Control Dynamics of fi-galactosidase in Relation to the Bacterial Cell Cycle

Transcription

1 European J. Biochem. 10 (1969) Control Dynamics of fi-galactosidase in Relation to the Bacterial Cell Cycle B. C. GOODWIN School of Biological Sciences, University of Sussex (Received January 24/June 25,1969) The dynamic behaviour of /3-galactosidase was studied in chemostat cultures of Escherichia coli B synchronized with respect to cell division by periodic phosphate feeding. Stable oscillations in the enzyme were observed when lactose was present at 0.020/,. Perturbation experiments were used to study the stability of these states, and transient high-frequency oscillatory responses were observed. These results show that the lactose control circuit has different dynamic states, which may be due to differences in the degree of damping or in the frequency of autonomous oscillations. The results are interpreted in terms of the interactions between the lactose control circuit and the cyclic physiology of the cell during the growth and division cycle. It has recently been observed that a number of enzymes are synthesized in a discontinuous manner during the cell cycle in microorganisms [l--51. Most of the enzymes which show such discontinuous or periodic synthesis in bacteria are known to be under control by feedback repression. A study of the dynamics of aspartate transcarbamylase synthesis in synchronized cultures ofbacillus subtilis by Masters and Donachie [6] suggested that the periodicity in this enzyme arose from oscillations in the feedback circuit controlling its synthesis. Such oscillations had been anticipated on theoretical grounds [7], and a model showing how timed, periodic enzyme synthesis could occur in bacteria during the cell cycle was suggested [S]. This model made a number of predictions regarding the dynamic behaviour of molecular control circuits in cells, and the results of experiments designed to investigate some of these are presented in this paper. MATERIALS AND METHODS Organisms E. coli B/I and E. coli 3300 were used in the experiments. Enzyme Assays The measurement of B-galactosidase activity was based upon the method of Cohn [9], and his units of enzyme activity were used. Enzyms. B-Galactosidase or B-D-galactoside galactohydrolase (EC ) ; aspartate transcarbamylase or carbamoylphosphate :L-aspartate carbamoyltransferase (EC ). Growth and Cell Counts Growth of the bacterial cultures was measured by absorbance at 540nm in an EEL photometer. An absorbance of 0.2 corresponds to about 2 ~ 1 0 ~ cellsld and a mass of about 0.35 mg-dry weight of bacteria per ml. Cell counts were made with a Coulter Counter Model "A", using a 30 p orifice tube. Culture Apparatus The bacteria were grown and synchronized in a chemostat by the method described previously [lo]. The medium used was M9 minimal salts with 0.2OlO glycerol as carbon source and buffered with 0.1 M Tris-C1, ph 7.4, in place of phosphate. The limiting nutrient was phosphate, which was added to the culture discontinuously, the period between additions being equal to the mean generation time of the culture as determined by the dilution rate (see [lo] for further details). RESULTS When E. coli B was grown on the minimal medium described above with 0.02 lactose present as well as 0.2 Ol0 glycerol, the behaviour of the cells and the enzyme P-galactosidase was as shown in Fig. 1. The two parts of the figure show data for two consecutive days, demonstrating the stability of the culture state under these conditions. This state can be maintained for days or weeks, and all perturbation experiments to be reported were carried out on such stable cultures. Cell numbers increased rapidly following the phosphate pulse, which entered the culture at the

2 Control Dynamics of P-Galactosidase European J. Biochem. 70-Galactosidase I ^. u Fig.1. Variations in cell number, turbidity (A,,,), and specific activity of B-galactosidme in a culture of E. coli B growing on M9 Tris, ph 7.4, in a chemostat. Phosphate (0.18 nim) was added at times marked by arrows on the abscissa, the interval between arrows being equal to the doubling time of the culture, 100 min x,cx 01 I I Fig. 2. Turbidity (A54o) and B-galactosidase activity of batch cultures growing asynchronously under phosphate limitation in M9 Tris, ph 7.4 with 0.Zo/, glycerol. One culture ( x ) had 0.2 O/, lactose, the other (0) had 0.02,lo lactose. Both cultures received a 0.2 mm pulse of phosphate at the arrow time shown by the arrow on the abscissa. The duration of the pulse was 3min, the amount of phosphate (KH,PO,) added being enough to increase the phosphate concentration in the culture by 0.18 mm. The degree of synchrony was about 60 Ole. Cell growth, as measured by turbidity (absorbance at 540 nm), was periodic, there being a characteristic delay following the addition of phosphate before the growth rate became greater than the washout rate. The rate fell again when phosphate became limiting; but as shown in the preceding paper, growth never ceased throughout the cycle. The specific activity of /3-galactosidase was strongly periodic, there being one oscillation per generation time, which in this experiment was 100 min. The oscillation had a definite phase relationship to the time of the phosphate pulse and hence to cell division. The first question to arise is whether or not the phosphate pulse is the immediate cause of the variation in /I-galactosidase activity observed in Fig. I. That is to say, is a change in phosphate concentration a sufficient condition for a change in activity of cells under the conditions of growth employed '2 Or is the observed oscillation in this enzyme the result of a number of factors such as repeated phosphate pulses, cell synchrony, and lactose concentration? If phosphate variation has a direct inductive effect upon the synthesis of 8-galactosidase in cells, then a phosphate pulse added to an asynchronous batch culture which had become phosphate-limited for growth should result in an inductive response. No such response was observed, as shown in Fig.2. The asynchronous cultures were grown on the same medium as that used in the chemostat, but with phosphate added at an initial concentration of 0.8 mm. One culture (crosses) had 0.20/, glycerol a/, lactose; the other (circles) had 0.2 Ol0 glycerol Ol0 lactose. When phosphate became limiting for growth, as shown by the flattening of the turbidity curve, a pulse of phosphate (KH,PO,) was added to each culture, increasing the concentration by 0.2 mm. These experiments show that a single phosphate perturbation of an asynchronous culture of E. coli B does not result in any significant change in the specific activity of,9-galactosidase, with either Ol0 lactose in the medium. Evidently the periodicity in the enzyme observed in Fig.l arises from some combination of factors involving cell synchrony, repetition of the phosphate pulse, and concentration of lactose in the medium. Repeated cycles of phosphate variation and cell division synchrony in the chemostat are not themselves sufficient to generate an oscillation in /3-galactosidase either, as is evident from the results shown in Fig.3 (open circles). The conditions of or

3 Vol. 10, No. 3, 1969 B. C. GOODWIN 517 growth in this experiment were identical to those of Fig.1, with the exception that lactose was present at 0.20/, instead of 0.020/, and the generation time was 80 min instead of 100 min. Cell counts varied exactly as in Fig.1 relative to the phosphate pulse, showing the same degree of division synchrony, but these have not been included in the figure. The increased lactose concentration had the effect of increasing the activity of the cells by a factor of somewhat less than 2 (about 215 units as compared with 120 units in Fig.l), and eliminating the periodicity in enzyme activity. Lactose not only in E. coli (there is no detectable enzyme activity in the absence of an inducer), but also gives rise to catabolite repression of this enzyme [Ill. The absence of an culture would be a mean value activity which is the average over the oscillation in single cells. This population average would appear as a more or less ((noisy but aperiodic level as in Fig.3. If the latter explanation is correct, then any perturbation which temporarily aligns in phase all the lactose control circuits (one per cell in the culture) should reveal a high frequency oscillation. Ideally the required stimulus would be one which rapidly and strongly repressed or induced the lactose control circuits in all the cells simultaneously and then was removed. As an approximation to this, a pulse of glucose (sufficient to produce an initial concentration of 0.080/,) was introduced into the culture. Such a pulse will give rise to a transiently 4. J. # J, 4. J I Fig.3. Two experiments carried out on the same culture, growing on M9 Tris with 0.20/, glycerol /, lactose, synchronized with respect to cell division by phosphate pulses added at times shown by arrows on the abscissa. Open circles: behaviour of,!i-galactosidase in the unperturbed culture. Closed circles : transient response of,6-galactosidase to 0.08 O/, glucose, added at the arrow observable periodicity in the synchronized culture in Fig.3 could be explained in at least two different ways: (a) The increased lactose level results in a damping of the oscillation observed on 0.020/, lactose; or (b) the increased lactose level results in an increased frequency of oscillation in the lactose control circuit in each cell, this increased frequency no longer being commensurate with the generation time and hence failing to lock synchronously with cell the division cycle [12]. The latter explanation implies that in each cell of the culture with 0.2 lactose there is a high frequency oscillation but this fails to synchronize with the cell cycle because there is no stable entrained relationship between the frequency of the lactose control circuit and that of cell division. Then the lactose control circuit of each cell would be oscillating freely at a relatively high frequency and, in the absence of any synchrony between the lactose control circuits of different cells, the net result in the increased level of catabolite repressor as a result of its degradation, thus causing a repression of the lactase control circuits. These will then relax back to their original state as the glucose is used up. The dynamics of the transient response may be expected to reveal some properties of the control circuit. The result of such perturbation is shown in the curve drawn through the solid circles in Pig. 3. The transient was not a smooth curve but showed a series of fluctuations or oscillations with a mean period of about 30 min initially, becoming somewhat longer and more irregular as the repression due to glucose wore off and the enzyme returned to its original aperiodic behaviour. This result suggests that the original aperiodicity may have been due to asynchrony in the control circuit; but it is also possible that the glucose was itself the exciting stimulus which elicited a train of damped oscillations in the circuits. There was no observed growth response of the culture to the glucose perturbation, phosphate limitation

4 518 Cont,rol Dynamics of p-galactosidase European J. Uiochem. remaining the growth-controlling factor. Cell synchrony was also undisturbed in this experiment, as in all perturbation experiments to be reported below. The stability of the oscillatory behaviour of p-galactosidase shown in Fig. 1 was investigated by perturbation experiments whose results are shown in Fig.4. The initial condition was the steady-state oscillation in the enzyme obtained with 0.020/, lactose in the medium and a doubling time of 100 min, one such cycle being shown before the perturbation was introduced. In the curve with open circles, the stimulus was a pulse of glucose making the culture 0.040/, in this sugar at the arrow. This had the effect of interrupting the steady state low frequency oscillation and producing a train of oscillations served only if the frequency of the B-galactosidase oscillation matches that of cell division has a further expected consequence. It should be possible to produce an aperiodicity in p-galactosidase by keeping the lactose concentration fixed at 0.020/, in the medium and increasing the generation time of the culture to a value which again produces an absence of stable entrainment between the supposed frequency of B-galactosidase oscillation and that of cell division. It is postulated that cells growing with 0.02O/, lactose in the medium have an oscillation in the lactose control circuit with a period of about min, since the oscillation is observed when the cells have a generation time in this range, If the generation time is increased to 160 min, say, then it IJ.1,.1. J J, Fig.4. Perturbation experiments carried out on a synchronized culture with an oscillation period 100 rnin, the cells growing on M9 Tris with 0.2 /0 glycerol O/, lactose, doubling time 100 min. Open circles: 0.Mo/,, glucose perturbation starting at arrow a8 marked. Closed circles: 0.1 mm isopropylthiogalactoside (IPTG) perturbation starting at arrow lasting for three generation times and with a mean period of about 40min. An initial net repression lasting for about 1.5 generation times was followed by a slight net overshoot of B-galactosidase above its original mean value of approximately 120 units/mg bacteria. By the fourth generation time after the perturbation the oscillation had returned to its original phase and frequency in relation to the cell cycle, the glucose presumably having been catabolised. Starting from the same initial condition, the response of p-galactosidase to a 0.1 mm pulse of isopropylthiogalactoside is shown in Fig. 4 by the dosed circles. This non-metabolized inducer caused an amplitude modification but no frequency change in the enzyme. The inductive influence gradually decayed as the inducer was washed out of the culture. The hypothesis that a population periodicity in b-galactosidase activity in a culture of bacteria synchronized with respect to cell division will be ob- may no longer be possible for a 100 min oscillation to be stably synchronized with the cell cycle, and the oscillation would then disappear at the population level. The result of this type of experiment is shown in Fig.5. The well-defined oscillation in,&galactosidase seen in Fig.1 is now absent, the enzyme showing irregular variations without any clear periodicity. Its behaviour is similar to that observed in Fig.3, open circles. This result is consistent with the above hypothesis, but there are alternative interpretations as will be discussed below. When no lactose was present in the medium but isopropylthiogalactoside was included at a concentration of 0.1 mm, there was a marked oscillation in fi-galactosidase as shown in Fig.6, lower curve. The phase of the oscillation relative to the phosphate pulse was the same as that in cultures with O.O2O/, lactose and a generation time of about loomin. When a small amount of lactose (0.005 /0) was included in the medium, the result was as shown in the curve with open circles: the amplitude of the

5 Voi. 10, No. 3, 1969 B. C. GOODWIN u I. J * 1, I Fig.5. Cell number, turbidity (Asao), and specific activity of 8-galactosidme in a synchronized culture growing on 1119 Trzs utith 0.20/, glycerol O/, lactose, doubling time 160 min p ;Galactosida;se X J J, J I J, i, _ m L k 200 * v).- t, R. - 0) U m._ m? e -._ m " B , Fig.7. Turbidity, cell number, a d /l-galactosidase activity of a synchronized culture of E. coli 3300 growing on M9 Tris with 0.Zo/, glycerol, doubling time ~ oscillations was reduced, but the phase and the mean values were unchanged. Cells exposed to a constant level of a gratuitous inducer should behave in a manner approximating that of a constitutive (i-) mutant. E. coli 3300 was used to see if a periodicity in p-galactosidase occurred in the absence of the inducibility property. The cells were grown with 0.2 Ol0 glycerol as the carbon source, and synchronized in the chemostat in the usual manner. This strain of E. coli was found to grow on M9 Tris more slowly than the B strain used in the other experiments, and a generation time of 2.75 h was used. The behaviour of the culture is shown in Fig.7. There was a marked periodicity in p-galacto-

6 520 Control Dynamics of P-Galactosidase European J. Biochern. sidase activity, with the same phase relation between the enzyme oscillation and the phosphate pulse as that observed in E. colib. Incidentally, the synchronization technique is seen to work with this KJ2 strain. The amplitude variation in turbidity was very small, but the degree of cell synchronization was about the same as that obtained with E. coli B. Another factor which could be involved in the periodicity in P-galactosidase observed under certain conditions in the synchronized culture is the synchronous replication of the structural gene, as suggested by Kuempel, Masters, and Pardee [2]. As a measure of this, the potential of the cells for synthesis of,8-galactosidase was determined by inductior of cells from a glycerol-grown, synchronized culture at different times in the cell cycle (Fig. 8). It is evident cycle, the synthesis of several enzymes being found to fall into this category (e. g. [1,2,13]). Since these processes are periodic and ordered relative to one another in time, a technical description of their dynamic behaviour is that they constitute a set of oscillations with stable phase and frequency relations. The temporal organization of the system arises from the stability of the relative phasing of the periodic or oscillatory processes. The purpose of the studies reported in this paper was to investigate the nature of the dynamic relationship between the lactose control circuit of E. coli and the cell cycle under particular growth conditions. For purposes of the present analysis, the cell can be conceptually divided into two dynamic systems : the lactose control circuit, and the rest of the cell. I Fig.8. Turbidity, cell number, and induced B-galactosidase activity in samples from a synchronized culture of E. coli B growing on M9 Tris with 0.2 O/, glycerol, doubling time 2 hours. The enzyme activity is that observed in samples after I0 min induction with 1 mm isopropylthriogalactoside, and I mm phosphate I that this potential for enzyme synthesis, expressed as the specific activity offl-galactosidase after 10 min of induction with 1 mm isopropylthiogalactoside in samples with phosphate added at 1 mm to eliminate the phosphate limitation, is markedly periodic. Furthermore, the phase of the periodicity is identical to that observed in Fig. 1. DISCUSSION The cycle of cell growth and division is by definition a periodic process in the sense that daughter cells pass through approximately the same set of physiological states as the mother cell in steady state growth of a culture. That there are changes of physiological state in growing and dividing cells is evident from the fact that different processes occur at different times in the cell cycle. The most obvious of these temporally ordered processes in bacteria are the initiation of DNA replication, septation, and cross-wall formation. It is now becoming evident that many other processes in bacteria are likewise temporally organized and periodic during the cell In this study, the rest of the cell was held in a particular, well-defined steady state by means of the synchronization technique described, while the dynamic state of the lactose control circuit, measured by P-galactosidase activity, was observed under different conditions. Some relationships between them can thus be deduced. With lactose as inducer, the lactose control circuit in E. coli is subject to two opposing forces. The inducer, an early product of lactose catabolism [14], causes derepression of the lactose operon and so acts as a positive feedback signal on the circuit. Dynamically this produces instability. Opposing this is the force due to catabolite repression, the catabolite repressor being presumed to be a late product of lactose catabolism, but whose identity has not yet been established. This negative feedback signal has a stabilizing effect on the circuit. The two forces together may be expected to give the circuit a propensity to go into a state of damped or undamped oscillations, as has been pointed out also by Knorre [15]. The frequency of an oscillation in the lactose control circuit will depend upon many contributing

7 Vol.10, No.3, 1969 B. C. GOODWIN 521 factors which together describe the control characteristics of the circuit, a major one being the rate of change of repression as a function of changes in inducer and co-repressor concentrations. The method of cell synchronization employed may be expected to produce marked periodicities in the physiological state of the cells, thus resulting in strong periodic intracellular signals affecting various control processes. Some of these will arise directly from the environmental phosphate periodicity, such as the intracellular phosphate levels involved in alkaline phosphatase control; but others will be less directly connected with the phosphate cycle. The results of the phosphate perturbation studies shown in Fig. 2 demonstrate that j3-galactosidase synthesis is not immediately affected by fluctuations in exogenous phosphate in an asynchronous culture. However, since in certain strains of bacteria phosphate starvation increases the amount of catabolite repression [IS], it seems possible that the repetitive, periodic phosphate-starvation regime used for synchronization sets up a periodic variation in catabolite repressor pool size in the cells, maximum pool size occurring just prior to the phosphate pulse. A second factor which would tend to produce a periodicity in specific enzyme activity in a synchronized population but not in an asynchronous one is the change in potential for j3-galactosidase synthesis occurring in cells when the structural gene is replicated. The data of Fig.8 show that this periodicity of potential has the same phase relation to the synchronizing phosphate pulse as the periodicity of,b-galactosidase activity observed in Fig. 1. Both of these factors may be involved in generating the stable oscillation in j3-galactosidase observed in synchronized cultures growing with 0.020/, lactose in the medium. The stability of this oscillation relative to the synchronized cell cycle is evident from the results shown in Fig.4, where the oscillation returns to its original frequency, amplitude, and phase values after the glucose perturbation. Either or both of these factors could account for the periodicity observed in /3-galactosidase activity in the constitutive mutant E. coli 3300, since catabolite repression continues to operate in cells which have lost the inducibility property [17]. No measurements of potential for j3-galactosidase synthesis were made in the constitutive, so it is not known if the structural gene replicates at the same time as in E. cozi B relative to the phosphate pulse, or at a different time. If periodic variations in catabolite repressor pool size do occur, then other enzymes such as tryptophanase and amylomaltase, which are also controlled by catabolite repression should show oscillations of the same type as /3-galactosidase. However, since they are controlled by different catabolites, one might expect that they would show different phase rela- tionships to the phosphate pulse, depending upon how the phosphate periodicity affects any particular catabolite repressor. The absence of an observable oscillation in the synchronized culture growing with 0.201, lactose in the medium can be explained in at least two different ways, as mentioned above. Either the lactose control circuit is dynamically stable at the steady state determined by the culture conditions and so no selfsustaining oscillations are possible ; or the frequency relation between the autonomously oscillating control circuit and the intracellular periodicity arising from synchronous lactose gene replication and/or that assumed to occur in catabolite repressor as a result of the phosphate periodicity is altered so that, under the interaction conditions occurring in this state, no stable, ordered relationship between the two frequencies is possible. The rate of decrease in the specific activity of /3-galactosidase during the transients following the glucose perturbations in Fig.3 and 4 is sometimes considerably greater than can be accounted for by dilution and washout. This indicates either that the enzyme is unstable under the conditions of phosphate limitation employed, or that there are experimental errors in the results. Occassionally the amplitude of an oscillation as drawn is not much greater than the confidence range of the assay procedure, so the estimate of frequency may be too high, especially in Pig.3. The oscillations in Fig.4 are more convincing. The estimated frequencies of the oscillations are in fact close to those observed by Knorre [I51 in perturbation experiments on asynchronous batch cultures of E. coli ML 30, where damped oscillations in j3-galactosidase were also observed when cells were transferred from 0.2 glucose to 0.2 Ol0 lactose. The interpretation of Knorre s experiments is subject to the same difficulty as that encountered above, in that it is not possible to decide whether individual control circuits were dynamically damped or whether the observed dying away of the oscillation arose from progressive asynchrony of control circuit behaviour. The experimental results of Fig.5 and 6 are likewise subject to the two alternative explanations considered above : dynamic stability of the control circuit, with damped osciilations foliowing a perturbation ; or interacting non-linear oscillators, with the stability of their phase relations depending upon their frequencies. It is also possible that under certain conditions one explanation is correct, and under others the alternative. The evidence available from these studies does not allow one to decide conclusively between them. However, it is evident that the dynamic properties of a control circuit are very important in determining what role it will play in relation to the growth and division cycle of the cell. A control circuit is not necessarily a dynamically

8 522 B. C. GOODWIN: Control Dynamics of /3-Galactosidase Xuropean J. Biocheiii. inactive element in the cell, serving no other purpose than the maintenance of a steady state level of products. It can also contribute to the temporal organization of the events occurring during the cell cycle by undergoing periodic variations with welldefined, stable phase and frequency relations to the cell cycle. The fact that the physiological cycle of cells synchronized by an environmental phosphate periodicity is certainly different from that of cells growing under constant nutrient conditions does not alter the general relevance of the results obtained on the dynamic behaviour of the lactose control circuit. It is evident that under all conditions of steady state growth the physiological state of cells must undergo a cycle, so that some set of control circuits must behave as self-sustaining oscillators (see also [IS]). Under the growth conditions employed in the chemostat, with 0.2 O/, glycerol always included in the medium, the lactose control circuit is not essential to the cell s physiology. It is perhaps for this reason that its dynamic behaviour can be altered relatively easily by perturbations and by changes in the growth conditions. A control circuit which is more directly involved in the essential physiological organization of the cell, such as the pyrimidine circuit involving aspartate transcarbamylase, would probably be much more tightly locked into the temporal organization of the cell cycle, and less easily perturbed. However, such a circuit might also show differences in oscillatory state under different growth conditions, depending upon the dynamic state of the circuit and its interactions with other control circuits which constitute the rest of the cell. There are, of course, other possible types of interaction between control circuits than those discussed above. I would like to acknowledge the excellent technical assistance of Miss Jacqueline Hall and very helpful discussions with Dr. Sofia NeEinovA. The research was supported by a grant from the Medical Research Council. REFERENCES 1. Masters, M., Kuempel, P. L., and Pardee, A. B., Biochem. Biophys. Res. Commun. 15 (1964) Kuempel, P. L., Masters, M., and Pardee, A. B., Biochem. Biophys. Res. Commun. 18 (1965) Ferretti, J. J., and Grey, E. D., Biochem. Biophys. Res. Commun. 29 (1967) Tauro, P., and Halvorson, H. O., J. Bacteriol. 92 (1966) Johnson, R. A., and Schmidt, B. R., Biochim. Biophys. Acta, 92 (1966) Masters. M.. and Donachie. W. D.. Nature. 203 (1966) \ I Goodwin, B. C., Temporal Organization in Cells. Academic Press, London Goodwin, B. C., Nature, 209 (1966) Cohn, M., Bacteriol. Rev. 21 (1957) Goodwin, B. C., European J. Biochern. 10 (1969) Magasanik, B., Cold Spring Harbor Symp. Quant. Biol. 24 (1961) Xinorsky, N., Non-linear Oscillations, Van Nostrand, New York Kogoma, T., and Nishi, A., J. Gen. Appl. Microbiol. 11 (1965) Burstein, C., Cohn, M., Kepes, A., and Monod, J., Biochim. Biophys. Acta, 95 (1965) Knorre, W. A., Biochem. Biophy8. Res. Commun. 31 (1968) McFall, E., and Magasanik, B., Biochim. Biophys. Acta, 55 (1962) Cohn, M., and Horibata, K., J. Bacteriol. 78 (1959) Goodwin, B. C., Symp. SOC. Gen. Microbiol. 19 (1969) 223. B. C. Goodwin School of Biological Sciences, University of Sussex Brighton (Sussex), Great-Britain