concentration of ATP from the energy charge. Measurements on batch cultures and cell resuspensions (Figs. 1-3, 5) show that the adenylate pool may drop to about 30% of its normal value before the energy charge is measurably affected. The same result was observed in cells grown under steady state conditions in an adenine-limited chemostat (Fig.  6). In the chemostat, the pool level was reduced to 60% of normal with very little effect on growth rate, and growth continued even when the pool level was 30% of normal. The similarity of the lowest pool level that supported growth in the chemostat and the level at which a normal value of energy charge could no longer be maintained in a batch culture starved for adenine (Fig. 2) is notable. We have never observed growth when the energy charge has decreased by as much as 10% from its value in a rapidly growing culture. It thus seems that a change of less than 10% in the energy charge affects 6935 does a 3-fold change in the concentration of ATP. This much greater sensitivity of metabolism and growth to changes in the value of the energy charge than to changes in ATP concentration agrees with predictions based both on metabolic functional needs and on the results of enzyme studies in uitro.
Although the range of adenine nucleotide pool levels over which the energy charge is stabilized is impressive, it must, of course, have limits.
The specific reasons for loss of stabilization when the pool falls to about 30% of its normal level are not known. The simplest explanation would be that the concentration of ADP drops to the point where it kinetically limits the regeneration of ATP. It is perhaps more likely that the effect is much more complex than this. The shapes of the curves for response of regulatory enzymes to variation in energy charge depend to some extent on virtual saturation of the ATP and ADP or AMP sites, and there may be differences between enzymes in responses to energy charge at nonsaturating nucleotide levels. Such levels are probably never reached in uiuo; we have not observed the pool to fall below about 50% of its normal value except in the adenine auxotroph.
Mechanisms for stabilizing the energy charge when the adenylate pool level is below 50% of normal will obviously not have evolved if the cell's homeostatic mechanisms maintain the pool level above 50%.
The effect of energy charge on growth is not, of course, exerted solely at the level of macromolecular synthesis. Rather, it must be the sum of effects on many processes, including the synthesis of the necessary building blocks, their activation (3), and their assembly into macromolecular products. When the energy charge fell in the adenine auxotroph as a result of extensive depletion of the adenylate pool, the rate of incorporation of labeled leucine into the protein continued nearly unabated while the charge decreased by about XI%, and then declined relatively slowly (Fig. 7). Presumably, the levels of intermediates fell during the period of declining protein synthesis.
The level of messenger RNA also probably decreased, since either the lower energy charge or the low concentration of ATP as a substrate would be expected to prevent synthesis of RNA at a significant rate. In any case, although some protein synthesis occurred after the charge fell below the normal range, the results were compatible with the generalization that growth does not occur under these conditions, since leucine incorporation reached a very low level in much less than one generation.
When cells were starved for glucose rather than for adenine, the rate of incorporation of leucine into protein fell much more sharply than when adenine was limiting (Fig. 9). This difference presumably reflects the general deficiency of metabolic intermediates in the glucose-starved cells. Since the decrease in size of the adenylate pool is relatively small in these cells, the difference between incorporation as a function of ATP concentration and as a function of energy charge is much less pronounced in the glucose-starved cells (Fig. 10) than in those starved for adenine (Fig. 8).
When cells of the adenine auxotroph were suspended in medium lacking both glucose and adenine, the energy charge stabilized at a level below the normal growth range, and the constancy of the adenylate pool level indicates that there was no net synthesis of nucleic acid. By addition of glucose, it was possible to break the usual covariance of ATP concentration and energy charge. The energy charge rose rapidly because of the availability of a metabolic energy source. With the energy over-all metabolism as reflected in growth more severely than charge in its normal range, macromolecular synthesis should by guest on March 24, 2020 http://www.jbc.org/ 6936 resume, and the expected decrease in the adenine nucleotide pool level is seen in Fig. 11. Protein synthesis also resumed, reaching nearly the rate seen in cells growing exponentially despite the rapid fall in the concentration of ATP. The rate of incorporation remained relatively high even after the concentration had fallen to about 10% of the level seen in growing cells, and its decline was approximately parallel with the decline in energy charge.
Since the first discussion of the adenylate energy charge as a metabolic control parameter, it has been emphasized that the effects of energy charge in uiuo on any reaction must be modulated by other regulatory effects (10). The uniqueness of energy charge regulation lies in its ubiquity; virtually all metabolic sequences either utilize or regenerate ATP, and all may be expected to be regulated by the energy charge. But for any reaction or any sequence, energy charge is only one of the control parameters.
Since the rate of the reaction or sequence is determined by interaction of many regulatory inputs, it cannot be predicted from knowledge of the extent of variation of any one of them. This situation is shown clearly in Fig. 11. When glucose was limiting, the energy charge was about 0.8 and there was virtually no incorporation of leucine. After addition of glucose, when the energy charge had fallen to 0.8 because of adenine deficiency (at about 30 min in the figure), the rate of leucine synthesis was over half as fast as in the growing culture. The difference between these two situations must be the much higher levels of all intermediates in the latter case. Biological growth involves a more complex coordination of chemical reactions than any other known event or process. Clearly, very many concentrations and other parameters must be within acceptable limits for growth to occur, and the rate of growth at any given moment will be a function of the values of whichever parameters happen at that time to deviate from their optimal values. Further, the optimal value for any parameter may depend to some extent on the present values of others. It is to be expected that the acceptable ranges of some parameters will be relatively wide, and those of others will be narrow, depending on the metabolic functions of the compounds involved. Because the energy charge is one of the regulatory inputs to most, and probably all, metabolic sequences, even a slight decrease in its value must lead, through complex networks of primary, secondary, and more indirect effects, to changes in the rates of most sequences and in the concentrations of many intermediates. As a consequence of the amplification that must result from these metabolic cascades, it may be expected that the acceptable ranges of the ATP:ADP and ATP:AMP ratios and of the energy charge should be extremely narrow.
That is, such an integrated process as growth should be strongly dependent on the value of the energy charge, and growth should cut off sharply as the energy charge declines past a narrow critical range. The results of the chemostat experiments (Fig. 6) are consistent with this expectation; growth was not observed at energy charge values detectably below normal. However, it is obvious that, when the energy charge is in the range compatible with growth, many other factors will affect both the ability of the cell to grow and the rate of growth. No net growth is possible, for example, in the absence of a nitrogen source, no matter how high the energy cha&ge or the concentrations of intermediates of glycolysis and the citrate cycle. In the chemostat experiments reported here, the only ultimate limiting factor is the supply of adenine nucleotides.
Thus, the situation is much simpler than in a glucose-limited culture, for example, where energy and nearly all metabolic intermediates will be in short supply. Even in the adenine-limited cultures, however, there will be secondary effects, and in view of the involvement of ATP in all metabolic sequences, they may be extensive.
The decrease in growth rate observed in the chemostat experiment may have resulted rather directly from effects of lowered adenylate concentration on membrane function or from a decrease in the ability of the cell to synthesize nucleic acids, histidine, or nucleotide cofactors (which incorporate some of the atoms of ATP), or, less directly, from lowered concentrations of other intermediates or from a slightly lowered energy charge. A complex interaction of several of these effects seems most likely. In particular, it is impossible to determine from our results whether a slight decrease in energy charge accompanied, and perhaps participated in causing, the decrease in growth rate. There is no evidence, and no basis for conjecture, as to whether the value of the energy charge, once it was in the range allowing growth, was a rate-limiting factor. Whatever the reasons for the decrease in growth rate as adenine became more severely limiting, the primary importance of the chemostat experiment lies in the observation that growth occurred over a 3-fold range of adenine nucleotide concentrations, but only when the energy charge was essentially constant.
Thus, when the only primary limiting factor is the availability of adenine nucleotides, the intracellular concentrations of these compounds can be forced far below the normal range without preventing growth; but growth appears to be possible only when the adenylate concentration ratios, and the energy charge, are maintained within narrow normal limits.