VOL.
9,
NO.
2,
JANUARY
20, 1 9 7 0
Replication of a Bacterial Plasmid and an Episome in Escherichia coli" Michael Bazaral and Donald R. Helinski
ABSTRACT :
Colicinogenic factor El (ColEJ deoxyribonucleic acid isolated from Escherichia coli consists predominately of closed circular double-stranded molecules of a molecular weight of 4.2 X lo6. During amino acid starvation the rate of synthesis of C o E l deoxyribonucleic acid became linear and increased 2.0-2.5-fold over the normal rate, whereas the rate of synthesis of chromosomal deoxyribonucleic acid progressively decreased. Since the time required to synthesize a complete COW, deoxyribonucleic acid molecule is short relative to the replication time of the chromosome, this result indicates that protein synthesis is not necessary for the initiation of synthesis of CoEl deoxyribonucleic acid. The replication of CoE1deoxyribonucleic acid in E . coli also was examined after a shift of medium from DzO, [15N]H.Cl dense medium to normal medium. At the end of the first generation after the medium shift, the ColEl deoxyribonucleic acid was distributed in a ratio of 1 :2 :1 among unreplicated, once-replicated, and at least twice-replicated classes, respec-
C
olicinogenic factor El (ColE1)is the extrachromosomal genetic element which determines the production of a specific antibiotic protein, colicin El, and confers upon its host bacterium immunity to this antibiotic. CoEl is not a fertility agent in Escherichia coli and promotes neither its own transfer, nor the transfer of chromosomal markers to noncolicinogenic bacteria. The conjugal transfer of CoWl is dependent on the presence of fertility factors such as the F factor (Nagel de Zwaig and Puig, 1964; Kahn and Helinski, 1964). Since the CoE1 factor does not appear to integrate with the bacterial chromosome, it is classified as a bacterial plasmid (Clark and Adelberg, 1962). ColEl isolated from E. coli consists of a closed circular duplex DNA molecule of a molecular weight of 4.2 x 106 (Bazaral and Helinski, 1968a). There are a minimum of four copies of this molecule per chromosome in colicinogenic E. coli cells. The E. coli fertility factor, F, also has been reported to consist of DNA in the form of closed circular duplex molecules with a molecular weight estimated to be 45 X 106 (Freifelder, 1968). The relationship of the F factor, a bacterial episome, to the host cell differs from that of ColEl in at least two significant respects; the F factor can promote bacterial mating and
* From the Department of Biology, University of California, San Diego, La Jolla, California 92037. Supported by U. S. Public Health Service Research Grant AI-07194 and National Science Foundation Grant GB-6297. M. B. acknowledges the support of a U. S. Public Health Service fellowship (No. l-Fl-GM-29,966-0lAl). D. R. H. is a U. S . Public Health Service Research Career Development awardee (NO.K04-6M07821).
tively. This observation is consistent with a mechanism of CoWl deoxyribonucleic acid replication involving the duplication of a molecule, the return of the copies to a pool of CoEl molecules, and subsequent selection of a molecule from the pool at random for the next duplication event. Since partially replicated molecules are not observed 15 min (1/6 generation) after the medium shift, the time necessary for a complete replication of CoEl deoxyribonucleic acid is substantially less than 15 rnin. F factor deoxyribonucleic acid was isolated as a closed circular duplex deoxyribonucleic acid form. The sedimentation coefficient of supercoiled F factor deoxyribonucleic acid was estimated by sucrose gradient analysis to be 80 S; corresponding to a molecular weight of 75 X lo6. The highest observed yield was 0.7 supercoiled F factor molecule per chromosome. F factor initiation continued at a decreasing rate in the absence of amino acids, suggesting that limited initiation of F factor synthesis can occur without protein synthesis.
the F factor can become genetically linked to the chromosome (Hayes, 1953). Both plasmids and episomes are present in approximately constant average numbers per cell during exponential growth (Bazaral and Helinski, 1968a,b; Wohlhieter et af., 1964). These observations in additicn to the fact that these elements do not segregate out of growing populations imply the existence of a mechanism of control of the replication of episomes and plasmids and their segregation between daughter cells during division. In this sense, the extrachromosomal elements may be analogous to the bacterial chromosome. This report is concerned with several characteristics of the mode of replication of CoWl and F factor DNA in E. coli. Experimental Section Bacterial Strains. The ColEl factor of C600 (CoE1) was originally derived from E . coli K-30. C600 is an E. coli K-12 strain requiring thiamine, threonine, and leucine for growth. A thymine-requiring mutant (C600 (COW,) thy-) of this strain was made by a modification of the procedure of Stacey and Simpson (1965). Approximately lo8 cells were incubated in 5 ml of supplemented Tris minimal medium containing 0.7 pg of N,N'-nitrosoguanidine, 42 pg of trimethopterin, and 20 pg of thymidine until visable grawth had occurred. Mutants were selected by replica plating on supplemented Tris plates with and without thymidine (5 pg/ml). From this population, a cell capable of growing on 2 pg/ml was selected. E. coli CR34, a C600 derivative requiring thymine, was
REPLICATION OF
C O ~ AEN D~ F
IN
E. c o l i
399
B I O C H E M I S T R Y
made F +by conjugation with YS40 ( F ) followed by nutritional counterselection against the donor. The F factor was originally derived from W1485 (F). The male specific phage MS2 used to detect the presence of the F factor was a gift from P. Bonner. Growth and Labeling Conditions. Tris-buffered minimal medium (DeWitt and Helinski, 1965) and M9 phosphatebuffered minimal medium (Kahn and Helinski, 1964) were used. For growth of E. coli C600 (ColE1)thy- Tris medium was supplemented with 5 pg/ml of thiamine, 2 pg/ml of thymine, 40 pg/ml of DL-leucine, 100 pg/ml of DL-threonine, and 1.0 ml of 1.0 M sodium phosphate (pH 7.0) per 1.; 0.4% glucose was used as the carbon source. When DNA was labeled with either [ 3H]methylthymine (128 mCi/mg) or [ I4C]2-thymine (0.45 mCi/mg), the final thymine concentration was adjusted to 2.0-2.5 pg/ml. Density labeling was performed in the same medium, with the NH4C1replaced with 200 pg/ml of [ l5NN]H4Cl and with S5% of the water by volume replaced with D,O. E. coli CR34 strains were grown under the same conditions listed above for C600, except that supplemented M9 phosphate medium was used as the basic salts medium in place of the Tris medium. Cell densities were measured using a Zeiss spectrophotometer at 660 mp. Direct Dye-Buoyant Centrifugation of Sarkosyl Lysates. DNA was fractionated by the direct dye-buoyant centrifugation procedure described previously (Bazaral and Helinski, 1968a). This procedure is based on the ethidium bromidecesium chloride centrifugation technique developed by Radloff et til. (1967). Sucrose Gradient Velocity Centrifugation. Sucrose gradient centrifugation was performed in a Spinco Model L2 or L4 ultracentrifuge in an SW25.1 swinging-bucket head at 25,000 rpm at 20“. A linear 5-20% sucrose gradient (30 ml) in TES buffer2 was used. Fractions (1 ml) were collected by drop, o r time, from the bottom of the tube. Alternatively, sucrose gradient centrifugation was performed using an SW65 swinging bucket rotor at 50,000 rpm at 20” and containing a 5-ml linear 5-20Z sucrose gradient in TES buffer. Fractions (0.14 ml) were collected by drop from the bottom of the tube. Cesiunr Chloride Density Equilibrium Separation of DensityLaheled DNA. To a total sample volume of 3.75 ml (in TES buffer), 4.75 g of CsCl was added. The solution was placed in a nitrocellulose tube, overlaid with light mineral oil, and centrifuged for 60 hr in a Spinco No. 40 rotor at 40,000 rpm at 20”. At the end of the run 0.10-ml fractions were collected by drop from the bottom of the tube. Counting of Radioisotopes. Samples of a volume less than 0.20 ml were prepared by a modification of the method of Bollum (1966). Samples were absorbed into 1 X 1 in. squares of Whatman No. 3MM filter paper, and washed successively with cold 10% trichloroacetic acid, 95% ethanol, and ether. Filters were commonly washed in batches of 30 by consecutive immersion in 500 ml of trichloroacetic acid, 500 ml of ethanol, 1 Tris minimal contained (per liter): 2.0 g of NHICI, 0.37 g of NaC1, 0.01 g of MgCh.6H~O,O.026g of NazSOa, 0.9 mg ofFeC18, and 100 mg of 1.0 M Tris-CI (pH 7.3). M9 minimal contained (per liter): 7.0 g of N L I ~ H P O3.0 I , g of KHIPOI, M MgSOA, and M FeCh. * TES bufkr contained: 0.05 M Tris-C1-0.05 M NaC1-0.005 M EDTA (pH 8.0).
400
B A Z A R A L
AND
HELINSKI
and 250 mi of ether for 15 min with occasional agitation. Samples of a volume larger than 0.20 ml were precipitated with 50 pg of bovine serum albumin and 10% trichloroacetic acid in a total volume of 1.2 ml at 0”. Precipitates were collected on 25-mm S & S Bactiflex B-6 nitrocellulose membrane filters, washed twice with 10 ml of cold 10% trichloroacetic acid, and dried in a 100” oven for 30 min. Dried filters from either preparation were immersed in scintillation vials containing 10 ml of toluene-2,5-diphenoxazole (0.4 wiv) and counted in a Beckman liquid scintillation counter. All significant samples were counted to at least 5z statistical accuracy. Reagents. All radioisotopes were purchased from the New England Nuclear Co. Sarkosyl NL30 was obtained from the Geigy Chemical Co. Ethidium bromide was a gift of the Boots Pure Drug Co. Ltd., Nottingham, England. D?O (99.5z mole % min) was purchased from the Matheson Chemical Co., and [15“jH4C1(95 mole % min) was purchased from the Isomet Corp. Lysozyme (“B” grade) and RNase (“A” grade) were purchased from the California Corporation for Biochemical Research.
z
z
Results Estimation of the Proportion of CoIE1 DNA which is 23 S. Since the isolation of CoEl DNA from E. coli by the procedure employed depended on the closed circular configuration of this DNA, it was necessary to show that the closed circular form of ColE, DNA is the major form, and not a by-product of the open circular or linear forms. This was accomplished by extracting DNA from colicinogenic bacteria with a technique which resulted in relatively large and fastsedimenting fragments of the bacterial chromosome, and then demonstrating that the major peak of low molecular weight, slow-sedimenting DNA is the 23S, or covalently closed form, of ColE, DNA. E. coli C600 (ColEJ thy- cells were labeled for five generations in 10 ml of supplemented minimal medium containing 0.3 mCi of [3H]thymineand 2 pg/ml of cold carrier thymine. At a cell density of 0.55 (OD660) the cells were harvested and lysed in sarkosyl as described in the Experimental Section. The lysate (2.5 ml) was added to a mixture consisting of 2.5 ml of TES, 3.5 ml of HsO, and 10.8 g of CsCI. The preparation was centrifuged 44 hr in a Spinco Type 50 rotor. At the end of this time 30 fractions were collected from the bottom of the tube. The DNA-containing fractions were identified by their viscosity, pooled, dialyzed, and applied to a sucrose gradient. As shown in Figure 1 , the major observed low molecular weight peak sedimented as the 23s closed circular ColE, DNA form with a small additional amount of 17s material, or the open circular CoZEI DNA form (Bazaral and Helinski, 1968a). The material under the 23s peak (fractions 14, 15, and 16) comprises 2% of the DNA applied to the sucrose gradient. Since approximately 5 0 z of the material in the 23s region represents contaminating DNA, it can be estimated that the ratio of 23s ColEl DNA to chromosomal DNA is approximately 1 % in this experiment. ColEl D N A Synthesis in the Absence of a Required Amino Acid. Although the mechanism of initiation of chromosomal synthesis in E . coli is not precisely elucidated, it has been established (Maaloe and Hanawalt, 1961; Lark et al., 1963) that protein synthesis is required. The continued synthesis of
V O L . 9, N O . 2, J A N U A R Y 2 0 , 1 9 7 0
1
I
1356,000counts/min 23s
12H
I
0
1
HOURS
8
t
I-
01 I
,
2
I
10
-7".
20
FRACTION NUMBER
1: Sucrose gradient analysis of a sample of DNA from E. that was purified by CsCl gradient centrifugation. A DNA solution (1 ml) containing 1.1 x 106 cpm and a p proximately 50 fig of DNA was applied to a 30-ml sucrose gradient, and centrifuged in a SW25.1 rotor for 8.75 hr at 20". At the end of the run 1.0-ml fractionswere collected, and the total radioisotope in FIGURE
co/i C600 (ColE1) thy-
each fraction was determined.
ColEl factor DNA in the absence of a required amino acid described below, therefore, demonstrates that the CoEl DNA synthesis is under the control of a mechanism distinguishable in some significant aspects from that responsible for the control of chromosomal DNA synthesis. The characteristics of thymine uptake in the absence of a required amino acid under the experimental conditions used is described in Figure 2. E. coli C600 (CoZE,) thy- was grown on 10 ml of supplemented Tris minimal media to an ODseo of 0.5, Cells were harvested, washed, and resuspended in 20 ml of medium complete except for leucine and containing 0.05 mCi of [3H]thymine in 40 pg of carrier thymine. A 1 0 4 sample of the culture was withdrawn and added to 400 pg of DL-leucine and the remaining 10 ml was incubated without leucine. The growth and incorporation of [3H]thymine into each culture was followed. The growth profile during amino acid starvation (Figure 2a) showed that cell growth is essentially stopped, and the uptake kinetics (Figure 2b) are consistent with the hypothesis that initiation of chromosome synthesis ceased (Maalae and Hanawalt, 1961 ; Lark et al., 1963). The kinetics of ColEl DNA synthesis in the absence of leucine also were measured. E. coli C600 ( C o E J thy- were grown to an OD6c0 of 0.425 in 30 ml of supplemented Tris minimal medium containing 2 pg/ml of thymine and a total of 0.01 mCi of ['Clthymine. Cells were harvested, washed, and resuspended in 40 ml of medium containing 1 pg/ml of thymine but lacking leucine. Pulses of 20-min duration of 0.1 mCi of [3H]thymine werea dded to 10-ml aliquots at 0, 70, and 130 min. At the end of the pulses the cells were chilled on ice. When the last pulse was completed the cells were subjected to the sarkosyl-ethidium bromide CsCl centrifugation procedure, and the amount of and 3H incorporated into ColEl DNA and chromosomal DNA was determined. As shown in Table I the 14Clabel in the chromosome and CoEl DNA was not lost during starvation. The synthesis of DNA during starvation (3H label) was normalized to the in each sample to minimize the effect of variation level of in yield among samples. It is apparent that relative to the first interval the rate of incorporation of [aH]thymine in
Growth and DNA synthesis of E. coli C600 (ColEJ thywith and without leucine. (A) Growth measured by ODs60with (-) and without (- - --) leucine. (B) DNA synthesis with (-) and without (- - - -) leucine. [ 3H]Thymineuptake was measured by adding 0.05-ml samples of the culture to 10 ml of water at 60",and collection of the cells on 0.45 f i pore Millipore filters, followed by two washes with 10 ml of water at 60".Filters were then dried, and the amount of radioisotope was determined. FIGURE 2:
chromosomal DNA was decreasing, as expected, while the rate of synthesis of CoEl DNA increased 2.0-2.5-fold and remained essentially constant when compared with either chromosomal DNA synthesis in the absence of leucine or C o E , DNA synthesis in the presence of leucine during the 0-20-min interval (Table 11). To preclude the possibility that the initial increment in the rate of CoE1 DNA synthesis was not directly related to starvation for leucine, 1 4C-prelabeled cells were pulsed as before with and without DL-leucine (40 pg/ml). The increment in rate of CoEl DNA synthesis was only observed in the absence of leucine (Table 11). The 25 depression in the rate of [ 3H]thymine incorporation into chromosomal DNA of the cells in the presence of leucine relative to the starved cells was consistently observed (Table 111) and is unexplained. Characterization of CoIEl DNA Synthesized During Amino Acid Starvation. To establish that the CoZE1DNA synthesized during amino acid starvation is the typical 23s form of CoZE, DNA, E. coli C600 (COEI)thy- cells were starved for leucine as described above and pulsed with [3H]thymine during the interval from 60 to 90 min after the removal of leucine. ColEl DNA was then purified by the sarkosyl-ethidium bromide techniq:ie and subjected to sucrose density gradient centrifugation. The sucrose gradient profile (Figure 3) was identical with that obtained for CoEl DNA synthesized during exponential growth on complete media (Bazaral and Helinski, 1968a). Experimental Approach to the Replication of ColEl DNA. A density label experiment similar to that of Meselson and
REPLICATION OF
C O ~ EA N~ D F
IN
E. c o l i
401
BIOCHEMISTRY
TABLE I: -
ColEl DNA and Chromosomal DNA Synthesis in the Absence of Leucine.
_____
____
_____
-
0-20 rnin Cpm
-~
963
'4C
ColEl 3,180 64,130
4 c
Chromosome 3H
130-1 50 min
83,170
~
R,
1,161 2.55
2.18
4,240 1 .oo
-
Cpm
1,280 2.55
3H
~
-~ -~
R,
Cpm
Rllb
__
70-90 min
3,280
63,860
70,700 0.37
0.21
30,812
19,383
a DNA was prelabeled with [14C]thyminefollowed by 20-min pulses with [3H]thymine during various periods as described in the text. b R, = 3H cpm/14C cpm normalized to R, = 1.0 for chromosomal DNA in the 0-20-min interval in the absence of leucine. The calculations do not take into account the increase in specific activity of the pulse. At the later times the specific activity of the [3H]thymineis increased (not more than 3 0 z ) by utilization of carrier thymine during the period preceding the pulse.
Stahl (1958) in which the density of the ColEl DNA as an integral unit was followed after a shift in the density of the medium was carried out to establish the existence of semiconservative replication of the Col factor, obtain some indication of the length of time necessary to replicate a CoEl DNA molecule and determine if each ColEl DNA molecule is limited to one round of replication per generation. Preliminary experiments demonstrated that 85 % DzO and 100% [ 'jN]H4C1isotope replacement in the minimal medium allowed exponential growth, and provided a density increment of 0.028 g/ml in the DNA compared with cells grown on normal medium. Ethidium bromide produces a density difference between supercoiled ColEl DNA and chromosomal DNA in buoyant CsCl of approximately 0.042 g/ml (Radloff et al., 1967; Bazaral and Helinski, 1968a); thus, a sufficient density difference remained to partly separate covalently closed ColEl DNA from chromosomal DNA, since in the extreme case the
103 I
density difference between heavy chromosome and normal ColEl DNA would be 0.014 g/ml. Since a broadening of the peaks in ethidium-CsCl was observed during the density shift experiments, further purification by sucrose gradient centrifugation was employed to ensure purity of the ColEl DNA fraction. A sample sucrose gradient profile of ColEl DNA and chromosomal DNA pools from a n ethidium bromide-CsC1 equilibrium centrifugation of DNA obtained from cells shifted from ljN, D 2 0 , and [ 'Elthymine medium to normal density [3H]thyminemedium for 40 rnin showed that pure 23s ColEl DNA could be isolated by this technique (Figure 4). When the purified 23s ColEl DNA (Figure 4, fractions 14, 15, and 16) and chromosomal DNA were separately subjected to CsCl preparative equilibrium centrifugation (Figure 5 ) , heavy, hybrid, and (for the ColE1 DNA) light DNA can be distinguished. The Kinetics of ColEl Replication Studied by Density Shift. To study the kinetics of ColEl DNA replication by density shift, E. coli C600 (CoE1)rhy- bacteria, preadapted to DzO, and 'jN were grown to a n of 0.45 on 50 ml of medium
I 1 '
6C0
I
TABLE 11: ColEl DNA and Chromosomal DNA Synthesis During 0 to 20 Min in the Presence and Absence of Leucine:
I
Complete Medium Minus Leucine
I
_ _ ~ _ _ _ _ _ _
Cpm FRACTION
3 : Sucrose gradient analysis of ColE, DNA labeled during amino acid starvation and purified by centrifugation in an ethidium bromide-CsC1 gradient.A 1. O m 1 sample of ColEl DNA in TES buffer containing approximately 0.3 fig of DNA and 4ooo cpm, derived from E . coli C600 (ColEJ thy- labeled during the period from 60 to 90 rnin after removal of leucine, was added to the sucrose gradient. Samples were centrifuged in a Spinco SW25.1 rotor for 8.75 hr, 20". (0-0) Col factor derived from amino acid starved cells; (x-x) 23s ColEl DNA derived from exponentially growing cells and run in an identical gradient in the same rotor. FIGURE
402
B A Z A R A L
AND
'4C
UVMBER
H E L I N S K I
Rnb
1,226
ColEi
Complete Medium ~~
Cpm 968
1 02
2 23 3H
4C Chromosome 3H
3,734
1,343
107,500
124,200 0 74
1 00
146,300
R,
134,100
~ _ _ _ _ _ _ _ _ ________ _
See footnote a, Table I. b R, = 3H cprn/'4C cpm normalized to 1.0 for chromosomal DNA in the absence of leucine. Q
V O L . 9, N O . 2, J A N U A R Y 2 0 , 1 9 7 0
TABLE 111: F Factor
DNA and Chromosomal DNA Synthesis During Amino Acid Starvation.. Complete Medium Minus Leucine
Complete Medium, 0-22 min CPm
0-22 min Cpm
Rnb
1,305
4 c
1,421
1 .oo
146,041
R,
CPm )
0.25
929
120,600
352
145,000
44,547
0.44
0.73 78)275
1.6%
Rn
1 159 0.39
108,900
1.1%
F factor/chromosome
0.89 2,137
11 8,722
'4c
60-82 rnin
1,944
0.89
Chromosome 3H
CPm
Rn
1,946
F factor
3H
30-52 rnin
0.26 16,100
1.3%
2.2%
DNA was prelabeled with [14C]thyminefollowed by 20-min pulses with r3H]thymine during various periods as described in the text. b R, = 3H cpm/14C cpm normalized to R, = 1.0 for chromosomal DNA in presence of leucine. As in the case of the data in Table I, no correction was made for increases in specific activity of [3H]thymine resulting from the utilization of carrier thymine during the course of the experiment. 0
containing [14C]thymine, DzO, and ISN. The cells were then harvested and resuspended in 75 ml of normal density medium containing 2 pg/ml of thymine and a total of 1.0 mCi of raH]thymine. The generation time of the bacteria measured by optical density was 105 min on heavy medium and 60 rnin on light medium. The apparent chromosomal duplication time 90 was rnin in light medium. At the times indicated in Figure 6,
P
6 3H: 9000 epm I4C: 2520 cpm m
h
51
10-ml samples were withdrawn and chilled. All samples were lysed and subjected to ethidium bromide-CsCl centrifugation, and the CoEl DNA samples were purified additionally by sucrose gradient centrifugation. Purified Col factor DNA and samples of chromosomal DNA from each time point were
'i
A
1'
4,
5
c)
l h
B
I4 ~
FRACTION
NUMBER
FIGURE 4: Sucrose gradients of chromosomal and ColEl DNA derived by cesium chloride-ethidium bromide density equilibrium centrifugation of sarkosyl lysates of density-shifted E. coli C600 (CoEl)thy- cells. (A) Chromosomal DNA prepared as described in *H; (----) lac.(B) ColEt DNA prepared as dethe text. (-) scribed in text. (-) 3H;(- - - -) lac. Both gradients were centrifuged in the same SW25.1 rotor for 8.75 hr, 20". Samples of 0.2 ml were taken from each fraction for determination of radioisotope.
Density equilibrium centrifugation of DNA derived from density-shifted E. coli C600 (CoE1) thy-. (A) Chromosomal DNA from the cesium chloride-ethidium bromide centrifugation of sarkosyl lysates. (0-0) 3H; (x----x) lac.(B) ColEl DNA after the additional purification step of sucrose gradient centrifugation. (0-0) 3H;(x- - - -x) W. Fractions are numbered in order collected from the bottom of the tube. Heavy, hybrid, and light DNA are indicated by H, Hy, and L, respectively. FIGURE 5 :
REPLICATION
OF
C O ~ EA' N D F
IN
E. c o l i
403
BIOCHEMISTRY
,?’:
I,
a‘
I 2 3
SAMPLE -_-6
4
56
NUPBE’ 5
HCL‘S
6: Growth of E. coli C600 (ColEI) thy- bacteria during the density-shift experiment. Bacteria were grown on heavy medium containing [14C]thymine,harvested, and washed 3.4 hr after initiation of the experiment, and then resuspended in normal density medium containing [ 3H]thymine as described in the text. Samples were taken at the numbered times shown in the figure. FIGURE
subjected to preparative CsCl equilibrium centrifugation. The chromosomal and CoEl DNA both were initially heavy (Figure 7, graphs I Cand le). Chromosomal DNA, during DNA synthesis in light media, shifted density in a predictable manner, and in one generation (Figure 7, graph 5c) most of the cells had replicated their chromosomes once while only a small proportion of cells had begun a second round of replication. The ColEl factor did not replicate in the same manner, however; second rounds of synthesis were apparent when little more than half of the chromosomal D N A replicated once (Figure 7, graph 4). Even after 1 and 1.5 chromosomal DNA replications, one-third of the Co& DNA remained unreplicated. In a similar experiment 15-min pulses of [ 3H]thymine were used during the growth of heavily labeled cells in normal medium to establish a n upper limit for the replication time of the complete CoI factor molecule. The experimental procedure was identical with the density shift experiment described above, except that after the cells were removed from the heavy 14Cmedium, they were resuspended in medium containing only unlabeled thymine (1.7 pgiml). Two 10-ml cultures were pulsed with 0.25 mCi of [3H]thymine for 15 min, one from the beginning of growth in normal medium, and the other beginning after 15-min growth in normal medium. In each case the [ 3H]thymine was incorporated into a symmetrical hybrid density peak of ColE, DNA. If the replication time of the complete CdE1 DNA molecule was as long as 15 min, the density of the DNA into which the tritium was incorporated would have been skewed toward full heavy density. The complete absence of such skewing suggests that the replication time for a colicinogenic factor is considerably less than 15 min. Characterization of F Factor DNA. At least some of the F factor DNA has been shown to be a covalently closed circular DNA molecule (Freifelder, 1968). Molecules of this configuration should be readily isolated using the rapid sarkosylethidium bromide-cesium chloride technique used for the ColE factors. As a preliminary step in the characterization of F factor DNA as isolated by the direct dye-buoyant density procedure, it was necessary to demonstrate that the host strain employed harbored no additional covalently closed circular episomal elements. To achieve this end the procedure was tested on a mixed suspension of E. coli CR34 and CR34 (F) strains differentially labeled with radioisotope. Cultures
404
B A Z A R A L
AND
ELI IN SKI
FRACTION N U M B E R
7: Density-equilibrium centrifugation of chromosomal and ColEl DNA derived from the density-shift experiment. The number
FIGURE
in the upper left corner of each graph indicates the time at which the sample was taken (see Figure 6), and “C” and “E” designate chromosomal and ColEl DNA, respectively. Fractions are numbered in order as collected from the bottom of the tube. (0----0) 3H;(x---x) lac. Total counts for each graph for 3H and lac,respectively: (IC) 0, 38,400; (le) 0,361; (2c) 12,100,49,900; (2e) 168,445; (3c) 46,300, 49,000; (3e) 663, 606; (4c) 22,433, 8,500;(4e) 279, 117; (5c) 132,200, 21,400; (5e) 993,190; (6c) 195,000,23,400;(6e) 5,019,650.
(10 ml) of CR34 and CR34 ( F )were labeled with 0.01 mCi of [14C]thymineand 0.1 mCi of [3H]thymine, respectively. After labeling, the two cultures were each washed once, then resuspended together. D N A from the cell mixture was then analyzed by dye-buoyant density centrifugation of a sarkosyl lysate of the mixture. The radioactivity profile resulting from the centrifugation is shown in Figure 8. The results indicate that the satellite DNA was derived exclusively from F-containing cells. Sucrose gradient analysis of satellite DNA derived from a similar experiment showed that the F factor DNA consisted of two homogeneous classes, sedimenting at approximately 80 and 48 S (Figure 9). Assuming that the 80s and 48s peaks correspond to the covalently closed and open circular forms, respectively, of Ffactor DNA, a molecular weight of 75 X lo6 can be estimated for the circular F factor (Bazaral and Helinski, 1968b). Contour length determinations on F factor D N A are consistent with this molecular weight estimate (B. Kline and D. R. Helinski, unpublished results). On the basis of a molecular weight estimate of 75 X lo6 for the F factor, the maximum observed yield of 2 . 2 x (Table 11) corresponds to 0.7 closed circular F factor isolated per chromosome, and is consistent with the expected number of one copy of the factor per chromosome. Initiation of F Factor Synthesis During Amino Acid Staruation. To determine the characteristics of initiation of F factor synthesis in the absence of protein synthesis, the kinetics of the synthesis of F factor and chromosomal DNA were
V O L . 9, N O . 2, J A N U A R Y 2 0 , 1 9 7 0
I
chromosome
I
z
P
v)
bz
FRACTION NUMBER
3
0 u
Dye-buoyant equilibrium centrifugation of [14C]thyminelabeled CR34 DNA and [ SH]thymine-labeledCR34 (F)DNA. Experimental details are presented in Results. (-) 3H; (- - -) 14C.
FIGURE 8:
measured. A culture (30 ml) of E. coli CR34 (F)was grown to an OD660 of 0.56 in supplemented M9 medium containing 0.015 mCi of [14C]thymine.The culture was then washed and resuspended in 40 ml of M9 medium lacking leucine and containing 1 pg/ml of thymine. The resuspended cells were then grown in four separate 10-ml cultures. To the first, 40 pg of DL-leucine and 0.1 mCi of [3H]thymine were added immediately after resuspension. Labeling in the absence of leucine was begun by the addition of 0.1 mCi of [3H]thymineto the remaining cultures beginning 0, 30, and 60 min after resuspension. All cultures were labeled for 22 min, and at the end of the labeling period were kept on ice until the end of the experiment. The growth of the culture, as measured by OD66O, during the experiment, was similar to the growth of C600 (ColE1) thy- shown in Figure 2. The 3H and 14C radioactivity in the chromosomal and F factor DNA were determined in each sample after sarkosyl lysis and dye-buoyant centrifugation. Synthesis of DNA during starvation (3H label) was normalized to the 14C prelabel to minimize the effect of variation in yield among samples. It is apparent that relative to the cells grown with leucine, the rate of synthesis of the chromosome was decreasing during the course of the experiment. Unlike the CoE1 factor, however, the rate of synthesis of F factor also was decreasing. If 20 min is sufficient time to complete any F factor DNA synthesis begun before amino acid starvation, F factor DNA synthesis must have been initiated in the absence of protein synthesis (if the synthesis of F factor DNA occurs at the same rate as chromosomal DNA, less than 2 min is required for the synthesis of the F factor under the growth conditions of these experiments). However, unlike C O R , DNA, the initiation of F factor DNA duplication is eventually markedly reduced in the absence of protein synthesis. Discussion The observation that in the natural host, E . coli, there are several Co& factors per chromosome suggested that the mechanism of replication of these factors might be different in certain respects from the replication of the chromosome. Experiments designed to directly characterize the replication of ColE1DNA show that in one generation, under conditions in which the chromosome is replicating normally, some of the C o E l DNA molecules replicate twice, and an approximately equivalent number do not replicate at all. The relative proportion of unreplicated to once-replicated to twice-replicated
9: Sucrose gradient analysis of E. coli CR34 ( F ) satellite DNA. (A) Mixed F factor DNA and CdE2 DNA marker DNA (25 S). (B) F factor DNA alone. Sucrose gradients (5 ml) were centrifuged using a Spinco SW65 rotor. Details of centrifugation are described in the Experimental Section. FIGURE
molecules after one generation is approximately 1 :2 :1. Certainly, if a steady state is maintained with respect to CoEl DNA content, then in any one generation the number of unreplicated molecules must be equal to the number replicated twice, and the experiment is thus internally consistent. The observed ratios at the end of one generation suggest a model in which the initiation of rapid synthesis of the CoEl DNA molecules occurs randomly with respect to the previous replication of the molecules, regardless of whether or not the molecule had been previously replicated in that generation. Calculations based on the assumptions that there are four CoEl DNA molecules in the replicating pool at the beginning of growth in normal media, and that the replication is unaffected by cell division predict that the proportion of unreplicated Co& DNA molecules at the end of one net doubling should be 21 %. In addition to these calculations, the kinetics of appearance of twice-replicated molecules are qualitatively in agreement with the random model. Nevertheless, the possibility that there is a very strict mechanism for the selection of molecules to be replicated which coincidently produces the same pattern of replication as a random process cannot be excluded. It is possible, however, to preclude all models in which the initiation of synthesis of each factor molecule occurs once and only once in a generation. In this case the kinetics of density shift would be identical with that of the chromosomal DNA. It is also possible to exclude a model in which replication takes place using a single template molecule, similar to the early replication of (PX-174 (Sinsheimer, 1968), since at least one-half of the molecules are replicated once or more during one generation. Clearly, these conclusions are based on the assumption that the replication of the CoEl factor DNA 8 Calculations are described in the thesis of M. B. submitted to the Graduate School of the University of California, San Diego, Calif,, in partial fulfillment of the Ph. D. degree.
REPLICATION OF
Col&
AND
F
IN
E. c o l i
405
BIOCHEMISTRY
after a density shift in the medium is not unlike the mechanism of replication under normal growth conditions. Support for this assumption comes from the observation that under the density shift conditions employed chromosomal DNA replication apparently proceeded normally. Several other conclusions may be drawn from the density shift experiment, although these merely confirm that the basic character of the ColEl DNA replication is like all other double-stranded DNA. Like chromosome replication, ColE, factor replication is semiconservative. The premature light DNA in the density shift experiment occurs only after an appreciable pool of hybrid DNA is accumulated, and is indicative of a second round of replication rather than conservative replication. Since the replication is semiconservative, there must be a mechanism for causing at least one single-strand break in order for strand separation to occur during replication, and there must also be a mechanism for repairing the break. The minimum replication time established by these experiments is much less than 15 min, and there is no reason to assume that the number of nucleotides per minute incorporated into ColEl DNA molecules in the act of replication differs significantly from the rate observed for the chromosome. The control of the synthesis of ColEl DNA is further distinguished from that of the chromosome by the fact that the ColE, DNA synthesis is increased in rate and continues linearly during amino acid starvation, whereas under these conditions the initiation of chromosome synthesis is prevented. A formal hypothesis that is consistent with all of these observations is that ColEl molecules are replicated randomly at intervals during one generation by a mechanism which is limited by the number, or the activity, of a stable replicator, rather than the number of templates. This replicator would be subject to regulatory control by the metabolic state of the cell, since the rate of ColEl DNA synthesis was increased during amino acid starvation. This hypothesis does not account for the regulation of the number of replicators during normal growth. The relationship between the proposed replicator and the protein material associated with ColEl DNA when gently isolated (Clewell and Helinski, 1969) is presently under study. Another important consideration in the maintenance of the Co/El factor in a population of cells is the mechanism of regulation of the segregation of Cof factor molecules. The F factor DNA molecules have been reported to segregate with those parental chromosome strands that they were associated with before replication of the chromosome (Cuzin and Jacob, 1965). The same mode of segregation cannot occur for all of Co/El DNA molecules in the cell since one-half of the molecules have either replicated twice or not at all in one generation. The only other direct studies of the replication of stable extrachromosomal genetic elements in bacteria were concerned with the replication of R factor DNA in Proteus mirabilis (Rownd et a/., 1966). These elements were shown to undergo replication similar to the Co/El of E. coli. F factor synthesis in the absence of amino acid was also
406
BAZARAL
AND HELINSKI
examined to determine if the control of different extrachromosomal genetic elements is similar. Unlike ColE, DNA, however, the F factor DNA initiation was depressed in the absence of protein synthesis, although not as rapidly, or to the same extent, as the initiation of chromosomal DNA synthesis. This dissimilarity in the control of synthesis of F and ColE, DNA might have been anticipated on the basis of the fact that there are multiple copies of the ColEl DNA per chromosome, as contrasted with one copy of the F factor DNA. Despite the apparent difference in the regulation of their initiation, chromosomal and F factor synthesis remain in proportion during amino acid starvation. One possibility suggested by this observation is that the initiation of F factor synthesis is dependent on chromosome synthesis, through a mechanism which does not require protein synthesis. Studies are in progress to examine the protein factors that may be involved in the initiation of CoE1 and F factor DNA synthesis in E. coli cells carrying these extrachromosomal circular DNA elements. References Bazaral, M., and Helinski, D. R. (1968a),J . Mol. Biol. 36, 185. Bazaral, M., and Helinski, D. R. (1968b), Biochemistry 7, 3513. Bollum, F. J. (1966), in Procedures in Nucleic Acid Research, Cantoni, G. L., Davies, D. R., Ed., New York, N. Y . , Harper & Row, p 296. Clark, A. J., and Adelberg, E. A. (1962), Aim. Rec. Microbiol. 16,289. Clewell, D., and Helinski, D. R. (1969), Proc. Natl. Acad. Sci. U. S. 62, 1159. Cuzin, F., and Jacob, F. (1965), C. R. Acad. Sci. Paris 260, 541 1. DeWitt, W., and Helinski, D. R. (1965), J. Mol. Biol. 13, 692. Freifelder, D. (1968),J . Mol. Biol. 35,95. Hayes, W. (1953), Cold Spring Harbor Symp. Quant. Biol. 18,75. Kahn, P., and Helinski, D. R . (1964),J. Bacteriol. 88, 1573. Lark, K. G., Repko, T., andHoffman, E. J. (1963), Biochim. Biophys. Acta 76,9. Maalpe, O., and Hanawalt, P. C. (1961), J . Mol. Biof. 3, 144. Meselson, M., and Stahl, F. W. (1958), Proc. Natf. Acad. Sci. U. S. 44,671. Nagel de Zwaig, R., and Puig, J. (1964), J . Geri. Microbiol. 36, 311. Radloff, R., Bauer, W., and Vinograd, J. (1967), Proc. Natl. Acad. Sci. 57, 1514. Rownd, R., Nakaya, R., and Nakamura, A. (1966), J . Mol. Biol. 17, 376. Sinsheimer, R. L. (1968), in The Molecular Biology of Viruses, Crawford, L. V., and Stoker, M. G. P., Ed., Cambridge, England, University Press, p 101. Stacey, K. A., andSimson, E. (1965),J . Bacteriol. 90,554. Wohlhieter, J. A., Falkow, S., CitareUa, R. V . , and Baron, L. S. (1964),J . Mol. Biol. 9, 576.