(1972) 64, 671-680
Replication during Development Caulobac ter crescen tus
Department of Biochemical Sciences Princeton University Princeton, N. J. 08540, U.S.A. (Received 23 August
Patterns of DNA replication and development have been determined in the dimorphic bacterium Caulobacter crescentzcs using a new method of cell synchrony. Characteristic DNA cycles were identified for the two cell types. The swarmer cell cycle is composed of G1, S and Gz periods of 65, 85 and 30 minutes, respectively, and the stalked cell cycle is composed of S and G2 periods of 90 and 30 minutes. Thus, the two cell types produced at division initiate DNA synthesis at different times : in the stalked cell chromosome replication begins immediately. while in the swarmer cell the onset of replication is delayed for approximately 65 minutes. Since the pre-synthetic gap in DNA replication corresponds to the time required for stalk formation by the swarmer cell, DNA synthesis is characteristic only of the stalked form of C. crescentus. The results suggest that there may be a structural requirement for initiation of DNA replication in these bacteria, and t.hat in the stalked cell this requirement for initiation has been satisfied at division, while in the swarmer cell further development is required.
1. Introduction The life cycle of t,he stalked bacterium Caulobacter crescentus is marked by a series of well-defined changes in cell structure and function which recommends it as a system for the study of development. Figure 1 outlines the growth pattern of I?. crescentucs and shows the three major morphological forms: (1) the motile swarmer cell which carries a single polar flagellum and pili to which infecting RNA phages attach, (2) the non-motile stalked cell which has formed an outgrowth of cell wall and membrane material, or stalk, at the point of flagellum attachment, and (3) the dividing cell which has retained the cellular stalk and developed a flagellum at the opposite pole (Poindexter, 1964). Stalk development by the swarmer cell is accompanied by loss of the two properties most characteristic of the swarmer: motility and susceptibility to infection by RNA phage (Schmidt, 1966; Shapiro & AgabianKeshishian, 1970). Also, as shown in Figure 1, all three forms possess an adhesive holdfast. This material is found at the base of the flagellum in the swarmer cell and at the tip of the stalk in the stalked cell (Poindexter, 1964). One striking feature of the above growth pattern is the polarity of cell development in C. crescentus and the resulting asymmetric division that produces one stalked cell t Present Brunswick,
address: Department New Jersey.
of Pharmacology, 671
FIG. 1. Growth and development of C. crescentus. (1, 2 and 3) are described in the text and in Poindexter
The characteristics (1964).
and one swarmer cell. The subsequent divisions of these two cell types illustrate that the growth of this bacterium is in reality composed of two related cell cycles, one in which a stalked cell develops into a dividing cell and the other in which a swarmer cell first forms a stalk and then develops into a dividing cell. The dividing cells produced by the two cycles are apparently morphologically and genetically identical, but the time required to form a dividing cell from a swarmer cell is substantially longer than that required to form a dividing cell from a stalked cell (Poindexter, 1964). Thus, cell growth and metabolism must be regulated to accommodate the extra time required for the development of a stalk by the swarmer cell. As a first step toward understanding the types of regulatory mechanisms required for development in C. crescentus, we began an investigation to determine how the timing of chromosome replication is co-ordinated with stalk formation and cell division. Using a new technique for obtaining cell synchrony, characteristic DNA cycles were identified for the two cell types. The DNA cycle in the swarmer cell is composed of a presynthetic gap (G1), a period of synthesis (S) and a postsynthetic gap (a,), while the stalked cycle contains only an S and G,. The length of G, in swarmer cells was found to approximate the time required for stalk formation.
C. crescentus, strain CB15 (ATCC 19089) was grown at 30°C in a water bath with gyrating shaking in minimal salts medium (Poindexter, 1964) which contained 0.2% glucose. The cells were synchronized by growing 25-ml. cultures of the cells in a 14-cm glass Petri plate to a density of approximately 3 x 10s cells/ml. The plate was carefully washed with preconditioned medium and incubated for 1 to 2 hr with the same medium. The plate was then washed again and swarmer cells were collected after an additional 10 min of incubation. The yield varied from 1 to 3 x IO7 cells/ml. Growth of these synchronous cells was followed either by viable counts or in the model B Coulter counter. Stalk formation was followed by staining with Gray’s stain (Society of American Bacteriologists, 1957). Since incorporation of exogenous thymine, thymidine and deoxyribose into the DNA of C. crescentus is negligible (unpublished data; J. Gerhardt, personal communication), DNA synthesis was followed by labeling cells with [3H]deoxyguanosine (Schwarz Bioresearch) or [3H]deoxyadenosine (New England Nuclear). Incorporation was stopped by precipitation with cold 50/, trichloroacetic acid in the presence of 100 pg bovine serum albumin/ml. Samples were centrifuged and resuspended in 1-O N-KOH to degrade RNA. After incubation for 15 hr at 37°C the samples were neutralized with 12 N-HCl, reprecipitated with cold 10% trichloroacetic acid which contained 100 rg per ml. deoxyadenosine or deoxyguanosine as carrier, chilled for 30 min at 4”C, and filtered on Whatman GFjA
glass filters which had been presoaked in 10% trichloroacetic acid plus carrier. The filters were then washed successively with 10% trichloroacetic acid containing carrier and 50$:& ethanol, dried and counted in toluene scintillation which contained O.4o/o 2,5-diphenyloxazole and 0.01 o/o 1,4-bis-[2-(5-phenyloxazolyl)]-benzene (Packard Instrument Company). RNA synthesis was followed by pulse-labeling cells with [3H]deoxyguanosine (see Table 1). The samples were precipitated with cold 5% trichloroacetic acid in the presence of 100 pg bovine serum albumin/ml., chilled for 30 min at 4°C filtered on Whatman GF/.A glass filters which had been presoaked with 10% trichloroacetic acid containing 100 pg cold deoxyguanosine/ml. and prepared for counting as described above. This method measured total incorporation into the macromolecular fraction. Incorporation into RX.4 was calculated by subtracting the counts incorporated into DNA (base-stable fraction, see above) during the same period of incubation.
3. Results (a) Cell synchrony and development The only method previously available for synchronizing C. crescentus is based on separation of the two cell forms by repeated centrifugation at 4°C (Stove & Stanier, 1962). To avoid possible physiological shock to the cells from this treatment, a new technique for obtaining synchronous cultures was developed. The method depends upon the adhesive properties of the holdfast which is located at the flagellated pole of the swarmer cell and at the distal end of the stalk of the stalked cell. During incubation in glass Petri plates, cells attach to the surface by this holdfast and continue to grow and divide. All three morphological forms may attach, but stalked cells ultimately predominate on the surface because of the orientation of the holdfast (Fig. 1). Thus, only swarmer cells are released into the medium at division and these cells are prevented from re-attaching by gentle agitation of the plate. Using plates prepared as described in Materials and Methods, a 25ml. culture containing 1 to 3 x lo7 swarmer cells per ml. was obtained from Petri plates after a lo-minute incubation. As shown below this is equivalent to 1112 of the time required for division of a stalked cell and l/l8 of the time required for division of the swarmer cell. In stained samples (Fig. 2) approximately 95 to lOOo/o of the released cells were swarmer cells, as indicated by the presence of a polar flagellum and the absence of a stalk. Swarmer cells obtained by this procedure were synchronized for both growth and development (Fig. 2). A first wave of cell division at HO&IO minutes produced a population of sibling swarmer and stalked cells. The cell number, determined either by viable counts or by particle counts using the model B Coulter counter, doubled at this division, indicating that all the ceils divided and were physiologically norma,l. A second synchronous wave of division normally occurred at 300f 10 minutes. This division amounted to a 50% increase in cell number and is attributed to the division of stalked cells 120 minutes after their production at the first division. This is consistent with Poindexter’s (1964) observation that the doubling time for stalked cells is shorter than the doubling time for swarmer cells. Also in agreement with sn average division cycle of 120 minutes for stalked cells and 180 minutes for swarmer cells in minimal glucose medium is the synchronous increase in cell number at 350 to 360 minutes; this presumably reflects division of the swarmer cells produced during the first synchronous division. The appearance of stalked cells was followed in synchronous populations of swarmer cells by examination of stained preparations (Fig. 2). Using this morphological assay
S. T. DEGNEN
FIG. 2. Growth and stalk formation in synchronous cultures of swarmer cells. Synchronous cultures were collected as described in Materials and Methods, and growth was followed by viable counts (-e-e--) and by particle counts (--A--A--, arbitrary units) with a model B Coulter counter. Stalk formation (-m-m-) was followed by examining the morphology of samples which had been treated with Gray’s stain (Society of American Bacteriologists, 1957).
for development, the time required for stalk formation under these conditions was estimated at about 60 minutes, a period approximating the difference between division times for swarmer and stalked cells. (b) Replication
in the swarwbercell cycle of C. crescentus
The cycle of chromosome replication during development of swarmer cells was determined in synchronous cultures which were collected as described above. The cells were transferred to a flask in a shaking water bath and samples were withdrawn from the culture at various times for determination of cell number and rate of DNA synthesis. The results from one of these experiments in which the cells mere pulsed with [3H]deoxyguanosine (Fig. 3(a)) showed that significant synthesis of DNS did not take place during the initial 65 minutes of incubation. This gap in replication (G,) was followed by an S period of approximately 80 minutes and a G, period of 30 minutes. Similar values were obtained when [3H]deoxyadenosine was the precursor (Fig. 3(b)). The pattern of DNA synthesis shown by the curves in Figure 3 has been observed repeatedly. Times have ranged from 60 to 70 minutes for G,, from 80 to 90 minutes for S and from 30 to 40 minutes for G,, depending to some extent on the exact length of time required for division. The average values of G,, S and G, are 65, 85 and 30 minutes, respect’ively. Since two cell types are present after division the pattern of DNA replication is more complicated, but the results shown in Figure 3(b) are most typical of our results
for the post-division cycle. The rapid increase in the rate of incorporation of [3H]deoxyadenosine into DNA at the time of division may result from initiation of chromosome synthesis in the newly-divided stalked cells. This interpretation is consistent with results presented below which indicate that stalked cells do not have a G,. The time between division and the second increase in the rate of DNA synthesis is approximately equal to the G, observed for swarmer cells in the first cycle of growth. This gap and the increased rate of synthesis which follows at 240 minutes probably reflect’ the initial periods in the G,-S-G, cycle of swarmer cells produced at 180 minutes. In another experiment the rates of both DNA and RNA synthesis were determined during the swarmer cell cycle. These results (Table 1) showed that the cells actively incorporated [3H]deoxyguanosine into the alkali-labile trichloroacetic acid-precipit able fraction at all stages of t,he cell cycle. This finding argues that the periods design ated as G, and G, in the DNA cycle are not artifacts of permeability or metabolism. TABLE Rates of RNA
Time after collection (min)
[3H]deoxyguanosine (cts/min RNA
incorporatedt X 10e3) DNA
40 50 70 90 110 130 150 160 170 180 190
39.0 32.5 47.3 53.4 49.6 46.3 39.2
1.06 1.48 2.59 3.63 3.71 4.41 3.71 3-36 2.62 1.44 2.19
.I Samples of 0.5 ml. from a synchronous population of swarmer cells were pulsed with 5 $X/ml. (6 Ci/m-mole) for 8 min at the times indicated. The reaction was stopped
of [3H]deoxyguanosine and the incorporation Methods. The rate of incubation.
into either RNA of synthesis is expressed
The presence of a G, period
was determined as described as cts/min incorporated/O.5
of (7, in the Swarmer
in Materials of culture/8
in the swarmer cell cycle was confirmed by a second technique which depends upon the coupling between completion of DNA replication and cell division in C. crescentus. When synchronous populations of cells are treated with an inhibitor of DNA replication at various times in the division cycle, the inhibitor halts completion of the cycle when it is added during the S period, but not when it is added after replication is completed. Thus, if the inhibitor is added
FIG. 3. Rate of DNA synthesis in synchronous cultures of swarmer cells. The rate of DNA synthesis during the cell cycle of synchronous swarmer cells (see Materials and Methods) was determined by incubating 0.5-1111. samples of the culture withdrawn at the times indicated with either (a) 17.5 pCi/ml. of [3H]deoxyguanosine (6 Ci/m-mole) for 8 min or (b) 16.1 $X/ml. of [3H]deoxya,denosine (19 Ci/m-mole) for 5 min. The reactions were stopped and the counts incorporated into DNA were determined as described in Materials and Methods. Cell growth, which is shown only in (a), was followed by particle counts (--e--e--) with the model B Coulter counter. The time for G1 was determined by estimating the period from the start of incubation to the midpoint in the curve of increasing rate of DNA synthesis, the time for S by estimating the period between the midpoint of the curve of increasing rate of DNA synthesis and the midpoint of the ourve of decreasing rate, and the time for Gz by estimating the period from the midpoint in the curve of decreasing rate of DNA synthesis to the synchronous cell division.
during G, the cells will proceed to divide at the normal time. This method has been used by Clark (1968) to determine the time between completion of chromosome replication and cell division in synchronous cultures of Escherichia coli B/r. Similar experiments on C. crescentus were performed by removing portions of cells from a synchronous population of swarmer cells at various times and treating them with hydroxyurea, an inhibitor known to stop DNA synthesis in these bacteria
without interfering with RNA synthesis (Degnen & Newton, 1972). Cell number was determined by viable counts or particle counts immediately after the addition of the inhibitor and at times after division had taken place in a control culture (Fig. 4). The G, periods of 36 minutes and 40 minutes calculated from the results of these two experiments are in good agreement with the values of 30 to 40 minutes for the same period determined by pulse-labeling (see Results, section (b)). (d) Replication
in the stalked cell cycle of C. crescentus
The DNA cycle in stalked cells of C. crescentus can be determined by pulse-labeling cellswhich are attached to the surface of a Petri plate and following the radioactivity in swarmer cells which are releasedinto the medium. This method dependsupon two facts: first, only those attached cells in the S period will be labeled and subsequently releaseswarmer cellswith radioactive DNA, and secondly, the oldest oellson the plate will divide first so that the time at which a cell is releasedinto the medium is inversely related to its age at the time of labeling.
FIG. 4. Cell division following inhibition of DNA synthesis in synchronous swarmer cell cultures. Synchronous cultures of swarmer cells of C. crescentus were prepared as described in Materials and Methods. At the times indicated in (a), l.O-ml. samples were removed and added to warmed flasks containing hydroxyurea so that the final concentration of inhibitor was 5 mg/ml. The final number of cells in each flask was then determined at 215 and 330 min; the averages of the two values were plotted in (a) as a function of the time at which the inhibitor was added. In each case a controlled culture to which no hydroxyurea has been added was included to determine the time of cell division (b). Cell number in the first experiment (-A-Aand ---A--) was determined by particle counts (Coulter model B counter) and in the second experiment (-O-O-and -a-m-) by viable counts. Values for Ge were estimated from the difference in the midpoints of the synchrony curve (b) and the curve for the final cell number following inhibition by hydroxyurea (a) ; these midpoints are indicated by vertical lines.
FIG. 5. Replication of DNA in attached stalked cells. Stalked cells attached to Petri plates of the type previously described for the collection of synchronous swarmer cells (Materials and Methods) were labeled for 7 min by the addition of 25 ml. of minimal salts-glucose medium which contained 0.75 mCi of [3H]deoxyguanosine (6 Ci/m-mole). The medium was then removed, the plate washed twice with 25 ml. of preconditioned medium, and another 25 ml. of preconditioned medium which contained 300 pg of unlabeled deoxyguanosine/ml. was added. The plate was agitated at a speed of 1.2 on a New Brunswick rotary water bath and the release of newly divided cells into the medium followed by viable counts ([email protected]
), and 0.2-ml. samples were withdrawn at the times indicated for determination of radioactivity (--- (:;--- \ -- ) illcorporated into DNA (see Materials and Methods).
Plates prepared as described in Materials and Methods were incubated with shaking for 120 minutes to prevent attachment of additional swarmer cells and t,o ensure that all attached cells were stalked (see Results, section (a)). The culture was then labeled for 8 minutes with [3H]deoxyguanosine and the release of cells into fresh medium which contained 300 pg unlabeled deoxyguanosine/ml. was followed. Figure 5 shows the number of viable cells and the amount of labeled DNA accumulated in the medium at different times after labeling the attached cells. A G, period of approximately 30 minutes was observed for the initial period during which bhe newly-divided swarmer cells were released into the medium with little or no incorporated [3H]deoxyguanosine. Substantial radioactivity was subsequently released with the progeny cells and the radioactivity continued t,o accumulate for approximately 90 minutes; this period corresponded to the S period for the attached stalked cells. Cells were released into the medium at a constant rate during the first G,, S and the second G, period (Fig. 5). The length of G, plus S is sufficient to account for the total division t.imc of 120 minutes for the stalked cell (Fig. 2). This suggests that these cells do not contain a significant G1, a conclusion which is supported by the length of the hiatus in accumulation of radioactivity beginning at 120 minutes (Fig. 5). This gap, which should be equivalent to G, plus G, for the attached cells, is equal to the length of the initial gap, or a period corresponding to G, alone. Thus, chromosome replication probably begins at the time of cell division, although it is not possible to tell from the present data. whether initiation takes place just before or just after division. A similar approach has been used by Helmstetter (1967) t,o determine the DNA cycle in Escherichia coli B/r cells which were attached to membrane filters. The present
technique (Fig. 5) differs in several respects, however. (1) The age distribution of the attached cells is random, as indicated by the constant rate at which swarmers were released during the first 180 minutes. (2) The measure of radioactivity in the medium at any one time is a cumulative value and does not represent the rate of DNA replicat’ion in the bound cells, and (3) attachment of the cells to the plate is asymmetric, since the cells adhere to the plate by the stalk. Since the mode of cell attachment to Petri plates allows release of only one cell t’ype into the medium at division, it should be possible to assess whether the segregation of chromosomes to progeny cells is random or non-random. In the above experiment the amount of radioactivity released during successive S periods should give some indication of the segregation pattern, This appoach is complicated, however. Release of swarmer cells is not quantitative because a small fraction of them reattach despite agitation of the plate during collection. Also, the large amount of [3H]deoxyguanosine incorporated into RNA (approximately 96% of the label) makes a complete chase difficult. Turnover of this RNA probably accounts for some continued labeling of DNA after the pulse and the acceleration in the rate of label released during S (Fig. 5).
4. Discussion Development in C. crescentus is polar andleads to a dividing cell which is structurally asymmetric. Thus, at division, two cell types are produced which have different growth cycles: the swarmer cell develops a stalk and divides in 180 minutes and the sibling stalked cell divides in 120 minutes (see Figs 1 and 2). The results described above (Figs 3, 4 and 5) show that chromosome replication is regulated in a special way to accommodate this pattern of growth. This is accomplished by controlling the times at which chromosome replication is initiated in the daughter cells at division, rather than by drastically altering the lengths of the S and G, periods (Fig. 6). As a result, chromosome synthesis is characteristic only of the stalked form of C. crescentus. The data used to construct the scheme given in Figure 6 are summarized from the patterns of DNA replication determined by pulse-labeling synchronous populations of swarmer cells (Fig. 3) and the pattern of labeling in progeny released from an asynchronous population of attached stalked cells (Fig. 5). The values assigned to S in the two cycles differ by 5 minutes, but it is difficult to know whether the time required for chromosome replication in the stalked cell cycle is in fact significantly longer; the determinations were made under quite different conditions and, as noted above (see Results, section (b)), the values for S varied by as much as 10 minutos in separate experiments. The presence of gaps in the DNA cycle is supported by two types of experiments. In the first it was shown that the periods of reduced incorporation into DNA could not be accounted for by changes in permeability alone; swarmer cells actively incorporated [3H]deoxyguanosine into RNA throughout the growth cycle (Table 1). In the second set of experiments the presence of G, was independently confirmed by the limited ability of synchronous cells to divide after the addition of an inhibitor of DNA synthesis to the culture at various times during the cell cycle (Fig. 4). The times calculated from these results are in good agreement with the G, period determined by pulse-labeling (see Results, section (0)). A value of 35 minutes has also been obtained from the time of continued cell division in asynchronous, exponentialiy 44
/ 6-b l 3 t-G,-+----S----+ b ,
(90) FIG. 6. Summary of the DNA cycles in swarmer given for G,, S end Ga are averages (see text).
(30) and stalked
cells of C. crescenlus.
growing cultures of C. crescentus after addition of selective inhibitors of chromosome replication (Degnen & Newton, 1972). This finding supports the conclusion from pulselabeling experiments that the length of G, in the swarmer cell cycle and the stalked cell cycle is the same. The immediate questions raised by these results are what mechanisms control the different timing of initiation of chromosome replication in the two cell types and how initiation in the swarmer cell is related to development, e.g. stalk formation ? Models of initiations based on diffusible cytoplasmic initiation proteins (Jacob, Brenner & Cuzin, 1963) predict the simultaneous start of DNA synthesis in both progeny cells. In C. crescentus, however, the stalked and swarmer cell share a common cytoplasm before division, and yet initiation of chromosome replication is delayed only in the swarmer cell (Fig. 6). This suggests that the time of initiation may not be controlled exclusively by soluble factors. Since stalked cells and swarmer cells are morphologically distinct, the differences between them could include differences in membrane associated sites needed for DNA synthesis. Thus, one possibility for the differential control of DNA synthesis would be a structural requirement for initiation which is satisfied in only the stalked cell at the time of division; further development would be needed to elaborate this required structure in the swarmer cell. This work the National (A. N.) from
was supported by grants from the American Cancer Institutes of Health (GM962), and a Career Development the National Institutes of Health.
Society (no. VC-35A), Award to one of us
REFERENCES Clark, D. J. (1968). Cold Spr. Hurb. Syrnp. Quant. B&L 33, 823. Degnen, S. T. & Newton, A. (1972). J. Back in the press. Helmstetter, C. E. (1967). J. Mol. Biol. 24, 417. Jacob, F., Brenner, S. t Cuzin, I?. (1963). Cold Spr. Harb. Sywq. Quunt. Biol. 28, 329. Poindexter, J. S. (1964). Bact. Rev.(1964). 28, 231. Schmidt, J. M. (1966). J. Gen. Microbial. 45, 347. Shapiro, L. & Agabian-Keshishian, N. (1970). PTOC. Nat. Acud. Sk., Wash. 67, 200. of Microbiological Methods, p. 26. New Society of American Bacteriologists (1957). M anual York: McGraw-Hill Book Co., Inc. Stove, J. L. & Stanier, R. Y. (1962). Nature, 196, 1189.