JOURNAL OF BACTEROLOGY, June 1972, p. 1135-1146Copyright i 1972 American Society for Microbiology

Vol. 110, No. 3Printed in U.S.A.

Effect of Growth Conditions on the Formationof the Relaxation Complex of Supercoiled ColEl

Deoxyribonucleic Acid and Protein inEscherichia coli

DON B. CLEWELL' AND DONALD R. HELINSKI

Department of Biology, University of California, San Diego, La Jolla, California 92037

Received for publication 31 January 1972

Colicinogenic factor El (ColEl) is present in Escherichia coli strain JC411(ColEl) cells to the extent of about 24 copies per cell. This number does notappear to vary in situations which give rise to twofold differences in theamount of chromosomal deoxyribonucleic acid (DNA) present per cell. If cellsare grown in the absence of glucose, approximately 80% of the ColEl moleculescan be isolated as strand-specific DNA-protein relaxation complexes. Whenglucose is present in the medium, only about 30% of the plasmid molecules canbe isolated as relaxation complexes. Medium shift experiments in which glu-cose was removed from the medium indicate that within 15 min after the shiftthe majority (>60%) of the plasmid can be isolated as relaxation complex. Thisrapid shift to the complexed state is accompanied by a two- to threefold in-crease in the rate of plasmid replication. The burst of replication and the shiftto the complexed state are both inhibited by the presence of chloramphenicol.Inhibition of protein synthesis in log cultures by the addition of chloramphen-icol or amino acid starvation allows ColEl DNA to continue replicating longafter chromosomal replication has ceased. Under these conditions, noncom-

plexed plasmid DNA accumulates while the amount of DNA that can be iso-lated in the complexed state remains constant at the level that existed prior totreatment. In the presence of chloramphenicol, there appears to be a randomdissociation and association of ColEl DNA and "relaxation protein" during or

between rounds of replication.

The colicinogenic factor El (ColEl), a bac-terial plasmid determining the production ofthe antibiotic protein colicin El, has been iso-lated from gently lysed cells as a supercoiledcircular deoxyribonucleic acid (DNA) molecule(molecular weight of 4.2 x 106) tightly com-plexed to a specific protein (5, 6). The isolatedcomplex has the property of undergoing aunique phenomenon when treated with certainionic detergents, proteases, heat, or alkali. Inresponse to any one of these treatments, thetwisted circular DNA in the complex under-goes a relaxation to the open circular state,and a strand-specific nick or gap is present inthe open circular DNA product (2, 5, 6). Sim-ilar complexes have been found in the ColE2,ColE3, and ColIb plasmid systems (2, 7, 8), as

I Present address: Departments of Oral Biology and Mi-crobiology, University of Michigan, Ann Arbor, Mich. 48104.

well as the F1 sex factor system (11) of Esch-erichia coli.

In this communication, data will be pre-sented dealing with some physiological aspectsof the ColEl relaxation complex. It will beshown that the number of plasmid moleculesin the complexed state differs depending onthe medium employed for bacterial growth.Furthermore, it will be demonstrated that thenumber of plasmid molecules per host chromo-some is not constant in different media butrather that there is a constant number per cell.Finally, data are presented on the synthesis ofplasmid DNA and relaxation protein underconditions of medium shifts, and inhibitionof protein synthesis.

MATERIALS AND METHODS

Materials. Reagents and sources were as follows:1135

on June 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

CLEWELL AND HELINSKI

Brij 58, from Atlas Chemical; sodium deoxycholate(DOC) from Mann Research Laboratories; sodiumdodecyl sulfate (SDS) from Fisher Scientific; Sar-kosyl NL30 (sodium dodecyl sarcosinate) from theGeigy Chemical Company; egg-white lysozyme fromWorthington Biochemical Corp.; CsCl (technicalgrade) from Penn Rare Metals, Division of KaweckiChemical Company; chloramphenicol (CAP) fromParke, Davis and Company; ethidium bromide was agift of the Boots Pure Drug Company Ltd., Not-tingham, England; [methyl-'H]thymine (18.2Ci/mmole) and [2-'4C]thymine (55.2 mCi/mmole)from New England Nuclear Corp.

Bacteria and media. E. coli K-12 strains JC411thy- (ColEl) (4) (a thymine-requiring strain isolatedby D. Kingsbury) and CR34 (ColEl) (12) were uti-lized in the present study. Each strain harbors theColEl plasmid DNA which was transferred by conju-gation from E. coli strain K-30. The tris(hydrox-ymethyl)aminomethane (Tris)-Casamino Acids me-dium used for these strains has been described indetail previously (5). Normally, 2 ug of thymine perml was present in the medium. Unless otherwisespecified, growth was performed at 37 C.

Preparation of cleared lysate. The lysing proce-dure has been described in detail previously (5, 6). Itessentially involved the lysis of ethylenediamine-tetraacetic acid (EDTA)-lysozyme spheroplastswith a detergent mixture of Brij 58 and DOC fol-lowed by a low-speed centrifugation of the lysate.The latter centrifugation removes about 95% of thetotal DNA leaving most of the plasmid DNA (bothcomplexed and noncomplexed) in the supernatantfluid. This supernatant fluid is referred to as acleared lysate.

Preparation and dye-buoyant centrifugation ofSarkosyl lysate. Two 15-ml log cultures of cells la-beled separately with [3H]thymine and ["4C]thy-mine were centrifuged, and the cell pellets were sus-pended together in 3.4 ml of 25% sucrose (containing0.05 M Tris, pH 8.0). To this suspension was added0.4 ml of lysozyme (5 mg/ml), followed by, after 5min of incubation at 25 C, 0.8 ml of Na2 EDTA (0.25M, pH 8.0). After 5 min of further incubation, a 1-mlportion was removed for lysis with Brij 58 while theremainder was lysed by adding 2 ml of 2% Sarkosyl.The Sarkosyl lysate was centrifuged to equilibriumin CsCl-ethidium bromide, fractionated, andcounted as described previously (6).Sucrose density gradients. Sucrose density gra-

dient centrifugation, dropwise fractionation of thegradients, and the counting of radioisotope were car-ried out as described in detail previously (6). Unlessdesignated otherwise, the gradients contained TES(0.05 M NaCl, 0.005 M Na2 EDTA, and 0.03 M Tris,pH 8.0). The high salt gradients contained 0.55 MNaCl, 0.005 M Na2 EDTA, and 0.03 M Tris, pH 8.0.Medium shifts. In experiments involving a me-

dium shift, cells were harvested by centrifugation ina Sorvall centrifuge (SS39 rotor) at 25 C for 4 min at10.000 rev/min; washed by resuspending in twice theoriginal volume of the medium; recentrifuged andresuspended in the new medium. The entire proce-dure required 12 to 15 min. Cell growth during the

shift was minimal as evidenced by less than a 10%increase in turbidity after the shift. Turbidity wasmeasured with a Klett-Summerson colorimeter.

RESULTSEffect of glucose on the level of complexed

ColEl DNA. It was observed that the per-centage of ColEl DNA in the form of a relaxa-tion complex of supercoiled DNA and proteinwas not constant when purified from an E. colistrain growing in different media. To deter-mine what effects glucose might have on thepercentage of ColEl DNA in the complexedstate, the following experiment was performed.E. coli strain JC411 thy- (ColEl) cells weregrown in two different 15-ml batches of Tris-Casamino Acids-glycerol medium. One batchcontained in addition, 0.5% glucose and [3H]-thymine (0.3 mCi); the other portion lackedglucose and contained ['4C]thymine (15 ,Ci).Each culture was harvested in log phase, afterwhich the cells were suspended and mixedtogether in a solution of 25% sucrose (0.5 MTris, pH 8.0). After preparation of lysozyme-EDTA spheroplasts, a 1-ml portion of themixed suspension was lysed with a Brij 58-DOC detergent mix, whereas the remainderwas lysed with Sarkosyl. A cleared lysate (seeMaterials and Methods) was prepared fromthe Brij 58-DOC lysate and subsequently ana-lyzed by sucrose density gradient centrifuga-tion. Prior to centrifugation, one portion wastreated with SDS while another portion servedas an untreated control. Since SDS convertsonly the supercoiled ColEl DNA that is in theform of relaxation complex to the open circularstate, the extent of conversion after the addi-tion of SDS is a quantitative measure of thelevel of complexed ColEl DNA. The result,which is shown in Fig. 1, indicates that,whereas the ColEl DNA extracted from thecells grown in the absence of glucose (the "C-label) is in the form of a relaxation complex ofsupercoiled DNA and protein to the generallyobserved extent of about 82% (5), ColEl DNAobtained from cells grown in the presence ofglucose (the 3H-label) was complexed to theextent of only about 33%.

This difference in the level of complexedColEl DNA was also observed upon examina-tion of total cellular DNA in the Sarkosyllysate. When the Sarkosyl lysate was centri-fuged to equilibrium in an ethidium bromide-CsCl buoyant density gradient, the profileshown in Fig. 2 was obtained. It has beendemonstrated previously (5) that either Sar-kosyl or the equilibrium centrifugation proce-

1136 J. BACTERIOL.

on June 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

COLE1 DNA IN E. COLI

N

Qx

a.)O)

cli

bx

a.)0

10 20 30 10 20 30

FRACTION NUMBERFIG. 1. Sedimentation analysis of cleared lysates ot cells grown in the presence and absence of glucose.

Strain JC411 thy- (ColEl) cells grown as described in the text in the presence and absence of glucose wereharvested and resuspended together. From a portion of the resuspended cells, a cleared lysate was preparedas described in Materials and Methods and diluted twofold with TES. One 0.3-ml portion was mixed with0.1 ml of 1% SDS, while another 0.3-ml portion was mixed with 0.1 ml of TES to serve as a control. Afterincubation at 25 C for 10 min, the samples were sedimented (from right to left) through 5 to 20o sucrosedensity gradients containing 0.55 M NaCI in an SW50.1 rotor (15 C) at 50,000 rev/min for 135 min. (0) 3H-labeled DNA prepared from cells grown in presence of glucose. (0) "4C-labeled DNA prepared from cellsgrown in the absence of glucose. (a) Untreated cleared lysate; (b) cleared lysate treated with SDS. NC indi-cates noncomplexed ColEl DNA; C indicates complexed ColEl DNA; RC indicates the relaxed state ofcomplexed ColEl DNA.

50-

40 12131 15 4

-30 2 33x Ix

020 I2

10 ~

10 20 ~ 30

FRACTION NUMBERFIG. 2. Dye-buoyant density gradient centrifuga-

tion of a Sarkosyl lysate of cells grown in the pres-ence and absence of glucose. A portion of the mixedcell suspension described in Fig. 1 was lysed withSarkosyl and centrifuged to equilibrium in a CsCI-ethidium bromide gradient in a Ti 60 rotor (15 C) at44,000 rev/min for 60 hr. (0) 3H-labeled DNA pre-pared from cells grown in the presence ofglucose; (0)"C-labeled DNA prepared from cells grown in theabsence of glucose. The numbers 1 to 5 designateregions where fractions were pooled for subsequentanalysis on sucrose density gradients.

dure itself promotes the relaxation of com-plexed ColEl DNA, with the result that non-complexed DNA appears in the dense bandcharacteristic of covalently closed circularDNA (13) while the relaxed DNA which re-mains associated with protein appears in the"light" portion of the large chromosomal DNAband (6). When DNA from the various poolsindicated in Fig. 2 were dialyzed and analyzedseparately on sucrose gradients (Fig. 3), themajority of the "4C-labeled ColEl DNA wasfound in the open circular 17S form and in the"light" portion of the noncovalently closedband (Table 1). However, the 3H-labeledColEl DNA appeared predominantly as 23Ssupercoiled DNA in the satellite band of theequilibrium gradient. Table 2A shows that theproportions of relaxed and supercoiled mole-cules correspond closely to that observed inthe SDS-treated cleared lysate (Fig. 1). Table2B summarizes the results of a similarly per-formed experiment, except that the isotopeswere reversed. Essentially identical resultswere obtained. The reduction in the amount ofplasmid DNA in the complexed state thusappears to be related to the presence of glu-cose. Although in these experiments glycerolwas present in addition to glucose, in otherexperiments where glycerol was not presentsimilar results were obtained. It was furthershown that glycerol itself is not involved indetermining the amount of plasmid DNA inthe complexed state, since cells grown in a

1137VOL. 110, 1972

on June 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

CLEWELL AND HELINSKI

2

4

2

to 4

x

0

-r' 2

4

2

8

4

x

4 20-

0L)

4 ¢

2~~~~~~~

2 17S Pool 5 4

, ,'

10 20 30

FRACTION NUMBERFIG. 3. Sedimentation analysis of the fractions

pooled after dye-buoyant density centrifugation ofthe Sarkosyl lysate of cells grown in the presenceand absence of glucose. Sedimentation was through 5to 20% sucrose density gradients as described in Fig.1. (0) 3H-labeled DNA prepared from cells grownin the presence of glucose; (0) "4C-labeled DNAprepared from cells grown in the absence of glucose.

Casamino Acids medium (lacking glucose)yield identical results regardless of whether or

not glycerol is present.The percentage of ColEl DNA molecules in

the complexed state is not a function of the

TABLE 1. Relative amounts of supercoiled andrelaxed plasmid DNA recovered after dye-buoyant

density centrifugation of a Sarkosyl lysatea

Amt of plasmidPecn-DNA (counts/

Pool Counts/min plasmid min)in p~ool DNAb Super- Re-

coiled laxed

A. 3H counts1 49,027 13.76 6,7462 683,099 0.08 5463 269,262 0.27 7274 110,641 0.27 2985 30,462 2.10 639Total 1,142,491 0.78 6,746 2,210

B. 'IC counts1 2,468 17.08 4212 87,710 0.42 3683 32,468 2.31 7504 11,924 4.59 5475 3,856 19.70 759Total 138,426 2.06 421 2,424

a DNA prepared and analyzed as described in Fig.2.

bBased on the amount of plasmid DNA in eachpool expressed as a percentage of the total countsrecovered in each sucrose density gradient.

TABLE 2. Percentage of complexed andnoncomplexed ColEJ DNA obtained from cleared

and Sarkosyl lysatesa

Plasmid DNA Plasmid DNA

nclard in crudeiclesated Sarkosyl

Determination lysate(So) Sateo(%)Com- Non- Com- Non-plxdcorn- plxdcorn-plexed plexed plexed plexed

A. Glucose present (3H) 33b 67b 29c 71cGlucose absent (WC) 82b 18b 88c 12c

B. Glucose present ("C) 21 79 20 80Glucose absent (3H) 88 12 87 13

a Same preparation of cells was used for both anal-yses as described in Fig. 1 and 2 and the text.

b Data taken from Fig. 1.c Data taken from Table 1.

growth rate of the cells. Experiments in whicha rich medium lacking glucose (CasaminoAcids-yeast extract) was used to allow a rela-tively rapid growth rate (45-min doublingtime) yielded data indicating a typically highpercentage of ColEl DNA in the form of a re-laxation complex. In contrast, cells growingrelatively slowly (90-min doubling time) in a

J. BACTERIOL.1138

on June 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

COLE1 DNA IN E. COLI

minimal medium but in the presence of glu-cose always yielded a low percentage ofplasmid DNA in the complexed state.

In addition to the striking effect of the pres-ence of glucose on the percentage of plasmidDNA in the complexed state, it became ap-parent from the above experiments that thepercentage of total DNA that is plasmid DNAin the case of glucose-grown cells was aboutone-half that found in the case when glucosewas absent (Table 1). This observation is alsoreflected in data of Fig. la in which the ratioof 3H to I4C in the ColEl DNA peak (ratio of3.9) is only about one-half that of the chromo-somal DNA (ratio of 7.4) indicated by the ratioof counts in the crude lysate. This observationhas been reproduced a number of times withsimilar results. From four experiments, av-erage values of 0.93 and 1.72 for the percentageof the total DNA representing plasmid havebeen obtained when glucose was present andabsent, respectively. It was determined, on thebasis of cell counts of cultures of equal tur-bidity (using both the Petroff-Hauser countingchamber and viable cell counting), that cellsgrown in a glucose-glycerol-Casamino Acidsmedium (doubling time of 45-60 min) haveapproximately twice the mass of cells grown ina glycerol-Casamino Acids medium (doublingtime of 80-100 min). On the basis of the spe-cific activity of isotope in the DNA [as deter-mined by the method of Burton (3)], the totalamount of DNA present per cell was estimatedto be 16.9 x 10-15 g/cell for cells growing inthe presence of glucose and 8.0 x 1015 g/cellfor cells growing in the absence of glucose. Theamount of DNA present can be expressed as4.0 and 1.9 genome equivalents per cell, re-spectively, assuming that the size of the E. coligenome is 2.5 x 109 daltons (or 4.2 x 10-15 g).These values are higher than those reported byCooper and Helmstetter (9) for E. coli strainB/r cells growing with doubling times of 50and 90 min.

Nevertheless, cells growing in CasaminoAcids-glucose medium, although approxi-mately twice the size of cells growing in Casa-mino Acids-glycerol medium, also have ap-proximately twice as much chromosomal DNAper cell. The fact that the ratio of plasmidDNA to chromosomal DNA in the former situ-ation is about one-half that in the latter caseindicates that, independent of the amount ofchromosomal DNA per cell, cells grown in thepresence of glycerol or glucose have the sameamount of plasmid DNA per cell. We havepreviously estimated that, under conditionswhere glucose was absent, the number of

plasmid molecules per chromosomal genomeequivalent is approximately 12 (6). Since thereare about two genome equivalents per cellunder these conditions, the average number ofplasmid molecules per growing cell shouldtherefore be 24 in Tris-Casamino Acids me-dium irrespective of the presence of glucose.

Immediate effects of the removal of glu-cose. A 30-ml culture of strain JC411 thy-(ColEl) cells was grown for several generationsin a medium containing 0.5% glucose and["4C]thymine (15 ,Ci). While in log phase theculture was divided into three equal portions,pelleted, and resuspended into three differentmedia (20 ml each). All three media containedequal amounts of [3H]thymine (0.8 mCi) butno "C-isotope. In addition, one of the media,the control, contained 0.5% glucose and thusresembled the prelabeling conditions. In theother two, glucose was absent, and in one caseCAP (30 gg/ml) was present. All three cellsuspensions were incubated for 15 min, atwhich time they were rapidly chilled on ice.Cleared lysates were prepared, treated or nottreated with SDS, and analyzed on sucrosegradients. The results are shown in Fig. 4. Onthe basis of the extent of SDS-induced relaxa-tion of the ColEl DNA, it is clear that in thecase where glucose was removed the majority(>60%) of the plasmid DNA was found in thecomplexed state. This case is in contrast tothose in which CAP was present or glucose wasmaintained in the medium. The ratio of 3H to"C provides a measure of the relative extentsof DNA replication after the medium shift.The ratios in each case for plasmid and chro-mosomal DNA are shown in Table 3. It is evi-dent that the removal of glucose from the me-dium resulted in greater than a onefold in-crease in the rate of synthesis of plasmid DNAcompared to that of the parallel culture whereglucose was present. When CAP was present,an increase in the rate of DNA synthesis wasnot observed. In all cases, the rate of replica-tion of chromosomal DNA did not vary.When cells labeled with [3H]thymine were

grown in the presence of glucose for severalgenerations and then starved for thymine foras long as 1 hr, no difference was observed inthe amount of complexed DNA compared tothat of the prestarved state. However, if glu-cose was removed during the thymine starva-tion, the plasmid DNA was then found pre-dominantly (-90%) in the complexed state.The presence of CAP inhibited this shift.To determine what happens in the reverse

situation, a 30-ml culture of strain JC411 thy-(ColEl) cells was grown for several generations

1139VOL. 110, 1972

on June 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

CLEWELL AND HELINSKI

8

4

Cl*j

0

x

020_

8

4

4

2

cm4 0

x

2 EXL-

qt.

4

2

l0 20 30 10 20 30

FRACTION NUMBERFic. 4. Sedimentation analysis of cleared lysates to determine the effect of the removal of glucose from

the medium. Cleared lysates were prepared from cells grown in the presence of glucose and then shifted to amedium lacking glucose in the presence and absence of CAP as described in the text. The cleared lysateswere incubated in the absence and presence of SDS and centrifuged (from right to left) as described in Fig. 1.(a) 3H-labeled DNA synthesized after the medium shift; (0) "4C-labeled DNA present prior to the mediumshift. Cleared lysates were prepared from the following cells. (a) Cells shifted into a medium identical (withthe exception of the isotope) to the preshift medium; (b) cells shifted to a medium lacking glucose; (c) cellsshifted to a medium lacking glucose but where CAP was present; d, e, and f correspond to the samples a, b,and c, respectively, except that these samples were treated with SDS prior to centrifugation. The designa-tions NC, C, and RC are described in Fig. 1.

TABLE 3. Effect of the immediate removal of glucosefrom the medium on the rate of replication of ColE)

DNAa

3H counts per min/P"Ccounts per min

Growth condition Chromo-somal PDasmidDNA

DN

Glucose maintained ....... 2.31 2.38Glucose removed .......... 2.68 6.42Glucose removed, chloram-

phenicol present ......... 2.00 1.64

" The data were taken from the experiment de-scribed in Fig. 4. Chromosomal DNA counts wereobtained from the crude lysates used for the prepara-tion of cleared lysates in this experiment.

in the absence of glucose (Casamino Acids andglycerol present) and in the presence of [3H]-thymine (0.6 mCi). In log phase, the culture

was divided into two 15-ml portions and glu-cose (final concentration of 0.5 M) was addedto one portion. After 15 min, the cells werechilled on ice and cleared lysates were pre-pared and analyzed on sucrose density gra-dients. The results are shown in Fig. 5. Treat-ment with SDS produced a similar amount ofrelaxation in each case; thus, the addition ofglucose did not appear to affect significantlythe amount of plasmid DNA in the complexedstate during the 15-min time period of theshift. Similar experiments have shown thateven after 30 min there is no noticeable effectof the addition of glucose on the level ofplasmid DNA in the complexed state. Thesedata suggest that the protein of the relaxationcomplex is stable and probably is diluted to alower level only as a result of cell division.Similarly performed double-labeling experi-ments indicate that replication during the first15 min after glucose addition proceeds at a

8 -% *

4.k~'R7~~~~~~~.6

140 J. BACTERIOL.

on June 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

COLE1 DNA IN E. COLI

02032o 0 3

FRATIO NC/\

0-

cell grc d RCNC4-C

2

10 20 30 10 20 30

FRACTION NUMBER

FIG. 5. Sedimentation analysis of cleared lysates

to determine the effect of the addition of glucose to

the medium. Cleared lysates were prepared from

cells grown in the absence of glucose and then

shifted to a medium containing glucose as describedin the text. The cleared lysates were incubated inthe absence and presence of SDS and centrifuged(from right to left) through a 5 to 20%o sucrose den-sity gradient as described in Fig. 1. Cleared lysateswere prepared from the following cells. (a) Cellsgrown in the absence of glucose; (b) cells grown inthe absence of glucose and then in the presence ofglucose for 15 min.; c and d correspond to the sam-

ples a and b, respectively, except SDS was added tothese samples prior to centrifugation as described inFig. 1.

rate not detectably different from that of a

control culture with glucose remaining absent.Effect of inhibition of protein synthesis on

plasmid DNA synthesis. A 40-ml culture ofstrain CR34 (ColEl) cells was grown for sev-

eral generations in glycerol-Casamino Acidsmedium containing [14C]thymine (30,uCi). Inlog phase, the cells were pelleted and resus-

pended in 80 ml of medium containing [3H]-thymine (1.6 mCi) in place of the [14C]thy-mine. A 20-ml sample of the freshly resus-

pended cells was removed to serve as a controlwhile 30 ,ug of CAP/ml was added to the re-

mainder. The control cells were allowed togrow for one generation (90 min) and thenquickly chilled on ice. From the CAP-treatedcells, 20-ml samples were removed and placedon ice after 90, 180, and 360 min. A clearedlysate was prepared from each sample, treatedwith SDS, and analyzed by sucrose densitygradient centrifugation.As shown in Fig. 6a, the majority of the

ColEl DNA in the one-generation control cellswere in the complexed state as expected. Inaddition the 3H to 14C ratio at this point rep-resents an extent of DNA replication approxi-mately equivalent to a doubling of the amountof DNA that was present prior to the medium

shift. Fig. 6b through d indicate that, whilereplication of plasmid DNA was occurring, therelative proportion of noncomplexed ColElDNA increased dramatically. Fig. 7 illustratesthat, while the synthesis of chromosomal DNAwas essentially at a halt at a level equivalentto about 60% that of the one-generation con-trol, the plasmid DNA continued to replicateat an approximately linear rate. By 360 min,the extent of plasmid DNA replication repre-sented slightly more than a complete doublingof ColEl DNA. The ratio of 3H to '4C withinthe complexed and noncomplexed DNA wasnot significantly different for the time pointstaken, suggesting that the association and dis-sociation of plasmid DNA into and from thecomplexed state is a reversible and randomevent with respect to unreplicated and newlyreplicated plasmid DNA. Using the totalamount of "4C-labeled plasmid DNA (noncom-plexed plus complexed) in each gradient (Fig.6) as a normalizing factor, it can be seen (Fig.8) that the amount of 3H-labeled DNA ap-pearing in the complexed state reaches anearly constant value, while the amount in thenoncomplexed state continues to increase.This observation is consistent with the notionthat the amount of protein with which plasmidDNA can become associated to form complexremains at a constant level in the presence ofCAP and is relatively stable. In a separate butsimilarly performed experiment, a sample ofcells was analyzed after 18 hr of incubation inthe presence of CAP. At this time, the amountof ColEl DNA synthesized was almost equiva-lent to three times the amount found in theone-generation control cells. A concentrationof 12 ,g of CAP/ml was equally as effective inthe inhibition of the synthesis of the super-coiled DNA-protein complex.

In consideration of the differences in thepercentage of plasmid DNA molecules in thecomplexed state depending on whether glucoseis present in the medium, it was of interest todetermine the effect of inhibition of proteinsynthesis on ColEl DNA synthesis in cellsgrown in the presence of glucose. The experi-ment was carried out essentially as describedfor the analagous experiment in the absence ofglucose described above except that since thedoubling time is shorter in the presence of glu-cose the one-generation control represented a60-min time period; and samples of the CAP-treated culture were removed after 60, 120,and 240 min. As in the previous case, the rela-tive amount of plasmid DNA found in thecomplexed state decreased dramatically duringCAP-treatment (Fig. 9). It is apparent that

VOL. 110, 1972 1141

on June 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

CLEWELL AND HELINSKI

3H40

C9 200

x

"- 400

20r

N

x

(L

10 20 30 10 20 30

FRACTION NUMBERFIG. 6. Sedimentation analysis of cleared lysates prepared from cells grown either in the absence or pres-

ence of CAP. Cells were grown in the presence of ['4C]thymine in the absence of glucose and then shifted toa medium containing CAP and [3H]thymine and lacking ["C]thymine as described in the text. Cleared ly-sates were prepared, treated with SDS as described in Fig. 1, and centrifuged through 5 to 20%o sucrose den-sity gradients in an SW65 rotor (15 C) at 50,000 rev/min for 135 min. (0) 3H-labeled DNA synthesized afterthe medium shift; (0) "4C-labeled DNA synthesized prior to the medium shift. Cleared lysates were preparedfrom the following cells. (a) Cells grown in the presence of ["4C]thymine and shifted to a medium identical(with the exception of the isotope) to the preshift medium for one generation (90 min); (b), (c), and (d) cellsshifted to a medium containing CPA for 90, 180, and 360 min, respectively. The designations NC and RC aredescribed in Fig. 1.

200 NC

15 ec*NC C

~~-~- A Chr

to V

5L X *~~~~~~~Chr5 0Ci

2 3 4TIME (GENERATION EQUIVALENTS)

FIG. 7. Extent of replication of DNA found in thecomplexed and noncomplexed state in cells grown ineither the absence or presence of CAP. The ratios of3H to "C were calculated from the data presented inFig. 6. The values obtained for chromosomal DNAwere based on the amount of 3H and 14C in thecrude lysates. NC indicates noncomplexed ColElDNA; C indicates complexed ColEJ DNA; and Chrindicates chromosomal DNA. The solid points repre-sent DNA from the one-generation control cells; theopen points correspond to the CAP-treated cells.

I') L)

2 3 4

TIME (GENERATION EQUIVALENTS)FIG. 8. Relative amounts of incorporation of [SH]-

thymine into complexed and noncomplexed ColElDNA in cells grown in the presence of CAP. Thedata were obtained from the experiment described inFig. 6. The NC and C designations are described inFig. 1.

this change was due to an accumulation ofnoncomplexed plasmid DNA, while theamount of DNA in the complexed state re-mained at a constant level (Fig. 10). Onceagain the ratio of 3H to 14C of the plasmid DNA

1142 J. BACTERIOL.

on June 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

COLE1 DNA IN E. COLI

3H20

C? 100

x

, 20

10

10 20 30 10 20 30

C*J

bx

La-

FRACTION NUMBERFIG. 9. Sedimentation analysis of cleared lysates prepared from cells grown in the presence of glucose and

either in the absence or presence of CAP. Cells were grown as described in Fig. 6 except that glucose waspresent in the growth medium. Cleared lysates were prepared, treated with SDS, and centrifuged (from rightto left) through 5 to 20%/o sucrose density gradients in an SW50.1 rotor (15 C) for 100 min. (0) [3H]thymine-labeled DNA synthesized after the medium shift; (0) [14C]thymine-labeled DNA synthesized prior to themedium shift. Cleared lysates were prepared from the following cells. (a) Cells grown in the presence of ["C]-thymine and shifted to a medium identical (with the exception of the isotope) to the preshift medium; (b),(c), and (d) cells shifted to a medium containing CAP for 60, 120, and 240 min, respectively. The NC and RCdesignations are described in Fig. 1.

in the complexed and noncomplexed states issimilar (Fig. 11). In contrast to the previousexperiment where glucose was not present, it isobserved from the 3H to 14C ratios (Fig. 10)that at the end of 240 min the extent ofplasmid DNA replication represented a 2.5-fold increase over that found in the one-gener-ation control. In this and other similarly per-formed experiments, the rate of plasmid DNAsynthesis appears to increase during the firstfew hours of treatment. As indicated in Fig.10, the replication rate after the first hour wastwo- to threefold greater than that observedduring the first hour.

Additional experiments have indicated thatplasmid replication continues under theseconditions for approximately 10 hr to a pointrepresenting a five- to sixfold increase in thenumber of plasmid molecules per cell. Experi-ments with CAP concentrations as high as 280,gg/ml yielded similar results.Experiments involving amino acid starva-

tion as a means of inhibiting protein synthesisprovided results very similar to those from theCAP experiments. Upon removal of CasaminoAcids, plasmid DNA replication continued andnoncomplexed DNA accumulated. When glu-cose was absent in such an experiment, the

25 .

20 .

it) qt

15 .

o10

1 2 3 4TIME (GENERATION EQUIVALENTS)

FIG. 10. Relative amounts of incorporation of [SH1-thymine into complexed and noncomplexed ColElDNA in cells grown in the presence of CAP. Thedata were obtained from the experiment described inFig. 9.

* NC

- -C

VOL. 110, 1972 1143

s .

on June 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

CLEWELL AND HELINSKI

30

0 NC

25-

20-

1(-) o

10 NC *

Chr A

Chr5

2 3 4

TIME (GENERATION EQUIVALENTS)

FIG. 11. Extent of replication of DNA found inthe complexed and noncomplexed states in cellsgrown in the presence of glucose and either in theabsence or presence of CAP. The ratios of 3H to 14Cwere calculated from the data presented in Fig. 9.The values obtained for chromosomal DNA were

based on the amount of 3H and 14C counts in thecrude lysates. NC indicates noncomplexed ColElDNA; C indicates complexed ColEJ DNA; Chr indi-cates chromosomal DNA. The solid points representthe one-generation control cells; the open points cor-respond to the CAP-treated cells.

extent of replication was limited to approxi-mately one doubling of plasmid DNA. How-ever, when glucose was present, replicationcontinued to the extent of more than a three-fold increase in plasmid DNA. In the lattercase, the rate of plasmid replication was abouttwofold faster than that observed in a one-

generation control culture, thus confirming an

observation previously reported by Bazaraland Helinski (1).

DISCUSSIONColEl DNA is present in host E. coli cells to

the extent of approximately 24 copies per

growing cell. This value does not change underconditions where the amount of chromosomalDNA per cell varies. This finding indicates a

fundamental difference between ColEl DNAand chromosomal DNA with regard to the con-trol and maintenance of the number of thesereplicons.

The percentage of ColEl DNA that can beisolated as a relaxation complex of supercoiledDNA and protein is much lower when the cellsare grown in a medium containing glucosethan in a medium lacking glucose. When cellsare grown in the presence of glucose for severalgenerations and then shifted to a mediumlacking glucose, there is a rapid increase in thepercentage of ColEl DNA that is found in thecomplexed state. This increase is accompaniedby a two- to threefold increase in the rate ofsynthesis of ColEl DNA. Although both phe-nomena are inhibited if CAP is present, thesedata do not necessarily indicate a direct rela-tionship between relaxation complex and thereplication of ColEl DNA. In fact, a burst ofplasmid DNA synthesis might be expected if arapid decrease in cell mass were to occur inorder to maintain a constant number ofplasmid molecules per cell. DNA synthesis isnot required for the synthesis of the plasmidDNA-protein complex since thymine starva-tion did not prevent the shift to the complexedstate when glucose was removed. Experimentsutilizing nalidixic acid to inhibit DNA syn-thesis yield similar results (Clewell, unpub-lished data).

In experiments in which cells were grown inthe absence of glucose and then shifted to amedium containing glucose, no noticeable dif-ference could be detected in the amount ofcomplexed plasmid DNA present after 15 min.This supports the notion that the protein inthe complex is stable physiologically and anydecrease in its level is the result of dilutiondue to cell division.The effect of glucose suggests the involve-

ment of catabolite repression in the regulationof synthesis of complexed ColEl DNA. This issupported by recent data demonstrating thatthe effect of glucose can be reversed by exoge-nous cyclic AMP (Katz, Kingsbury, and Helin-ski, manuscript in preparation; Clewell, un-published data).

Inhibition of protein synthesis by CAP treat-ment or starvation for amino acids resulted ina gradual halt in chromosomal DNA synthesisconsistent with completion of a round of repli-cation. In contrast, however, plasmid DNAsynthesis continued to various extents de-pending upon the particular conditions em-ployed. In all cases, the synthesis of the pro-tein component(s) of the relaxation complexwas inhibited, whereas the protein present be-fore inhibition appeared to be stable. PlasmidDNA that was synthesized accumulated asnoncomplexed, covalently closed molecules.On the basis of data obtained from density-

1144 J. BACTERIOL.

on June 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

COLE1 DNA IN E. COLI

shift and amino acid starvation experiments, itwas previously proposed (1) that a randomselection of ColEl DNA molecules to be repli-cated in a cell takes place by a mechanisminvolving a limited number, or activity, of astable replicator(s) and that this replicatorwould be subject to regulatory control by themetabolic state of the cell. It has been demon-strated that ColEl DNA interacts specificallywith a protein substance (relaxation protein),the level of which depends on the metabolicstate of the cell. It has been suggested (5, 6)that this protein may be involved in the initia-tion of plasmid DNA replication on the basisof the assumption that a covalently closed cir-cular DNA molecule must at some time ac-quire a nick to allow separation of the tem-plate strands during semiconservative replica-tion and the potential role of the relaxationcomplex in the nicking event. It was also pro-posed that plasmid molecules found in thecomplexed state represent a "repressed" (orresting) condition with regard to replication.This was based on the assumption that if thetravel time for replication of a ColEl DNA mol-ecule is similar to that of the host chromo-some, then it would take approximately 4 to 7sec to complete a round of ColEl DNA repli-cation. Thus, in a randomly growing culture atany particular instant, the vast majority ofplasmid molecules are not in the process ofreplication. It is conceivable that the ColElDNA complex is part of an inactive replicatorthat is activated in vivo by a specific signal.During replication of ColEl DNA in the pres-ence of CAP or under conditions of starvationfor amino acids, it has been shown thatplasmid replication continues and the ColElDNA freely associates with or dissociates fromthe protein component(s) of the relaxationcomplex as predicted (1) for the proposed rep-licator.The timing of the ColEl DNA initiation

event does not appear to be tightly coupled tothe stage of replication of the host chromo-some since the initiation of plasmid DNA syn-thesis continues long after chromosomal DNAsynthesis has been stopped by CAP treatment.The increase in replication rate during aminoacid starvation (in the presence of glucose) (1)would also appear to reflect a replication con-trol process that at least under certain condi-tions can be uncoupled from the controlprocess for chromosomal DNA replication.However, it seems reasonable that undernormal growth conditions there is coordinationbetween host chromosome and plasmid DNAsynthesis since a constant number of plasmid

copies per cell is maintained during growth. Itis possible that the inhibition of protein syn-thesis may in fact produce a condition thatmore closely resembles the process of plasmidreplication during conjugal transfer of ColElDNA.Although the ColEl DNA relaxation com-

plex appears to have certain properties thatsuggest a role for this complex in the initiationof plasmid DNA replication, direct evidencefor this role has not been obtained. Using theextended CAP treatment which allows con-tinued synthesis of plasmid DNA in the ab-sence of chromosome synthesis, it has beenpossible to mutagenize ColEl DNA selectivelyand obtain conditional plasmid DNA replica-tion mutants (Kingsbury and Helinski, manu-script in preparation). An investigation is cur-rently underway to determine if any of theseplasmid DNA replication mutants are alteredwith respect to the properties of the relaxationcomplex of supercoiled ColEl DNA and pro-tein.

ACKNOWLEDGMENTSThis work was supported by Public Health Service re-

search grant AI-07194 from the National Institute of Allergyand Infectious Diseases and by National Science Foundationgrant GB-6297. Donald R. Helinski is a Public HealthService Research Career Development Awardee (K04-6M07821).We are indebted to Bernard Ashcraft and Joan Gains for

their invaluable technical assistance. We also thank DavidKingsbury, whose initial observation of different levels ofrelaxation complex in cells grown in different media led usto the study of the glucose effect, and Stephen Cooper andLeonard Katz for helpful discussions.

LITERATURE CITED

1. Bazaral, M., and D. Helinski. 1970. Replication of a bac-terial plasmid and an episome in Escherichia coli.Biochemistry 9:399-406.

2. Blair, D. G., D. B. Clewell, D. J. Sherratt, and D. R.Helinski. 1971. Strand-specific supercoiled DNA-pro-tein relaxation complexes: comparison of the com-plexes of bacterial plasmids ColE, and ColE2. Proc.Nat. Acad. Sci. U.S.A. 68:210-214.

3. Burton, K. 1956. A study of the conditions and mecha-nism of the diphenylamine reaction for the colori-metric estimation of deoxyribonucleic acid. Biochem.J. 62:315-323.

4. Clark, A. J., and A. D. Margulies. 1965. Isolation andcharacterization of recombination-deficient mutantsof Escherichia coli K12. Proc. Nat. Acad. Sci. U.S.A.53:451-459.

5. Clewell, D., and D. Helinski. 1969. Supercoiled circularDNA-protein complex in Escherichia coli: purificationand induced conversion to an open circular DNAform. Proc. Nat. Acad. Sci. U.S.A. 62:1159-1166.

6. Clewell, D., and D. Helinski. 1970. Properties of a su-percoiled deoxyribonucleic acid-protein relaxationcomplex and strand specificity of the relaxationevent. Biochemistry 9:4428-4440.

7. Clewell, D., and D. Helinski. 1970. Evidence for theexistence of the colicinogenic factors E2 and E, as

1145VOL. 110, 1972

on June 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

CLEWELL AND HELINSKI

supercoiled circular DNA-protein relaxation com-

plexes. Biochem. Biophys. Res. Commun. 40:608-613.8. Clewell, D., and D. Helinski. 1970. Existence of the col-

icinogenic factor-sex factor ColIb-P9 as a supercoiledcircular DNA-protein relaxation complex. Biochem.Biophys. Res. Commun. 41:150-156.

9. Cooper, S., and C. Helmstetter. 1968. Chromosome rep-

lication and the division cycle of Escherichia coli B/r.J. Mol. Biol. 31:519-540.

10. Jacob, F., S. Brenner, and F. Cuzin. 1963. On the regu-lation of DNA replication in bacteria. Cold SpringHarbor Symp. Quant. Biol. 28:329-349.

11. Kline, B. C., and D. Helinski. 1971. F, sex factor ofEscherichia coli. Size and purification in the form of astrand-specific complex of supercoiled deoxyribo-nucleic acid and protein. Biochemistry 10:4975-4980.

12. Okoda, T., K. Yanigisawa, and F. J. Ryan. 1960. Selec-tive production of thymine-less mutants. Nature(London) 188:340.

13. Radloff, R., W. Bauer, and J. Vinograd. 1967. A dye-buoyant-density method for the detection and isola-tion of closed circular duplex DNA: the closed cir-cular DNA in HeLa cells. Proc. Nat. Acad. Sci.U.S.A. 57:1514-1522.

1146 J. BACTERIOL.

on June 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from